The Research Of ANN-LDWPSO Harmonic Detection Algorithm In Photovoltaic Systems

Harmonic detection technology is such an important guarantee that the PV systemhas a safe and stable operation. This thesis study the photovoltaic system harmonicdetection algorithm, it focuses on the following tasks:Firstly, it introduces the of impact and the importance of electric power quality toharmonics, and a detailed discussion and comparison of all harmonic detectionalgorithm applicability in photovoltaic systems. Ultimately,we selected ArtificialNeural Network harmonic detection algorithm. ANN harmonic detection algorithmswork for each harmonic detection.Tthe complexity and discontinuity of photovoltaicsystems leads to the complexity of harmonic frequency circuit. The Hopfield NeuralNetwork algorithm and BP will certainly contain more intermediate layers, causing alarge amount of computation and slow convergence or even convergence speed. Theaccuracy of the algorithm is not guaranteed. Through the analysis of the topology,performance, and the calculation of different ANN algorithms,comparing againstcharacteristics of PV systems, we chose the Adaline linear Adaptive ANN,which hasthe easiest topology,the best accuracy and the fastest response.Secondly, in order tomake up a single ANN harmonic detection’s shortcomings, which is a slowconvergence or unideal convergence precision or even no convergence and an easyway to fall into local minimum problem, we carry out analysis of Particle SwarmOptimization. The integration of the two algorithms is designed to improve ANNalgorithm, using linear decreasing weight PSO (LDWPSO) to adjust the weight valueof each layer of ANN. Based on linear decreasing weight particle swarm of artificialneural network algorithm (ANN-LDWPSO), we can get the weight vector when theerror converges to a minimum,and using it to calculate the amplitude and phase angleof harmonic. Successfully detected all the harmonics in photovoltaic systems,whitch isproviding a reliable guarantee for the future.Thirdly, the single Adaline and ANN-LDWPSO harmonic detection algorithm modules are programmed in the simulation.By comparing the performance of both, we get the result about the ANN-LDWPSO harmonic detection algorithm both in the response rate and the convergenceprecision is better than the single Adaline neural network algorithm.The conclusionsverify the correctness and feasibility of the ANN-LDWPSO algorithm.Meanwhile, wehave a in-depth research and analysis to the working principle of the structure of themain components of photovoltaic systems and various parts of the equivalent circuit.Based on that,we construct an optimal mathematical model and a simulation diagramof photovoltaic systems using Matlab to build. Then taking advantage of the PWMcontroller to make up the point of PCC voltage.The simulation results show that thequality of PCC voltage is improved, and the ANN-LDWPSO harmonic detectionalgorithm for the photovoltaic control system harmonic detection research opened up anew field.Finally, on the theoretical basis,we design the testing of the electrical part and themonitoring page photovoltaic solar detect bench,whicht is fused into the Wind-and-Sun controller.The monitoring results once again proved the versatility and practicalityof harmonic detection algorithm.