Research On Evaluation Method Of The Express Way Socioeconomic Benefit Based On The Neuro-Fuzzy System

The scientific evaluation for express way socioeconomic benefit is an important warrant for project investment decision-making, even a more important measuring way of judging an project operation effective or not. By now, many international and domestic organizations and scholars have carried out relative researches and have obtained good results. However, as to the remaining specific evaluation results, the defects are always existing in the singleness of an evaluating object, or not absolute quantitativity of evaluation analyzing method, or a relatively random index value which results in a untruthful evaluating conclusion. The paper Research on Evaluation Method of the Express Way Socioeconomic Benefit Based on the Neuro-Fuzzy System serves for overcoming the above defects.Firstly, the common methods of building index system are studied. Having analyzing the producing mechanism of express way socioeconomic benefit, a comprehensive evaluation index system has been set up , and calculation method for evaluating index has also been studied.At present, the existing defects and problems in the relative evaluations are mainly caused by the evaluating models which lack of knowledge-concluding ability from data samples and information losing in calculating. Aiming at this, a new evaluation model Roughset-Neuro Fuzzy System (RS-NFS) is proposed with improvement in both the deciding method of member function and extracting method of initial rules. The new model can conclude knowledge, automatically decide single-index evaluation member function's and has simplified the structure of fuzzy system by applying of rough-set theory to extract initial rules. Besides, the model system's parameter-adjusting ability is improved by genetic algorithm with control and over-parameter-adjustment is prevented effectively. In order to verify the model's effectiveness and reliability, the model is verified with samples, and the result shows that RS-NFS model is effective and reliable. Definitely, the application of the new model will effectively solve the conclusion distortion in previous evaluations.In the end, we have developed the experiment system ESNFS1.0 by Python and C++ for the evaluating model and carried real case studies with the system. The results show that RS-NFS has good realistic function.