Short Term Load Forecasting of Distribution Feeder Using Artificial Neural Network Technique

This paper explains the load forecasting technique for prediction of electrical load at Hawassa city. In a deregulated market it is much need for a generating company to know about the market load demand for generating near to accurate power. If the generation is not sufficient to fulfill the demand, there would be problem of irregular supply and in case of excess generation the generating company will have to bear the loss. Neural network techniques have been recently suggested for short-term load forecasting by a large number of researchers. Several models were developed and tested on the real load data of a Finnish electric utility at Hawassa city. The authors carried out short-term load forecasting for Hawassa city using ANN (Artificial Neural Network) technique ANN was implemented on MATLAB and ETAP. Hourly load means the hourly power consumption in Hawassa city. Error was calculated as MAPE (Mean Absolute Percent-age Error) and with error of about 1.5296 % this paper was successfully carried out. This paper can be implemented by any intensive power consuming town for predicting the future load and would prove to be very useful tool while sanctioning the load.


Introduction
The most used thing in today's world is energy. We use energy in various forms in our day to day life, electricity, electricity, solar energy, wind energy, chemical energies in form of batteries and many other forms. Sometimes we are extravagant and sometimes we are careful. But to provide users uninterrupted supply of electricity there must be proper evaluation of present day and future demand of power. That's why we need a technique to tell us about the demand of consumers and the exact capability to generate the power and this need load forecasting techniques [1].
Electric load forecasting is the process used to forecast future electric load, given historical load, weather information along with current and forecasted weather information. Load forecast accurate models for electric power load forecasting are essential to the operation and planning of a utility company. Load forecasting helps an electric utility to make important decisions including decisions on purchasing and generating electric power, load switching, and infrastructure development. Load forecasts are extremely important for energy suppliers, financial institutions, and other participants in electric energy generation, transmission, distribution, and markets. The Energy Management System demands accurate load forecasting and short term Load Forecasting provides better and truthful results [2].
Load forecasting is however a difficult task as the load series is complex pattern and exhibits several levels of system seasonably.
Several conventional techniques had been used for the load forecasting. However the advantages of those techniques have led to the use of the artificial intelligence techniques [3]. An artificial neural network (ANN) is the piece of a computing system designed to simulate the way the human brain analyzes and processes information. It is the foundation of artificial intelligence (AI) and solves problems that would prove impossible or difficult by human or statistical standards. And the artificial neural network is an attempt to simulate the network of neurons that make up a human brain so that the computer will be able to learn things and make decisions in a humanlike manner. ANNs are created by programming regular computers to behave as though they are interconnected brain cells [4].
Short-term electrical load forecasting is vital for the efficient operation of electric power systems. A power grid integrates many stakeholders who can be affected by an inaccurate load forecast. Power generation utilizes 24-hour or 48-hour ahead forecasts for operations planning, i.e. to determine which power sources should be allocated for the next 24 h transmission grids need to know in advance the power transmission requirements in order to assign resources end users utilize the forecast to calculate energy prices based on estimated demand. Contingency planning load shedding management strategies and commercialization strategies are all influenced by load forecasts. Forecast errors result in increased operating costs [2]. predicting a lower load than the actual load results in utilities not committing the necessary generation units and therefore incurring higher costs due to the use of peak power plants ;on the other hand, predicting a higher load than actual will result in higher costs because unnecessary baseline units are started and not used. Reliable forecasting methods are essential for scheduling sources and load management [5]. The key function of an electric power company is to supply customers with high quality electric energy in secured and economical manner. In order to do so, an electric power company faces economical and technical problems in planning, control and operation of electric power system. For the purpose of optimal planning and operation of electric power system, there is need for proper evaluation of present day and future electric power load [1,6].
Load forecasting has always been important for planning and operational decision conducted by utility companies. However, with the deregulation of the energy industries, load forecasting is even more important Load forecasting for vision load demand prerequisite is the most imperative key for power system planning. The ability of the generation, transmission and distribution capacities are strictly dependent on the precise energy and load forecasting for that system. Power system expansion planning starts with a forecast of anticipated future load requirements. Estimates of both demand and energy requirements are crucial to valuable system planning [7]. An electric load forecasting is used by an electric power company to anticipate the amount of electric energy needed to supply so as to meet up the demand [8].

Theoretical Background
Load forecasting in power system is an important subject and has been studied from different point of view in order to achieve better load forecasting results [9]. Techniques such as regression analysis, expert system, artificial neural network and multi-objective evaluations have been used based on different choices of inputs and available information. Distribution system load forecasting has been challenging problem due to its spatial diversity and sensitivities to land usage and customer habits.
Different tools have been developed to assist utilities to simulate and estimate the future land, usage land and load growth in their territory, so that distribution system planners can plan according to their goal and interests. Many factor need to be considered for this purpose, namely.
• Land usage in the future Electric load forecasts can be divided into three categories based on the planning perspective of the duration: • Short Term Load Forecasts (STLF): This is usually from one hour to one week.
• Medium Term Load Forecasts (MTLF): This is usually from one week to a year.
• Long Term Load Forecasts (LTLF): This is longer than a year.
Short-term load forecasting at the distribution level predicts the load of substations, feeders, transformers, and possibly customers with a typical forecasting horizon ranging from half an hour to one week. High quality load forecasting is important for the planning and operation of distribution systems. Fig. 1 presented the view of Hawassa distribution substation. Journal of Informatics Electrical and Electronics Engineering (JIEEE) A2Z Journals system, there is need for proper evaluation of present status and future electric power load. Other problems, which are faced by the system like over loading and under loading, no proper power arrangement and the capacity of power distribution substation is not equal to power demand. Ref. [8] focused on the study of electrical power load at Hawassa city distribution feeder and the appropriate forecasted power loads by studying the short term forecasting by using ANN method. Also it develops the load curve on the 24 hours [8]. Ref. [9] was introduced the efficient approach to short-term load forecasting at the distribution level. Ref. [10] was presented on the title of short term load forecasting using a hybrid model based on support vector regression. Ref. [11], introduced on the title of Short Term Load forecasting using SVM based PUK kernel. In this paper, Support Vector Machine (SVM) model based on Pearson VII universal kernel known as PUK kernel has been proposed for Short Term Load Forecasting (STLF). Also it was introduced the Machine learning algorithms have been applied in two parts. The performance is verified with simulation on a case study based on 6 weeks' load pattern of distributed grid station in Pakistan. The forecasting results of PUK based SVM model are proved to be better than the other models and indicate that the forecasting precision of the proposed methods is much efficient to other methods.

Characteristics of Power System Load
In this paper a number of techniques are included and analyses the implementation of the following methodologies [12][13][14][15][16].
• Test data input • Neural network formation • Train neural network • Next iteration of artificial neural network • Error calculation • Error comparison • Output of neural network • Load prediction of desired period In general forecasting methods can be divided into two broad categories: Parametric methods and artificial intelligence based methods. Based on analyzing qualitative relationships between the load and the factors affecting the load, the parametric methods formulate mathematical or statistical models of load. Then the parameters of the built model are estimated from historical data.

Data Collection and Preparation
To perform this work, the historical data (load data and temperature) were collected from Hawassa city. The historical load data of the distribution feeder was obtained from the transmission and distribution office of Hawassa city and the temperature was obtained from the internet, and Hawassa city metrology agency office. The data collected is the actual load data of the 132/33/15 kV substation, more focusing in the 33kV and 15 kV in the Hawassa city for the different month of 2019 Gagarin calendar (2011 Ethiopian calendar). The load data of first 21 days of a month and temperature will be used for training the network and the load data and temperature for the remaining days in the month will be used for the network validation.

Input Data
The input data is of one week winter (02/05/2011 to 08/05/2011), One week summer (25/12/2011 to 01/11/2011) and One Holiday of 2011 EC of electrical load of the distribution feeder. Load data was recorded for Hawassa city, so there is 15 days of data are selected from a year. This data was used for training and testing of the 24-hour load forecasting using ANN. The load data was collected from a database containing data recorded from power meters located at the Hawassa city substation and organization office. Table 1 presented the Hawassa city distribution feeder name with its rating. The output of summer and winter measured data from typical electrical load profile of Hawassa City for Monday is presented by the figures 2 and 3. Table 4 presented the Maximum load profile for distribution feeders (DF).

Factors Affecting Short Term Load Forecasting
These factors can be categorized as Time factor, weather, economy, Humidity and random disturbances. In this research paper these factors and their impact on consumption of electric power and their significance in short term load forecasting is evaluated. Generally the factors affecting to the load forecasting are concluded in the figure 4.

Artificial Neural Network
An artificial neural network (ANN) is the piece of a computing system designed to simulate the way the human brain analyzes and processes information. It is the foundation of artificial intelligence (AI) and solves problems that would prove impossible or difficult by human or statistical standard [13,[17][18][19][20].
The use of artificial neural networks (ANN or simply NN) has been a widely studied electric load forecasting technique. Neural networks are essentially non-linear circuits that have the demonstrated capability to do non-linear curve fitting. Artificial Neural Networks are mathematical tools originally inspired by the way human brain process information. It consists of an interconnected group of artificial neurons and processes information using a connectionist approach to computation. Artificial Neural Networks (ANN) is a soft technique used in various optimization processes. This method is able to perform non-linear modeling and adaptation. It does not require assumption of any functional relationship between load and weather variables in advance.
The most popular artificial neural network architecture for electric load forecasting is back propagation. Back propagation neural networks use continuously valued functions and supervised learning.

ANN Forecasting Using Model
The developed forecasting model which considers temperature and humidity as input data since the load demand does not depend only on temperature and humidity and in order to account for other factors. Figure 5 presented the model arrangement of ANN for load forecasting.

Mathematical Model of a Neuron
A neuron is an information processing unit that is fundamental to the operation of a neural network. The three basic elements of the neuron model are: 1. A set of weights, each of which is characterized by a strength of its own. A signal xj connected to neuron k is multiplied by the weight wkj. The weight of an artificial neuron may lie in a range that includes negative as well as positive values.
2. An adder for summing the input signals, weighted by the respective weights of the neuron.
3. An activation function for limiting the amplitude of the output of a neuron. It is also referred to as squashing function which squashes the amplitude range of the output signal to some finite value The architecture of the ANN comprises of: • Input layer: Contains neurons equal to the number of inputs.
• Hidden layer(s): The number of hidden layers and the number of neurons in each layer depends on the complexity.
• Output layer: Usually it has one neuron, and its output ranges from 0 to 1, that is, greater than 0 and less than 1. But multiple outputs can be present.

Network Structure
Network is consists of an input layer, a hidden layer, and an output layer. The choice of one hidden layer is a combination of historical success with acceptable predictions. The architecture of the ANN comprises of • Output layer: Usually it has one neuron, and its output ranges from 0 to 1, that is, greater than 0 and less than 1. But multiple outputs can be present [15,[21][22][23][24][25][26][27][28][29][30].

Journal of Informatics Electrical and Electronics
Engineering (JIEEE) A2Z Journals

Performance Metrics
There are a number of error measurements that are relevant for quantifying the performance of the model. The most widely reported error in neural network literature is the MAPE, given in the below equation, where, A is a (1xN) set of actual values, F is a (1xN) set of forecast values, and N is the number of points being forecast, which in this case is 24.

ANN Evaluation Process
Contrary to feed forward networks, Recurrent Neural Networks (RNNs) are models with bidirectional data flow. While a feed forward network propagates data linearly from input to output, RNNs also propagate data from later processing stages to earlier stages. RNNs can be used as general sequence processors. Figure 6 presented the evaluation process of the ANN.

Calculation of absolute error
This method has been implemented in MATLAB. The forecasting performance is evaluated by using the standard mean absolute percentage error (MAPE).
where, L(t) is actual data and La(t) is predicted data

Short Term Load Forecasting In Hawassa City
Short term load forecasting is basically is a load predicting system with a leading time of one hour to seven days, which is necessary for adequate scheduling and operation of power systems. For proper and profitable management in electrical utilities, short-term load forecasting has lot of importance [31][32][33][34][35][36][37][38].
Some of the techniques that have been proposed and implemented to create STLF for the planning of distribution feeder are: • Similar-day approach

Factors
Amongst several factors which impact consumption, there are two high level components which are highly significant [39][40][41][42][43][44][45]: • Seasonality • Weather These factors can be expressed as: where, C is consumption, S is season/TOD, W is weather, and ε is error or residue.

Mean Absolute percentage error (MAPE)
The

Results and Discussion
In this section, we make use of the empirical electric load data of Hawassa city to examine the performance of the proposed methods by using the ANN. The short term load forecasting is performed for planning of the distribution feeder. The 24 hours actual versus predicted ANN based load forecast for output in MW for feeder A is shown in figure 7. Figure 8 presented the simulation results of the feeder A. result will be not zero that means it has forecasted result around 5.43MW and the maximum load forecasting for the feeder A is 18.9 MW at the time of 20:00 the forecasting power feeder of the year is more accurate because the MSE and MAPE will be level value that means MAPE is1.0e+04 , 0.0024, 1.5296. Figure 9 presented the distribution feeder A to D, actual & forecasting values comparison. we see the 11 th hour the actual power in the feeder will be large amount but the forecasted value will be less than the actual value. During this time the large amount of power will be save because the forecasted process input will be consid-   Table 5 presented the summer and winter time holidays load forecasting. The actual and forecasted distribution feeder load comparison for holiday summer season is presented by the figure 16. Figure   17 presented proposed technique. Therefore, it is important to plan the distribution feeder at the substation level. Figure 18 presented the actual and forecasting load of distribution feeder load comparison for holiday winter season. Figure 19 presented the simulation result of the feeder on the holiday winter season. The forecasted result will be more standardized because of the result relation between actual and forecasted value and the MAPE of error will be under the IEEE standard. The distribution feeder at the 18 th hour had actual data less than the maximum forecasted result which is around 20.2MW. It means for the summer time the distribution feeder required to distribute the forecasted amount of power according to the MATLAB simulation. It is important to plan the distribution feeder at the substation level. Table 6 presented the MAPE for the feeder on the case study. Table 7 presented the distribution feeder load forecast.

Economic Analysis
Economic analysis of the forecasting distribution feeder to calculate the cost of each forecasted feeder based on the forecasting deviation (FD) and distribution feeder MAPE are calculated based on the simulation result output. Figure 8 presented the economic analysis of the forecasted feeders.

Conclusion
The main purpose of this study is to investigate an intelligence method for the short term load forecasting, by using three layer feed-forward and back-propagation neural networks for Hawassa city, Ethiopia. It takes into account the effect of the amount of period of disconnected time of load. The results show that neural networks can be considered an efficient and applicable method to short term load forecasting. Its forecasting reliabilities were evaluated by computing the mean absolute error between the exact and predicted values and compare the result of mean absolute error between the three neural networks. The results suggest that ANN model with the developed structure can perform good prediction with least error. Finally, this neural network could be an important tool for the short term load forecasting. Results also show that a simple ANN-based prediction model, appropriately tuned, can outperform other more complex models.