基于传感器信号和神经网络的汽油机失火状态识别方法

State Recognition Method for Gasoline Engine Misfire Based on Sensor Signal and Neural Network

提出以传感器信号作为特征输入向量,建立BP神经网络模型进行失火状态识别的方法。汽油机运行状态可以通过发动机转速、进气歧管绝对压力、节气门开度等主要参数来表征,利用解码器,获取发动机稳定和瞬态空载工况下正常状态和1缸、1、4两缸失火状态的上述传感器信号数据,以此作为训练样本,建立了BP神经网络模型进行失火状态的识别。结果表明,此方法能正确识别出正常状态、一缸和两缸失火状态,并且测试方案更简单,成本更低。

A state recognition method for misfire is presented, this method is based on BP neural network and sensor signals as characteristic input vectors. The gasoline engine operating status can be reflected by engine speed, intake manifold absolute pressure, throttle position and such main parameters, using decoder to collect above sensor signals data under normal unload condition, cylinder 1 misfire condition and cylinder 1 -4 mis- fire condition, setting the data as training samples and designing a BP neural network model for misfire state recognition. The result shows that this method can correctly identify the normal state, 1 cylinder misfire and two cylinders misfire states, moreover, this test scheme is simpler and more cost-effective.