基于ARIMA-KF模型的船舶系统设备状态参数预测

针对目前大多数预测模型在船舶智能运维领域应用过程中存在的预测精度偏低、模型不易构建等问题,提出了自回归积分滑动平均模型(Auto-Regressive Integrated Moving-Average Model,ARIMA)和卡尔曼滤波(Kalman-filter,KF)相结合的船舶系统设备状态参数组合预测模型—ARIMA-KF模型。该模型首先构建了自回归积分滑动平均(ARIMA)单步和多步预测模型；然后利用卡尔曼滤波(KF)算法对ARIMA预测模型参数值进行寻优,得到ARIMA-KF组合预测模型；最后,基于组合模型对船舶海水冷却系统状态参数进行预测,将预测值与实船获取的实际值进行对比及误差分析。结果表明,采用基于ARIMA-KF组合模型比单一的ARIMA模型预测精度提高3%左右。研究结果对船舶系统设备的健康管理和视情维修具有一定的指导意义。

Research on Main Parameter of Equipment Based on ARIMA-KF

In order to solve the problems of low prediction accuracy and difficulty in Model construction during the application of most prediction models in the field of intelligent operation and maintenance in ships, a combined prediction model of ship system and equipment state parameters, ARIMA-KF model, which combines autoregressive integral moving average model(Auto-Regressive Integrated Moving-Average Model,ARIMA) and Kalman filter(KF), is proposed. Firstly, the single-step and multi-step prediction models of ARIMA are constructed. Secondly, the Kalman filter algorithm is used to optimize the parameter values of the ARIMA prediction model, and the ARIMA-KF combined prediction model is obtained. Finally, the state parameters of Marine seawater cooling system are predicted based on the combined model, and the predicted values are compared with the actual values obtained by the actual ship and the error analysis is made. The results show that the prediction accuracy of the combined ARIMA-KF model is about 3% higher than that of the single ARIMA model.The results of this study have certain guiding significance for the health management and condition-based maintenance of ship system and equipment.