Research On Modeling And Solution Algorithms Of Modern Power System Optimization

With the development of power industry, the modern power system has become more tremendous, and its framework and modes of operation get to be more complex in the extreme. With the emergence of new nature and requests, it is difficult to solve the power system optimization for the conventional optimization models and algorithms. It is, therefore, necessary to development perfect model and practical algorithm in the field of power system optimization according to the characteristic of modern power system. This dissertation devotes to the interrelated problems of power system optimization, including the study on efficient algorithms of the power flow and its application, the research on problem formulation of short-term power scheduling, the research on problem formulation of reactive power optimization (RPO), and the research on solution algorithm based on particle swarm optimization (PSO). Above all, main work that has been carried out in this dissertation is as following. The optimal hydrothermal scheduling is formulated with limited-energy units'consideration. For making full use of the nature renewable energy source, the hydro unit is firstly deal with based on the peak-shaving method of hydro unit to find its work location in the load curve. Search technology with the adaptive step is adapted to improve the speed of convergence. As for the thermal unit commitment, a discrete binary PSO (DPSO) method is improved for solving it. The proposed method alters the updating rule of the particles position, introduces the No-hope/Re-hope criterion and heuristic pseudo-mutation mechanism to iterative process. As overcome the demerit of focusing on the local optimum and ensure state variables valid. The feasibility of the proposed method is demonstrated by two systems, and the results show that it is advantageous in terms of the solution quality and computation efficiency. According to the characteristic of modern power grid, an advanced model of optimal reactive power planning is presented. In the problem formulation, the objective function includes energy loss cost, power loss, and the total capacitor cost comprises the purchase cost and the installment cost. Discrete nature of capacitors and transformer taps, and harmonic distortion effects are all taken into consideration in the mathematic formulation. A fully scheme of advanced PSO is proposed for solving the RPO, which includes population initialization, fitness calculation, non-uniform mutation mechanism. A stage-coding strategy is used based on the criteria that different attention, such as different mutation operator and different calculation precision, should be given to different voltage level control variable. There is some shortcoming in solving of RPO problem in a large-scaled network, such as huge time-cost and low precision, i.e. The pseudo parallel PSO and adaptive changing population size technology are presented to avoid being trapped by local optimal and elevate the efficiency of solution approach. To improve the speed of calculation, a decomposing optimization is used according to power grid of natural management region. The simulation results of three power systems illustrate the validity of solution approaches. In reactive power dispatch problem, a multi-objective model of reactive power optimization (RPO) is proposed, which involves the voltage stability view. The proposed model takes into account of power loss minimization, voltage stability margin maximization in its objective function. The least singular value of converged load flow Jacobian is employed to measure voltage stability. To solve the multi-objective optimization problem, two methods have been employed. One is transforming multi-objective problem into single-objective problem, then solve the latter using single-objective algorithm. While getting no knowledge of the preference of sub-objective, a novel multi-objective PSO based on the distance of fitness space is presented and applied to multi-objective RPO. Test results are used to show the accuracy of the algorithm. In recent years, the increasing levels of high voltage direct current (HVDC) in power systems draw many engineers'attention. However, the research work mostly focused on the aspect of design and control of HVDC system. In this dissertation, the reactive power optimization and economic operation problem of AC/DC hybrid power system is developed. A mathematical formulation of this problem is proposed, in which the objective function is the sum of power loss in AC power network and that in DC power network. In this formulation, the control variables include not only conventional AC variables, but also some DC variables such as the control voltage, the control current and control power in conversion station. A modified PSO-based approach is also proposed to solve the AC/DC RPO problem. The simulation results prove that the study of AC/DC hybrid system is very significant. Finally, conclusions about the dissertation are summarized and the further work is pointed out.