Harmonic Source Recognition Technology Based On Blind Source Separation

With the continuous development of electricity market reform,the significant need for the power companies and customers is to achieve the security and quality of supply.Harmonics are a prominent problem due to the connection of a large number of power electronic devices.Therefore,to appropriately estimate harmonic levels and quantitatively determine the responsibility at each node has become an urgent issue.First of all,a multi-index evaluation system is formed for the evaluation of harmonic pollution.Next,the selection principle of the evaluation system is determined.The combination weighting method is applied to evaluate harmonic pollution level,where the weight of each index is determined by the optimization model.Last,the fuzzy mathematics method is used to comprehensively evaluate the harmonic pollution level.A case of a traction substation with 24 hours data of the high-voltage side is statistically studied,and the harmonic pollution level is divided into five levels for the evaluation.The results show the reliability of the evaluation method.The Norton equivalent model for harmonic responsibility determination is also studied,which combines the independent component analysis.The quantitative determination of the harmonic responsibility at the point of common coupling of customer side and system side is realized through this model.The harmonic voltage and harmonic current at the point of common coupling are regarded as variables for measurement,and then separated by using Independent Component Analysis(ICA)algorithm.With the introduction of the correction factor,the parameters of Norton equivalent circuit model and the estimated values of the harmonic current waveforms on both sides are obtained.Finally,the harmonic responsibility of customer side and system side can be calculated by performing the superposition theorem.Besides,compared with the traditional method,this proposed method has higher precision,and the validity is verified.For the multi-harmonic power system with unknown topology,the minimum condition entropy and ICA algorithm are combined to form a multi-harmonic source location and responsibility determination method.The ICA algorithm is used to reconstruct the injection current of each harmonic source and the estimation of the mixing matrix.By calculating the conditional entropy between the reconstructed harmonic current and the harmonic voltage of each measuring point,the harmonic source is located based on the principle of minimum conditional entropy,and the harmonic pollution responsibility of each node is calculated by the mixing matrix.The verification analysis is performed in IEEE-14 system,and the results show that the proposed method can achieve accurate locating of harmonic sources and determination of harmonic responsibility.