输电网规划方案的多层级组合模糊综合评价方法研究

发布时间：2018-11-05 20:54

【摘要】：面对众多规划方案,专家和决策者的主观评价缺乏客观依据,传统技术经济最优不够全面,亟需研究能够有效处理多准则约束的综合评价理论,而传统综合评价理论在指标体系、评价方法、指标权重、模糊处理以及模型求解等方面存在不足,本文基于此进行了深入探究,概括如下:首先,本文在研究工作中,考虑输电网规划工作的实际特点,结合项目背景,以广东电网《电网规划指标体系研究报告》为基础,搭建一套合适的综合评价指标体系,该体系含6个战略层指标、16个目标层指标和60个操作层指标,在此基础上详细分析各底层指标的意义、计算方法、所需数据、评价判据和评分标准等。其次,本文阐述了现有主、客观权重求取方法的优缺点,结合数理统计和线性空间,通过在超空间构建优化数学模型,基于组合权重向量与原有主、客观权重向量偏差最小的思想提出一套新的组合权重求取方法,该方法能有效融合专家经验和数据信息,实现指标合理赋权,并用“平均贴近度”检验组合权重向量的合理性。然后,考虑到规划方案存在层次性和模糊性,本文提出多层级组合模糊综合评价方法:构建“模糊关系矩阵”到“决策数据矩阵”的转换规则,实现合理赋权;基于组合模糊隶属度子集和各单一模糊隶属度子集偏差最小思想构建模糊关系矩阵;考虑最大隶属度原则有效度,通过加权算术平均型合成算子,实现对评价结果的合理分析。再是,本文对粒子群算法(IM_PSO)做了改进,并以此求解组合权重和组合模糊隶属度子集的优化数学模型。本文通过提出新的自适应惯性权重、线性变化学习因子、邻近粒子信息、混沌差分初始化以及高斯变异算子,使改进粒子群算法(IM_PSO),无论是寻优精度,还是跳出局部极值,抑或算法稳定性,均有大幅度的性能提升。最后,本文通过对某省级三个输电网规划方案的实证研究,证明本文所提算法是合理有效的,并对全文工作进行总结和展望。
[Abstract]:In the face of many planning schemes, the subjective evaluation of experts and decision makers is lack of objective basis, the traditional technical and economic optimization is not comprehensive enough, so it is urgent to study the comprehensive evaluation theory which can deal with multi-criteria constraints effectively. However, the traditional comprehensive evaluation theory has shortcomings in index system, evaluation method, index weight, fuzzy processing and model solving. Based on this, this paper makes a deep exploration, which is summarized as follows: first, in the research work, Considering the actual characteristics of transmission network planning and combining with project background, a suitable comprehensive evaluation index system is set up on the basis of Guangdong Power Grid "Research report on Power Network Planning Index system". The system consists of 6 strategic layer indexes. On the basis of 16 target layer indexes and 60 operational layer indexes, the significance, calculation method, needed data, evaluation criteria and scoring criteria of each underlying index are analyzed in detail. Secondly, this paper describes the advantages and disadvantages of the existing subjective and objective weight calculation methods, combined with mathematical statistics and linear space, through the construction of an optimal mathematical model in hyperspace, based on the combination of weight vector and the original main, The idea of minimum deviation of objective weight vector is put forward in this paper. This method can effectively combine expert experience and data information, realize reasonable weight of index, and test the rationality of combination weight vector with "average closeness degree". Then, considering the hierarchy and fuzziness of the planning scheme, this paper puts forward a multi-level combined fuzzy comprehensive evaluation method: constructing the conversion rule from "fuzzy relation matrix" to "decision data matrix" to realize reasonable weight; The fuzzy relation matrix is constructed based on the idea of minimum deviation between the combined fuzzy membership subset and the single fuzzy membership subset, and considering the validity of the maximum membership principle, the reasonable analysis of the evaluation results is realized through the weighted arithmetic average type composition operator. Secondly, the particle swarm optimization algorithm (IM_PSO) is improved to solve the optimal mathematical model of combination weight and combined fuzzy membership subset. In this paper, a new adaptive inertial weight, linear variable learning factor, adjacent particle information, chaotic differential initialization and Gao Si mutation operator are proposed to improve the particle swarm optimization (IM_PSO) accuracy. Still jump out of the local extremum, or the stability of the algorithm, there is a significant improvement in performance. Finally, this paper proves that the proposed algorithm is reasonable and effective through the empirical study of three provincial transmission network planning schemes, and summarizes and prospects the full text of the work.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM715

【相似文献】