Research And Application Of The Discrete Data Fitting Model

The essay is based on the project of researching the virtual instrument for measuring the motor parameters, focusing on the study of the curve method of small sample data that is in discrete distribution. Using the error analysis and data fitting technology, the essay proposes a discrete data fitting model which can detect outline points from the testing data and the results are visible.The project probes into the theory and method of three-phase squirrel-cage asynchronous motor parameters testing. It researches a new virtual instrument that can analyze motor parameters intelligently by using the technology of artificial intelligence, computer processing and high-speed interfaces. The virtual instrument can increase the standardization's degree of the test. The essay focuses on electrical test data post-processing and analysis, namely the curve fitting problem of motor performance parameters curve and motor characteristic curve. The main contents are divided into three parts as follows.Firstly, based on an in-depth study of fitting data algorithms, the essay mainly analyzes the disadvantages of the LSM(Least Squares Method) and the LSVM(Least Squares Vector Machine). Because of its conciseness and convenience, the LS (Least Squares) is always used to draw the parameter curves and performance curves. However, the LS gives a seriously distorted curve with an apparent apophysis, in the case of high order fitting. We can use LSVM to solve the problem, but the outline points will influence the resultant curve.Secondly, in order to solve the above problem, the essay attempts to build a new model using LMS(Least Median Squares) to analyze the error points, to pretreat the dynamic measuring errors and then to get the fitting curves of testing data.Thirdly, this model is used in curve fitting of electromotor parameters testing which includes load testing and no load testing. The results show that this model can erase the influence of outline points, improve the effects of data curve fitting and reflect the characteristic of the motor. Finally, the conclusion of the study is made, and the direction of further research is pointed out.