Research On Harmonic Prediction Method For Photovoltaic Power Plants Based On Improved MKELM Model

As a key node in the regional power grid,the detection and management of harmonic content in PV power plants has always been the focus of research in the field of PV power plant operation.A large number of impact loads and distributed new energy access to the grid bring more complicated power quality problems,most of the existing reactive power compensation devices are statically installed,and the compensation devices cannot achieve the estimated harmonic ceiling to complete a single installation and commissioning,which brings great challenges to the prediction of harmonic pollution traceability,spatial and temporal trend flow analysis and safety warning responsibility division,real-time accurate prediction of each harmonic component in PV power plants,which is important for harmonic pollution It is important to predict the harmonic pollution and the dynamic operation and tracking of harmonic suppression equipment.The current common harmonic prediction models still have poor prediction accuracy,delay lag and other problems.It cannot provide a reasonable basis for the future input of harmonic suppression equipment quantity or capacity,therefore,this thesis conducts an in-depth study on the existing problems of harmonic content prediction in PV power plants,and the main research contents are as follows:Firstly,for the problems of modal mixing,frequency confusion and poor detection a ccuracy of harmonic signals in the detection process,a variational modal decomposition a lgorithm optimized by multi-scale permutation entropy(MPE-VMD)is proposed to utilizethe anti-interference advantage of multi-scale alignment entropy in processing non-stationa ry signals,and the intrinsic mode functions functions in the variational modal decompositi on algorithm are screened and reconstructed according to different entropy values to effec tively separate multi-frequency harmonics into the intrinsic mode functions functions of co rresponding frequencies,so as to realize the extraction and fusion of different characteristi c components of harmonic numbers,reduce the complexity of operation and improve the accuracy of harmonic detection in PV power plants.Secondly,on the basis of multi-scale permutation entropy optimization,considering the high frequency complexity of the harmonic signals of PV power plants,the difficulty of the general limit learning machine in selecting the hyper-parameters of harmonic signals,and the inability of the classifier to cope with the influence brought by parameter perturbations,the Harris Hawk algorithm is introduced to optimally find the kernel function parameters and regularization coefficients in the kernel limit learning machine,and determine the optimal range of the kernel function parameters and regularization coefficients The training model is established and imported into the PV plant harmonic data for sample training to ensure the prediction accuracy of the model.Finally,the variational mode decomposition algorithm optimized by multi-scale permutation entropy is introduced into the optimal training model to build a PV power plant harmonic content prediction model.The error evaluation index is introduced,and through fast Fourier transform spectrum analysis and error comparison,it is demonstrated that the harmonic content prediction model established in this thesis can offset the effects brought about by the disturbance of each harmonic component in the PV power plant and achieve the function of accurate prediction of different harmonic sources.