Evaluation of City–Industry Integration Development and Regional Differences under the New Urbanization: A Case Study of Sichuan

Abstract

As an important focus of China’s new urbanization construction, the policy of “city–industry integration” has played a key role in promoting the sustainable development of urban construction. This paper integrates the three major elements of social service, ecology and environment, and science and technology into the traditional indicator system of “industry, city, and people”. Firstly, the PSO-AHP-EM model was constructed to empower the indicators subjectively and objectively, and the degree of city–industry integration was sorted by the GRA-TOPSIS model. Secondly, the Gini coefficient and Theil index were used to further explore the non-equilibrium of city–industry integration development. Finally, an empirical study was conducted on 18 prefecture-level cities in Sichuan Province from 2010 to 2019, and their development laws were analyzed. The results show that: (1) The weights obtained by the PSO-AHP solution were better than the weights obtained by AHP. (2) The level of city–industry integration in Sichuan was not high, and most cities need to be further developed. (3) The regional difference of city–industry integration was relatively small, and the main source of the difference was Chengdu Plain Economic Zone. This paper puts forward some suggestions to provide scientific support for the evaluation of city–industry integration.

1. Introduction

In recent years, China’s urbanization process has been accelerating, and the urbanization rate has increased from 49.95% in 2010 to 60.6% in 2019. However, the separation of industrial development and urban construction has become more serious at the same time, resulting in a series of urban problems to be solved urgently. If the development of industry lacks the supporting role of the city, it will cause urban diseases such as home–work separation, traffic congestion, and environmental pollution [1,2,3]. In the end, it will only lead to the urban area evolving into an industrial zone, becoming a “sleeping city” or even a “ghost city” [4]. If the urban development loses the driving role of the industry, it will lead to problems such as industry hollowing-out, insufficient stamina of urban development, and decline in residents’ living standards [5,6]. As a result, the Chinese government has gradually realized the problem of urban construction and thus put forward the strategic goal of “city–industry integration”. In 2014, the State Council issued the National New Urbanization Plan (2014–2020), which takes “city–industry integration” as the leading idea for the development of new urbanization. This idea takes the city as the carrier to develop the industrial economy, and takes industry as the guarantee for the promotion of urban renewal, so as to achieve a high-quality development model [7].

To achieve the integration of industry and city, the basic role of traditional theories [8] and the promotion of innovative theories are very important. The way in which to meet the inherent requirements of matching industrial space and social space is the core issue in terms of promoting the realization of China’s new urbanization construction goals. With the emergence of new topics in urban development such as innovation-driven approaches and replacing old growth drivers with new ones, the exploration of city–industry integration under the new situation should be further deepened [9]. Under the severe situation of fundamental changes in the speed, structure, and driving force of economic development, it is of great significance to evaluate the development level of city–industry integration, as well as its spatial differentiation [10].

This paper sorts out the main influencing factors discussed by scholars and comprehensively evaluates the development level from the six aspects of industry, city, people, service, ecology, and science. This part includes three steps: (1) establishing an index system with the support of a big data index database, (2) combining PSO-AHP and EM to build a new empowerment method, and (3) improving TOPSIS with GRA to obtain a multi-attribute decision-making method. At the same time, the Gini coefficient and Theil index are introduced to measure the regional difference of city–industry integration development. This paper focuses on the balanced development of city–industry integration, which plays a key role in realizing high-quality urbanization. This is particularly the case for Sichuan, a southwest province with unbalanced and insufficient development [11].

The research on city–industry integration in Sichuan Province is mostly focused on the problems and development paths of a specific city or industrial agglomeration area [11,12,13]. There are few quantitative evaluations of city–industry integration, and only Gan [14] has conducted a relevant analysis based on the coupling coordination degree approach. Thus, the existing studies have not fully explored the indicators and methods for assessing the city–industry integration in Sichuan, and further quantitative analysis is needed to evaluate the region.

The following are the novelties: (1) Establishing a comprehensive index system of city–industry integration with the help of big data. The main tool is a browser automation framework named Selenium. (2) Analyzing the development level of city–industry integration with a comprehensive method, including PSO-AHP, EM, and GRA-TOPSIS. (3) Introducing the calculation of regional differences into the evaluation of city–industry integration and more intuitively exploring the source of uncoordinated development.

The remainder of this paper is as follows: Section 2 outlines the literature and the research framework of this paper. Section 3 covers the materials and methods. Section 4 covers the results and discussion. Section 5 covers the conclusions of the study and suggestions.

2. Background

2.1. Literature Review

So far, the research results of city–industry integration have focused on qualitative analysis [8] and development concepts [15]. The literature review is summarized into three aspects: (1) indicators of city–industry integration evaluation; (2) methods of city–industry integration evaluation; (3) regional differences of urbanization.

2.1.1. Indicators of City–Industry Integration Evaluation

At present, the evaluation of city–industry integration is generally based on analyzing its connotation or influencing factors, as well as considering the availability of indicators, in order to construct an indicator system. Industrial development, urban construction, and people’s livelihood are the most basic elements, and thus many scholars have built an index system from these three aspects [16,17,18,19]. On this basis, Jia supplemented the allocation of resources [20]; He expanded the people’s life into population, employment, and income and life, expanding urban construction into land and environmental facilities, as well as social development [21]; and Tang added the degree of spatial integration [22]. Some scholars also made a comprehensive description by paying attention to the natural environment [23], basic social services [24], and scientific and technological innovation [25].

On the basis of the previous research on the indicators of city–industry integration, scholars usually separately discussed the three elements of social service, ecological environment, and scientific and technological innovation. Therefore, this paper incorporates these three elements into the main index categories to realize a comprehensive evaluation of city–industry integration.

2.1.2. Methods of City–Industry Integration Evaluation

The quantitative measurement of city–industry integration is one of the key points and difficulties in the research of city–industry integration.

Table 1. Evaluation methods of city–industry integration.

Authors	Main Method
Zhang and Shen, etc. [26]	AHP
Su et al., etc. [31]	Fuzzy comprehensive analysis
Wang et al., etc. [32]	Factor analysis
Yang and Fang, etc. [27]	Principal component analysis
Zhang et al., etc. [33]	EM
Tang, etc. [22]	GRA
Gan and Mao, etc. [34]	TOPSIS

shows the common methods. In general, scholars have mostly used analytic hierarchy [26], principal component analysis [27,28], the coupled coordination relationship model [14,29], factor clustering analysis [30], gray association model [22], and fuzzy hierarchy [31]. Due to the large differences in the study areas, the index systems and methods of city–industry integration evaluation have not yet formed a consensus [8].

For the evaluation of the above city–industry integration, most scholars still choose the single weighting method. The latest trend is to combine subjective weighting and objective weighting [35], which can avoid one-sidedness to some extent, and improve the scientificity of weighting. In addition, this paper constructs the GRA-TOPSIS model, which constructs a new relative closeness based on Euclid distance and gray correlation degree [14], and has strong practicality for evaluating the degree of city–industry integration.

2.1.3. Regional Differences of Urbanization

Due to differences in geographical location, transportation conditions, and industrial structure, regional development differences are common. Up to now, regional development differences mainly focus on quantitative research on the spatiotemporal evolution [36], the convergence of regional development differences [37], and the causes and influencing factors of regional development differences [38]. Among them, the quantitative measurement and analysis of the spatiotemporal evolution have always been the focus of regional difference research. The research scale mainly involves multiple levels such as global [39], national [40], regional [41], and developed urban agglomerations [42], and gradually expands to the provincial [36] and municipal [43] levels, but the overall research on the differences between western regions is relatively weak [44]. The research methods are mainly divided into mathematical–statistical methods such as coefficient of variation [36,39,43], Gini coefficient [36,42,43], and Theil index [36,38,43], and there are also spatial measurement methods represented by Moran index [45], SDE [46], and ESDA [46,47].

Compared with using a single statistical indicator to measure regional development differences [42], this study conducts a more comprehensive study based on the evaluation value of city–industry integration. At the same time, the economic zones of Sichuan are used as the basic research units to further study the problem of unbalanced development.

2.2. Research Framework

The framework of this paper is expressed in Figure 1. This paper takes the development law of city–industry integration as the research goal, focusing on three main problems: (1) establishing a comprehensive index system of city–industry integration through the web crawler and related literature; (2) the combination of PSO-AHP-EM and GRA-TOPSIS used to calculate the degree of city–industry integration; (3) the Gini coefficient and Theil index, applied to analyze the regional difference and evolutionary trajectory. On the basis of the above technical route, the panel data of 18 cities in Sichuan Province from 2010 to 2019 (most recently available data as of 2021) were evaluated as a case study, and the results of the integration of industry and city were discussed and suggested.

3. Materials and Methods

3.1. Evaluation Index System

The closeness of industry and city has a direct impact on urban development, but it has not yet been included in the overall indicator system of urban development. This shows that the measurement research of city–industry integration is not perfect, and can be promoted with the help of “big data” and “cloud platforms” [48]. On the basis of Liu’s literature review on city–industry integration [8], this paper summarizes 6 index categories, as shown in Table 2. Through network crawlers and literature review, the number of indicators was controlled between 20 and 30, and a scientific development evaluation system of city–industry integration was constructed.

3.1.1. Web Crawler

This paper used crawler technology to access the index database named after urban development in the regional decision-making template for the China Economic and Social Big Data Research Platform (https://data.cnki.net (accessed on 24 January 2022)) [49]. A test tool named Selenium in Python was used to simulate launching the Chrome browser, opening the web page, and clicking to select [50], as shown in Figure 2. With the help of the navigation bar of the indicator category contained in the drop-down box of the indicator library (the URL pointed to by each node contained the corresponding name of the indicator within its range), the URL of each node was queued and accessed from top to bottom. The names of the crawled indicator formed new child nodes, which were grouped into the parent node where the corresponding indicator category was located and saved in XML format. In this way, the most comprehensive indicator collection work was obtained, including the classification of indicator categories where indicators were located. The main information for crawling is shown in

Table A1. Crawling results of indicators.

Data-Bind	Metric Classification	The Number of Indicators
Urban development	People	14
	National accounts	12
	Labor employment	50
	Investment in fixed assets	20
	Industry	35
	Finance	16
	Foreign economic relations and trade	6
	Total retail sales of consumer goods	14
	Tourism	4
	Scientific and technological innovation	4
	Resource environment	24
	Public service	50
	Infrastructure	45

.

Table 2. Indicators sources of the index system.

Target Layer	Criteria Layer	Indicators	Indicator Sources
City integration production	Economy and industry	Per capita GDP (+)	1. Literature research: Jia et al. [20]; Tang [28]; Zhou et al. [51] 2. Web crawler
		The proportion of the secondary industry output value to the GDP (+)
		The proportion of the tertiary industry output value to the GDP (+)
		The gross industrial output value of enterprises above scale (+)
	Habitat and employment	Household deposit balance (+)	1. Literature research: Zhang and Shen [26]; Gan et al. [14]; Yan [52] 2. Web crawler
		The proportion of employed persons in the secondary industry (+)
		Number of registered unemployed rate in urban areas (+)
		Per capita housing area in urban areas (+)
	Social services	Number of hospital beds per 10,000 people (+)	1. Literature research: Li and Zhang [16]; Li and Zhang [18]; Zhou et al. [51] 2. Web crawler
		Number of autobuses per 10,000 people (+)
		Culture sports and the media expenditure (+)
		Number of schools per 10,000 people (+)
	Urban construction	Investment in fixed assets in the construction of municipal public facilities (+)	1. Literature research: He and Xia [21]; Yang and Fang [27]; Cong et al. [10] 2. Web crawler
		The proportion of industrial land (−)
		The proportion of land used for public facilities (+)
		Per capita urban road area (+)
	Ecology and environment	Expenditure on energy saving (+)	1. Literature research: Yang and Fang [27]; Yan [52] 2. Web crawler
		The comprehensive utilization rate of industrial solid waste (+)
		Wastewater discharge (−)
		The reduction rate of energy consumption per unit of GDP (+)
	Science and technology	The number of people with a college degree or above per 10,000 people (+)	1. Literature research: Zhen and Zhu [25]; Yan [52] 2. Web crawler
		Science and technology expenditure accounts for the proportion of fiscal expenditures (+)
		R&D input intensity (+)
		Applications granted (+)
		The full-time equivalent of R&D personnel (+)

Table 2. Indicators sources of the index system.

Target Layer	Criteria Layer	Indicators	Indicator Sources
City integration production	Economy and industry	Per capita GDP (+)	1. Literature research: Jia et al. [20]; Tang [28]; Zhou et al. [51] 2. Web crawler
		The proportion of the secondary industry output value to the GDP (+)
		The proportion of the tertiary industry output value to the GDP (+)
		The gross industrial output value of enterprises above scale (+)
	Habitat and employment	Household deposit balance (+)	1. Literature research: Zhang and Shen [26]; Gan et al. [14]; Yan [52] 2. Web crawler
		The proportion of employed persons in the secondary industry (+)
		Number of registered unemployed rate in urban areas (+)
		Per capita housing area in urban areas (+)
	Social services	Number of hospital beds per 10,000 people (+)	1. Literature research: Li and Zhang [16]; Li and Zhang [18]; Zhou et al. [51] 2. Web crawler
		Number of autobuses per 10,000 people (+)
		Culture sports and the media expenditure (+)
		Number of schools per 10,000 people (+)
	Urban construction	Investment in fixed assets in the construction of municipal public facilities (+)	1. Literature research: He and Xia [21]; Yang and Fang [27]; Cong et al. [10] 2. Web crawler
		The proportion of industrial land (−)
		The proportion of land used for public facilities (+)
		Per capita urban road area (+)
	Ecology and environment	Expenditure on energy saving (+)	1. Literature research: Yang and Fang [27]; Yan [52] 2. Web crawler
		The comprehensive utilization rate of industrial solid waste (+)
		Wastewater discharge (−)
		The reduction rate of energy consumption per unit of GDP (+)
	Science and technology	The number of people with a college degree or above per 10,000 people (+)	1. Literature research: Zhen and Zhu [25]; Yan [52] 2. Web crawler
		Science and technology expenditure accounts for the proportion of fiscal expenditures (+)
		R&D input intensity (+)
		Applications granted (+)
		The full-time equivalent of R&D personnel (+)

Published by CNKI, the China Economic and Social Big Data Research Platform is a large-scale statistical database that integrates statistical data resource integration, deep data mining analysis, and quick search of multi-dimensional statistical indicators, which brings together important official data released in China over the years.

3.1.2. Theoretical Analysis

The factors affecting city–industry integration mainly include industrial development, urban functions, and residents’ needs. However, complete living facilities, good ecological environment, and abundant scientific resources can enhance the city’s carrying capacity and promote the construction of an innovative ecological chain. Through the previous combing of relevant literature, it can be seen that some scholars have included social services [21], ecological environment [24], and science and technology [25] into the index category separately.

We chose to comprehensively consider these three elements together and construct a city–industry integration index system on the principle of scientificity, systematicity, and accessibility.

3.2. Normalization

To prevent the influence of dimension and numerical size on the evaluation, the data were standardized by the range method.

$$\mathrm{Positive} \mathrm{indicator}: y_{ij}=\frac{x_{ij}-min\left(x_{ij}\right)}{max(x_{ij})-min\left(x_{ij}\right)}$$

(1)

$$\mathrm{Negative} \mathrm{indicator}: y_{ij}=\frac{max\left(x_{ij}\right)-x_{ij}}{max(x_{ij})-min\left(x_{ij}\right)}$$

(2)

where x_ij is the value of indicator j in year i, y_ij is the normalized value. $max\left(x_{ij}\right)$ is the maximum value of indicator j, and $min\left(x_{ij}\right)$ is the minimum value of indicator j. When the indicator generates positive or negative contributions to the system, Equations (1) and (2) are applied, respectively.

3.3. Combination Empowerment

To prevent the final evaluation results from being inconsistent with the facts, the weights in the evaluation indicator system consist of subjective weight and objective weight [53]. The detailed generation process is described as follows.

3.3.1. Subjective Weight Definition (PSO-AHP)

PSO is a global optimization algorithm that is derived from the study of the predation behavior of flocks of birds. Its basic idea is to randomly initialize a group, and then correct the speed and position of individuals in the group through an iterative process until the end of the algorithm [54]. As a very representative subjective empowerment method, AHP is convenient and fast to calculate, is not limited by objective data, and is widely used in various fields of research. However, there are certain deficiencies in the AHP model consistency test. When the consistency test is not satisfied, the expert judgment matrix needs to be adjusted, and it is impossible to retain the expert preference. In addition, once the judgment matrix is determined, the value of the consistency indicator function cannot be improved [55].

The PSO-AHP method was presented in 2011 [56] and is now widely used in weight determination [57]. The goal function is to minimize the consistency index of the AHP, and the weight calculation of the evaluation index is combined with the consistency test of the judgment matrix. It is possible to obtain the optimal consistency index test value without changing the judgment matrix, as well as to maintain the original information of the judgment matrix to the greatest extent. Thus, the solved weight value can better reflect the real preference of the decision maker and is more reasonable.

• Establish a hierarchical model: target layer; criterion layer; indicator layer.
• Construct a comparison matrix A_k. Using the nine-scale method for scaling, the values of the matrix elements are compared to reflect the relative importance of several factors. Get the comparison matrix.

  $$A=\left[\begin{array}{cccc}{a}_{11}& a_{12}& \cdots & a_{1n}\\ a_{21}& a_{22}& \cdots & a_{2n}\\ ⋮& ⋮& ⋮& ⋮\\ a_{n1}& a_{n2}& \cdots & a_{nn}\end{array}\right]$$

  (3)
• Determination of objective function.
  • For the judgment matrix A_k, let the single ordering weight of each element be ${\omega }_k$, $k=1~n$. If the judgment matrix A_k satisfies $a_{ij}=\raisebox{1ex}{${\omega }_i$}\!\left/ \!\raisebox{-1ex}{${\omega }_j$}\right.\left(i,j=1~n\right)$, then A_k has complete consistency, that is, the consistency index is equal to 0.

    $${\displaystyle \sum }_{k=1}^n\left(a_{ik}{\omega }_k\right)={\displaystyle \sum }_{k=1}^n(\frac{{\omega }_i}{{\omega }_k}){\omega }_k=n{\omega }_i$$

    (4)

    $${\displaystyle \sum }_{i=1}^n{\displaystyle \sum }_{k=1}^n\left(a_{ik}{\omega }_k\right)-n{\omega }_i=0$$

    (5)
  • Obviously, the smaller the value at the left end of Equation (4), the higher the degree of the consistency of A_k. When Equation (5) is established, A_k has complete consistency. Therefore, the weight calculation and optimization of each element can be transformed into the following optimization problem:

    $$\mathrm{The} \mathrm{objective} \mathrm{function}: Min{F}_{CI}\left({\omega }_k\right)={{\displaystyle \sum }}_{i=1}^n\left|{{\displaystyle \sum }}_{k=1}^n\left(a_{ik}{\omega }_k)\right)-n{\omega }_i\right|/n$$

    (6)

    $$\mathrm{The} \mathrm{constraints}: \{\begin{array}{c}{\displaystyle \sum }_{k=1}^n{\omega }_k=1\hfill \\ {\omega }_k>0\left(k=1~n\right)\end{array}$$

    (7)
  • In Equation (6), $F_{CI}\left(n\right)$ is the consistency index function, and when the minimum value is obtained, the corresponding weight value is the optimal weight corresponding to matrix A_k.

    $${\omega }_i^A={\displaystyle \sum }_{k=1}^n{\omega }_k • w_i^k$$

    (8)

    $$F_{CI}^A\left({\omega }_i^A\right)={\displaystyle \sum }_{k=1}^n{\omega }_k • F_{CI}^k\left({\omega }_i^k\right)$$

    (9)

    where ${\omega }_i^A$ is the medium weight of the lower element to the target layer, and $w_i^k$ is the weight of the element at the lower level relative to the element at the upper level. $F_{CI}^A\left({\omega }_i^A\right)$ is the consistency indicator calculated by the final weight, and when its value is less than 0.1, it can be considered that the total sort result has satisfactory consistency.
• Determination of objective function. The algorithm flow is shown in Figure 3.
  • Determine the parameters of the PSO algorithm, namely, the number of populations n, the maximum number of iterations N, and the learning factors c₁ and c₂; the variation range of v_m is the inertia coefficient $\left[-v_{max},v_{max}\right]$. In this article, n = 100, N = 500, c₁ = c₂ = 2, −v_max = −0.001, and v_max = 0.001.
  • Normalize the weights to generate the initial solution of the particles. The feasible solution is substituted into Equation (5) to calculate the fitness of the initial particles, and the global optimal particles are selected from them. Update iterates over particles with Equations (10) and (11).

    $$v_{in}{}^{k+1}=v_{in}{}^k+c_1{r}_1\left(p_{in}{}^k-x_{in}{}^k\right)+c_2{r}_2\left(p_{gn}{}^k-x_{in}{}^k\right)$$

    (10)

    $$x_{in}{}^{k+1}=x_{in}{}^k+v_{in}{}^k+1 \left(i=1,2\dots m;n=1,2,\dots N\right)$$

    (11)
  • After updating the fitness of the particles, compare and select the optimal position of the particles and the global. If the iteration termination condition is met, the iteration terminates, outputs the optimal solution obtained by the model, and moves on to the next step. If the iteration termination condition is not met, jump back to the first step until the iteration termination condition is met.
  • Substitute the obtained optimal value into the objective function and calculate the consistency ratio value corresponding to the judgment matrix. If the consistency requirement is met, the next step is conducted; otherwise, adjust the judgment matrix and re-execute the loop. Finally, the global optimal position and corresponding weight value, and consistency indicator function value MinF_CI are output.

3.3.2. Objective Weight Definition (EM)

To improve the data accuracy, EM is used to modify the weights obtained by AHP [58]. The smaller the information entropy of the index, the greater the variation degree of the index value, the more information it provides, the more importance it can play in the comprehensive evaluation, and the larger its weight.

• Calculate the characteristic weight.

  $$p_{ij}=\frac{x_{ij}}{{{\displaystyle \sum }}_{i=1}^n{x}_{ij}}$$

  (12)
• Calculate the information entropy e_j and difference coefficient g_j of the jth evaluation index.

  $$e_j=-\frac{1}{\ln n}{\displaystyle \sum }_{i=1}^n{p}_{ij}\ln p_{ij}$$

  (13)

  $$g_j=1-e_j$$

  (14)
• Calculate the weight of the jth index.

  $${\omega }_j=\frac{g_j}{{{\displaystyle \sum }}_{i=1}^m{g}_j}$$

  (15)

3.3.3. Blends Weighs

For the robustness of the weights, a combination of subjective and objective assignments is required. This paper adopts the weight combination method proposed by Shang [59].

• Represent weights as a linear combination of subjective weights ${\omega }_a$ and objective weights ${\omega }_e$.

  $$\omega =\alpha {\omega }_a+\left(1-\alpha \right){\omega }_e$$

  (16)

  where $\alpha$ is the proportion of subjective weights to the combined weights, and $\left(1-\alpha \right)$ is the proportion of objective weights to the combined weights.
• The goal is to minimize the square sum of the deviation between the combined weight and the subjective weight, and the square sum of the deviation between the combined weight and the objective weight. Moreover, substitute the objective function into Equation (16).

  $$\mathrm{The} \mathrm{objective} \mathrm{function}: minz={{\displaystyle \sum }}_{i=1}^n\left[{\left(\omega -{\omega }_a\right)}^2+{\left(\omega -{\omega }_e\right)}^2\right]$$

  (17)

  $$\underset{\alpha }{min}{\displaystyle \sum }_{i=1}^n\left\{{\left[\left(\alpha {\omega }_a+\left(1-\alpha \right){\omega }_e-{\omega }_a\right)\right]}^2+{\left[\left(\alpha {\omega }_a+\left(1-\alpha \right){\omega }_e-{\omega }_e\right)\right]}^2\right\}$$

  (18)
• Take the derivative of Equation (18) with respect to $\alpha$ and let the first derivative be zero, then obtain $\alpha =0.5$, and substitute it into Equation (16).

  $$\omega =0.5\times {\omega }_a+0.5\times {\omega }_e$$

  (19)

The derivation result shows that the best result is that the subjective weights and objective weights each account for 50%, when the sum of the squares of the deviations between the combined weights and the subjective and objective weights, respectively, is minimized. This shows that the subjective and objective perceptions of the importance of the indicators are the same.

3.4. Evaluation Method (GRA-TOPSIS)

The purpose of the GRA technique is to measure the relationship between elements on the basis of similarity [60]. It has significant advantages in solving complex decision-making problems characterized by vague, incomplete, and inaccurate information [61]. TOPSIS is a management decision-making method for comparing and ranking multidimensional vectors, in which Euclidean distance calculation is the core of the model. This method can systematically analyze the gap between the object and the ideal state, truly reflect the problem, and provide clear information for decision makers [62]. However, TOPSIS cannot accurately reflect the changing trend of data series, and GRA ignores the correlation situation between the scheme and the ideal solution [63]. By defining the gray association coefficient of GRA, the limitations of TOPSIS can be improved, and the purpose of correcting the relative proximity can be achieved [35]. The procedure of the GRA–TOPSIS method is as follows.

• Determine the positive ideal solution Z⁺ and negative ideal solution Z⁻ of the weighted normalized matrix.

  $$Z^{+}=\left(z_1{}^{+},z_2{}^{+},\dots ,z_n{}^{+}\right)=\omega$$

  (20)

  $$Z^{-}=\left(z_1{}^{-},z_2{}^{-},\dots ,z_n{}^{-}\right)=0$$

  (21)

  where $z_j{}^{+}=max{\mathrm{z}}_{ij}={\omega }_j, z_j{}^{-}=min{z}_{ij}=0, j\in N$.
• Calculate the Euclid distance to the positive ideal solution di⁺ and negative ideal solution di⁻ for each scheme.

  $$d{i}^{+}=\sqrt{{{\displaystyle \sum }}_{j=1}^n{(zij-z{j}^{+})}^2} \mathrm{and} d{i}^{-}=\sqrt{{{\displaystyle \sum }}_{j=1}^n{(zij-z{j}^{-})}^2}$$

  (22)
• Calculate the gray correlation coefficient matrix between each scheme and the positive R⁺ and negative ideal solution R⁻.

  $$R^{+}={(r_{ij}{}^{+})}_{m\times n} \mathrm{and} R^{-}={(r_{ij}{}^{-})}_{m\times n}$$

  (23)

  $$r_{ij}{}^{+}=\frac{\underset{i}{min}\underset{j}{min}\left|z_j{}^{+}-z_{ij}\right|+\epsilon \underset{i}{max}\underset{j}{max}\left|z_j{}^{+}-z_{ij}\right|}{\left|z_j{}^{+}-z_{ij}\right|+\epsilon \underset{i}{max}\underset{j}{max}\left|z_j{}^{+}-z_{ij}\right|}$$

  (24)

  $$r_{ij}{}^{-}=\frac{\underset{i}{min}\underset{j}{min}\left|z_j{}^{-}-z_{ij}\right|+\epsilon \underset{i}{max}\underset{j}{max}\left|z_j{}^{-}-z_{ij}\right|}{\left|z_j{}^{-}-z_{ij}\right|+\epsilon \underset{i}{max}\underset{j}{max}\left|z_j{}^{-}-z_{ij}\right|}$$

  (25)

  where $\epsilon$ is the resolution coefficient ($\epsilon \in \left[0, 1\right]$). We take the value of 0.5 [64].
• Calculate the gray correlation degree of each scheme with the positive ideal solution r⁺ and the negative ideal solution r⁻.

  $$r^{+}=\frac{1}{n}{{\displaystyle \sum }}_{j=1}^n{r}_{ij}{}^{+} \mathrm{and} r^{-}=\frac{1}{n}{{\displaystyle \sum }}_{j=1}^n{r}_{ij}{}^{-}$$

  (26)
• Dimensionless processing is performed on Euclid distance and correlation degree.

  $$D_i{}^{+}=\frac{d_i{}^{+}}{max{d}_i{}^{+}} \mathrm{and} D_i{}^{-}=\frac{d_i{}^{-}}{max{d}_i{}^{-}}$$

  (27)

  $$R_i{}^{+}=\frac{r_i{}^{+}}{max{r}_i{}^{+}} \mathrm{and} R_i{}^{-}=\frac{r_i{}^{-}}{max{r}_i{}^{-}}$$

  (28)
• Combine dimensionless distance and dimensionless correlation. The larger the values of D_i⁻ and R_i⁺, the closer the scheme is to the positive ideal solution, and so the combined formula can be determined as Equation (30).

  $$P_i{}^{+}=\alpha D^{-}+\beta R^{+} \mathrm{and} P_i{}^{-}=\alpha D^{+}+\beta R^{-}$$

  (29)

  $$Q{i}^{+}=\frac{P_i{}^{+}}{P_i{}^{+}+P_i{}^{-}}$$

  (30)

  where $\alpha =\beta =\frac{1}{2}$. Sorting the cases according to the size of the relative closeness, the closer the degree is to one, the better the scheme is.

3.5. GINI

The Gini coefficient is an effective statistical tool for equilibrium analysis and an indicator of the degree of difference [42]. In this paper, the Gini coefficient is used to reflect the regional difference of the city–industry integration degree, and its size is proportional to the size of the regional difference [65]. The calculation process is as follows.

• Suppose X represents a non-negative random variable with a probability distribution function of $F\left(X\right)=P \left(X\le x\right)$, then there is a corresponding quantile function $F^{-1}\left(p\right)=\inf \left\{x: F\left(X\right)\ge P\right\}$. If the form of the quantile function is determined as $G_F\left(p\right)=1-\frac{L_F\left(p\right)}{p}$, the functional form of the Gini coefficient is obtained.

  $$G_F={{\displaystyle \int }}_0^1\left(1-\frac{L_F\left(p\right)}{p}\right)\phi \left(p\right)dp$$

  (31)

  where $\phi \left(p\right)=2p$, and $\phi \left(p\right)\in \left[0, 1\right]$.
• Equation (31) is the continuous function form of the Gini coefficient, but the city–industry integration degree of each city from 2010 to 2019 calculated in this paper is discrete. Therefore, Fan’s calculation formula is adopted [65]:

  $$G=\frac{2}{k}{\displaystyle \sum }_{i=1}^{k}i{e}_i-\frac{k+1}{k}$$

  (32)

  where $e_i=\frac{y_i}{{{\displaystyle \sum }}_{i=1}^k{y}_i}, e_1<e_2<\cdots <e_k$. k represents the number of regions; y_i represents the composite score of the ith region; and e_i represents the overall score of the ith region as a share of the population, in order from lowest to highest.

3.6. Theil Exponential Decomposition

The Gini coefficient is often used to calculate inequality in general and does not explain the source of inequality (from within or between regions) [66]. Therefore, it is combined with the Theil index, an indicator of income level difference based on the concept of information entropy [67] to reflect the source of regional difference in the degree of city–industry integration. Its size also has a positive correspondence with the regional difference.

$$T=\frac{1}{n}{\displaystyle \sum }_{i=1}^n\frac{y_i}{y}\ln \left(\frac{y_i}{y}\right)$$

(33)

$$T=T_b+T_w={\displaystyle \sum }_{k=1}^k{y}_k\ln (\frac{y_k}{\raisebox{1ex}{$n_k$}\!\left/ \!\raisebox{-1ex}{$n$}\right.})+{\displaystyle \sum }_{k=1}^k{y}_k({\displaystyle \sum }_{i\in g_k}\frac{y_i}{y_k}\ln \frac{\raisebox{1ex}{$y_i$}\!\left/ \!\raisebox{-1ex}{$y_k$}\right.}{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$n_k$}\right.})$$

(34)

T is the Theil index, and its value range is [0, 1]. y_i represents the city–industry integration degree of each city, y is the sample average, y_k is the integration degree of each region, and g_k represents the number of groups. Decomposing the Theil index into inter-group difference T_w and intra-group difference T_b can more intuitively reflect the uneven characteristics of city–industry integration in various regions.

4. Case Study

4.1. Research Case Area

Sichuan Province, as a populous province, occupies a prominent position in the overall situation of national urbanization development. In 2017, its urbanization rate exceeded 50% for the first time, which is a large gap compared with other developed regions, and many problems need to be solved urgently. Sichuan attaches significant importance to the integration of industry and city as an important measure to promote the construction of new urbanization. However, statistics show that the level and growth rate of urban development in different regions of the province are quite different.

Therefore, Sichuan is selected as the research object to help the construction of new urbanization in Western China, so that the impact can be radiated to the whole country. Sichuan Province is in the hinterland of Southwest China, with a total area of 486,000 square kilometers, including 18 prefecture-level cities and three ethnic minority autonomous prefectures. Considering the availability and continuity of data, three ethnic autonomous prefectures were excluded. The detailed location of the study area and urbanization rate are shown in Figure 4.

4.2. Data Acquisition and Result

The data used in this study were obtained from the Sichuan Statistical Yearbook (2011–2020), the China City Statistical Yearbook (2011–2020), the China Urban Construction Statistical Yearbook (2011–2020), and statistical yearbooks of every city. The missing data were filled by interpolation and linear regression.

By inviting 12 experts from the Center for Rural Construction Integrated Management to score the AHP judgment matrix, the final weight allocation result was obtained.

4.2.1. Result of Indicator Data Processing

The result pairs of the PSO-AHP and AHP solutions constructed above are shown in

Table 3. Comparison table of PSO-AHP and AHP judgment matrix calculation results.

Model	Judgment Matrix	W1	W2	W3	W4	W5	W6	Consistency Indicator Function Value
PSO-AHP	B	0.2661	0.1944	0.1681	0.1650	0.1328	0.0736	0.0455
	C1	0.3623	0.1453	0.2974	0.1950			0.0242
	C2	0.2326	0.1534	0.3478	0.2662			0.0367
	C3	0.2495	0.1236	0.3990	0.2279			0.0322
	C4	0.3623	0.1092	0.1966	0.3319			0.0288
	C5	0.4113	0.1526	0.1076	0.3285			0.0228
	C6	0.0978	0.3761	0.1972	0.1666	0.1623		0.0225
AHP	B	0.2490	0.1917	0.1645	0.1604	0.1221	0.1123	0.0625
	C1	0.3539	0.1642	0.2918	0.1902			0.0552
	C2	0.2263	0.1825	0.3322	0.2589			0.0777
	C3	0.2427	0.1517	0.3910	0.2147			0.0804
	C4	0.3507	0.1345	0.1935	0.3214			0.0800
	C5	0.4012	0.1498	0.1274	0.3216			0.0682
	C6	0.1169	0.3690	0.1936	0.1642			0.0450

. The results show that the PSO-AHP weight optimization model had a smaller value of the consistency index function than the traditional AHP model. According to the theory of analytic hierarchy, the smaller the CI value, the higher the consistency of the judgment matrix, and the more reasonable the corresponding weights. Therefore, the PSO-AHP model can keep the original information of the judgment matrix to the greatest extent when the judgment matrix was determined. That is, the solved weight values can better reflect the real preferences of the decision maker, and the obtained weights are more reasonable.

Use Equation (16) to couple the weights obtained by PSO-AHP and EM to obtain the comprehensive weights, as shown in

Table 4. Evaluation index weights in 2019.

Criteria Layer	Indicators *	PSO-AHP	EM	Comprehensive Weight
Economy and industry	PG	0.0964	0.0231	0.0598
	PSOVG	0.0387	0.0104	0.0245
	PTOVG	0.0791	0.0181	0.0486
	GIOVEAS	0.0519	0.0340	0.0430
Habitat and employment	HDB	0.0452	0.0802	0.0627
	PES	0.0298	0.0107	0.0202
	PUTP	0.0676	0.0221	0.0449
	PCHA	0.0517	0.0111	0.0314
Social services	NB	0.0419	0.0285	0.0352
	NA	0.0208	0.0391	0.0300
	CSME	0.0671	0.0680	0.0676
	NS	0.0383	0.0201	0.0292
Urban construction	IFAC	0.0598	0.0947	0.0773
	PII	0.0180	0.0134	0.0157
	PBF	0.0324	0.0552	0.0438
	PURA	0.0548	0.0095	0.0322
Ecology and environment	EEA	0.0546	0.0271	0.0408
	CURISW	0.0203	0.0054	0.0128
	WD	0.0143	0.0139	0.0141
	RRECPG	0.0436	0.0196	0.0316
Science and technology	NP	0.0072	0.0382	0.0227
	SEAPUFE	0.0277	0.0270	0.0283
	RDII	0.0145	0.0614	0.0379
	AG	0.0123	0.1718	0.0921
	FERDP	0.0119	0.0975	0.0547

* The indicators are expressed in abbreviations.

.

4.2.2. Result of the Development of City–Industry Integration

• Overall results and discussion of city–industry integration.

Sorting the level of city–industry integration in 18 prefecture-level cities in Sichuan Province can clearly reflect the changes of city–industry integration. As shown in

Table 5. Changes in the development of city–industry integration.

Region	2010		2011		2015		2018		2019
	Score	Ranking	Score	Ranking	Score	Ranking	Score	Ranking	Score	Ranking
Chengdu	0.6282	1	0.5970	1	0.6228	1	0.6390	1	0.6300	1
Zigong	0.3835	4	0.3745	7	0.3790	6	0.3682	13	0.3718	9
Panzhihua	0.3975	3	0.3963	4	0.3910	4	0.4062	2	0.3917	3
Luzhou	0.3529	16	0.3395	17	0.3570	13	0.3657	14	0.3639	14
Deyang	0.3827	6	0.3755	6	0.3793	5	0.3851	5	0.3976	2
Mianyang	0.4186	2	0.4077	3	0.4164	3	0.3891	3	0.3786	7
Guangyuan	0.3634	12	0.3494	13	0.3498	15	0.3864	4	0.3912	4
Suining	0.3548	14	0.3523	11	0.3707	8	0.3719	10	0.3547	18
Neijiang	0.3559	13	0.3468	15	0.3539	14	0.3352	18	0.3612	16
Leshan	0.3832	5	0.3770	5	0.3764	7	0.3698	11	0.3714	10
Nanchong	0.3784	7	0.3749	9	0.3671	11	0.3724	8	0.3711	11
Meishan	0.3644	11	0.3677	2	0.4251	2	0.3730	7	0.3759	8
Yibin	0.712	8	0.3558	10	0.3680	10	0.3739	6	0.3797	6
Guang’an	0.3665	10	0.3749	16	0.3432	16	0.3624	16	0.3669	12
Dazhou	0.3446	17	0.3677	14	0.3366	18	0.3697	12	0.3609	17
Ya’an	0.3708	9	0.3558	8	0.3689	9	0.3720	9	0.3908	5
Bazhong	0.3414	18	0.3735	18	0.3378	17	0.3646	15	0.3643	13
Ziyang	0.3540	15	0.3735	12	0.3597	12	0.3482	17	0.3616	15

, with the socio-economic development, the level of city–industry integration development in the 18 cities showed a relative upward trend during the decade. According to the GRA-TOPSIS evaluation principle, the closer it is to one, the higher the degree of city–industry integration. However, except for Chengdu, the degree of city–industry integration was less than 0.5, which indicates that the overall level of city–industry integration in Sichuan is not high, and there is still much room for improvement.

As the capital city of Sichuan Province, Chengdu’s development level has always been significantly higher than other cities. However, its index has been rising more slowly, which is related to its need to drive the whole Chengdu Plain Economic Zone. Moreover, with the construction of the Chengdu Metropolitan Area, Chengdu will further exert its influence on Deyang, Meishan, and Ziyang. Mianyang is the China Science and Technology City, its development level is stable at the forefront, and it was successfully selected as a national demonstration zone for city–industry integration in 2016. However, its city–industry integration has slightly decreased from 2018, which is related to the business environment reform in the Hi-tech Zone from the second half of 2017. Therefore, there is a reverse change in the values of some indicators of urban construction as well as ecology and environment in Mianyang. Panzhihua, famous for its secondary industry, has been steadily ranked in the top five, while scientific and technological innovation has developed rapidly.

Bazhong is backward due to its geographical location in remote mountainous areas, unclear advantageous industries, and low scientific level. However, the degree of city–industry integration has risen from 0.3414 to 0.3643, and the ranking has also risen from 18 to 13. This means that Bazhong, located in the geometric center of Chengdu, Chongqing, and Xi’an, relies on its regional advantages to promote its stable development in all aspects. Although the level of urbanization and economic development in Neijiang has improved, it has been labeled as a “large agricultural city and weak industrial city”. Therefore, the development level of city–industry integration is slow, and the ranking has not been significantly improved. Dazhou, as an underdeveloped area in China, is backward in terms of its productivity and innovation ability, but it is rich in steel resources and has enormous potential for economic and social development. On the one hand, Dazhou has vigorously developed steel in urban areas in recent years, resulting in an improper discharge of “three wastes”, and greater constraints on the development of upstream and downstream industries. On the other hand, Dazhou’s public construction has not been perfect, and people’s livelihood cannot be better guaranteed. Thus, the development of Dazhou’s city–industry integration has shown repeated up-and-down fluctuations.

2.

Evaluation results and discussion of each subsystem.

The difference of the city–industry integration degree indicates that the development of city–industry integration also has the property of regional differences. According to the Function orientation of five economic zones in Sichuan, this paper divided Sichuan into the Chengdu Plain Economic Zone, the South Sichuan Economic Zone, the Northeast Sichuan Economic Zone, and Panzhihua, as shown in Figure 5. The development level of city–industry integration is affected by many aspects, and GRA-TOPSIS was used to calculate the closeness of six subsystems in each city. The results of 2010 were compared with those of 2019, as shown in

Table 6. The development level of each index category.

Region		Chengdu Plain Economic Zone	South Sichuan Economic Zone	Northeast Sichuan Economic Zone	Panzhihua
Economic industry	2010	2.8091	1.2005	1.3908	0.4190
	2019	2.9977	1.2744	1.4264	0.4108
	Change	0.1886	0.0739	0.0356	−0.0082
Habitat and employment	2010	3.5818	1.5358	2.0731	0.3790
	2019	3.6058	1.6039	2.2388	0.4060
	Change	0.0240	0.0681	0.1657	0.0270
Social services	2010	3.0621	1.3298	1.6086	0.3882
	2019	2.9938	1.3373	1.6316	0.6313
	Change	−0.0683	0.0075	0.0230	0.2431
Urban construction	2010	3.1851	1.3388	1.7931	0.3819
	2019	3.2248	1.3354	1.8320	0.3837
	Change	0.0397	−0.0034	0.0389	0.0018
Ecology and environment	2010	3.1164	1.6486	2.0382	0.3973
	2019	2.9344	1.8503	1.9834	0.3614
	Change	−0.1820	0.2017	−0.0548	−0.0359
Science and technology	2010	2.6773	1.0293	1.0989	0.2893
	2019	2.7376	1.0297	1.0845	0.2644
	Change	0.0603	0.0004	−0.0144	−0.0249

. In general, the whole of Sichuan has a certain growth in habitat and employment. As the dynamic system of city–industry integration, people provide labor and technical support. The rising score of this subsystem indicates that the development of city–industry integration in Sichuan has driven most people.

The Chengdu Plain Economic Zone has decreased in terms of social services and ecology and environment, which shows that this zone is prone to lose coordination efforts on service and ecology while rapidly developing industrial transformation and urban construction. The South Sichuan Economic Zone has decreased in terms of urban construction, while the Northeast Sichuan Economic Zone has decreased in terms of ecology and environment, as well as science and technology. These two economic zones are both eco-cultural tourism zones, with industrial development in southern Sichuan and agricultural development in northeast Sichuan, indicating that both economic zones still need to strengthen their industrial transformation. Although Panzhihua’s overall level is high and has increased, it has decreased in three major categories: industry and economy, ecology and environment, and science and technology. This prompted Panzhihua to focus on synergistic development of industrial restructuring and ecological protection when building Vanadium and Titanium New City.

4.2.3. Differences in the Development Level of City–Industry Integration

As Panzhihua is the only city in the Panxi Economic Zone in this paper, the study of its regional differences was not studied. It can be seen from Figure 6 that during the sample investigation period, the Gini coefficient and Theil index had strong consistency in the changing trend, and their increase and decrease were synchronous, indicating that both indicators can corroborate each other in measuring the imbalance of city–industry integration. The Gini coefficient of the city–industry integration degree between economic zones fluctuated between 0.16 and 0.2, and the Theil index fluctuated between 0 and 0.02. From the Gini coefficient, the development difference between economic zones was within a reasonable range, and the fluctuation range was small. This shows that while regional differences have expanded after slight fluctuations in the past decade, Sichuan remains focused on overall balanced development.

Decomposing the Theil Index can effectively derive the sources of regional development differences, as shown in

Table 7. Theil index and structural decomposition of development of city–industry integration in each region.

Year	Overall Difference	Inter-Regional Differences	Intra-Regional Differences	Chengdu Plain Economic Zone	South Sichuan Economic Zone	Northeast Sichuan Economic Zone
2010	0.0047	0.0005 (10.73%)	0.0042 (89.27%)	0.0041 (87.80%)	0.0000 (0.40%)	0.0001 (1.07%)
2011	0.0052	0.0006 (11.32%)	0.0046 (88.68%)	0.0045 (86.45%)	0.0001 (1.62%)	0.0000 (0.61%)
2012	0.0067	0.0014 (20.69%)	0.0053 (79.31%)	0.0052 (77.55%)	0.0000 (0.66%)	0.0001 (1.10%)
2013	0.0065	0.0013 (20.51%)	0.0051 (79.49%)	0.0051 (78.69%)	0.0000 (0.38%)	0.0000 (0.42%)
2014	0.0053	0.0014 (26.23%)	0.0039 (73.77%)	0.0038 (71.99%)	0.0000 (0.67%)	0.0001 (1.11%)
2015	0.0061	0.0010 (16.91%)	0.0050 (83.09%)	0.0049 (80.63%)	0.0001 (1.49%)	0.0001 (0.97%)
2016	0.0074	0.0015 (19.67%)	0.0060 (80.33%)	0.0058 (78.37%)	0.0001 (1.24%)	0.0001 (0.72%)
2017	0.0072	0.0015 (20.24%)	0.0057 (79.76%)	0.0056 (78.23%)	0.0001 (1.26%)	0.0001 (0.27%)
2018	0.0052	0.0014 (26.76%)	0.0038 (73.24%)	0.0037 (71.17%)	0.0001 (1.32%)	0.0000 (0.74%)
2019	0.0052	0.0009 (14.65%)	0.0044 (83.35%)	0.0043 (82.55%)	0.0001 (1.07%)	0.0001 (1.74%)

. Overall, the development level of city–industry integration in Sichuan Province had a slight difference, and the spatial imbalance was mainly due to intra-regional differences. From 2010 to 2019, the Theil index increased from 0.0047 to 0.0052, and the difference gradually expanded. The intra-regional difference index increased from 0.0042 to 0.0044, always much higher than the inter-regional difference index. Therefore, the main reason for the spatial imbalance of city–industry integration in Sichuan was the internal differences of various economic zones. From the difference of city–industry integration degree within the economic zone, the Chengdu Plain Economic Zone had the largest value, while the other two economic zones had the smallest and most stable values. Moreover, the contribution rate of the Theil index of the Chengdu Plain Economic Zone was the highest, which decreased from 87.80% to 82.55%. It indicates that the unbalanced development of city–industry integration in the Chengdu Plain Economic Zone has been alleviated to a certain extent, but the internal differentiation is still serious. This is mainly because the development level of Chengdu, Deyang, and Mianyang is much higher than that of the other five cities. South Sichuan and northeast Sichuan economic zones had the lowest contribution rate, and the development level of city–industry integration was balanced.

4.3. Policy Suggestions

(1)

The development level of city–industry integration is not only affected by a single factor. In the development of urbanization, we should pay attention to the comprehensive coordination of industry, city, residents, environment, and science. People are an intermediate bridge connecting industries and cities, and the development level of people’s lives is an important measure of the quality of city–industry integration. Therefore, urbanization development should pay attention to increasing population welfare expenditure and improving the medical and health environment. At the same time, urbanization development should abandon the extensive industrial development mode with high pollution and high energy consumption and instead pay attention to the construction and protection of the urban ecological environment. The development concept of innovation, coordination, green, openness, and sharing is the fundamental idea for promoting the high-quality development of new urbanization in an all-around way.

(2)

Given the imbalance of regional development, it is necessary for the government to fundamentally play a role in resource allocation and provide a fair development environment for different cities, strengthening regional cooperation, cultivating urban agglomeration and support, and coordinating cross-regional development in order to weaken the regional differences of city–industry integration development. We took intraregional cooperation as a breakthrough in the reduction of the imbalance within the Chengdu Plain Economic Zone. The South Sichuan Economic Zone will further build a multi-center urban agglomeration integrated innovation and development pilot zone to improve its strategic position. Moreover, the Northeast Sichuan Economic Zone will provide full play to its eco-tourism strengths for industrial transformation, improving the scientific and technological innovation ability of south Sichuan and northeast Sichuan economic zones, establishing a cross-regional coordinated development mechanism, and realizing sustainable development.

5. Conclusions and Prospects

5.1. Conclusions

This paper conducted a comprehensive study on the development of city–industry integration under the background of new urbanization. To fully reflect the development of city–industry integration, social services, ecology and environment, and science and technology were integrated into the discussion of city–industry integration. We combined web crawler with literature research to build an index system based on multi-source data [49]. Because AHP cannot improve consistency after a given judgment matrix, the PSO algorithm was introduced to optimize its weights [57], and the EM model was added to obtain real and credible weights. GRA-TOPSIS was used to evaluate the level of city–industry integration from multiple attributes. Finally, to explore the degree and source of unbalanced development, the Gini coefficient and Theil index were used for calculations. Taking 18 prefecture-level cities in Sichuan Province from 2010 to 2019 as an example, the main conclusions are as follows:

(1) By comparing the consistency index function values obtained by PSO-AHP and AHP, it can be concluded that the PSO-AHP model can achieve the optimization of the weight values when the judgment matrix has been determined [55]. (2) At present, although the overall level of city–industry integration development in Sichuan Province is still relatively low, it shows an upward trend. Only Chengdu’s development level is much higher than the average [14], and the index scores of each city fluctuate up and down. (3) Dividing Sichuan province into different economic zones, from the scores of each subsystem and the relative changes of scores, there are certain regional differences in the development of city–industry integration. The overall imbalance of the development level of city–industry integration in Sichuan is small. The main reason for the imbalance is the significant differences within the Chengdu Plain Economic Zone. The methods and conclusions proposed in this study enrich the theory of city–industry integration and play a supporting role in the formulation of related policies.

5.2. Shortcomings and Prospects

The research on the development of city–industry integration is a dynamic work and needs to be further improved.

5.2.1. Study Limitations

(1)

Since the indicator data are characteristic of Sichuan Province, and the development of city–industry integration in each area has its unique characteristics, if the indicator system is applied to other regions or extended to the whole country, the evaluation indicators and corresponding data may need to be adjusted according to the conditions of the study area.

(2)

Although the evaluation system constructed in this paper is comprehensive, the evaluation indicators only select typical indicators, which simplify the complexity of the problem. At the same time, given the accessibility and validity of the data, this paper temporarily does not consider the data of the three autonomous prefectures in Sichuan Province, which has a certain impact on completely reflecting the overall level of Sichuan.

5.2.2. Future Research

(1)

With the updating and supplementation of statistical data, as well as the enrichment and refinement of statistical indicators, the index system of city–industry integration in Sichuan Province can be further improved. On the basis of refining the evaluation indexes, it will be of more reference value to measure the city–industry integration in areas with the same stage and type of urbanization development.

(2)

At present, there are few studies on regional differences for new urbanization, and less on city–industry integration. Therefore, in the future, more methods for regional differences can be introduced into the study of city–industry integration. This paper adopted mathematical statistics methods and spatial measurement methods such as the Moran index that can be introduced in the future, providing further development of the combined use of these two types of methods for spatial imbalance analysis.