IJSRD || National Conference on Emerging Trends, Challenges & Opportunities in Power Sector || March 2017 || ISSN: 2321-0613 ©IJSRD 2017 Published by IJSRD 1 Short Term Load Forecasting using Neuro fuzzy Gunjan Dave 1 Sweta J. Shah 2 Yogesh Patel 3 Merolina Christie 4 Urvish Mewada 5 1 Student 2,3,4,5 Assistant Professor 1,2,3,4,5 Department of Electrical Engineering 1,3,4,5 Ahmedabad Institute of Technology, GTU, India 2 IITE, Indus University Abstract— Optimal day to day operation of electric power generating plant is very essential for any power utility constitution to reduce input costs and the prices of electricity. So, to generate reasonably the required power, one needs to forecast the future electricity demands since power generation relies heavily on the electricity demand. Load forecast has three different horizons: short term forecast, medium term forecast and long term forecast. The Short-Term Load Forecasting (STLF) provides information for utility program system planners so that they can come up with a short-term solution to protect the transmission and distribution systems and to better serve the clients. This article presents the development of Adaptive Neuro Fuzzy Interface System (ANFIS) based short-term load forecasting model. The fusion of neural networks and fuzzy logic in neuro-fuzzy models achieves readability and learning ability at once. This article presents prediction of electric load by considering various information like time, temperature, humidity, day and historical load data. Historical load data is taken from MGVCL and weather data is taken from the website www.timeanddate.com. Key words: Adaptive Neuro-Fuzzy Interface System, Load Forecasting, Short Term Load Forecasting I. INTRODUCTION There is a growing trend towards unbundling the electricity system. This is continually confronting the different sectors of the industry (generation, transmission, and distribution) with increasing need on plan management and operations of the network. The operation and planning of a power utility company requires an adequate model for electric power load forecasting. Load forecasting plays a key role in serving an electric utility to make important decisions on power, load switching, electric potential control, network reconfiguration, and infrastructure development. [1] It helps in deciding and planning for maintenance of the power systems. By understanding the demand, the utility can know when to carry out the maintenance and ensure that it has the lower limit impact on the consumers. [2] Load forecasting is the predicting of electrical power required to meet the short term, medium term or long term demand. The reasons why businesses need STLF include energy purchasing, unit commitment, reduce spinning reserve capacity, T&D (transmission and distribution) operations and demand side management. Forecasted values are sent to day ahead planning system by demand side one day in advance. [3] This article presents the development of soft computing based short-term load forecasting model which forecast the electric load. Traditional methods have the inherent inaccuracy of load prediction and numerical instability. Further, the non-stationarity of the load prediction process, coupled with complex relationship between weather variables and the electric load render such traditional techniques ineffective as these methods assume simple linear relationships during the prediction process. [4] The prime inherent advantage associated with the soft computing techniques of not requiring a mathematical model has been a motivating factor for consideration in our present work. In soft computing technique fusion of neural networks and fuzzy logic in neuro-fuzzy models achieves readability and learning ability at once. So, this article presents the development of an Adaptive Neuro Fuzzy Inference System (ANFIS) based short-term load forecasting model which forecast the electric load. II. BASICS OF ANFIS The model obtained with neural network is not understandable in terms of physical parameters (black box model) and it is impossible to interpret the result in terms of natural language. On the other hand, the fuzzy rule base consists of if-then statements that are almost natural language, but it cannot learn the rules itself. To obtain a set of if-then rules two approaches are used. First, transforming human expert knowledge and experience, and second, automatic generation of the rules. The second method is intensively investigated. The fusion of neural networks and fuzzy logic in neuro-fuzzy models achieves readability and learning ability (extracting rules from data) at once. On 1993, Roger Jang [5] developed the ANFIS technique that could overcome the shortcoming of the ANNs and fuzzy systems. Neuro-fuzzy approaches have been widely applied to the short-term load forecasting (STLF). Adaptive Neuro-Fuzzy based Inference System (ANFIS), an integrated system, comprising of Fuzzy Logic and Neural Network can address and solve problems related to non-linearity, randomness and uncertainty of data. In this article the ANFIS model to STLF is presented. The fuzzy part of the ANFIS is constructed by means of input and output variables, membership functions, fuzzy rules and inference method. The training inputs are also called energy drivers and are variables that can affect the output, such as, in case of the energy consumptions: the daily production, the climatic data, the day of the week, etc. The membership functions of the system are the functions that define the fuzzy sets. The fuzzy rules have a form of if-then rule and define how the output must be for a specific value of membership of its inputs. In general, the fuzzy systems have different kind of inference methods but ANFIS is based on a particular type of fuzzy system with Takagi-Sugeno rules as inference method. FIS basically consist of five subcomponents a rule base (covers fuzzy rules), a database (portrays the membership functions of the selected fuzzy rules in the rule base), a decision-making unit (performs inference on selected fuzzy rules), fuzzification inference and defuzzification inference.