摘要
大型风电机组传动系统运行工况复杂,运维人员无法实时了解其运行状态。针对这一情况,提出一种基于深度神经网络模型(DNN)的状态监测方法。首先,BP算法训练DNN模型时容易陷入局部最小值和过拟合,因此,将麻雀搜索算法(SSA)与BP算法结合,提出一种SSA优化BP算法训练DNN模型的方法。然后,采用风电机组SCADA系统数据建立DNN模型来估计风电机组传动系统的关键运行参数,利用估计参数的残差信息构建T2统计量应用于状态监测,并给出T2统计量报警限的确定方法。最后,将提出的DNN训练方法与其他方法进行对比并将该方法应用于某风电机组传动系统状态监测。结果表明,SSA优化BP算法能够有效避免局部最小值和过拟合,该状态监测方法能够提前预警风电机组传动系统的异常状态。
Operation condition of large wind turbine drive system is complex,and the operation and maintenance personnel cannot understand its operation state in real time.In order to solve this problem,a condition monitoring method based on deep neural network model(DNN)is proposed in this paper.First of all,BP algorithm is prone to fall into local minimum value and over-fitting when training DNN model.To solve this problem,combining sparrow search algorithm(SSA)with BP algorithm,and a method of training DNN model by SSA optimized BP algorithm was proposed.Then,a DNN model was established based on SCADA system data of wind turbine to estimate the key operating parameters of wind turbine drive system.The residual error of key parameter was used to construct T2 statistics for condition monitoring,and the method to determine the alarm limit of statistics was given.Finally,the DNN training method proposed in this paper was compared with other algorithms,and the condition monitoring method was applied to the condition monitoring of the drive system of a wind turbine.The results show that the BP algorithm optimized by SSA can effectively recede local minimum and over-fitting.The condition monitoring method can give an alarm of the abnormal condition of wind turbine drive system in advance.
作者
王印松
石建涛
WANG Yinsong;SHI Jiantao(Department of Automation,North China Electric Power University,Baoding 071003,China)
出处
《电力科学与工程》
2021年第9期26-34,共9页
Electric Power Science and Engineering
关键词
风电机组
深度神经网络
麻雀搜索算法
状态监测
wind turbine
deep neural network
sparrow search algorithm
condition monitoring
