INCORPORATING WIND POWER FORECAST UNCERTAINTIES INTO STOCHASTIC UNIT COMMITMENT USING OPTIMIZATION TECHNIQUES
Abstract
The impact of large amounts of wind has complicated implications to Unit Commitment (UC) and Economic Dispatch (ED).The main objective of a UC, namely a Security Constrained Unit Commitment (SCUC), is to obtain a UC schedule with the minimum production cost and not compromising the system’s security. Normally, the SCUC’s constraints include the load balance, the reserve requirement, ramp rate limits, minimum up and down time limits and network constraints.In this project, Neural Network (NN) based Prediction Intervals (PIs) are implemented using Particle Swarm Optimization (PSO) based nonparametric Lower–Upper Bound Estimation (LUBE) method for wind power forecast uncertainty quantification. The Monte Carlo simulation method is used to generate wind power scenarios. Then the wind power scenarios are incorporated into a SCUC model. The heuristic Genetic Algorithm (GA) is utilized to solve the stochastic SCUC problem. GA can be implemented using MATLAB software.
Introduction
The inherent intermittency and variability of renewable resources such as wind require that current industry practices, such as Unit Commitment (UC) and Economic Dispatch (ED), be altered to accommodate large amounts of renewable generation. With sudden weather changes, the output power of a wind farm can drop most part of its power or even drop to zero. The decision to start or shut down thermal units for the next operating hours must take into account the inherent uncertainty of wind power forecasts. The error inherent to wind power forecasting can make the power operation costly prohibitive and/or unreliable.UC problems incorporated with wind generation uncertainties become a more important and challenge task in smart grid applications. The only solution for variability and limited predictability is to use conventional generation units to compensate.
Literature review
Kazarlis et al. (1996) developed a GA based solution to the UC problem. GAs are general purpose optimization techniques based on principles inspired from the biological evolution using metaphors of mechanisms such as natural selection, genetic recombination and survival of the fittest. Using the varying quality function technique and adding problem specific operators, satisfactory solutions to the unit commitment problem is obtained.
C.P. Cheng et al. (2000) described an application of a combined Gas and Lagrangian Relaxation (LR) method for the UC problem. The Lagrangian Relaxation and Genetic Algorithms (LRGA) incorporate GA into LR method to update the lagrangian multipliers and improve the performance of LR method in solving combinatorial optimization problems such as the UC problem.
Swarup and Yamashiro (2002) developed the solution methodology of UC using GA. Problem formulation of the UC takes into consideration the minimum up and down time constraints, startup cost and spinning reserve, which is defined as minimization of the total objective function while satisfying the associated constraints.
Srinivasan and Chazelas (2004) developed an efficient algorithm for aiding unit commitment decisions. An Evolutionary Algorithm (EA) with problem specific heuristics and genetic operators is employed to solve the problem. The initial random population is seeded with good solutions using a priority list method to increase the speed of convergence and improve efficiency of the algorithm.
Ummels et al. (2007) described a new simulation method that can fully assess the impacts of large-scale wind power on system operations from cost, reliability, and environmental perspectives. The method uses a time series of observed and predicted 15-min average wind speeds at foreseen onshore and offshore wind farm locations. Unit Commitment and Economic Dispatch (UC-ED) tool is adapted to allow for frequent revisions of conventional generation unit schedules, using information on current wind energy output and forecasts for the next 36 h.
Wang et al. (2008) developed a SCUC algorithm which takes into account the intermittency and volatility of wind power generation. The UC problem is solved in the master problem with the forecasted intermittent wind power generation. Next, possible scenarios are simulated for representing the wind power volatility. The initial dispatch is checked in the sub problem and generation re-dispatch is considered for satisfying the hourly volatility of wind power in simulated scenarios.
T. Logenthiran and D. Srinivasan (2010) developed PSO based heuristic optimization algorithms to solve the UC problem. This PSO algorithm is indeed capable of obtaining higher quality solutions efficiently in solving UC problems.
Sturt and Strbac (2012) described an efficient formulation of the SUC problem that is designed for use in scheduling simulations of single bus power systems. Unlike traditional SUC techniques, this formulation use the quantile based scenario tree structure that avoids the need for exogenous operating reserves.
Jiang et al. (2012) developed a robust optimization approach to accommodate wind output uncertainty with the objective of providing a robust unit commitment schedule for the thermal generators in the day-ahead market that minimizes the total cost under the worst wind power output scenario. Robust optimization models the randomness using an uncertainty set which includes the worst-case scenario, and protects this scenario under the minimal increment of costs. It introduces a variable to control the conservatism of this model, by which it can avoid over-protection. By considering pumped-storage units, the total cost is reduced significantly.
HaoQuan et al. (2014) developed the nonparametric neural networkbased prediction intervals (PIs) for forecast uncertainty quantification. Wind power scenarios are incorporatedinto a stochastic security-constrained unit commitment (SCUC)model. The heuristic genetic algorithm is utilized to solve the stochastic SCUC problem.