Maximizing efficiency of energy consumption forecasting using a machine learning approach / by Mohamed Mohamed Ramadan Ali Alsoul ; Supervised Prof. Hegazy Mohamed Zaher, Prof. Naglaa Ragaa Saeid Hassan, Dr. Eman Mostafa Ahmed.
Material type:
TextLanguage: English Summary language: English, Arabic Producer: 2024Description: 91 Leaves : illustrations ; 30 cm. + CDContent type: - text
- Unmediated
- volume
- تعظيم كفاءة التنبؤ لاستهلاك الطاقة باستخدام منهجية تعلم الآله [Added title page title]
- 006.31
- Issues also as CD.
| Item type | Current library | Home library | Call number | Status | Barcode | |
|---|---|---|---|---|---|---|
Thesis
|
قاعة الرسائل الجامعية - الدور الاول | المكتبة المركزبة الجديدة - جامعة القاهرة | Cai01.18.05.Ph.D.2024.Mo.M (Browse shelf(Opens below)) | Not for loan | 01010110092377000 |
Browsing المكتبة المركزبة الجديدة - جامعة القاهرة shelves Close shelf browser (Hides shelf browser)
| No cover image available | No cover image available | No cover image available | No cover image available | No cover image available | No cover image available | No cover image available | ||
| Cai01.18.05.Ph.D.2024.Al.O On black widow optimization / | Cai01.18.05.Ph.D.2024.If.M. A Multi-Objective De Novo Programming Under Uncertainty / | Cai01.18.05.Ph.D.2024.Is.I An intelligent approach for solving optimization problems / | Cai01.18.05.Ph.D.2024.Mo.M Maximizing efficiency of energy consumption forecasting using a machine learning approach / | Cai01.18.05.Ph.D.2024.Mo.O On Multi-Objective Fractional Programming Problem / | Cai01.18.05.Ph.D.2024.Ra.M A modified harris hawks algorithm for solving optimization problems / | Cai01.18.05.Ph.D.2024.Sa.P A proposed differential evolution algorithm for solving optimization problems / |
Thesis (Ph.D)-Cairo University, 2024.
Bibliography: pages 85-91.
Optimization extensively applied in various disciplines, ranging from engineering, computer science, finance, and decision-making. It is widely regarded as a fundamental technique with diverse applications. In the context of engineering and industry, optimization is employed to reduce costs and energy consumption or to maximize profits, output, performance and efficiency.
Household energy consumption patterns are constantly evolving influenced by variables such as temperature, humidity and time of day. Predicting how much energy each family uses is really hard because everyone uses energy in different ways. Smart meters and home displays are on the rise.
In this thesis, it was applied machine learning models for energy consumption in household and commercial sector, with a specific focus on identifying potential optimizations to reduce energy usage and maximize its efficiency, sustainability and preservation. With the emergence of machine learning, the ability to accurately predict future energy consumption has significantly reached applying two kinds of machine learning models. Computational intelligence methods showed better accuracy than statistical methods, especially for parameters with greater variability in source data. Thus, this thesis introduced two suggested models:
The first model is about "Optimizing Homes Energy Consumption Utilization Using H-R Predictor" which it was applied using REFIT data collected from UK household smart meters to analyze energy consumption.
The suggested prediction H-R algorithm is one of achievements of this thesis which it included two suggested unsupervised machine learning models (LSTM and XGBoost) to train data for forecasting and controlling the energy consumption. The performance of these models was evaluated using R2 as a metric to measure the percentage of variance in the dependent variable that can be predicted.
The second model is about "A hybrid Firefly based Gravity Search Algorithm (F-GSA) for Optimizing Energy Consumption". The proposed model was developed to forecast the energy consumption in homes.
Hybrid firefly-based Gravity Search Algorithm (F-GSA) enhance energy consumption predictions by combining strengths, providing efficient projections and historical data insights, offering a promising solution for tackling energy consumption forecasting challenges.
The research models were assessed using root mean squared error (RMSE), mean absolute deviation (MAD) and R-Squared (R²), also known as the coefficient of determination, in order to directly compare them to the energy readings in the data. MAD is utilized to quantify the error in predicting energy consumption in 10-minute intervals (equivalent to six errors per hour). On the other hand, R² is employed to evaluate the overall performance of the models in predicting energy consumption over a specific time period, such as a day or month.
يتم تطبيق الأمثلية (Optimization) على نطاق واسع في مختلف التخصصات مثل الإقتصاد والهندسة وعلوم الكمبيوتر والمصارف المالية وغيرها بهدفتعظيم الربحية أو تقليل التكاليف والمخاطر أو المزيج بين كلاهما. وتتضمن الامثلية تطبيقات متنوعة ومعقدة ومن ثم تحتوي على عدد كبير من المعلمات (Parameters) هكذا يصعب حل مثل هذه المشكلات بالطرق التقليدية. تقدم خوارزميات التعلم الآلي ML وخوارزميات Metaheuristic حلول جيدة وفعالة قريبة من المثالية من خلال البحث بذكاء فى مساحة الحل لتلك الأنواع من المشكلات.
ومع تطور أنماط استهلاك الطاقة المنزلية باستمرار متأثرة بمتغيرات مثل درجة الحرارة والرطوبة خلال فترات اليوم أو خلال الفترات الموسمية يعتبر التنبؤ بكمية الطاقة التي تستخدمها كل أسرة أمرًا صعبًا لأن الجميع يستخدم الطاقة بطرق مختلفة. يهدف هذا البحث إلى استكشاف مجال استهلاك الطاقة الكهربائية في القطاع المنزلي والتجاري من خلال التعلم الآلي (ML) مع التركيز بشكل خاص على تحديد التحسينات المحتملة لتقليل استخدام الطاقة وتعظيم كفاءتها واستدامتها والحفاظ عليها.
مع ظهور طرق التعلم الآلي واستخدام العدادات الذكية (Smart meters) تحسنت القدرة على التنبؤ باستهلاك الطاقة في المستقبل بدقة عالية وبشكل كبير نتيجه استخدام طرق الذكاء الإصطناعى وتعلم الالة لتدريب النماذج عليها واستخدامها كأداه قوية للتنبؤ حيث تظهر طرق الذكاء الاصطناعى دقة أفضل من الطرق الإحصائية التقليدية وخاصة بالنسبة للمعلمات ذات التباين الأكبر في بيانات المصدر
Issues also as CD.
Text in English and abstract in Arabic & English.
There are no comments on this title.