A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia
Maximum Demand (MD) management is essential to help businesses and electricity companies saves on electricity bills and operation cost. Among different MD reduction techniques, demand response with battery energy storage systems (BESS) provides the most flexible peak reduction solution for various m...
| Main Author: | |
|---|---|
| Format: | Thesis (University of Nottingham only) |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://eprints.nottingham.ac.uk/73229/ |
| _version_ | 1848800769846804480 |
|---|---|
| author | Ng, Rong Wang |
| author_facet | Ng, Rong Wang |
| author_sort | Ng, Rong Wang |
| building | Nottingham Research Data Repository |
| collection | Online Access |
| description | Maximum Demand (MD) management is essential to help businesses and electricity companies saves on electricity bills and operation cost. Among different MD reduction techniques, demand response with battery energy storage systems (BESS) provides the most flexible peak reduction solution for various markets. One of the major challenges is the optimization of the demand threshold that controls the charging and discharging powers of BESS. To increase its tolerance to day-ahead prediction errors, state-of-art controllers utilize complex prediction models and rigid parameters that are determined from long-term historical data. However, long-term historical data may be unavailable at implementation, and rigid parameters cause them unable to adapt to evolving load patterns. Hence, this research work proposes a novel incremental DB-SOINN-R prediction model and a novel dynamic two-stage MD reduction controller. The incremental learning capability of the novel DB-SOINN-R allows the model to be deployed as soon as possible and improves its prediction accuracy as time progresses. The proposed DB-SOINN-R is compared with five models: feedforward neural network, deep neural network with long-short-term memory, support vector regression, ESOINN, and k-nearest neighbour (kNN) regression. They are tested on day-ahead and one-hour-ahead load predictions using two different datasets. The proposed DB-SOINN-R has the highest prediction accuracy among all models with incremental learning in both datasets. The novel dynamic two-stage maximum demand reduction controller of BESS incorporates one-hour-ahead load profiles to refine the threshold found based on day-ahead load profiles for preventing peak reduction failure, if necessary, with no rigid parameters required. Compared to the conventional fixed threshold, single-stage, and fuzzy controllers, the proposed two-stage controller achieves up to 6.82% and 306.23% higher in average maximum demand reduction and total maximum demand charge savings, respectively, on two different datasets. The proposed controller also achieves a 0% peak demand reduction failure rate in both datasets. The real-world performance of the proposed two-stage MD reduction controller that includes the proposed DB-SOINN-R models is validated in a scaled-down experiment setup. Results show negligible differences of 0.5% in daily PDRP and MAPE between experimental and simulation results. Therefore, it fulfilled the aim of this research work, which is to develop a controller that is easy to implement, requires minimal historical data to begin operation and has a reliable MD reduction performance. |
| first_indexed | 2025-11-14T20:56:50Z |
| format | Thesis (University of Nottingham only) |
| id | nottingham-73229 |
| institution | University of Nottingham Malaysia Campus |
| institution_category | Local University |
| language | English |
| last_indexed | 2025-11-14T20:56:50Z |
| publishDate | 2023 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | nottingham-732292023-07-22T04:40:19Z https://eprints.nottingham.ac.uk/73229/ A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia Ng, Rong Wang Maximum Demand (MD) management is essential to help businesses and electricity companies saves on electricity bills and operation cost. Among different MD reduction techniques, demand response with battery energy storage systems (BESS) provides the most flexible peak reduction solution for various markets. One of the major challenges is the optimization of the demand threshold that controls the charging and discharging powers of BESS. To increase its tolerance to day-ahead prediction errors, state-of-art controllers utilize complex prediction models and rigid parameters that are determined from long-term historical data. However, long-term historical data may be unavailable at implementation, and rigid parameters cause them unable to adapt to evolving load patterns. Hence, this research work proposes a novel incremental DB-SOINN-R prediction model and a novel dynamic two-stage MD reduction controller. The incremental learning capability of the novel DB-SOINN-R allows the model to be deployed as soon as possible and improves its prediction accuracy as time progresses. The proposed DB-SOINN-R is compared with five models: feedforward neural network, deep neural network with long-short-term memory, support vector regression, ESOINN, and k-nearest neighbour (kNN) regression. They are tested on day-ahead and one-hour-ahead load predictions using two different datasets. The proposed DB-SOINN-R has the highest prediction accuracy among all models with incremental learning in both datasets. The novel dynamic two-stage maximum demand reduction controller of BESS incorporates one-hour-ahead load profiles to refine the threshold found based on day-ahead load profiles for preventing peak reduction failure, if necessary, with no rigid parameters required. Compared to the conventional fixed threshold, single-stage, and fuzzy controllers, the proposed two-stage controller achieves up to 6.82% and 306.23% higher in average maximum demand reduction and total maximum demand charge savings, respectively, on two different datasets. The proposed controller also achieves a 0% peak demand reduction failure rate in both datasets. The real-world performance of the proposed two-stage MD reduction controller that includes the proposed DB-SOINN-R models is validated in a scaled-down experiment setup. Results show negligible differences of 0.5% in daily PDRP and MAPE between experimental and simulation results. Therefore, it fulfilled the aim of this research work, which is to develop a controller that is easy to implement, requires minimal historical data to begin operation and has a reliable MD reduction performance. 2023-07-22 Thesis (University of Nottingham only) NonPeerReviewed application/pdf en cc_by https://eprints.nottingham.ac.uk/73229/1/Thesis%20-%20NG%20RONG%20WANG_Corrected.pdf Ng, Rong Wang (2023) A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia. PhD thesis, University of Nottingham. electricity power; maximum demand; battery energy storage system; power supply |
| spellingShingle | electricity power; maximum demand; battery energy storage system; power supply Ng, Rong Wang A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia |
| title | A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia |
| title_full | A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia |
| title_fullStr | A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia |
| title_full_unstemmed | A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia |
| title_short | A novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in Malaysia |
| title_sort | novel dynamic maximum demand reduction controller of battery energy storage system for educational buildings in malaysia |
| topic | electricity power; maximum demand; battery energy storage system; power supply |
| url | https://eprints.nottingham.ac.uk/73229/ |