Pattern Recognition And Comprehensive Evaluation Method Based On Interval Numbers

Pattern recognition method and comprehensive evaluation method have been widely used together. But in the practice, some index values are of uncertainty because of the ambiguities of some characteristic indexes and the measurement error of the index values. The interval number is a simple and effective method to describe this kind of uncertainty. Therefore, this dissertation investigates the issues related to pattern recognition and comprehensive evaluation when their characteristic indexes or the acquired data is interval number.After analyzing the existed distance definition of the interval number, a new distance measure is given based on the midpoint and width of the interval number and the corresponding effect factors, i.e. taking into account the information uncertainty caused by the midpoint and width of the interval number and the extent of their effects on real applications. Based on the definition, its rationality has been proved with some related properties obtained. On this basis, the closeness principles are established from which a pattern recognition method is derived based on the interval number. Finally, this pattern recognition method is used in the analysis of the electromagnetic environment during examinations. Its feasibility and effectiveness of the method are illustrated.In comprehensive evaluation problems, the evaluation results are closely related to the mentality of the judges. The same judge may have a different mentality in different environments. Therefore, the mentality of the judge must be considered during the comprehensive evaluation. The mentality index and the subsequent reordering of the interval number are used for the comprehensive evaluation. Together with the interval number, a comprehensive evaluation method is proposed. An example is given to show the practicability of this method and analyze the dynamic comprehensive evaluation results of the judge in the process.