Leg Conversion Method for the Continuous Control of a Railway Vehicle Propulsion Inverter

Abstract

In this study, we propose a whole-leg scheme for the continuous control of a railway vehicle propulsion inverter. The three-leg method is commonly used to obtain the output AC power from the input DC voltage in a railway vehicle propulsion inverter. When the power semiconductor of an inverter operating in a three-leg manner fails, a problem occurs in the power supply. Therefore, this study proposes a method that automatically converts into the two-leg method without replacing the power semiconductor when it fails while the inverter is driven with the three-leg method. This leg conversion method can prevent problems in supplying power by converting the power that can be obtained from an inverter operating with the three-leg method into that obtained from an inverter operating with the two-leg method. In addition, to verify the effectiveness of the leg conversion method, a demonstrative experiment was conducted after a simulation, and an actual inverter was fabricated. Consequently, the experimental results confirmed that in an inverter operating with the three-leg method, even if the power semiconductor failed, it could be converted for operation with the two-leg method to supply power normally.

1. Introduction

In recent years, resource depletion owing to environmental pollution and the excessive usage of fossil fuels has emerged as a major issue. Research on eco-friendly methods and carbon reduction methods that use renewable energy is being conducted. Methods of using alternative energy are applied in various fields, and much research is being conducted with eco-friendly methods, including in transportation. Representatively, many studies using electricity and hydrogen have been conducted and are currently widely used. In addition, eco-friendly construction methods are widely applied and used in railroads. When energy that is obtained in an eco-friendly manner is converted into the electricity required by railroad cars, it is converted and used with an inverter for propulsion [1,2,3,4,5,6].

If the power semiconductor used in a railway vehicle propulsion inverter fails, the control of the inverter malfunctions, resulting in a problem in the supply of power in places where power is needed. In severe cases, trains that are in operation may stop or be derailed. When such problems occur, they can lead to great loss of life, so appropriate countermeasures are needed [7,8,9,10].

For the normal operation of an inverter with a smooth and efficient power supply, the power semiconductor, which is where problems often occur, must be directly replaced; this is often cumbersome [11,12,13,14,15].

The proposed scheme enables normal power supply by applying the two-leg scheme to an inverter operating in the existing three-leg scheme. Inverters operating with a two-leg method have been applied in various ways, such as in electric motors, and many studies have been conducted. However, the leg transfer method proposed in this paper refers to a method of converting an inverter operating in the three-leg mode to the two-leg mode when a short-circuit fault occurs. With this method, the power supply can be continuously maintained by diverting the current in the leg where the power semiconductor has failed to the DC-link capacitor. To verify the leg conversion method, simulation experiments were conducted, and inverters were fabricated by using PSIM to conduct demonstrative experiments.

Consequently, when the three-leg operation method is converted into the two-leg method, the power that can be obtained from the three-leg operation method can be obtained, and the power supply can be made sustainable until it is applied to a load-coupled inverter.

The remainder of this paper is organized as follows. Section 2 describes the circuits of the inverters of the three-leg, two-leg, and leg conversion methods. Section 3 compares and analyzes the effects of the three-leg and two-leg systems through simulations of inverters operating with the leg conversion method. In Section 4, we present a demonstrative experiment that was conducted after designing an inverter based on the leg conversion method and analyze the results. Finally, Section 5 presents our concluding statements.

2. Inverter Structure and Control Method for Applying the Leg Conversion Method

The inverters used in railway vehicles convert an input DC voltage into an output AC power. The three-leg method is used as a typical operation method for such inverters. In addition, an inverter can be operated by using the two-leg method. In this section, before verifying the leg conversion method, the circuit and control methods of an inverter operating with the three-leg and two-leg methods used in railway vehicles are explained, and the circuit of an inverter operating with the leg conversion method is described.

2.1. Structure and Control Method of a Three-Leg Method Based Inverter

Figure 1 shows a circuit diagram of an inverter that uses the three-leg method. Six IGBTs convert the input DC power into output AC power. Figure 2 demonstrates the control block diagram of a three-leg inverter. The difference between the reference current and the measured current $v_{c\_ref}$ is converted into a control quantity by using a PI controller. Each PWM waveform is signaled to the IGBT such that the inverter can convert the input DC power into the output AC power [16,17,18,19].

2.2. Structure and Control Method of a Two-Leg Method Based Inverter

Another type of inverter based on the two-leg method uses two IGBTs less than those in the existing three-leg method, but it can obtain the same power as an inverter that operates with the three-leg method.

Figure 3 shows a circuit diagram of an inverter that uses the two-leg method. Unlike the three-leg method, the inverter operating with the two-leg method uses four IGBTs to obtain the output AC power from the input DC power. If one of the previously used phases is connected to the center of the two DC-link capacitors and appropriate control is performed, the output AC power can be obtained from the input DC power in the same manner as in the three-leg method. In this study, the IGBT of phase C of the three phases is assumed to be short-circuited; the control block diagram is demonstrated in Figure 4.

As shown in Figure 4, after passing through the PI controller, the control amount in phase C, $v_c$, is subtracted from the control amount in phase A, $v_a$, and the control amount in phase B, $v_b$, which are converted into $v_{at} \mathrm{and} v_{bt},$ respectively [20,21,22,23,24,25,26,27]. By using this to generate a PWM waveform and send a signal to the IGBT, the inverter can convert and supply power normally. The voltage in each phase of the control method can be expressed as follows [28]. Based on the calculated values, three-phase voltage and current waveforms are obtained by using the SPWM control method.

$$v_{at}=v_a-v_c$$

(1)

$$v_{bt}=v_b-v_c$$

(2)

2.3. Structure and Control Method of the Proposed Inverter

The leg conversion method used in this study is a two-leg method that converts a leg corresponding to a leg in an inverter driven with the three-leg method during IGBT failure.

Figure 5 presents a circuit diagram of the proposed inverter. Figure 6 shows a circuit diagram of the inverter’s operation before and after a short circuit fault. Figure 7 shows the current flow diagram of an inverter with leg conversion applied.

The method proposed in this paper is a method for the normal operation of the inverter after an short circuit fault occurs in an IGBT. A gate drive that was used in an experiment generated an error signal according to the short circuit current and overvoltage. Through this, it was possible to determine whether a short circuit fault existed in the transistor [29,30,31,32].

Basically, a circuit that can be separated from the main circuit is needed when a short circuit occurs in an IGBT. Accordingly, $sw{A}_b,sw{B}_b,\mathrm{and} sw{C}_b$ are placed in each IGBT so that the leg can be separated from the main circuit when a problem occurs in the power semiconductor. These switches are composed of switches with b-contact characteristics so that current can flow at all times. In addition, a continuously controllable circuit is required so that current can flow normally even in the leg in which the power semiconductor has a problem. Accordingly, as shown in Figure 7, $sw{A}_a,sw{B}_a,\mathrm{and} swC\_a$ are installed and connected to the neutral point of the DC-link capacitor. The switch used at this time is composed of a switch with a contact characteristic so that current cannot normally flow.

In this paper, it is assumed that only the C-phase IGBT has failed. Accordingly, $swC\_b$ opens, the main circuit of the C-phase IGBT is separated, and $swC\_a$ closes, so the C-phase can be connected to the neutral point of the DC-link capacitor.

3. Simulation Verification of the Leg Conversion Method for Coping with Power Semiconductor Failure

In this study, a simulation was conducted to verify the possibility of operating an inverter with the leg conversion method when the power semiconductor of a propulsion inverter used in a railway vehicle failed. The circuit used in the simulation was that described with the proposed inverter circuit diagram shown in Figure 8, and the parameters of the inverter were set as listed in

Table 1. Summary of various parameters used in the simulations.

Parameter	Unit	Value
Voltage	600	[V]
Inductor	150	[mH]
Capacitor	10	[μF]
Resistor	500	[Ω]
DC-link capacitor	4700	[μF]
DSP	TMS320F28335
IGBT	SKM100GB12T4
Gate driver	SKHI 22BH4 R

. An experiment was conducted to convert the inverter’s operation from the three-leg method into the two-leg method, assuming that a problem occurred in phase C of the power semiconductor [33,34,35].

Figure 9 shows various waveforms before and after leg conversion when the power semiconductor failed according to the application of the leg conversion method. In this study, the power semiconductor failure was assumed to occur in phase C. As shown in Figure 8, the failure of the power semiconductor in phase C occurred at approximately 0.525 [s]. As shown in Figure 9a, the current flows of $I_{sw5},I_{sw6},\mathrm{and} I\_DC\_C$ changed according to the time before and after 0.525 s when the power semiconductor of phase C failed. Before 0.525 s, the currents of $I_{sw5} \mathrm{and} I_{sw6}$ flowed, and only the current of $I\_DC\_C$ flowed when the power semiconductor failed. However, as shown in Figure 9b,c, which depict the waveforms of the output voltage and current of the inverter, the waveforms at approximately 0.525 s were all the same. In other words, when an inverter operating with the leg conversion method was applied in a propulsion inverter for railway vehicles, it produced sufficient performance to avoid power supply interruption due to the continuous control according to the power semiconductor failure occurring in one phase. An additional experiment was conducted to determine whether the output voltage and current values obtained from the inverter operating with the leg conversion method were identical to those of inverters operating with the three-leg and two-leg methods.

Figure 10 shows the various waveforms of an inverter operating with the three-leg method. As shown in Figure 10a, $I_{sw5} \mathrm{and} I\_sw6$, the IGBT currents of phase C flowed. Furthermore, because the current of $I\_DC\_C$ did not flow, the inverter was operating with the three-leg method. As shown in Figure 10b,c, the inverter supplied the output voltage and current normally.

Figure 11 shows various waveforms of an inverter operating with the two-leg method. As shown in Figure 11a, $I_{sw5} \mathrm{and} I\_sw6$, the IGBT currents of phase C, did not flow. Furthermore, because the current of $I\_DC\_C$ flowed, the inverter operated with the two-leg method. As shown in Figure 11b,c, the inverter normally supplied the output voltage and current. The magnitudes of the output voltage and current supplied by the inverter were the same for the three-leg and two-leg methods. In addition, the output voltage and current of the inverter operating with the leg conversion method when the power semiconductor failed were the same. To verify this, based on the simulation results, a demonstrative experiment was conducted after designing an inverter to apply leg conversion to an actual inverter in the case of power semiconductor failure.

4. Results and Discussion

Based on a comparative analysis of the results of the simulation experiments, a demonstrative experiment was conducted to prove the actual effect of an inverter operating with the leg conversion method. For this, an inverter was manufactured based on the parameters presented in Table 1 and the circuits used in the simulations. Continuous control of railway vehicle propulsion inverter to demonstrate the effectiveness of the inverter operation based on the leg conversion method, an experiment was conducted employing the three-leg, two-leg, and leg conversion methods.

Figure 12 shows a self-made inverter based on Table 1 and Figure 8 to verify the its operation with the leg conversion method. First, an experiment was conducted while assuming that a problem occurred in phase C of the power semiconductor in an inverter operating with the three-leg method.

Figure 13 shows the waveforms when current control was performed at a voltage of 600 V. As shown in the first waveforms in Figure 13a,b, the current flowed in the area connected to the DC-link capacitor when changing from three-leg to two-leg operation. When operated with the leg conversion method, no current flowed in the area connected to the neutral point of the DC-link capacitor. However, when operated with the two-leg method, the problematic leg’s side current was blocked, and the current flowed to the DC-link capacitor. As shown in

Table 2. Delay time when the leg system was converted.

	600 [V], 100 [mA]	600 [V], 300 [mA]
a	35.8 [ms]	34.5 [ms]
b	34.4 [ms]	34.2 [ms]
c	34.8 [ms]	35.7 [ms]

and Figure 10, the delay time during leg conversion occurred at approximately 37.23 ms, on average, when converting from three-leg to two-leg operation; however, no problems occurred in the normal state. In addition, an additional experiment was conducted to check whether the output values were the same as those when operating with the three-leg and two-leg methods.

Figure 14 shows the input DC voltage and three-phase output current waveforms of inverters operating with the three-leg and two-leg methods when an input DC power of 600 V was applied. As shown in

Table 3. Voltage and current values recorded during operation.

	Three-Leg Method	Two-Leg Method
Input Voltage	598.6 [V]	598.7 [V]
Output Current_a	203.36 [mA]	204.7 [mA]
Output Current_b	215.57 [mA]	216.31 [mA]
Output Current_c	214.25 [mA]	216 [mA]

, when the inverter was driven with the two-leg method, the same value of the three-phase output current could be obtained as in the three-leg method. The results of the simulation and verification experiments on the inverter indicated that the same output value as that of the inverter operating with the three-leg method could be obtained when it was operated with the two-leg method, and continuous control was possible throughout.

Accordingly, the results determined that no problems occurred in the supply of power when a problem occurred in the power semiconductor of the inverter, and the operation was converted from the three-leg to the two-leg method. Consequently, even when a power semiconductor fails while supplying power from a propulsion inverter operating with the three-leg method in a railway vehicle, its operation is directly converted into the two-leg method, and the power is supplied while the inverter operates normally.

5. Conclusions

In this study, a leg conversion method for continuous control in a propulsion inverter in a railway vehicle was applied and its performance was verified. Prior to verifying the leg conversion method, the structure and control method of the three-leg and two-leg methods, which are representative operating methods for inverters, were described, and a method for reducing the ripple effect of the failure of the power semiconductor in an inverter was described. To verify this possibility, a simulation experiment and a demonstrative experiment of an inverter operating with the leg conversion method were conducted, and the results were compared and analyzed.

The results confirmed that the waveforms of the three-phase voltage and current supplied to the system before and after the failure of the power semiconductor in phase C were the same. That is, it can be seen that by using the leg-switching method, interruption of the power supply due to the occurrence of a short circuit can be avoided. Through this, it is possible to solve various problems, such as the suspension of operation and derailment, which may occur in railroad cars due to power supply interruption in the event of a short circuit accident. Finally, through the experiment on conversion from the three-leg method to the two-leg method, it was confirmed that a control delay occurred when the leg was converted, but no problem with power supply occurred in the normal state before and after the leg conversion. It is judged that additional research on the delay problem is needed in the future.

Accordingly, it is judged that if the leg-switching method proposed in this paper is applied even in the event of a short circuit accident in the power semiconductor of a propulsion inverter in a railway vehicle, it will be possible to minimize the loss due to the interruption of the power supply and prevent disruptions in train operation.