基于PSO—BP优化算法的变频液压系统故障检测

Frequency Conversion of Hydraulic System Failure Detection Based on PSO--BP

变频液压系统控制信号组成复杂，故障检测采集的数据量庞大，常常出现故障检测数据串扰噪声加大而不清晰；传统的变频液压系统的故障检测主要依靠主观经验与参数测量的方法来进行故障诊断，这些方法对数据串扰造成的影响没有去除，经常会出现漏诊与虚诊现象；为此提出一种PSO-BP优化算法的变频液压系统故障检测方法，将变频液压控制系统的故障参数输入到经过优化后的神经网络进行故障诊断模型训练，克服故障数据串扰模糊不清晰的缺点；实验证明，经过优化后的神经网络模型可以将变频液压系统的故障检测准确率大幅度地提高，对实际的液压系统生产有指导意义。

The hydraulic system of variable frequency fault detection problem accurately. The hydraulic system of variable frequency control signal complex, fault detection of the data collected huge amount, often appear fault detection data crosstalk noise and not clear increase. The traditional frequency hydraulic system fault detection rely mainly on the subjective experience and parameters measurement of the methods for fault diagnosis, these methods of data crosstalk, the impact of not remove, often appear misdiagnosis and virtual diagnosis phe- nomenon. Therefore put forward a kind of PSO--BP frequency hydraulic system fault detection method, variable frequency hydraulic control system fault input parameters through the optimized to the neural network fault diagnosis model training, overcome fault data crosstalk fuzzy not clear faults. Experiments show that after the optimal neural network model variable frequency hydraulic system can be the fault detection accuracy greatly improved, the actual hydraulic system production is of guiding significance.