基于GNN的全寿命期砼泵液压系统故障分析

液压系统是机、电、液耦合的复杂系统,实践表明工程机械有70%的故障是由液压系统引起的,液压系统的故障诊断已成为国内外学者研究的热点,其中智能化诊断已成为当前发展趋势,以神经网络应用最为广泛。然而研究发现,各类BP网络模型在样本点的选取上均没有考虑液压系统寿命周期不同对故障诊断所造成的影响,为解决这个问题,构建了全寿命期砼泵液压系统故障分析模型,在神经网络分析的基础上考虑设备寿命期对系统故障分析的影响,并结合遗传算法对BP神经网络进行优化。

Life-circle Fault Diagnosis Based on GNN for Concrete Pump's Hydraulic System

Hydraulic system is a complex system combined with machine, electricity and liquid coupling. Practice indicates that about 70% failure in construction machinery is caused by the hydraulic system. Now the research on fault diagnosis of hydraulic system perch in a fledging period, and intelligent fault diagnosis has become the current development trend while neural network is most widely used. Researches found that the selection of samples in some BP models failed to consider the life cycle of hydraulic system during fault diagnosis. Facing to the problem, I put forward a life-circle fault diagnosis model of concrete pump’s hydraulic system based on GNN. In the model, I consider the influence of equipment’s life period on the basis of the neural network analysis and the genetic algorithm to optimize the BP neural network.