應用進化規劃多目標演算法於配電系統饋線重構之研究

A Study on the Applications of Multi-Objective Evolutionary Programming on Distribution System Feeder Reconfiguration Problems

許富淵(Fu-Yuan Hsu)

指導教授 ： 蔡孟伸

國立臺北科技大學/機電學院/機電科技研究所/博士(2012年)

https://doi.org/10.6841/NTUT.2012.00701

摘要

配電系統饋線重構問題是一種藉由尋求改變開關狀態的策略來達成改善配電系統運轉性能的最佳化問題，因此已有許多研究提出以軟性計算 （Soft Computing） 的技巧來處理這樣的問題。在這些研究當中，基因演算法 （Genetic Algorithm, GA）或者進化規劃法（Evolutionary Programming, EP）是最常應用於此類最佳化問題的軟性計算方法。這些方法是仿效生物演化方式來搜尋最佳解，在演化過程中是利用交配/突變 （Crossover/Mutation） 等基因運算 （Gene Operations） 來產生新的個體。但是針對輻射狀架構之配電系統，因為具有由拓樸所衍生之等式限制條件 （Equality Constrains in the Topology），所以利用 GA 或 EP 來解決配電系統饋線重構問題時，常因基因運算而產生不符合拓樸限制條件的解。因此針對染色體，在演算法當中必須加入一個檢查與處理的機制。但不合法染色體的處理機制會耗費大量的演算法執行時間 （Processing Time），而導致演算法執行效率降低。另一方面，以往這些有關於配電系統饋線重構的研究，以單一目標的問題作為考量居多，這樣可迅速求得最佳或次佳解。但此方式往往忽略系統運作的實際情資，進而使該解成為不符合現實之無效解。通常配電系統調度人員期望在允許的情況下，重構策略可多方面改善配電系統的運轉能力，換言之，配電系統饋線重構問題應該以多目標 （Multi-Objective）為考量。因此本論文依照配電調度人員對於核心技術的期望，發展適合輻射狀配電系統之多目標饋線重構演算法，演算法核心方法係利用仿效生物演化方式的演算法來求解饋線重構之最佳解。為了要避免基因運算所導致之不合法解的產生，本論文發展一種以“有條件式突變 （Conditional Mutation）”基礎的 EP來達成；另一方面，針對多目標而言，配電系統調度人員可能對於饋線重構解的需求有二：1)多目標整合解；2)最佳化解群。對於此兩個需求，本論文分別發展以 “灰關聯度 （Grey Relational Analysis, GRA）” 配合極差方案 （Inferior Solution） 與極優方案 （Superior Solution） 之方法，以及引用 “非支配排序基因演算法-II （Non-dominated Sorting Genetic Algorithm-II, NSGA-II）” 之適應值計算方法來結合本論文所提出的 EP 發展出 EP-GCRA 與 NSEP 來達成。

關鍵字

輻射狀配電系統 ； 配電系統饋線重構 ； 多目標最佳化問題 ； 軟性計算 ； 進化規劃法 ； 交配/突變 ； 有條件突變 ； 灰關聯分析法 ； 極差方案與極優方案 ； 非支配排序

並列摘要

The feeder reconfiguration problem of power distribution system is an optimization problem which can achieve the improvement of the operational performance on the power distribution system by searching a strategy of changing the switches states. Many studies were proposed by utilizing the soft computing techniques to solve this problem. In these studies, Genetic Algorithm (GA) and Evolutionary Programming (EP) are two popular methods that have been applied for this optimization problem. The operators used in GA or EP emulate the biological evaluation to solve the optimization problems. In the traditional evolutionary process, the new individual can be generated by the gene operations, such as, crossover and mutation. However, for the problems that deal with radial power distribution systems, cares must be taken due to its topological constraint. Thus, when the GA or EP was applied to solve the feeder reconfiguration problem of power distribution system, the solutions that do not satisfy the topological constraints of the problem may be generated by genetic operations. A mechanism of chromosome validation process must be used in these algorithms before calculating the fitness values. However, the validation process takes the largest portion of the processing time. As a result, the performance of these algorithms is reduced. On the other hand, most of previous studies only considered single objective when deal with the feeder reconfiguration problems of power distribution systems. Nevertheless, this consideration ignores the some information in the real system. In reality, the power distribution system dispatchers expect that the different proposed strategies can be chosen based on their experience and conditions. Therefore, the feeder reconfiguration problem of power distribution system must take the multi-objective into consideration. A multi-objective feeder reconfiguration algorithm for redial power distribution system is developed in this dissertation. The emulation of biological evaluation algorithm is applied to solve the multi-objective feeder reconfiguration problems. In order to avoid the illegals be generated by the genetic operations, an Evolutionary Programming based on the “Conditional Mutation” is proposed in this dissertation. When solving the multi-objective problems, two approaches can be applied: 1) Integrating multi-objective into single-objective. 2) Identifying all solutions by considering all objectives. In this dissertation, the EP-GCRA and the NSEP are proposed respectively. The EP-GCRA algorithm applies the grey relational analysis by proposing the concept of inferior and superior solutions that integrate all the objectives for the multi-objective problems. For the second approach, the NSEP algorithm that utilizes the fitness value calculation used in the NSGA-II and the special “Conditional Mutation” operator proposed in this dissertation are developed to identify the optimal solution set. The EP approach is applied for these two algorithms. The results show that the application of EP performs better than traditional GA approaches.

並列關鍵字

Radial Power Distribution System ； Multi-Objective Optimization Problem ； Soft Computing ； Evolutionary Programming (EP) ； Crossover/Mutation ； Conditional Mutation ； Grey Relational Analysis (GRA) ； Inferior Solution ； Superior Solution ； Pareto Non-dominated Sorting ； NSGA-II ； EP-GCRA ； NSEP

參考文獻

[68] K. L. Wen, T. C. Chang and J. H. Wu, “Data preprocessing on grey relational analysis,” Journals on Grey Systems, Vol. 11, No. 1, 1999, pp. 139-141.

[1] S. Toune, H. Fudo, T. Genji, Y. Fukuyama and Y. Nakanishi, “Comparative study of modern heuristic algorithms to service restoration in distribution systems,” IEEE Transactions on Power Delivery, Vol. 17, No. 1, January 2002, pp. 173-181.

[2] M. Paar and P. Toman, “Utilization of particle swarm optimization algorithm for optimization of MV network compensation,” Power Tech, 2007 IEEE Lausanne, Switzerland, July 1-5 2007, pp. 1991-1995.

[3] Th. Back, G. Rudolph and H. P. Schwefel, “Evolutionary programming and evolution strategies: Similarities and differences,” In Proceedings of the Second Annual Conference on Evolutionary Programming, La Jolla, CA, 1993, pp. 11-22.

[5] W. H. Chen and M. S. Tsai, “A novel approach to multi-objective network reconfiguration,” in Proc. Int. Conf. Advanced Power System Automation and Protection, Jeju, Korea, October 25-28 2004, pp. 503-506.