IWOA和BP神经网络的压缩机故障检测算法

针对压缩机故障难以检测与分类的问题,传统的方法是采用BP神经网络的检测方法,利用从压缩机语音信号中提取的识别特征来预测压缩机故障类型,该方法具有良好的可行性,但是BP神经网络容易陷入局部最优值的情况,从而导致了预测精度较低和稳定性较差的问题。因此,本文提出了改进的鲸鱼优化算法(IWOA)和BP神经网络的模型,首先通过优化WOA中的搜索猎物被执行的概率和包围猎物向最优个体聚集的过程,从而提高了IWOA的全局搜索能力和收敛速度,其次将IWOA对BP神经网络的权值和阀值进行深度寻优,从而提高了BP神经网络的预测精度和稳定性,最后将该模型运用到压缩机故障检测实验中。实验结果表明,与其他模型相比,证明了IWOA和BP神经网络提高了预测精度,且具有良好的稳定性。

Compressor fault detection algorithm based on IWOA and BP neural network

Aiming at the problem that compressor failure is difficult to detect and classify, the traditional method is the BP neural network method, which uses the recognition features extracted from the compressor voice signal to predict the compressor failure type.This method has good feasibility, but it is easy to fall into the situation of local optimal value, which leads to the problem of low prediction accuracy and poor stability. Therefore, this paper proposes an improved whale optimization algorithm (IWOA) and BP neural network model.First, by optimizing the probability of the searched prey being executed and the process of surrounding the prey to the optimal individual, the global search ability and convergence speed of IWOA is improved; Second, IWOA deeply optimizes the weights and thresholds of the BP neural network to improve the prediction accuracy and stability of the BP neural network;Finally, the model is applied to the compressor fault detection experiment, and the experimental results show that compared with other models, the IWOA and BP neural network improve the prediction accuracy and have a good stability.