CAJ | 학술논문

为实现故障预测的实时性，克服设备实际数据处理能力的局限，采用了线性时变参数离散灰色模型（linear time-varying parameters discrete grey model ，TDGM ），并对其建模所用的最小二乘法采用递推算法，形成了递推建模法，解决了实时建模预测所面临的建模序列取舍、计算和存储量巨大等难题，并采用视情剔除旧数据的方法克服了递推建模法存在的“数据饱和”现象，完成了在线故障预测系统（online fault prediction sys-tem ，OFPS）的初步设计。实例仿真验证表明：该系统运行稳定可靠，计算和存储量在可承受范围之内，预测效果良好，可以应用到实际设备中。
위실현고장예측적실시성，극복설비실제수거처리능력적국한，채용료선성시변삼수리산회색모형（linear time-varying parameters discrete grey model ，TDGM ），병대기건모소용적최소이승법채용체추산법，형성료체추건모법，해결료실시건모예측소면림적건모서렬취사、계산화존저량거대등난제，병채용시정척제구수거적방법극복료체추건모법존재적“수거포화”현상，완성료재선고장예측계통（online fault prediction sys-tem ，OFPS）적초보설계。실례방진험증표명：해계통운행은정가고，계산화존저량재가승수범위지내，예측효과량호，가이응용도실제설비중。
In order to realize real-time fault prediction and overcome the limitations of actual data pro-cessing capacity of the equipment ,this paper applies a recursive algorithm to the least square method used for establishing a linear time-varying parameters discrete grey model (TDGM ) .This is called re-cursive modeling method .T his method can solve the puzzles in the choice of modeling data sequence and in the huge amount of computation and storage and so on .Besides ,the method of condition-based old data removal is used to deal with the data saturation caused by the recursive modeling method . The above efforts leads to the preliminary design of an online fault prediction system (OFPS) .The simulation results show that the OFPS is stable and reliable in operation with a tolerable range of com-putation and storage and a good prediction effect .T herefore ,it can be applied to practical devices .