Research On Principle And Engineering Applications Of Improved Flower Pollination Optimization Algorithm

Optimization problems can be widely found in diverse fields,such as scientific,social,economic,engineering,and so on.Therefore,the researches on efficient optimization methods have great theoretical and practical significance.In recent decades,meta-heuristic algorithms have increased widely interests and become an important branch of optimization methodology.Flower pollination algorithm is a nascent meta-heuristic algorithm which is inspired by the pollination behaviour of angiosperm.At present,the research on this algorithm is still in its infancy and not thorough enough.The research results are still very few,scattered and lack of systematicness.This thesis focuses on the performance improvements and engineering applications of the flower pollination algorithm,in order to develop its theories and to extend its applications.The main research work and contributions of this paper can be summarized as follows:Firstly,to improve its performance for solving unconstrained optimization problems,an orthogonal flower pollination algorithm is proposed by embedding the quantization orthogonal crossover operator combined with the range analysis method into the frame of the basic flower pollination algorithm.Numerical experiments tested on typical benchmark functions show the proposed algorithm has better performance than the basic algorithm.Secondly,parameter tuning methods based on flower pollination algorithm for three types of controllers,i.e.classical PID controller,PID-based cascade controller,and fractional order PID controller,are investigated.For classical PID controller,a parameter tuning method based on the orthogonal flower pollination algorithm and performance index in time domain is proposed and examined using five different industrial process control models.Tests results suggest that the proposed method is better than the traditional Z-N tuning method and the improved particle swarm optimization taken from literature.For PID-based cascade controller,a tuning method which simultaneously tunes the parameters of the inner and outer loop controllers based on the orthogonal flower pollination algorithm is proposed.Simulation studies on three control processes with different characteristics such as stable,unstable,and integrating processes,and comparisons with the basic flower pollination algorithm and the genetic algorithm demonstrate the effectiveness and superiority of the proposed method.Due to its simplicity,effectiveness,and universality,the proposed technique may serve as an alternative for the design of cascade control systems.For fractional order PID controller,a tuning method using the orthogonal flower pollination algorithm based on the Bode's reference model is proposed.During the controller design procedure,the IAE between the step responses of the designed system and the reference model is used as the minimization goal and the orthogonal flower pollination algorithm plays the role of an optimizer.Tests on two examples suggest that the orthogonal flower pollination algorithm is effective and performs better than the real code genetic algorithm,the artificial bee algorithm,the teaching learning based algorithm,and the basic flower pollination algorithm on the accuracy of solution,convergence speed,and stability.Thirdly,parameter estimation methods based on flower pollination algorithm for two types of systems,i.e.chaotic systems and photovoltaic(PV)systems are explored.For the former,a hybrid flower pollination algorithm which well combines the good global exploration ability of the original flower pollination algorithm and the powerful local exploitation ability of the Nelder-Mead simplex method together by integrating these two methods in a very simple way is proposed.The experimental results tested on three typical chaotic system parameter estimation problems with three unknown parameters,including the Lorenz system,the Rossler system,and the Lorenz system under the noise condition demonstrate that the algorithm can estimate the unknown parameters efficiently and accurately.Meanwhile,compared with some algorithms from literature,the proposed algorithm is energy efficient and superior.For the latter,a hybrid flower pollination algorithm which combines the basic flower pollination algorithm with the Nelder-Mead simplex method and the generalized opposition-based learning mechanism is proposed.The experiments based on the I-V data of a solar cell and a PV module taken from literature clearly demonstrate the effectiveness of this algorithm.The comparisons with some other published methods demonstrate that the proposed algorithm is superior than most reported algorithms in terms of the accuracy of final solutions,convergence speed,and stability.Furthermore,the tests on three PV modules of different types(Multi-crystalline,Thin-film,and Mono-crystalline)suggest that the proposed algorithm can obtain superior results at different irradiances and temperatures.The proposed algorithm can serve as a new alternative for parameter estimation of solar cells/PV modules.Next,constrained optimization methods based on flower pollination algorithm are researched.Two widely-used and effective constraint handling methods,i.e.the Deb's heuristic rules and the ?-constraint method are respectively combined with the basic flower pollination algorithm and then tested on 13 benchmark functions.However,the performances of both these two algorithms are shown to be not sufficient.Meanwhile,when being combined with flower pollination algorithm,Deb's heuristic rules and the?-constraint method are respectively suitable for problems without equality constraints and with equality constraints.Then,a hybrid flower pollination algorithm for solving constrained optimization problems is proposed.In the proposed algorithm,firstly a population initialization method based on good point set and the generalized opposition-based learning mechanism are introduced into the basic algorithm to enhance its search performance,and then the Deb's heuristic rules and the ?-constraint method are combined together to treat the constraints.Finally,the performance of the proposed algorithm is examined on the benchmark functions and some typical practical engineering constrained problems including the structural design problems of the tension/compression spring,the welded beam,the pressure vessel,the speed reducer,and the parameters optimization problem of multi-pass turning.The numerical experiments suggest that the proposed algorithm has a good performance.Finally,the application of flower pollination algorithm on the solving of the permutation flow shop scheduling problems(PFSPs)with the objective to minimize the makespan which belongs to the combinatorial optimization field is explored.A hybrid flower pollination algorithm is proposed for solving this problem.This algorithm utilizes a largest-order-value(LOV)rule based random key encoding scheme to convert the continuous solutions in flower pollination algorithm into a discrete job permutation,utilizes the Nawaz-Enscore-Ham(NEH)heuristic method to improve the quality of the initial population,and utilizes a local search strategy based on the insertion neighborhood structure to enhance its exploitation ability.Experiments tested on two famous benchmark suits and comparisons with some other advanced algorithms from literature suggest that the proposed algorithm is effective and competitive thus it can be used as a new alternative for solving PFSPs.Overall,this paper carried on some positive explorations for the performance improvements and engineering applications of the flower pollination algorithm,and provides some new and effective methods for the solving of some practical engineering problems.Therefore,the work of this paper has great theoretical and practical values.