针对泵压式供应系统液体火箭发动机的故障监控问题，建立了用于发动机故障检测的非线性动态数学模型，设计并实现了发动机系统的广义卡尔曼滤波器。利用新息序列的统计特性，进行了发动机故障新息检测算法的仿真研究，讨论了降低滤波器计算费用的方法以及置信度、自由度对检测算法性能的影响。本文的工作为进一步研究发动机故障在线实时检测算法奠定了重要基础。 Aiming at the fault monitoring problem of liquid propellant rocket engine in pump pressure supply system, a nonlinear dynamic mathematical model for engine fault detection was established and a generalized Kalman filter of engine system was designed and implemented. Based on the statistic characteristics of the new interest series, the simulation research on the detection algorithm of the engine failure is carried out. The methods of reducing the computational cost of the filter and the influence of confidence and degree of freedom on the performance of the detection algorithm are discussed. The work in this paper lays an important foundation for further research on online real-time detection algorithm of engine fault.