发电机组电气故障智能检测技术研究

电气故障是发电机组正常运行的重大隐患,及时检出电气故障具有重要意义。针对发电机组的电气故障检测问题,提出一种基于回声状态网络的智能检测方法,可以实现发电机组电气故障的自动实时检测。此方法中,将发电机组的常见10类故障作为输入,进而纳入回声状态网络中进行学习和训练,并根据网络分析结果做出故障判断。实验过程中,通过多种传感器配合人工巡检采集各变量数据,通过多变量时间序列数据在回声状态网络中的训练和学习,形成故障检测的最终结果。实验结果表明,该文提出的智能检测算法可以对发电机组的运行状态进行自动实时的故障检测。该文所提出的方法具有非人工、智能化、实时性的特点。

Electrical fault is a major hidden danger to the normal operation of generator sets,so it is of great significance to detect electrical faults in time.Aiming at the problem of electrical fault detection of generator set,an intelligent detection method based on echo state network is proposed,which can realize automatic real-time detection of generator set electrical fault.In this method,10 kinds of common faults of generator sets are taken as input,and then incorporated into the echo state network for learning and training,and the fault judgment is made according to the results of network analysis.In the course of the experiment,through a variety of sensors with manual inspection to collect variable data,and through the training and learning of multi-variable time series data in the echo state network,the final result of fault detection is formed.The experimental results show that the intelligent detection algorithm proposed in this paper can automatically and real-time detect the fault of the generator set.The method proposed in this paper is non-manual,intelligent and real-time.