A CSDM-Oriented Approach To Weight Determination For Analytic Hierarchy Process And Its Application Study

According to the principal of Analytic Hierarchy Process (AHP), the weight ranking will be given by using the method. So constructing judgment matrices (JM) and computing weights of alternatives is important. Since the method of AHP was introduced, the JMs constructed by experts always can not meet the test of logical consistency is a problem of AHP. The problem is a direct influence on determining weights of alternatives. According to the principal of classical AHP, the JM that does not pass the test should be rectified till meets the test. The logically inconsistent problem of judgments is one of interest when scholars do research on AHP. Although scholars presented their many methods, the paper deems those methods do not solve the inconsistent judgments of JM at all. Scores of years ago, the research on the theory of complex system attracted many scholars'attention, and lots of fruits were developed. What is more important, the fruits brought us new angles of view and approaches that can be used to tackle the problem. In consequence, based on the theory of complex system, the paper discusses the essential reasons that engender the logically inconsistent problem. Then, it is to explore a complex-system-decision-making (CSDM) oriented approach to weight determination for analytic hierarchy process.Based on the complex system theory of meta-synthesis from qualitative analysis to quantitative analysis, the paper deems the classical AHP has two drawbacks in constructing and processing JMs, and the drawbacks are the main reasons that cause the logical inconsistency. The first one is that the JM constructed by the reciprocal relation is incapable of reflecting more information of experts'subjective evaluations. The second is that the weight determination approach does not solve the objectively existed problem of background changing. Therefore, evaluations made by the classical AHP and current modified AHP approaches are probably incorrect. To overcome these two drawbacks, the new concept of entire judgment matrix (EJM) is presented, according to the complex system theory of meta-synthesis from qualitative analysis to quantitative analysis. Based on the technique of artificial neural network (ANN) and the back-propagation (BP) algorithm, an EJM-based approach to determine weights of alternatives on AHP index hierarchy is presented to obtain weights with respect to each of index. The experimental simulation analysis verifies that the given approach is able to obtain ranking weights more approximate to the objective, and proves to be more reasonable than the current AHP approaches.For computing the weight with respect to each of the criteria, the method of...