Recently, the smart grid technologies have been developed rapidly recently, which an important component is the so called demand response management (DRM). With the help of a DRM program, a utility company can adjust the power demand and electricity price to reduce the cost of power generation and consumption. However, there are many problems in DRM need to be solved. For example, to solve the problem of optimizing a generator's power (GP), the conventional methods such as economic dispatch (EDP) may reduce the profit of the utility company. To solve the problem of optimizing a consumer's power (CP), the existing smart pricing strategies may reduce the long-term benefits of the customers. This dissertation aims to develop a two-stage game model to increase the profit of the utility company and while increase the long-term benefit of the customers. For solving the GP. It is critical for the power generator and utility company to allocate the power demand properly, but the profit for the utility company may be reduced. To solve the CP, it is difficult for the customers to achieve a long-term beneficial power-usage-pattern with myopic pricing strategies. The stability of the smart grid and the benefit of the customers may also be reduced due to the myopic pricing strategies. It is difficult for the utility company to use the existing methods (e.g., EDP) to order an optimal power demand from the power generators to earn the maximum profit. There are two issues that are needed to be solved in the GP. First, the weight function for the utility company and power generators in the GP is not established properly in the existing methods. For example, the value of the weight function for the utility company and power generators are usually the same in an EDP method. However, in a smart grid, the utility company has the privilege to demand the power while the power generators must follow the demand. Hence the value of weight function for the utility company should be greater than the one for the power generators in a GP. Second, the optimal demand for the utility company is most likely not the optimal generation for the generators. The imbalanced power will increase the generation cost significantly. It is also difficult for a utility company to maintain an efficient DRM for a long-time period by using the existing smart pricing strategies. Applying incentive is the major solution for the utility company to influence the power demand of a customer. However, the traditional pricing strategies are shortsightedly designed, by which the long-time efficiency for the DRM is reduced. For example, the trigger punishment strategy applies a punishing price to a customer for a long period when a non-cooperation behavior is detected. During the punishment period, the customer chooses its power consumption freely since the punishment will be applied anyway. Such selfish behaviors reduce the long-term efficiency for the DRM and the stability of the smart grid. In this dissertation, we propose a two-stage game model to solve the GP and CP to increase the long-term efficiency for the DRM, maintain the stability of the smart grid, and also increase the profit of the utility company. In the first stage, a Stackelberg game model is applied to solve the GP, in which the utility company is the leading player while the generators are the following players. We prove that the GP for the following players is a convex problem mathematically. The following players achieve the Nash equilibrium (NE) state by choosing the unique optimal generation. The leading player reacts with this unique generation to achieve the optimal profit. Both the leading and following players reach an agreement in the NE state, in which they have no motivation to deviate the optimal actions. A genetic algorithm is developed to obtain the optimal demand for the leading and following players. In addition, we introduce a power balance constraint to the leading and following players to avoid the cost caused by the imbalanced power. By applying the constraint, the generated power is equal to the demand all the time. The smart grid will not need to store the excessive power in the energy storage unit or send the power back to the power generators to keep them idling. The cost is avoided and the efficiency of the DRM is increased. In the second stage, a repeated game model is applied to solve the CP, in which the customers are the players. The strategy for the players is to minimize the individual power consumption of each customer. The utility function for the players is the cost of the customers. The objective for the players is to minimize the cost. In this work, we prove that the NE state exists for the repeated game. However, it has been shown that in the NE state, the players' myopic behaviors may reduce the benefits for the entire group of players. To avoid the loss, we use a genetic algorithm to find the Pareto-efficient solution for the players, in which no player can increase its benefit by reducing other players' benefit. We apply a Tit-for-Tat (TFT) smart pricing strategy to increase the punishment strength from the utility company. Once an irrational behavior from a player is detected, a punishment will be applied to the player for a short period of time. The player can choose to cooperate or not during the punishment period. Compared to the existing smart pricing strategies, the long-term benefit for the smart grid is increased by applying the TFT strategy to the customers. The numerical simulations in different scenarios are conducted to evaluate the performance of the proposed two-stage game framework by using MATLAB. All the parameters and constraints of the related components are from the Department of Energy's report and the Oasisui online database. Five power generators, one utility company, and one hundred customers have been used in the simulations. Compared with the existing solutions (e.g., EDP and gaming optimization), the cost in power consumption is reduced by 6% percent while the profit for power generation is increased by 8% percent in our test scenarios. With the help of the proposed model, we enhance the efficiency for the DRM. The peak-to-average ratio (PAR) of the power demand of our work is compared with the EDP method. The effect of the PAR is studied. The numerical results show that the proposed model has a similar PAR to that of the EDP method, which implies that the proposed model has no negative influence on the stability of the smart grid. The punishing effort of the TFT strategy is compared with the trigger strategy (TP) to study the punishment influence on the customers. The numerical results show that the customers who are applied with the TFT strategy are more willing to cooperate with the utility company. The impact of the power loss ratio and different types of customers is also simulated and analyzed. The simulation results show that the players with a greater transmission loss ratio are more willing to cooperate. The customers that are associated with a greater linear dissatisfaction coefficient are more concerned about the dissatisfaction cost. The customers with greater price-sensitive coefficients are more concerned about the consumption cost. In summary, compared to the existing solutions, the proposed two-stage game model improves the performance of the DRM while maintain the stability of the smart grid. We also discuss the future research issues in the related areas. / A Dissertation submitted to the Department of Electrical and Computer Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Summer Semester 2018. / May 8, 2018. / Includes bibliographical references. / Ming Yu, Professor Directing Dissertation; Xiuwen Liu, University Representative; Leonard Tung, Committee Member; Petru Andrei, Committee Member.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_647221 |
| Contributors | Fan, Huipu (author), Yu, Ming (professor directing dissertation), Liu, Xiuwen, 1966- (university representative), Tung, Leonard J. (committee member), Andrei, Petru (committee member), Florida State University (degree granting institution), College of Engineering (degree granting college), Department of Electrical and Computer Engineering (degree granting departmentdgg) |
| Publisher | Florida State University |
| Source Sets | Florida State University |
| Language | English, English |
| Detected Language | English |
| Type | Text, text, doctoral thesis |
| Format | 1 online resource (81 pages), computer, application/pdf |
Page generated in 0.0016 seconds