Research On Safety Monitoring Statistic Model Of Concrete Dam Based On BP Neural Network

Since 1949, many weir dams was constructed, these projects play an important role in flood control, power generation, irrigation and so on. But because large-scale hydraulic structures, complicated geological conditions, structure in a variety of forms, changed operating conditions, beared heavier loads, construction lasted longer and complex, improper operation and management after the completion, structure aging and other reasons, safety monitoring of water conservancy engineering is gradually valued. The theory and method of dam safety monitoring was researched deeply which not only play an important role in the monitoring of dam safety conditions, give full play to the engineering benefit under the premise of the dam operation safely,but also has far-reaching significance of develop of dam theory and improve the level of construction and management.At present, there are so many ways to process dam deformation monitoring data, which include statistical model, deterministic modelc and mixed models. New methods have been proposed, such as neural network model and wavelet model. In recent years,PSO Fuzzy Clustering Algorithm, Improved Particle Swarm Optimization, Shuffled Frog Leaping Algorithm, PSO bionic algorithm, Ant Colony Algorithm and so on were introduced process dam deformation monitoring data, and significant progress has been obtained. These methods all have their own advantages and disadvantages, such as, the mapping relation of dam influential factors can be described by statistical model, analytical calculation was carried out on the variable, the deformation law of dam was quantitative analysis, But a function of model need artificial regulation; BP artificial artificial neural network (ANN)model which better than statistical model in sides of accuracy and efficiency prediction, but it falling into minimal part easily and poor generalization and soAccording to the BP ANN model characteristics of more accurate measurement and the statistical model characteristics of concise and practical, the paper puts forward a new method of dam safety monitoring data processing, which is the statistical model based on BP ANN.Using the difference between the calculated values of BP ANN model and the measured values as the effect quantities of statistical model. Thus the difference statistical model of dam deformation was calculated.Then combined calculated value of the difference statistical model with BP ANN model calculation values, the statistical model based on BP ANN calculation values was obtained. When predicting, combined The BP ANN model of prediction wih predictive value of the difference statistical model, which is the predictive value of the statistical model based on BP ANN.The model is used for dam monitoring data analysis, build the statistical model, the BP ANN model and the statistical model based on BP ANN of 16# and 18# dam section according to the monitoring data of Xinanjiang gravity dam. And analyze the feasibility and superiority of this method.The primary contents of the paper are as follows:(1) Discussed are the basis principle and method of establishing statistical models and the BP ANN of main observation vailables of concrete darn.(2) The statistical model was build based on SPSS18.0, and the BP ANN model was build based on MATLAB 2012. According to modeling and forecasting results in the two models, BP ANN model is better than the statistical model in the aspect of modeling and prediction accuracy.(3) Through analysis of the advantages and disadvantages of the statistical model and BP ANN, build the statistical model based on BP ANN.(4)Build the statistical model, the BP ANN model and the statistical model based on BP ANN of 18* dam section, through comparde of 18# dam section of three kinds of model, the statistical model based on BP ANN of the modeling and forecasting accuracy is superior to the statistical model and BP ANN model. To further verify the validity and superiority of this method.In a word, the statistical model based on BP ANN which mix of the precision of the BP ANN model and the analytical expression of the statistical model. The precision of prediction is better than single neural network model and statistical model.