A SOLUTION TO THE DILEMMA PROBLEM IN THE VALUATION OF ENGINEERING GEOLOGICAL ENVIRONMENTAL QUALITY

In the valuation of engineering geological environmental quality, there is the weight dilemma problem in the traditional weighted composite index method, fuzzy comprehensive evaluation method, grey clustering method and etc., however, there is the dilemma problem of selecting training samples and testing samples in the artificial neural network method in this paper, sensitive unit, insensitive unit and similar-insensitive unit are firstly put forward according to the classification of the units of an engineering geological environmental system. An insensitive unit is defined as the unit whose quality category is constant while the maximum ratios of factor weights change from K=1～K=.Based on the concept of insensitive unit, the standard training set is defined as the input-output pairs of insensitive units. The paper has proposed a new artificial neural network method based on standard training set, trying to solve the weight dilemma problem in the traditional evaluation method and the dilemma problem of selecting training sample and testing sample in the artificial neural network method.