Spatial Simulation Of Vulnerability To Re-Poverty Based On Improved PSO-SVM Model

In 2020,the Fifth Plenary Session of the 19 th CPC Central Committee clearly proposed "consolidating and expanding the results of poverty eradication and promoting the strategy of rural revitalization".After the war on poverty,China has entered the post-poverty alleviation era,and consolidating and expanding the results of poverty eradication has become an important task.In this context,the advantages of conducting forward-looking research on re-poverty vulnerability have been highlighted,and it is of great practical significance to make prediction and analysis on regional re-poverty vulnerability in order to formulate targeted policies,improve the effective utilization of regional resources,reduce the risk of returning to poverty or new poverty,and consolidate and expand the achievements of poverty eradication and promote the implementation of rural revitalization strategy.Based on the theoretical framework of "Exposure-Sensitivity-Adaptability",this paper comprehensively evaluates the re-poverty vulnerability of the old revolutionary region of the Left and Right River in Guangxi for five periods from 2000 to 2020.Based on the theoretical framework of "exposure-sensitivity-adaptability",we conducted a comprehensive evaluation of the re-poverty vulnerability of the old revolutionary region of the left and right rivers in Guangxi for five periods from 2000 to 2020,analyzed its spatial and temporal pattern characteristics,explored its main influencing factors,and introduced the normal cloud model and particle redistribution mechanism to improve the Particle Swarm Optimization(PSO),and further optimized the Support Vector Machines(SVM)model.model(SVM)parameters,compared the prediction simulation results of four algorithm models,namely,SVM model,PSO-SVM model,CPSO-SVM model and RPSO-SVM model,and determined that RPSO-SVM model is the optimal and most suitable algorithm model,and applied it to the prediction of the old revolutionary areas of the left and right rivers in Guangxi The predictive simulations of re-poverty vulnerability in 2025,2030 and 2035 were conducted,and the characteristics of its spatio-temporal pattern changes were analyzed.Its main conclusions are as follows:(1)The comprehensive evaluation results based on the "exposure-sensitivity-adaptability" re-poverty vulnerability index system show that the overall re-poverty vulnerability of the old revolutionary areas of the Left and Right River in Guangxi from 2000 to 2020 shows a decreasing trend.The spatial heterogeneity of different levels of re-poverty vulnerability is greater with the change of time.In terms of spatial clustering,the high-high and low-low clustering areas of re-poverty vulnerability in the old revolutionary areas of the left and right rivers in Guangxi have obvious trends of change in the five periods from 2000 to 2020,with the high-high clustering areas showing spatial changes of decrease and the low-low clustering areas showing spatial changes of increase.-From 2000 to 2015,the high-high agglomerations showed little change and were concentrated in the central part of the study area,and by 2020,the high-high agglomerations decreased significantly and were more fragmented.-high agglomerations decreased significantly and were mainly scattered in the northwest;from 2000 to 2020,low-low agglomerations showed a trend of first decreasing and then increasing,where the The clustering characteristics of the low-low agglomerations are all more prominent,and the clustering characteristics of the low-low agglomerations in the south only gradually come to the fore in 2020.In terms of changing trends,from 2000 to 2015,the re-poverty vulnerability projected in the east-west and north-south directions in the old revolutionary areas of the left and right rivers within Guangxi showed a trend of first rising and then falling,indicating that the high areas of re-poverty vulnerability in these four periods were concentrated in the central region and the low value areas were scattered around;in 2020 the spatial characteristics of re-poverty vulnerability turned to The spatial characteristics of re-poverty vulnerability in 2020 turn to a trend of decreasing from west to east and from north to south,and the decrease is larger,indicating that the high value areas of re-poverty vulnerability are in the west and north,and the differences of re-poverty vulnerability in the east-west and north-south directions are larger.(2)Diagnosis of the main influencing factors of re-poverty vulnerability in the old revolutionary areas of the Left and Right River in Guangxi was conducted by means of geographic detectors,and the diagnostic results showed that the main influencing factors of re-poverty vulnerability varied with the change of time,and the main influencing factors ranked in the top 5 in terms of the mean value of the explanatory power of each influencing factor in the five periods from 2000 to 2020were(0.9923)= number of hospital beds(0.9923)> land reserve resources per capita(0.9922)> disposable income per rural resident(0.9919)> forest land area per capita(0.9915),indicating that savings balance per capita,land reserve resources per capita,disposable income per rural resident,and forest land area per capita have the strongest explanatory power for re-poverty vulnerability.The stronger the explanatory power of re-poverty vulnerability,the greater the impact on re-poverty vulnerability.(3)The constructed SVM model,PSO-SVM model,CPSO-SVM model,and RPSO-SVM model were tested and simulated to compare the algorithms in terms of their result accuracy(curve fit),fitness value,and running time,and the results showed that the curve fit was: RPSO-SVM> CPSO-SVM> PSO-SVM> SVM The mean relative error of prediction(MAPE)is SVM(0.0984)> PSO-SVM(0.0928)>CPSO-SVM(0.0564)> RPSO-SVM(0.0349),the algorithm optimization speed is RPSO > CPSO > PSO,and the running time is SVM > CPSO-SVM > PSO-SVM >RPSO-SVM.RPSO-SVM.the results demonstrate that the RPSO-SVM model is more suitable and has more accurate and better results for re-poverty vulnerability prediction simulation.(4)The RPSO-SVM model was applied to predict and simulate the future re-poverty vulnerability of the old revolutionary region of the Left and Right River within Guangxi in 2025,2030 and 2035,and the prediction results showed that the mean relative error(MAPE)of the prediction results in 2025,2030 and 2035 were3.97%,4.04% and 4%,respectively,which further verified that the introduction of the particle The effectiveness of the particle swarm algorithm improved by introducing the redistribution mechanism in the parameter optimization of the support vector machine model is further verified,and it also shows that the RPSO-SVM model has strong reliability and is suitable for the prediction simulation of re-poverty vulnerability.(5)The spatial autocorrelation method and Arc GIS trend analysis tool were used to analyze the spatio-temporal pattern of the predicted results of re-poverty vulnerability in 2025,2030 and 2035 in the old revolutionary areas of the Left and Right River in Guangxi,and the results showed that: the proportion of different levels of re-poverty vulnerability in the old revolutionary areas of the Left and Right River in Guangxi from 2025 to 2035 Although there is spatial heterogeneity in the change trend,the values of different levels of re-poverty vulnerability all show a decreasing trend,and the spatial extent of different levels of re-poverty vulnerability also changes dynamically in different degrees.From 2025 to 2035,the re-poverty vulnerability of the old revolutionary areas of the left and right rivers in Guangxi is decreasing in both east-west and north-south directions,indicating that the re-poverty vulnerability of the old revolutionary areas of the left and right rivers in Guangxi may show the spatial characteristics of high west and low east,high north and low south in these three future periods.This implies that after 2020,the focus of consolidating and expanding the results of poverty eradication,poverty return risk monitoring and poverty prevention and control should be adjusted to the western,northern and northwestern directions of the old revolutionary areas of the Left and Right River in Guangxi.