{"title":"基于人工神经网络和非线性Backstepping控制器的分段太阳能光伏模拟器","authors":"Ambe Harrison, Njimboh Henry Alombah","doi":"10.3103/S0003701X23600285","DOIUrl":null,"url":null,"abstract":"<p>The current state of affairs on the Photovoltaic emulator (PVE) is facing two main challenges: complexity in resolving the nonlinear equations of the photovoltaic (PV) and the problem of effective control of the PVE power conversion stage (PCS). In this paper, a new power electronics-based PVE is proposed to emulate the dynamic and static characteristics of the PV cell/module. The nonlinear equations of the PV cell/module are resolved using a new piecewise segmentation technique, involving the splitting of the current-voltage (<i>I–V</i>) curve into twelve linear segments associated with the letters a to m (a–m). Based on input environmental conditions, a trained artificial neural network (ANN) is constructed to assist the linearization process by predicting the current-voltage boundary coordinates of these segments. By the use of simple linear equations with the boundary coordinates, a reference voltage is then generated for the PVE. A nonlinear backstepping controller is designed to exploit the PVE reference voltage and stabilize the PCS. The stability of the controller is verified by Lyapunov laws. Optimal performance and control of the PCS were ensured by resorting to particle swarm optimization (PSO). The overall system has been investigated in the MATLAB environment with major tests including the response to fast-changing irradiance and temperature, the EN 50530 test, and the response to change in the load. The proposed PVE revealed a satisfactory dynamic performances in mimicking the PV characteristics. Furthermore, the accuracy of the PVE as a function of the mean absolute percentage error (MAPE) was found less than 0.5% even for the worst case of environmental conditions. Experimental validation of the proposed PVE under real environmental conditions further validated its good dynamic and static robustness.</p>","PeriodicalId":475,"journal":{"name":"Applied Solar Energy","volume":null,"pages":null},"PeriodicalIF":1.2040,"publicationDate":"2023-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A New Piecewise Segmentation Based Solar Photovoltaic Emulator Using Artificial Neural Networks and a Nonlinear Backstepping Controller\",\"authors\":\"Ambe Harrison, Njimboh Henry Alombah\",\"doi\":\"10.3103/S0003701X23600285\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>The current state of affairs on the Photovoltaic emulator (PVE) is facing two main challenges: complexity in resolving the nonlinear equations of the photovoltaic (PV) and the problem of effective control of the PVE power conversion stage (PCS). In this paper, a new power electronics-based PVE is proposed to emulate the dynamic and static characteristics of the PV cell/module. The nonlinear equations of the PV cell/module are resolved using a new piecewise segmentation technique, involving the splitting of the current-voltage (<i>I–V</i>) curve into twelve linear segments associated with the letters a to m (a–m). Based on input environmental conditions, a trained artificial neural network (ANN) is constructed to assist the linearization process by predicting the current-voltage boundary coordinates of these segments. By the use of simple linear equations with the boundary coordinates, a reference voltage is then generated for the PVE. A nonlinear backstepping controller is designed to exploit the PVE reference voltage and stabilize the PCS. The stability of the controller is verified by Lyapunov laws. Optimal performance and control of the PCS were ensured by resorting to particle swarm optimization (PSO). The overall system has been investigated in the MATLAB environment with major tests including the response to fast-changing irradiance and temperature, the EN 50530 test, and the response to change in the load. The proposed PVE revealed a satisfactory dynamic performances in mimicking the PV characteristics. Furthermore, the accuracy of the PVE as a function of the mean absolute percentage error (MAPE) was found less than 0.5% even for the worst case of environmental conditions. Experimental validation of the proposed PVE under real environmental conditions further validated its good dynamic and static robustness.</p>\",\"PeriodicalId\":475,\"journal\":{\"name\":\"Applied Solar Energy\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.2040,\"publicationDate\":\"2023-10-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Applied Solar Energy\",\"FirstCategoryId\":\"1\",\"ListUrlMain\":\"https://link.springer.com/article/10.3103/S0003701X23600285\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"Energy\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Applied Solar Energy","FirstCategoryId":"1","ListUrlMain":"https://link.springer.com/article/10.3103/S0003701X23600285","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"Energy","Score":null,"Total":0}
A New Piecewise Segmentation Based Solar Photovoltaic Emulator Using Artificial Neural Networks and a Nonlinear Backstepping Controller
The current state of affairs on the Photovoltaic emulator (PVE) is facing two main challenges: complexity in resolving the nonlinear equations of the photovoltaic (PV) and the problem of effective control of the PVE power conversion stage (PCS). In this paper, a new power electronics-based PVE is proposed to emulate the dynamic and static characteristics of the PV cell/module. The nonlinear equations of the PV cell/module are resolved using a new piecewise segmentation technique, involving the splitting of the current-voltage (I–V) curve into twelve linear segments associated with the letters a to m (a–m). Based on input environmental conditions, a trained artificial neural network (ANN) is constructed to assist the linearization process by predicting the current-voltage boundary coordinates of these segments. By the use of simple linear equations with the boundary coordinates, a reference voltage is then generated for the PVE. A nonlinear backstepping controller is designed to exploit the PVE reference voltage and stabilize the PCS. The stability of the controller is verified by Lyapunov laws. Optimal performance and control of the PCS were ensured by resorting to particle swarm optimization (PSO). The overall system has been investigated in the MATLAB environment with major tests including the response to fast-changing irradiance and temperature, the EN 50530 test, and the response to change in the load. The proposed PVE revealed a satisfactory dynamic performances in mimicking the PV characteristics. Furthermore, the accuracy of the PVE as a function of the mean absolute percentage error (MAPE) was found less than 0.5% even for the worst case of environmental conditions. Experimental validation of the proposed PVE under real environmental conditions further validated its good dynamic and static robustness.
期刊介绍:
Applied Solar Energy is an international peer reviewed journal covers various topics of research and development studies on solar energy conversion and use: photovoltaics, thermophotovoltaics, water heaters, passive solar heating systems, drying of agricultural production, water desalination, solar radiation condensers, operation of Big Solar Oven, combined use of solar energy and traditional energy sources, new semiconductors for solar cells and thermophotovoltaic system photocells, engines for autonomous solar stations.