Study On Nonlinear Prediction For Reservoir Bank Collapse In Fuling Area Of The Three Gorges Project

The Three Gorges Project, which has been focused upon for a long time all over the world, formally started to be constructed on December 14th, 1994. The second stage of dam works of Three Gorges Project with the span of 1.6km has been closed crest completely on October 26th, 2002. The altitude of the whole dam has risen to 185m as designed. After the impoundment of the Three Gorges Reservoir, water level has increased to 135m in 2003, then has reached up to 156m in 2006. Furthermore, it will arrive at 175m in 2009 under normal operation. After storing water, the glory sight of Safety of Water Filling will come true and realize the dreams of generations. The Three Gorges Project on the Yangtze River is the largest water conservancy and hydroelectricity project in the world with comprehensive benefits, which will benefit offspring forever and will provide important effect for the flood control, irrigation and power generation in the Yangtze Rive drainage area. However, it will arouse and aggravate geology disasters of bank collapse in the Three Gorges Region on the Yangtze River.After the construction of reservoir on huge watercourse, the natural environment will change greatly along the Yangtze River. During water storing, the rising of water level in reservoirs will increase the erosion base level and enhance the ground water level. Meanwhile it will arouse dynamic changes of Hydrological factors of watercourse. In addition, the area along river will suffer fiercely from alteration. After water storing, affection from reservoir water will swell on slope stability of reservoir banks along the Yangtze River ranged from 135m to 175m. Several factors will accelerate the bank collapse, such as increasing the span of dipping in water, accreting the fluctuation of the water level, strengthening underground water dynamical effect. Bank collapse on reservoir bank will bring about graveness influence to people who live along the reservoir banks. Whereas, correlative Departments of State were required to make out some special subject research, such as the geological investigation to the bank collapse, prediction for reservoir bank collapse, prevention and cure for bank collapse in the Three Gorges Reservoir. In this case, this paper especially makes research on the geology conditions of bank collapse and prediction for reservoir bank collapse. Obviously, the theme selection and research of this paper is of scientific significance, at the same time it is of extensive practicability.The Three Gorges Reservoir is the typical watercourse reservoir, and the submerged scope is very spacious. 26 countries in Hubei Province and Chongqing city will be influenced, and the water level will be raised to one hundred meters. The sluice of reservoir will be attempered by the function between storage in winter and release in summer. After the construction of the Three Gorges Reservoir, the water level of the dam will fluctuate between 145m and 175m. The span of water level will change up to 30m at most. For the date of the third stage water storing being approaching, the stability of bank collapse in the Three Gorges Reservoir is more and more important at high pitch than ever before.Reservoir bank collapse is the severe geological disaster, which will influence the security of the Three Gorges Project. For a long time, the research on reservoir bank collapse is focused on soil bank collapse in plain region at home and broad, and research on the watercourse bank collapse by the scientists and scholars seem little, especially in the area of Three Gorges Reservoir. Therefore, the traditional methods of collapse prediction are unfit for the reservoir bank of river channel in mountain area, which are only suit to the wash and abrasion type of reservoir bank. Accordingly, suitable prediction model and method systems should be established which is fit for the type of the reservoir bank collapse in the area of Three Gorges Project.Owning to the bank collapse of reservoir banks distributed widely, the stability conditions of banks in the area of Three Gorges Reservoir will be sharply deteriorated after water storing. In order to find out the rule of bank collapse, and to classify and count the forecasting parameters of bank collapse in the Yangtze Three Gorges Project Region, several methods were adopted as following: investigating the collapse geology conditions, analyzing the influence factors of collapse classifiably, synthetically studying on the pattern of bank collapse, the prediction methods for bank collapse and making out the study on how to assign a value of prediction parameter of bank collapse. Meanwhile the reservoir bank collapse was predicted by the traditional methods such as"ЕГKaЧУТИH method"and some empirical methods. Study results indicated that the traditional prediction methods could not objectively and correctly predict and evaluate the geology disaster of bank collapse.Based on the scientific research idea of"the general system science"and"the geology process mechanism analyzing and quantitative evaluation", this paper brought forward that the reservoir bank collapse were a long geological course related to multitudinous factors, which was a puzzle with high nonlinear and indeterminacy. This kind of puzzle should be studied by nonlinear methods to insure the research achievements was objective and correct, which was the essential cause that the traditional can not make the prediction objectively to the reservoir bank collapse. A new way to research by nonlinear theory was put forward in this paper to make out the study on prediction for reservoir bank collapse. The nonlinear methods have powerful capabilities in the data mining and knowledge discovery aspects. Based on them, a prediction model with well mapping ability established, the research achievements gained from which can closely approach the target to be studied.The nonlinear prediction of reservoir bank collapse was studied for the first time based on the rough set theory (RS), the BP neural network and extenics methodology in this paper. Firstly, the influencing factors of reservoir bank collapse were disposed with by the rough set theory, which realized the attribute reduction to those influencing factors. Then the weight coefficients of the attributes were figured out by the concept of the significance of attributes. In the process, the significance was normalized to weight coefficients. Therefore, the most simplified and reasonable parameters model can be provided for the prediction studies for reservoir bank collapse based on the BP neural network and extenics methodology. Finally, the nonlinear theories based on the BP neural network and extenics methodology was applied to predict the reservoir bank collapse in Fuling area of the Three Gorges Project.Research achievements indicated that it was feasible to make out the prediction for reservoir bank collapse based on the nonlinear theories, and the prediction results can reflect the stability situation of bank slope after the water storing conditions. Therefore, the new prediction methods by the nonlinear theories can solve the puzzles of practical projects. Furthermore, research achievements can provide scientific reference for the design of prevention and cure engineering of reservoir bank collapse.By means of the deep and system research on reservoir bank collapse in the type of watercourse reservoir-the Three Gorges Project, the primary results and conclusions were gained as following:1. Based on the on the spot in-situ geology survey of the reservoir bank collapse in the area of three Gorges Project in Yangtze River, a great deal of geological data was collected by a group of 6 experts in 90 days. The geological conditions of Reservoir Bank Collapse in Fuling Area of the Three Gorges Reservoir were analyzed and were concluded. Several conditions included in the bank collapse research such as the engineering geological condition, the morphological and structural features of bank slope, the composing materials, the development situation of the collapses in existence, the failure modes of bank collapse, measured correlative parameters of collapse prediction, etc. were all detailedly surveyed. So the full and accurate basic materials and scientific geology basis can be provided for prediction of Reservoir Bank Collapse in Fuling Area of the Three Gorges Project.2. The prediction methods for bank collapse and their applicability have been systematically summarized in this paper. Moreover, the geological condition of bank collapse in the area of the Three Gorges Reservoir was further perfected. It was discovered by research that the existing prediction methods for bank collapse were basically fit for soil bank collapse in plain region, but were not suitable for bank collapse prediction in the watercourse reservoir.3. The prediction of reservoir bank collapse is a puzzle with high nonlinear. It only can be studied reasonably and objectively by nonlinear theories. So a prediction model should be established according to the nonlinear theories to simulate and approach the real stability situation of the bank slope. The traditional methods were applied to make the prediction of reservoir bank collapse at first in the paper, and the prediction results by traditional methods can be consulted and compared in the nonlinear prediction.4. Soil bank slopes and rock bank slopes both exist in reservoir bank slopes in Fuling area. The latter is a great majority. According to the stability situation in existence and the research achievements of experts from home and abroad, the bank slopes are divided into four levels: good stability, relatively good stability, relatively bad stability and bad stability. Taking rock bank slopes in Fuling area as research objects, prediction of reservoir bank collapse under 156m and 175m water level was made out by the new research methods of nonlinear theories.5. For the first time, the reservoir bank collapse was studied by nonlinear prediction models, which were based on RS Rough Set theory, the BP neural network and extenics methodology in this dissertation. Different bank slopes with different Structural feature and modes of bank collapse were predicted under different water levels. This model is transplantable expediently, not only can predict the watercourse bank collapse in the Three Gorges Reservoir Region, but also can predict the watercourse bank collapse in other mountainous areas.6. The influencing factors to rock bank slopes were disposed with by rough set theory, so the achievements from attribute reduction were gained in this paper. In addition, the significance of attributes was figured out at first through the significance concept of attributes that belong to rough set theory. Then the significance was normalized to weight coefficients, which can provide the most simplified and reasonable parameters models for further research on nonlinear prediction of reservoir bank collapse.7. The incidence of the reservoir bank collapse and the prediction research on slope stability of rock bank could only considered 6 and 5 factors separately. The redundant attributes were deleted from the decision information table, the number of the factors after reduction was separately decreased from 11 to 6, and the other was decreased from 13 to 5. It reflects that the rough set theory has powerful capabilities in data mining and knowledge discovery aspects. Theoretical analysis demonstrates that once a reasonable decision information table has been built to make out attributes reduction by using Rough Set theory, the simplest evaluation model can be realized and established.8. The achievements of attributes reduction based on rough set theory, and the Matlab 7.3 toolbox provided a reliable and efficient platform to establish the nonlinear prediction model for reservoir bank collapse based on BP neural network theory. Output and input patterns, the number selection of nodes in hidden layers, weight functions, transfer functions, error functions, training algorithms and const value functions of BP neural networks were probed and studied systemically and detailedly in this paper, which can provide firm theory basis for bank collapse prediction by BP neural networks. On Matlab7.3 platform, the number of input nodes, hidden nodes and output nodes of BP Neural Network can be selected at discretion, so were training algorithms and functions. Research results indicated that when the number of the hidden nodes is about twice than that of the input nodes, the BP neural Networks would have the optimal performance to achieve the rapid convergence effect. Therefore, the globally optimal solution can be gotten.9. By using the gradient descent algorithm with boundary condition,"trainrp", the 6-11-8 model of width prediction for reservoir bank collapse based on BP Neural Network was established, which could make the neural networks have the optimal performance. Moreover, the 5-11-4 model of stability prediction for bank slope based on BP neural network was built too by using Levenberg-Marquardtr back-propagation-algorithm, which will make the BP neural networks have the optimal performance.10. The weight of BP network model was initialized by weight vectors figured out based on rough set theory, which can make the neural networks have optimal performance to achieve the rapid convergence effect, effectively avoiding to immerse to minimum and optimize the networks. The width prediction for reservoir bank collapse and stability prediction both based on BP neural network were analyzed. Considering the field investigation, they were more reliable and objective than the results predicted by the traditional prediction. Moreover, the prediction results were more close to the real situation of bank slope.11. In order to validate nonlinear theory scientifically, extenics methodology, another nonlinear theory, was applied to the prediction research on reservoir bank collapse in this paper. The evaluation indexes system was composed by the properties sets after attributes reduction based on rough set. Weight coefficients calculated by rough set theory were applied to extenics prediction for reservoir bank collapse. Besides research on reservoir bank collapse, extenics evaluation application was programmed to be a visual software in Microsoft VC++6.0 program design language. The newest soft version is v2.3 with high accuracy and expeditiousness by constantly correcting. It is also a very important innovation of the paper. Research indicated that the prediction results of reservoir bank collapse, based on extenics methodology, were more accordable with the real stability situation of bank slopes.12. Compared with the two nonlinear prediction methods for reservoir bank collapse, it can be known that the prediction results by extenics and BP neural network are similar and accordant with the fact. It is validated that the nonlinear theories applied to bank collapse prediction were feasible and reasonable, for they have powerful nonlinear disposing capabilities. Prediction results indicated that it was improbably to predict and evaluate the type of watercourse bank collapse in traditional methods. Furthermore, research achievements of bank collapse prediction could be applied to the protection and cure engineering design for reservoir bank collapse in Fuling area and other reservoir banks according as science reference, which can provide scientific reference and basis.13. According to the accurate prediction results by the nonlinear methods, it is feasible and scientific to apply nonlinear theories to establish the prediction model for reservoir bank collapse. In a word, the ideas and methods of research are innovative points of this dissertation. Therefore, it is worthy of application and popularizing to make prediction for reservoir bank collapse in nonlinear theories. Besides these, it can be known from the research in this paper that the thesis selection and research of this dissertation are scientific and valuable practically.