| Summary, etc. |
In this dissertation, augmented Lagrange Hopfield network based methods are proposed for solving short-term optimal generation scheduling problems including economic dispatch, unit commitment, and hydrothermal scheduling. The objective of the economic dispatch (ED) problem is to determine the output power of each generating unit so as load demand and other constraints of the network and the generators are satisfied at minimum cost. In this dissertation, different ED problems are solved. Multi-objective fuel constrained ED problem is to simultaneously minimize operating cost and emission of generating units subject to power balance, fuel delivery, fuel storage, operating limits, fuel delivery limits, and fuel storage limits constraints. Combined heat and power ED problem is to minimize total operating cost of generating units subject to power balance, heat balance, and generating limits constraints. In this problem, the maximum power and heat outputs of cogeneration units depend on their heat»power feasible region which is a trade-off between power and heat production. ED problem with prohibited operating zones is to minimize total generating cost of generating units while satisfying power balance, generating limits, prohibited zones, spinning reserve, and ramp rate constraints. ED problem with piecewise quadratic cost function is to minimize total generating cost of generating units among available fuel types for each unit subject to power balance and generating limits constraints. Hydrothermal system ED problem which is the ED problem applied for both thermal and hydro units is to minimize total operating cost of thermal generating units while satisfying power balance, continuity of reservoir, and generating limits constraints. For solving the multi-objective fuel constrained ED, combined heat and power ED, and hydrothermal system ED problem, an augmented Lagrange Hopfield network (ALHN) is proposed. ALHN is a continuous Hopfield network with its energy function based on augmented Lagrange function. In ALHN, unit and system constraints can be easily handled by sigmoid function of Hopfield network and augmented Lagrange function, respectively. Moreover, ALHN is a recurrent network with parallel processing, thus it is very fast in solving very large-scale ED problems. For solving the ED problem with prohibited operating zones or piecewise quadratic cost function, heuristic search is used to handle the non-convex in these problems before using ALHN for solving final ED. For off-line planning, unit commitment (UC) is used to schedule generating units based on load forecast for a specified planning horizon. The UC problem is typically a large-scale, nonlinear, and mixed-integer problem. In this dissertation, the UC problem is to minimize the generator fuel cost, start up and shutdown costs of thermal units subject to power balance, spinning reserve, generating limits, minimum up and down times, operating ramp rates, and start up and shutdown ramp rates constraints. Three solution methods based on augmented Lagrange Hopfield network are proposed for solving the UC problem including enhanced augmented Lagrange augmented Hopfield network (ALAHN), improved merit order and ALHN (IMO-ALHN) and ALHN based Lagrangian relaxation (ALHN-LR). In the enhanced ALAHN method, ALAE-IN which is a hybrid of discrete and continuous Hopfield neural network with its energy based on augmented Lagrangian function is used for finding primary unit scheduling of generating units satisfying power balance and spinning reserve neglecting minimum up and down time constraints; heuristic search is used for repairing minimum up and down time constraint violations; and finally ALHN is used for solving ED problem. For the IMO-ALHN method, IMO which is merit order of generating units based on their average production cost enhanced by heuristic search is used to find unit scheduling of generating units satisfying power balance, spinning reserve and minimum up and down time constraints so that the total cost is minimized, ALHN is applied for solving ED problem, and heuristic search is used for repairing ramp rate constraint violations if the feasible solution of the ED problem is not found. In the ALHN»LR method, an improved LR (ILR) is applied using improved adjustment factors for updating Lagrangian multipliers and neglecting duality gap. The methodology of the ALHN-LR method is similar to that of the IMO-ALHN method, in which ILR with heuristic search is used for finding unit scheduling of generating units. In many power systems, the hydrothermal scheduling (HTS) is used to optimally and simultaneously schedule both thermal and hydro generating units including pumped»storage units. The objective of the HTS problem with pumped-storage units is to minimize the generator fuel cost and start up costs of thermal units while satisfying power balance, spinning reserve, generating limits, minimum up and down times, generating ramp rates, on/off line minimum level, limited fuel, environmental emission, transmission line, water discharge, and water balance constraints. Two solution methods based on augmented Lagrange Hopfield network including IMO-ALHN and ALHN-LR are proposed for solving this problem. In the IMO-ALHN method, IMO is used for committing thermal, hydro and pumped-storage units satisfying power balance, spinning reserve, minimum up and down times, limited fuel, water discharge and water balance constraints, ALHN is used for solving constrained ED problem, and heuristic search is applied for repairing ramp rate, emission and transmission constraint violations if the feasible solution of the constrained ED problem is not found. The ALHN-LR method also solves the problem in the similar manner to IMO-ALHN, in which ILR with heuristic search is used similar to IMO. In deregulated power systems, the centralized optimal generation scheduling needs reformulation to profit-based self-scheduling optimization problem. The profit-based DC (PBUC) problem is to maximize the total profit of the generation company defined as the difference between the total revenue and total cost subject to inequality constraints including power demand, reserve, minimum up and down times, and operating limits constraints. In this dissertation, an IMO-ALHN solution method is proposed. Here, IMO is used to find unit scheduling of generating units so as to maximize the company profit satisfying minimum up and down times constraints, and ALHN is used for solving optimal power dispatch for profit maximization subject to inequality constraints of power demand, reserve, and operating limits. The proposed solution methods have been tested on various test systems for ED, DC and HTS problems. The test results have shown that the proposed solution methods are much efficient than other methods reported in the literature in terms of less total cost and faster computational time. Therefore, the proposed methods are very favorable for large-scale practical optimal generation scheduling problems |
AI Search