Optimization Algorithms And Models For Power System Dispatch

With the development of economy, power system has been confronted with some signif-icant changes, such as network size becomes larger and load becomes more uncertain, which brings new challenges for designing an efficient algorithm to obtain the dispatch solution. The traditional optimization algorithm, which although have a long period of development and im-provement, are still experiencing many deficiencies when facing these new changes in power system. On the other hand, the world has been confronted with the era of energy crisis and global warming as the increasing depletion of the traditional energy, such as coal, oil, natural gas, etc. Thus, renewable energy has been largely penetrated into power system, which need-s reliable models to measure its profit and risk with respect to its stochastic and intermittent. Most researchers just consider the profit brought by renewable energy, while the risk caused by renewable energy has been neglected. On the whole, the complexity of power system dispatch is mainly reflected in the nonconvex, high-dimensional and multi-model objective function; a large number of equality and inequality constraints; and many stochastic and uncertain factors, which brings significant challenges on model constructing and algorithm designing. Therefore, how to construct a reliable and flexible model for assessing power system dispatch under un-certain environment, and design an efficient and applicable optimization algorithm to obtain the dispatch solution are of great practical significance for power system operation. This paper does some research work on these issues.(1) In order to solve traditional power system dispatch problem, this paper proposes a vari-ant of group search optimizer (GSO), called self-learning group search optimizer (SLGSO), to overcome the problem of slow convergence without significantly impairing the global search feature of GSO. SLGSO is inspired from the fact that animals could efficiently search for food based on the past successful experience. Simulation results obtained based on the benchmark functions demonstrate that SLGSO performs better than GSO. In addition, SLGSO has been successfully used for optimizing hydrothermal dispatch problem.(2) In order to well balance the algorithm’s local search ability and global search ability, this paper presents a novel evolutionary predator and prey strategy (EPPS). The EPPS is based on a dynamic predator-prey model stemmed from the study of animal group living behaviors. In the model, concepts of experienced predators, strategic predators, the prey and its safe location are firstly developed to simulate three typical animal behaviors:scanning, hunting and escaping. To validate the applicability and practicability of EPPS, experiments were undertaken on a set of 20 benchmark functions and three real-world problems, respectively. The results show that EPPS has a more superior performance in comparison with other recently developed methods reported in the literature. Moreover, EPPS has also provided new idea for designing intelligence algorithm.(3) In order to reliably assess the problem of stochastic and uncertain economic dispatch, this paper proposes a multi-objective mean-variance-skewness (MVS) model. The MVS mod-el considers the maximization of both the expected return and skewness while simultaneously minimizing the risk, which is formulated as a competing and conflicting three-objective opti-mization problem. Then we propose a multi-objective optimization algorithm, multiple preys based evolutionary predator and prey strategy (MPEPPS), to provide Pareto solutions, which show the trade-off relationship among the expected return, the skewness and the risk of the dis-patching objective. Subsequently, a multi-criteria decision making method, the technique for order preference by similarity to an ideal solution (TOPSIS), is applied for determining the fi-nal dispatch solution. The objective of this paper is to develop a reliable model to assess the stochastic dispatch problem from the perspective of economics and reliability of power system operation, and propose an efficient algorithm to obtain a solution that considers all of the possi-ble load and wind power simultaneously. Simulation results based on a modified IEEE 30-bus power system demonstrate the reliability and effectiveness of the MVS and MPEPPS in solving stochastic dispatch problem.(4) In order to solve the problem of optimal power flow considering wind power integrated (OPFWP), this paper presents a probabilistic interval optimization (PIO) model. The PIO model considers the profit and risk simultaneously with consideration of the environment of uncertain wind power, where the profit is manifested by the net decrease of generation cost between the same power system with and without wind power integrated and the risk is measured by the distribution probability of wind power. In addition, the PIO model does not need to repeated sample the wind speed to simulate the actual wind power distribution or solve a multi-objective optimization to obtain the trade-off solution between the profit and risk. Then the OPFWP is solved by a newly proposed evolutionary algorithm, named evolutionary predator and prey strategy (EPPS). Numerical tests based on a modified IEEE 30-bus system demonstrate the effectiveness of the PIO model, which can provide scientific reference to evaluate renewable energy.