Coordinate Transformation Method For Reducing The Correlation Of Variables And Its Application In Ant Colony Algorithm

Intelligent swarm algorithm is the cornerstone of artificial swarm,and the correlation be-tween decision variables is an important part of research on intelligent swarm algorithm.An effective method to reduce the correlation between decision variables can enhance the algo-rithm’s global optimization ability and provide tools for the development of intelligent swarm algorithms.The coordinate transformation method(RCTM)is proposed to reduce the correlation be-tween decision variables.The algorithm mainly adopts the idea of coordinate transformation,and it is important to select a matrix as the transformation matrix.Through experimental anal-ysis,RCTM can effectively reduce the overall correlation between variables in the case of di-agonal hexagram.In general case,when the number of solutions is close to dimensions,the probability of reducing the overall correlation is greater than 90%.Ant colony algorithm(ACO)is a typical and widely used intelligent colony algorithm,it is mainly used to solve discrete problems and complex continuous optimization problems.How-ever,ACO_Rhas the defect that it can not jump out of the local optimal solution.RCTM can effectively reduce the overall correlation between variables,therefore,we bring RCTM into ACO_Rand present a new algorithm which is named as ant colony optimization algorithm on the continuous domain based on coordinate transformation(CTACO_R).CTACO_Ris tested by 12standard test functions.Experimental results show that CTACO_Rcan reduce the correlation be-tween decision variables when dealing with high-dimensional functions,besides,it can expand the search domain so that the algorithm can jump out of the local optimal solution and improve the accuracy of optimization results.In particular,when functions are linearly independent,the solution space after coordinate transformation can also be more in line with the feature of variable independence.So,CTACO_Ris more effective when dealing with linearly independent functions.Therefore,CTACO_Rcan optimize ACO_Rfrom the above two aspects.Therefore,it can improve the global search ability and enhance optimization ability.