Abstract
Battery energy storage systems can assist distribution network operators (DNOs) to face the challenges raised by the substantial increase in distributed renewable generation. A challenge is that these resources are intermittent and often ‘invisible‘ to the DNO. If not monitored, the aggregate size of small embedded generation resources can cause thermal wearing of distribution assets and voltage excursions, especially in sunny/windy periods with insufficient local demand. Several developers of energy storage solutions, with technologies such as lithium-ion (Li-ion) batteries, offer their products to address peak shaving, frequency and voltage control needs within the network. Once deployed within the energy network batteries experience capacity degradation with usage, these companies will need to incorporate methods from prognostics and health management (PHM) in order to better manage their products. The main deliverable of this project is validation of data analysis, based on relevance vector machine, to predict the remaining useful life of Li-ion batteries. The accuracy of the predictions for different batteries is all within 10 cycles (within 8.5% relative error). These results confirm the importance of PHM methods within a distribution system operator model, where lifecycle management of critical sub-systems and systems will become increasingly important to network operators.
Highlights
Recent years have seen large penetration of distributed energy resources (DERs) in medium-voltage (MV) and low-voltage (LV) distribution grids, as part of decarbonisation agenda of the power generation sector
We present this concept with preliminary experimental results using the state-of-art machine learning technique, relevance vector machine (RVM), to predict the remaining useful life (RUL) of Li-ion batteries
Energy storage systems are expected to play a key role in energy systems and may help distribution network operators (DNOs) address the challenges introduced by the development of renewable resources
Summary
Recent years have seen large penetration of distributed energy resources (DERs) in medium-voltage (MV) and low-voltage (LV) distribution grids, as part of decarbonisation agenda of the power generation sector. Frequency control is supported by expensive peaking generators providing ‘spinning reserve’ and voltage is kept within acceptable technical limits by the operation of capacitor banks or on-load-tap-changers (OLTC). Frequent use of these measures though may reduce the lifetime of this equipment. As the physical mechanisms that lead to ‘cell ageing’ are complex to understand, we need to develop tools that can monitor the asset’s health and predict failure, in order to increase the reliability and resilience of the overall system This drives us to develop a more integrated and holistic energy storage management system, using historical data and machine learning techniques. Due to its predictive capability, the proposed algorithm is expected to be used in further work in the asset management system for large scale energy storage networks
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
