排序方式: 共有3条查询结果,搜索用时 15 毫秒
1
1.
2.
Particle swarm optimisation (PSO) is an evolutionary metaheuristic inspired by the swarming behaviour observed in flocks of birds. The applications of PSO to solve multi-objective discrete optimisation problems are not widespread. This paper presents a PSO algorithm with negative knowledge (PSONK) to solve multi-objective two-sided mixed-model assembly line balancing problems. Instead of modelling the positions of particles in an absolute manner as in traditional PSO, PSONK employs the knowledge of the relative positions of different particles in generating new solutions. The knowledge of the poor solutions is also utilised to avoid the pairs of adjacent tasks appearing in the poor solutions from being selected as part of new solution strings in the next generation. Much of the effective concept of Pareto optimality is exercised to allow the conflicting objectives to be optimised simultaneously. Experimental results clearly show that PSONK is a competitive and promising algorithm. In addition, when a local search scheme (2-Opt) is embedded into PSONK (called M-PSONK), improved Pareto frontiers (compared to those of PSONK) are attained, but longer computation times are required. 相似文献
3.
This paper presents a simplified high-frequency model for three-phase, twoand three-winding transformers. The model is based on the classical 60 Hz equivalent circuit, extended to high frequencies by the addition of the winding capacitances and the synthesis of the frequency-dependent short-circuit branch by an RLC equivalent network. By retaining the T-form of the classical model, it is possible to separate the frequency-dependent series branch from the constant-valued shunt capacitances. Since the short-circuit branch can be synthesized by a minimum-phase-shift rational approximation, the mathematical complications of fitting mutual impedance or admittance functions are avoided and the model is guaranteed to be numerically absolutely stable. Experimental tests were performed on actual power transformers to determine the parameters of the model. EMTP simulation results are also presented 相似文献
1