摘 要:研究试验数据处理问题对电机系统的分析具有重要意义。在数据处理中,通常采用最小二乘支持向量机(LS-SVM)的方法进行电机试验数据拟合。但当电机试验数据集中存在异常点时,拟合曲线和真实曲线会出现较大误差,影响试验结果的准确性。针对这一问题,该文将测量不确定度理论应用在电机试验数据拟合中,构建了相应的数学模型,并通过小样本电机试验数据的拟合对该模型进行了试验分析。结果表明,采用测量不确定度理论的方法后,使用最小二乘支持向量机的方法取得了更优拟合效果,具有较高的工程应用价值。 关键词:测量不确定度;数据拟合;最小二乘支持向量机;试验数据
Abstract: To research the problem of test data processing is significant to the analysis of system in the motor. The method of least squares support vector machines(LS-SVM) is usually used to fit test data during the data processing in motor. However, if there are outliers of test data set in the motor, it will appear a large difference between the fitting curves and the real curves and affect the accuracy of test results. So it used the theory of uncertainty in data fitting and constructed a corresponding mathematical model which making an experimental analysis with small sample data fitting to solve this problem. The results shows that, through using the method of LS-SVM after using the theory of adopting the uncertainty, it achieved a better fitting result which has a high value in engineering. Key words: uncertainty; data fitting; LS-SVM; test data
|