摘要
目的识别除湿机的性能状态和预测吸附剂的剩余寿命。方法针对除湿机故障过程缓变的特点,提出一种基于数据驱动的遗传神经网络模型。首先,为解决设备失效程度划分模糊的问题,由5个热力参数组成反映吸附剂劣化程度的特征向量,关联分析得到除湿机的5类故障模式。其次,利用遗传神经网络建立状态参数和故障模式的映射关系。最后,对表征设备吸附能力的主参数进行外推预测。结果训练好的诊断网络可准确地识别出设备的劣化程度及其演变过程,预测网络的预测精度非常高。结论该方法可有效地实现对除湿机的故障诊断与预测。。
Objective To identify performance state of dehumidifier and predict residual life of adsorbent. Methods A data driving-based genetic neural network model was proposed in view of the slow variable failure process of dehumidifier. Firstly, to solve the issue fuzzy division of equipment failure degree, 5 failure patterns of the dehumidifier were obtained with correlation analysis according to characteristic vectors formed by 5 thermodynamic parameters which reflect adsorbent degradation. Secondly, the mapping relationship between state parameters and failure patterns was established with genetic neural network. Finally, principal parameters reflecting adsorption capacity of equipment were predicted by extrapolation. Results The diagnosis network could determine deterioration degree and evolution process of equipment accurately. It possessed very high accuracy in network prediction. Conclusion This method can effectively finish fault diagnosis and prediction of dehumidifier.
出处
《装备环境工程》
CAS
2017年第1期78-83,共6页
Equipment Environmental Engineering
关键词
遗传算法
神经网络
除湿机
露点温度
genetic algorithm
neural network
dehumidifier
dew point temperature