基于脉冲耦合神经网络和粒子群优化算法的水轮发电机组故障在线录波检测  

Online Wave Recording Detection of Hydro-generator Set Faults Based on Pulse Coupled Neural Network and Particle Swarm Optimization Algorithm

作　　者：陈勇旭 万科 CHEN Yongxu;WAN Ke(CHN Energy Dadu River Houziyan Power Generation Co.,LTD.,Kangding Sichuan 626005,China)

机构地区：[1]国能大渡河猴子岩发电有限公司,四川康定626005

出　　处：《微电机》2025年第1期26-32,共7页Micromotors

摘　　要：提出基于脉冲耦合神经网络和粒子群优化算法的水轮发电机组故障在线录波检测方法,实时检测水轮发电机组的工作情况,识别隐性异常,防止出现严重故障,保证水轮发电机的平稳运行。利用故障录波器通过不同的采样频率分段记录水轮发电机组故障前后的电气量作为故障录波数据,采用粒子群优化算法,以记录的水轮发电机组故障录波数据波形SINR(信干噪比)最大为目标进行波形优化,降低外界环境各种干扰源及噪声的影响;将优化后的故障录波波形作为PCNN网络的输入数据,构建脉冲耦合神经网络,根据网络神经元所携带脉冲生成器的激活情况和脉冲的释放情况,判断出水轮机是否存在故障、故障的类型和严重程度等信息,实现水轮发电机组故障在线录波检测。实验表明,应用该方法可以及时、精准地检测出水轮发电机组故障类型,且在故障捕捉率方面有着较好表现。Proposed an online fault recording detection method for hydroelectric generators based on pulse coupled neural network and particle swarm optimization algorithm,which could detect the working condition of hydroelectric generators in real time,identify hidden anomalies,prevent serious faults,and ensure the smooth operation of hydroelectric generators.Using a fault recorder to segment and record the electrical quantities before and after faults in a hydroelectric generator set at different sampling frequencies as fault recording data,the particle swarm optimization algorithm was used to optimize the waveform of the recorded fault recording data of the hydroelectric generator set with the goal of maximizing the SINR(signal-to-noise ratio),reducing the influence of various external interference sources and noise;Using the optimized fault recording waveform as input data for the PCNN network,a pulse coupled neural network was constructed.Based on the activation and release of the pulse generator carried by the network neurons,information such as whether there is a fault in the water turbine,the type and severity of the fault were determined,and online fault recording detection of the water turbine generator set was achieved.The experiment shows that the application of this method can timely and accurately detect the types of faults in hydroelectric generators,and has good performance in fault capture rate.

关 键 词：水轮发电机组 脉冲耦合神经网络 录波数据 灰度化 故障检测 粒子群算法 

分 类 号：TM312[电气工程—电机]