基于动态权重优化的风电机组齿轮箱轴承温度预测模型  

Temperature Prediction Model for Wind Turbine Gearbox Bearings Based on Dynamic Weight Optimization

作　　者：吴九牛 翟广宇 李德仓 高德成 蒋维栋 WU Jiuniu;ZHAI Guangyu;LI Decang;GAO Decheng;JIANG Weidong(Gansu Institute of Metrology,Lanzhou 730050,China;School of Economics and Management,Lanzhou University of Technology,Lanzhou 730050,China;Mechatronics T&R Institute,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]甘肃省计量研究院,兰州730050 [2]兰州理工大学经济管理学院,兰州730050 [3]兰州交通大学机电技术研究所,兰州730070

出　　处：《轴承》2024年第9期100-107,共8页Bearing

基　　金：国家自然科学基金资助项目(71861026)。

摘　　要：为准确预测风电机组齿轮箱轴承的温度状态,结合灰色预测GM(1,N)模型、BP神经网络模型和支持向量回归模型,提出了一种动态权重优化的组合预测模型。通过对3种预测模型的理论分析选择了各自合理的模型结构,并用粒子群算法优化模型参数;预处理齿轮箱轴承温度的原始数据后用指数平滑法确定各单一模型的动态权重系数,建立齿轮箱轴承温度的组合模型;通过滑动窗口法统计分析齿轮箱轴承预测温度的残差,判断齿轮箱轴承的运行状态。研究结果表明:组合模型的各项评价指标均优于单一预测模型,决定系数为0.9772,预测效果更加稳定准确,能够及时监测齿轮箱轴承温度的变化情况。In order to accurately predict the temperature state of wind turbine gearbox bearings,a combined prediction model based on dynamic weight optimization is proposed by combining grey prediction GM(1,N)model,BP neural network model and support vector regression model.Through theoretical analysis of three prediction models,the reasonable structure is selected for each model,and the model parameters are optimized by particle swarm algorithm.After preprocessing the original temperature data of the bearings,the dynamic weight coefficient of each single model is determined by exponential smoothing method,and the combined model for temperature of the bearings is established.The residuals of predicted temperature of the bearings are analyzed statistically by sliding window method,and the operating state of the bearings is judged.The research results demonstrate that the evaluation indexes of combined model are all better than those of single prediction model,with a determination coefficient of 0.9772,the prediction effect is more stable and accurate,and the temperature change of the bearings can be monitored in time.

关 键 词：滚动轴承 风力发电机组 温度 预测 灰色系统 神经网络 支持向量回归预测法 

分 类 号：TH133.33[机械工程—机械制造及自动化] TH17[动力工程及工程热物理—流体机械及工程] TK83