ABC算法优化SVR的磨损故障预测模型

目的为了提高故障预测的精度,针对支持向量回归SVR(Support vector machine for regression,SVR)参数选择困难的问题,提出一种采用人工蜂群(artificial bee colony,ABC)算法优化支持向量回归(SVR)的故障预测模型(ABC-SVR)。方法该模型先对样本数据进行重构,然后将故障预测误差(适应度)作为优化目标,通过ABC算法寻优找到最优的SVR参数,建立故障预测模型。最后通过实例仿真验证模型的优越性。结果采用ABC算法优化的SVR故障预测模型进行时间序列预测,能够较好地跟踪发动机滑油金属元素浓度的变化过程,并且能够提前2个取样时间预测异常情况的出现。结论 ABC-SVR模型有效解决了SVR参数选择难题,能够更加准确地表现故障变化规律,提高了故障预测精度。

Wear Fault Prediction Model Based on SVR Optimized by ABC

Objective To improve the prediction accuracy of wear faults, a wear fault prediction model (ABC-SVR), which was based on support vector machine for regression (SVR) optimized by artificial bee colony(ABC) algorithm was proposed. Methods The model reconstructed the time series of wear faults and took the wear fault prediction accuracy as the optimization objective to find out the optimal SVR parameters by ABC algorithm and build prediction model of wear faults. Finally, the simulative contrasting experiment was applied to test the performance of the model. Results Time series prediction with SVR forecasting model optimized by ABC algorithm could track the concentration change process of metallic element in engine lubricating oil and predict the presence of the abnormal situation ahead of 2 sampling time. Conclusion ABC-SVR solves the problem of SVR parameter optimization, can describe the complicated change rules of wear faults accurately, and improves the accuracy of wear faults prediction.