Formal Modeling of IoT-Based Distribution Management System for Smart Grids

Abstract

The smart grid is characterized as a power system that integrates real-time measurements, bi-directional communication, a two-way flow of electricity, and evolutionary computation. The power distribution system is a fundamental aspect of the electric power system in order to deliver safe, efficient, reliable, and resilient power to consumers. A distribution management system (DMS) begins with the extension of the Supervisory Control and Data Acquisition (SCADA) system through a transmission network beyond the distribution network. These transmission networks oversee the distribution of energy generated at power plants to consumers via a complex system of transformers, substations, transmission lines, and distribution lines. The major challenges that existing distribution management systems are facing, maintaining constant power loads, user profiles, centralized communication, and the malfunctioning of system equipment and monitoring huge amounts of data of millions of micro-transactions, need to be addressed. Substation feeder protection abruptly shuts down power on the whole feeder in the event of a distribution network malfunction, causing service disruption to numerous end-user clients, including industrial, hospital, commercial, and residential users. Although there are already many traditional systems with the integration of smart things at present, there are few studies of those systems reporting runtime errors during their implementation and real-time use. This paper presents the systematic model of a distribution management system comprised of substations, distribution lines, and smart meters with the integration of Internet-of-Things (IoT), Nondeterministic Finite Automata (NFA), Unified Modeling Language (UML), and formal modeling approaches. Non-deterministic finite automata are used for automating the system procedures. UML is used to represent the actors involved in the distribution management system. Formal methods from the perspective of the Vienna Development Method-Specification Language (VDM-SL) are used for modeling the system. The model will be analyzed using the facilities available in the VDM-SL toolbox.

A Formal Model for Reliable Data Acquisition and Control in Legacy Critical Infrastructures

“…Nonetheless, formal methods have also been used to model the infrastructure in the case of smart grids. Still, these efforts have focused on understanding the system behavior during runtime [33]. To the best of our knowledge, the only real example of enabling the deployment of a SCADA system while making use of formal models to ensure resiliency can be found in [34].…”

Hybrid Deep Learning Applied on Saudi Smart Grids for Short-Term Load Forecasting

“…The Saudi electric power distribution network is extremely complicated and unsuitable for 21st-century demands. A lack of automated analysis, limited visibility, the use of mechanical switches that cause delayed response times, and a lack of situational awareness are just a few of the flaws that are raised in the current electrical system [1,2]. Furthermore, the power and transportation industries have been major sources of greenhouse gas emissions on the planet [3].…”

“…The SG paradigm is a modern method of electricity transmission, and when it is checked, updated, and integrated with a bidirectional flow of energy and data, it can be employed. The conventional power system has emerged with the development of the SGs due to the rise of smart sensors, and information and communication technology [1]. According to the Saudi Vision 2030, new technologies have been used to create and improve the infrastructure of today's power systems in many ways.…”

“…Unsupervised learning and supervised learning are two types of machine learning algorithms. A supervised learning method is applied to a dataset comprising features, each of which has a label [1]. Another type is the unsupervised learning method, which learns from valuable qualities of the dataset structure and many features in the dataset [3].…”

“…The typical neural network has one input layer, one hidden layer, and one output layer. Because it contains more layers and calculations, the structure of DL is so much more complicated than other neural network models [1]. DL is a multi-layered computational model that uses features as inputs to represent data in various representations.…”