Application case 3: The optimal control of the new product RMPOS combustion boiler

RMPOS is the latest high-tech product developed by Beijing CIMAC Technology Co., Ltd., It is based on advanced neural network technology, modern control theory and nonlinear algorithms, through establishing a dynamic multi-objective optimization controller, intelligent optimization and adjustment DCS set parameters and control deviation, to achieve optimal control of boiler combustion, thereby improving thermal efficiency, reduce pollution emissions, with significant economic and social environmental benefits.

一. System features and optimization results

A.System features：

● System configuration is simple: just need a PC and system data communication tools

● As a software product, no loss, less maintenance

B. Typical optimization results：

• Improve combustion efficiency of 0.5% to 2.5%, fuel savings,
• NOx emissions by 10% to 50%, while reducing emissions of SOx and COx, is conducive to environmental protection
• Reduce the carbon content of fly ash
• Improve the quality of reheat steam temperature control, reducing both sides of the deviation
• improve operational automation level
• Reduce the labor intensity of operating personnel
• Extend equipment life
• reduce accidents and improve safety productivity

二. Economic Analysis

As a 300000 kilowatts coal-fired units, through optimized control system, if the boiler thermal efficiency rises by 1% a year, as the direct economic benefits for the 1.38 million Yuan, not including incidental indirect benefits. At the same time there is this project good social benefits, saving valuable coal resources, reducing air pollution and are environmentally friendly.

三. Power Plant Example: Before and after optimization system into the boiler thermal efficiency of about 2%

四. Project Background of RMPOS

Plant is a complex system, may occur at any unit load changes and other confounding factors, it is difficult to rely on artificial continuous operation will be in the best state of the control unit, which is the most typical boiler combustion control. Fuel cost is the main cost of power generation , power plant flue gas emissions is the main source of pollution , and how skillfully continuously adjust the amount of wind , bellows and furnace pressure and other variables , and always will be the best run state boiler control , which for the operating personnel technical level and workload requirements are very high . In fact , DCS system, there are thousands of data points , in part to participate in the control , in part for operating personnel to monitor the operational status of the unit , their role does not really play out. Boiler efficiency is a complex curved surface, the impact of many factors, and these factors are subject to change, by operating personnel continuously looking for the highest point of the surface is impossible, even if the operating personnel know the highest point, asking him to simultaneously adjust many parameters, which is unrealistic, with the help of RMPOS optimization control system, you can easily do it.

五. RMPOS principle

Artificial neural network is a simulation of biological neurons in the brain cell structure and function constitutes an information processing system with self-learning, self-identification, adaptive characteristics, can be arbitrary precision approximate any nonlinear continuous function, with a strong for complex environments and the ability to multi-objective control requirements for optimal control of complex systems.                               

六. Structure and working process

RMPOS as a secondary control system, in addition to the necessary data communication, no need to add new hardware. RMPOS from DCS control system parameters in real time, through the analysis and optimization, obtained set-point or control deviation of control parameters affecting thermal efficiency of the boiler, and recommended valuesfor these parameters in an open loop or closed loop back to the DCS. Parameter returns once every 15 seconds, fully meet the control requirements of real-time dynamic scene. In addition, in order to ensure safety in production, the system also has the data validation feature, if there is deviation from the normal input parameters, the system will automatically exit, and gives alarm.