Research Of Constrained Optimization Problems Based On Artificial Bee Colony Algorithm

In reality, many problems can be transformed into a constrained optimization problem to deal with, but solving this problem is often the focus and difficult point the work. Artificial bee colony algorithm puts forward a new idea to solve the constrained optimization problems. Like other swarm intelligence algorithm, it has some advantages such as parallel, robustness and strong commonality.The quality of the initial honey affects the performance of the ABC algorithm greatly. Even many of the researches of the ABC algorithm requires it to be feasible. But the traditional random search is difficult to obtain satisfactory results, especially for the strong constraint optimization problems which accounts for a very small proportion in the feasible region. In order to improve the quality of the initial stage and the enhance the optimum efficiency of the ABC algorithm, we establish RCGABC algorithm model in this article. It can select ratio coefficient flexibly based on the characteristics of the problems, and then generating high quality initial feasible honey that we need. Firstly, we use the Monte Carlo method to simulate the proportion of the feasible region of the problem in the search field approximately, and then set the control parameters. Secondly, we use the improved ABC algorithm to generate a feasible honey set Si. Thirdly, according to the random compression search algorithm that we established in this article, we expand the source set S1 to be the feasible source set S2. Lastly, We use clustering algorithm to classify the feasible source set S2. And then according to the distance of the source, we can select the optimal nectar through the greedy criterion. In order to verify the efficiency of the algorithm, we select two complex test functions in this article. For the calculation of the time and the diversity of nectar, we use Random method, ABC algorithm, RCGABC algorithm and stochastic compression method for matlab programming experiment respectively. Through comparing the analysis of the results, we can prove high computational efficiency of the RCGABC algorithm we proposed.To the constrained optimization problem of continuous function, we establish the improved hybrid artificial bee colony algorithm to solve the model. Firstly, we use the RCGABC algorithm to generate the initial feasible nectar set. Secondly, we establish the adaptive neighborhood search method for the gather honey bee and observation bee, which is as follows:The mproved neighborhood search operation add the global optimal nectar guidelines (XgiJ-XiJ) to the original neighborhood search formula. Meanwhile, we design the adaptive acceleration factor Ψ according to the structure characteristics of the honey source set and design the adaptive scaling factor φ based on the evolution algebra of the algorithm. And then we can control the searching step of the neighborhood nectar and the optimal nectar. Thirdly, we use the bee colony selection mechanism which is established based on the simulated annealing strategy, to enhance the developing abilities of the algorithm, and to accelerate the convergence speed of the algorithm. Lastly, we set up repairing operator based on the LCSA and RCSA algorithm to ensure the feasibility and the diversity of honey in the iterative process, and then the search efficiency of the algorithm can be improved. In order to verify the efficiency of this algorithm, we select 7 different types of the test functions to carry on programming experiments. And then we compare the results that generated from the program repeatedly with the penalty function ABC algorithm and penalty function GA algorithm. Thus we can verify the quality of the optimal solution and the convergence speed of the algorithm in this paper is all the best.For the combined constrained optimization problem, we build NDABC algorithm to get the model in the paper. Firstly, we use the way serial number encodes to express variable set of the optimization problem, and then generate full permutation of the serial number set as the initial source algorithm. According to the characteristics of the problem, we generate the corresponding actual combination by using greedy algorithm and tabu search strategy. Then we establish the adaptive neighborhood searching method for the Serial number coded honey source. According to the adaptive serial number operation number that set by the iterative process of the algorithm and the characteristics of the problem, we can select the corresponding method. The adaptive operation is shown in the following type:In the formula, cycle is the current evolution algebra and Maxcycle is the maximum evolution algebra. And MEmrise is the maximum number of operation growth, MEmin is the minimum position operand. Adaptive operation ensures the algorithm to have stronger development ability in the early stage and better local search ability in the late stage. To verify the solving efficiency of the NDABC algorithm for combined optimization problem, we choose the fuel transportation problem, knapsack problem and flight landing scheduling problem. And based on their characteristics, we establish suitable encoding and decoding operator and adaptive neighborhood search operator. Then we establish fine tuned operator for the flight landing scheduling model and use matlab programming to solve it. And then we compare the result with other documents. We can prove the calculation efficiency of the NDABC algorithm built in this article for combined optimization problem is really high.