基于随机森林的喷水推进装置控制回路故障检测方法

Fault Detection Method for Control Loop of Water Jet Propulsion Device Based on Random Forest

为了及时准确发现喷水推进装置发生的故障,提升其运行的安全性,提出一种数据驱动的故障检测方法。将随机森林引入喷水推进装置故障检测领域,解决当前闭环控制回路故障检测缺乏自动化、智能化手段的问题;采用特征重要性度量策略对用于构建随机森林模型的特征参数进行评估,在满足故障检测精度的需求下降低特征参数维度,提升故障检测的实时性;基于网格搜索机制对随机森林的超参数进行优化,提升故障检测的准确率。通过基于AMESim的仿真试验和基于NICompact RIO的实物试验综合验证所提方法的有效性,结果表明该方法能有效提升喷水推进装置的可靠性,具有较强的工程应用价值。

In order to timely and accurately detect faults in the water jet propulsion system and improve its operational safety,a data-driven fault detection method is proposed.Random forest is introduced into the field of water jet propulsion device fault detection to solve the problem of lack of automation and intelligent means in current closed-loop control circuit fault detection.The feature importance measurement strategy is adopted to evaluate the feature parameters used to construct the random forest model,reduce the feature parameter dimensionality while meeting the requirements for fault detection accuracy,and improve the real-time performance of fault detection.Based on the grid search mechanism,the hyper parameters of random forests are optimized,and the accuracy of fault detection is improved.The effectiveness of the proposed method is comprehensively verified by simulation experiments based on AMESim and physical experiments based on NI CompactRIO.The experimental results indicate that the proposed method can effectively improve the reliability of waterjet propulsion devices and has strong engineering value.