基于贝叶斯网络的风力发电机故障诊断方法

The Method for diagnosis of Wind Turbine Based on Bayesian network

摘要:

贝叶斯网络是概率统计学的重要分支,具有强大的不确定性问题处理能力,适用于复杂系统的故障诊断。风力发电机系统维护成本较高,为减少维修成本,需要进行准确的故障定位。文章对基于贝叶斯网络的故障诊断方法进行了研究,介绍了贝叶斯网络故障诊断模型的建立过程,并着重介绍了诊断算法推导和计算过程。利用历史故障统计数据建立了风力发电机系统贝叶斯网络MATLAB模型,主要包括网络结构有向无环图和条件概率分布参数等内容。最后,模拟了两种故障,分别采用贝叶斯网络方法和相关性矩阵方法进行故障诊断,通过对两种方法诊断结果的比较,前者具有更好的故障分辨率,可有力支持复杂系统的维护保障、降低维修成本。

Abstract:

Bayesian network is an improtant variety of probability statistics,Because of its powerful capabilities of dealing with the uncertainty problem,it is suitable for diagnosis of complex system.The maintenance cost of wind turbine is very high,people needs to locate the faults accurately to reduce the maintenance cost. This paper studied the method of diagnosis based on bayesian network,described the details of the method for building the bayesian network diagnosis model ,especially for the derivation and calculaiton.It built the bayesian network MATLAB model of wind turbine using the historical fault data,this model contains Directed acyclic graph and Conditional probability distribution. Finally,this paper simulated two kinds of faults in the wind turbine system,two methods were used to diagnose these fault,One is the Bayesian network model and the other is the Dependency matrix.The result is that the Bayesian network method has the better fault resolution more than the Dependency matrix method,this could provides strong support to reducing maintenance cost of the complex system .