Research On Plus-strand RNA Genetic Algorithms

DNA computing is a novel research field. There are many theoretical and practical research reported. By simulating DNA operations and encoding as DNA strands, the complex knowledge can be expressed as DNA molecular and accessed the information toward the optimum in the evolution. However, the DNA genetic algorithm also has weak local searching ability and tends to premature. This dissertation studies on the improvement of genetic algorithm by simulating the replication and recombination process of biological plus-strand RNA and to solve the nonlinear, constrained problems in chemical engineering. The main contents are as follows.(1) Inspired by the DNA genetic algorithm and the biological plus-strand RNA, a novel RNA-GA approach is derived by adding a new operation named "Replication-Recombination" based on plus-strand RNA structure. The replication and recombination process in plus-strand RNA synthesized is favorable to improve the population diversity mainly due to the strict complementary principle. Several test functions and the RBF neural network optimization problem for the soft-sensor modeling of a continuous stirred tank reactor are solved to show the validity of this approach.(2) A hybrid optimization algorithm, plus-strand RNA computing based genetic algorithm with sequential quadratic programming (SQP), is proposed to solve complex nonlinear optimization problems. The comparison results of typical benchmark functions show the efficiency and practicability of the HPRNA-GA. An optimal gasoline blending short-term scheduling solution is given by the HPRNA-GA. The total profit is superior to the RNA genetic algorithm (RNA-GA) and the particle swarm optimization (PSO).(3) Inspired by the mechanism of the biological RNA, a new RNA genetic algorithm for optimal design of multipass heat exchanger networks is proposed. Various genetic operators inspired from the biological DNA are designed to enhance the search ability of the algorithm. To show the reliability of this approach it is used to total annual cost for different heat exchanger network problems taken from open literature. The comparison results show that the proposed approach can produce significant capital cost savings compared with other methods.