基于冗余信息量的设备行为可预测性的研究

Equipment Behavior Predictability Evaluation Based on Redundancy

为了定量评估机械信号的可预测性,从而正确预测机械运行状态,文中提出了一种应用Bootstrap技术和冗余信息量相结合来评估各种类型机械信号的新方法.根据对周期信号、准周期信号、混沌信号、随机白噪声信号以及一组实际振动数据的冗余信息量的估算结果,得出信号的最大可预测极限,从而在很大程度上为提高机械设备长周期运行提供了依据.研究结果表明,该方法能定量地表示机械信号间的相互依赖关系,为准确预测机械运行状态提供了充分的依据.

In order to evaluate the predictability of mechanical signal quantitatively for the condition of machine behavior, a new method based on bootstrap technique and redundancy is proposed to deal with all types of mechanical signals. According to the analysis of computational results of the redundancies of four simulating time series and a set of practical vibration data, the optimum prediction precision was obtained. The results show that the method can quantitatively express the dependent relationship among mechanical signals and can exactly provide sufficient evidence for the future prediction of equipment behavior.