International Journal of Electrical & Computer Sciences IJECS-IJENS Vol: 12 No: 01 7 125601-7474 IJECS-IJENS © February 2012 IJENS I J E N S Abstract— Power flow is very important tool for analysis power systems. One of the best power flow methods is Newton Rahpson. The current algorithm of Newton Rhapson power flow still used rectangular limit (P min -P max /Q min -Q max ) to represent generator capability curve (GCC). Using rectangular limit is not optimum because it is ignore some are inside GCC. Although less optimum the rectangular limit is still used in many power systems applications, because using GCC as limit in power flow needs complicated mathematical equation. Neural Network (NN) is one of the best methods in imitating a curve. In this paper NN is employed to imitate GCC to limit the operating point of generator in power flow using security check algorithm. The Advantages of using GCC based on NN as power flow limit is can minimize the complicated mathematical equations. Also the algorithm is very simple and accurate especially in representing the operating point near steady state limit. Index Terms— Newton-Raphson, Power-Flow, Generator capability Curve, Neural-Network, Constructive back- propagation. I. INTRODUCTION ower flow is very important tool for the analysis power systems and it is used in operational and planning. The objective of power flow is calculating unspecified bus voltage angles and magnitudes, active and reactive powers, as well as line loadings and their associated real and reactive losses for certain generation and load conditions. Since the middle of the last century, many methods were proposed to solve this problem. In 1954 Dunstan [1] was demonstrated a digital method for solving power flow problem. In 1956 Ward and Hale [2] also have credited the successful digital formulation and solution of the power flow problem. Still relating to computer application, In 1967 Tinney and Walker [3] introduced the optimally ordered and sparsity-oriented programming techniques to improve the computational time in computer application. In 1974 Brian Stott was introduced Newton–Raphson (NR) method [4], as well as the decoupled [5] and fast decoupled (FD) power flow approaches [6]. The NR usually converges faster than other methods, but it takes longer computational time per iteration. The including [3] in Newton based method was made the Newton’s methods became standard application in industry[7]. Nowadays, the power systems become bigger and more complicated [8-12], The Newton based technique was developed and combined with other methods to get better performance [13-14]. The influenced of equipment especially power electronics component was made the power quality has a harmonics inside, the [15] was used the decoupled Newton to get faster process iteration in accounting converter model in iterative process. Until Now, the Newton based technique still can solve some power system problem, but the operating point of power source (generator) is not enough if it just be operated using rectangular constraint. The proposed method in this paper is concern in accounting any kind of limit generation, not only line and curve but also combination of line and curve with discontinuities. Actually the proposed methods can be applied in any kind of curve, but in this paper only simulated the GCC model [16] for limit operating point of generator in power flow application. The Neural Network (NN) [17-18] is employed to imitate the GCC to eliminate the complicated of mathematical equation. The security check algorithm also developed to make a simple in checking limit. The organization of the rest of the paper is as follows. In Section II, how GCC based on NN can easily corporate in NR power flow will be described, starting from overview NR power flow, how to imitated GCC using NN and how to security check algorithm is including in NR power flow. In Section III, the IEEE data test 30 bus is used to verify the proposed methods. Finally, a conclusion is given in Section IV. II. METHODOLOGY A. Overview of Newton Raphson Power Flow NR widely used for solving simultaneous nonlinear algebraic equation by successive approximation procedure Improved Algorithm of Newton Raphson Power Flow using GCC limit based on Neural Network Mat Syai’in #1, 2 , Adi Soeprijanto #1 , #1 Department of Electrical Engineering, Sepuluh Nopember Institute of Technology (ITS) Surabaya-Indonesia #2 Department of Marine Electrical Engineering, Surabaya Shipbuilding State Polytechnic (PPNS-ITS) Surabaya-Indonesia matt.syaiin@gmail.com P