【毕业论文】多目标期望控制型式负荷分配问题的动态加权优化方法.doc

哈尔滨工业大学本科毕业设计（论文） 哈尔滨工业大学本科毕业设计（论文） - PAGE \* MERGEFORMAT IV- 摘 要 节能减排计划下的电力系统负荷分配问题，是一个需要同时考虑发电企业发电成本、污染治理成本和电网公司购电成本的多目标优化问题，难以应用经典数学算法来解决。 针对此问题，本文首先对节能减排下负荷分配问题建立多目标期望控制模型，并对其进行简化，即只考虑单时段以及网损率为定值。提出一种动态加权的优化算法，将多目标模型转化为单目标优化问题。此种算法在每次迭代过程中能够自动根据期望值偏差率调整相对权值，在一定程度上避免了人为赋值的主观弊端，加速算法的收敛。对于加权后的单目标优化问题，本文采用了一阶梯度法和二阶梯度法配合寻优，方法简单，不需要太多的数据支持，便于理解与实现。 针对各算法进行了详细的设计，并编制相应的计算机程序。算例验证了多目标期望控制思想的正确性，同时也发现程序的适应性不足，存在修正不完全的缺陷，可通过选择适应性更好的算法加以改善。 关键词：节能减排；负荷分配；多目标；期望控制；动态加权 Abstract The load distribution under energy conservation and emission reduction program should take generation cost, pollution-control cost and purchase cost of grid companies into account, which is a multi-objective optimization problem. It is difficult to find the global optimum to solve this kind of problems. Aiming at the problem, this paper first builds a multi-objective expected control model of the load distribution under energy conservation and emission, and simplifies it, which considers only in single-period and let the network loss rate be constant. A dynamic weighting algorithm is proposed to transform the multi-objective model into single objective optimization problem. The algorithm can automatically adjust relative weights according to deviation rate to the expected values. It can avoid the drawbacks of human subjective assignment and accelerate the convergence to some extent. For the weighted single objective optimization problem, in this paper, the first-order gradient algorithm and the second-order gardient algorithm used cooperatively, is simple and doesn’t require too much data to support, easy to understand and realize. It designs the algorithms in detail and edits programs in the paper. Numerical examples have demonstrated the correctness of the multi-objective expected control thought, but also found lack of adaptability of the program. Although there is a case that can’t modify completely, it can be improved by replacing the gardient algorithms of some adaptive algorithms. Keywo