基于瞬时转速和机器学习的船用柴油机健康状态评估

以Z6170型船用中速柴油机为对象，进行故障模拟试验，参考已有文献，提取瞬时转速多维特征参数，形成不同状态下的瞬时转速特征参数样本，通过t-分布邻域嵌入算法（t-SNE）进行特征向量的维数约简，采用四种不同算法完成柴油机总体运行状态的自动识别，再采用随机森算法建立故障缸识别模型。研究结果表明：结合柴油机工作过程，提取瞬时转速多维特征参数，利用随机森林算法，建立柴油机总体状态评估与故障缸判定模型，可较好的评估其健康状态。

Evaluation of the Health of Marine Diesel Engines Based on Instantaneous Angular Speed and Machine Learning

Taking a Z6170 type marine medium-speed diesel engine as an object, some fault simulation tests were conducted. The mul-tidimensional characteristic parameters of instantaneous speed were extracted to form the samples of instantaneous speed characteristic parameters in different states of the diesel engine referring to the past literature. The dimension reduction of characteristic parameters was carried out by t-distributed stochastic neighbor embedding algorithm (t-SNE), and the overall operation state of the diesel engine was automatically identified by four different algorithms. Then the random forests algorithm was used to establish the fault cylinder identification model. It has shown if the multi-dimensional characteristic parameters of instantaneous speed are extracted combined with the working process of the diesel engine, the model of overall state evaluation and fault cylinder determination of the diesel en-gine can be established by using the random forest algorithm, it is promising to better evaluate the health state of the marine diesel engine by using the instantaneous speed.