| It is generally known that the basic requirement for power system to operate safely and steadily is to maintain the balance between the power generation and the demand in real time. Traditionally, generation side resources are primarily used to track the fluctuations of loads. In principle, any measure that can be taken by generating units to ensure power balance can also be taken by loads. However, the demand has so far been largely underutilized. It is only when the imbalance is severe and cannot be remedied quickly or economically enough using generation side resources, the demand side resources are then used. With the development of society, the conflicts between energy shortage and environmental pollution are becoming more serious, promoting the change of the structure of the electrical sources. The proportion of thermal power units with high capacity, as well as nuclear power units is increasing, so the cost of reserve provided by generation side is increasing obviously. In recent years, renewable power generation, such as wind power generation and solar power generation, has seen a tremendous growth all over the world. The substantial increase in the proportion of power produced from renewable generation has a significant impact on generation side’s ability to track the changes in demand. This is due to the variability and uncertainty impacts of renewable generation. Furthermore, renewable power generation make the demand of the reserve for power system increase. In a word, the traditional measures used to ensure power balance are meeting great challenges. At the same time, power system’s communication ability is improved greatly, and the controllability of scattered electric loads is enhanced. By adjusting the operation states of loads actively, it is possible to implement cooperative generation-load control across multiple time scales.In power system there are many loads with energy storage property, such as air conditioners, refrigerators, water heaters, heat pumps, and plug-in electric vehicles, which can be actively incorporated to participate in power balance control. By adjusting the operation states of these loads, the change of loads’aggregated power can deliver a reliable resource to the power system while causing an imperceptible effect on end-use performance. This paper takes air-conditioning loads as an example, making loads provide system-level services for power system. The main contributions and innovations are described as follows:1) Methods for modulation of an air-conditioning load group’s aggregated power characteristics are proposed based on establishing the model of an individual air conditioner as well as the aggregated model of a load group. Firstly, the dynamics of a single air conditioner is modelled which comprises two types of expressions. The former is a much easier model describing by a first-order differential equation while the latter is a much complicated model describing by a three order differential equation. The latter model is more precise than the former model. The load group’s aggregated model comprises the Monte Carlo model and the state space model. The Monte Carlo model can be used to describe the operation of the air-conditioning load group, while the state space model can be used to estimate the air-conditioning load group’s operating state. Secondly, a state estimation method based on Kalman filter is proposed, providing credible data for analysis or control of the air-conditioning load group’s operating state. Lastly, the mechanism of diversity loss caused by inappropriate control schemes is analyzed, basing on which two kinds of load control strategies are proposed to modulate aggregated power characteristics of an air-conditioning load group. At the same time, another kind of load control strategy is borrowed from the literature which can also get nice aggregated power modulation characteristics of an air-conditioning load group. The impact of communication delay was evaluated on the control effect. A segmental load recovery scheme was proposed to avoid "recovery peak" in the all three control strategies. Simulation results verify the effectiveness of the proposed aggregated power characteristics modulation methods for an air-conditioning load group.2) A load control mechanism for air-conditioning loads to participate in primary frequency control is designed. Through adjusting air conditioners’setting temperatures according to the deviation of frequency, the aggregated power of air-conditioning load group can be changed, providing primary frequency reserve for power system. Two kinds of load control strategies for load groups responding to frequency events are proposed basing on the above basic frequency responsive load control mechanism. The former strategy is a decentralized load group control, by application of load group’s diversity recovery mechanism, load group’s diversity loss can be avoided. At the same time, a segmental load recovery scheme is proposed to avoid "recovery peak". The latter strategy is a coordinated load control for load groups. By application of load group’s diversity recovery mechanism, a decentralized air-conditioning load group control strategy with maintenance of load diversity is achieved. However, there is an unwanted parasitic "power peak" in the strategy. To eliminate the negative effects, a coordinated load control strategy is presented. The "power peak" in the decentralized control is used as the power control objective for the load group of another load group which implemented distributed load control, the control signals can be calculated using the dichotomy method. Then, a fast and stable response avoiding large oscillations and power peak is achieved for the load group combining the original two load groups. Simulation results verify the effectiveness of the proposed load control strategies for air-conditioning load group responding to frequency events.3) To incorporate and maximize the utility of load reserve, a cooperative generation-load frequency control strategy is proposed accounting for power network constraints. A decentralized control scheme is applied to the load side, which adjusts parameter settings to regulate its power demand in response to frequency measurement signal across primary frequency control time scale. Sampling operation states of air-conditioning load individuals across secondary frequency control time scale, the load aggregation model is identified using ARMAX technique through which power demand changes of air-conditioning load groups are estimated. A secondary frequency controller is designed on the model prediction control framework, producing control signals for regulating units by accounting for the load side control response. Frequency control aims at keeping power balance between generation and load sides while power network constraints should be satisfied. Negative impacts of load control are quantified on branch power limits with insecure load nodes identified and blocked from participating in frequency control. Simulations are conducted on New England 10-generator 39-bus system, verifying the effectiveness of the proposed cooperative generation-load frequency control strategy considering power network constraints.4) Air-conditioning load groups possess the potential to participate in active power balance control for power system. Two kinds of dispatch models for power system with air-conditioning load groups are proposed basing on the design of the dispatch architecture. Firstly, a random sampling algorithm was designed to track the power control objective of an air-conditioning load group. Then, an optimal dispatch model for load aggregator as well as a dynamic economic dispatch model with air-conditioning loads are respectively constructed. The former model is for the air-conditioning load groups on one load aggregator in response to the assigned dispatch signals, which can be solved by mixed integer programming. The latter model’s optimization objective is the minimum of both generation costs and load control costs. The constructed model is formed as a double objective optimization problem, which can be solved by the reference target method. One of the object function (the minimum of load control costs) is put into constraints, and the model can be converted to a single objective optimization problem. Simulations are conducted on IEEE30 system, verifying the effectiveness of the proposed random sampling algorithm for tracking the power control objective of a air-conditioning load group as well as the dispatch and control models for power system integrating air-conditioning load groups.5) In the distributed load control strategy, measurement data and control demands are transferred among loads, load aggregators and the dispatch center. The load aggregators are responsible for coordinating loads and the dispatch center, transmitting aggregated operating states of various load groups to the dispatch center and explaining load control demands assigned from the dispatch center. Take into account the highly distributed nature of loads, there are massive information needed to deal with for distributed load control strategy, so a cloud computing architecture is proposed for distributed load control to implement the distributed hierarchical control structure of load by flattening computing mode, and the cloud computing architecture directly responds the service requests asked by loads, load aggregators and the dispatch center. Using Cloudsim software, the cloud computing architecture for distributed load control is simulated, and the cloud computational amount of various controllable loads and storage tasks are quantified. Simulation results show that cloud computing platform possesses following superiorities:fast computing speed, strong storage capacity, easy to expand, and the ability to suit to the demand of distributed load control. Cloud computing process simulations are conducted based on Hadoop Map/Reduce. |
PDF Full Text Request