Mitsubishi Heavy Industries, Ltd. (MHI) examined the feasibility of the virtual measurement of power plant water quality using software sensors developed in collaboration with the University of Tokyo. MHI developed a software sensor model that follows changes in plant operation that combines genetic algorithm-based variable selection, a nonlinear regression model and an adaptive model, and verified that the chloride ion concentration, which had been difficult to continuously measure, could be virtually measured through analysis using actual plant data.
The Mitsubishi Heavy Industries Group has been expanding its digital solution business that provides maintenance and operation supports using AI and IoT. Going forward, MHI will develop applications that incorporate software sensors to build a power plant water quality diagnosis system that can monitor state quantities regardless of operator skill.
