Study On Rough Set Attribute Reduction Algorithm Based On QPSO And Analysis Of Basin Monitoring Sites Network

For realizing scientific utilization and development, we should analyze data collected by all kinds of hydrological monitoring sites in basin, in order to understand the inherent law of basin and the relationship of various characteristics. On the one hand, because of the difference of equipment and observation condition of hydrological monitoring sites, there are many rivers without complete data, and there are other rivers which have long and complex data, so the data is hard for research and application use. On the other hand, the location of a hydrological monitoring site is always selected by experience, and it’s possible that the distribution of monitoring sites cannot reflect the basin, or monitoring sites are located too close, which will bring duplicate construction. For this reason, the data of basin has the characteristics of complexity, subjectivity and uncertainty. We need uncertainty analysis method to process the data of basin.The traditional attribute reduction algorithms of Rough Set have kinds of restrictions and insufficiency, so the algorithms cannot be used in reduction of basin data successfully. Based on the attribute reduction algorithms of Rough Set, this paper uses the PSO to improve attribute reduction algorithms, getting a new attribute reduction model, which will be used in reduction of basin data combining the instance. It will provide a workable method for reduction of basin data.The research mainly contains the following aspects.This paper introduces the related concepts of Rough Set and all kinds of attribute reduction algorithms systematically, and analyzes the insufficiency in reduction of basin data of all kinds of attribute reduction algorithms. Then use the PSO to improve the reduction algorithm combined with the characteristics of Rough Set theory.Detail the construction of attribute reduction model from particle coding, fitness function design, particle update and particle code conversion based on quantum PSO. Through examples and standard UCI data sets, we analyze and check the correctness and applicability of the model.Combined with practical data, attribute reduction model is applied to basin data reduction. With the model, seven hydrological monitoring sites are chosen to represent the original fourteen hydrological monitoring sites. Then through the BP neural network prediction, the effectiveness of the proposed results is tested. The reduction result improves the utilization rate of basin data and provides reference on reasonable layout of hydrological monitoring sites.