This work presents a decision making methodology which is developed for the effective energy exchange between off-grid stations with Renewable Energy Sources (RES) for power generation. When multi-source energy stations (nodes) are connected, a network is formed where the decisions should be made according to the overall supply and demand of energy. Therefore, each station needs to be aware of the status of the network and to be able to share information related to the availability of its resources and its load demand. The local energy management remains a priority for each individual node whereas part of the load demand can be served when an excess of energy exists in another node of the network. Each node is able to exchange information with the monitoring and control station that supervises the operations of the entire network as part of the operations domain of the grid.
The main operation objectives are to exchange power based on demand response strategies, to maximize the usage of available stored power at network level and to utilize the total amount of available renewable energy. The desired features for the overall behaviour of the network is to ensure the security of supply, to reroute the energy based on dynamically evolving conditions and to provide an automatic decision support mechanism based on available network resources. The presented methodology facilitates the deployment of a flexible modular decision-making mechanism for networked systems that rely on a multi-agent communication structure using an Internet of Things (IoT) architectural reference model. The rules of operation are represented by propositional logic and the implementation is realized by an agent-based structure, which is executed at each node. Finally the raw data of each node are transformed into useful information which is considered by the agent and the algorithm which is responsible for the energy management within the network. The decision making framework is able to gather and act on energy and resource-related information in an automated fashion with the goal to improve the efficiency, reliability and sustainability of the production and distribution of energy. The energy sharing is based on a load balancing approach with priorities that are dynamically updated based on the status of each node and the internal handling of its energy resources and local load demand. Although this methodology is shown by a case study which is related to the energy domain, it can be easily extended to other networked systems where exchange of resources is necessary.
The decision making methodology is developed with implementation considerations for the operation of a smart grid that involves three autonomous stations with local energy storage capabilities using a lead-acid battery stack. Furthermore, there are hydrogen generation and long term hydrogen storage options. Each station has a PV array, wind generators and a diesel generator for backup. Also at one of the stations there is a Polymer Electrolyte Membrane (PEM) fuel cell and a PEM electrolyzer for the utilization and production of hydrogen respectively. In order to make this network as self-sustained as possible the main target is to minimize the use of the diesel generator. Therefore, the priority for the energy exchange is firstly to serve the local load demand and secondly to avoid the use of the generator by utilizing the available energy in the network. The protection of the subsystems within each station is also considered as respective indications for the status of the energy storage is available to the supervisory control station. This isolated grid is located at Xanthi, Greece and its operation is monitored using a Supervisory Control and Data Acquisition (SCADA) system while the Machine to Machine (M2M) communication for the decision making is implemented by the IoT enabled architecture using the agent-based middleware. The performance of the involved nodes is explored and the overall operation of the grid is presented whereas the response of each node is monitored online. An industrial grade agent-based IoT architecture can increase the resiliency and flexibility of such networks whereas the dynamic rerouting of energy can significantly improve the robustness and optimum use of the available resources.