基于支持向量机的大型水泵性能预测

Large-scale water pump performance forecast based on support vector machines

为了预测大型水泵在运行中的性能,采用了统计学习理论中的核心算法——支持向量机。针对某一全调节轴流水泵,建立了水泵扬程计算模型和效率计算模型,并绘制出了不同叶片角度时的性能曲线。结果表明,基于支持向量机建立的水泵性能计算模型具有一定的简洁性,只需要知道少量的训练样本就能建立数学计算模型;它克服了传统的实验方法获取性能曲线费用高,也克服了神经网络方法中"过学习"现象和"欠学习"现象的出现;该模型可以很好地预测大型水泵的综合性能,提高水泵的使用效率。

In order to forecast the large-scale water pump running performance, the kernel algorithm of Statistical Learning Theory （SLT）, Support Vector Machine （SVM）, is applied. For an all-adjusted axial flow pump, the calculation model of pump lift and efficiency calculation model are set up, and the performance curves at the different angles, of blade are drawn. The calculation result indicates that the calculation model of water pump performance based on SVM is pithily and the model can be set up with a small amount of train samples. It overcomes the defects that the high cost in obtaining performance curves with traditional tests and the appearance of over-fitting and under-fitting fitting in the ANN. The model can forecast the comprehensive performance of large-scale water pump and increases the rate of utilization of the pump.