Research And Application Of Widely Linear Total Least Squares Adaptive Filtering Algorithm

Considering both input errors and response errors,total least squares is the best technique.Total least squares minimizes disturbances in the input and output data.In this paper,a widely linear total least squares adaptive filtering algorithm(ACTLMS)is proposed based on TLS.The specific research contents are as follows.The widely linear model based unbalanced power system frequency estimation technique is studied.First of all,the widely linear model,the complex number statistical model,and the complex voltage signal of the three-phase power system are briefly introduced.Based on the widely linear adaptive frequency estimation technique of balanced power system,the derivation process of adaptive frequency estimation based on widely linear least squares technique for unbalanced power system is given.A widely linear total least squares adaptive filtering algorithm is studied.First,a full study of the total least squares(TLS)problem was conducted.Then we proposed a widely linear total least squares adaptive filtering algorithm(ACTLMS),and gave a detailed formula derivation process of ACTLMS.Although the input and output have interference,the performance of the ACTLMS algorithm is obviously superior to the ACLMS.Since the assumption that the input and output are noisy is realistic,this ACTLMS algorithm has widely applicability.An analysis and study of the performance of a widely linear total least squares adaptive filtering algorithm from the three perspectives of convergence,local stability and filter coefficients.We find that the ACTLMS algorithm will be globally asymptotically convergent.Finally,using the matlab programming environment,ACTLMS algorithm was simulated for different environments.The performance of ACTLMS algorithm is obviously better than that of ACLMS.Especially when the SNR is low,the ACTLMS algorithm has better noise immunity than the ACLMS algorithm.The application of widely linear least squares adaptive filtering algorithm in unbalanced power system is studied.Multi-noise voltage measurement for unbalanced three-phase power systems introduces a widely linear prediction model.In order to obtain the frequency estimation in an adaptive way,based on the recursive minimization of the Rayleigh quotient associated with the model,a gradient-decreasing ACTLMS is proposed.The sliding window frame makes the proposed ACTLMS frequency estimation method as its scalar version and its noiseless version of the ACLMS method less sensitive to measurement noise and higher-order harmonic distortion,while also providing more degrees of freedom to control the tracking speed and estimation Accuracy.Finally,using the simulation environment of matlab,simulation and support analysis of synthetic and real-world noise unbalanced power systems are performed.The application of total least squares enhanced intelligent DFT technology in three-phase power system is studied.For a balanced three-phase power system,we introduce a complex total least squares(CTLS)framework that makes the original SDFT algorithm robust to noise and harmonics.The performance advantages of the proposed CTLS framework are verified by simulating various power system conditions and measuring the real world.For unbalanced three-phase power systems,we have expanded the Intelligent SDFT technique originally designed for non-nominal frequency estimation of real-valued single-phase voltages to handle complex-value switching voltages in unbalanced power systems.We introduce an total least squares framework,and the SDFT algorithm is more robust when unbalanced power systems are polluted by noise and unwanted harmonic pollution.Finally,the performance advantages of the proposed TLS-SDFT technology are verified by matlab simulation and actual measurement.