Nowadays, companies and researchers are developing devices that are connected over the Internet to create new services for users through the collection and analysis of data obtained from sensors. The information obtained from the sensors is collected in the cloud. However, there is a new approach called edge computing whose goal is to process information at the edge of a network instead of processing it in the cloud. Edge computing, combined with machine learning algorithms, has become a powerful tool to optimise tasks in both industrial processes and everyday life. This combination allows decision making in real-time since the data processing is carried out in the place where it is being acquired. In this paper we propose a Non-Intrusive Appliance Load Monitoring (NIALM) which has two functions: a) send detailed energy consumption information to the data server only when it is necessary, and to process the information using an intelligent algorithm based on an Artificial Neural Network to recognise when and how much energy the appliances are consuming. The PCB design of the board includes the ESP12-S microchip. We evaluated the Evolutionary Hyperplane Neural Network against the Evolutionary Spherical Neural Network to decide the best algorithm for our proposed method. The Evolutionary Artificial Neural Networks are trained using the Differential Evolution Algorithm. According to the numerical experiments, the Evolutionary Hyperplane Neural Network showed a better performance of classification up to 82%.