International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 10 | Oct 2024
p-ISSN: 2395-0072
www.irjet.net
Effective Prediction of Electric Power Consumption using Random Forest and XGBoost Regressor Sasikala M1, Krithik U2, Kishore Kumar B P2, Pandi Selvam P2 1Professor and Faculty Mentor, Dept. of Computer Science and Engineering, K L N College of Engineering,
Pottapalayam, Sivagangai, Tamil Nadu, India. 2B. E Student, Dept. of Computer Science and Engineering, K L N College of Engineering, Pottapalayam, Sivagangai,
Tamil Nadu, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Accurate prediction of electric power
like peak usage hours and extreme weather events to make its results easy to interpret. The system adjusts the calculated values by 10% in case of poor accuracy predictions and by 5% in case of moderate accuracy predictions for optimum accuracy of its predictions so that the produced energy is always kept up with its demand. This adaptive and scalable solution promises a more efficient way of handling power consumption, where energy sustains themselves across different regions and networks.
consumption is very important for efficient energy management and cost savings. Traditional methods are often known to not handle intricate power consumption patterns. In this work, a machine learning-based approach has been followed using the Random Forest and XGBoost Regressor algorithms to predict electric power consumption. The system applies hourly power consumption data combined with weather conditions and seasonal factors like temperature, rainfall, and humidity for predictions. If the accuracy is very low, the system adds 10%, if it is moderate then 5% on the predicted value. Such adjusted predictions it issues as its output; thus, ensuring that energy supplied equals demand. The management of energy, reduction in imbalances of supply, and healthy productive usage of energy will thus be improved.
2. LITERATURE SURVEY [1] Modeling and Forecasting Electricity Consumption Amid the COVID-19 Pandemic: Machine Learning vs. Nonlinear Econometric Time Series Models-Lanouar Charfeddine*, Esmat Zaidan*, Ahmad Qadeib*, Alban Hamdi Bennasr*, Ammar Abulibdeh. The paper compares machine learning and nonlinear econometric models for electricity consumption forecasting using data from the COVID-19 pandemic. It emphasizes how weather factors, like temperature and rainfall, affect energy usage and shows that the best accuracy is provided by machine learning models. However, accuracy can decrease when applying to years with changed weather conditions because it depends on past seasonal data.
Key Words: Random Forest, XGBoost Regressor, Electric Power Consumption Prediction, Machine Learning, Energy Forecasting.
1. INTRODUCTION Accurate electric power usage prediction is crucial for efficient energy management, cost-reducing savings, and promotion of sustainability. However, against the backdrop of the growth in world energy demand, traditional methods for forecasting often cannot cope with the complex factors influencing power consumption as advanced by various meteorological and seasonal and regional factors. Traditional methods for such forecasting are usually less accurate in non-linear usage data of power. These are the challenges that this project addresses by using Random Forest and XGBoost Regressor, two machine learning algorithms for the power consumption forecast. The system considers past records of the power usage together with weather data, comprising elements of temperature and rainfall, which give it predictive accuracy. Powerful pre-processing techniques would overcome outlier and missing data complications during the prediction in order to present reliable outcomes resulting from imperfect data from the real world. It also autonomously detects dominant factors
© 2024, IRJET
|
Impact Factor value: 8.315
[2] Spatio-temporal Granularity Co-optimization Based Monthly Electricity Consumption Forecasting-Kangping Li*, Yuqing Wang*, Ning Zhang*, Fei Wang*, Chunyi Huang. The paper recommends an increase in spatio-temporal granularity for enhanced monthly electricity consumption forecasting. When high-resolution load data from meters is used to optimize forecasting accuracy, the integration of information from space and time benefits results. The method developed for short-term forecasting did not discuss or test it on monthly forecasting. While assuming no noise in the load data, it may limit the application into real-world scenarios. [3] Power Source Flexibility Margin Quantification Method for Multi-energy Power Systems Based on Blind Number Theory-Bai Xiao*, Jialiang Wang*, Zhiwen Xiao*, Gangui
|
ISO 9001:2008 Certified Journal
|
Page 697