排灌机械工程学报2025，Vol.43Issue(6)：588-594,7.DOI:10.3969/j.issn.1674-8530.23.0280
多策略改进的SSA-VMD和BiLSTM的水泵机组特征识别
Multi-strategy improved SSA-VMD and BiLSTM for water pump unit feature extraction
摘要
Abstract
An ISSA combined with VMD and multidimensional statistical indexes feature extraction method for water pumping units was proposed,which applied ISSA to BiLSTMto improve the accuracy of subsequent pumping unit fault feature classification.Firstly,ISSA was used to determine the optimal decomposition parameters of VMD to extract high-quality modal components.Subsequently,the multi-dimensional statistical indexes of the modal components were calculated,based on which the compre-hensive feature vectors were constructed.Finally,the multidimensional feature vectors were input into the fault diagnosis model for feature classification.The experimental results show that the accuracy of the proposed fault diagnosis method is as high as 98.75%under the given data set.At the same time,in order to explore the good effect of the proposed model,SSA,WOA,and GTO are introduced for com-parative analysis.Under the same model conditions,the effect of ISSA algorithm is significantly better than other optimisation algorithms,and at the same time,the effect of constructing the feature vectors by using multidimensional statistical indexes is also significantly better than that of using a single index as the fault feature,which provides an effective solution for dealing with actual vibration signals.关键词
水泵机组/特征提取/改进的麻雀搜索算法/多维特征指标/故障诊断Key words
water pump unit/feature extraction/improved sparrow search algorithm/multidimensional feature index/fault diagnosis分类
农业工程引用本文复制引用
张兼博,曾云,李想..多策略改进的SSA-VMD和BiLSTM的水泵机组特征识别[J].排灌机械工程学报,2025,43(6):588-594,7.基金项目
云南省揭榜挂帅科技项目(202204BWO50001) （202204BWO50001）
