Improved GA-BP Model Establishment And Analysis Of Deformation Prediction Of Coal Industry Square

Mine buildings within the industrial square for safety production of mine is of greatsignificant, but inevitable uneven settlement may affect the normal use of the buildings,and even can cause tremendous disasters and accidents. So it is necessary to monitorsettlement deformation of the mine buildings systematically and regularly. Base on theanalysis of the observational data,and then predicting subsidence deformation of the futurefor a period of time, and take the appropriate measures to deal with it.BP neural network is more widely used as a kind of prediction methods, but it hassome problem that network’s structure is difficult to reasonable design and easy to fall intolocal optimum. Simple genetic algorithm (SGA) is easy to premature convergence, it canadaptively adjust the crossover probability and mutation probability Pc Pm measure torealize the improvement of the simple genetic algorithm, by using the improved geneticalgorithm first on BP network structure optimization design, and then on the BP networkweights and threshold for global optimization,so as to achieve the improved GA-BP model.The programming is realized by using MATLAB, the settlement monitoring data ofYihai company Dameigou coal mine industrial square auxiliary shaft winch room as thebasis, first the original monitoring data interpolation pre-processing in order to actualizeequal time interval sequence, and then the simulation prediction with improved GA-BPmodel and several other methods. The results show that the improved GA-BP model in thecumulative settlement prediction and single stage settlement prediction are the best, canmeet the requirements of settlement deformation prediction of coal mine industrial squarebuilding.