Article Preview
Top1. Introduction
Electricity has now become a part of everyone’s life. In India, demand is increasing day by day and also facing problems such as power quality issue, lack of latest technology, losses etc. Stress on existing energy resources has also increased drastically due to increase in electricity demand and thus, leads to shortage of power. Now-a-days demand of CPP is increasing day by day due to its various advantages such as high efficiency, low environmental impact, low investment cost, simple operation and many more.
The load demand prediction and generation of CPP must be accurate. Electrical load Prediction is the process of future load prediction in the blocks of hours, days, weeks, months or for a year. It is mainly done in order to prognosticate the power consumption of electrical utilities and help them to plan their future decisions related to Generation, Transmission and Distribution system. Electrical load prediction also helps in managing demand and supply of electricity at every moment which is managed by utilities. Load forecasting is used for power supply planning, transmission & distribution planning, demand side management, maintenance and financial planning. So accurate load prediction is very essential for the effective working of the power system otherwise it may leads to the equipment’s failure and supply loss (Jaswal, 2013).
An author (Alam et al., 2013; Oak & Patil, 2016) presented case study of the captive power plant in Maharashtra and Bangladesh respectively. A study by (David et al., 2004) has shown that the scenario of captive power plants in India can be broadly assigned to (a) manage the backup power system, (b) checks the quality of the supply, (c) ensures the benefits of co-generation process of industries, and (d) also generate electricity at lower costs than the high industrial tariffs. In view of designing the Captive power generation from coolant jet author (Pooja et al., 2018) suggested some guidelines for executing the captive hydro power plant at low head of coolant jet for developers.
There are various factors which affect load forecasting such as weather variables (temperature, rain, wind and humidity), holidays, festivals, economic growth and new load demand (Gupta et al., 2013). The two main classification of load forecasting techniques are parametric and non-parametric. Examples of parametric technique are auto regressive moving average (ARMA), linear regression, general exponential technique and stochastic time series techniques. These techniques are also known as statistical or traditional techniques. The main limitations of traditional methods are its capability to cover up the changes due to environment. However, this can be is resolved by using non- parametric techniques. This method is also known as modern techniques as it uses artificial intelligence. Modern techniques include Neuro-Fuzzy Method, Artificial Neural network (ANN) method, Genetic algorithm (GA), Fuzzy logic method, and Particle swamp optimization (PSO). Among these methods of artificial intelligence based technique called artificial neural network is appeared to be most appropriate and has received attention of many researchers (Bello & Harrison, 2015; Hippert, Pedreira, & Souza, 2001; Karthikeyan et al., 2019). ANN has the capacity to make decisions for uncertain environment, to solve complex problems, image recognition, and prediction capability. Thus ANN has emerged as a useful techniques as compared to the other traditional techniques. (Abiodun et al., 2018; Araque et al., 2017).