Realization of long-term stable operation of waste-to-energy plants with AI technology

Abstract

Waste-to-energy plants are expected to become a decentralized energy facility due to their ability to generate stable power while sanitarily disposing of waste. Hitachi Zosen has been conducting research and development on the advanced control technology of waste-to-energy plants with AI technology, and has achieved certain results. In this paper, we improved our AI technology by adding a model that predicts future combustion images from actual combustion images in the last several tens of minutes. Furthermore, we developed an improved combustion control system that enhances the accuracy of the existing AI model with sensor values and additional control methods for ACC.

From 2021 to 2022, we conducted a demonstration test of this system at the Mito City Incineration Plant. As a result, the frequency of combustion deterioration decreased significantly and the number of manual operations was reduced by 86.4%. We also confirmed that stable operation of the incinerator was achieved for approximately 73 days using ACC+AI control, contributing to the streamlining of operational management.