为了实时、准确、可靠地预测轨姿控发动机试验中推力矢量的变化情况,本文在分析支持向量机(SVM)预测算法缺陷的基础之上,提出并建立了一种自适应能力较强的故障预测模型——ARVM(Adaptive Relevance Vector Machine),并将其应用于某型轨姿控发动机高模试验推力矢量参数预测中.研究结果表明,ARVM预测模型在稀疏性和算法精度方面均体现出较高的优越性,能够很好地预测轨姿控发动机试验推力矢量的变化趋势. In order to predict the change of thrust vector in the track-and-nose control engine test in real time, accurately and reliably, based on the analysis of the defects of SVM prediction algorithm, this paper proposes and establishes a fault with strong self-adaptability Which is used to predict the thrust vector parameters of a high-mode test of a rail-orbit-controlled engine.The results show that the ARVM prediction model shows better performance in both sparsity and algorithm accuracy High superiority, can well forecast the trajectory control engine test thrust vector trend.