内燃机机体振动信号时域识别方法研究

Study of Time-Base Recognition Method for Vibation Signal From Internal Combustion Engine Cylinder Block

根据内燃机在工作过程中 ,其机体在不同时刻受到不同激振力作用这一特点 ,对机体振动信号在时域上进行了分析处理与识别。首先 ,对内燃机在工作过程中的机体振动信号进行小波降噪处理 ,从而突出了其机体在不同时刻受到激振的结果。然后 ,对机体在不同时刻所受到激振的结果分别建立二阶AR模型 ,以模型系数作为模式识别参数。最后 ,采用感知器神经网络对模型参数进行识别与分类。由于只对振动信号进行处理与识别便可得到与不同激振力相关的信息 ,因此这种处理与识别方法具有形式简单 ,便于实现的特点 ,因而 ,对实现内燃机在线控制 ,监测与故障诊断均有重要的应用价值。

According to that the cylinder block stands different excition forces at different time during an internal combustion engine runs.The vibration signals from a cylinder block are analyzed and identified on the time base.Firstly,the vibration signals are denoised by wavelet transformation in the working process of an engine.Secondly,according to the results of exciting force on the cylinder block at different time,the quadratic AR models are set up respectively,and then take model coefficients as the model recognition parameters.At last,the parameters are recognized and classified by apperceive neural network.For its ability to get information related to different exciting forces just by processing and recognizing the vibration signals,this method is simple and realizable.Therefore it is valuable for controlling cngine online,and also for crror monitoring and faults diagnosis.