基于多特征参量的回转支承智能健康状态评估

Intelligent health condition assessment on slewing bearing based on multiple characteristic parameters

为了提高回转支承运行可靠性,及时发现其潜在的失效,实施良好的设备维护与管理,有必要对其进行健康状态评估。选取表征回转支承健康状态的温度和扭矩作为特征参量,建立了一种采用遗传算法优化动态递归Elman神经网络的回转支承多参量健康状态评估模型,并利用3 MW变桨回转支承疲劳寿命实验数据对该模型进行了网络训练和测试。结果表明,该模型评估结果与实验值相符,可准确地对回转支承进行健康状态评估。

It was necessary to estimate their health conditions of slewing bearing,improve its operational reliability,and find out the potential fault and implement good equipment maintenance and management.The temperature and the torque were selected as characteristic parameters,they reflected slewing bearing status.Then,a multi-parameter health state evaluation model was established with Elman dynamic recursive neural network based on genetic algorithm（GA）.Finally,the model was trained and tested by test data from 3 MW variable pitch slewing bearing fatigue life experiments.Results showed that the evaluation results from the model were in agreement with experiment data,and the model could preciously assess the health state of slewing bearing.