تشخیص فروپاشی ولتاژ با استفاده از بهینه سازی کلونی مورچه برای برنامه های کاربردی شبکه های هوشمند

Load demand levels in power networks are continuing to increase especially considering the penetration of Plug-in Hybrid Electric Vehicles, and as a result bus voltages are requiring more support from generators elsewhere in the network in order to maintain the voltage within specified limits. If the network is unable to support the increasing demand, bus voltage will begin to degrade until the point of voltage collapse which can lead to catastrophic network failure. Previous studies have shown that evolutionary computing techniques are effective methodologies for locating voltage collapse points. Ant Colony Optimization techniques allow for the optimization of many independent parameters simultaneously, the loading parameter for each bus is considered in this work. In this study, Ant Colony Optimization is applied to detect voltage collapse conditions in power networks, to obtain faster computing time with the future goal of providing online detection and prediction for use in smart grids. Two case studies are considered in this study to assess the performance of the proposed detection algorithm; the first case study includes 9-bus system while the other case study involves IEEE 118-bus system. Results obtained from both cases and conclusions drawn are also presented.

The ever-increasing demands in modern power networks have resulted in several large blackouts in recent years. Given that these demands increase show no signs of slowing, there is a need for further study into the stability of power systems to avoid more blackouts in the future. Voltage collapse has been identified as the cause of many of these power system events. It is characterized as a gradual decay of bus voltage magnitudes until a sharp drastic drop occurs. This drastic drop, or “collapse point,” has been tied to the occurrence of saddle-node bifurcation, and is discussed in Section 2. There have been many studies conducted to identify these collapse points using the direct method [1], [2] and [4] as described in Section 3. The continuation power flow method [3], [4] and [5] was introduced as an alternative to the direct method as a way to deal with numerical difficulties associated with the direct method. Study [7] has begun using Genetic Algorithms (GA), such as Particle Swarm Optimization (PSO) and Differential Evolution (DE), in conjunction with these methods to locate the collapse points, or Ant Colony Optimization (ACO) to determine maximum loadability levels [8]. The ACO, introduced by Dorigo in 1996 [9], is an analog to the foraging habit of ants in nature. This algorithm, like many other GA-based algorithms, relies on the past experiences, or solutions of previous iterations, to move towards a globally best solution based on a describing function or fitness value. This algorithm has an advantage over other GA-based algorithms in that it can optimize a large number of independent parameters to create an optimal solution. The ACO is further described in Section 4. This paper utilizes an ACO algorithm to locate the voltage collapse point in power systems. The voltage collapse point is commonly tied to a loading parameter λ*, at which the system experiences a bifurcation [1] and [2]. This parameter is a single scalar value used to equally load all buses. The proposed approach utilizes the ACO algorithm's ability to optimize groups of independent parameters and thus λ* becomes a vector of scalar values each associated with a single system parameter. The proposed approach is described in detail in Section 5. The proposed approach is then applied to two case studies involving an IEEE 9-bus system and an IEEE 118-bus system while results are summarized in Section 6.

In this paper, Ant Colony Optimization-based algorithm was developed to identify and detect voltage collapse points in electric power network. As demonstrated through two case studies involving IEEE 9-bus and IEEE 118-bus systems, the proposed method has the ability to locate the closest saddle-node bifurcation point in an electric power system. It does this by applying loading values to the system until a bifurcation occurs. The proposed method has an advantage over other methods used in [1], [2], [3], [4], [5] and [7] since it does not assume uniform loading of all buses. Therefore it is capable of identifying not only the weak buses in the power system but also the system parameters that drive those buses to voltage collapse. However the proposed method does not provide the same degree of accuracy for the loading values as determined by other methods. This limitation is due to the discrete nature of the states of the construction graph and the limitation on the step size (resolution) of λ*. The speed of the algorithm is dependent on the resolution and range of λ* since each of these parameters increase the number of states in a given stage of the construction graph.