Voiceprint Detection Technology for Main Equipment Faults in Pumped Storage Power Plants Based on Small Sample Anomaly Feature Mining Algorithm
Due to the difficulty in accessing some equipment in pumped storage power plants,it is difficult to collect ef-fective signals.In order to reduce the missed and false detection rates in equipment fault detection,a new fault detection method for main equipment in pumped storage power plants is proposed by combining small sample anomaly feature mining algorithm with voiceprint recognition technology.Select a piezoelectric accelerometer as the voiceprint acquisition sensor de-vice,set relevant parameters,and determine the sensor placement position through Pearson correlation coefficient.Use the voiceprint data acquisition module to collect operating status data.Combining EMD with wavelet transform to denoise the collected operational status data of the main equipment and reduce fault detection errors.Using a small sample anomaly fea-ture mining algorithm for anomaly data mining on denoised datasets,combining the mining results with voiceprint data pro-cessing to achieve main device fault detection.The experimental results show that the proposed method not only reduces the missed detection rate and false detection rate of fault detection,but also overcomes the influence of interference signals,with high practical value.
small sample anomaly feature mining algorithmvoiceprint recognition technologypumped storage power plantsfault detectionvoiceprint acquisition sensor
国家科技期刊平台
NSTL
万方数据
