论文部分内容阅读
针对Kalman预测在非线性系统故障预报中预测误差较大的问题,提出一种基于支持向量机预测新息的Kalman预测方法.根据未知非线性系统的典型变量分析子空间模型进行Kalman预测,采用支持向量机时间序列预测算法预测未来时刻的新息,利用新息进行Kalman单步和多步预报.在连续搅拌反应器上的仿真研究表明:所提出方法能准确地预测较长时间段内故障过程的劣化趋势,预知可能发生的故障,使操作人员有时间采取必要措施消除故障隐患.
Aiming at the large prediction error of Kalman prediction in nonlinear system fault prediction, a Kalman prediction method based on SVM prediction is proposed.According to the Kalman prediction of typical variable analysis subspace model of unknown nonlinear system, The vector machine time series prediction algorithm predicts the new interest in the future and uses the new information to carry out the Kalman single-step and multi-step forecasting.The simulation on the continuous stirred reactor shows that the proposed method can accurately predict the fault process over a long period of time The trend of deterioration, predict the possible failure, so that operators have time to take the necessary measures to eliminate potential problems.