Research On The Unit Commitment Problem With Security-Constraints And Wind Power

With the development of national economy and reform of power utilities, it is more and more important to guarantee the safety, economy, and reliability of power system, which make the operation dispatch decision-makings ever more complicated. Optimal power flow (OPF) and unit commitment (UC) problem are the key issues of operation dispatch. This dissertation, guided by the OPF and UC problem, performs meticulous and thorough work of research. The main contents and contributions of this dissertation are summarized as follows:(1) Research on the optimal power flow (OPF) problem with different network models. The OPF problem with ac power flow is an accurate representation, which may be solved after a larger number of iteration in a long execution time. On the contrary, the OPF based on the dc power flow has the fast convergence characteristic and short execution time. However, the line losses can not be taken into consideration. Therefore, the second-order and considering voltage magnitude second-order network model are introduced into the OPF problem. The OPF problem with the second-order or considering voltage magnitude second-order network model not only can achieve the accurate representation of line losses, but also the OPF with second-order network model can be formulated as the convex programming so that the fast convergence to the global optimum can be guaranteed. Numerical results show the advantages of the second-order or considering voltage magnitude second-order network model in terms of the convergence characteristic and execution time.(2) Research on the conventional UC problem and the optimization method. The UC problem is a large-scale, non-convex, and nonlinear mixed-integer program (MIP), which is very difficult to find the exact solution. In this dissertation, the second-order network model is introduced in the conventional UC problem to model the line losses. The mixed-integer non-convex UC problem is formulated as the mixed-integer convex program (MICP) problem, in which the ramp rate limits and their influence on the spinning reserve requirement are also modeled. The MICP base UC is optimized by the branch and bound-primal dual interior point method. During the branch and bound stage, the best first search (BeFS) and depth first search (DFS) are combined to accelerate the search process. In order to achieve the goal of finding a fast, near-optimal, and feasible solution for the large-scale UC calculation, some accelerating strategies (the heuristic, partition of the studied period, and the modified branch and bound) are developed. The numerical results indicate that the MICP based UC model is reasonable and the proposed approach is efficient by the comparison of results with previously reported method and is promising for large-scale UC calculation.(3) Research on the security-constrained unit commitment (SCUC) problem. The power is transmitted over the transmission line, the security constraints of the network may result in the infeasibility of the schedule obtained from the conventional UC problem. In order to guarantee the secure operation of power system, it is necessary to add the network constraints into the UC problem. However, the introduction of network constraints aggravates the scale and complexity of UC problem. An efficient approach based on the branch and bound-interior point method is proposed to solve the SCUC problem with ac constraints via the second-order network model. The solution process of the SCUC problem is divided into two stages. In the first stage, the SCUC problem with second-order or considering voltage magnitude second-order model is decomposed into single-hour SCUC problem to avoid the tremendous computation burden, of which the single-hour SCUC problem with second-order model is formulated as the MICP. The MIP or MICP based the single-hour SCUC problem is optimized by branch and bound-interior point method. This stage is called single-hour SCUC based subproblem. In the second stage, the units'generation level should be corrected by ac constraints according to the commitment schedule obtained in the first stage. This stage is called economic dispatch based subproblem. For testing the efficiency of proposed approach, a modified IEEE 118-bus with 36 thermal units, a modified IEEE 118-bus with 54 thermal units, and a modified IEEE 678-bus with 170 thermal units are analyzed over a period of 24 hours.(4) Research on the unit commitment with wind power generation and energy storage system. Possible scenarios, which are generated by the Latin hypercube sampling combined with Cholesky decomposition method, are simulated for representing the wind power volatility. The scenario reduction technique, based on probability metrics, is employed to decrease the number of scenarios so that the computation burden can be reduced. In order to achieve the higher system flexibility and economy, the energy storage system is also introduced into UC problem, and the corresponding deterministic and stochastic model are established, which are optimized by branch and bound-interior point method. The numerical results show that the wind power volatility will influence the unit commitment problem greatly. The more volatile the wind power, the larger the total operation cost. Even the UC problem may be infeasible. Improving the ramping capabilities of thermal units or the introduction of energy storage system not only can reduce the operation cost, but also can alleviate the impact of wind power volatility on the UC problem.(5) Try to adopt Trust-Tech to find the feasible solution and optimal solution of the optimization problem. In order to locate the feasible solutions, the constraints of optimization problem are converted to the dynamic system named AS-QGS, which is solved by the Pseudo-Transient method. The stable equilibrium points of AS-QGS are the feasible solutions of the primal optimization problem. If the optimal solution should be found, the constrained optimization problem is converted to the unconstrained optimization problem by the sequential unconstrained minimization technique, and the corresponding dynamic system is established. The stable equilibrium point of this dynamic system is the optimal solutions of original constrained optimization. The numerical results indicate that Trust-Tech has a good application in the optimization problem.