Application And Research Of Intelligent Optimized Active Disturbance Rejection Control Algorithm In APF

APF?Active Power Filter,APF?is to improve the power quality,harmonic elimination control means,and having a conventional passive filter power system parameters compared to unaffected,while the advantages of fast dynamic response studies increasingly pay attention to personnel.For APF harmonic current conventional tracking control is poor,slow response of DC voltage,and large overshoot problems,taking ADRC intelligent optimization strategy.Active disturbance rejection control?ADRC?has no dependence on the precise mathematical model of the system.All uncertain disturbance factors?both internal and external?that act on the controlled object become"undefined disturbances".Estimate the characteristics of compensation.And it effectively solves the contradiction between"fastness"and"overshoot"in the traditional control algorithm,and can guarantee the control accuracy under the unknown strong nonlinearity,coupling and uncertain disturbance.However,there are insufficient researches on optimization problems and many parameters that are difficult to adjust.In this paper,the Artificial Neural Network?ANN?and Particle Swarm Optimization?PSO?are used as intelligent algorithms to optimize the active disturbance rejection control,and the control effect is verified based on APF.Firstly,the Active Power Filter?APF?application occasions,principle steps,control methods are introduced in this paper,and the mathematical model of three-level APF in each coordinate system is detailed,and the dq coordinate system is analyzed.Download the control system model,and then derive the ADRC-based control system.Then,the detailed calculation formulas of the Tracking Differentiator?TD?of the ADRC algorithm,the Nonlinear State Error Feedback?NLSEF?and the Extended State Observer?ESO?algorithm are derived,and the specific meanings of the corresponding parameters are explained.The current loop artificial neural network algorithm is optimized to compensate the extended state observer.A trained nonlinear mapping network with a certain structure and approximation ability is used to compensate for an unknown part of the model,which can reduce the burden on the observer,improve its observation accuracy,and improve control quality.In this paper,the control effect comparison between neural network training and the presence or absence of ANN compensation is completed,and the feasibility and superiority of ANN to optimize ADRC is verified on APF.Due to the numerous ADRC parameters,the optimized loop particle swarm optimization algorithm used in the voltage loop to optimize the NLSEF parameter?_{0 1},?₀₂ in ADRC was studied indirectly to improve the control accuracy.This paper has completed the description of the PSO algorithm's principle steps,algorithm optimization comparison,performance test comparison,comparison of APF direct current side control effects under different index function constraints,and analyzed the APF control effect and dynamic performance test by optimizing parameters,eliminating many parameters The process of manually adjusting the parameters of the ADRC control algorithm,and handing over the complex adjustment to a computer simulation system,reflects the feasibility and superiority of the intelligent algorithm to optimize the ADRC.The paper has 56 pictures,13 tables,and 83 references.