When the thermal power plant is equipped with the "Intelligent Central Hub"—Gansu Energy and Chemical Group Jingmei (Baiyin) Thermal Power Embedded Intelligent Monitoring and Control System Construction Scan

New Gansu · Gansu Daily reporter Du Xueqin

Not long ago, the list for the 24th China Automation and Digitized Process Manufacturing “New Quality Award” was released, and Jiingmei (Baiyin) Thermal Power Co., Ltd., under Gansu Energy and Chemical Industry Group, was named on it. This is another top science-and-technology award that Jingmei (Baiyin) Thermal Power has received.

Located in the Yindong Industrial Park in Baiyin City, Jingmei (Baiyin) Thermal Power has 2 sets of 350-megawatt ultra-supercritical coal-fired combined heat and power units using air-cooling, which are the primary heat-supply heat sources for the Baiyin District of Baiyin City.

Founded in May 2012, this power plant is not very large. So why has it frequently won industry science-and-technology awards in the past two years?

Behind the awards is a set of embedded intelligent monitoring and supervision system developed independently by the company. In 2022, Jingmei (Baiyin) Thermal Power launched the construction of a smart power plant. At present, the company has successfully created the first modern smart power plant in China based on Safety Zone I of the plant control system.

“Our smart system deeply integrates comprehensive sensing, multi-system coordination, and big-data analytics—like installing an ‘intelligent central hub’ in the power plant.” Cui Guanhua, deputy director of the production technology department of Jingmei (Baiyin) Thermal Power, said.

Jingmei (Baiyin) Thermal Power’s centralized control room.

Start-up and grid connection: upgrading from a “manual gearbox” to “intelligent driving”

In traditional thermal power enterprises, the generation peak period is when operations personnel are the busiest.

On the morning of March 26, in the centralized control room of Jingmei (Baiyin) Thermal Power Co., Ltd. of Gansu Energy and Chemical Industry Group, the reporter witnessed a scene different from that of traditional thermal power plants: with the high-voltage power-production period in full swing, the large screens showed constantly changing data, operations personnel calmly monitored the panel, and the entire production process looked orderly and efficient.

“Previously, when we heard the command, ‘Unit start-up, prepare for grid connection!’ the atmosphere in the entire centralized control room would instantly become tense.” Liu Yuchen, head of the thermal control class in the production technology department of Jingmei (Baiyin) Thermal Power, said. In the past, from boiler ignition to successful grid connection, it required at least three experienced operators to focus intensely, completing hundreds of precise operations within several hours; the whole process was long and full of pressure.

Now, the situation is completely different. Through the renovation of the embedded intelligent monitoring and supervision system, operators are more like “supervisors.” With just 1 person watching key nodes, once the system confirms the commands at those nodes, subsequent complex processes are automatically executed by the system. From ignition, temperature ramp-up, and pressure ramp-up, to start-up switching and grid connection, it achieves full-process “intelligent driving.”

“It’s like installing an intelligent driving system in a car.” Liu Yuchen said. The system can control every parameter smoothly and precisely along the optimal route. By shortening the cold-state start-up to grid connection and decommissioning and tuning process by 4 hours, “more importantly, the intelligent program strictly ‘follows the rules,’ avoiding human mis-operations.”

Precision has also improved significantly. In the critical feedwater control stage, the system achieves full-process automatic adjustment. Even when switching feedwater pumps, water-flow fluctuations are firmly controlled within a very small range. Liu Yuchen said, “Operators’ workload and mental pressure have come down, while safety and economic performance are actually up.”

Jingmei (Baiyin) Thermal Power’s centralized control room.

Smart early warnings make hidden risks “visible and controllable”

197 megawatts, 235 megawatts, 201 megawatts… The continuously flashing numbers on the large screens in Jingmei (Baiyin) Thermal Power’s centralized control room are the generation commands sent to the plant by the State Grid.

“Almost every minute it changes.” Cui Guanhua said. During normal operation, monitoring personnel need to keep an eye on hundreds or thousands of parameters on the screen at all times, judging the equipment’s health status from subtle changes. However, the objective correlation factors among all parameters cannot be identified by manual means alone; the difficulty is very high. Even if operators stay highly focused, they still cannot predict and diagnose early faults in the units.

In 2025, Jingmei (Baiyin) Thermal Power connected a unit operating condition early warning and diagnosis system to its existing intelligent detection system.

“We ‘feed’ the data accumulated over the past 10-plus years into artificial intelligence. Through data iteration, we form our own expert rule library and data model library.” Cui Guanhua said. This system makes production data accumulated over more than ten years truly “come alive,” turning it into continuously improving safety productivity.

The unit operating condition early warning and diagnosis system can scan and analyze all operating data non-stop 24 hours a day, automatically discover abnormal correlations between parameters, issue warnings at the budding stage of a fault, and push the issue location, possible causes, and handling recommendations to operators through a clear visualization interface.

“What’s most impressive is that it can learn the experience of our veteran experts.” Cui Guanhua introduced. The system provides convenient tools to turn the valuable experience of excellent monitoring personnel into rules that the system can recognize and execute, so knowledge can be continuously accumulated and passed on.

In addition to accurately locating single-point problems, this early warning system can also analyze linkage risks across the entire system. Cui Guanhua said, “Now equipment operation is safer, and staff workload is also lower.”

Jingmei (Baiyin) Thermal Power Group.

An “embedded” path: letting intelligence take root in safe soil

For thermal control personnel responsible for maintaining and controlling the control systems, modifying and optimizing control logic used to be a “high-risk” task. Usually, it could only be done when the unit was shut down, with high trial-and-error costs and few training opportunities.

Now they have a safe “practice ground”—a digital twin system. The system can replicate, with high precision, a virtual environment identical to the real control system. In this “parallel world,” thermal control personnel can boldly try any logic modifications and tests, without affecting actual production at all.

“Here, our operators can rehearse various accident response procedures, while thermal control personnel can go deep into the system’s underlying layer and repeatedly conduct logic optimization experiments.” Cui Guanhua said. Now, the efficiency and safety factor of corporate new-hire training and veteran experts’ study of new algorithms are no longer comparable to before.

Unlike many intelligent solutions that “patch” outside existing systems, Jingmei (Baiyin) Thermal Power has chosen a more challenging but more thorough “embedded” route—deeply implanting intelligent algorithm models into the core of the control system (Safety Zone I) internally, like “chips.”

“Essentially, it’s installing ‘intelligence’ directly into the power plant’s ‘nervous central hub’ for operation.” Cui Guanhua further explained. This not only preserves the original control system’s stability and reliability, and realizes real-time seamless linkage between intelligent analysis and automatic control, but more importantly, all data computations and command generation are completed inside the control network, ensuring the highest safety grade—truly achieving “data never leaves the domain, and intelligence takes deep root.”

Editor-in-charge: Yang Chenyu

A wealth of information and precise insights—available on the Sina Finance APP