Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN)
Artificial Neural Network (ANN) in Mixed-Integer Linear Programming (MILP) technique for load scheduling of appliances in a single smart home system. This proposed method is achieved through backpropagation method of ANN tools in MATLAB which is the central mechanism by which neural networks learn t...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/93120/1/SitiHajarJoharryMSKE2020.pdf |
| _version_ | 1796865559153868800 |
|---|---|
| author | Joharry, Siti Hajar |
| author_facet | Joharry, Siti Hajar |
| author_sort | Joharry, Siti Hajar |
| collection | ePrints |
| description | Artificial Neural Network (ANN) in Mixed-Integer Linear Programming (MILP) technique for load scheduling of appliances in a single smart home system. This proposed method is achieved through backpropagation method of ANN tools in MATLAB which is the central mechanism by which neural networks learn to predict the next day load consumption of a home and promptly inserting the output to the MILP which would optimize the process of load scheduling. The integration of ANN with MILP can contribute to the precision of load scheduling. Having said that, to obtain the day ahead energy consumption, the annual data of the home is extract and injected in ANN as input and target classes. Hence, with the process of backpropagation, energy consumption is predicted while taking into consideration the Mean Squared Error (MSE) of the model. This prediction is then incorporated in the programming of MILP for optimization of load scheduling. The performance of the model is then evaluated by comparing before and after the optimization process. A total load of the appliance has been reduced from 51.24 kW/day to 44.84 kW/day. Furthermore, the overall cost of the electricity bill has been reduced from $3.98/day to $2.45/day. Therefore, the deduction of 38.44% of electricity bills makes the proposed method notably applicable and best to use in real time situation. |
| first_indexed | 2024-03-05T20:58:51Z |
| format | Thesis |
| id | utm.eprints-93120 |
| institution | Universiti Teknologi Malaysia - ePrints |
| language | English |
| last_indexed | 2024-03-05T20:58:51Z |
| publishDate | 2020 |
| record_format | dspace |
| spelling | utm.eprints-931202021-11-19T03:23:51Z http://eprints.utm.my/93120/ Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) Joharry, Siti Hajar TK Electrical engineering. Electronics Nuclear engineering Artificial Neural Network (ANN) in Mixed-Integer Linear Programming (MILP) technique for load scheduling of appliances in a single smart home system. This proposed method is achieved through backpropagation method of ANN tools in MATLAB which is the central mechanism by which neural networks learn to predict the next day load consumption of a home and promptly inserting the output to the MILP which would optimize the process of load scheduling. The integration of ANN with MILP can contribute to the precision of load scheduling. Having said that, to obtain the day ahead energy consumption, the annual data of the home is extract and injected in ANN as input and target classes. Hence, with the process of backpropagation, energy consumption is predicted while taking into consideration the Mean Squared Error (MSE) of the model. This prediction is then incorporated in the programming of MILP for optimization of load scheduling. The performance of the model is then evaluated by comparing before and after the optimization process. A total load of the appliance has been reduced from 51.24 kW/day to 44.84 kW/day. Furthermore, the overall cost of the electricity bill has been reduced from $3.98/day to $2.45/day. Therefore, the deduction of 38.44% of electricity bills makes the proposed method notably applicable and best to use in real time situation. 2020 Thesis NonPeerReviewed application/pdf en http://eprints.utm.my/93120/1/SitiHajarJoharryMSKE2020.pdf Joharry, Siti Hajar (2020) Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN). Masters thesis, Universiti Teknologi Malaysia, Faculty of Engineering - School of Electrical Engineering. http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:135986 |
| spellingShingle | TK Electrical engineering. Electronics Nuclear engineering Joharry, Siti Hajar Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) |
| title | Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) |
| title_full | Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) |
| title_fullStr | Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) |
| title_full_unstemmed | Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) |
| title_short | Load scheduling for smart home using day-ahead prediction from Artificial Neural Network (ANN) |
| title_sort | load scheduling for smart home using day ahead prediction from artificial neural network ann |
| topic | TK Electrical engineering. Electronics Nuclear engineering |
| url | http://eprints.utm.my/93120/1/SitiHajarJoharryMSKE2020.pdf |
| work_keys_str_mv | AT joharrysitihajar loadschedulingforsmarthomeusingdayaheadpredictionfromartificialneuralnetworkann |