Support for building monitoring models by AI

In addition to the wide variety of equipment that must be monitored at thermal power plants, past operating behavior is rarely reproduced because operating conditions change from moment to moment due to fluctuations in power generation output and operating environment. Furthermore, it is rare for the same defective event to occur again. Therefore, extensive equipment monitoring is required.

In general, in order to achieve extensive equipment monitoring using AI, personnel with specialized knowledge of both thermal power plants and AI need to build a huge amount of AI models and to maintain and manage the models after operation starts. Therefore, securing the personnel and the effort required to build, maintain, and manage the AI model were issues.

Therefore, as an optional software to expand the functionality of Online Anomaly Detection System using AI provided by Azbil Corp. (hereinafter referred to as Azbil), we jointly developed an optional software by combining our operational know-how regarding thermal power plants and Azbil's know-how regarding AI development and operation.

With the jointly developed an optional software, most of the AI models are automatically built, making it possible to further streamline the build of AI models. Additionally, because the types of information that can be entered have increased dramatically, it is now possible to monitor wide areas, which was difficult to do in the past, and to detect signs of abnormalities that were previously unrecognizable.

We will promote the expansion and introduction of these systems to other companies, as these systems are expected to streamline operational management and equipment management at thermal power plants and large-scale factories.