Research On Application Of Improved Hybrid Rice Optimization Algorithm In Multiobjective Clustering

With the rapid development of modern technology in the world,the ability to analyze data in the era of big data has gradually increased.Clustering,often referred to as unsupervised classification,is a widely used and common data analysis technique.In recent years,the clustering data analysis method based on the optimization of the objective function value in the intelligent optimization algorithm has become a research hotspot.Among them,the multi-objective optimization clustering technology,which has the advantages of overcoming the limitation of clustering and getting rid of the sensitivity to the initial class center,has been applied to the fields of image segmentation,remote sensing image and genetic information engineering.However,there are problems such as complex operation structure,insufficient uniformity of solution set distribution and low operating efficiency in multi-objective optimization theory,so its related theory needs to be further explored.The hybrid rice optimization algorithm that inherently has a sorting breeding mechanism and individual diversity is a heuristic algorithm that simulates three-line hybrid rice breeding methods.It has a simple structure and is easy to implement,making it a potential multi-objective optimization algorithm with good performance.In this thesis,a multi-objective Hybrid Rice Optimization Algorithm is presented,which is simple in structure and efficient in operation.The main work of this thesis is as follows:1.In this thesis,a Hybrid Rice Optimization Algorithm with improved partitioning mechanism is designed.The greedy strategy adopted by the basic hybrid rice optimization algorithm at the beginning of the design may cause its convergence to be too fast.When its structural characteristics and the shortcomings of the greedy strategy in the three-line division are deeply studied and explored,the breeding division mechanism is improved.The improved method in this paper can alleviate the contradiction between convergence and diversity when the original algorithm is optimized,and the comprehensive ability of the algorithm in global and local search will be improved.By conducting experiments on the CEC2015 benchmark test function,its improvement in function optimization performance can be demonstrated.2.The above improved algorithm has been expanded into a multi-objective hybrid rice optimization algorithm.The implementation principle of the algorithm mainly uses the non-dominated hierarchical sorting search Pareto solution and the sorting lineage characteristics of the hybrid rice optimization algorithm,and its excellent multi-objective optimization ability has been confirmed by the common multi-objective benchmark function test.3.The clustering method based on the multi-objective hybrid rice optimization algorithm is implemented.There are many types of clustering effectiveness indicators,and only two specific evaluation indicators are selected for multi-objective optimization.Combined with the multi-objective equilibrium strategy,the limitation of the evolutionary clustering method with a single evaluation index as the objective function is alleviated,so that the accuracy of the clustering method based on the intelligent optimization algorithm is improved.In summary,the main research content of this thesis is the clustering method based on the improved hybrid rice optimization algorithm.The clustering performance of the algorithm is tested through 6 sets of UCI public data sets,and it is also applied to real remote sensing image clustering to explore the Applicability of the algorithm in the field of remote sensing images.The experimental results show that the clustering method based on the improved hybrid rice optimization algorithm has the ability to avoid complex traditional clustering design forms and local optimal sensitivity issues.Clustering method based on multi-objective hybrid rice optimization algorithm has higher operating efficiency and higher clustering accuracy,and has broad application prospects in the field of remote sensing images.