A New Intelligent LFC Design in a Deregulated Environment Fatemeh Daneshfar and Hassan Bevrani Department of Electrical and Computer Engineering University of Kurdistan Sanandaj, Iran daneshfar@ieee.org Abstract— In this paper, an intelligent approach based on XCSR (accuracy-based learning classifier system with continuous- valued inputs) method is proposed for the load-frequency control (LFC) system using a modified traditional frequency response model suitable for a bilateral-based deregulation policy. Model independency and flexibility in specifying the control objectives; cause it as an interesting solution for the LFC design in new power system environment. To demonstrate the capability of the proposed solution, a simulation on a 3-area power system with possible contract scenarios is given. Keywords- Load frequency control; learning classifier systems; XCSR; deregulated environment I. INTRODUCTION In a deregulated environment, the load-frequency control (LFC) design has an important role to enable power exchanges and to provide better conditions for the electricity trading. It is treated as an essential auxiliary service to keep the electrical system reliability at a suitable level [1]. However usually the load frequency controllers used in the industry are proportional-integral (PI) type that are designed for a specific operating points, these conventional LFC designs are not usable for large-scale power systems with nonlinearities, undefined and uncertain parameters, also if the nature of the disturbance varies, they may not perform as expected [1]. Then as a result, adaptable and flexible controllers like intelligent controllers [2-8] are more suitable than the classical ones, for the LFC problem in deregulated environments. XCSR is a “continuous-valued input, Learning Classifier System (LCS)”. It is one of the intelligent approaches which have received little attention in the area of power system control. It is a machine learning approach for producing adaptive, flexible systems in an unknown environment using reinforcement learning (RL), evolutionary computing and other heuristics [9]. All evolutionary computing approaches like genetic algorithms (GA) mainly are based on searching a problem space by producing and developing an initially random individuals (of solutions) such that fitter individuals (solutions) are generated over time [10]. The RL is also a machine learning technique which the agent (learner) interacts with the environment and learns through mapping state and action combinations to their utility with the aim of being able to maximize future environment reward [10]. XCSR is a system with a knowledge base of rules, where the rules are usually in the RL traditional production system form of “IF state THEN action”. Evolutionary computing techniques are used to generate and search the space of legal rules, whilst RL techniques are used to assign rewards for the existing rules, and therefore guiding the search process to find better and fitter rules [10]. In this paper, an XCSR based control structure design is described and applied to a modified dynamical model for a general control area in the deregulated environment with bilateral contract policy introduced in [1]. It has an intelligent controller that receives the area control error (ACE) and its deviation (∆ACE) signals and provides generator set point signal (∆P c ), using XCSR method; then it is distributed among the different units under control using fixed participation factors. The above technique has been applied to the LFC problem in a three-control area power system as a case study. In the performed simulation, the test power system is considered as a collection of control areas interconnected with high-voltage transmission lines (tie-lines). The organization of the rest of the paper is as follows. In Section 2, a brief discussion on a test system for LFC synthesis problem is given. An explanation on XCSR method and how a load–frequency controller can work within this formulation is provided in Section 3. In Section 4, a case study of a three- control area power system, for which the above architecture is implemented for, is discussed. Simulation results are provided in Section 5, and paper is concluded in Section 6. II. TEST SYSTEM Here, to illustrate the effectiveness of the proposed control strategy for LFC design, a generalized dynamical model for a control area in restructured power system is used [1]. The modified LFC block diagram for control area i can be obtained as shown in Fig. 1. A power system in a deregulated environment includes separate generation, transmission and distribution companies with an open access policy. In an open energy market, a distribution company (Disco) has the freedom to contract with any available generation company (Genco) in its own or another control area (there can be various ICEE2012 1569538443 1