Research On Slope Stability Evaluation System Using Case-based Reasoning

The research of slope stability evaluation is fundamental to the research of Slope Engineering, however, the complexity, fuzziness, inadequacy and uncertainty of Slope Engineering bring difficulty to the slope stability evaluation. Presently there are many various methods about slope stability evaluation, and every kind of method has some hypothesizes and is confined, which brings many inconveniences and limites to apply these methods. Consequently, new methods are needed to introduce. Case-based reasoning (CBR) is, in essence, the use of past experiences as a basis for dealing with novel problems, which has been used successfully in many fields. The slope Engineering depending on engineering practice experiences in deeply sense is similar to CBR. The traditional rule express of knowledge is difficult, but CBR keep away from the bottle neck of rule express of knowledge by recalling past cases. So, CBR is introduced into slope stability evaluation in this paper, which has feasibility and significant meaning.Firstly, this paper discussed the present state of slope stability evaluation research and its tendency, many kinds of slope stability estimation methods and their character. At the same time, this paper discussed the present state, history and work theory of CBR, and also introduced case representation methods, case organization methods and case searching methods. Secondly, this paper had a research about slope instability model classification, influenced factors of slope stability, attributes weights calculating methods and a new data unification method. Thirdly, this paper constructed a slope stability evaluation system using case-based reasoning and put forward a design idea and a design flow about this system, and constructed the frame of this system, and also introduced compositive modules and their function. What's more, this paper brought forward slope case representation, which adapted the forms of radical tables, a record is a slope case. The paper built a slope case base by using relational database and employ database management system to manage and maintain the case base. In addition, a adjusting and learning method was given. Fourthly, this paper discuss several kinds of case searching methods and similarity degree calculating methods. There are two practice methods to case searching, one is calculating similarity degree directly, the other is to reduce the searching range of target case firstly and then in which to calculate the similarity degree and finally give an order to these similarities degree in order to find out the best similar slope sound case. Traditional similarity calculation methods lack flexibleness, so, many newtheories are introduced into in this paper, such as analogy analysis theory, grey-relational theory, BP neural network theory and fuzzy neural network . The several theories are suitable to be used in some fields such as slope engineering which posses many complex and uncertainty factor. This paper gave the several theories a brief introduction and built respective case searching methods combined with the character of slope engineering, and applied them with some practical slope cases and attained some results which are a little different. But the whole effects is not bad, users can make a choice and a synthesis judgement among these results caused by the several case searching methods according to the practical slope, and then instruct engineering practice.Introduce CBR into slope stability evaluation and construct a slope stability evaluation system using case-based reasoning, which is no doubt to provide a new way for slope stability evaluation.The research is sponsored by Huo Yingdong Education Funds and Ministry of Education for cadre in the universities.