基于BAS-BP模型的谐波检测

Research on Harmonic Detection Method Based on BAS-BP Neural NetWork

有多种方法可以检测电力系统中的谐波,当前最常用的是基于瞬时无功功率理论的谐波检测算法,但是瞬时无功功率理论涉及到多次坐标变换,计算量大且谐波检测存在延时.近年来,随着神经网络研究的不断深入,开始采用基于BP神经网络的模型进行谐波检测,可以减少系统复杂度,检测速度快,适应能力强.但这种方式的缺点是随机初始化权阈值,使得BP神经网络存在在训练过程中陷入局部最优的问题,最终导致模型精确度不高.为此,采用天牛须搜索(Beetle Antennae Search)优化BP神经网络,并将训练好的BAS-BP模型运用于谐波检测.通过Simulink仿真对比BAS-BP模型与BP模型和ip-iq模型,论述了BAS-BP模型在谐波检测方面的可行性与优越性.

There are a variety of methods to detect harmonics in power systems.At present,the most commonly used method is harmonic detection algorithm based on instantaneous reactive power theory.However,instantaneous reactive power theory involves multiple coordinate transformations,which requires a large amount of calculation and has a delay in harmonic detection.In recent years,with the deepening of neural network research,the harmonic detection based on BP neural network model can greatly reduce the complexity of the system,with fast detection speed and strong adaptability.The disadvantage is that the weight threshold is randomly initialized,which makes the BP neural network fall into the problem of local optimum in the training process,and finally leads to the low accuracy of the model.For this reason,this paper used beetle antennae search to optimize BP neural network,and applied the trained BAS-BP model to harmonic detection.Through Simulink simulation,the BAS-BP model is compared with BP model and ip-iq model,and the feasibility and superiority of BAS-BP model in harmonic detection are discussed.