THE DEVELOPMENT OF BIG DATA PLATFORM FOR PREDICTING THE PERFORMANCE AND OPERATION OF BATTERY ENERGY STORAGE SYSTEM
Battery Energy Storage System (BESS) is a technology that utilizes a collection of battery cells to meet the needs of the electricity network. One of the components of BESS is Battery Manegement System (BMS) that is meant to predict operating parameters and battery performance such as State of Charg...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/79410 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Battery Energy Storage System (BESS) is a technology that utilizes a collection of battery cells to meet the needs of the electricity network. One of the components of BESS is Battery Manegement System (BMS) that is meant to predict operating parameters and battery performance such as State of Charge (SoC) and State of Health (SoH). The large number of cells in BESS makes the prediction more complex. Big data technology can be a solution to solve these challenges.
Utilizing big data technology, using two architectures namely the Smart Grid Architectural Model (SGAM) and the Cross Industry Standard Process for Data Mining (CRISP DM) so that it becomes a big data platform. The big data platform ranges from data sources to SoC and SoH predictions. Several previous studies have discussed this topics but have not discussed the relationship between the infrastructure resources of the big data platform and the performance of the big data platform in the form of its execution time.
In this final project, a data platform was developed for SoC and SoH predictions with research objects in the form of fifteen battery cells. The results of the Final Project show that the SoC prediction model produces an error of 2.2% using the Artificial Neural Network (ANN) method an of 2.3% for SoH predition using the Recurrent Neural Network (RNN) method. Both values are still within the standard 5% error for a battery. Furthermore, it is also found that the infrastructure resources of the big data platform in the form of CPU has power function of ????=1557.8 ?????0.269, whilst RAM resembles linear function of ????=?3????+1168. It is also concluded that performance difference between two system configurations form linearity with ????=0.0002?????29.558y. |
|---|