System Identification Methods Based On Exponential Criterion Functions

System modeling and model identification are the basis of control system analysis and design.Establishing a mathematical model of the system through measurement data is an effective method of in-depth study of the system.In recent years,system identification technology has been widely used in various fields of power systems,especially in the mechanism research and simulation of power system,automatic control,adaptive prediction and prediction,fault diagnosis of power electronic converters and modeling of photovoltaic grid-connected inverters,it has been very successfully applied.The criterion function is an important basis for obtaining the identification algorithms.By updating the value of the criterion function,the accuracy of the corresponding identification algorithm can be measured in real time.The criterion function is generally composed of the sum of the squares of the model and the actual process error,this paper combines this criterion function with the exponential function to obtain the exponential criterion function,and studies the recursive and iterative identification algorithms based on this criterion function for different systems.The main research work of this paper is as follows.(1)For the controlled autoregressive system,the exponential criterion function based recursive gradient algorithm is derived using the defined exponential criterion function and the gradient search.In order to further improve the accuracy of parameter estimation,this paper uses the Newton search principle to propose a Newton recursive algorithm based on the exponential criterion function.(2)For the controlled autoregressive system,the exponential criterion function based gradient iterative algorithm and the exponential criterion function based Newton iterative algorithm are studied respectively by using the defined exponential criterion function.This paper combines the multi-innovation identification theory and proposes an exponential criterion function based multi-innovation gradient iterative algorithm for the controlled autoregressive system.This iterative identification algorithm realizes the dynamic update of the data in the data window by using the moving data window,so that the algorithm can estimate the system parameters in real time.(3)For the finite impulse response moving average system,that is,the finite impulse response system in which the noise term is a moving average process,the exponential criterion function based recursive extended gradient algorithm and the exponential criterion function based extended Newton recursive algorithm are derived by using the hierarchical identification principle.(4)For the finite impulse response moving average system,the exponential criterion function based gradient iterative algorithm and the exponential criterion function based Newton iterative algorithm are studied respectively by using the hierarchical identification principle.The exponential criterion function based multi-innovation extended gradient based iterative algorithm is obtained by using the multi-innovation identification theory.For all the algorithms proposed above,this paper gives detailed derivation steps and flow charts.In order to illustrate the effectiveness of the proposed algorithms,this paper gives the numerical simulation results.At the same time,the proposed algorithms are compared with other algorithms,and the performance of different algorithms under different simulation conditions is studied by controlling the experimental conditions in the simulation.In the final summary and outlook section,the main characteristics of the proposed algorithm and the directions that need to be further explored in the future are summarized.