Successful Test Of Automated Combustion Tuning System Using AI Technology

Verification Testing Performed at Coal-fired Power Plant, Operated by the Taiwan Power Company

A newly developed system from Mitsubishi Hitachi Power Systems, Ltd. (MHPS) which aims to automate combustion tuning in boilers using artificial intelligence (AI) has undergone verification testing in Taiwan, and demonstrated results that are comparable to those that could be achieved by a highly experienced engineer. The system was installed at Taiwan Power Company's Linkou Thermal Power Plant, and verification was carried out on-site. The new automated boiler combustion adjustment system is a core element of a comprehensive system, currently under development, that will eventually enable AI-controlled operation of thermal power plants. The latest test results mean development can now proceed toward further advances, including improvements to the system's precision and expansion of its applicable functions.

MHPS is developing AI-based boiler control systems to support power companies globally. By analyzing the large volumes of complex digital data acquired during boiler operation, these systems can provide a range of functions such as cost optimization (operating costs, maintenance management costs, etc.) and early detection of anomalies.

Prior to the current verification testing program, MHPS had provided power generation equipment to the Linkou Thermal Power Plant, overseeing delivery of three sets of coal-fired supercritical-pressure boilers and steam turbines with outputs of 800 megawatts (MW) each. The first set has already started commercial operation, and the second is scheduled to go into service in spring 2017.

In conventional combustion tuning testing, a variety of processes are optimized by adjusting parameters -- such as flue-gas emission characteristics, combustion balance, steam temperature characteristics, and boiler efficiency -- set by an experienced engineer. During the recently completed verification testing in Taiwan, an AI system that had learned changes in process values was called on to propose the optimal parameters. The results confirmed that the AI system proposed parameters essentially no different from those set by an experienced engineer, and the various process values were well-balanced.

MHPS will continue to enhance the AI-based combustion tuning system at the Linkou Thermal Power Plant. It will also apply the results obtained through the verification testing program in Taiwan to MHPS digital solutions portfolio, for future proposals to other potential users.