Method for intelligently realizing online control and optimization of incinerator based on AI

Abstract

According to the method for intelligently realizing online control and optimization of the incinerator based on the AI, the technical means of deep learning, reinforcement learning, a multi-mode integration model and the like are utilized, deep analysis and mining are carried out on operation data of the incinerator, an intelligent control model is established, all the operation data of the incinerator are analyzed in real time, combustion parameters are dynamically adjusted, and the control accuracy of the incinerator is improved. The online intelligent control and optimization of the incinerator are realized, so that the efficient, stable and safe operation of the incinerator is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of waste incineration power generation, and in particular to a method for realizing online control and optimization of an incinerator based on AI intelligence. Background Art

[0002] The current waste incineration industry has entered a stock competition market, and the requirements for refined management and control are becoming increasingly important. In order to comprehensively enhance the company's core technology competitiveness, it is necessary to combine the different working conditions of the boiler, incorporate the best combustion adjustment method into the database, and use the AI ​​big data system in the forefront of technology to timely optimize the boiler combustion conditions, thereby improving the unit utilization efficiency, increasing corporate revenue, and reducing production costs.

[0003] Regarding incinerator control optimization technology, foreign manufacturers such as Japan, Europe and the United States have long-term technical accumulation and practical experience in the field of incinerator design and control, and have formed a relatively mature technical system. These technologies can achieve effective control of key parameters, ensure efficient and full combustion of garbage, and effectively inhibit the generation of harmful substances, meeting strict environmental emission standards. However, in actual applications, the control optimization technology of foreign manufacturers also has some limitations.

[0004] These technologies often have high requirements for the quality of garbage. Since the garbage classification system abroad is relatively complete and the garbage composition is relatively simple, its control optimization technology can play a better effect. However, in China, the garbage classification system is still imperfect, the garbage composition is complex and changeable, the calorific value fluctuates greatly, the moisture content is high, and the classification is not thorough. This leads to the phenomenon of "not adapting to the local conditions" in the actual application of foreign technologies, resulting in adaptability problems when foreign technologies are applied in China, such as incomplete combustion, increased equipment failure rate, and reduced energy efficiency.

[0005] These have greatly limited the widespread application and popularization of the above technologies. Compared with foreign manufacturers, domestic manufacturers pay more attention to technology research and development and application in combination with domestic actual conditions, and are constantly exploring and innovating in incinerator control optimization technology. However, they are basically technical routes based on mechanism models and expert experience, that is, through in-depth research on the operating mechanism of the incinerator, the corresponding mathematical model and control strategy are established, hoping to achieve precise control of the incinerator. The problem with this technical route is that the actual operation rate is often low when the domestic garbage calorific value fluctuates greatly and the operating conditions are complex.

[0006] With the development of artificial intelligence technology, it is necessary to adopt cutting-edge technology for independent innovation, apply artificial intelligence technology to the field of incinerator control optimization, and explore the implementation of incinerator control systems that are more adaptable to the current situation of poor domestic garbage classification and large fluctuations in calorific value, thereby improving incineration efficiency and environmental protection performance, and reducing technology dependence and operating costs.

[0007] The incineration system of a garbage incinerator includes a furnace body, a burner, and auxiliary equipment. The furnace body holds the garbage and controls the surface temperature; the burner provides heat to cause the garbage to undergo pyrolysis and oxidation reactions; and the auxiliary equipment handles the in and out of the garbage and the emission of smoke. The system converts garbage into harmless substances and reduces environmental pollution.

[0008] Furnace body: This is the main part of the incineration system, usually made of refractory materials, insulation materials and steel plates. There is a furnace inside the furnace body, which is used to hold the garbage to be incinerated and convert it into harmless substances through high-temperature incineration. The surface temperature of the furnace body is usually controlled at a low level to reduce heat loss and prevent burns.

[0009] Burner: The burner is a key device in the incineration system. It is responsible for providing the heat required to burn the garbage. The burner usually uses natural gas, fuel oil, etc. as fuel to produce high-temperature flames through combustion. These flames come into contact with the garbage and heat it to a high enough temperature to cause the organic matter in it to undergo pyrolysis and oxidation reactions, and eventually convert it into gas and solid residue.

[0010] Auxiliary equipment: The incineration system also includes some auxiliary equipment, such as feeding device, slag discharge device, flue gas treatment device, etc. The feeding device is responsible for feeding the garbage into the furnace, and the slag discharge device is used to discharge the solid residue generated during the incineration process. The flue gas treatment device is used to treat the flue gas generated during the incineration process to reduce its pollution to the environment.

[0011] During the incineration process, the garbage is heated by high-temperature flames in the furnace, and the organic matter in it undergoes pyrolysis and oxidation reactions to generate gas and solid residue. The gas is discharged through the flue, purified by the flue gas treatment device, and discharged into the atmosphere. The solid residue is discharged through the slag discharge device and can be further processed or utilized. Summary of the invention

[0012] To overcome the above problems, the purpose of the present invention is to provide a method for online control and optimization of an incinerator based on AI intelligence, to realize online intelligent control and optimization of the incinerator, thereby achieving efficient, stable and safe operation of the incinerator.

[0013] The present invention is implemented by the following scheme: a method for online control and optimization of an incinerator based on AI intelligence, the method comprising the following steps:

[0014] S1. Data access work: The measurement point data of the distributed control system DCS is transmitted to a server through the OPC communication protocol; the measurement point data includes key parameters of temperature, pressure and flow; at the same time, the flame monitoring video stream is accessed through the RTSP protocol to obtain real-time combustion status information;

[0015] S2. Model training and iteration: Collect at least three months of historical measurement point data and combustion status information, and perform deductions in a waste incineration simulation environment to obtain the initial model and its parameters; then, run the initial model in a production environment, but do not control the incineration equipment for the time being; during this period, the initial model automatically learns, observes and compares the differences between the instructions given by the initial model and the instructions of the operator, and learns the operator's operating rules in combination with the performance of the core indicator of main steam flow; through this process, the iterative model and its parameters are obtained;

[0016] S3, control closed loop stage: open the control authority of the iterative model so that it can coordinately control the equipment such as pusher, grate, fan, air chamber damper, and lime slurry regulating valve. Through the feedback of actual control performance, further iterate the model to obtain the reverse control model. This stage will ensure the effectiveness and stability of the model in actual operation.

[0017] S4. Continuous evolution of the system: Equipped with a trigger model, the trigger model can automatically discover the operational differences between the operator and the AI ​​intelligent counter-control model and accumulate the difference information. When the difference exceeds a certain limit, the trigger model will automatically start and stop the operation of the incineration equipment; this process will enable the model to continuously adapt to new operating environments and conditions, maintaining its efficiency and advancement;

[0018] S5. Analyze various operating data of the incinerator in real time through the reverse control model and the trigger model, dynamically adjust the combustion parameters, and realize online intelligent control and optimization of the incinerator; the combustion parameters include: air supply volume, feeding speed, and grate cycle.

[0019] Furthermore, the method also includes: through the technical means of perception, prediction and control at three levels, the pusher, grate, fan, air chamber damper and flue gas system objects are realized to realize the intelligent operation, thereby improving the main steam stability, increasing the power generation per ton and reducing the consumption of consumables.

[0020] Furthermore, the technical means at the perception level include garbage source tracing and combustion status identification;

[0021] Tracing the waste source can accurately trace the fermentation status of each grab bucket of the incinerator by sensing the location code and partition management information. It can also realize the real-time perception of the change of waste quality and provide important input data for control optimization.

[0022] Combustion status identification uses flame monitoring and measurement point data to identify whether there is biased burning, the location of the fire line, and the thickness of the material layer. This information is crucial for adjusting combustion parameters and optimizing the combustion process.

[0023] Furthermore, the technical means at the prediction level are: through the early prediction of the key indicators such as main steam flow, main steam pressure, main steam temperature, NOX and SO2, the changing trend of the incinerator operation status can be predicted, so as to adjust the control mode in time to avoid the occurrence of abnormal operation; it can significantly improve the stability and efficiency of the incinerator and reduce the failure rate.

[0024] Furthermore, the technical means at the control level are: collaborative control. This collaborative control strategy aims to achieve intelligent collaboration among pushers, grates, air chamber dampers, fans, lime slurry regulating valves, and denitrification agent unloading valves. By optimizing the control parameters and logical relationships of these components, the overall operating state of the incinerator can be accurately controlled. It can improve the stability of the main steam, reduce energy consumption and unit consumption of consumables, and improve the economic benefits of the incinerator.

[0025] The beneficial effects of the present invention are:

[0026] 1. Clever redefinition of technical goals

[0027] The traditional control optimization problem is transformed into: learning from the "old masters", that is, letting the AI ​​system learn the operating methods of excellent operators under similar working conditions, so that it can learn patterns from historical data.

[0028] 2. Multi-modal and multi-scale fusion technology

[0029] 1) Based on flame monitoring and measurement point data, combined with video recognition technology and deep learning technology, identify whether there is biased burning, fire line location, combustion status and other operating condition information;

[0030] 2) Combine real-time data and historical data to find the relationship between measurement points from the data and make advance predictions on the change trend of measurement points;

[0031] 3) Further combine operating procedures and large model technology to allow the AI ​​system to acquire professional knowledge of incinerator operations, combine knowledge and dynamic data, and thus achieve an operating level that is no less than that of experienced operators.

[0032] 3. Furnace-bin coordination technology

[0033] Introducing intelligent technology for furnace-bin collaboration to achieve collaborative optimization of the garbage bin and incinerator. Traditional methods often only focus on the optimization of the incinerator itself, while ignoring the impact of the garbage bin on the incineration process. Through the access of garbage bin coding information and partition management information, combined with spatial restoration technology and machine learning technology, it is possible to predict the fermentation status of each grab bucket garbage in advance, thereby providing guidance for the adjustment of furnace equipment such as grates. This innovation breaks through the limitations of traditional methods, achieves the goal of optimizing the incineration process from the source, and will effectively improve the stability and efficiency of the incinerator.

[0034] 4. Adaptive Technology

[0035] In addition to the above technologies, adaptive technology is introduced to take into account the working conditions, and a model system combined with reinforcement learning is established. By analyzing the operating methods of senior operators in historical data, the system has the ability to adapt to different working conditions. Taking into account the differences in combustion control requirements in different seasons and working conditions, adaptive technology can automatically adjust control parameters and strategies according to changes in real-time data and historical data to adapt to various complex working conditions. This innovation makes the technical solution more flexible and intelligent, and can cope with various changes and challenges. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic flow chart of the method of the present invention.

[0037] Figure 2 It is a schematic diagram of the hardware architecture used in the method of the present invention. DETAILED DESCRIPTION

[0038] The present invention will be further described below in conjunction with the accompanying drawings.

[0039] See also Figure 1 As shown, a method for online control and optimization of an incinerator based on AI intelligence of the present invention comprises the following steps:

[0040] S1. Data access work: The measurement point data of the distributed control system DCS is transmitted to a server through the OPC communication protocol; the measurement point data includes key parameters of temperature, pressure and flow; at the same time, the flame monitoring video stream is accessed through the RTSP protocol to obtain real-time combustion status information;

[0041] S2. Model training and iteration: Collect at least three months of historical measurement point data and combustion status information, and perform deductions in a waste incineration simulation environment to obtain the initial model and its parameters; then, run the initial model in a production environment, but do not control the incineration equipment for the time being; during this period, the initial model automatically learns, observes and compares the differences between the instructions given by the initial model and the instructions of the operator, and learns the operator's operating rules in combination with the performance of the core indicator of main steam flow; through this process, the iterative model and its parameters are obtained;

[0042] S3, control closed loop stage: open the control authority of the iterative model so that it can coordinately control the equipment such as pusher, grate, fan, air chamber damper, and lime slurry regulating valve. Through the feedback of actual control performance, further iterate the model to obtain the reverse control model. This stage will ensure the effectiveness and stability of the model in actual operation.

[0043] S4. Continuous evolution of the system: Equipped with a trigger model, the trigger model can automatically discover the operational differences between the operator and the AI ​​intelligent counter-control model and accumulate the difference information. When the difference exceeds a certain limit, the trigger model will automatically start and stop the operation of the incineration equipment; this process will enable the model to continuously adapt to new operating environments and conditions, maintaining its efficiency and advancement;

[0044] S5. Analyze various operating data of the incinerator in real time through the reverse control model and the trigger model, dynamically adjust the combustion parameters, and realize online intelligent control and optimization of the incinerator; the combustion parameters include: air supply volume, feeding speed, and grate cycle.

[0045] The method also includes: realizing intelligent operation of pushers, grates, fans, air chamber dampers, and flue gas system objects through technical means at three levels of perception, prediction, and control, thereby improving main steam stability, increasing power generation per ton, and reducing consumables.

[0046] The technical means at the perception level include garbage source tracing and combustion status identification;

[0047] Tracing the waste source can accurately trace the fermentation status of each grab bucket of the incinerator by sensing the location code and partition management information. It can also realize the real-time perception of the change of waste quality and provide important input data for control optimization.

[0048] Combustion status identification uses flame monitoring and measurement point data to identify whether there is biased burning, the location of the fire line, and the thickness of the material layer. This information is crucial for adjusting combustion parameters and optimizing the combustion process.

[0049] The technical means at the prediction level are: through the early prediction of the key indicators of main steam flow, main steam pressure, main steam temperature, NOX, and SO2, the changing trend of the incinerator operation status can be predicted, so as to adjust the control method in time to avoid the occurrence of abnormal operation; it can significantly improve the stability and efficiency of the incinerator and reduce the failure rate.

[0050] The technical means at the control level are: collaborative control. This collaborative control strategy aims to achieve intelligent collaboration among pushers, grates, air chamber dampers, fans, lime slurry regulating valves, and denitrification agent unloading valves. By optimizing the control parameters and logical relationships of these components, precise control of the overall operating state of the incinerator can be achieved. The stability of the main steam can be improved, and energy consumption and unit consumption of consumables can be reduced, thereby improving the economic benefits of the incinerator.

[0051] See also Figure 2 As shown, the hardware architecture adopted by the method of the present invention includes: a data collector, an inference accelerator, a main server, a backup server and a container host;

[0052] 1) Data collector

[0053] Data is exchanged with the original distributed control system DCS through communication methods such as OPC or MODBUS. The cost of accessing the system is low and there is no need for on-site parking and modification.

[0054] 2) Inference Accelerator

[0055] The built-in inference accelerator can effectively accelerate large model inference and improve efficiency and performance.

[0056] 3) Primary Server

[0057] The main server provided for production operation is mainly used for algorithm execution, big data computing, model training, middleware management, etc.

[0058] 4) Backup Server

[0059] Real-time hot standby of production environment data, real-time monitoring of host health, and smooth switching if any failure occurs.

[0060] 5) Container Host

[0061] Container deployment and coordination, container lifecycle management, dynamic expansion or reduction of container data, and real-time monitoring of resource usage.

[0062] The present invention aims to solve the problems of large time lag, large fluctuation, multi-modal and multi-variable, and complex system coordination, and the technical solution of the present invention has achieved innovation in many aspects.

[0063] 1. Clever redefinition of technical goals

[0064] Faced with large time lags and large fluctuations, it is almost impossible to use traditional control methods to deal with time lags of more than 50 minutes and complex fluctuations. The traditional control optimization problem is transformed into: learning from "old masters", that is, letting the AI ​​system learn the operating methods of excellent operators under similar working conditions, so that it can learn rules from historical data.

[0065] 2. Multi-modal and multi-scale fusion technology

[0066] 1) Based on flame monitoring and measurement point data, combined with video recognition technology and deep learning technology, identify whether there is biased burning, fire line location, combustion status and other operating condition information;

[0067] 2) Combine real-time data and historical data to find the relationship between measurement points from the data and make advance predictions on the change trend of measurement points;

[0068] 3) Further combine operating procedures and large model technology to allow the AI ​​system to acquire professional knowledge of incinerator operations, combine knowledge and dynamic data, and thus achieve an operating level that is no less than that of experienced operators.

[0069] 3. Furnace-bin coordination technology

[0070] Introducing intelligent technology for furnace-bin collaboration to achieve collaborative optimization of the garbage bin and incinerator. Traditional methods often only focus on the optimization of the incinerator itself, while ignoring the impact of the garbage bin on the incineration process. Through the access of garbage bin coding information and partition management information, combined with spatial restoration technology and machine learning technology, it is possible to predict the fermentation status of each grab bucket garbage in advance, thereby providing guidance for the adjustment of furnace equipment such as grates. This innovation breaks through the limitations of traditional methods, achieves the goal of optimizing the incineration process from the source, and will effectively improve the stability and efficiency of the incinerator.

[0071] 4. Adaptive Technology

[0072] In addition to the above technologies, adaptive technology is introduced to take into account the working conditions, and a model system combined with reinforcement learning is established. By analyzing the operating methods of senior operators in historical data, the system has the ability to adapt to different working conditions. Taking into account the differences in combustion control requirements in different seasons and working conditions, adaptive technology can automatically adjust control parameters and strategies according to changes in real-time data and historical data to adapt to various complex working conditions. This innovation makes the technical solution more flexible and intelligent, and can cope with various changes and challenges.

[0073] In summary, the present invention achieves innovation through clever redefinition of technical objectives, multimodal and multi-scale fusion technology, furnace-bin collaborative technology, adaptive technology, etc., which will improve the stability and efficiency of the incinerator and bring more efficient and environmentally friendly solutions to the waste disposal industry.

[0074] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made according to the scope of the patent application of the present invention should fall within the scope of the present invention.

Claims

1. A method for online control and optimization of an incinerator based on AI intelligence, characterized in that: The method comprises the following steps: S1. Data access work: The measurement point data of the distributed control system DCS is transmitted to a server through the OPC communication protocol; the measurement point data includes key parameters of temperature, pressure and flow; at the same time, the flame monitoring video stream is accessed through the RTSP protocol to obtain real-time combustion status information; S2. Model training and iteration: Collect at least three months of historical measurement point data and combustion status information, and perform deductions in a waste incineration simulation environment to obtain the initial model and its parameters; then, run the initial model in a production environment, but do not control the incineration equipment for the time being; during this period, the initial model automatically learns, observes and compares the differences between the instructions given by the initial model and the instructions of the operator, and learns the operator's operating rules in combination with the performance of the core indicator of main steam flow; through this process, the iterative model and its parameters are obtained; S3, control closed loop stage: open the control authority of the iterative model so that it can coordinately control the equipment such as pusher, grate, fan, air chamber damper, and lime slurry regulating valve. Through the feedback of actual control performance, further iterate the model to obtain the reverse control model. This stage will ensure the effectiveness and stability of the model in actual operation. S4. Continuous evolution of the system: Equipped with a trigger model, the trigger model can automatically discover the operational differences between the operator and the AI ​​intelligent counter-control model and accumulate the difference information. When the difference exceeds a certain limit, the trigger model will automatically start and stop the operation of the incineration equipment; this process will enable the model to continuously adapt to new operating environments and conditions, maintaining its efficiency and advancement; S5. Analyze various operating data of the incinerator in real time through the reverse control model and the trigger model, dynamically adjust the combustion parameters, and realize online intelligent control and optimization of the incinerator; the combustion parameters include: air supply volume, feeding speed, and grate cycle.

2. According to claim 1, a method for online control and optimization of an incinerator based on AI intelligence is characterized in that: The method also includes: realizing intelligent operation of pushers, grates, fans, air chamber dampers, and flue gas system objects through technical means at three levels of perception, prediction, and control, thereby improving main steam stability, increasing power generation per ton, and reducing consumables.

3. The method for online control and optimization of an incinerator based on AI intelligence according to claim 2 is characterized in that: The technical means at the perception level include garbage source tracing and combustion status identification; Garbage traceability can accurately trace the fermentation status of each grab bucket of the incinerator by sensing location coding and zoning management information. It can realize real-time perception of changes in garbage quality and provide important input data for control optimization. Combustion status identification can identify the presence of biased burning, fire line location, and material layer thickness through flame monitoring and measurement point data. This information is crucial for adjusting combustion parameters and optimizing the combustion process.

4. The method for online control and optimization of an incinerator based on AI intelligence according to claim 2 is characterized in that: The technical means of prediction are: by analyzing the main steam flow, main steam pressure, main steam temperature, NO X , and the early prediction of key indicators of SO2 can predict the changing trend of the incinerator's operating status, so as to adjust the control method in time to avoid the occurrence of abnormal operation; it can significantly improve the stability and efficiency of the incinerator and reduce the failure rate.

5. The method for online control and optimization of an incinerator based on AI intelligence according to claim 2 is characterized in that: The technical means at the control level are: collaborative control. This collaborative control strategy aims to achieve intelligent collaboration among pushers, grates, air chamber dampers, fans, lime slurry regulating valves, and denitrification agent unloading valves. By optimizing the control parameters and logical relationships of these components, precise control of the overall operating state of the incinerator can be achieved. The stability of the main steam can be improved, and energy consumption and unit consumption of consumables can be reduced, thereby improving the economic benefits of the incinerator.

Images