果蝇算法优化的相关向量机在电子装备中的应用

针对电子装备渐变故障预测问题,提出一种基于果蝇算法优化相关向量机的故障预测方法。该方法将原始时间序列数据进行相空间重构处理,并基于折交叉验证和果蝇算法优化相关向量机模型的核函数参数,从而建立故障预测模型,并以某型雷达发射机速调管监测数据对模型性能进行了验证。实验结果表明,相比已有方法,该方法在全局优化、收敛速度、预测精度以及预测可靠性方面都具有一定优势。

Application of RVM Based on FOA to Fault Prognostic of Electronic Equipment

To solve the gradually varied fault prognostic problem of electronic equipment,a method of Relevance Vector Machine (RVM)based on fruit Fly Optimization Algorithm (FOA)is presented. The original time sequence data are reconstructed in the phase space as input,and the kernel function parameter of RVM model is optimized based on FOA and k-fold cross validation to establish the prediction model. And the performance of the proposed model is validated by radar transmitter fault prediction experiment. The results demonstrate that the presented method has better global optimization,convergengce speed,prediction accuracy and reliability than the existed methods.