Qinghai-tibet Railway Wind Monitoring And Warning System, Wind Speed Prediction Algorithm

Strong-wind is a kind of the disaster weather which affects the security and stability of the railway transport. With the rapid development of the high-speed railway, train safety is under the serious hidden danger brought by the powerful crosswind.Train accidents that caused by strong-wind, have occurred occasionally around the world.Qinghai-Tibet Railway from Geermu to Lasa locates in the hinterland of Qinghai-Tibet Plateau, where extreme weather events,particularly the strong-wind,occur frequently due to complex climate and that bring serious hidden-troubles to the security of the railway transport. For most wind monitoring stations along the Qinghai-Tibet Railway, the maximum wind speed was over 25m/s in history, even exceeded 40m/s, which leaded to enormous destruction to the railway bridge, railway line, vehicles and communications equipment. Consequently, as an urgent demand for railway departments to make scientific guidance,it has socio-economic significance and applicable value in engineering to research on the forecast algorithm and the evolvement law of the wind along the railway.Wind speed along the Qinghai-Tibet Railway shows the randomicity and dynamic correlation. Accordingly, the thesis proposes the forecast algorithm based on the theory of Time Series Analysis by summarizing the foreign and domestic research results.Moreover, the thesis puts forward two kinds of prediction algorithm to further improve the predict effect:(1)SVM-ARMA algorithm, which is the combination of Support Vector Machine with Time Series Analysis Theory, builds the Time Series Analysis model in the sense of Structural Risk Minimization Principle; (2) ARMA-SMFKF algorithm, which is the combination of Time Series Analysis Theory with Kalman Filtering Theory, is an automatic error correction algorithm which made use of new information brought from observed value;For the purpose of strengthening the ability of tracing mutational sites.The above two algorithms have tested by wind speed sequence of Qinghai-Tibet Railway and results show that:Compared with ARIMA algorithm, accuracy of SVM-ARMA algorithm on short-term wind speed prediction has improved;The algorithm of ARMA-SMFKF has substantially improved the prediction accuracy,especially the accuracy of sequence of mutational sites.Furthermore,the problem of prediction delay has also been significantly improved. Two kinds of optimization algorithms have achieved good prediction effect and they have great value to application and research.In the end, the thesis has designed and developed a set of software system for the purpose of forecasting wind speed of the Qinghai-Tibet Railway, which is based on Matlab and SQL Server Database.It provides wind information inquiry function of the 52 wind monitoring stations along Qinghai-Tibet railway, and using above algorithms achieved the function of wind speed time series stationarity testing and processinging, pattern recognition, model-order determination, model parameter estimation, wind prediction and so on.