进行了拖拉机田间作业故障跟踪试验 ,得到了拖拉机现场使用的故障数据 ,用非齐次泊松过程建立了拖拉机故障过程的数学模型 ,并用神经网络方法给出了模型的参数估计 ,讨论了拖拉机故障过程的改善与劣化与故障强度函数的关系 ,得到了拖拉机使用初期的故障强度曲线 ,给出了平均故障间隔时间的估计值 . The tractor field operation fault tracking test was carried out, the tractor field fault data was obtained, the mathematic model of tractor failure process was established by using nonhomogeneous Poisson process, and the parameter estimation of the model was given by using neural network method. The tractor failure The relationship between the improvement and deterioration of the process and the failure strength function, the initial failure strength curve of the tractor is obtained, and the estimation of the mean time between failure (MTBF) is given.