基于机器学习的水液压高速开关阀退化趋势预测

高速开关阀以其结构简单、响应速度快、抗污染能力强、稳定性好等优点得到了广泛的应用。水液压高速开关阀的工作介质黏性低,更容易因性能退化发生故障。提出了一种基于机器学习的水液压高速开关阀性能退化状态识别及退化趋势预测方法。搭建了高速开关阀性能测试试验台,将电流信号的变化作为高速开关阀的性能退化指标。根据高速开关阀性能退化程度,将其退化状态定义为正常期、退化期和严重退化期3个阶段。采用BP神经网络(BPNN)方法对高速开关阀的退化状态进行了识别,并采用粒子群优化长短期记忆模型(PSO-LSTM)方法对高速开关阀的退化趋势进行了预测。使用高速开关阀的性能退化试验数据对提出模型的有效性进行了检验,结果表明该方法具有较高的预测精度。

Prediction of Degradation Trend of Water Hydraulic High Speed On/Off Valve Based on Machine Learning

High speed on/off valve has been widely used because of its simple structure, fast response, strong anti-pollution ability and good stability. Because of the low viscosity of the working medium, the water hydraulic high speed on/off valve is more likely to fail due to performance degradation. This study presents a method of high speed on/off valve performance degradation state identification and degradation trend prediction based on machine learning. An high speed on/off valve performance test bench is built, and the change of current signal is taken as the performance degradation index. According to the degradation degree of high speed on/off valve performance, the degradation state is defined as normal period, degradation period and severe degradation period. BPNN is used to identify the degradation state of high speed on/off valve, and PSO-LSTM method is used to predict the degradation trend of high speed on/off valve. The effectiveness of the proposed model is tested by using performance degradation test data of high speed on/off valve. The results show that the method has high prediction accuracy.