基于机器学习的燃气轮机控制系统分层故障诊断

针对传统算法在实际应用中存在网络规模庞大、学习训练时间过长和知识“组合爆炸”而导致网络组织失败等问题,提出基于变分模态分解（Variational Mode Decomposition,VMD）、自回归模型（Autoregressive Model,AR）和轻量型梯度提升机（Light Gradient Boosting Machine,LightGBM）算法的燃气轮机控制系统分层故障诊断方法。首先对原始信号进行VMD AR建模,获取最具表征性的故障特征向量;然后根据不同层级特点和故障类型设计多个LightGBM模型;最后采用贝叶斯优化算法优化分层诊断模型中的超参数,并将待测信号输入模型进行故障诊断。结果表明：所提算法不仅能够达到95%以上的故障识别率,而且具有更强的泛化能力。

Hierarchical Fault Diagnosis of Gas Turbine Control System based on Machine Learning

Aiming at the problems of large network scale,long learning and training time and easy knowledge combination explosion in practical applications of traditional algorithms which lead to network organization failure,a hierarchical fault diagnosis method of gas turbine control system based on variational mode decomposition (VMD),autoregressive model (AR) and light gradient boosting machine (LightGBM) algorithms was proposed.Firstly, the original signal was modeled by VMD AR to obtain the most representative fault feature vector.Secondly,multiple LightGBM models were designed according to different hierarchical characteristics and fault types.Finally,the Bayesian optimization algorithm was used to optimize the hyperparameters of the hierarchical diagnosis model,and the signal to be tested was input into the model for fault diagnosis.The results show that the proposed algorithm can not only achieve a fault recognition rate of more than 95%, but also has stronger generalization ability.