基于概率神经网络的汽车发动机失火故障诊断  

Engine Misfire Diagnosis Based on Probabilistic Neural Network

作　　者：王子健[1] 王德军[1] 

机构地区：[1]吉林大学通信工程学院,长春130012

出　　处：《吉林大学学报（信息科学版）》2016年第2期229-236,共8页Journal of Jilin University（Information Science Edition）

基　　金：国家自然科学重点基金资助项目(61034001)

摘　　要：为准确诊断汽车发动机常发生的单缸失火和双缸失火故障,利用概率神经网络分析发动机转速与曲轴位移角度诊断发动机失火故障。在AMEsim软件环境下搭建一款四缸发动机模型,利用故障注入的方式模拟发动机失火,提取发动机转速和曲轴角度位移数据,在Matlab环境下进行数据处理与分组,建立概率神经网络PNN(Probabilistic Neural Network)进行训练与测试。实验结果表明,发动机转速与曲轴转角位移能有效反应发动机真实运行情况,训练好的PNN可对发动机单缸、双缸失火进行准确的诊断和定位。该方法具有简洁、经济、高效和准确度高等优点。Due to the phenomenon of engine misfire in single cylinder and dual cylinders often occured in vehicles, PNN（ Probabilistic Neural Network ） was used to detect the engine misfire by analyzing the engine rotary velocity and crankshaft angular displacement. Under the circumstance of AMEsim software, we construct a kind of four cylinders in line engine and use the injecting approach to simulate engine misfire. Then, extract the engine velocity and crankshaft angular displacement, put these data in the Matlab software to process and classify them. A PNN for training and testing is established. The experiment results indicate that the engine rotary velocity and crankshaft displacement reflect the real operation condition of the engine, the tested PNN can diagnose and probe single cylinder or double cylinders misfire in the engine. This technique has series of advantages of simplicity, economy, efficiency and high accuracy.

关 键 词：AMEsim发动机模型 故障注入 失火故障诊断 概率神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]