162 results about "Combinatorial optimization problem" patented technology

The invention discloses a thermal process model parameter identification method through an improved hybrid particle swarm algorithm. The method comprises the following steps: 1) determining an identification system structure and parameters to be identified; 2) obtaining input / output data for identification; and 3) carrying out the improved hybrid particle swarm algorithm to obtain an optimal solution. The identification problem of a thermal process model is converted into the combinatorial optimization problem of parameters; effective searching is carried out on a parameter space through the improved hybrid particle swarm algorithm to obtain optimal estimation of system model parameters; compared with a basic particle group algorithm, the method introduces selection, hybridization and mutation mechanisms in a genetic algorithm, thereby keeping population diversity and preventing the algorithm from being trapped in the local optimal solution; the idea of vaccine extraction and vaccination in artificial immunity is introduced, so hat algorithm search speed is improved; improved adaptive mutation is adopted, so that diversity of particles is kept more reasonably; and through introduction of a simulated annealing idea, the method has probabilistic leap capability in the searching process and prevents the searching process from being trapped in the local optimal solution.