Fault Diagnosis Of Dual-PWM Wind Power Grid-connected Converters Based On Estimated Instantaneous Amplitude Of Phase Current

Wind power generation technology is one of the most important developments of clear energies currently.In wind generation system,the dual-PWM grid-connected converter plays a significant role for energy conversion.Due to the influence of harsh operation environment,such as high temperature,high humidity and high salt-fog,the failure rate of current sensor and power switch,namely,insulated gate bipolar transistor(IGBT),are increasing continuously.Specially,because of the influence of noise and harmonic,it is hard to realize the fault feature extraction of the current sensor incipient fault.As the wind speed always changes randomly when the wind power generation system operate,the open-circuit fault detection results of the power switch IGBT are prone to false alarm.In addition,affected by the complexity of multiple fault mode,it is difficult to precisely realize the detection and location of faults,especially when certain IGBTs suffer from open-circuit at the same time.As a result,it is meaningful to research the diagnosis of the current sensor fault and the IGBT open-circuit fault,which makes important contribution for improving the operation efficiency and ensuring the operation safety of the wind power generation systems.In this paper,based on the analysis of phase current instantaneous amplitude,the diagnosis of the current sensor fault and IGBT open-circuit fault of wind energy conversion system are realized successfully.In order to accurately estimate the instantaneous amplitude of phase current that usually fluctuates with severe operation environment,the Hilbert Transform(HT)algorithm and corresponding improved algorithms are presented.Utilizing the estimated phase current instantaneous amplitude,the problems of current sensor incipient fault,IGBT open-circuit robust detection and multiple IGBT open-circuit diagnosis are studied.The main research achievements are as follows:For current sensor fault and IGBT open-circuit fault,fault characteristics based on the instantaneous amplitudes of phase currents are analyzed respectively.Firstly,based on MATLAB / Simulink software,the simulation of wind energy conversion system,where a permanent magnet direct generator is connected to the grid-connected converter directly,is built,and the output three-phase current characteristics of the system are analyzed.Considering the sinusoidal wave characteristics of the measured three-phase current of the wind power grid-connected converter,the Hilbert Transform(HT)algorithm is applied to estimate the phase current instantaneous amplitudes.Finally,according to simulation analysis results,it is shown that the fault feature extraction approach that is proposed utilizing the instantaneous amplitude of phase current is feasible for the grid-connected converter fault diagnosis.To deal with the current sensor incipient fault of wind power grid-connected converters,the fault diagnosis approach consisting of moving window principal component analysis and Kullback-Leibler divergence is proposed.Combined with the constructed data matrix that is composed of three-dimensional instantaneous amplitudes,the principal component analysis algorithm and the KL divergence are utilized for extracting small fault feature for current sensor early small faults.Finally,through analyzing the circuit mechanism of grid-connected converter,the theoretical estimation model of current sensor fault amplitude,that is related with variable KL divergence,is established to realize the online estimation of small gain fault amplitude.The simulation results show that the proposed method can effectively detect,locate and estimate the fault amplitude of current sensor under the condition with different severity.Considering the influence of variable wind speed,a novel open-circuit fault diagnosis approach that is based on the estimated instantaneous amplitude is presented to reduce false positive rate.In this paper,the phase current instantaneous amplitude is utilized for the first time to deal with open-circuit fault.In addition,in order to continuously and smoothly estimate the instantaneous amplitude of sinusoidal signal whose amplitude and frequency changes with time,the sliding window algorithm is introduced to improve the traditional Hilbert transform,where the improved HT algorithm can eliminate the influence of boundary effects.Simultaneously,to reduce various disturbance influence including measurement noise,circuit harmonics and random wind speed changes and so on,the current signal is firstly preprocessed through park transform and low-pass filtering.At last,according to the preset fault detection threshold,the open-circuit faults of both the grid side and generator side are detected simultaneously and robustly.Combined with mean phase current method,location of upper and lower bridge arms of open-circuit fault can be carried out.The simulation results show that compared to the typical current average approach,the proposed instantaneous amplitude method can detect open-circuit faults with lower false positive and false negative rates.The Park's vector instantaneous amplitude based open-circuit fault diagnosis approach is proposed to deal with the multiple open-circuit fault of IGBTs.For further improving the estimation accuracy and real-time characteristics,a novel improved Hilbert transform algorithm is presented utilizing the wave extension approach,where the wave extension is realized through the phase prediction of Park's vectors.In order to accurately detect and locate the multiple open-circuit faults that contains various fault modes,a logic strategy that contains multiple different fault features and certain preset thresholds is constructed,where the fault features are extracted through changing current phase series and computing instantaneous amplitude mean values.According to the simulation results,it is shown that the proposed multiple fault diagnosis approach can simultaneously detect and locate the failure IGBTs when multiple IGBTs suffer from open-circuit fault at the same time.Certainly,for the multiple open-circuit fault diagnosis of the grid side and the generator side,the proposed approach are all feasible and robust.