Combustion control optimization method and system for coal-fired boiler

Abstract

The invention provides a combustion control optimization method and system for a coal-fired boiler, and relates to the technical field of combustion optimization of coal-fired boilers. The method comprises the steps that operation record data of the coal-fired boiler are collected, the operation record data comprise combustion control record data and combustion output record data, a plurality of candidate rollback time periods are recognized from the combustion control record data, dominant parameter recognition is conducted on each candidate rollback time period, dominant control parameters of each candidate rollback time period are determined, and the dominant control parameters of each candidate rollback time period are determined; the method comprises the steps of obtaining a plurality of candidate fallback time periods, constructing a state fallback feature of each candidate fallback time period, performing background grouping on the plurality of candidate fallback time periods according to the state fallback features, generating a plurality of fallback background sets, constructing a fallback structural domain of each fallback background set, and performing combustion control optimization on the coal-fired boiler based on the plurality of fallback structural domains. According to the method, the combustion control process is optimized by analyzing the combustion state change characteristics in the combustion adjustment process, so that the requirements of low-nitrogen combustion and energy-saving operation are better met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal-fired boiler combustion optimization, and in particular to a combustion control optimization method and system for a coal-fired boiler. BACKGROUND

[0002] As an important energy conversion device in thermal power units, the combustion operation state of a coal-fired boiler directly affects the pollutant emission level and energy utilization efficiency. Under the background of increasing environmental protection and energy saving requirements, the operation control of a coal-fired boiler needs to consider the control of nitrogen oxide emissions and the improvement of fuel utilization efficiency while ensuring the safe and stable operation of the unit. In engineering practice, the combustion process is usually optimized by adjusting the air distribution, coal supply and related operation parameters to achieve the goals of low-nitrogen combustion and energy saving and consumption reduction.

[0003] However, the combustion process of a coal-fired boiler has characteristics such as multi-parameter coupling and frequent changes in operating conditions, and the mutual influence between different adjustment measures is significant. In actual operation, when the combustion state deviates, the combustion process needs to be corrected by adjusting related parameters, but since the combustion state is affected by multiple factors, part of the operating state is difficult to restore to the original stable level after adjustment, which may cause the deterioration of emission indicators or the increase of energy consumption level. The existing control methods mainly focus on adjusting a single parameter or local indicators, and do not consider the overall characteristics of the combustion behavior and its evolution process under the combined state of multiple parameters.

[0004] Therefore, there is an urgent need for a method that can analyze the change characteristics of the combustion state of a coal-fired boiler during the adjustment process to support the optimization control of the combustion process and provide a basis for reducing pollutant emissions and improving operating efficiency. SUMMARY

[0005] To solve the above technical problems, the present application provides a combustion control optimization method and system for a coal-fired boiler, and the technical solutions adopted are as follows: The present application provides a combustion control optimization method for a coal-fired boiler, comprising: Collecting operation record data of the coal-fired boiler, the operation record data including combustion control record data and combustion output record data, determining a plurality of control parameters related to the operation control of the coal-fired boiler, and identifying a plurality of candidate rollback periods from the combustion control record data; Identifying the dominant parameters of each candidate rollback period based on the combustion output record data, determining the dominant control parameters of each candidate rollback period, and constructing the state rollback characteristics of each candidate rollback period in combination with the combustion control record data and the combustion output record data; The state rollback features of the plurality of candidate rollback periods are background grouped to generate a plurality of rollback background sets, rollback domains of each rollback background set are constructed, and the coal-fired boiler is controlled and optimized based on the plurality of rollback domains.

[0006] Preferably, identifying the plurality of candidate rollback periods from the combustion control record data comprises: A plurality of control parameter sequences of the combustion control record data are constructed, a plurality of sequence segments of each control parameter sequence are extracted to correspond to sliding difference features respectively, and state labels of each sequence segment are identified according to the sliding difference features, wherein the state labels include stable states and change states; Each control parameter sequence is traversed based on the state labels of the sequence segments, and a plurality of local rollback segments in each control parameter sequence are extracted based on a preset rollback reference structure, wherein the local rollback segments are composed of stable segments, change segments and rollback segments; The local rollback segments of the plurality of control parameter sequences are fused by period overlap to generate a plurality of candidate rollback periods corresponding to the combustion control record data.

[0007] Preferably, the dominant parameter of each candidate rollback period is identified based on the combustion output record data, and the dominant control parameter of each candidate rollback period is determined, comprising: State platform features of each combustion output parameter in each candidate rollback period corresponding to stable segments, change segments and rollback segments are extracted from the combustion output record data, change deviation values and regression deviation values of each combustion output parameter are determined according to the plurality of state platform features, and the irreversible degree of the combustion output parameter is calculated based on the change deviation values and the regression deviation values; The dominant output parameter of the candidate rollback period is determined according to the irreversible degree, a combustion output sequence of the dominant output parameter is constructed, control response detection of each control parameter is performed based on the combustion output sequence, control response parameters of each control parameter with respect to the combustion output sequence are calculated, and the dominant control parameter of the candidate rollback period is determined according to the plurality of control response parameters.

[0008] Preferably, the state rollback features of each candidate rollback period are constructed in combination with the combustion control record data and the combustion output record data, comprising: A rollback state window of the candidate rollback period is determined according to the state platform features of the change segments, a plurality of rollback state sequences of the rollback state window are extracted from the combustion control record data, combustion background features of the candidate rollback period are extracted from the combustion control record data, and the state rollback features of the candidate rollback period are generated in combination with the dominant output parameter and the dominant control parameter of the candidate rollback period.

[0009] Preferably, the rollback domains of each rollback background set are constructed, comprising: Based on combustion background characteristics, dominant output parameters, and dominant control parameters, multiple candidate rollback time periods are grouped to generate multiple rollback background sets. The rollback label for each group of state rollback features in the rollback background set is determined based on the irreversibility of the combustion output parameters. The rollback label includes reversible rollback and irreversible rollback. Based on the backoff labels, the envelope structure sample and structure verification sample of each backoff background set are determined. Based on the multiple backoff state sequences in the envelope structure sample, the envelope structure range of each control parameter is determined. Based on the multiple envelope structure ranges, multiple initial structural domains of the backoff background set are constructed. Structural verification is performed on each initial structural domain based on multiple backtracking state sequences in the structural verification sample. The false alarm index of each initial structural domain is calculated, and the backtracking structural domain of the backtracking background set is determined from multiple initial structural domains based on the false alarm index.

[0010] Preferably, optimizing combustion control of a coal-fired boiler based on multiple back-off structural domains includes: Acquire real-time operating data of coal-fired boilers, construct real-time status features based on real-time operating data, match real-time status features with combustion background features of multiple backtracking background sets, and determine the target backtracking structure domain corresponding to the real-time operating data. Extract the real-time state sequence from the real-time operation data, perform irreversible structural proximity detection on the real-time state sequence based on the target back-off structural domain, generate operation risk parameters for the real-time operation data, and generate a real-time control optimization strategy for the coal-fired boiler based on the real-time operation data and the target back-off structural domain.

[0011] A second aspect of the present invention provides a combustion control optimization system for a coal-fired boiler, and a combustion control optimization method for a coal-fired boiler for implementing the above-mentioned claims, comprising: The rollback identification module is used to collect the operation record data of the coal-fired boiler, including combustion control record data and combustion output record data, to determine multiple control parameters for the operation control of the coal-fired boiler, and to identify multiple candidate rollback periods from the combustion control record data; The rollback feature extraction module is used to identify the dominant parameters for each candidate rollback period based on combustion output recording data, determine the dominant control parameters for each candidate rollback period, and construct the state rollback features for each candidate rollback period by combining combustion control recording data and combustion output recording data. The combustion control optimization module is used to group multiple candidate back-off periods into backgrounds based on state back-off characteristics, generate multiple back-off background sets, construct a back-off structure domain for each back-off background set, and optimize the combustion control of the coal-fired boiler based on multiple back-off structure domains.

[0012] The present invention has the following beneficial effects: This invention, through joint analysis of combustion control and output data recorded during the operation of a coal-fired boiler, constructs a state retreat feature reflecting the characteristics of combustion state changes, focusing on combustion regulation behavior and its retreat process. It then groups and analyzes retreat processes with similar regulation characteristics under different operating backgrounds, forming corresponding retreat structure descriptions. By introducing a method to characterize the irreversible characteristics of the retreat process, this invention can identify unfavorable combustion state characteristics easily formed under specific operating conditions and regulation modes from the perspective of multi-parameter combination states. Furthermore, it characterizes the evolution range of the combustion state based on the retreat structure domain. This invention, starting from the combustion state change process, holistically characterizes combustion regulation and retreat behavior, helping to identify and avoid unfavorable operating areas that may lead to irrecoverable combustion states during combustion control. This provides a decision-making basis for optimizing coal-fired boiler combustion control strategies, improving the rationality and stability of coal-fired boiler operation control. Attached Figure Description

[0013] Figure 1 This is a schematic flowchart of a combustion control optimization method for a coal-fired boiler, provided as one embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of a combustion control optimization system for a coal-fired boiler, provided as one embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.

[0016] Please see Figure 1 The diagram illustrates a flow chart of a combustion control optimization method for a coal-fired boiler according to one embodiment of the present invention, the method comprising the following steps: Step S1: Collect the operation record data of the coal-fired boiler, including combustion control record data and combustion output record data, determine multiple control parameters for the operation control of the coal-fired boiler, and identify multiple candidate rollback periods from the combustion control record data.

[0017] Specifically, operational record data generated during the long-term operation of coal-fired boilers will be collected. This includes at least combustion control record data characterizing the operation of various adjustable parameters during the combustion process and the combustion background of the coal-fired boiler. It also includes combustion output record data reflecting the combustion results and operational effectiveness of the coal-fired boiler. Combustion control record data includes, for example, control parameters related to air and fuel supply such as primary air volume, secondary air volume, oxygen quantity, pressure, combustion temperature, coal feed rate, and coal feeder speed, used to regulate the combustion process; and combustion background parameters characterizing the coal-fired boiler under different operating conditions, such as unit load, the operational status of the coal mill combination, and the operating status of auxiliary systems such as denitrification devices and soot blowing devices. Combustion output record data includes, for example, output parameters related to pollutant emissions such as nitrogen oxide concentration and carbon monoxide concentration; output parameters related to thermal performance such as flue gas temperature and furnace temperature; and other output parameters that characterize the combustion state and operational effectiveness.

[0018] Based on the above data, and according to the operating characteristics and control requirements of coal-fired boilers, several control parameters that can characterize the combustion regulation behavior of coal-fired boilers are determined. Further analysis of the combustion control recording data identifies the time intervals during operation where the coal-fired boiler adjusts from a relatively stable state and subsequently experiences a pullback. Time intervals that meet preset conditions are identified as candidate pullback periods, resulting in multiple operating segments that characterize the adjustment and pullback behavior during combustion.

[0019] In one implementation, the process of identifying candidate rollback periods includes: Multiple control parameter sequences of combustion control recording data are constructed to characterize the temporal changes of each control parameter during the operation of the coal-fired boiler. Multiple sequence segments corresponding to each control parameter sequence are extracted, and the state labels of each sequence segment are identified based on the sliding difference features. The state labels include stable state and changing state.

[0020] In this process, multiple control parameter sequences are first processed using a sliding window, with each sliding window denoted as a sequence segment. The mean parameter value of each sequence segment is calculated, and the difference between the mean parameter value of the current segment and the mean parameter value of the previous segment is recorded as the sliding difference feature of the current segment. This method determines the sliding difference feature of each control parameter sequence across multiple sequence segments. For state label identification, if the sliding difference feature of a sequence segment is greater than a preset stability threshold, it is marked as a changing state; otherwise, it is marked as a stable state. The preset stability threshold can be reasonably set based on the stable operating states of various control parameters during normal operation of the coal-fired boiler. Its purpose is to identify a significant deviation in a control parameter caused by the equipment itself or certain operating events during the operation of the coal-fired boiler, followed by a regression process where the controller adjusts its operating conditions back to near the point of deviation.

[0021] Each control parameter sequence is traversed based on the state labels of the sequence segments. Multiple local back-back segments are extracted from each control parameter sequence based on a preset back-back reference structure. The back-back reference structure is set as a continuous state of stable-changing-stable. Each extracted local back-back segment consists of a stable segment, a changing segment, and a back-back segment, representing the process of the control parameter changing from a stable state to a changing state, such as increasing or decreasing, and then backing back to a stable state. After traversing each control parameter sequence, continuous segments that conform to the preset back-back reference structure are extracted to obtain multiple local back-back segments of the control parameter sequence.

[0022] Considering that multiple variables typically change synchronously within the same operational event during operation—for example, oxygen adjustment corresponds to airflow changes, and coal feed may also show slight correlation—it's possible that multiple control parameters exhibit regression within a given time period. To address this, local regression segments of multiple control parameter sequences are overlapped and merged. When the time overlap between two local regression segments exceeds a certain proportion, they are merged. For example, local regression segments with a time overlap greater than 70% are merged into a new segment. The final output consists of multiple candidate regression time periods corresponding to the combustion control record data, representing typical operational segments in coal-fired boiler operation where control parameter adjustments are accompanied by regression behavior.

[0023] Step S2: Identify the dominant parameters for each candidate rollback period based on combustion output recording data, determine the dominant control parameters for each candidate rollback period, and construct the state rollback features for each candidate rollback period by combining combustion control recording data and combustion output recording data.

[0024] Specifically, for each candidate rollback period identified above, based on the combustion output record data within the corresponding period, the characteristics of changes in combustion output parameters before and after the rollback period and during the rollback process are analyzed, and the degree of response of different combustion output parameters to the rollback behavior is analyzed, thereby determining the dominant combustion output parameter that plays a major role in the change of combustion state within the candidate rollback period.

[0025] Furthermore, by combining combustion control recording data, the correlation between the adjustment behavior of each control parameter and the change of the dominant combustion output parameter during the candidate rollback period is analyzed, thereby determining the dominant control parameter corresponding to the candidate rollback period. Based on the dominant control parameter and the corresponding combustion output change characteristics, a state rollback feature is constructed to characterize the combustion state change process during the candidate rollback period, so as to depict the overall performance of the combustion state during the adjustment and rollback process.

[0026] In one implementation, the process of determining the dominant control parameters for the candidate rollback period includes: Preferably, the dominant parameters for each candidate rollback period are identified based on combustion output recording data, and the dominant control parameters for each candidate rollback period include: Based on the combustion output record data, the state plateau features of each combustion output parameter in each candidate rollback period are extracted for the stable segment, the changing segment, and the rollback segment, respectively, representing the representative level of the combustion output parameter within the corresponding segment. In this embodiment, for the state plateau features of the stable segment and the rollback segment, the statistical characteristics of the combustion output parameter within the segment, such as the mean or median, can be directly used for characterization. For the state plateau features of the changing segment, the level of the most severe deviation during the change is used for characterization, such as the peak / valley value of the combustion output parameter in the changing segment.

[0027] After extracting multiple state platform features for candidate rollback periods, the variation deviation and regression deviation values ​​of each combustion output parameter are further calculated, and the irreversibility of the combustion output parameter is calculated based on these values. In this embodiment, the absolute value of the difference between the state platform features of the stable segment and the variation segment is taken as the variation deviation value, and the absolute value of the difference between the state platform features of the stable segment and the rollback segment is taken as the regression deviation value. These represent the degree of deviation of the combustion output parameter from the stable segment in the variation segment, and the degree of non-recovery from the stable segment in the rollback segment, respectively. Finally, the ratio of the regression deviation value to the variation deviation value is used as the irreversibility of the combustion output parameter to characterize whether the combustion output parameter can recover to its original operating level after adjustment and rollback. The greater the irreversibility, the more difficult it is to return to the previous state after deviation.

[0028] Based on this, the dominant output parameter for the candidate rollback period is determined according to the irreversibility, and a combustion output sequence of the dominant output parameter is constructed. Specifically, the combustion output parameter with the highest irreversibility can be used as the dominant output parameter for the candidate rollback period. The combustion state sequence of this combustion output parameter is determined based on the combustion output record data, characterizing the dynamic process of the combustion output parameter changing over time. Furthermore, control response detection is performed on each control parameter based on the combustion output sequence, and the control response parameter of each control parameter with respect to the combustion output sequence is calculated.

[0029] It is worth noting that if the irreversibility of multiple combustion output parameters is low during the process of determining the dominant output parameters, it indicates that the control behavior or operation event corresponding to the candidate rollback period is in a reversible state, that is, after the deviation occurs, it can return to the stable state before the deviation. In this case, the corresponding candidate rollback period can be used as a normal sample for comparison.

[0030] The calculation of control response parameters includes correlation analysis of the control parameter sequence and the combustion output sequence. Considering the response lag phenomenon in coal-fired boiler systems, the lag correlation between the two sequences can be calculated. For example, the correlation coefficient can be calculated at different lag periods, and the largest correlation coefficient can be selected as the control response parameter with respect to the combustion output sequence. Finally, the control parameter that responds most significantly to changes in the dominant output parameter, i.e., the control response parameter with the largest value, is selected as the dominant control parameter for the candidate backoff period, thereby achieving the identification of the dominant control parameter for the candidate backoff period based on combustion output record data.

[0031] Based on the above-mentioned feature information, the state rollback features of each candidate rollback period are constructed by combining combustion control record data and combustion output record data. This includes extracting combustion background features of the coal-fired boiler in the candidate rollback period based on combustion control record data, and extracting multiple state points in the candidate rollback period that can represent the combustion system being in an irreversible state.

[0032] Specifically, the rollback state window for candidate rollback periods is determined based on the state platform characteristics of the changing segments. The rollback state window represents a small segment of the system where the deviation is most severe, or where the deviation has taken shape. For example, in the process of stabilization, rise, peak, fall, and stabilization, the period near the peak represents a period where the system is experiencing severe deviation. The rollback state window can be located based on the state platform characteristics of the changing segments. The period within the target length centered on the dominant output parameter is recorded as the rollback state window. Based on combustion control recording data, the sampled values ​​of multiple data sampling points within the rollback state window with respect to different control parameters are extracted, thus obtaining a rollback state sequence of multiple state points, representing the combustion state of the system at a specific moment.

[0033] The combustion background characteristics include multiple operational background-related information of the coal-fired boiler during the candidate rollback period, including but not limited to the unit load during this period (e.g., the rated load value reached), the coal mill combination status (e.g., which coal mills are operating), and the operating status of the soot blowing and denitrification units (e.g., whether they are in operation or shut down). Finally, by combining the combustion background characteristics, dominant output parameters, dominant control parameters, and multiple rollback state sequences of the candidate rollback period, a state rollback characteristic for the candidate rollback period is constructed to characterize the combustion state rollback characteristics of the coal-fired boiler under specific operating conditions and regulation modes.

[0034] Step S3: Group multiple candidate rollback periods into background groups based on state rollback characteristics to generate multiple rollback background sets, construct a rollback structure domain for each rollback background set, and optimize combustion control of the coal-fired boiler based on multiple rollback structure domains.

[0035] Specifically, based on the state regression features extracted from each candidate regression period, multiple candidate regression periods are grouped, and those with the same or similar operating backgrounds and regression features are assigned to the same regression background set. For each regression background set, a corresponding regression structure domain is constructed based on the state regression features of the different candidate regression periods it contains. This domain characterizes the range of combustion state changes under specific operating backgrounds and adjustment methods. During the operation of the coal-fired boiler, the real-time operating status of the boiler can be determined based on real-time operating data, and the regression background set matched under the real-time operating status can be identified. A comparative analysis is performed between the real-time operating status and the regression structure domain of the matched regression background set, providing a basis for optimizing and adjusting the combustion control strategy. This supports the coal-fired boiler in meeting low-NOx combustion requirements while also considering operating efficiency and energy consumption levels.

[0036] In one implementation, the fallback structure domain for each fallback context set includes: Based on combustion background characteristics, dominant output parameters, and dominant control parameters, multiple candidate rollback periods are grouped to generate multiple rollback background sets. For the background grouping process, for discrete features such as the operating status of auxiliary equipment, multiple discrete results can be directly determined, such as the status of the soot blowing device and denitrification device being running / not started, and which specific combustion output parameters and control parameters correspond to the dominant output parameters and dominant control parameters, etc. For continuous features such as unit load, a discrete interval division method can be used, for example, using 10% of the rated load as a load interval. Multiple candidate rollback periods are grouped in this way to generate multiple rollback background sets.

[0037] Based on this, considering the differences in the irreversibility of the dominant output parameters in some candidate rollback periods, the multiple candidate rollback periods in the rollback background set can be further divided into sample types. The rollback label of each group of state rollback features in the rollback background set can be determined according to the irreversibility of the combustion output parameters. For example, candidate rollback periods with the irreversibility of the dominant output parameters less than a preset reversible threshold are recorded as reversible rollbacks, and those otherwise are recorded as irreversible rollbacks. The preset reversible threshold can be reasonably set based on the actual operating requirements of the coal-fired boiler. When the irreversibility is less than the preset reversible threshold, it indicates that this degree of deviation from the load operating requirements exists.

[0038] Based on the above division, envelope structure samples and structure verification samples are determined for each backtracking background set using backtracking labels. The envelope structure samples include multiple irreversible backtracking samples, and the structure verification samples include multiple reversible backtracking samples. Based on the multiple backtracking state sequences in the envelope structure samples, statistical analysis is performed on the value range of each control parameter to determine the value range of each control parameter as the envelope structure range. Multiple initial structural domains of the backtracking background set are then constructed based on these multiple envelope structure ranges.

[0039] For multiple initial structural domains, based on candidate quantile values Determine the local range of each control parameter in the initial structural domain, specifically, with 1- Determine the percentile value of the control parameter within the envelope structure range, for example when When the value is 0.05, a local range is generated based on the 5th and 95th percentiles of the envelope structure range of the control parameter. The initial structural domain below can be used for Multiple values ​​are used to construct multiple initial structural domains. Structural verification is performed on each initial structural domain based on multiple backtracking state sequences in the structural verification samples. The false alarm index for each initial structural domain is calculated, including the percentage of samples belonging to the initial structural domain in multiple reversible backtracking samples. Specifically, for any reversible backtracking sample and initial structural domain, for multiple backtracking state sequences of the reversible backtracking sample, if the values ​​of multiple control parameters in the state sequence all fall within the range of the corresponding control parameters in the initial structural domain, the backtracking state sequence is considered to belong to the corresponding initial structural domain. If the percentage of backtracking state sequences belonging to the corresponding initial structural domain in the reversible backtracking sample is greater than a preset ratio, then the initial structural domain corresponding to the reversible backtracking sample data is considered to be the one that eliminates interference from some noise points.

[0040] After determining the false alarm index of each initial structural domain, the back-off structural domain of the back-off background set is determined from multiple initial structural domains based on the false alarm index. Specifically, the initial structural domain with the lowest false alarm index is taken as the back-off structural domain of the back-off background set.

[0041] In one implementation, combustion control optimization of a coal-fired boiler based on multiple backoff structural domains includes: Real-time operating data of the coal-fired boiler is acquired, and real-time state features are constructed based on this data. These real-time state features are then matched with combustion background features from multiple backtracking background sets to determine the target backtracking structural domain corresponding to the real-time operating data. In this process, the aforementioned method for analyzing and processing the operating record data of the coal-fired boiler is used to determine the combustion background information of the coal-fired boiler under the real-time operating data, as well as the corresponding dominant output parameters and dominant control parameters. Combined with the combustion background information, the operating background corresponding to the real-time operating data is matched with the combustion background features of multiple constructed backtracking background sets to determine the target backtracking structural domain corresponding to the current operating state of the coal-fired boiler.

[0042] By monitoring the initial structural domain in real time, potential change segments can be identified. Once a change segment is detected within a certain time period for the coal-fired boiler, multiple real-time state sequences are extracted based on real-time operating data. Irreversible structural proximity detection is then performed on these real-time state sequences based on the target fallback structural domain. During this process, proximity state characterization is performed based on the ingress state of the real-time state sequences. Specifically, the reference range of the target fallback structural domain with respect to multiple control parameters is determined, and the out-of-bounds state of each control parameter in the real-time state sequence with respect to the target fallback structural domain is identified. When a control parameter falls within the corresponding reference range in the target fallback structural domain, it is recorded as an out-of-bounds parameter. The total percentage of out-of-bounds parameters in the real-time state sequence is statistically analyzed as an operational risk parameter, characterizing the proximity of the current operating state of the coal-fired boiler to the irreversible structure. Finally, a real-time control optimization strategy for the coal-fired boiler is generated based on the real-time operating data and the target fallback structural domain.

[0043] For example, when the operational risk parameter exceeds a preset risk threshold, an operational warning is issued, and multiple control parameters belonging to the out-of-bounds state are identified. Based on the reference range of these parameters in the target backoff structural domain, the adjustment range of these multiple out-of-bounds control parameters is determined, providing a basis for the optimized combustion control of coal-fired boilers. This invention starts from historical operational data of coal-fired boilers, systematically analyzes the backoff behavior that may occur during combustion regulation, and identifies irreversible characteristic structures that easily lead to an unrecoverable combustion state under specific operational backgrounds and control conditions. In actual operation of coal-fired boilers, once some irreversible combustion states are formed, they are likely to cause an increase in pollutant emission levels or a deterioration in energy consumption indicators, thus deviating from the operational goals of low-NOx combustion and energy saving. This invention, by characterizing the evolutionary characteristics of the combustion process from the perspective of multi-parameter combination states, achieves the identification and constraint of the aforementioned irreversible characteristic structures. This approach helps to avoid the formation of unfavorable states in advance during combustion regulation, providing data support and decision-making basis for the optimization and adjustment of combustion control strategies, thereby improving the safety and stability of coal-fired boiler operation.

[0044] Please see Figure 2 The diagram illustrates a structural schematic of a combustion control optimization system for a coal-fired boiler according to one embodiment of the present invention, comprising: The rollback identification module is used to collect the operation record data of the coal-fired boiler, including combustion control record data and combustion output record data, to determine multiple control parameters for the operation control of the coal-fired boiler, and to identify multiple candidate rollback periods from the combustion control record data; The rollback feature extraction module is used to identify the dominant parameters for each candidate rollback period based on combustion output recording data, determine the dominant control parameters for each candidate rollback period, and construct the state rollback features for each candidate rollback period by combining combustion control recording data and combustion output recording data. The combustion control optimization module is used to group multiple candidate back-off periods into backgrounds based on state back-off characteristics, generate multiple back-off background sets, construct a back-off structure domain for each back-off background set, and optimize the combustion control of the coal-fired boiler based on multiple back-off structure domains.

[0045] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A combustion control optimization method for a coal-fired boiler, characterized by, The method comprises the following steps: Collecting operation record data of a coal-fired boiler, the operation record data comprising combustion control record data and combustion output record data, determining a plurality of control parameters related to operation control of the coal-fired boiler, and identifying a plurality of candidate rollback periods from the combustion control record data; Identifying a dominant parameter for each candidate rollback period based on the combustion output record data, determining a dominant control parameter for each candidate rollback period, and constructing a state rollback feature of each candidate rollback period in combination with the combustion control record data and the combustion output record data; Grouping the plurality of candidate rollback periods according to the state rollback feature, generating a plurality of rollback background sets, constructing a rollback domain of each rollback background set, and optimizing combustion control of the coal-fired boiler based on the plurality of rollback domains.

2. A combustion control optimization method for a coal-fired boiler according to claim 1, characterized in that, The method of identifying a plurality of candidate rollback periods from the combustion control record data comprises the following steps: Constructing a plurality of control parameter sequences of the combustion control record data, extracting a plurality of sequence segment sliding difference features corresponding to each control parameter sequence, and identifying a state label of each sequence segment according to the sliding difference features, wherein the state label comprises a stable state and a change state; Traversing each control parameter sequence based on the state label of the sequence segment, and extracting a plurality of local rollback segments in each control parameter sequence based on a preset rollback reference structure, wherein the local rollback segment is composed of a stable segment, a change segment and a rollback segment; Fusing the local rollback segments of the plurality of control parameter sequences by period overlap, and generating a plurality of candidate rollback periods corresponding to the combustion control record data.

3. A combustion control optimization method for a coal-fired boiler according to claim 2, characterized in that, The method of identifying a dominant parameter for each candidate rollback period based on the combustion output record data comprises the following steps: Extracting, according to the combustion output record data, a state platform feature corresponding to each combustion output parameter in a stable segment, a change segment and a rollback segment in each candidate rollback period, determining a change deviation value and a regression deviation value of each combustion output parameter according to a plurality of state platform features, and calculating an irreversibility of the combustion output parameter based on the change deviation value and the regression deviation value; Determining a dominant output parameter of the candidate rollback period according to the irreversibility, constructing a combustion output sequence of the dominant output parameter, detecting a control response parameter of each control parameter based on the combustion output sequence, and determining a dominant control parameter of the candidate rollback period according to a plurality of control response parameters.

4. A combustion control optimization method for a coal-fired boiler according to claim 3, characterized in that, The method of constructing a state rollback feature of each candidate rollback period in combination with the combustion control record data and the combustion output record data comprises the following steps: Determining a rollback state window of the candidate rollback period according to the state platform feature of the change segment, extracting a plurality of rollback state sequences of the rollback state window according to the combustion control record data, extracting a combustion background feature of the candidate rollback period from the combustion control record data, and generating the state rollback feature of the candidate rollback period in combination with the dominant output parameter and the dominant control parameter of the candidate rollback period.

5. A combustion control optimization method for a coal-fired boiler according to claim 4, characterized in that, The method of constructing a rollback domain of each rollback background set comprises the following steps: Grouping the plurality of candidate rollback periods based on the combustion background feature, the dominant output parameter and the dominant control parameter to generate a plurality of rollback background sets. The backtracking label of each group of state backtracking features in the backtracking background set is determined according to the irreversibility of the combustion output parameter, and the backtracking label includes reversible backtracking and irreversible backtracking; The envelope structure sample and the structure verification sample of each backtracking background set are determined based on the backtracking label, the envelope structure range of each control parameter is determined according to a plurality of backtracking state sequences in the envelope structure sample, and a plurality of initial structure domains of the backtracking background set are constructed based on a plurality of envelope structure ranges; The structure verification of each initial structure domain is performed based on a plurality of backtracking state sequences in the structure verification sample, the false positive index of each initial structure domain is calculated, and the backtracking structure domain of the backtracking background set is determined from the plurality of initial structure domains based on the false positive index.

6. A combustion control optimization method for a coal-fired boiler according to claim 5, characterized in that, The combustion control optimization of the coal-fired boiler based on the plurality of backtracking structure domains includes: Real-time operation data of the coal-fired boiler is acquired, real-time state features are constructed according to the real-time operation data, the real-time state features are matched with the combustion background features of the plurality of backtracking background sets, and a target backtracking structure domain corresponding to the real-time operation data is determined; A real-time state sequence of the real-time operation data is extracted, irreversible structure proximity detection is performed on the real-time state sequence based on the target backtracking structure domain, an operation risk parameter of the real-time operation data is generated, and a real-time control optimization strategy of the coal-fired boiler is generated based on the real-time operation data and the target backtracking structure domain.

7. A combustion control optimization system for a coal-fired boiler, characterized by, The system is used to implement the combustion control optimization method of the coal-fired boiler according to any one of claims 1-6, and includes: A backtracking identification module is configured to collect operation record data of the coal-fired boiler, the operation record data including combustion control record data and combustion output record data, determine a plurality of control parameters related to operation control of the coal-fired boiler, and identify a plurality of candidate backtracking time periods from the combustion control record data; A backtracking feature extraction module is configured to identify dominant parameters for each candidate backtracking time period based on the combustion output record data, determine dominant control parameters of each candidate backtracking time period, and construct state backtracking features of each candidate backtracking time period in combination with the combustion control record data and the combustion output record data; A combustion control optimization module is configured to group the plurality of candidate backtracking time periods according to the state backtracking features, generate a plurality of backtracking background sets, construct backtracking structure domains of each backtracking background set, and perform combustion control optimization of the coal-fired boiler based on the plurality of backtracking structure domains.

Images