Energy Storage System Control Scheme based on Pretrained Adaptive Dynamic Programming for PSOs

With the increasing requirement of power the storage of energy becomes more and more challenging. To tackle this problem control schemes for energy storage systems are introduced. Over the years, various techniques were implemented in control schemes to improve its performance. In recent years, machine learning is widely used in control schemes for energy storage systems, one of those machine learning techniques is Adaptive Dynamic Programming (ADP). ADP is a type of algorithm that allows the subject to improve its behavior in an unknown domain while interacting over a period of time. ADP is an online self-learning algorithm that uses the data collected over a period of time to improve itself. This study proposes an optimal energy control scheme for energy storage systems using ADP. The algorithm will take the parameters such as load demand, power consumed in the time period, and the cost of electricity in the given location as its input and the algorithm’s final objective is to improve the performance index of the energy storage system, the better the performance index the lower the electricity cost. This method will provide an optimum performance index for the energy storage system based on the input parameters presented to the algorithm. To improve the efficiency and to reduce the time consumption for self-learning of ADP algorithm pretraining was required. To retrain the ADP algorithm another algorithm is used i.e. Particle Swarm Optimization (PSO) was used. The PSO algorithm can find the optimum solution in a defined solution space. The pretrained ADP model provides an optimum performance index for the energy storage systems which reduces the cost of electricity.