Review on Soft Computing-Based Controllers for Frequency Regulation of Diverse Traditional, Hybrid, and Future Power Systems

Abstract

In recent decades the power system has become a complex network, to design a load frequency control (LFC) requires solving a complex equation. Optimisation techniques are essentially required to optimise the parameters of different controllers used for LFC issues in the power system. In a unified power system, the LFC is examined from all angles using different optimisation strategies to optimise the conventional PI, PID, cascaded, and fuzzy controllers as well as recently designed controllers. This manuscript specifically reviews the use of soft computing techniques in the frequency regulation of the power system with single/multiple areas that include conventional, renewable, and combinations of both, with FACTS devices and certain energy storage devices such as superconductor magnetic energy storage (SMES) and battery sources. Furthermore, deregulated power systems and microgrids are also considered for the study. To regulate LFC under various disturbances such as generation rate constraints (GRC) and dead band control, a few additional control approaches are utilised. Models of the power system are discussed and analysed. In addition, the merits and drawbacks of the studied techniques/structures that address design and implementation issues—as well as control issues that relate to the LFC problems—have been discussed.

1. Introduction

1.1. Motivation

The infiltration of renewable energy sources (RESs), the adoption of novel ideas like the smart grid, and the computerisation of power system regulators are based on apprehensive communication technologies and contributing to increase in the complexity of power systems [1]. The aforesaid factors have an unswerving impact on the security, stability, and functioning of the power systems. The voltage and frequency are the most crucial power system indices. Frequency regulation in PS has received a lot of consideration due to its importance [2]. Primary, secondary, and tertiary control are the standard divisions of frequency regulation. The governor droop often implements the principal frequency regulation, resulting in steady stated errors. The primary frequency control loop is accountable for recording the frequency variation before the under/over frequency protection relays are activated. The secondary control for controlling frequency in power systems is called AGC, often referred to as LFC. It has two main objectives: (i) keep the frequency within a desired range; and (ii) control the exchange of power via significant tie-lines between the two control areas.

The uncertainty of active power production is greatly raised as the level of infiltration of RESs—such as wind farms and solar plants in power networks—increases, resulting in frequency variations. Along with the stochastic nature of demand, this rise in active power fluctuation would cause the power system's frequency to oscillate sharply. Therefore, more reliable and ideal LFC techniques are required for use in future power systems in order to address such issues.

In interconnected power systems, there have been numerous control strategies proposed for LFC. Four categories of strategies can be made: (i) the design of PID controllers for regulating frequency and tie-lines power flows is the primary emphasis of classical control approaches; (ii) modern controls methods comprise best control strategy, i.e., adaptive control and sliding mode control; (iii) intelligent control strategies, i.e., fuzzy and neural network based controller; and (iv) soft computing-based methods for fine-tuning the parameters of controllers, which researchers have given a lot of attention to recently.

1.2. Contributions

Power system operation and control is a broad area of study in the area of power systems that include controls for both frequency and voltage stability. These days, researchers are paying a lot of attention to the problem of frequency management because of the reasons outlined in Section 1.1. However, this topic encompasses a number of in-depth and specific subjects, power system inertia support techniques, main frequency control, secondary frequency control, and tertiary frequency control, as well as protection during frequency emergencies.

The LFC deals with regulating the frequency due to minute disruptions like load fluctuations and variations in RESs. In this study, we focus on the key details, namely LFC models, control techniques, and optimisation techniques used for controller parameter tuning. For researchers, frequency response models are crucial. Therefore, this paper introduces the most significant frequency response (FR) models and categorises them into two primary classes, namely models with emerging technologies and model structures (referred to as conventional). This study evaluates the LFC control strategies of the last ten years. The article emphasises the benefits and drawbacks of each control technique and optimisation technique in this review. Moreover, a comparison of a few techniques is shown clearly. Additionally, this study introduces the research directions and gaps, which might serve as a useful road map for researchers.

The suggested optimisation techniques for the controller parameters tuning for LFC are highlighted in this review. Furthermore, a survey of the proposed frequency response model for power systems is conducted. In addition, LFC models found in contemporary power systems, microgrids, and upcoming smart grid models are investigated. The same goes for trends and potential avenues for future research. The frequency response modeling of various power system topologies is initially discussed due to its significance in LFC investigations.

Table 1. Representation of power systems: conventional, hybrid, and future power systems.

	Power System Models for Optimal Frequency Regulation
S. No.	Conventional Power System	RESs Integrated Power System	Islanded Microgrid(IMG)/Smart Grid
1	Single area	Single-area	Single-area IMG
2	Multi-area (Two area, Three Area, Four area, Etc.)	Multi-area (Two area, Three Area, Four area, Etc.)	Two-area IMG
3	Deregulated PS system	Deregulated PS	Distributed energy sources (DGs)
4	PS with HVDC	PS with HVDC	---
5	PS with FACTS devices	PS with FACTS devices	--
6	PS with energy storage	PS with energy storage	--

illustrates how LFC models are split into various categories—namely, traditional, hybrid, and future LFC models. Based on their configurations, the typical LFC model structures are examined.

As a result, a thorough review of dual-area, conventional three-area and multi-area PS—as well as FR models of signal control area—are conducted. Many distinct system models and formations have been published in the literature, taking into account various generating unit types as thermal, nuclear, hydro, and gas. The formations of present and future PS models for FR are then split into main categories: AGC models with high voltage direct current (HVDC) links and FACTS devices, LFC of deregulated PS, LFC models with significant penetration of distributed generation and RESs, LFC in microgrids, and LFC models appropriate for smart grids.

2. Survey of Various LFC Models and Optimisation Techniques

In this section, the term conventional power systems refers to power systems where electricity is produced using fossil fuels. In fact, the most renowned power plants for these systems are thermal, hydroelectric, and nuclear generating units. PS are typically split into single-area, two-area, three-area, four-area, and multi-area power systems as per their size and interconnections. The power system models given in Table 1 are designed for conventional as well as the deregulated power systems. The primary distinction between a traditional monopolistic electricity market and the emerging competitive deregulated market is that in the former, electricity is only considered to be a component of the energy supply, whereas in the former it is viewed as a service and thus marketed similarly to other commodities. Several FR models, control strategies, and optimisation techniques for LFC models are suggested in the literature. According to no free lunch theorem, no optimisation technique can be used to solve all problems that mean a particular optimisation technique will be better suited for one kind of solution but may not be suitable for another. This means that there is always a search to find a suitable optimisation technique for a particular type of optimisation problem. A thorough assessment of the optimisation techniques used for optimizing the different controller parameters of PS models for LFC is provided in the following sections.

2.1. Single-Area Power System

Single area power system is a closely knit electrical area in which numbers of generators form a cohrent group. Figure 1 shows the representation is single area power system that consists of a governor, turbine, and generator. The proposed AGC method for SAPS and optimisation techniques are investigated in

Table 2. Comparison of latest control strategy applied for LFC in the single-area power system.

Ref. No.	Power System Type	Sources	Devices	Optimisation Technique	Controller
[3]	Conventional	Thermal Unit	----	Improved Particle Swarm Optimisation (IPSO)	PID
[4]	Conventional	Thermal, Hydro, and Gas units	----	Genetic Algorithm (GA)	PI
[5]	Conventional	Thermal Unit	----	Quadratic Regulator Approach with Compensating Pole (QRAWCP)	PID
[6]	Conventional	Thermal, Hydro, and Gas units	HVDC Link	Modified PSO	PID
[7]	Conventional	Thermal Unit	----	Iterative Linear Matrix Inequality	iterative PID (IPID)
[8]	Conventional	Thermal Unit	----	Ant Colony Optimisation (ACO)	PID Controller
[9]	Conventional	Thermal Unit	----	Elephant Herding Optimisation (EHO)	PID
[10]	Hybird power plant	Thermal, wind units	----	Bat Inspired Algorithm	LADRC
[11]	Microgrid	Reheat thermal, diesel engine generation (DEG), wind	Battery energy storage (BESS), Fuel cell (FC)	Grey Wolf Optimiser (GWO)	Fuzzy PID
[12]	Microgird	DEG, PV, wind	BESS, FC	GA	MPC and PID
[13]	Microgrid	Wind turbine, BDG	Mini Pumped hydro energy storage and BESS	Coyote-Optimisation Algorithm	PI controller

.

The design and implementation of LFC for single-area power systems (SAPS) is the subject of the first studies on frequency control. In the literature [14,15,16,17,18], a number of SAPS models with LFC control techniques are examined. A dynamic mathematical model of frequency response for SAPS made up of thermal power plants is designed and a robust frequency control technique is suggested in [19]. Single-area LFC models for power networks with several energy sources—including hydro, gas, and thermal sources—is presented in [17]. A well-described frequency response LFC model of hydroelectric power systems is found in [20]. The improved PSO (IPSO) in Reference [3] is used for the AGC issues of SAPS.

2.2. Two-Area Power System

A two-area power system is constructed using two different control areas connected through a tie-line. The linearised two area power system is presented in Figure 2. A summary of AGC used in two-area power systems (TAPS) is provided in [21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38]. References [21,22] study the influence of tie-lines models on the LFC of TAPS. In [25], the LFC models for TAPS that take voltage control loop effects on the frequency response are developed. In [27,28,29], frequency response models for TAPS that take into account the nonlinearities of the GRC and the GDB are proposed. References [23,38] present a discussion on how to simplify the frequency response model by reducing its complexity. The multi-source, two-area LFC models that take nonlinearities into account are highlighted in [29]. LFC models of TAPS with parametric and nonparametric disturbances are discussed in references [27,29,36]. In [25], an LFC strategy is proposed for TAPS linked by HVDC/DC transmission lines. References [39,40] present TAPS frequency response models made up of reheat-thermal turbines coupled by AC/DC links. For thermal–thermal two-area power systems, load frequency control techniques that account for communication channel delay are proposed in [21,22,23,24,25]. Two-area power systems reheat thermal turbine frequency model with governor dead-band zone is shown in [30]. GRC non-linearity for the reheat thermal turbine-governor system in TAPS is taken into consideration in [41,42]. Considering the non-linearity of hydro power plants, the LFC strategy for hydro–hydro power systems is presented in [17,43,44]. SMES units are offered as LFC models for TAPS in [45,46,47]. In view of the influence of batteries and SMES, a frequency regulator for TAPS is provided in [48]. In reference [49], an LFC model of traditional two-area PS is proposed, together with the involvement of energy storage systems and electric vehicles. In [50], the LFC models take into account the electrical load's stochastic nature. In [51,52], the LFC model accounts for the uncertainty associated with RESs. The various optimisation techniques used for tuning the different controller parameters and other details for two-area PS are discussed in

Table 3. Comparison of latest strategy applied for LFC in the different power system models.

Ref. No.	Power System Type	No. of Areas	Sources	Devices	Optimisation Technique	Controller
[53]	Conventional	1,2	Hydro-thermal gas Units	HVDC	Differential Evolution (DE) algorithm	I, PI, PID
[54]	Conventional	2	Thermal unit	----	Particle Swarm Optimisation	PID controller
[36]	Conventional	2	Thermal	----	Differential Evolution	2-DOF PID
[55]	Conventional	2	Hydro-thermal gas	----	Teaching Learning Based Optimisation (TLBO)	PID controller
[56]	Conventional	2	Thermal	HVDC	Multi-Verse Optimiser	Fuzzy-PID controller
[26]	Conventional	2	Hydro-thermal gas	UPFC and SMES	Firefly Algorithm	Fuzzy-PID
[54]	Conventional	2	Hydro-thermal gas	(GCSC)	Sine Cosine Algorithm	Optimised fuzzy fine-tuning controller
[55]	Convetional	2	DEG and TTG	Additional inertia	Imperialist Competitive Algorithm (ICA)	FOPID controller
[56]	Conventional	2	Hydro-thermal gas units	HVDC, SMES	Bull-Lion Optimisation	PI-PDF controller
[57]	Deregulated	2,3	Thermal–hydro	--	Imperialist Competitive Algorithm (ICA).	PID Controller
[58]	Deregulated	4	Thermal–hydro	----	Binary Coded Genetic Algorithm (BGA) and Real-Coded Genetic Algorithm (RGA)	PID controller
[59]	Deregulated	2	Thermal–thermal	TCSC	Quasi-Oppositional Harmony Search (QOHS)	PID controller
[60]	Deregulated	2	Thermal-gas power plants	TCPS, SMES	Teaching Learning Based Optimisation and Pattern Search	`TID controller
[61]	Deregulated	4	Thermal units	----	Water Cycle Algorithm (WCA)	CC-TI-TID
[62]	Hybrid PS	2	Thermal–thermal, PV power model	----	Genetic Algorithm (GA)	Fuzzy logic controller (FLC), PID
[63]	Hybrid PS	2	Diesel generator, photovoltaic power model	----	Bacterial Foraging Algorithm	PI, PID, Fuzzy Logic Controller
[64]	Hybrid PS	2	Thermal, wind, photovoltaic	--	Salp Swarm Algorithm (SSA)	PID controller
[65]	Hybrid PS	2	Thermal and wind	----	Coyote Optimisation Algorithm (COA)	Cascaded PDn-PI
[66]	Microgrid	2	Diesel generator, wind and solar	SMES, BESS	Social-Spider Optimiser (SSO)	PID controller
[67]	Microgrid	2	BDG, wind, tidal units	----	Yellow Saddle Goatfish Algorithm (YSGA)	PIFOD-(1 + PI)
[68]	Microgrid	1,2	Thermal	----	Electro-Search Optimisation (ESO) with Balloon Effect	Adaptive controller

.

For deregulated electricity networks, various control strategies have been presented as the restructuring concept is adopted. In [69,70,71,72,73,74,75,76,77,78,79,80], LFC issues for deregulated power networks are discussed. A power system is separated into sections under deregulation so that each section has its own administrator. An independent system operator (ISO) is responsible for overseeing power distribution companies (DISCOs), power transmission companies (TRANSCOs), and power generation companies (GENCOs) in this regard [70,71,72,73]. The GENCOs may or may not take part in LFC service in this new setting. In deregulated power networks, the provision of auxiliary services is based on a competitive electricity market [74,75]. For several kinds of deregulated electricity systems, frequency response models have been established in the literature [70,71,73,74,75,76,77,78]. LFC strategy for deregulated power systems with only thermal units are proposed in [75,76,77]. References [60,81] offer LFC strategies for deregulated hydro power networks. An LFC framework for deregulated multi-source power system is depicted in [82]. For LFC research, a restructured TAPS with both a thermal–gas and thermal–nuclear system is depicted in [83,84].

2.3. Three Area Power System

LFC modelling for three-area interconnected power systems (TAIPS) are presented in [85,86,87,88,89,90,91,92,93]. An LFC model for a TAIPS is provided in [94], where steam-hydro power units are taken into account in the first and second areas, while a steam power unit is the only source of energy in the third area. In [93,95,96], a thermal power system with three control areas is examined. In three-area interconnected power systems, radial and ring networks among the various control areas are explored in [97]. In [72,98,99], a three-area power system LFC model that takes GDB and GRC nonlinearities into account is proposed. Reference [100] investigated the effects of communication latency on LFC in TAIPS. The effects of parametric uncertainty on LFC of three-area linked power systems are highlighted in References [98,101]. Three-area thermal power systems frequency control model is proposed in [102,103]. The LFC for hydropower systems with three areas is depicted in [58,104]. Three-area hydro-thermal power system load frequency controllers are presented in [101]. The LFC for multi-source power systems that take into account thermal, gas, and hydro energy sources presented in [105].

2.4. Four-Area Power System

To keep the frequency within a permitted range, large power systems are typically separated into various control regions. The LFC issues in four-area interconnected power systems (FAIPS) has been presented in [100,103,106,107,108,109,110]. An LFC for FAIPS with hydro power units as the first effort in this sector is presented in [73]. Furthermore, the frequency response models of FAIPS that are appropriate for LFC are presented in [111]. In [110], an LFC strategy for interconnected multi-area power systems considering nonlinearities like GRC and GDB is presented. Using a fuzzy control for an LFC model, the uncertainties of the power system characteristics are taken into account in [112]. A FAIPS with various energy sources and turbines—including hydro, gas, non-reheat thermal, and reheat thermal power plants—is presented in [95]. LFC scheme for FAIPS using various structures, such as longitudinal and ring connections is proposed in [109,110]; in this model, three area contain thermal units and one area contains a hydro unit.

2.5. Multi-Area Power System

In nature, power systems are linked together. A larger system known as the multiarea system is created by connecting various single areas together. A six area power system consisting diverse power generating sources is presented in [113,114,115,116]. AGC models are provided in [116]. For various restructured power system designs, such as five-area and multi area power systems [75,79,107,117,118]. In reference [119], a TAIPS with two GENCOs and DISCOs in each area was designed. In [120], the idea of a restructured TAIPS that is vehicle-to-grid (V2G) enabled is put forth. References [121,122] presents a discrete mode LFC method for a deregulated TAIPS. Furthermore, handling deregulation of the triple-area and four-area power systems is shown in [71,102].

2.6. Microgrid and Smart Grid

A microgrid (Figure 3) is a tiny power grid that can function on its own or in tandem with other microgrids. Energy generation that is distributed, dispersed, decentralised, or embedded refers to the use of microgrids. LFC models of fuel cell, wind, and solar-powered electric hybrid power systems is presented in [123,124,125,126,127,128,129,130]. Modelling the LFC takes into account a microgrid and hybrid system made up of PV, WTG, micro turbines, fuel cells (FC), capacitive energy storage (CES), and aqua electrolysers (AE). Active power flow and frequency stabilisation is achieved using different optimisation based PID controller [131,132]. In [133,134], they explore how electric water heaters, demand response, and electric vehicles affect the LFC of microgrids. For LFC research, nonlinear microgrid models are also proposed in [135].

Due to its many benefits, smart grid control themes have recently attracted a lot of attention from researchers. For use in upcoming smart grid systems, several new LFC techniques have been devised in [126,136,137,138,139,140,141,142,143,144]. LFC models that take electric cars’ (Evs’) contribution into account are presented in [139]. Similarly, a coordination mechanism between heat pump water heaters and electric vehicles is recommended for LFC in upcoming smart grids [140]. Adaptive dynamic demand response, which is seen as LFC has been given significant smart grid functionality in [126,136]. Frequency response models that take into account the impact of various storage types have been proposed in [13,145,146] and evaluated. In [143,147,148,149], a novel plug-in EV frequency response model is put forth for the primary and secondary frequency control levels. Future smart grids may experience issues with potential cyber-attacks on LFC systems, as stated in [142,150,151,152].

3. Soft-Computing-Based Controller for LFC

In order to maintain a steady system frequency, a power generating module has the capacity to modify its active power output in response to a detected system frequency deviation from a setpoint. Furthermore, the LFC is classified as: (i) primary frequency control; (ii) secondary frequency control; and (iii) tertiary control.

(i)

Primary frequency control: Changing a generating unit's power versus frequency in accordance with its static generation characteristic as stablished by the speed governor settings constitutes primary control. Restoring generation and demand balance within the synchronous area at a frequency other than the nominal value is the goal of primary control. This is done at the expense of the rotating masses of the producing units and associated motors' kinetic energy. After the generation and demand balance has been disturbed, the primary control action time is 0 to 30 s.

(ii)

Secondary frequency control: Utilizing a central regulator, secondary control modifies the active power set points of generating sets that are subject to secondary control in order to simultaneously restore the system frequency to its set-point value and power interchanges with neighbouring control areas to their planned values. Secondary control assures that the complete reserve of primary control power activated will be made accessible once again by changing the operating points of individual producing units. A few minutes pass more slowly under secondary management than under primary control. It starts acting roughly 30 s after a disruption or event and finishes within 15 min.

(iii)

Tertiary control: Tertiary controls are any automatic or manual adjustments made to the operating points of the participating generating units in order to replenish a sufficient secondary control reserve or to provide the desired (from an economic perspective) distribution of this reserve among the active generating units. Altering the set operating points of thermal power plant generation sets, around which primary and secondary control acts, connecting or disconnecting pump storage hydro power stations operating in an intervention mode, changing the power interchange programme, and controlling load are all examples of tertiary control.

Due to its great advantages, revolutionary soft-computing-based control system design has attracted a lot of attention from researchers over the past 15 years. The distinguished advantages of using soft computing techniques include their less solution costs, assurance of a solution and their viability. They can deal with difficult, nonlinear, and uncertain technical problems. In contrast to other methods, numerous researches have supported the viability of controller constructed on soft computing technique.

The LFC’s parameters have been optimised using soft computing approaches in order to obtain good dynamic response. To best tune the gains of the controllers, lots of evolutionary optimisation algorithms have been applied. The classification of the various soft-computing-based controllers is presented in the Figure 4.

3.1. Conventional Controller

The conventional control phenomena rely upon the traditional integer order (IO) controllers, which are frequently used to leverage the governor and minimise the ACE/μACE in the power system to improve FR. The conventional controller is designed with various combinations such as Integral (I), Proportional Integral (PI), Proportional Integral Derivative (PID), Proportional Integral Derivative with filter coefficient (PIDN). The conventional PID and PIDN controllers are presented in Figure 5a,b. Several research papers used I/PI/PID controllers in conventional, deregulated, microgrid, and smartgrid environments to deal with LFC issues. A basic integral controller tuned with Bacterial Foraging-Based Optimisation [153], iterative linear matrix inequalities techniques by the mixed H₂/H_∞ technique based decentralised proportional-integral (PI) controller [154], biogeography-based optimisation (BBO) tuned PID and fuzzy-PID four-area power system [155], conventional PID controller tuned by various optimisation techniques (i.e., hybrid genetic algorithm-simulated annealing (GA-SA), and chaotic optimisation algorithm (COA) and ant colony optimization are proposed in [156,157]; and Hosoke-Jeeve’s optimisation tuned double mode PI controller [158] tuned AGC are examined for regulating the system frequency and tie-line power fluctuations. Reference [159] proposes a frequency-regulation PID regulator. In order to resolve the LFC issues in connected power systems, GA is utilised in power systems for LFC controller gains as well as controller tuning [160]. A model based on population and inspired by the cooperative behaviour of fish schooling or bird flocking—PSO—is frequently used [57]. A PSO-based PID controller is used in [3,161] to overcome the LFC problem in single-area power systems. Similar to this, a PSO-based PID controller has addressed the LFC issue in interconnected power systems with different power units, including thermal, hydro, and gas turbines in [45]. A combination of two algorithms PSO-GSA is used for tuning the parameters of PI-PD controller in multi-area hybrid power system [162]. In [163], hybrid PSO strategies combining other soft computing techniques are also suggested for LFC.

Recently, various novels proposed soft computing techniques addressing the LFC issue in both traditional and contemporary power systems have been created consider the algorithm for differential evolution (DE) [119,164,165]. The flower pollination algorithm-based PID controller is proposed in for LFC in three area power system [99]. The QOHS algorithm has been employed for FR in single-area and three-area thermal-generation-based power system [89], simlarly the QOHS algorithm has been used for hybrid diesel-tidal generation based PS [166]. A Teaching Learning Based Optimisation (TLBO) tuned PIDD controller is proposed for two-area PS in [48]. A modified bias (MB) and coefficient diagram method (CDM)-based PID regulator has been used for FR in an independent microgrid (MG) system under full unfavourable conditions and Grey Wolf Optimiser algorithm is used for tuning the controller parameters [167]. A Bat-Inspired Algorithm (BA) optimised PD controller has been deployed for single-area PS considering the high penetration of wind [10]. A GWO algorithm is considered for optimizing the PID regulator parameters for solar- thermal generation based three-area PS [168]. A bacterial foraging optimisation based integral controller is applied to a three-area power system in [169]. Mouth-Flame Optimisation (MFO) tuned PIDF controller is used for renewable based two-area hybrid power systems [170].

3.2. Fractional Order Controller

Researchers have used so many optimisation techniques-based FO controllers to handle LFC problems since they have advantages when compared with the IO-based classical controller. The FOPID controller is an improvement over the conventional PID controller that offers non-integer integro-differential order selection in addition to the traditional PID controller gains. The FOPID model is presented in Figure 6. A Lion Algorithm-based FO-based PI controller has been proposed for LFC in dual-area system made up of hydro-thermal-gas power generating units in [171]. Alomoush introduced the use of FOPID controllers in LFC problems in 2010 and established a performance comparison between the use of FOPID controllers and IOPID controllers in dual-area conventional power networks [172]. The parameters of the controller were tuned with the help of MATLAB optimisation toolbox. Furthermore, the LFC of a single-area conventional system with a non-integer BB-BC algorithm-based FOPID controller is presented in article [173]. Reference [76] provides the design of a Bacterial-Foraging-based FOPID controller for FR of a reconstructed triple-area thermal power system. The LFC of a two-area reconstructed PS utilising a decentralised BB-BC based FOPID controller is investigated in [174] with an expansion of the use of FOPID controller. How the FOPID controller compares to existing IO-based I/PI/PID controllers with respect to the dynamic performance of the system has also been examined. The paper [175] proposes a reheat, non-reheat, hydro turbine based LFC single-area PS using a FOPID controller. The proposed controller's sensitivity is established during parametric uncertainty. Furthermore, the impact of SMES on two agent restructured PS using bat optimisation tuned FOPID controller has been investigated in [176]. Grasshopper optimisation algorithm (GOA)-based FOPID has been deployed for LFC in higher-order power systems [177]. A FOPI controller is proposed for frequency regulation of a dual-area restructured power system [178].

A novel reduced order LFC technique that uses a non-integer FOPID controller is presented in [179]. Utilising a newly created fractional order IDF (FOIDF), the triple-area multi-source reconstructed power network's system frequency and tie-line power are stabilised. A tuned IO-based PIDF controller is used to compare the system dynamics [180]. Furthermore, an effective LFC of a single-area microgrid utilising a self-tuned non-integer FOPID controller [181]. The design and implementation of a MOEO-optimisation-based FOPID controller for the LFC in the isolated renewable hybrid microgrid is proposed in [182]. A modified Black Hole Optimisation algorithm based stochastic non-integer controller Fractional-Order Fuzzy proportional-integral-derivative (MOFOFPID) is built up for a hybrid microgrid in [132].

3.3. Tilt Integral Derivative (TID)/Integral-Tilt Derivative (I-TD) Controller

Tilt integral derivative (TID) as a PID-type feedback control system compensator is offered, replacing the proportional component with a tilting component with a transfer function s^−1/n. A better feedback controller is achieved because the whole compensator's transfer function comes closer to an optimal transfer function [183]. The TID controller is presented in Figure 6. An intelligent, efficient, and reliable LFC scheme is necessary in today's complex and integrated power systems using RESs to maintain frequency and tie-line power flow within a tolerable range. A novel LFC approach using a dual-stage controller designed using fractional-order tilted-integral-derivative (TID) and integer-order PD controllers is proposed [30]. Furthermore, SSA has been deployed to tune the gains of the controller. Additionally, an optimal LFC structure summarizing the idea of fractional-order based (TID) controller. The parameters of controller are optimised using same SSA algorithm in [184]. In-depth analysis has been provided to support the CC-TID controller's superiority over other widely used state-of-the-art controllers in terms of several performance indices for the LFC in hybrid energy distributed power system. Crow search algorithm is used with chaotic mapping (CCSA) for optimizing the controller gains. In [185], a combination of FOPID and TID controllers is presented for the LFC in multi-area power system. In addition, the controller and SMES unit gains are tuned through a new manta ray foraging optimisation algorithm (MRFO). In [186], a TID controller has been proposed for LFC of a two-area interconnected power system. The optimum value of gains of the TID controller is calculated using constrained nonlinear optimisation. a new integral-tilt-derivative (I-TD) controller (Figure 7), optimised by a prevailing heuristic optimisation and known as ‘Water Cycle Algorithm’ (WCA), is anticipated for the LFC of a two-area, thermal-hydro-nuclear power units [187].

3.4. Cascaded and Dual-Stage Controller

Cascaded controllers were used by numbers of researchers for minimizing the frequency tie line power deviation. In cascade control, there are two controllers, one controller located inside the feedback loop of the other. The output of the first controller serves as the set-point for the second controller. This type of controller handles the distrubances more effectively. This type of controller can respond to disturbances more efficiently. When there are multiple measurements but only one control variable is accessible, cascade control is employed. The basic configuration of cascade PI-PID and PI-PD controller is presented in Figure 8 and Figure 9. Dash et al. present a Bat Algorithm (BA)-based PD-PID controller for the LFC system in [188]. Reference [189] used a Grey Wolf Optimisation-tuned PD-PID controller to regulate the frequency of three-area thermal reheat system. In a dual-area thermal-hydro with DG system, the use of a SCO-based PD-PID controller for the LFC issue of hybrid power system is explored in [190]. Performance of the suggested controller in comparison to a PIDF controller is also presented. Reference [191] provides an illustration of the cascaded PI-PD controller's involvement in the LFC scheme four-area thermal plant. In this context, a triple-area thermal-hydro-gas plant uses hybrid stochastic fractal search and pattern search optimisation based cascaded PI-PD controller with plug in EV proposed in [192]. A Marine Predator Algorithm (MPA)-based cascaded PIDA (Proportional-Integral-Derivative-Acceleration) controller was used to deal with LFC issues in [193]. In [135], a GOA-optimised FOPID-(1+PI) regulator is employed to regulate the frequency and tie-line power in single-area multi source microgrid.

A new Stochastic Fractal Search (SFS)-optimised I-PDF controller was designed in [194] to manage frequency fluctuation in triple-area (wind, thermal, and hydro) power plants. In a traditional dual-area thermal network, a Whale Optimisation Algorithm (WOA)-optimised PIDN-FOPD controller is employed for the same described problem [195]. The WOA-optimised PIDN-fractional order integral-derivative (PIDN-FOID) controller is now first equipped for LFC issue in dual-area thermal-gas system as a result of this extension. By evaluating the performance of various non-cascaded I/PI/PIDN controllers, the superiority of the controller is determined [195]. A novel multi verse optimisation (MVO to tune the combination of FOPI and FOPD controller for multi-area power system consisting hydro, thermal, and gas units in each area and the system is examined in two prospects—i.e., with and without HVDC link—to realise practical situation [196]. An Imperialist Competitive Algorithm (ICA)-tuned fuzzy PIDN-fractional order integral (FPIDN-FOI) regulator for minimizing the frequency oscillation in multi-area system is presented in [197]. The equilibrium-optimiser-based cascade FO-3DOF-PID and tilt-integral-derivative (TID) regulator for LFC in hybrid power system consisting renewable energy source integrated has been proposed in [198].

A recently designed sine-cosine algorithm optimised FOPI-FOPD regulator is used in a dual-area and triple-area reconstructed PS for AGC context by extending the FO based cascaded controller and analysis is also done on the proposed controller's superiority with regard to settling time and peak overshoot, undershoot of frequency deviation [199]. According to [82], a Volleyball Premier League (VPL)-optimisation tuned deregulated dual-area plant was accomplished using s two-degree-of-freedom proportional-integral-FO-based proportional-derivative with derivative filter (TDF-PI-FOPDN) and a fair comparison has also been demonstrated with an IO-based integral and PID controller. For triple-area LFC systems, the fundamental TDF-based cascaded (TDF-CC) controller is proposed in [200] and its performance is compared with cascaded/non-cascaded I/PI/ PID/TID/PI-PD controllers. The same Salp Swarm Algorithm (SSA)-optimised cascaded CC-TID controller is used for the same context with the increase of the control areas [201]. The Artificial Hummingbird Optimiser Algorithm (AHA)-tuned tilt fractional-order integral-derivative with fractional-filter (TFOIDFF) is proposed for renewable integrated hybrid power system in [202]. In [203], a new cascaded SSA tuned fuzzy-PI-TI controller is deployed for AGC of hybrid PS. In [204], a novel cascade fuzzy-noninteger CFPDμF-PI control strategy has been proposed to manage the frequency deviation that occurs due to the existence of RES’s units in the power system. In the CFPDμF-PI controller, FPDμF as a primary and PI as a secondary controller and Slime Mold Algorithm (SMA) are used collectively as the optimiser for controller gains. A Harris’ Hawks Optimisation-tuned FO3DOFTID controller is proposed for multi-area PS in [205].

3.5. AI Approaches

Artificial neural network (ANN), adaptive neuro fuzzy (ANFIS), and fuzzy logic-based controllers are commonly designed for the different LFC strategy.

3.5.1. ANN Approach

An extended control strategy for artificial neural networks (ANNs), centered on the idea of connected information processing phenomena, links the received net input neuron and output neural signal in a non-linear (connected with weighted sum) connection. In order to produce better results in complicated power networks, ANN can be used for linear and non-linear LFC problems. With appropriate controller management, the ANN technique has been used in linear and non-linear power systems to restore the system frequency as proposed in [104,206]. In [207], an application of ANN for three area PS is illustrated in [207]. A linearised reheat thermal TAPS LFC strategy with ANN support has been built in [208]. Reference [209] shows the implementation of the artificial model predictive (AMP) method for controlling the frequency of the dual-area linearised power system. In [210], they discuss the simulation of a fuzzy logic technique for reinforced learning for triple-area restructured power networks, and it is shown how to use the multi-surface perceptron neural structure (MLPNN) for the LFM problem for the dual-area restructured power network. Reference [211] takes into account poolco and bilateral transactions in traditional multi-area electric power networks when studying triple-surface feed-forward artificial neural networks for frequency stabilisation. A PSO-based ANN controller for LFC of an EV-integrated microgrid is presented in [161].

3.5.2. Neuro Fuzzy and Adaptive Neuro Fuzzy Approach

The use of the neuro-fuzzy approach for LFC has been demonstrated by a number of researchers. Studies [212,213], which concentrated on a dual-area LFC analysis under a recurring fuzzy neural network based feed-forward controller, have acknowledged the involvement of neuro-fuzzy networks in LFC issues. Dual-area thermal-hydro plant simulation results plotted under a hybrid neuro-fuzzy controller are proposed in [213]. For conventional FAIPS, Gheisarnejad et al. [111] analysed the Fractional order-fuzzy PID control technique and examined its performance for a two-area conventional power system; an intelligent neuro-fuzzy technique has also been examined. It is framed in [214] how a fuzzy-neural net tuned controller is used for the LFC issues of a dual-area reheat turbine network. In [215] an ANFIS approach with SMES-TCPS is proposed for LFC of three unequal area of power.

Similar to this, researchers have grown interested in the adaptive neuro-fuzzy interface topology (ANFIT) that is based on the conventional feed-forward technique and has no synaptic weight [122]. In [58], the functionality of an ANFIT-based controller for the AGC of a three-unequal thermal-hydro plant has been presented. The paper [216] provides an illustration of the use of an adaptive ANFIT-PID controller in a single-area electric power network. A design of an automatic power frequency regulator using an integrated four-area network of renewable resources (wind and solar PV) for frequency stabilisation is presented in [217]. In reference [218], they introduce a relative performance analysis of the ANFIT controller for the LFM problem of a six-area network consisting of super-capacitor (SC) and FACTS devices. A PSO-tuned FO-Fuzzy-PID controller is applied to a hybrid power system in [219] they present how a non-integer controller is applied to the controlled system’s frequency.

3.5.3. Fuzzy Approach

Due to its sensible wide operating range and incredibly rapid response time, an efficient fuzzy logic approach for regulating system dynamics has rekindled attention among researchers. For non-linear multi-area power system frequency regulation, Chang et al.’s initial study in [220] monitored the PI controller using fuzzy logic automatic power frequency management (APFM) of linear or non-linear power system. It is discussed in [49,112,117,221] and involves monitoring various traditional as well as non-integer controllers with fuzzy logic assistance. An ideal decentralised Water Cycle Algorithm (WCA)-tuned type-II FLC for frequency stabilisation of conventional electric power systems with GRC and GDB nonlinearities is presented in [222]. A design of the online control of type-II fuzzy technique being used for the LFC of a dual-area restructured power network is illustrated in [223,224]. Reference [225] uses a relative analysis of type-I and type-II fuzzy approaches for the APFM of a four-area classical network. The Big-Bang Big-Crunch algorithm is used for tunning the membership function and controller parameters of the proposed controller. Furthermore, a multi-area electric power network’s optimal LFC operation employing a neuro-fuzzy adaptive gain scheduling controller technique is examined in [58]. In [225], a type-II fuzzy system is developed for the RES’s integrated microgrid. A type-2 (T2) fuzzy controller has been suggested for LFC of two-area PS with the considering the GRC. The performance of the controller is compared with PI controller and type-1 fuzzy controller [75]. To address the LFC problem, ref. [226] proposes an adaptive fractional order (FO)-fuzzy-PID controller for LFC of a renewable hybrid power system. Furthermore, the parameters of controllers have been tuned using a TLBO algorithm.

The benefits of fuzzy technique in a traditional system lead to support in green microgrid power system. The management of secondary load frequency for an island micro-power network has been the focus of the authors of [227] using the PSO tool. For the LFC problem of the dual-area solar-PV-wind based linked microgrid power system, a nominal fuzzy-PID controller has been achieved [184]. MPA algorithm is used to tune the scaling factor of the controller. In paper [228], it is explained how to determine the best adaptive neuro-fuzzy PID controller parameters by using the improved sine-cosine technique (ISCA). In [229], the frequency regulation of hybrid microgrid made up of RESs and energy storing devices has been maintained using type-II fuzzy-PID controller. The proposed controller has been tuned by Moth Swarm Algorithm.

3.6. Other Modern Controllers

In AGC and microgrid networks, a few recently developed controllers (

Table 4. Merits and demerits of controllers.

Controllers	Merits	Demerits
Conventional Controller	These methods are well investigated in the available literature. Smile to design.	It delivers larger frequency deviation. Fails to perform better under the different operating conditions.
Fractional Order Controller	Using the FO differential equation, real-world system characterisation is simple. Robustness, stability, and load rejection capability is more than classical IO controllers. The higher order can be reduced to low order in FO based controller.	Takes more time to achieve a stable state. Numbers of control parameters need to tune optimally for better dynamics response.
Tilt Integral Derivative/Integral- Tilt Derivative Controller	Robustness, stability, and load rejection capability is more than classical IO controllers. It offers easier tuning, enhanced disturbance rejection capacity, and demonstrates outstanding sturdiness to parameter variability	Takes more time to achieve a stable state. Numbers of control parameters need to tune optimally for better dynamics response.
Cascade/Dual-Stage Controller	Provides an additional sensor to help in disturbance reduction before the output quality is final. Due to the larger number of tuning nubs compared to non-cascaded controllers improves the system dynamics.	More time is required because controller has more tuning parameters. High level of tolerance for sensor error is required. Investigation of cyber-attack robustness required. Selection of primary controller and secondary controller is an important factor to demonstrate superior system responses.
Fuzzy Logic Based Controller	These controllers are cheaper to develop, they cover a wider range of operating conditions. They are more readily customizable in natural language terms. Adaptable, simple knowledge base design. The concepts of control and supervision are defined in the same language. When the final user is not a control engineer, end-user interpretation is simpler. Simple calculation toolboxes and specialised integrated circuits are widely accessible. In rule bases, consistency, redundancy, and completeness can be verified (knowledge acquisition supervision). This might enhance user interpretability and accelerate automated learning. Due to its capacity for universal approximation, FLC can incorporate a standard design (PID, state feedback) and fine-tune it to specific plant nonlinearities.	Retuning takes a lot of time even if applied to a similar plant in other location. Instinctive fuzzy PID-type design does not exactly perform better to its well-tuned conventional complement.
ANN Controller	The ANFIS offers various benefits, such as the capacity for rapid learning, adaptability, and the ability to capture the nonlinear structure of a process. Simple calculation toolboxes and specialised integrated circuits are widely accessible.	Retuning is time consuming, even when applied to the same plant at another location.

) have proven to be reliable for LFC schemes. In order to decrease frequency oscillation in a single-area hybrid power network used sliding mode controllers in [186]. The paper [230] proposed an integral higher degree SMC controller to properly address the LFC problem for the triple-area hydro power plant in order to improve the SMC controller action. For independent renewable microgrid systems, a unique sliding mode consensus controller (SMCC) has been designed [231]. Reference [132] provides an illustration of how the H-infinity controller is set up and how it is used for LFC in a single-area distributed microgrid network powered by SMES and UCs. An iterative PID-H-infinity (IPID-H-infinity) controller for LFC of hybrid renewable microgrid system was recently proposed by Pandey et al. [232].

The two-degree-of-freedom (2DOF) and three-degree-of-freedom (3DOF) PID controllers have been used in several studies. The output signal is produced by the DOF-PID controller block based on the difference between a reference signal and a measured system output. According to the given setpoint weights, the block computes a weighted difference signal for each of the proportional, integral, and derivative actions. The sum of the proportional, integral, and derivative actions on the corresponding difference signals—with each action weighted in accordance with the gain parameters—results in the block output. The derivative action is filtered by a first-order pole. Gains on controllers can be adjusted manually or automatically. The Simulink Control DesignTM programme is necessary for automatic tuning. The fundamental diagram of 2DOF-PID and 3 DOF-PID is presented in the Figure 10 and Figure 11. A novel Seagull Optimisation Algorithm (SOA) tuned with three-degrees-of-freedom (3DOF) is suggested for multi-area (thermal, hydro, and nuclear units) power system [233].

A hybrid FOPID controller made by combining the FOPID and TID controller is proposed for LFC issues of multi-area PS. A Manta Ray Foraging Optimisation (MRFO) algorithm is used for optimizing the parameters of proposed controller [234]. Furthermore, in [235] a two-degree-of-freedom FO-based PID controller with AGC is discussed. The performance of the proposed controller is compared to that of the FOI, FOPI, and FOPID controllers.

4. Future Scope

The modern power system is experiencing structural modifications as a result of difficulties with environmental effects, growing industrialisation, higher demand, and supply for loads, as well as a better quality of life. India pledged to cut its carbon footprint by 33–55% from 2005 levels by 2030 as part of the Paris Climate Agreement and would work to produce 40% of its total electricity output from non-conventional sources in the same year. As a result, the system becomes oscillatory due to an increase in the penetration level of probabilistic RESs and system inertia, which causes changes in the system's frequency and interchangeable tie power. However, frequency oscillation that exceeds the rated limit causes power outages and so-called blackouts. It is abundantly obvious that the existence of renewable energy could cause frequency regulation issues for the modern electric power network of the future. Therefore, a more sophisticated control system is needed to deliver consistent electricity with better-coordinated frequency regulation while taking into account new renewable resources. Based on the observed literature survey, the following list of possible prospective studies in the area of LFC problems needs to be investigated:

• Examine some recently discovered renewable resources and how they fit into LFC.
• The LFC problem in restructured renewable microgrids needs to be further investigated.
• The existing controllers tuned with novel optimisation techniques for fine-tuning the controller gains that improve the frequency profile need to be investigated.
• Recently considered controllers that have been applied for LFC on any power system need to be investigated.

The contribution of upcoming restructured power market in microgrid LFC is an essential structure where GENCO and DISCO will be taken care of. For stable microgrid operation that takes into consideration the dynamics of restructured microgrids, additional work is needed to create improved LFC control techniques:

• Create an innovative approach to optimal-robust control that can manage system dynamics, parametric variation, and variations in power output.
• Create some cutting-edge fractional order (FO) and dual/triple-stage cascaded controllers to improve system performance.
• Future load forecasting and prediction are essential for microgrid functioning. As a result, LFC system could leverage deep learning-based encryption algorithm.
• Look into a viable control strategy to locate and isolate LFC network faults.
• Create an ideal control method for real-time LFC.
• It is necessary to investigate more communication-based LFC techniques for power networks.

To achieve LFC, various power system controllers and components exchange control signals or system frequency deviation signals. These signals are transmitted across communication infrastructure as data packets. As a result, the underlying communication is crucial for the operation of LFMC activities. A strong and trustworthy communication framework is necessary for the LFC to operate steadily and smoothly.

There have been various communication standards and protocols proposed for smart grids. Since it suggests an object-oriented modelling for all the components/domains of the power system, the IEC 61850 standard series has emerged as the front-runner among them [231]. This IEC 61850 object-oriented modelling technique is applied for power system devices in the organisation of data, the configuration of objects, and the mapping of those objects onto protocols to ensure their consistency and interoperability. IEC 61850 was initially designed for substation automation, but in more recent iterations it has been embraced for power utility automation [236].

The effects of LFC methods on single- and multi-area power systems' susceptibility to cyber-attacks have been examined. Utilizing cutting-edge information and communication technologies, single- and multi-area microgrid networks have grown dramatically in order to deliver independent operation, management, and flexibility. The need for protection for such networks from cyber-attack is increased because they are not entirely secure. The evaluation of unwanted data attack effects on tie-line power exchange is important. With the intention of diminishing cyber-attacks, the proper cyber security systems should be implemented.

Numerous factors—including environmental concerns, issues with fossil fuels, energy system security, problems with the economy, and operating costs—are changing the way the power system looks. Numerous nations have made the decision to increase the proportion of RESs such as wind and solar in their energy systems. Numerous issues arise as a result of growing the proportion of RESs in power systems, including a decrease in the system's overall inertia, an increase in the power imbalance during short-term operation, and an increase in frequency and tie-line power oscillations. The system oscillates and the frequency oscillation (highest frequency deviation) rises when the overall inertia is reduced, among other issues. The frequency oscillations directly increase when the power imbalance in instant operation increases. It is obvious that frequency control issues would arise in both present and future power systems. These systems would struggle to maintain sufficient damping and inertia. In order to control such frequency problems, appropriate control mechanisms and ancillary services such as main and secondary reserves are required. The following list of research holes in the area of LFC that require additional study is based on the literature review:

• Increasing the robustness of LFC-related control approaches.
• Optimal-robust control techniques for LFC that can manage changes in both plant parameters and power output.
• Developing new LFC objective functions that can enhance the performance of power systems, taking into account the dependability of LFC loops
• Improving the capacity of LFC models to handle cyber-attack glitches.
• Suggesting appropriate control strategies that can identify the sensor faults and isolate them from LFC loops.
• Developing new fault diagnosis techniques that work with LFC.
• Recommending appropriate control techniques to identify and isolate sensor defects in LFC loops.
• Developing new control strategies that utilise wide variation in the plant paramters and load demand.
• Developing control strategies for power systems modelled without assumptions.

The potential for demand-side participation to provide some ancillary services for independent system operators, such as primary and secondary storage and LFC services, is now being investigated in great detail. Realistic participation strategies are needed to minimise the oscillatory increase in frequency. For contemporary power systems, a suitable coordination between demand-side and generation-side participation in LFC can be suggested as a promising area of study. To support demand-side participation in LFCs, further research into the infrastructure of contemporary power systems is also needed. Where extensive market research is required for infrastructure and supporting services, electric vehicles and other demand side factors can contribute to Smart Load LFC. Therefore, new strategies are needed to reduce the oscillatory increase in frequency.

5. Conclusions

The most recent advances in LFM approaches for various traditional and renewable energy-based power systems and renewable integrated microgrid/smart girds are surveyed in this article. Single- and multi-area power systems are covered. An effort has been made to obtain a critical remark on precise single/multi-area power system frequency response models, LFC for RESs-based power systems, and their relative analysis in conventional, deregulated, and microgrid power domains due to the significant frequency stabilisation feature in modern power systems. There has been some discussion of the mathematical modelling related to changes in frequency deviation, tie-line power deviation, tie-line error, and control error of conventional/microgrid power networks under the LFC scenario. Additionally, the detailed applications of classical controllers, FO controllers, tilt derivative, tilt integral controllers, recently invented cascaded controllers, as well as other recently developed and used controllers for the LFC are examined. Different controllers’ merits and demerits have also been examined. Additionally, various current soft computing strategy are examined for the LFC problem, including optimisation algorithms, fuzzy logic method, artificial neural network, and other recent approaches. Finally, the future scope and research gaps of this domain have been identified.

The research makes it clear that LFC is essential for future power systems that include renewable energy resources. The recent trend involves using various computing techniques to boost LFC performance and accuracy. Finally, a few areas for further research in the field of contemporary LFC systems are discussed. To enable large-scale application, integrated modelling and application studies that pair these LFC approaches with feasible communication technologies are needed.