传统的线性神经网络,其激励函数是固定函数,所以不能对间谐波的频率进行调整,而基波频率微小的偏差将导致谐波检测出现较大的误差。笔者在采用BP神经网络的高精度基波检测的基础之上,采用激励函数参数可调的线性神经网络来分析间谐波。理论分析和仿真结果都表明,无论有没有噪声,文中提出的方法都具有较高的检测精度,自适应能力较强,其中检测与偶数次谐波比较接近的间谐波的优势更明显。 The traditional linear neural network whose excitation function is a fixed function can not adjust the frequency of the interharmonics. However, the slight deviation of the fundamental frequency leads to a large error of the harmonic detection. Based on the high-precision fundamental detection of BP neural network, the author uses the linear neural network with adjustable excitation function parameters to analyze inter-harmonics. Both theoretical analysis and simulation results show that the proposed method has higher detection accuracy and stronger adaptive ability with or without noise, and the detection of the interharmonics with the even harmonics is more obvious.