一种基于小样本数据的装备故障预测方法

A Method for Equipment Fault Prognosis Based on Small Samples

针对目前装备故障预测面临的数据小样本问题以及现有数据驱动方法的不足,提出了一种基于相关向量机的装备故障预测方法,并通过故障预测仿真实例验证了方法的有效性和合理性。理论分析及应用结果表明:与神经网络、支持向量机等方法相比,该方法在保持较高预测精度的同时,在输出形式、参数设置等方面具有优势,应用前景广阔。

Aiming at the problem of small samples in equipment fault prognosis and the defects of the existing data-driven methods, a meth- od for fault prognosis based on relevance vector machine was proposed. Example calculation results demonstrate the rationality and the va- lidity of the proposed method. The theoretical analysis and the results of example calculation show that, compared with support vector ma- chine, the proposed method is better in terms of output form and parameter setting while retain high prediction accuracy, and is more suit- able for on-line prognosis.