Research And Application Of Biogeography-based Optimization

Inspired by the subject of Biogeography, Biogeography-Based Optimization algorithm(BBO) is proposed. It is a bionic optimization algorithm which is derived from Migration patterns and variation regularity of species survival. It is a new evolutionary algorithm based on biological evolution theory and species migration principle. Since proposed, BBO algorithm shows a good optimization performance in a series of benchmark test functions with constraint condition and unconstraint condition, and BBO algorithm has caused broad attention from scholars both at home and abroad. BBO algorithm, like other intelligence optimization algorithms, has some defects, such as jumping into local optimum and poor exploration competence. Therefore, to enhance the computing performance of BBO algorithm, five BBO-based improved algorithms were presented according to two improvement ideas. And five improved algorithms are applied to system identification, TSP, scheduling problem and planting planning problem. Therefore, the main research is the improvement and application research of BBO algorithm, including the following few aspects.(1) For improvement the disadvantages of jumping in local optimum in late iterations, an improved BBO algorithm(IBBO) based on migration operation and variation operation is presented. To improve BBO’s local search capability, BBO combining with simulated annealing(SABBO) is presented. To balance the search ability and exploitation ability of BBO algorithm, BBO with differential evolution(DEBBO) is proposed. For solving poor population diversity of BBO algorithm, hybrid Biogeography-Based Optimization algorithm(HBBO) is proposed. To quick up the convergence rate and precision of BBO algorithm, BBO algorithm based on PSO(PSOBBO) is proposed. Test functions with different characteristics are chosen to test the each improved algorithm. Test results indicate that, both in the global convergence result and rapidity of convergence, the optimization effect of five kinds of improved algorithm was better than all other traditional evolutionary algorithms.(2) To solve system identification problem, the identification problem is changed into parameter optimization problem of continuity interval. Then IBBO and IDEBBO are applied to system identification. Hammerstein model based on IBBO and Wiener model based on IDEBBO are taken as examples to be calculated. By comparison with other evolutionary algorithms, the results show that the algorithm can improve the precision of parameters identification and obtain good identification performance. It also has verified the validity and feasibility of IBBO and IDEBBO.(3) To solve traveling salesman problem and scheduling problem. A discrete SABBO algorithm based on real coding is proposed for path optimization of TSP problem. The typical standard problems are simulated. The results indicate that the suggested algorithm is valid. A real coded PSOBBO algorithm is presented for scheduling optimization scheme of scheduling problem. Two standard JSP problems are tested. And the results demonstrate that the algorithm is viable and valid.(4) To solve the long-standing imbalance supply of raw tomato during tomato sauce season in Xinjiang. By studing the relationship between planting area of seeding time and tomato yield of production period, we can construct nonlinear mathematical model of tomato planting plan. After transferring plan problem into combinational optimization problem, we present a real coding HBBO algorithm for the problem. Simulation is based on the dates provided by a Xinjiang Ketchup factory. The results show that the algorithm can realize the balance between tomato yield and ketchup factory capacity. Experiment results showed effectiveness and rationality of tomato planting plan model and viability of the proposed algorithm. By comparison to other evolutionary algorithms, it reflects the algorithm has good convergence. Since optimization effect of this method to solving large-scale planning is not nice. Therefore, a discrete BBO algorithm is proposed for solving large-scale tomato planting planning problem. The simulation results show that this method can achieve a balanced supply of raw tomato.