| Electrical quantities and accuracy of various other data are directly related to the proper working of substation automation equipment in substation automation systems.And detect the bad data of the substation can help to understand the operation of power system equipment. Machine learning methods are more intelligentto detect bad data than traditional method. Machine learning methods are able to reduce false detection and missed detection of bad data. The main work of this paper is as follows:Firstly, this paper set forth the theory of machine learning and machine learning strategies briefly.This paper is focusing on support vector machine of statistical learning.Secondly, this paper analyzes the source of bad data by scheduling automation systems and substation automation systems. Elaborateresidual detection method of traditional bad data detection method, and point out its shortcomings.The estimator is establishedon support vector machine regression method by comparing the difference between the measured data and the estimated data.Machine learning methods are more intelligent compared with the traditional method.Finally, a set of substation equipment state detection software is developed basing on automated information detection project. This softwaredetects bad data firstly and then detectsthe unbalance of three-phase current and voltage. In the end switching state and its monitored state is detected. The software is very reliable and more economicalcompared to the manual inspection of electrical substation equipment. |
