基于信息融合技术的电梯控制系统故障诊断

摘要

This paper proposed an method which based on information fusion fault diagnosis for the elevator control system because of its difficult and low diagnosis rate, while optimizing the network learning rate and slopes of activation functions that improve on the hack-propagation (BP) neural network and then made the network had more capable of nonlinear classification and higher convergent efficiency. Meanwhile, regarded elevator control signals and running parameters as the characteristics of neural network classifier, and then applied the Dempster-Shafer (D-S) evidence theory fusing the result of multiple classifiers.Through making the basic reliability distribution of the evidence theory not completely depending on the expert subjective valuations, the impersonal realized valuation. The results of the experiment indicate that the method has high stability and can effectively improve the diagnosis accuracy of the elevator control system.%针对电梯控制系统故障诊断困难及诊断准确率不高的问题,提出一种基于信息融合技术的故障诊断方法.通过优化网络学习率与激活函数的倾斜度,对BP神经网络进行改进,使得网络非线性分类能力更强,收敛效率更高.利用电梯控制开关量信号和电梯运行模拟量参数作为神经网络分类器的特征,应用D-S证据理论合成法则将多个分类器的结果进行融合判决,使证据理论的基本可信度分配不再完全依赖专家进行主观化赋值,实现了赋值的客观化.实验结果表明,该方法具有较高的稳健性,能有效提高电梯控制系统故障诊断准确率.