论文部分内容阅读
针对标准差分进化算法收敛速度慢,容易陷入局部最优从而导致收敛精度不高的缺点,提出将DE/rand/1和DE/best/1线性加权相结合以及自适应重构交叉概率因子的改进差分进化算法。该算法中变异策略采用将DE/rand/1和DE/best/1通过线性模拟退火加权策略相结合,交叉因子则根据进化代数自适应重构,使得算法在初期重视全局搜索能力以找到全局最优可能解,后期重视局部收敛速度,以提高算法寻优能力和收敛速度。最后将该算法和其他改进差分进化算法用于城市供水水处理过程的加药凝絮参数辨识中,仿真结果表明,该算法相对于其他3种算法具有更快的收敛速度和更好的收敛精度,所得模型对检验数据的误差平方和很小,表明该模型准确可靠,为投药过程的前馈反馈控制和水厂的优化运行打下了良好基础,具有很好的实际意义。
Aiming at the shortcomings of the standard differential evolution algorithm, such as slow convergence speed and easy falling into local optimum, which leads to poor convergence accuracy, the paper proposes a combination of DE / rand / 1 and DE / best / 1 linear weighting and the improvement of adaptively reconstructed cross-probabilistic factor Differential evolution algorithm. In this algorithm, DE / rand / 1 and DE / best / 1 are combined by linear simulated annealing weighted strategy. The crossover factor is adaptively reconstructed based on evolutionary algebra, which makes the algorithm focus on the global search ability in the initial stage to find the global optimum You may solve the problem, pay attention to the local convergence speed later, in order to improve the algorithm optimization ability and convergence speed. Finally, this algorithm and other improved differential evolution algorithm are applied to the identification of dosing flocculent parameters in urban water supply process. The simulation results show that the proposed algorithm has faster convergence speed and better convergence accuracy than the other three algorithms , The sum of squares of the error of the obtained model to the test data is very small, which shows that the model is accurate and reliable, which lays a good foundation for the feedforward feedback control of dosing process and the optimized operation of waterworks, which has very good practical significance.