Research On Photovoltaic Cell Model Parameter Identification And Maximum Power Point Tracking Method Based On Improved Knowledge Acquisition And Sharing Algorith

Photovoltaic power generation is essential to the "dual carbon" strategy.As a prerequisite for many studies,the accurate modeling of photovoltaic power generation systems is essential.At the same time,studying maximum power point tracking methods is crucial in reducing the cost of power generation.Therefore,this paper focuses on PV cell parameter extraction and maximum power point tracking techniques.Firstly,a multi-strategy gaining-sharing knowledge-based algorithm(MSGSK)based PV cell model’s parameter identification method is proposed to address the weak local exploitation capability exposed by the gaining-sharing knowledge-based algorithm(GSK)in extracting PV cell parameters.An adaptive parameter mechanism and a chaotic elite learning strategy are designed to improve the convergence speed;a backtracking difference variation strategy is designed to balance the overly intense local exploitation after introducing the two strategies.To verify the effectiveness of MSGSK,experiments on PV model parameter identification are introduced,and the results show that MSGSK improves the convergence and accuracy of GSK.Secondly,a PV cell model’s parameter identification method based on reinforcement learning gaining-sharing knowledge-based algorithm(RLGSK)is designed to improve the performance of identification results.Considering that the MSGSK parameter adaptive mechanism is too rigid in its adjustment method,the individual triage mechanism is optimized through the reinforcement learning mechanism to adjust the GSK algorithm parameters flexibly.The optimal allocation of computational resources is achieved through a performance-based population size reduction mechanism.Experiments are introduced to identify the parameters of the PV model to verify the effectiveness of RLGSK.The results show that RLGSK improves robustness based on improved accuracy and convergence.It is demonstrated that RLGSK achieves optimal allocation through component analysis experiments.On the other hand,a maximum power point tracking(MPPT)method based on the differential evolutionary gaining-sharing knowledge hybrid algorithm(HDEGSK)is designed to address the real-time and accuracy challenges faced in practical applications of PV power generation systems.The initial values of the traditional MPPT algorithm are optimized by HDEGSK,considering that the traditional MPPT algorithm converges quickly but with sensitive initial values.HDEGSK algorithm effectively couples the GSK with the differential evolution update method through an improved individual shunt mechanism.To verify the effectiveness of HDEGSK,PV MPPT experiments are introduced,and the results show that the traditional algorithm optimized by HDEGSK significantly improves the real-time performance and accuracy.Finally,a "one-step" MPPT technique further reduces power losses.A performance-oriented differential evolutionary gaining-sharing knowledge hybrid algorithm(PG-HDEGSK)is proposed,which introduces a performance-oriented adaptive parameter adjustment mechanism and a population size reduction mechanism based on HDEGSK to optimize the allocation of computational resources.It was demonstrated experimentally through parameter identification in terms of performance: PG-HDEGSK > RLGSK > HDEGSK > MSGSK > GSK.A power prediction model with multi-peak characteristics was established by analyzing parameter identification results under different light intensities.In order to verify the validity of the prediction model,MPPT experiments were carried out by the constant voltage and constant current methods,and the results showed that the "one-step" MPPT technique has a very high tracking accuracy,and the tracking error is further reduced.