Research On Fault Warning Technology Of Track Circuit In Microcomputer Monitoring

With the rapid development of information technology and automatic fault detection, automatic fault identification and alarm has been used in various industrial production field. With the rapid development of domestic railway industry, especially the high-speed railway, China has become the world's longest operating mileage of high-speed railway opened in the state. So as to ensure safe and efficient operation of railway equipment, safe and reliable operation of track circuit, it is particularly important. How to identify and track circuit fault diagnosis rapidly and effectively has become one of the important factors to guarantee the safe operation of the railway.At present, most of the domestic track circuit fault diagnosis is the use of experienced staff to carry out the artificial identification so as to achieve the purpose of alarm, the degree of automation is low, the work efficiency is also limited. So in order to realize the automation of track circuit fault diagnosis and alarm, this thesis proposed two kinds of diagnostic algorithm, In the first part, according to the local weighted regression fitting principle, the machine learning algorithm is used to validate the algorithm, and the curve fitting method is used to achieve the goal of recognition, the advantage is that it can be seen more intuitive similarity,but the deficiency is that each time can only carry on the recognition of two curves, it can't simultaneously carry on the similarity recognition of multiple curves, the efficiency is relatively low, and there will be a part of the miscarriage of justice. The second algorithm draw lessons from the support vector machine algorithm in face recognition to achieve the ability of classification prediction, the technology is applied to the automatic classification recognition and prediction of track circuit fault curve, which is used to realize the automatic alarm of track circuit fault diagnosis. This method based on support vector machine fault curve fault classification prediction technology, Matlab as the main means to achieve the technology in the track circuit fault curve identification of the feasibility, that will be divided into three main steps, data characteristics of track circuit fault curve extraction, the characteristic value of the curve is the calculation of singular value decomposition, the training classification model. First, realize the normal curve and the curve of fault classification and prediction of two types, that is, the advantages of support vector machines, two classification prediction, and then combined with the voting algorithm mechanism to achieve a multi classification prediction of a variety of fault curves. The main technologies involved are studied, The feature extraction of the curve and the curve classifier are studied in detail, and finally achieved a higher recognition rate.The innovation point of this paper is:First, the extraction of numerical characteristics of fault curve singular value of this feature for the classifier of support vector machine training.Second, the least square support vector machine is used to classify the fault curve feature value, that is, the classification of singular value is used to predict the final realization of many types of track circuit fault curve.Finally, based on the feature extraction of the fault curve and the singular value decomposition, the data set classification prediction of the fault curve feature of the track circuit is realized. In contrast, locally weighted fitting algorithm, the classification ability of support vector machine high, based on the Matlab experimental platform, the classification and prediction of various types of track circuits based on the least squares support vector machine is verified, reached a higher classification prediction rate, of course, this study can also draw on some of today's more advanced technology and algorithms, such as big data and intelligent data analysis technology, to achieve a better classification of the ability to predict the fault and the more detailed fault distribution, to provide more reliable guarantee for the safe operation of railway track circuit in the future.