Genetic Aided Hydro-Thermal Generation Scheduling

• Main Authors: Chen Po-Hung, 陳柏宏
• Other Authors: Chang Hong-Chan
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/11465516106761172287

碩士 === 國立臺灣科技大學 === 電機工程技術研究所 === 86 === This dissertation presents novel solution algorithms andresults based on the genetic algorithm(GA) for solving thehydro-thermal generation scheduling(HTGS) problem. The culmin-ation of this research is a complete and efficient HTGS softw-are package for system operation planning needs. The objective of hydro-thermal generation scheduling is tominimize the total fuel costs of apower system while satisfy-ing various physical and operational constrains. Basically,hydro-thermal generation scheduling is a non-linear programm-ing problem and is difficult to solve by a mathematical prog-ramming method. Previous efforts have been to simplify the fo-rmulations to make the problem solvable with optimization alg-orithms which belong to the class of greedy search algorithms.However, these solution algorithms usually got stuck at a loc-al optimun rather than at the global optimun. In this research, we apply the genetic algorithm which is a general-purposeglobal optimization technique to the HTGS problem. One of theadvantages of the new proposed approach is the use of stochas-tic operators instead of deterministic rules, thus allowing itto escape from a local optimun, to obtain the global optimun. The HTGS problem involves economic dispatch, thermal unitcommitment, and hydroelectic scheduling as subproblems. In theenconomic dispatch subproblem, a salient feature oh the propo-sed approach is that the solution time grows approximately li-nearly with problem size other than exponentially. This featu-re is particulary attractive in large-scale problems. In thethermal unit commitment subproblem, the difficult minimaluptime/downtime constrains are embedded and satisfied through-out the proposed encoding and decoding algorithms. Therefore,the global optimun of the problem can be approached withrather high probability. In the hydroelectric scheduling subp-roblem, complete solution algorithms and encoding/decoding te-chniques are proposed for solving different types of hydro pl-ants involving hydraulically independent plants, hydraulicall-y coupled plants, and pump-storage plants. In the hydro-thermal generation scheduling problem, sched-uling of hydraulically coupled plants has been reckoned as oneof the most difficult parts. Both electric and hydraulic coup-lings create a multi-dimensional, non-linear programming prob-lem. Conventional solution methods have used successive appro-ximation technique to make the problem solvable by decomposingthe multi-dimensional problem into several one-dimensional pr-oblems. However, this technique is not very efficient and notflexible enough. In view of this, this dissertation presentsan efficient approach to the hydraulically coupled plants sch-eduling problem. In the proposed GA approach, the hydraulical-ly coupled plants which are located on the same river are sol-ved concurrently. The most difficult water balance constrainscaused by hydraulic coupling are automatically satisfied. finally, representative test examples based on the actualTaipower system are presented and analyzed to illustrate thecapability of the proposed approach in practical applications.Test results show the attractive properties of the GA approac-h, which is a highly optimal solution and more robust converg-ence behavior.