TheApplication Of Forecasting Model Based On PCA-ELM In Coal Mine Water Burst Prediction

Coal mine water burst forecast is a complicated problem. It includes manyfactors including hydrological geology, engineering geology, rock mechanics andmany other factors. Coal mine water burst is affected by many changing factors,traditional methods have many difficulties to accurately predict coal mine water burst.In this regard, the prediction method is proposed based on the principal componentanalysis (PCA) and extreme learning machine (ELM).Through the existing research of coal mine water burst forecast methods,pointed out that those methods has advantages and shortcomings. Combinemechanism of mine water burst, the latest method named ELM to predict coal minewater burst is selected, it can reach the requirements of real time, accuracy andreliability.Previous works indicated that when there are too many factors, the informationthat those factors contains generally overlap. If ELM is directly used for networkprediction, these overlapping information can reduce the operation rate. Therefore,PCA is selected to pretreat samples, Meanwhile, it is possible to furthest retain theoriginal information, so as to shorten the neural network model of training time andimprove the forecast performance.Forecasting model is constructed, input of network is principal componentanalyzed with SPSS software, and the network structure of ELM is designed. Byexperiments, the data of coal mine water burst is elected, and the network structure isdetermined, it includes the selection of the number of nodes in the hidden layer andexcitation function. The forecasting method is applied to predict coal mine water burst,obtain the experimental results. Comparing with the previous water burst predictionmethods, its performance is tested. The results of coal mine water burst predictionmethod proposed show that it has better effect.