Functional and non-functional testing of state estimation algorithm for edge deployment

Abstract

State estimation (SE) is a process in electrical distribution systems that provides estimates of the operating condition of the system by using available measurements. It helps DSOs to monitor distribution grid conditions and take necessary control actions. In modern grids, operation has become more complex due to the growing integration of renewable energy sources, distributed generation, and highly dynamic loads. This increase in complexity makes system monitoring more critical and at the same time more challenging, particularly due to the variability of load and generation. Hence, maintaining accurate and reliable state estimation is essential to ensure stable and secure grid operation. Traditionally, SE has been executed in centralized control centers. While this approach is very common, these solutions may lead to limitations such as communication delays, reduced scalability, and lower resilience when fast decisions are required closer to the grid edge. Another drawback in centralized systems is that they are more prone to cybersecurity risks than distribution systems where part of the decision-making is solved locally on the edge. Recent developments in edge computing provide an opportunity to move certain monitoring and control functions closer to where data is generated. In power systems, this creates the requirement to execute state estimation in a more distributed manner. However, much of the existing research focuses mainly on algorithmic accuracy or theoretical performance, and less attention has been given to how SE algorithms behave when deployed in practical computing environments that resemble real operation conditions such as limited computational power, time restrictions, and measurement alignment. This thesis focuses on the deployment and testing of a state estimation algorithm within a simulated edge-based environment. A testing framework was developed in which the SE algorithm is implemented in GNU Octave, and Python is used for data generation, orchestration, and handling communications. Measurement data, which include voltage, current, and temperature values, are transmitted through MQTT messaging to emulate real-time grid operation. The system components are containerized using Docker and managed with Docker Compose to create a modular environment for testing. In this study, functional and non-functional tests were performed to assess the SE algorithm’s performance and robustness. The results show that the developed framework is capable of systematic testing of SE algorithms under realistic communication and deployment conditions and confirms that the SE algorithm is suitable for edge-based distribution grid monitoring.