Combustion regime awareness hierarchical system and method

Abstract

This application relates to a combustion situation perception and classification system and method, comprising: a root cause diagnosis module, used to calculate the current violation degree of at least one physicochemical constraint based on the current operating parameters when any operating parameter of the target combustion device meets a preset abnormal state alarm condition, and output the current root cause diagnosis result according to a preset violation pattern library; a trend prediction module, used to determine the target physicochemical constraint, and output the current trend prediction result when the violation degree time series corresponding to the target physicochemical constraint meets a preset deterioration condition; and a decision output module, used to generate a target control strategy and control the target combustion device. This solves the problems of low accuracy of diagnostic results, incomplete fault tracing, and poor reliability of identifying unknown abnormal conditions in related technologies when diagnosing boiler combustion conditions. It can significantly improve the accuracy of diagnostic results, reduce the probability of false alarms and missed alarms, and improve the timeliness and rationality of boiler combustion control.

Description

Technical Field

[0001] This application relates to the field of mechanical technology, and in particular to a combustion situation perception and classification system and method. Background Technology

[0002] Currently, in the fields of thermal power generation and industrial heating, the operation and management of combustion equipment highly rely on operators' situational awareness capabilities regarding massive amounts of real-time monitoring data. Taking a 300MW circulating fluidized bed boiler as an example, the number of measuring points on a single unit typically exceeds 2,000, covering multiple dimensions such as temperature, pressure, flow rate, flue gas composition, coal quality analysis, steam parameters, and feedwater parameters. Experienced operators can quickly sift through the massive amount of measuring data to form a complete situational understanding of the boiler's combustion status. This entire understanding process consists of four steps: identifying parameter anomalies, determining the root cause, predicting development trends, and formulating control measures. However, this manual situational awareness model has two fundamental limitations: first, judgment ability is entirely dependent on the operator's personal experience and on-site condition, making it difficult to standardize and systematically pass on relevant knowledge; second, for the same type of abnormal operating conditions, different shift supervisors may give significantly different fault attributions and handling solutions, and the uncertainty of decision-making directly affects operational safety and economic benefits.

[0003] In related technologies, when diagnosing faults in boiler combustion conditions, the main approach is to establish a physical knowledge neural network model that includes mass conservation equations and energy conservation equations to simulate boiler operation, and then combine it with a generative adversarial network for fault diagnosis. In this approach, physical knowledge serves to constrain the training of the neural network, i.e., a physical knowledge-assisted learning method. The final fault diagnosis conclusion is output entirely by the black box of the neural network.

[0004] However, when diagnosing boiler combustion conditions, the relevant technologies rely on physical conservation laws only as training aids rather than direct diagnostic tools, lack multi-constraint collaborative diagnostic capabilities, cannot distinguish the causal direction of deviations, and are highly dependent on historical fault data. This results in low accuracy of diagnostic results, incomplete fault tracing, and poor reliability in identifying unknown abnormal operating conditions, which urgently need to be addressed. Summary of the Invention

[0005] This application provides a combustion situation perception and classification system and method to solve the problems of low accuracy of diagnosis results, incomplete fault tracing, and poor reliability of identification of unknown abnormal conditions when related technologies are used to diagnose faults in boiler combustion conditions. It can significantly improve the accuracy of diagnosis results, reduce the probability of false alarms and missed alarms, and improve the timeliness and rationality of boiler combustion control.

[0006] The first aspect of this application provides a combustion situation awareness and classification system, including: The data sensing module is used to collect the current operating parameters of the target combustion device when the combustion situation sensing level of the target combustion device is at the data sensing level. The anomaly detection module is used to switch the combustion situation perception level from the anomaly detection level to the root cause diagnosis level if any operating parameter meets the preset abnormal state alarm conditions when the combustion situation perception level is at the anomaly detection level. The root cause diagnosis module is used to calculate the current violation degree of at least one physical and chemical constraint based on the current operating parameters when the combustion situation perception level is at the root cause diagnosis level. When the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold for a duration of a preset duration, the module outputs the current root cause diagnosis result according to the preset violation mode library and controls the combustion situation perception level to switch from the root cause diagnosis level to the trend prediction level. The trend prediction module is used to determine the target physical and chemical constraints whose current violation degree is greater than the corresponding preset violation threshold when the combustion situation perception level is the trend prediction level. When the violation degree time series corresponding to the target physical and chemical constraints meets the preset deterioration conditions, the module outputs the current trend prediction result and controls the combustion situation perception level to switch from the trend prediction level to the decision output level. The decision output module is used to generate a target control strategy based on the current root cause diagnosis results and the current trend prediction results when the combustion situation perception level is the decision output level, and to control the target combustion device according to the target control strategy.

[0007] Optionally, in some embodiments, it further includes: The non-perceptive module is used to obtain the percentage of the number of acquisition signals of the acquisition components in the target combustion device when the combustion situation perception level is non-perceptive, and to control the combustion situation perception level to switch from non-perceptive level to data perception level when the percentage of the number of acquisition signals reaches a preset ratio.

[0008] Optionally, in some embodiments, the anomaly detection module includes: The detection unit is used to compare each operating parameter in the current operating parameters with the corresponding preset operating parameter threshold to obtain the parameter comparison result when the combustion situation perception level is an abnormal detection level. The first control unit is used to switch the combustion situation perception level from the abnormal detection level to the root cause diagnosis level when any operating parameter in the parameter comparison results meets the preset abnormal state alarm conditions.

[0009] Optionally, in some embodiments, the root cause diagnosis module includes: The first determining unit is used to determine the current physical and chemical constraints when the combustion situation perception level is the root cause diagnosis level. The calculation unit is used to calculate the current violation degree of at least one physical and chemical constraint based on the current physical and chemical constraints and the current operating parameters. The first output unit is used to output the current root cause diagnosis result according to the preset violation pattern library when the duration of the current violation degree of any physical or chemical constraint being greater than the corresponding preset violation threshold reaches a preset duration. The second control unit is used to switch the combustion situation perception level from the root cause diagnosis level to the trend prediction level when the duration of the current violation degree of any physical or chemical constraint being greater than the corresponding preset violation threshold reaches a preset duration.

[0010] Optionally, in some embodiments, the trend prediction module includes: The second determining unit is used to determine the target physical and chemical constraints whose current violation degree is greater than the corresponding preset violation threshold when the combustion situation perception level is the trend prediction level. The second output unit is used to output the current trend prediction result when the violation time series corresponding to the target physicochemical constraints meets the preset deterioration condition; The third control unit is used to switch the combustion situation perception level from the trend prediction level to the decision output level when the time series of the violation degree corresponding to the target physical and chemical constraints meets the preset deterioration conditions.

[0011] Optionally, in some embodiments, the decision output module includes: The generation unit is used to generate target control strategies based on the current root cause diagnosis results and the current trend prediction results when the combustion situation perception level is the decision output level. The control unit is used to control the target combustion device according to the target control strategy.

[0012] A second aspect of this application provides a combustion situation perception and classification method, including the following steps: When the combustion status perception level of the target combustion device is at the data perception level, the data perception module collects the current operating parameters of the target combustion device. If any operating parameter meets the preset abnormal state alarm conditions when the combustion situation perception level is at the abnormal detection level, the abnormal detection module controls the combustion situation perception level to switch from the abnormal detection level to the root cause diagnosis level. When the combustion situation perception level is at the root cause diagnosis level, the root cause diagnosis module calculates the current violation degree of at least one physical and chemical constraint based on the current operating parameters. When the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold for a duration that reaches a preset duration, the current root cause diagnosis result is output according to the preset violation mode library, and the combustion situation perception level is switched from the root cause diagnosis level to the trend prediction level. When the combustion situation perception level is at the trend prediction level, the trend prediction module determines the target physical and chemical constraints whose current violation degree is greater than the corresponding preset violation threshold. When the violation degree time series corresponding to the target physical and chemical constraints meets the preset deterioration conditions, the current trend prediction result is output, and the combustion situation perception level is controlled to switch from the trend prediction level to the decision output level. When the combustion situation perception level is at the decision output level, the decision output module generates a target control strategy based on the current root cause diagnosis results and the current trend prediction results, and then controls the target combustion device according to the target control strategy.

[0013] Optionally, in some embodiments, before collecting the current operating parameters of the target combustion device by means of the data sensing module when the combustion status perception level of the target combustion device is at the data sensing level, the method further includes: When the combustion situation perception level is at the non-perceptive level, the non-perceptive module obtains the percentage of the number of acquisition signals of the acquisition components in the target combustion device, and controls the combustion situation perception level to switch from the non-perceptive level to the data perception level when the percentage of the number of acquisition signals reaches the preset ratio.

[0014] Optionally, in some embodiments, when the combustion situation awareness level is at the anomaly detection level, if any operating parameter meets a preset abnormal state alarm condition, the anomaly detection module controls the combustion situation awareness level to switch from the anomaly detection level to the root cause diagnosis level, including: When the combustion situation awareness level is at the abnormal detection level, the parameter comparison results are obtained by comparing each operating parameter in the current operating parameters with the corresponding preset operating parameter threshold. When any operating parameter in the parameter comparison results meets the preset abnormal state alarm conditions, the control combustion situation awareness level is switched from the abnormal detection level to the root cause diagnosis level.

[0015] Optionally, in some embodiments, the root cause diagnosis module is configured to, when the combustion situation awareness level is at the root cause diagnosis level, calculate the current violation degree of at least one physicochemical constraint based on the current operating parameters, and when the duration for which the current violation degree of any physicochemical constraint exceeds the corresponding preset violation threshold reaches a preset duration, output the current root cause diagnosis result according to a preset violation pattern library, and control the combustion situation awareness level to switch from the root cause diagnosis level to the trend prediction level, including: When the combustion situation awareness level is at the root cause diagnosis level, determine the current physical and chemical constraints; Based on the current physical and chemical constraints, calculate the current degree of violation of at least one physical and chemical constraint according to the current operating parameters; When the current violation degree of any physical or chemical constraint exceeds the corresponding preset violation threshold for a duration that reaches the preset duration, the current root cause diagnosis result is output according to the preset violation mode library, and the combustion situation perception level is switched from the root cause diagnosis level to the trend prediction level.

[0016] Therefore, the embodiments of this application have at least the following beneficial effects: (1) This application embodiment establishes for the first time a six-level quantitative classification system for the operational status awareness of carbon-containing fuel combustion devices. It systematizes the complete cognitive process of operators from no awareness to taking action into six technical levels that can be automatically executed by the control system. Each level has clearly defined data input requirements, information processing logic, output data format, and technical judgment conditions for jumping to the next level or falling back to the next level. Among them, the L0 no-awareness level clarifies for the first time when the control system "should realize that it does not yet have the data basis to make a status judgment", so that the entire six-level system forms a closed loop from "cannot be perceived" to "complete perception".

[0017] (2) The embodiments of this application realize automatic cognitive transition from L2 to L3. The automatic transition from L2 to L3 is triggered by the violation of the physical constraint equation exceeding the diagnostic threshold and the duration exceeding the preset time window. This determination enables the computing device to automatically execute according to objective physical judgment criteria, without relying on the subjective judgment of the operator, and can fundamentally eliminate the uncertainty of human beings in the cognitive transition process.

[0018] (3) In the embodiment of this application, the trend perception uses the time derivative of the constraint violation degree as a quantitative indicator, so that the trend judgment is transformed from the subjective feeling of the operators into a mathematical indicator that can be calculated, measured and set with clear trigger thresholds, so that the trend warning has objectivity and reproducibility.

[0019] (4) The embodiment of this application forms an L5 closed-loop control and feedback monitoring mechanism. The L5 module automatically generates control commands based on root causes and trends and sends them to the actuator. After execution, it returns to L1 to continuously monitor the control effect, thereby forming a complete closed loop of "perception → diagnosis → prediction → decision → re-perception".

[0020] (5) The embodiments of this application establish a structured data flow mechanism between layers, and define a standard data packet transmission format between adjacent layers to ensure the completeness and consistency of information in the process of passing information level by level.

[0021] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0022] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a block diagram of a combustion situation awareness and classification system according to an embodiment of this application; Figure 2 This is a flowchart illustrating a level six situational awareness method according to an embodiment of this application; Figure 3 This is a flowchart of an automatic transition determination method according to an embodiment of this application; Figure 4 This is a schematic diagram illustrating a constraint violation pattern matching logic according to an embodiment of this application; Figure 5 This is a schematic diagram of a six-level situational awareness closed-loop control logic according to an embodiment of this application; Figure 6 This is a flowchart of a combustion situation perception and classification method provided according to an embodiment of this application. Detailed Implementation

[0023] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0024] The combustion situation perception and classification system and method of this application are described below with reference to the accompanying drawings.

[0025] Before introducing the combustion situation perception and classification system of this application, let me briefly introduce the design motivation of this application and the method of fault diagnosis of boiler combustion conditions in related technologies.

[0026] Specifically, the design motivation of this application stems from in-depth observation of the current operation and management status of combustion devices. During the operation of a 300MW circulating fluidized bed boiler, the CO concentration suddenly rose from below 50ppm to 150ppm. Faced with the same anomaly, shift supervisor A believed that "the total air volume is insufficient," shift supervisor B believed that "the secondary air ratio is not good," and shift supervisor C believed that "it may just be a drift of the measuring probe." These three different judgments correspond to three different approaches. It can be seen that the situational awareness process of the operators—from seeing the phenomenon of rising CO to analyzing the cause of incomplete combustion—is essentially a process of cross-verification using multiple physical laws. If this reasoning logic can be transformed into the automatic calculation of constraint equations and the automatic matching of violation patterns, the control system can autonomously complete this cognitive leap. Therefore, this application deconstructs this cognitive process into six clear levels, L0-L5, each level corresponding to a working mode of the control system, with automatic switching between levels triggered by technical conditions.

[0027] Among related technologies, a boiler status monitoring and anomaly alarm scheme is proposed. The main idea is as follows: First, acquire boiler operational monitoring data and maintenance personnel data. Then, determine the type of boiler status anomaly based on the monitoring data. These anomaly types include burner anomalies, fuel and air supply anomalies, and boiler body anomalies. Second, determine different alarm levels based on the type and number of anomalies. Finally, classify alarms by combining the relative positional distribution data of maintenance personnel between the boiler room and the monitoring screen. The technical route of this scheme can be summarized as "parameter acquisition → anomaly classification → graded alarm." From a situational awareness perspective, this scheme essentially enhances and refines the anomaly detection layer. By introducing personnel location awareness and alarm grading, it improves the accuracy of anomaly alarms.

[0028] Among related technologies, a scheme for data-driven model training based on physical knowledge has been proposed. The main idea is as follows: First, acquire historical operating data of industrial boilers and perform data preprocessing. Second, establish a physical knowledge neural network model containing mass conservation equations and energy conservation equations, and simulate the boiler's operating behavior through this physical knowledge neural network. Third, establish an improved adversarial generative network model, using adversarial training between the generator and discriminator to generate training data to expand the sample set. Finally, use the training sample set to train the diagnostic model, and use the test set for fault diagnosis testing. The role of physical knowledge (mass conservation equations and energy conservation equations) is to constrain the neural network training, i.e., a physical knowledge-assisted learning method. It is embedded in the network structure or loss function to guide the network to learn mapping relationships that conform to physical laws.

[0029] Among related technologies, a fault diagnosis scheme based on structural mechanism models has also been proposed. The main idea is to adopt a dual-drive approach of "structural mechanism model + operational data". Here, "structural mechanism" mainly refers to calculations at the model level of the boiler thermal system. For example, it can verify the consistency of operational data based on overall thermal balance and material balance, or deduce internal state variables that cannot be directly measured.

[0030] However, when diagnosing faults in boiler combustion conditions, the relevant technologies do not address the cognitive leap from "discovering an anomaly" to "knowing why the anomaly occurred." The anomaly classification is mainly based on the equipment location where the anomaly occurred rather than on the physical root cause of the anomaly. It can only indicate that "a certain type of equipment may have a problem," rather than "which physical law has been violated and what problem the violated pattern reveals." Secondly, the final decision-maker of the scheme is still the trained neural network itself. The conservation equation does not directly participate in the judgment logic of diagnosis, which leads to three derivative problems: (1) It needs to rely on a large amount of balanced historical fault data to train the generator and the diagnostic network. For fault modes that have never appeared in the training data, the method does not have reliable diagnostic capabilities. For newly built combustion devices or old devices with limited operating data records, sufficient historical fault data often does not exist; (2) The diagnostic output is in the form of fault category labels, which cannot give a physical causal chain explanation of "which conservation law was violated, what the direction of the violation is, and why this violation mode uniquely corresponds to this root cause"; (3) The scheme is a single-layer diagnostic architecture that only performs fault classification and does not have the hierarchical progression and closed-loop control capabilities from no perception to decision perception. Furthermore, the thermal system mechanism of the scheme can only impose constraints on network training at the modeling level (i.e., it remains at the system-level thermal equilibrium level). The distributed control system (DCS) or programmable logic controller (PLC) and other control devices on site can only realize basic threshold comparison and over-limit alarm functions. They have not gone deep into the material conservation level of individual chemical elements, nor have they established a logic for judging the coordinated violation between multiple constraint equations. For example, the scheme cannot distinguish between two completely different abnormal situations: "carbon conservation violation alone" and "carbon conservation and energy conservation violation simultaneously in the same direction".

[0031] In summary, the relevant technologies have the following shortcomings when diagnosing faults in boiler combustion conditions: (1) Lack of a systematic situational awareness classification system: The operation monitoring of the combustion device is regarded as a binary judgment of normal and abnormal, or as a classification problem of multiple types of faults. Neither of these systems systematizes the complete cognitive process of the operator into a quantifiable, assessable, and automatically executable technical architecture; (2) The cognitive level leap is entirely dependent on human experience: The most critical step from "discovering the abnormality" to "knowing why the abnormality is" is left entirely to the individual operator. Different values ​​of the same abnormal condition may lead to completely different judgments; (3) The physical conservation law is only used as a training aid rather than a direct diagnostic tool. It lacks the ability to conduct multi-constraint collaborative diagnosis, cannot distinguish the causal direction of the deviation, and generally relies on historical fault data; (4) It does not have the high-order cognitive reasoning ability of the operator to directly locate the root cause of the fault from the parameter abnormality, does not have the ability to autonomously reason and locate the fault, and cannot replace the human to complete the key judgment process.

[0032] Based on the aforementioned problems, this application provides a combustion situation awareness and grading system. This system, through a root cause diagnosis module, calculates the current violation degree of at least one physicochemical constraint based on the current operating parameters when any operating parameter of the target combustion device meets a preset abnormal state alarm condition. When the duration for which the current violation degree of any physicochemical constraint exceeds a preset violation threshold reaches a preset duration, the system outputs the current root cause diagnosis result based on a preset violation pattern library. Furthermore, a trend prediction module identifies the target physicochemical constraint whose current violation degree exceeds the corresponding preset violation threshold, and outputs the current trend prediction result when the time series of the violation degree corresponding to the target physicochemical constraint meets a preset deterioration condition. Finally, a decision output module generates a target control strategy based on the current root cause diagnosis result and the current trend prediction result, and controls the target combustion device according to the target control strategy. This solves the problems of low accuracy, incomplete fault tracing, and poor reliability in identifying unknown abnormal conditions in related technologies when diagnosing boiler combustion conditions. It significantly improves the accuracy of diagnostic results, reduces the probability of false alarms and missed alarms, and enhances the timeliness and rationality of boiler combustion control.

[0033] Specifically, Figure 1 This is a block diagram of a combustion situation perception and classification system provided in an embodiment of this application.

[0034] like Figure 1 As shown, the combustion situation awareness and classification system 10 includes: a data perception module 100, an anomaly detection module 200, a root cause diagnosis module 300, a trend prediction module 400, and a decision output module 500.

[0035] The system includes: a data sensing module 100, used to collect the current operating parameters of the target combustion device when the combustion status sensing level is at the data sensing level; an anomaly detection module 200, used to switch the combustion status sensing level from the anomaly detection level to the root cause diagnosis level when any operating parameter meets a preset anomaly alarm condition when the combustion status sensing level is at the anomaly detection level; and a root cause diagnosis module 300, used to calculate the current violation degree of at least one physicochemical constraint based on the current operating parameters when the combustion status sensing level is at the root cause diagnosis level, and when the current violation degree of any physicochemical constraint exceeds the corresponding preset violation threshold for a preset duration, and then, based on a preset violation pattern library... The system outputs the current root cause diagnosis result and controls the combustion situation awareness level to switch from the root cause diagnosis level to the trend prediction level. The trend prediction module 400 is used to determine the target physicochemical constraint whose current violation degree is greater than the corresponding preset violation threshold when the combustion situation awareness level is at the trend prediction level. When the violation degree time series corresponding to the target physicochemical constraint meets the preset deterioration condition, the system outputs the current trend prediction result and controls the combustion situation awareness level to switch from the trend prediction level to the decision output level. The decision output module 500 is used to generate a target control strategy based on the current root cause diagnosis result and the current trend prediction result when the combustion situation awareness level is at the decision output level, and to control the target combustion device according to the target control strategy.

[0036] Among them, the preset abnormal state alarm conditions and preset deterioration conditions can both be conditions preset by the user, which can be conditions obtained through a limited number of experiments or conditions obtained through a limited number of computer simulations; the preset violation threshold can be a threshold preset by the user, which can be a threshold obtained through a limited number of experiments or a threshold obtained through a limited number of computer simulations; the preset duration can be a duration preset by the user, which can be a duration obtained through a limited number of experiments or a duration obtained through a limited number of computer simulations, and there is no specific limitation.

[0037] Specifically, in this embodiment, the current operating parameters of the target combustion device can first be collected by the sensor group in the data sensing module 100. These current operating parameters include fuel input parameters, flue gas output parameters, working fluid side parameters, and solid product parameters, presented in numerical and trend curve forms. Secondly, in this embodiment, the anomaly detection module 200 can compare the current operating parameters with preset operating parameter thresholds to obtain parameter comparison results. When any operating parameter in the parameter comparison results meets a preset abnormal state alarm condition, an abnormal state alarm is triggered. Furthermore, in this embodiment, the root cause diagnosis module 300 can calculate the current violation degree of at least one physical and chemical constraint based on the current physical and chemical constraint conditions (i.e., the pre-established physical and chemical constraint equation set) and the current operating parameters. The current violation degree is the absolute value of the normalized deviation between the theoretical value on the left side of each physical and chemical constraint equation and the measured value on the right side. When the current violation degree of any physical and chemical constraint exceeds the corresponding preset violation threshold for a duration exceeding the preset threshold, the corresponding constraint is determined to be violated. Based on the type and direction of the violated constraint, the corresponding abnormal root cause is determined, completing the automatic transition from the abnormal perception state to the root cause perception state. Next, in this embodiment, the trend prediction module 400 can calculate the time derivative of the constraint violation degree. When the time derivative indicates that the violation degree is continuously increasing, a trend deterioration warning signal is output. Finally, in this embodiment, the decision output module 500 can retrieve a matching target control strategy from a pre-established intervention strategy library based on the abnormal root cause and the trend deterioration warning signal, and send it to the control unit (i.e., the execution mechanism). In this application, the target combustion device is a boiler, including but not limited to a pulverized coal boiler (PC) and a circulating fluidized bed boiler (CFB).

[0038] Therefore, the embodiments of this application can transform the situational awareness ability of operators—from seeing data, detecting anomalies, judging causes, predicting trends, and taking action—into standardized technical functions that the control system can automatically execute. The operational situational awareness capability of the combustion device is quantified into six technical levels, L0 to L5: L0 (no awareness, insufficient sensor coverage), L1 (data display), L2 (anomaly alarm), L3 (root cause diagnosis, physical constraint violation determination), L4 (trend prediction, violation degree derivative analysis), and L5 (closed-loop control, automatic output of control commands). Each level has clearly defined input, output, and automatic switching conditions, and strict automatic transition judgment conditions exist between levels, forming a complete closed loop from raw sensor signals to control commands from the actuator.

[0039] Furthermore, to enable those skilled in the art to better understand the application prerequisites of the combustion situation awareness and classification system 10 of this application, the following description is provided in conjunction with specific embodiments.

[0040] As one possible implementation, some embodiments further include: a non-perceptive module, used to acquire the percentage of the number of acquisition signals of the acquisition components in the target combustion device when the combustion situation perception level is non-perceptive, and to control the combustion situation perception level to switch from the non-perceptive level to the data perception level when the percentage of the number of acquisition signals reaches a preset ratio.

[0041] The preset ratio can be a ratio pre-set by the user, or it can be a ratio obtained through a limited number of experiments, or it can be a ratio obtained through a limited number of computer simulations. No specific limitation is made here.

[0042] Specifically, when the combustion situation awareness level is at the non-perceptive level, this embodiment of the application can acquire analog signals (4-20mA current signals or thermocouple millivolt signals) or digital signals (Modbus RTU / TCP or OPC UA protocol) through the acquisition component in the non-perceptive module. After performing signal conditioning, analog-to-digital conversion and communication protocol parsing on the analog or digital signals, it provides calibration data to the upper-layer module. Based on the calibration data, it determines the proportion of the number of acquired signals (i.e., the online coverage rate of the core measurement points). When the proportion of the number of acquired signals reaches a preset ratio (the default value is 95%), it controls the combustion situation awareness level to switch from the non-perceptive level to the data awareness level. When the proportion of the number of acquired signals is less than the preset ratio, the control device determines that the current data foundation is insufficient to support meaningful situation judgment and controls the combustion situation awareness level to remain at the non-perceptive level. In this embodiment, the non-sensing module consists of sensor groups and measuring instruments installed on the target combustion device body and various auxiliary systems. The acquisition components include temperature sensors (thermocouples and resistance thermometers, arranged in layers according to the furnace height), pressure transmitters (arranged in various air ducts, flues, and steam-water pipelines), flow meters (air volume measuring device, steam flow measuring device, coal feed measuring device), flue gas analyzers (arranged at economizer outlet, furnace outlet, before and after the Selective Catalytic Reduction (SCR) device, and before and after the Flue Gas Desulfurization (FGD) device, etc., to measure O2, CO, SO2, and NOx concentrations), online coal quality analyzers (arranged at the coal feeder outlet, based on the principle of Laser-Induced Breakdown Spectroscopy (LIBS) or Prompt Gamma Neutron Activation Analysis (PGNAA)), and online fly ash carbon content detection devices, etc.

[0043] Therefore, the embodiments of this application can maintain an unperceived state when the signal acquisition integrity is insufficient. By statistically analyzing the proportion of acquired signals in real time and comparing it with a preset ratio, the system can automatically complete the step-by-step switching from the unperceived level to the data perception level after the signal integrity meets the standard. This ensures that the following control logic is activated only when the data is reliable, avoiding misjudgment of combustion conditions due to missing measurement points, signal failure, etc., and improving the stability and accuracy of the operation of this application.

[0044] Furthermore, to enable those skilled in the art to further understand the specific implementation principles of the data sensing module 100, anomaly detection module 200, root cause diagnosis module 300, trend prediction module 400, and decision output module 500 in the combustion situation perception and classification system 10 of this application, the following description is provided in conjunction with specific embodiments.

[0045] First, the specific implementation principle of the data sensing module 100 in the combustion situation perception and classification system 10 of this application will be introduced.

[0046] Specifically, when the combustion status perception level of the target combustion device is the data perception level, the embodiment of this application can collect the current operating parameters of the target combustion device from the sensor group in real time according to the preset cycle through the data perception module 100, and organize the current operating parameters into the following five parameter domains: (1) The fuel side parameter domain includes: coal mass flow rate, coal elemental analysis data (including: received carbon content, sulfur content, nitrogen content, oxygen content and hydrogen content), received low heating value, received ash content and dry ash-free volatile matter; (2) The air side parameter domain includes: total air volume, primary air volume, secondary air volume, air volume of each air duct and corresponding air temperature; (3) The flue gas side parameter domain includes: economizer outlet, furnace outlet, concentration of each component of flue gas before and after the selective catalytic reduction denitrification device SCR, and before and after the flue gas desulfurization device FGD (including: , , and (3) Flue gas temperature field distribution, flue gas pressure and flue gas flow rate (the above measuring points are in an "or" relationship, and all or some of the measuring points are selected for collection according to the actual boiler measuring point configuration); (4) Working fluid side parameter domain includes: main steam temperature, pressure and flow rate, reheat steam temperature, pressure and flow rate, feedwater temperature and pressure; (5) Solid product parameter domain includes: fly ash mass flow rate and its carbon content, bottom ash or slag mass flow rate and its carbon content. Among them, the preset cycle is usually 1s~5s, which matches the typical data refresh cycle of DCS or PLC. In addition, the data sensing module 100 can present the above current operating parameters in the form of numerical values ​​and trend curves on the control device interface, and send structured data packets to the anomaly detection module 200.

[0047] Secondly, the specific implementation principle of the anomaly detection module 200 in the combustion situation perception and classification system 10 of this application is introduced.

[0048] In one possible implementation, in some embodiments, the anomaly detection module 200 includes: a detection unit, configured to compare each operating parameter in the current operating parameters with a corresponding preset operating parameter threshold to obtain a parameter comparison result when the combustion situation perception level is an anomaly detection level; and a first control unit, configured to control the combustion situation perception level to switch from the anomaly detection level to the root cause diagnosis level when any operating parameter in the parameter comparison result meets a preset abnormal state alarm condition.

[0049] Among them, the preset operating parameter thresholds can be the design operating range given by the equipment manufacturer, the safety limits stipulated by industry technical regulations, and statistical thresholds derived from statistical analysis of historical data during the stable normal operation period of the boiler.

[0050] Specifically, when the combustion situation awareness level is at the anomaly detection level, this embodiment of the application can use the detection unit in the anomaly detection module 200 to compare each operating parameter in the current operating parameters with its corresponding preset operating parameter threshold to obtain the parameter comparison result. Then, the first control unit in the anomaly detection module 200 triggers an alarm, generates and sends an anomaly data packet to the root cause diagnosis module 300 when any operating parameter in the parameter comparison result meets a preset anomaly state alarm condition (i.e., any operating parameter is greater than its corresponding preset operating parameter threshold). Simultaneously, it controls the combustion situation awareness level to switch from the anomaly detection level to the root cause diagnosis level. The anomaly data includes an anomaly parameter identifier, the current operating parameter, the preset operating parameter threshold, and the duration of exceeding the limit.

[0051] Next, the specific implementation principle of the root cause diagnosis module 300 in the combustion situation perception and classification system 10 of this application will be introduced.

[0052] In one possible implementation, in some embodiments, the root cause diagnosis module 300 includes: a first determining unit, configured to determine the current physical and chemical constraints when the combustion situation perception level is at the root cause diagnosis level; a calculation unit, configured to calculate the current violation degree of at least one physical and chemical constraint based on the current physical and chemical constraints and current operating parameters; a first output unit, configured to output the current root cause diagnosis result according to a preset violation pattern library when the duration for which the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold reaches a preset duration; and a second control unit, configured to control the combustion situation perception level to switch from the root cause diagnosis level to the trend prediction level when the duration for which the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold reaches a preset duration.

[0053] Specifically, when the combustion situation perception level is at the root cause diagnosis level, this embodiment of the application can determine the current physicochemical constraints through the first determining unit in the root cause diagnosis module 300. The current physicochemical constraints include the following seven built-in general physicochemical constraint equations: C1 carbon conservation, C2 sulfur conservation, C3 nitrogen conservation, C4 stoichiometry, C5 energy conservation, C6 solid product mass conservation, and C7 working fluid side mass conservation.

[0054] Furthermore, in this embodiment of the application, the calculation unit in the root cause diagnosis module 300 can calculate the current violation degree of at least one physical and chemical constraint based on the current physical and chemical constraints and the current operating parameters. The current violation degree is... : ; in, The degree of violation at present; This represents the theoretical value on the left-hand side of the general physical and chemical constraint equation; The measured value on the right side of the general physical and chemical constraint equation is given.

[0055] It should be understood that the C1 carbon conservation constraint is: the mass of carbon entering the furnace should be equal to the carbon in the flue gas ( carbon addition The carbon in the coal (including fly ash carbon and bottom ash carbon) indicates a carbon imbalance, which may stem from deviations in coal feed measurement, lags in coal element analysis, or sampling errors in flue gas / ash. In this embodiment, the carbon conservation constraint equation for C1 can be set as: ; in, The current degree of violation of the carbon conservation constraint; Coal feed rate (kg / s); Carbon content of the fuel as received (%) In the flue gas Mass flow rate (kg / s); In the flue gas Mass flow rate (kg / s); The mass flow rate of fly ash (kg / s); The mass flow rate of the bottom ash is (kg / s). Carbon content (%) of fly ash; The carbon content (%) of the bottom ash.

[0056] It should be understood that the C2 sulfur conservation constraint is: fuel sulfur should be equal to flue gas sulfur. The sulfur in the solid sulfate, combined with the sulfur in the sulfur-containing solid sulfate, indicates a sulfur imbalance, suggesting a problem with the feed rate or quality of the desulfurizing agent. In this embodiment, the C2 sulfur conservation constraint equation can be set as: ; in, The current degree of violation of the sulfur conservation constraint; The sulfur content of the fuel as received (%). In the flue gas Mass flow rate (kg / s); The total ash and slag mass flow rate (kg / s); The mass fraction (%) of calcium sulfate in the ash residue.

[0057] It should be understood that the C3 nitrogen conservation constraint is: fuel nitrogen plus air nitrogen should equal the sum of all forms of nitrogen in the flue gas, used to assist in the determination. Anomaly. In this embodiment, the conservation constraint equation for nitrogen (C3) can be set as: ; in, This represents the current degree of violation of the nitrogen conservation constraint. Nitrogen content (%) of fuel received; Air mass (kg / s); The mass fraction of nitrogen in the air (%) The mass of flue gas is (kg / s). The mass fraction (%) of nitrogen in the flue gas.

[0058] It should be understood that the C4 stoichiometric constraint is: actual oxygen consumption (air supply) Reduce smoke The stoichiometric coefficient should be equal to the theoretical oxygen demand calculated based on fuel elemental analysis, used to determine combustion completeness. In this embodiment, the C4 stoichiometric coefficient constraint equation can be set as: ; in, The current degree of violation of the stoichiometric constraint; The amount of oxygen fed into the furnace (kg / s); Free oxygen content in exhaust gas (kg / s); The amount of oxygen not consumed during incomplete combustion (kg / s). It can be calculated based on fuel elemental analysis (C / H / S / O) and combustion reaction equations, where, The calculation formula is: ; in, The oxygen content (%) of the fuel received.

[0059] It should be understood that the C5 energy conservation constraint is: the heat released by the fuel should equal the sum of the heat absorbed by the working fluid, the heat loss from flue gas, the physical heat loss from ash and slag, and the heat dissipation loss. Energy imbalance indicates abnormal conditions or leaks in the heated surface. In this embodiment, the C5 energy conservation constraint equation can be set as: ; in, The current degree of violation of the energy conservation constraint; The lower heating value of the fuel is calculated as received (kJ / kg). The total mass of steam produced by the boiler (kg / s); Enthalpy rise of vapor (kJ / kg); Heat loss from flue gas (%) Physical heat loss of ash and slag (%) For heat loss (%).

[0060] It should be understood that the mass conservation constraint for C6 solid products is: the amount of ash fed into the furnace should equal the amount of fly ash plus the amount of bottom ash / slag. Imbalance indicates sampling or metering deviation in ash and slag. In this embodiment, the mass conservation constraint equation for C6 solid products can be set as: ; in, The current degree of violation of the mass conservation constraint of solid products; Ash content (%) is the mass fraction of the fuel received.

[0061] It should be understood that the C7 working fluid side mass conservation constraint is: the total feedwater volume should equal the main steam flow rate plus the reheat steam flow rate plus the blowdown flow rate plus the leakage flow rate. The working fluid side mass imbalance directly points to leakage at the heating surface, which is one of the most reliable safety diagnostic constraints. In this embodiment, the C7 working fluid side mass conservation constraint equation can be set as: ; in, The current degree of violation of the mass conservation constraint on the working fluid side; Total boiler feedwater (kg / s); Main steam output (kg / s); The reheat steam flow rate is (kg / s). This represents the total boiler blowdown volume (kg / s).

[0062] Furthermore, at the current degree of violation of any physical or chemical constraint Greater than the corresponding preset violation threshold δ i The duration reaches the preset duration (i.e., the trigger time window). In this embodiment, the first output unit in the root cause diagnosis module 300 can output the current root cause diagnosis result according to the preset violation pattern library. Simultaneously, the second control unit in the root cause diagnosis module 300 controls the combustion situation perception level to switch from the root cause diagnosis level to the trend prediction level. The preset violation pattern library contains twenty-two constraint violation patterns. As shown in Table 1, Table 1 is a constraint violation pattern table provided in this embodiment. This embodiment lists the following five representative constraint violation patterns to illustrate the construction method and judgment logic of the preset violation pattern library.

[0063] Table 1

[0064] It should be understood that, as shown in Table 1, in Mode P01 (single constraint violation): the right side of the carbon conservation constraint is too small (the measured total carbon is lower than the theoretical total carbon), while the other constraints are normal. In this case, a single constraint violation is only marked as abnormal and the root cause is not directly determined, because the decrease in carbon content may be due to changes in coal quality or deviations in fly ash sampling. The coal quality analysis report needs to be manually checked. In Mode P08 (multiple constraints in the same direction violation): the carbon conservation constraint and the energy conservation constraint are simultaneously violated in the same direction (both right sides are smaller than the left side). In this case, the violation of both constraints in the same direction uniquely determines that the coal feed measurement system is too high (both carbon and energy are proportional to the coal feed, so they must deviate in the same direction and proportionally). It is recommended to calibrate the coal feed measurement device. Mode P12 (Multi-Constraint Collaborative Violation Type): Oxygen metering ratio constraint is violated (actual oxygen consumption is less than theoretical oxygen demand), carbon conservation constraint is normal but CO concentration increases. In this case, insufficient oxygen consumption but closed carbon balance indicates that the CO increase is not caused by insufficient total air volume, but by some fuel not being fully in contact with oxygen. This is judged as uneven mixing of fuel and air (secondary air ratio deviation leads to local fuel-rich areas). It is recommended to adjust the secondary air damper opening distribution. Mode P16 (CFB-Specific Sequential Causal Type): Material inventory-bed pressure constraint is violated first (bed pressure difference continues to rise), followed by energy conservation constraint violation (bed temperature drops). In this case, the order of constraint violations reveals the direction of the causal chain: material changes first → heat transfer conditions change → temperature responds later. This is judged as a flow-dominant deviation caused by abnormal material circulation loop. It is recommended to check the fluidizing air volume of the return leg and the slag discharge status of the bottom ash cooler. Mode P20 (PC-specific thermodynamic coupling type): A coordinated violation of energy conservation constraints and furnace heat transfer distribution constraints (energy imbalance accompanied by a systematic shift in the radiation / convective heat absorption ratio). In this case, it is determined to be slagging on the water-cooled wall. It is recommended to locate the slagging area based on the decrease in radiation heat absorption zone and arrange corresponding soot blowers to enhance cleaning. All of the above processes are automatically executed by the control device according to objective physical judgment criteria, without relying on the subjective judgment of operators. Furthermore, the constraint violation judgment used in this application embodiment does not rely on machine learning, artificial neural networks, or any form of historical fault data training model.

[0065] Furthermore, when the target combustion device is a circulating fluidized bed boiler, the root cause diagnosis steps in this application embodiment further include the calculation of C8 solid material inventory-bed pressure dynamic constraints and the calculation of C9 in-furnace desulfurization-combustion coupling constraints; when the target combustion device is a pulverized coal boiler, the root cause diagnosis steps in this application embodiment further include the calculation of C10 furnace heat transfer distribution constraints and the calculation of C11 burner zone stoichiometric zoning constraints.

[0066] In addition, the specific implementation principle of the trend prediction module 400 in the combustion situation perception and classification system 10 of this application is introduced.

[0067] As one possible implementation, in some embodiments, the trend prediction module 400 includes: a second determining unit, configured to determine a target physicochemical constraint whose current violation degree is greater than a corresponding preset violation threshold when the combustion situation perception level is a trend prediction level; a second output unit, configured to output the current trend prediction result when the violation degree time series corresponding to the target physicochemical constraint meets a preset deterioration condition; and a third control unit, configured to control the combustion situation perception level to switch from the trend prediction level to the decision output level when the violation degree time series corresponding to the target physicochemical constraint meets the preset deterioration condition.

[0068] Specifically, when the combustion situation perception level is the trend prediction level, this embodiment of the application can determine the target physicochemical constraint whose current violation degree is greater than the corresponding preset violation threshold through the second determination unit in the trend prediction module 400; and select the violation degree time series within the most recent sliding time window through the second output unit in the trend prediction module 400, and obtain the time derivative of the violation degree by linearly fitting the violation degree time series using the least squares method. / When the time derivative of the degree is violated When the value is positive and continues to increase, the violation time series is determined to meet the preset deterioration condition. Simultaneously, a trend deterioration alarm and the current trend prediction result are output. At this point, the third control unit in the trend prediction module 400 controls the combustion situation perception level to switch from the trend prediction level to the decision output level. The default length of the sliding time window is 300 seconds.

[0069] Finally, the specific implementation principle of the decision output module 500 in the combustion situation perception and classification system 10 of this application is introduced.

[0070] In one possible implementation, in some embodiments, the decision output module 500 includes: a generation unit, used to generate a target control strategy based on the current root cause diagnosis result and the current trend prediction result when the combustion situation perception level is the decision output level; and a control unit, used to control the target combustion device according to the target control strategy.

[0071] Specifically, when the combustion situation perception level is at the decision output level, this embodiment of the application can use the generation unit in the decision output module 500 to retrieve a matching target control strategy from a preset intervention strategy library based on the current root cause diagnosis result and the current trend prediction result. The target control strategy is then converted into an executable output signal (i.e., a target control command, including: valve opening analog output, inverter frequency setting, or equipment start / stop switch output) and sent to the control unit (i.e., the actuator) in the policy output module 500. The control unit controls the target combustion device according to the target control strategy and returns to the data perception module 100 to continue monitoring the control effect after execution. The preset intervention strategy library is a pre-established mapping table, where each entry contains the correspondence between a root cause type and one or more specific control strategies, as well as applicable preconditions.

[0072] Therefore, this application embodiment systematizes the situational awareness process of operators of carbon-containing fuel combustion devices into a quantifiable, reproducible, and automatically executed hierarchical system by the control system. It can accurately define the data input type, information processing depth, and output granularity of each of the six levels from L0 to L5, so that the situational awareness capability of each level can be independently quantified and evaluated. It can realize the automatic start-up of L0→L1 and the automatic leap from L2 to L3, that is, the cognitive leap process of "detecting anomalies → judging the root cause" which is completely completed by humans in related technologies is transformed into a technical judgment process that can be automatically executed by the control system. It can establish a structured data flow mechanism between levels, so that the information output by the lower level after processing can be transmitted to the higher level in the form of data packets with complete information and standardized format. Ultimately, the control system can autonomously complete the complete cognitive closed loop from no perception to decision perception, eliminating the uncertainty of human experience judgment.

[0073] Furthermore, to enable those skilled in the art to better understand the combustion situation awareness and classification system of this application, the following description is provided in conjunction with specific embodiments.

[0074] Example 1: CFB Combustion Deviation: Full-chain Diagnosis and Closed-Loop Control of Combustion Deviation in Circulating Fluidized Bed Boilers. Unperceived Level: A 300MW subcritical circulating fluidized bed boiler with a rated evaporation capacity of 1025t / h has an online coverage rate of 98% for core measuring points (greater than the preset ratio of 95%). The control switches from the unperceived level to the data-perceived level for combustion status awareness. Data-perceived Level: Bed temperature 860℃, furnace outlet... The concentration is 3.5%, CO concentration is 150 ppm (normally should be below 50 ppm), bed pressure difference is 9.2 kPa, coal feed rate is 42 kg / s, and main steam flow rate is 390 t / h. Anomaly detection level: Detection of severely exceeded CO concentration limits (upper limit threshold 50 ppm) triggers an audible and visual alarm and generates an anomaly data packet (anomaly_param_id="CO", anomaly_value=150, threshold_value=50, exceed_duration=45s). Simultaneously, the combustion status perception level switches from anomaly detection level to root cause diagnosis level. Root cause diagnosis level: The violation of the stoichiometric constraint C4 is 0.08 (greater than the corresponding preset violation threshold of 0.05), and the violation of the carbon conservation constraint C1 is 0.03 (within the normal range). The violation mode CO↑+O2 normal → is judged as uneven fuel-air mixing, with the root cause pointing to secondary air ratio deviation (diagnostic confidence level 92%). The trigger time window is met at 60 seconds. t min Given a 60-second condition, the system outputs the current root cause diagnosis result based on a preset violation pattern library and switches the combustion situation awareness level from the root cause diagnosis level to the trend prediction level. Trend prediction level: The time derivative of the C4 violation degree over the past 120-second time window is calculated to be +0.0004 s. - ¹(positive value), and the second derivative is positive, satisfying the preset deterioration condition, outputting the current trend prediction result (i.e., trend deterioration warning): It is predicted that if no intervention is made, the CO concentration will further rise to above 200ppm in about 5 minutes, and the combustion situation perception level will be switched from the trend prediction level to the decision output level. Decision output level: Two matching strategies (adjusting the opening of the upper secondary damper or adjusting the opening of the lower secondary damper) are retrieved from the intervention strategy library. Based on the specific deviation feature of high CO concentration but normal bed temperature, the target control strategy of "increasing the opening of the upper secondary damper by 5%" is selected, generating the corresponding valve opening control command and sending it to the control unit (i.e., the secondary damper actuator). After execution, the data perception level is returned for monitoring: After about 2 minutes, CO begins to decrease from 150ppm, and after 5 minutes, it stabilizes in the normal range of 35ppm, and the closed-loop control is successful.

[0075] Example 2: PC Slagging: Early Diagnosis and Location of Slagging on Water-Cooled Walls of Pulverized Coal Boilers. In a 600MW supercritical tangential pulverized coal boiler, the flue gas temperature at the furnace outlet was 25°C higher than that under clean operating conditions at the same load. The heat flux density calculated from the water-cooled wall temperature measurement data was approximately 18% lower than under clean operating conditions, and the superheater desuperheating water flow rate was approximately 30% higher than the normal value under the same load. Root Cause Diagnosis Level: The violation of energy conservation constraint C5 is 0.06 (greater than the corresponding preset violation threshold of 0.05); Further analysis of furnace heat transfer distribution constraints: The average radiative heat absorption ratio of the entire furnace decreased from the design value of 48% to 37%, while the convective heat absorption ratio increased from the design value of 52% to 63%. This systematic shift in heat transfer distribution ratio is a typical thermodynamic characteristic of slagging on water-cooled walls—the slagging layer increases the thermal resistance of the water-cooled walls, weakens radiative heat transfer, and causes more heat to be carried by the flue gas to the convective heating surface; Zoning analysis by furnace height: In the upper front wall area of ​​the furnace (elevation approximately 35-45m), the decrease in radiative heat absorption is the largest (approximately 28% lower than the design value), thus the main slagging area is located on the upper front wall of the furnace. Trend Prediction Level: Ash and Sludge Thermal Resistance growth rate A value greater than zero and a positive second derivative indicate that the thermal resistance of the ash deposits is in a phase of accelerated growth, and it is predicted that the degree of heat transfer degradation will reach a level requiring mandatory intervention in approximately 48 hours. Decision output level: Based on the slagging location results, a target control strategy is generated, and a control command is output to trigger the soot blowers in the upper area of ​​the front wall to start in a programmed sequence.

[0076] Example 3: CFB Deep Peak Shaving: Adaptive Constraint Relaxation Diagnosis under Deep Peak Shaving Conditions. A 300MW circulating fluidized bed boiler participated in deep peak shaving of the power grid, with the load decreasing from 100% rated load to 30% rated load. Anomaly Detection Level: Under low load conditions, the threshold values ​​of each parameter are automatically switched to the preset values ​​for the corresponding load segment. For example, the normal operating range of furnace outlet oxygen at low load is widened from 3%-4% to 4%-8%, and the bed temperature is widened from 850-890℃ to 800-880℃, thereby avoiding false alarms triggered by normal fluctuations under low load. After running at 30% load for about 40 minutes, the bed pressure difference continuously increased from 6.5kPa to 8.2kPa, exceeding the upper limit of the threshold of 7.5kPa for this load segment. Root Cause Diagnosis Level: The violation rate of the material inventory-bed pressure constraint C8 is 0.12 (greater than the corresponding preset violation threshold of 0.06), while the carbon conservation constraint C1 and energy conservation constraint C5 are both within the normal range. This violation pattern indicates that the bed material inventory is accumulating net, but the energy input and output on the combustion side remain balanced. This is determined to be a localized bed material accumulation caused by a decrease in fluidization quality under low load, which is a unique risk of deep peak shaving conditions. Trend Prediction Level: It is predicted that if no intervention is taken, the bed material accumulation will further deteriorate within two hours. Decision Output Level: It is recommended to appropriately increase the primary air fluidization volume to improve the fluidization quality under low load.

[0077] Example 4: PC Fuel Switching: Adaptive Situational Awareness of Fuel Switching (Coal Type Change). A 600MW supercritical pulverized coal boiler, originally burning the design coal type (received carbon content 58%, lower heating value 22MJ / kg), switched to an economic coal type (received carbon content 50%, lower heating value 19MJ / kg) due to a change in fuel supply. Anomaly Detection Level: Approximately 30 minutes after the switch, the oxygen content at the furnace outlet was detected to continuously decrease from 3.5% to 2.1% (below the lower limit), while the main steam temperature decreased by approximately 10°C. Root Cause Diagnosis Level: The violation rate of carbon conservation constraint C1 is 0.03 (within the normal range, indicating correct fuel metering), and the violation rate of energy conservation constraint C5 is 0 (within the normal range, indicating overall energy balance between input and output). However, the measured oxygen consumption of oxygen metering ratio constraint C4 is 7% higher than the theoretical oxygen demand calculated based on the elemental analysis of the new coal type. This indicates that the online elemental analysis data of the new coal type (especially carbon content and calorific value) still deviates from the actual coal fed into the furnace. The violation mode is determined as "online coal quality analysis data lags behind actual coal type changes." Trend Prediction Level: Trend analysis shows that as the mixing ratio of the designed coal type and the new coal type in the coal bunker gradually stabilizes, the C4 violation rate is slowly decreasing. Decision Output Level: Currently, combustion adjustments need to be made temporarily based on real-time feedback of operating parameters. The accuracy of online coal quality analysis data will be restored after the coal type transition in the coal bunker is completed.

[0078] Example 5: Systematic verification was completed on a 3MWth circulating fluidized bed combustion test bench, with 13 typical deviation conditions artificially set in the test. On the one hand, the diagnostic accuracy of this application embodiment reached 93%, correctly diagnosing 12 conditions, which is far superior to the 62% accuracy of the single-parameter threshold method in related technologies (accurately diagnosing 8 conditions, and outputting ambiguous conclusions for 5 other conditions with multiple causes); on the other hand, the median response time of the automatic transition from L2 to L3 level in this application embodiment was 45s, compared to the median time of 4min20s for manual judgment by operators under the same conditions, reducing the response time by approximately 83%. In addition, this application embodiment can predict the deterioration of the condition and issue an early warning 5-15 minutes in advance; and for the two composite deviation conditions that never appeared in the training samples, this application embodiment can accurately diagnose them, while the data-driven comparison method in related technologies all misdiagnosed them.

[0079] Therefore, this application embodiment transforms the operational status perception process of carbon-containing fuel combustion devices into a standardized technical process that the control system can automatically execute, realizing a six-level cognitive closed loop from raw sensor signals to control commands from actuators. It is applicable to various scenarios of operation monitoring, anomaly diagnosis, and optimized control of combustion devices using carbon-containing substances as the main fuel in the power industry and industrial heating field.

[0080] Furthermore, to enable those skilled in the art to better understand the combustion situation awareness and classification system of this application, the following is combined with... Figures 2 to 5 The specific embodiments shown will be described in detail.

[0081] Figure 2 A flowchart of a six-level situational awareness method provided in an embodiment of this application; like Figure 2 As shown, this level six situational awareness method includes the following steps: S201, L0 Sensor Group L0 Non-sensing layer.

[0082] S202, L1 Data Acquisition L1 Data Awareness Layer.

[0083] S203, L2 Anomaly Alert L2 layer.

[0084] S204, L3 Root Cause Diagnosis L3 root cause diagnosis layer, which includes the ConstraintViolation Calculation Unit, and references the Constraint Equations (7universal), Violation Pattern Library (22 patterns), and Intervention Strategy Library.

[0085] S205, L4 Trend Prediction. L4 Trend Prediction Layer.

[0086] S206, L5 Decision Control, L5 Decision Perception Layer.

[0087] S207, Actuators.

[0088] Specifically, in this six-level situational awareness method, the automatic L0→L1 activation and L2→L3 automatic transition mechanisms are both triggered by objective technical conditions and do not rely on manual judgment. L3 root cause diagnosis can call upon physicochemical consistency diagnostic methods from relevant technologies as the underlying diagnostic engine, automatically activating after an anomaly alarm is triggered and completing the L2→L3 automatic transition based on constraint violation. The quantitative criterion for L4 trend perception uses the time derivative of constraint violation as the judgment indicator, calculating the time derivative of constraint violation and outputting trend warnings. L5 closed-loop control and feedback monitoring automatically generate control commands based on root causes and trends, and return to L1 monitoring after execution. The structured data flow mechanism between levels: information is transmitted between adjacent levels through standardized field data packets.

[0089] Furthermore, Figure 3 This is a flowchart of an automatic transition determination method provided in an embodiment of this application.

[0090] like Figure 3 As shown, the automatic transition determination method includes the following steps: S301, L2 abnormal alarm.

[0091] S302, Determine if the constraint is violated. > If satisfied, proceed to step S303.

[0092] S303, Determine if the condition is satisfied: t > 0. If satisfied, proceed to step S304.

[0093] S304, L2→L3 automatic transition completed.

[0094] S305, violates pattern matching (22 types).

[0095] S306, Determine if the following condition is met: Multi-constraint coordination violation? If the multi-constraint coordination violation is met, proceed to step S307; otherwise (i.e., if the single-constraint violation is met), proceed to step S308.

[0096] S307 outputs the unique root cause plus the confidence level.

[0097] S308, mark as abnormal and recommend manual inspection.

[0098] Furthermore, Figure 4 This is a schematic diagram illustrating a constraint violation pattern matching logic provided in an embodiment of this application.

[0099] like Figure 4 As shown, the constraint violation pattern matching logic is as follows: the current violation degree of the carbon conservation constraint is... Current violation of the sulfur conservation constraint Current violation of the nitrogen conservation constraint Current violation of stoichiometric ratio constraint Current violation of the energy conservation constraint Current violation of the mass conservation constraint of solid products Current degree of violation of the mass conservation constraint on the working fluid side The data is input to the root cause diagnosis module 300. This module, on the one hand, associates with a preset violation pattern library containing 22 types of constraint violation patterns (P01-P22) to complete pattern matching. On the other hand, it outputs a complete diagnostic result including the root cause of the fault, the direction of causal transmission, the confidence level of the judgment, and the corresponding treatment suggestions.

[0100] Furthermore, Figure 5 This is a schematic diagram of a six-level situational awareness closed-loop control logic provided in an embodiment of this application.

[0101] like Figure 5 As shown, the six-level situational awareness closed-loop control logic is as follows: the operating parameters of the target combustion device (i.e., the carbon-containing fuel combustion device) are input into the situational awareness system covering levels L0 to L5. The situational awareness system outputs control commands to four major execution units: coal feed rate adjustment, air volume adjustment, desulfurizing agent adjustment, and soot blower control. After the four types of control mechanisms complete the parameter adjustment, they directly feed back to the target combustion device to form a closed-loop control link. On the other hand, they transmit the adjusted equipment operating status back to the situational awareness system, realizing full-process dynamic perception, hierarchical judgment, and continuous adaptive correction. This is supplemented by a dual-layer closed-loop operation mechanism of multi-execution mechanism collaborative control and real-time reverse feedback of operating conditions, ultimately achieving integrated intelligent closed-loop management and control of boiler combustion, desulfurization, and ash accumulation. In this embodiment, the situational awareness system is organically composed of six-level processing modules and three knowledge base units.

[0102] Therefore, the combustion situation perception and grading system proposed in this application can, through the root cause diagnosis module, calculate the current violation degree of at least one physicochemical constraint based on the current operating parameters when any operating parameter of the target combustion device meets the preset abnormal state alarm condition. When the duration for which the current violation degree of any physicochemical constraint exceeds the corresponding preset violation threshold reaches a preset duration, the system outputs the current root cause diagnosis result based on a preset violation pattern library. Furthermore, the trend prediction module determines the target physicochemical constraint whose current violation degree exceeds the corresponding preset violation threshold, and outputs the current trend prediction result when the time series of the violation degree corresponding to the target physicochemical constraint meets the preset deterioration condition. Finally, the decision output module generates a target control strategy based on the current root cause diagnosis result and the current trend prediction result, and controls the target combustion device according to the target control strategy. This solves the problems of low accuracy, incomplete fault tracing, and poor reliability in identifying unknown abnormal conditions in related technologies when diagnosing boiler combustion conditions. It can significantly improve the accuracy of diagnosis results, reduce the probability of false alarms and missed alarms, and improve the timeliness and rationality of boiler combustion control.

[0103] Next, the combustion situation perception and classification method proposed in the embodiments of this application is described with reference to the accompanying drawings.

[0104] Figure 6 This is a flowchart of the combustion situation perception and classification method proposed in the embodiments of this application.

[0105] like Figure 6 As shown, the combustion situation awareness and classification method includes the following steps: When the combustion status perception level of the target combustion device is at the data perception level, the data perception module collects the current operating parameters of the target combustion device. If any operating parameter meets the preset abnormal state alarm conditions when the combustion situation perception level is at the abnormal detection level, the abnormal detection module controls the combustion situation perception level to switch from the abnormal detection level to the root cause diagnosis level. When the combustion situation perception level is at the root cause diagnosis level, the root cause diagnosis module calculates the current violation degree of at least one physical and chemical constraint based on the current operating parameters. When the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold for a duration that reaches a preset duration, the current root cause diagnosis result is output according to the preset violation mode library, and the combustion situation perception level is switched from the root cause diagnosis level to the trend prediction level. When the combustion situation perception level is at the trend prediction level, the trend prediction module determines the target physical and chemical constraints whose current violation degree is greater than the corresponding preset violation threshold. When the violation degree time series corresponding to the target physical and chemical constraints meets the preset deterioration conditions, the current trend prediction result is output, and the combustion situation perception level is controlled to switch from the trend prediction level to the decision output level. When the combustion situation perception level is at the decision output level, the decision output module generates a target control strategy based on the current root cause diagnosis results and the current trend prediction results, and then controls the target combustion device according to the target control strategy.

[0106] Optionally, in some embodiments, before collecting the current operating parameters of the target combustion device by means of the data sensing module when the combustion status perception level of the target combustion device is at the data sensing level, the method further includes: When the combustion situation perception level is at the non-perceptive level, the non-perceptive module obtains the percentage of the number of acquisition signals of the acquisition components in the target combustion device, and controls the combustion situation perception level to switch from the non-perceptive level to the data perception level when the percentage of the number of acquisition signals reaches the preset ratio.

[0107] Optionally, in some embodiments, when the combustion situation awareness level is at the anomaly detection level, if any operating parameter meets a preset abnormal state alarm condition, the anomaly detection module controls the combustion situation awareness level to switch from the anomaly detection level to the root cause diagnosis level, including: When the combustion situation awareness level is at the abnormal detection level, the parameter comparison results are obtained by comparing each operating parameter in the current operating parameters with the corresponding preset operating parameter threshold. When any operating parameter in the parameter comparison results meets the preset abnormal state alarm conditions, the control combustion situation awareness level is switched from the abnormal detection level to the root cause diagnosis level.

[0108] Optionally, in some embodiments, the root cause diagnosis module is configured to, when the combustion situation awareness level is at the root cause diagnosis level, calculate the current violation degree of at least one physicochemical constraint based on the current operating parameters, and when the duration for which the current violation degree of any physicochemical constraint exceeds the corresponding preset violation threshold reaches a preset duration, output the current root cause diagnosis result according to a preset violation pattern library, and control the combustion situation awareness level to switch from the root cause diagnosis level to the trend prediction level, including: When the combustion situation awareness level is at the root cause diagnosis level, determine the current physical and chemical constraints; Based on the current physical and chemical constraints, calculate the current degree of violation of at least one physical and chemical constraint according to the current operating parameters; When the current violation degree of any physical or chemical constraint exceeds the corresponding preset violation threshold for a duration that reaches the preset duration, the current root cause diagnosis result is output according to the preset violation mode library, and the combustion situation perception level is switched from the root cause diagnosis level to the trend prediction level.

[0109] It should be noted that the foregoing explanation of the combustion situation perception and classification system embodiment also applies to the combustion situation perception and classification method of this embodiment, and will not be repeated here.

[0110] According to the combustion situation perception and classification method proposed in this application, the root cause diagnosis module calculates the current violation degree of at least one physicochemical constraint based on the current operating parameters when any operating parameter of the target combustion device meets the preset abnormal state alarm condition. When the duration for which the current violation degree of any physicochemical constraint exceeds the corresponding preset violation threshold reaches a preset duration, the current root cause diagnosis result is output based on a preset violation pattern library. The trend prediction module determines the target physicochemical constraint whose current violation degree exceeds the corresponding preset violation threshold, and outputs the current trend prediction result when the time series of the violation degree corresponding to the target physicochemical constraint meets the preset deterioration condition. The decision output module generates a target control strategy based on the current root cause diagnosis result and the current trend prediction result, and controls the target combustion device according to the target control strategy. This solves the problems of low accuracy, incomplete fault tracing, and poor reliability in identifying unknown abnormal conditions in related technologies when diagnosing boiler combustion conditions. It significantly improves the accuracy of diagnosis results, reduces the probability of false alarms and missed alarms, and enhances the timeliness and rationality of boiler combustion control.

[0111] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0112] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0113] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0114] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0115] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

Claims

1. A combustion situation perception and classification system, characterized in that, include: The data sensing module is used to collect the current operating parameters of the target combustion device when the combustion status sensing level of the target combustion device is at the data sensing level; An anomaly detection module is used to control the combustion situation perception level to switch from the anomaly detection level to the root cause diagnosis level if any operating parameter meets the preset abnormal state alarm conditions when the combustion situation perception level is at the anomaly detection level. The root cause diagnosis module is used to calculate the current violation degree of at least one physical and chemical constraint based on the current operating parameters when the combustion situation perception level is the root cause diagnosis level. When the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold for a duration of a preset duration, the module outputs the current root cause diagnosis result according to the preset violation mode library and controls the combustion situation perception level to switch from the root cause diagnosis level to the trend prediction level. The trend prediction module is used to determine the target physical and chemical constraint whose current violation degree is greater than the corresponding preset violation threshold when the combustion situation perception level is the trend prediction level, and output the current trend prediction result when the violation degree time series corresponding to the target physical and chemical constraint meets the preset deterioration condition, and control the combustion situation perception level to switch from the trend prediction level to the decision output level. The decision output module is used to generate a target control strategy based on the current root cause diagnosis result and the current trend prediction result when the combustion situation perception level is the decision output level, and to control the target combustion device according to the target control strategy.

2. The combustion situation perception and classification system according to claim 1, characterized in that, Also includes: The non-perceptive module is used to obtain the percentage of the number of acquisition signals of the acquisition components in the target combustion device when the combustion situation perception level is non-perceptive, and to control the combustion situation perception level to switch from the non-perceptive level to the data perception level when the percentage of the number of acquisition signals reaches a preset ratio.

3. The combustion situation perception and classification system according to claim 1, characterized in that, The anomaly detection module includes: The detection unit is used to compare each operating parameter in the current operating parameters with the corresponding preset operating parameter threshold to obtain the parameter comparison result when the combustion situation perception level is the abnormal detection level. The first control unit is used to control the combustion situation awareness level to switch from the abnormal detection level to the root cause diagnosis level when any operating parameter in the parameter comparison results meets the preset abnormal state alarm condition.

4. The combustion situation perception and classification system according to claim 1, characterized in that, The root cause diagnosis module includes: The first determining unit is used to determine the current physical and chemical constraints when the combustion situation perception level is the root cause diagnosis level. The calculation unit is used to calculate the current violation degree of at least one physical and chemical constraint based on the current physical and chemical constraints and the current operating parameters. The first output unit is used to output the current root cause diagnosis result according to the preset violation pattern library when the duration of the current violation degree of any physical or chemical constraint being greater than the corresponding preset violation threshold reaches the preset duration. The second control unit is used to control the combustion situation perception level to switch from the root cause diagnosis level to the trend prediction level when the duration for which the current degree of violation of any physical or chemical constraint is greater than the corresponding preset violation threshold reaches the preset duration.

5. The combustion situation perception and classification system according to claim 1, characterized in that, The trend prediction module includes: The second determining unit is used to determine the target physical and chemical constraint whose current violation degree is greater than the corresponding preset violation threshold when the combustion situation perception level is the trend prediction level. The second output unit is used to output the current trend prediction result when the violation time series corresponding to the target physicochemical constraint satisfies the preset deterioration condition; The third control unit is used to control the combustion situation perception level to switch from the trend prediction level to the decision output level when the violation time series corresponding to the target physical and chemical constraints meets the preset deterioration condition.

6. The combustion situation perception and classification system according to claim 1, characterized in that, The decision output module includes: The generation unit is configured to generate the target control strategy based on the current root cause diagnosis result and the current trend prediction result when the combustion situation perception level is the decision output level. The control unit is used to control the target combustion device according to the target control strategy.

7. A method for classifying combustion situation perception, characterized in that, The method is applied to the combustion situation awareness and classification system as described in any one of claims 1-6, wherein the method includes the following steps: When the combustion status perception level of the target combustion device is at the data perception level, the data perception module collects the current operating parameters of the target combustion device. If any operating parameter meets the preset abnormal state alarm conditions when the combustion situation perception level is at the abnormal detection level, the anomaly detection module controls the combustion situation perception level to switch from the abnormal detection level to the root cause diagnosis level. When the combustion situation perception level is at the root cause diagnosis level, the root cause diagnosis module calculates the current violation degree of at least one physical and chemical constraint based on the current operating parameters. When the current violation degree of any physical and chemical constraint is greater than the corresponding preset violation threshold for a duration of a preset duration, the current root cause diagnosis result is output according to the preset violation pattern library, and the combustion situation perception level is controlled to switch from the root cause diagnosis level to the trend prediction level. When the combustion situation perception level is the trend prediction level, the trend prediction module determines the target physical and chemical constraint whose current violation degree is greater than the corresponding preset violation threshold. When the violation degree time series corresponding to the target physical and chemical constraint meets the preset deterioration condition, the current trend prediction result is output, and the combustion situation perception level is controlled to switch from the trend prediction level to the decision output level. When the combustion situation perception level is at the decision output level, the decision output module generates a target control strategy based on the current root cause diagnosis result and the current trend prediction result, and controls the target combustion device according to the target control strategy.

8. The method according to claim 7, characterized in that, Before collecting the current operating parameters of the target combustion device through the data sensing module when the combustion situation perception level of the target combustion device is at the data sensing level, the method further includes: When the combustion situation perception level is at the non-perceptive level, the non-perceptive module obtains the percentage of the number of acquisition signals of the acquisition components in the target combustion device, and when the percentage of the number of acquisition signals reaches a preset ratio, controls the combustion situation perception level to switch from the non-perceptive level to the data perception level.

9. The method according to claim 7, characterized in that, The method of switching the combustion situation awareness level from the anomaly detection level to the root cause diagnosis level when any operating parameter meets a preset abnormal state alarm condition, via the anomaly detection module, includes: When the combustion situation awareness level is the anomaly detection level, the parameter comparison results are obtained by comparing each operating parameter in the current operating parameters with the corresponding preset operating parameter threshold. When any operating parameter in the parameter comparison results meets the preset abnormal state alarm condition, the combustion situation awareness level is controlled to switch from the abnormal detection level to the root cause diagnosis level.

10. The method according to claim 7, characterized in that, The root cause diagnosis module is used to calculate the current violation degree of at least one physicochemical constraint based on the current operating parameters when the combustion situation perception level is at the root cause diagnosis level, and when the current violation degree of any physicochemical constraint exceeds the corresponding preset violation threshold for a preset duration, output the current root cause diagnosis result according to the preset violation pattern library, and control the combustion situation perception level to switch from the root cause diagnosis level to the trend prediction level, including: When the combustion situation awareness level is the root cause diagnosis level, determine the current physical and chemical constraints. Based on the current physical and chemical constraints, calculate the current degree of violation of at least one physical and chemical constraint according to the current operating parameters; When the duration for which the current violation degree of any physical or chemical constraint exceeds the corresponding preset violation threshold reaches the preset duration, the current root cause diagnosis result is output according to the preset violation pattern library, and the combustion situation awareness level is controlled to switch from the root cause diagnosis level to the trend prediction level.

Images