| 摘要:为了实现柴油机智能化故障诊断，提出一种基于PCA-PNN算法的柴油机故障诊断方法。该方法兼具主成分分析（principal component analysis， PCA）法降低数据维数和概率神经网络（probabilistic neural network， PNN）法计算速度快、容错率低、稳定性好的特点。利用AVL BOOST软件建立柴油机仿真模型，并进行有效性验证，采集包含12个柴油机故障特征参数的195组样本数据集。故障诊断试验表明：PCA-PNN柴油机故障诊断算法简洁、易于推理，诊断准确率为94.87%、运行时间为0.673 s，准确率更高、诊断速度更快，为探索柴油机智能化故障诊断提供新的技术路径。 |
| 关键词: 柴油机 主成分分析法 概率神经网络算法 故障诊断 |
|
| Diesel Engine Fault Diagnosis Research Based on PCA-PNN Algorithm |
| XU Hongming,SUN Fei,WANG Lin |
| School of Maritime, Zhejiang Institute of Communications, Hangzhou 311112, China;Ningbo C.S.I. Power&Machinery Group Co., Ltd., Ningbo 315000, China |
| Abstract:In order to realize intelligent fault diagnosis of diesel engines，a fault diagnosis method for diesel engine based on PCA-PNN algorithm is proposed.This method combines the principal component analysis（PCA）method with probabilistic neural network（PNN），which has the advantages of high computation speed，low fault tolerance and good stability.Diesel engine model was established by AVL BOOST software，and the validity of the model was verified.195 groups of sample data including 12 fault characteristic parameters of diesel engine were collected.The experiments of fault diagnosis show that：PCA-PNN fault diagnosis algorithm of diesel engine is simple and easy to reason，the diagnosis accuracy rate is 94.87% and the running time is 0.673 s，the accuracy rate is higher and the diagnosis speed is faster，it provides a new technical path for exploring intelligent fault diagnosis of diesel engines. |
| Key words: diesel engine principal component analysis probabilistic neural network algorithm fault diagnosis |
