Optimal wind-thermal generating unit(风火电机组协调优化策略)

NOMENCLATURE Additional up reserve requirement considering wind power generation. Additional down reserve requirement considering wind power generation. Number of states saved at each hour in the HDP algorithm. Percentage of maximum unit capacity. Maximum ramp-down rate for thermal unit i. Maximum down reserve contribution of thermal unit i. Down reserve contribution of thermal unit i at hour t. Operation cost function of thermal unit i. Total operation cost over the scheduling horizon. Index for thermal units. Index for wind units. Number of thermal units in the system. Number of wind units in the system. Generation of thermal unit i at hour t. Upper generation limit of thermal unit i. Maximum generation of thermal unit i at hour t. Lower generation limit of thermal unit i. Minimum generation of thermal unit i at hour t. System load demand at hour t.
T
Manuscript received November 15, 2005; revised October 13, 2006. Paper no. TEC-00386-2005. The author is with the Department of Electricalቤተ መጻሕፍቲ ባይዱEngineering, Ming Hsin University of Science and Technology, Hsin-Chu 30401, Taiwan, R.O.C. (e-mail: cclung@.tw). Digital Object Identifier 10.1109/TEC.2007.914188
USRB v (t) vI j vO j vR j α%, β % ε ε∗ ϕj (•)
Upper generation limit of wind unit j. Actual generation of wind unit j at hour t. Available generation of wind unit j at hour t. Total actual wind generation at hour t. Total available wind generation at hour t. Coefficient of additional up (or down) reserve requirement (linear model). Startup ramp rate limit of thermal unit i. Number of time intervals (hours). System ramping down capacity at hour t. Time period that thermal unit i had been continuously down till period t. Minimum down time of thermal unit i. Minimum up time of thermal unit i. Time period that thermal unit i had been continuously up till period t. System ramping up capacity at hour t. Scheduled state of thermal unit i for hour t (1: unit i is up, 0: unit i is down). Maximum ramp-up rate for thermal unit i. Up reserve contribution of thermal unit i at hour t. Maximum up reserve contribution of thermal unit i. System up spinning reserve requirement not considering wind power generation. Wind speed at hour t. Cut-in wind speed of wind unit j. Cut-out wind speed of wind unit j. Rated wind speed of wind unit j. Coefficients of additional up (or down) reserve requirement (second-order model). Predetermined resolution in the DSM algorithm (refined economic dispatch). Predetermined resolution in the DSM algorithm (approximate economic dispatch). Wind power curve of wind unit j. I. INTRODUCTION
0885-8969/$25.00 © 2008 IEEE
274
IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 23, NO. 1, MARCH 2008
been reduced rapidly, installation of WTGs as fuel savers is economically and environmentally attractive in windy regions. It would be beneficial to increase the power supply capacity in Taiwan by the installation of power plants using wind energy sources [2]. Therefore, it is very important to understand the wind generator capacity in production cost analysis for giving a good indicator to site new wind power plants in isolated systems, such as the Peng–Hu power system. However, the intermittency and unpredictability of wind power generation creates difficulty in control of frequency and scheduling of generation. It can be expected that many problems will arise in the renewable-energybased hybrid power system, particularly in system operation and ancillary services [3]–[9]. Daily generation scheduling is a critical task in a modern energy management system. Due to the uncertain nature of wind power, it is widely believed that large wind penetrations would put an increased burden on system operations. One of these issues is the provision of emergency reserve for the system security. In general, the largest proportion of the emergency reserve is carried to cover the loss of the largest generation unit in the system. However, with wind power penetrations increasing in isolated power systems, scheduling of additional emergency reserves will be needed to maintain an adequate level of supply reliability. In addition, maintaining sufficient emergency reserve across several units in the system is much more capable of responding to frequency deviations and system load pickup following a contingency for an isolated system [10]. Apart from the up spinning reserve requirements, there are strong demands for enough down spinning reserve requirements to satisfy the sudden rise of wind power generation at low system load times to avoid the forced shutdown of thermal units. The load variations, which are combined with wind power fluctuations, have to be absorbed by the system ramping capacity. The importance of wind–thermal coordination problem is, thus, likely to increase, and more advanced algorithms for solving the wind–thermal generation scheduling problem are worth developing to operate an isolated hybrid power system reliably and efficiently. The unit commitment (UC) problem is a very complex nonlinear mixed integer programming problem. The problem is further complicated by the wind–thermal coordination scheduling imposed by the adding of additional reserve requirements. Because of the relationship between the system spinning reserve requirements and the total actual wind power generation, both of them should be considered at the same time, not separately. These complex conditions make it very difficult to solve the wind– thermal coordination problem to achieve optimal utilization of wind energy sources in isolated systems. The literature on application of optimization methods for generation scheduling is vast. The UC methods reported in the literature, including priority list [11], dynamic programming [11], branch-and-bound [12], Lagrangian relaxation (LR) approach [13], genetic algorithm (GA) [14], and evolutionary programming (EP) [15], could be used to solve the extended unit commitment problem. Among all, the dynamic programming (DP) approach is of particular interest because of its ability to solve problems of a variety of sizes and to be relatively easy to handle a number of operating constraints. In this paper, a hybrid dynamic programming (HDP)