一种新的基于人工神经网络的综合集成算法

New algorithm of meta-synthesis based on artificial neural network

针对现有多目标综合评价中所存在的主观随机性和各种评价方法结果的差异性问题,依据层次分析法(AHP)、主成分分析法(PCA)和人工神经网络(ANN),提出了一种新的从定性到定量转换的综合集成算法,并用于多目标综合评价。该算法基于人工神经网络理论,吸收了层次分析法确定权值的技术和主成分分析法提取主因素的优点,建立了一个新的多目标综合评价模型———AHP PCA ANN模型。介绍了它的组成原理,给出了具体构成方法,描述了各个步骤的主要任务,通过全面质量管理综合评价的实例证明了有效性和可靠性。

Aimed at discrepant problems of every evaluated method's results and subjective randomicity which exist in present multiobjective comprehensive evaluation and based on analytic hierarchy process(AHP), principal component analytic (PCA) and articfical neural network(ANN), a new algorithm of metasynthesis from qualitiative to quantitative is proposed and used to comprehensive evaluation. Based on the theory of artificial neural network, the algotithm absorbs the skill which is used to determine the power by AHP and the excellence which is used to get principal component by PCA. The model of a new multiobjective comprehensive evaluation——AHP-PCA-ANN is proposed. Its constitutional principle and method are given, the main tasks of every step are described.Finally, by concrete example of the total quality management comprehensive evaluation in a certain construction bank of Hunan, its feasibility and effectiveness are proved.