Intelligent control method and system for boiler denitration

Abstract

The invention relates to the technical field of intelligent control, and comprises an intelligent control method and system for boiler denitration, and the method comprises the following steps: obtaining a NOx concentration calculation difference value of a current period and a previous period, generating a jump concentration judgment result, screening a response region according to the judgment result in combination with a disturbance ratio, calculating a response time delay, and generating time delay data, fitting lagging and residual errors to generate a correlation coefficient, extracting a forward relation microcell, calculating and correcting ammonia injection parameters, and generating a regulation and control record. According to the method, through ratio analysis of concentration jump and disturbance deviation, local anomaly recognition precision is enhanced, response area recognition stability is optimized in combination with point position set statistics, the reaction dynamic state is defined through time sequence tracking of ammonia spraying starting and NOx response, the time efficiency control force is improved and adjusted, and an error correlation path is established through linear fitting of residual errors and lags; and analyzing a transmission distortion source, and implementing proportion and concentration joint correction in combination with a hysteresis relationship, so that accurate dosing of ammonia water is realized, and the efficiency and dynamic stability of boiler flue gas denitration are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control, and particularly to a method and system for intelligent control of boiler denitration. Background Art

[0002] The technical field of intelligent control includes the perception and analysis of the operating state of industrial systems to achieve dynamic adjustment and control of various production processes. The core content is to use computer technology, control theory, and sensing and detection means to collect information, make decision-making judgments, and execute control on the target system, covering multiple directions such as industrial automation, process control, and energy management, and having the characteristics of systematicness, multi-variable coupling, and dynamics. Especially in the operation control of energy equipment, intelligent control pays special attention to the adaptive adjustment of complex working conditions and the optimization control of operation efficiency, and is often applied to fields such as power plant boilers, combustion devices, and chemical reaction processes.

[0003] Among them, the intelligent control method for boiler denitration refers to a process control method that based on the real-time monitoring of the nitrogen oxide emission characteristics during the boiler combustion process, and uses set rule parameters to control the operation of the denitration system, covering the real-time collection of the boiler combustion status and flue gas components, and making logical judgments and proportional adjustments to the ammonia injection amount and ammonia injection position through a set program flow to regulate the selective catalytic reduction reaction process. Specifically, it takes the nitrogen oxide concentration data in the flue gas and the furnace load situation as the judgment basis, and uses control parameters to adjust the actions of the system actuators to complete the adjustment and control of the denitration agent dosing amount.

[0004] In the existing boiler denitration process, it generally relies on a single-variable control path based on concentration or load, lacking the ability to identify the dynamic linkage of emission behaviors. During the point identification process, it mainly relies on fixed threshold rules, resulting in weak local anomaly identification ability and easy misjudgment of the response area under the background of complex disturbances. The lag of the ammonia injection reaction mainly relies on empirical parameter estimation, which fails to reflect the time matching between the ammonia injection action and the actual emission reaction, causing the accumulation of control delays and affecting the effectiveness of real-time adjustment. The residual control is biased towards static error correction, lacking a structural analysis and traceability mechanism for the error source, and unable to identify the adjustment imbalance problem caused by uneven system responses. The ammonia injection adjustment generally adopts a single proportional control method, ignoring the coupling relationship between dynamic response and local residuals, resulting in local over-dosing or under-dosing under multi-variable changes, restricting the efficient operation of the denitration process. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art, and to propose a method for intelligent control of boiler denitration.

[0006] In order to achieve the above purpose, the present invention adopts the following technical solution: A method for intelligent control of boiler denitration, including the following steps: S1: Obtain the current NOx concentration value at the NOx detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling period, calculate the concentration difference, compare it with the jump recognition threshold, and generate a jump concentration determination result; S2: According to the jump concentration determination result, calculate the disturbance deviation by combining the air flow velocity value and the standard flow rate reference value, judge whether the deviation value exceeds the reference, synchronously calculate the ratio of the concentration difference to the disturbance deviation, screen the area where the ratio exceeds the limit, and generate a spray ammonia response area recognition result; S3: Based on the spray ammonia response area recognition result, extract the spray ammonia start time and the time point when the NOx concentration first drops, calculate the response delay value, and generate NOx response delay data; S4: According to the NOx response delay data, count the difference between the response delay value and the theoretical response time as the lag value, collect the real-time NOx residual value at the corresponding point, perform a linear fit of the lag value and the real-time residual value, and judge the co-directional change relationship to generate residual lag correlation data; S5: Determine the residual lag correlation data, extract the spray ammonia amount and the liquid supply concentration value in the positive relationship micro-region, calculate the spray ammonia amount adjustment ratio and the concentration adjustment ratio in combination with the real-time NOx residual value, and perform a correction action to generate a spray ammonia regulation adjustment record.

[0007] As a further solution of the present invention, the jump concentration determination result includes a jump point position identifier, a jump amplitude level, and a jump trend direction. The spray ammonia response area recognition result includes a high response area position, a response intensity level, and a response influence range. The NOx response delay data includes a response time difference sequence, a response start time set, and a response end time set. The residual lag correlation data includes a linear fit slope value, a change direction consistency index, and a lag and residual fitting degree. The spray ammonia regulation adjustment record includes a spray ammonia adjustment ratio, a liquid supply concentration adjustment ratio, and an adjustment applicable area.

[0008] As a further solution of the present invention, the steps for obtaining the jump concentration determination result include: S111: Based on the current NOx concentration value at the NOx flue gas detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling period, calculate the difference between the two, and use the absolute difference between the current NOx concentration value and the NOx concentration value in the previous period for the difference calculation to obtain the concentration change difference; S112: According to the concentration change difference, compare it item by item with the jump recognition threshold, identify the concentration change difference items greater than the jump recognition threshold, and use the formula: ; Calculate the multi-period composite concentration fluctuation value , and obtain the jump concentration judgment index value, where Represents the NOx concentration value of the current cycle, Represents the NOx concentration value of the previous cycle, Represents the average value of the jump recognition threshold; S113: According to the magnitude relationship between the jump concentration judgment index value and the jump recognition threshold, mark the time points and corresponding detection point serial numbers where all jump concentration judgment index values are greater than the jump recognition threshold, and generate a jump concentration determination result.

[0009] As a further solution of the present invention, the steps for obtaining the ammonia injection response area recognition result include: S211: Based on the jump concentration determination result, combine the air flow velocity value of each detection point and the corresponding standard flow velocity reference value, calculate the difference between the two, take the absolute value to obtain the flow velocity disturbance amount value of each detection point, and calculate the disturbance degree through the ratio of the disturbance amount value to the reference value to obtain the disturbance deviation ratio; S212: According to the disturbance deviation ratio, and combine with the concentration change difference of the corresponding detection point for joint processing, using the formula: ; Calculate the concentration disturbance composite ratio , where, Represents the NOx concentration value at the i-th moment, Represents the NOx concentration value at the previous moment, Represents the air flow velocity value at the i-th point, Represents the standard flow velocity reference value at the i-th point; S213: According to the concentration disturbance composite ratio, compare with the preset concentration disturbance recognition threshold, screen all spatial regions where the ratio exceeds the threshold, and record the corresponding detection point position information and time information to establish an ammonia injection response area recognition result.

[0010] As a further solution of the present invention, the steps for obtaining the NOx response delay data include: S311: Based on the point set in the ammonia injection response area recognition result, collect the continuous time series values of the NOx concentration corresponding to each point after the ammonia injection action, judge the numerical fluctuation direction between each time series point according to the change trend of the NOx concentration value in each point collection sequence, identify the moment when it first changes from a non-decreasing state to a decreasing state, and obtain the first NOx concentration decrease time point; S312: According to the first NOx concentration decrease time point, combine with the trigger time of the ammonia injection action at each point, calculate the time difference between the two respectively, record the time interval data corresponding to each point, and obtain the point response delay value; S313: According to the point response time delay value, arrange the response times of each point after the ammonia injection action trigger in the order of point number and time sequence to form a cross-point response time set, and establish NOx response time delay data.

[0011] As a further solution of the present invention, the steps for obtaining residual lag correlation data include: S411: According to the response time delay values of each point in the NOx response time delay data, combined with the current liquid supply condition, calculate the theoretical response time of each point under the current liquid supply condition, and subtract the response time delay value of each point from the theoretical response time to obtain the response lag value of each point; S412: Combining the response lag values of each point, collect the difference between the NOx real-time concentration and the reference concentration before ammonia injection corresponding to each point after the end of the current cycle, construct a set of NOx real-time residual values corresponding to each point, and use the formula: ; Calculate the residual lag slope value , where represents the response lag value of the i-th point, represents the NOx real-time residual value of the i-th point, represents the average value of the response lag values of all points, represents the average value of the NOx real-time residual values of all points, represents the number of points; S413: According to the residual lag slope value, sort out the point slope value and the co-directionality judgment result, and output the correlation intensity between lag and residual of each point in time sequence to establish residual lag correlation data.

[0012] As a further solution of the present invention, the steps for obtaining the ammonia injection regulation adjustment record include: S511: According to the residual lag correlation data, screen the point area with a positive slope value, identify the micro-area with a positive residual lag relationship in the current cycle, extract the ammonia injection amount data and the liquid supply concentration value corresponding to the points in the micro-area, and obtain the ammonia injection and liquid supply data of the positive micro-area; S512: Based on the ammonia injection and liquid supply data of the positive micro-area, combined with the NOx real-time residual value of the corresponding point, calculate the deviation degree of the current liquid supply parameter on the residual response, calculate the ammonia injection amount adjustment ratio and the liquid supply concentration adjustment ratio, and generate ammonia injection regulation deviation ratio data; S513: According to the ammonia injection regulation deviation ratio data, perform the correction operation on the ammonia injection amount and the liquid supply concentration of the corresponding point, record the parameter change value and the correction direction before and after the adjustment, and establish the ammonia injection regulation adjustment record.

[0013] A boiler denitration intelligent control system includes: The jump recognition module obtains the NOx concentration values at the detection point in the current and previous cycles, calculates the difference and compares it with the jump recognition threshold, screens the jump points, and generates the concentration jump determination result; The response recognition module obtains the air flow velocity value and the flow velocity reference value according to the concentration jump determination result, calculates the disturbance deviation ratio, determines the response points, and generates the ammonia injection response area recognition result; The time delay determination module extracts the ammonia injection start time and the NOx concentration decrease time according to the ammonia injection response area recognition result, calculates and determines the response time delay value, and generates the NOx response time delay data; The residual correlation module obtains the response lag value and the NOx residual value based on the NOx response time delay data, performs linear fitting, extracts the slope to judge the co-directional change relationship, and generates the residual lag correlation data; The regulation and adjustment module extracts the ammonia injection amount and the liquid supply concentration according to the residual lag correlation data, calculates the adjustment data in combination with the residual value and executes the correction action, and generates the ammonia injection regulation and adjustment record.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, through the real-time recognition of the change trend of the flue gas NOx concentration, combined with the response ratio relationship between the concentration mutation amplitude and the air flow disturbance intensity, the high-sensitivity recognition of the emission state is realized, the local response ability under the non-steady state condition is enhanced, and the statistical mechanism of the continuous point set is introduced in the comparison of the concentration change and the disturbance difference ratio, so that the recognition of the response area is more group-based and structured, avoiding the interference caused by the misjudgment of isolated points. By tracking the delay relationship between the ammonia injection start and the NOx concentration decrease in time series, a stable reaction dynamic evaluation system is formed, providing clear time limit support for the subsequent adjustment actions. The linear fitting of the residual and the theoretical lag further reveals the potential lag source of the control deviation, and the change of the slope and directionality maps the transmission distortion in the control chain, optimizing the evaluation method of the error correlation structure. After determining the influence area of the NOx concentration residual, the linkage mechanism of proportional adjustment and concentration correction is introduced to realize real-time differential regulation based on the response lag, effectively controlling the redundancy of resource input in the non-target area, improving the overall reaction efficiency, forming high adaptability and strong coupling response ability, and enhancing the stability and energy saving of the denitration process. Description of the Drawings

[0015] Figure 1 is the main step flow chart of the present invention; Figure 2 is the flow chart for obtaining the jump concentration determination result of the present invention; Figure 3 is the flow chart for obtaining the ammonia injection response area recognition result of the present invention; Figure 4 is the flow chart for obtaining the NOx response time delay data of the present invention; Figure 5 This is the flowchart for obtaining residual lag correlation data of the present invention; Figure 6 This is the flowchart for obtaining the record of ammonia injection regulation adjustment of the present invention. Specific embodiments

[0016] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0017] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. In addition, in the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0018] Please refer to Figure 1 , a method for intelligent control of boiler denitration, comprising the following steps: S1: Obtain the current NOx concentration value at the NOx flue gas detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling period, calculate the concentration difference, and compare it one by one with the jump identification threshold. When the concentration difference is greater than the jump identification threshold, mark it as a concentration jump point and generate a jump concentration determination result; S2: According to the jump concentration determination result, record the current air velocity value at the position of the concentration jump point, combine it with the standard flow velocity reference value under the current boiler load state, calculate the absolute difference between the air velocity value and the reference value to obtain the disturbance deviation value, calculate the ratio between the concentration difference and the disturbance deviation value, compare it with the ammonia injection response ratio threshold, and count the set of points higher than the ammonia injection response ratio threshold to generate an ammonia injection response area identification result; S3: Based on the set of points in the ammonia injection response area identification result, collect the continuous time series values of the NOx concentration corresponding to the points after the ammonia injection action, extract the ammonia injection start time and the time point when the NOx concentration first decreases, calculate the time difference between the two and define it as the response delay value, and generate NOx response delay data; S4: Based on the response delay values of each point in the NOx response delay data, obtain the theoretical response time of the current liquid supply reaction. Statistically calculate the difference between each response delay value and the theoretical response time as the lag value. Collect the real-time NOx residual values at the corresponding points after the end of the current cycle, perform a linear fitting judgment on the lag value and the real-time residual value, extract the slope value and judge the co-directional change relationship, and generate residual lag correlation data; S5: Determine the micro-regions with a positive relationship in the residual lag correlation data, extract the ammonia injection amount and liquid supply concentration values within the current cycle, calculate the ammonia injection amount adjustment ratio and concentration adjustment ratio in combination with the real-time NOx residual value, and perform a correction action to generate an ammonia injection regulation adjustment record.

[0019] The jump concentration determination result includes the jump point position identifier, jump amplitude level, and jump trend direction. The ammonia injection response area recognition result includes the high response area position, response intensity level, and response influence range. The NOx response delay data includes the response time difference sequence, response start time set, and response end time set. The residual lag correlation data includes the linear fitting slope value, change direction consistency index, and lag and residual fitting degree. The ammonia injection regulation adjustment record includes the ammonia injection adjustment ratio, liquid supply concentration adjustment ratio, and adjustment applicable area.

[0020] Please refer to Figure 2 , step S1 is: S111: Based on the current NOx concentration value at the NOx flue gas detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling cycle, calculate the difference between the two. Use the absolute difference between the current NOx concentration value and the NOx concentration value in the previous cycle for the difference calculation to obtain the concentration change difference; Based on the current NOx concentration value at the NOx flue gas detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling cycle, obtain the corresponding values of each detection point in the two cycles, and perform a difference calculation operation on each pair of data separately, taking its absolute value as the concentration change difference in this sampling cycle. Specifically, when executing, taking the detection point numbered A1 as an example, the current cycle concentration value is 165 mg / m³, and the previous cycle concentration value is 160 mg / m³, then its difference calculation is , similarly, the current value of the A2 detection point is 172 mg / m³, and the previous cycle is 170 mg / m³, and the calculated difference is , perform this operation to complete the difference acquisition of all monitoring points, and obtain a set of concentration change difference data. The results are shown in the following table: Table 1 NOx Concentration Change Difference Table of Monitoring Points

[0021] As shown in Table 1, after comparing the current NOx concentration value at each detection point with that of the previous cycle, a clear change difference is formed. This difference is the core indicator in the subsequent jump recognition operation to obtain the concentration change difference.

[0022] S112: According to the concentration change difference, compare it item by item with the jump recognition threshold to identify the concentration change difference items greater than the jump recognition threshold, using the formula: ; Calculate the multi-cycle composite concentration fluctuation value , and obtain the jump concentration judgment index value, where represents the current cycle NOx concentration value, represents the NOx concentration value of the previous cycle, represents the average value of the jump recognition threshold; According to the concentration change difference, perform joint analysis and processing on the data of each point with the set jump recognition threshold respectively. During the judgment process, enhance the recognition of concentration mutations by adding a square root of the squared difference. Taking point A1 as an example, the current value is 165 and the value of the previous cycle is 160. After calculation, we get , , and the total is . For point A3, the current value is 169 and the previous cycle is 165. The calculation result is . Apply the above operations to all detection points to obtain the following results: Table 2 Calculation Table of Jump Concentration Judgment Index

[0023] As shown in Table 2, for each point judgment value, when the result is greater than 3.5 mg / m³, it is determined as a jump item. The threshold setting of 3.5 mg / m³ is determined based on the NOx emission fluctuation range in different load stages under the actual boiler operation conditions. Specifically, it refers to the average difference fluctuation range of NOx concentrations in two adjacent time periods within the load change cycle. This range usually concentrates between 2.8 and 3.2 mg / m³. To improve the sensitivity of identification and exclude the interference of minor disturbances in some stable operation periods, the system adjusts the threshold upward to 3.5 mg / m³ to ensure that it can respond to transient fluctuations of 5 mg / m³ and above. In actual operation, this threshold will be dynamically fine-tuned with the change of boiler load. When the load increases, the NOx emission amplitude increases, and the threshold will be slightly increased with the increase of the main steam flow rate, and vice versa. The final selected value is set at the middle and upper positions of the monitored average difference under typical high-load conditions, which is representative and robust. During the judgment process, all parameters are directly obtained through data collection and calculation without manual setting. The benefit of the formula is that it strengthens the fluctuation sensitivity at the numerical level by superimposing the difference and the square root of the square difference, thereby providing a clear judgment boundary for the behavior of exceeding the threshold and obtaining the jump concentration judgment index value.

[0024] S113: According to the magnitude relationship between the jump concentration judgment index value and the jump identification threshold, mark all time points and corresponding detection point serial numbers where the jump concentration judgment index value is greater than the jump identification threshold to generate the jump concentration determination result; Make a step-by-step judgment based on the jump concentration judgment index value and the set jump identification threshold of 3.5 mg / m³. For the monitoring points with judgment values higher than this threshold, perform jump marking identification. When performing the judgment operation, directly record the sampling period moment and the point number for the points with comparison values greater than the threshold. The judgment values of A1, A3, and A4 are 6.5, 4.5, and 6.5 respectively, all of which are greater than the threshold, so they are identified as jump points. The value of A2 is 0.5, which is lower than the threshold and is not marked. In the result record, it is necessary to organize the time series and the point number to form the judgment result item, and finally output the structured jump concentration result set to generate the jump concentration determination result.

[0025] Please refer to Figure 3 , the steps of S2 are as follows: S211: Based on the jump concentration determination result, combine the air flow velocity value of each detection point with the corresponding standard flow velocity reference value, calculate the difference between the two, take the absolute value to obtain the flow velocity disturbance amount value of each detection point, and calculate the disturbance degree through the ratio of the disturbance amount value to the reference value to obtain the disturbance deviation ratio; Based on the determination result of the jump concentration, it is necessary to clarify the difference between the air flow velocity associated with each monitoring point and the standard flow velocity reference value. In practice, the air flow velocity can be collected through the wind speed measurement points arranged in the boiler furnace. For example, if the air flow velocity collected by a certain measurement point within the sampling period is 12.4 m / s and the standard flow velocity reference value at this point under steady-state operation is 10.0 m / s, then the flow velocity disturbance value of this measurement point is 12.4 - 10.0 = 2.4 m / s. To obtain the degree of disturbance, it is necessary to construct a disturbance deviation ratio, which can be obtained by comparing the disturbance quantity with the standard flow velocity reference value, that is: 2.4 / 10.0 = 0.24. If the disturbance deviation reference value set by the system is 0.2, then this ratio is judged to be in an over-limit state. The specific judgment criteria are divided according to the following intervals: the disturbance deviation ratio less than or equal to 0.1 is regarded as the stable interval, 0.1 to 0.2 is the critical fluctuation interval, and exceeding 0.2 is the abnormal disturbance interval. This type of interval needs to be included in the subsequent joint identification. If the jump concentration determination result of the collection point is marked as a jump point at the same time point, the concentration change difference is the current concentration minus the concentration at the previous moment. Assuming that its current NOx concentration is 105 mg / m³ and the previous cycle is 95 mg / m³, then the difference is 10 mg / m³, which can be directly incorporated into the subsequent calculation to form a cross-data pair. This data pair is the corresponding result pair of the disturbance deviation and the concentration change, and can form double-parameter input items corresponding to multiple detection points, and establish a disturbance deviation ratio.

[0026] S212: According to the disturbance deviation ratio, and combined with the concentration change difference of the corresponding detection point for joint processing, using the formula: ; Calculate the concentration disturbance composite ratio , where represents the NOx concentration value at the i-th moment, represents the NOx concentration value at the previous moment, represents the air flow velocity value at the i-th point, represents the standard flow velocity reference value at the i-th point; According to the co-variation trend between the disturbance deviation ratio and the concentration change difference, it is necessary to establish a coupling relationship measurement index for cross-parameter values. Here, the method of constructing a concentration disturbance composite ratio is adopted. During the implementation process, for each detection point, it is necessary to obtain the absolute difference between the NOx concentration value at the current moment and the NOx concentration value at the previous moment, and at the same time, obtain the current air flow velocity value and the standard flow velocity reference value at the corresponding point, and then perform actual operations based on the formula. Assuming that the corresponding parameters of a certain monitoring point are: , , , , then the substitution into the calculation process is as follows: Calculate the numerator part: ; Calculate the denominator part: ; Calculate the composite ratio: ; The result shows that the composite ratio of the concentration perturbation at this detection point in the current cycle is 0.117. Its result can be further summarized to determine whether it is within the scope of the linked ammonia injection response. The operating data of the monitoring points are listed as follows: Table 3 Operating Parameter Table of Detection Points in the Ammonia Injection Area

[0027] Table 3 lists the experimental data. All the above data are obtained from actual test acquisitions. Based on this data, a composite ratio group of multiple detection points can be further formed as the input basis for ratio overlimit judgment.

[0028] S213: According to the composite ratio of the concentration perturbation, compare with the preset concentration perturbation identification threshold, screen all spatial regions where the ratio exceeds the threshold, and record the corresponding detection point position information and time information to establish the identification result of the ammonia injection response area; According to the composite ratio of the concentration perturbation, perform positioning and screening operations in the time dimension and the spatial detection point dimension. In specific implementation, the composite ratio judgment results of each detection point can be called, and the identification threshold is set to 0.10. Determine whether it is greater than this threshold. If the condition is met, extract the corresponding point number and the information of the sampling time point to form a data marking structure. This structure can be used for response area mapping. For example, for detection point A1 in Table 1, because its composite ratio 0.117 > 0.10, it is marked as a point in the response area. At the same time, record its sampling period as "t4" and the spatial position label as "A1". In this way, the data screening operation of the ammonia injection interval is completed and the identification result of the ammonia injection response area is established. The setting of the above identification threshold is based on the coordinated fluctuation between the air flow perturbation and the sudden change of NOx concentration. By extracting the composite ratio value ranges of the stable section and the perturbation section under multiple representative operating cycles in the actual working condition respectively, and calculating the cross-fluctuation boundary interval, finally, the average value at the intersection of the two value ranges is selected as the standard line of the identification threshold. Specifically, the minimum composite ratio in the perturbation interval is 0.102, the maximum in the stable interval is 0.098, and the critical value is set to 0.10 at their intersection. This threshold fluctuates with the changes of the boiler load level, air flow density and sampling frequency within the detection period, and fine-tuning intervals can be set separately under different load levels to cover the numerical variation ranges under different operating states, so that the identification boundary always remains in a state interval with high distinguishability.

[0029] Please refer to Figure 4 , the steps of S3 are: S311: Based on the point set in the ammonia injection response area identification result, collect the continuous time series values ​​of NOx concentration corresponding to each point after the ammonia injection action, judge the value fluctuation direction between each time series point according to the change trend of the NOx concentration value in the collection sequence of each point, identify the moment when the state changes from the non-decreasing state to the decreasing state for the first time, and obtain the time point when the NOx concentration first decreases; Based on the point set in the ammonia injection response area identification result, each identified point is marked and associated with the concentration value time series channel in the monitoring database according to the number. The NOx concentration data within 30 minutes after the ammonia injection action is continuously collected from each channel. The sampling frequency is set to once per minute, that is, 30 groups of NOx concentration time series values ​​are obtained for each point. For example, the sequence collected at point A001 within 30 minutes after the ammonia injection is {129, 127, 128, 124, 122, 121...}. This sequence is used as the basis for subsequent analysis of this point. Next, the actual start time of the ammonia injection action at each point needs to be determined. The start time data comes from the control system execution signal record. For example, the start time of the ammonia injection action at point A001 is 13:02. This information is accurately recorded in seconds. At the same time, the NOx concentration time series value is timestamped. The first drop is judged by matching the rows and further performing the "first drop" judgment in the NOx concentration sequence corresponding to each point. During the judgment process, the difference between the first group of data after the ammonia injection action is started and the previous item is compared backward one by one. If the current value shows a downward trend of less than or equal to negative 2ppm compared with the previous value, it is judged to be the first drop. For example, the value at 13:03 in the A001 point sequence is 127, and the value at 13:04 is 124, then the first drop time is considered to be 13:04. The judgment rule sets the drop amplitude threshold to 2ppm, which is the system default setting value. According to the historical equipment response range setting, the reasonable range is 1.5ppm to 3ppm. In this embodiment, the value is uniformly taken as 2ppm. The first NOx concentration drop time point of all ammonia injection response points can be obtained through this rule. Some sample data are shown in the following table: Table 4 Example of concentration change time series at ammonia injection response point

[0030] As shown in Table 4, the first drop time of point A001 occurs in the second minute after the ammonia injection. Compared with other points, a basis for subsequent analysis can be established.

[0031] S312: Calculate the time difference between the first drop time of NOx concentration and the trigger time of ammonia injection at each point, record the time interval data corresponding to each point, and obtain the point response delay value; According to the first NOx concentration drop time point obtained above, combined with the ammonia injection action trigger time corresponding to the point, calculate the time interval between the two point by point as the basic data of the point response delay. For example, for point A001, the ammonia injection starts at 13:02 and the concentration drops for the first time at 13:04, then the response delay is 2 minutes. During this process, an "absolute time difference" operation needs to be performed on each pair of time values, and the unit is converted to seconds. The execution method is to take the difference between the two timestamps and convert it to the form of seconds. The time difference is the point response time delay. For example, for A001, the time difference is 13:04 - 13:02 = 2 minutes = 120 seconds. Further introduce the NOx concentration change intensity factor, and superimpose the 1 / 2 average weight term of the total amplitude of the concentration change on the response time difference. The concentration change value is obtained by taking the absolute value of the difference between consecutive time series values and then summing them up. For example, the 30-minute sampling sequence of point A001 is {129, 127, 128, 124...}, then the difference sequence is {2, 1, 4,...}. Assuming the total difference obtained after summing is 48 ppm, take half of it as 24 and superimpose it on the time difference to get the corrected response time delay of this point as 120 + 24 = 144 seconds. Then introduce the reference fluctuation amounts M and N for average square root correction. Let the average concentration difference of point A001 in the 3 minutes before ammonia injection be M = 5 ppm, and in the 3 minutes after ammonia injection be N = 12 ppm, then the correction term is √(5² + 12²) = 13.0 ppm. The final response time delay value is 144 ÷ 13.0 ≈ 11.08 seconds. This value is used to represent the actual effective delay time of ammonia injection response. For the purpose of unified management, retain 2 decimal places to form a unified data format for output, recorded as 11.08. After processing all points in this way, a set of point response time delay values can be formed.

[0032] S313: According to the point response time delay values, arrange the response times of each point after the ammonia injection action trigger in the order of point number and time sequence to form a cross-point response time set, and establish NOx response time delay data; According to the point response delay values obtained in the previous paragraph, sort each point in ascending order of the number, and at the same time perform sorting in combination with the response time values to establish a dual index, so as to form a traceable response process time series. Then record the response duration required for each point after ammonia injection in a unified table structure. For example, the response delays of points A001 and A002 are 11.08 seconds and 12.75 seconds respectively. This type of data is recorded in the system database fields to facilitate the unified analysis of the response behavior trends of each point. At the same time, all point response time data can be classified and marked into three intervals: within 30 seconds, 30 to 60 seconds, and more than 60 seconds. For example, it is set that within 30 seconds is "fast response", 30 to 60 seconds is "medium response", and more than 60 seconds is "slow response". Record it in the unified output table in combination with the field attributes to enable the system to trace and quickly locate high-delay points. For example, the response delay of point A001 above is 11.08 seconds, which is classified into the fast response category, and the record field in the subsequent output data structure is "fast response".

[0033] Please refer to Figure 5 , and the S4 step is as follows: S411: According to the response delay values of each point in the NOx response delay data, in combination with the current liquid supply condition, calculate the theoretical response time of each point under the current liquid supply condition, and subtract the response delay value of each point from the theoretical response time to obtain the response lag value of each point; Based on the response delay values of each point in the NOx response delay data, it is necessary to further refine the acquisition of the response delay values. First, for the set of points indicated by the data source, clarify the corresponding number of points and their spatial positions. For example, a total of 9 points are sampled from the upper, middle, and lower layers of the boiler furnace, and their NOx response delay values are recorded respectively in different working condition cycles. This value is obtained from the time difference between the ammonia injection trigger time and the first NOx drop time defined in the previous paragraph. For example, the ammonia injection action of point A occurs at the 350th second, and the first recorded NOx drop corresponds to the 362nd second, then its response delay value is 12 seconds, and so on for the remaining points. For the theoretical response time under the liquid supply condition, it is necessary to first collect parameters such as the current ammonia injection flow rate, ammonia diffusion rate, and spatial distribution. For example, the current liquid supply flow rate is 80 L / h. According to historical operation records and experimental fitting relationships, the theoretical diffusion reaction time under this condition is 10 seconds. The reaction time is determined by the distance from the ammonia injection inlet to the target point, ambient temperature, mixing state, etc. For example, the lower layer point is within 3 meters of the ammonia injection inlet, and the diffusion rate at a temperature of 60 °C is 0.3 m / s, and the time obtained is seconds, then the theoretical response time is recorded as 10 seconds. Perform the operation for each point. For example, the response delay of the above point is 12 seconds, then the lag value is 2 seconds. If the response delay of point B is 9 seconds, then the lag value is -1 second. In actual operation, the response delay and theoretical time of each point are recorded through a data table. See Table 5: Table 5 Data Table for Calculating Lag Values

[0034] As shown in Table 5, to obtain the lag value, it is necessary to combine the time series response and the theoretical time calculated from the working conditions to jointly execute the difference calculation process. The response lag values of each point obtained are the core data required for the next step of analysis.

[0035] S412: Combine the response lag values of each point, collect the difference between the NOx real-time concentration of each point after the end of the current cycle and the reference concentration before corresponding ammonia injection, and construct the NOx real-time residual value set corresponding to each point. Use the formula: ; Calculate the residual lag slope value , where represents the response lag value of the i-th point, represents the NOx real-time residual value of the i-th point, represents the average value of the response lag values of all points, represents the average value of the NOx real-time residual values of all points, represents the number of points; According to the response lag values of the above-mentioned each point, it is necessary to further collect the NOx concentration data after the end of the current working condition cycle. This data is from the actual NOx emission values recorded at the detection points. For example, the NOx concentration at the end of the current cycle at Point A is 85 mg / m³, and the corresponding reference concentration at this point before ammonia injection is 90 mg / m³, resulting in a real-time residual value of -5 mg / m³. Perform this operation for all points to obtain the residual value of each point , and at the same time record the response lag value of the point as . Subsequently, construct two sets of sequence data and , perform a linear fitting operation, obtain the total number of points , and substitute them into the lag values in the above table : 2, -1, 5, corresponding to the residual values : -5, -2, -10, calculate the average value of each group , . Further calculate the numerator part: ; ; The denominator part is: ; Therefore, the slope , and the result is -1.33, which is a negative value, indicating that the more lagged the response, the more the residual tends to negative growth, and the two show a downward trend in the same direction.

[0036] S413: Based on the residual lag slope value, organize the point slope value and the co-directionality judgment result, output the correlation strength between the lag and the residual of each point in time sequence, and establish the residual lag correlation data; Based on the residual lag slope value obtained above, continue to perform data organization operations. First, establish a one-to-one correspondence between the slope results of all points and the corresponding lag values and residual values. For example, the lag value of point A is 2 seconds, the residual is -5 mg / m³, and the fitting slope is -1.33; for point B, it is -1 second, -2 mg / m³, -1.33; for point C, it is 5 seconds, -10 mg / m³, -1.33, forming three data sets. Then, perform time sorting according to the collection order, and generate sequence identifiers according to the point numbers. Finally, form a structured record table containing point numbers, response time delays, lag values, NOx residuals, and slope values with these data, which is used to track the long-term trend and regional consistency of the lag effect, obtain the residual lag correlation data. This result provides a data basis for identifying the NOx ammonia injection adjustment deviation effect and can be used for subsequent calls in the judgment of the ammonia injection rhythm control in the region.

[0037] Please refer to Figure 6 , and the steps of S5 are as follows: S511: According to the residual lag correlation data, screen the point regions with positive slope values, identify the micro-regions with positive residual lag relationships in the current cycle, extract the ammonia injection amount data and the liquid supply concentration values of the points corresponding to the micro-regions, and obtain the ammonia injection and liquid supply data for the positive micro-regions; When screening the positive relationship micro-regions according to the residual lag correlation data, it is necessary to call the residual and lag pairing data of all points in the previous cycle. First, extract the data points with a residual lag correlation coefficient greater than 0 from the point response data collected in the entire cycle. For example, among points A, B, and C, the residual of point A is 5 ppm, the lag is 6 s, the residual of point B is -2 ppm, the lag is 3 s, and the residual of point C is 3 ppm, the lag is 5 s. Judge its slope value through linear regression. If the slopes of points A and C are both greater than 0, then determine them as the set of positive micro-region points. Subsequently, extract the total ammonia injection amount and the liquid supply concentration values of points A and C in the current cycle from the ammonia injection record system. For example, the ammonia injection amount of point A is 3.5 L / min, the corresponding concentration is 28%, and the ammonia injection amount of point C is 4.2 L / min, the concentration is 31%. The above extraction results form a data set to form the ammonia injection and liquid supply data for the positive micro-regions; this data can be used as the basic data set for all adjustment ratio calculations and subsequent execution of correction actions, and is used to be superimposed and compared with the NOx residual trend to judge whether it constitutes an adjustment basis and perform parameter correction processing, and finally obtain the ammonia injection and liquid supply data for the positive micro-regions.

[0038] S512: Based on the forward micro-area ammonia injection liquid supply data and the real-time residual value of NOx at the corresponding point, the deviation degree of the current liquid supply parameter to the residual response is calculated, the ammonia injection amount adjustment ratio and the liquid supply concentration adjustment ratio are calculated, and the ammonia injection control deviation ratio data is generated; After obtaining the forward micro-area ammonia injection supply data, it is necessary to combine the ammonia injection parameters of each point in the current cycle with the NOx residual. The calculation of the ammonia injection adjustment ratio is based on the ammonia injection change amplitude corresponding to each unit residual change. For example, the ammonia injection amount of point A is 3.5L / min, the NOx residual is 5ppm, the ammonia injection amount of point C is 4.2L / min, and the residual is 3ppm. The corresponding ammonia injection increment under unit residual is 0.7 and 1.4L / min / ppm. Combined with the current liquid supply concentration of 28% and 31% respectively, the concentration adjustment amplitude is calculated proportionally to obtain the composite offset degree between concentration and ammonia injection amount; in order to further reflect the above adjustment The numerical basis of the adjustment operation is set as 0.6L / min / ppm for the baseline ammonia injection amount adjustment threshold and 2% for the concentration adjustment threshold. By comparing with the above calculated values, it is found that the ammonia injection deviation at point A exceeds the threshold but the concentration adjustment does not exceed the limit, while point C exceeds the limit at the same time. Therefore, point C is included in the control target list, and it is recorded that it needs to perform dual parameter adjustment operations. This process relies on the participating items in the data set, compares the gap between the degree of deviation and the threshold point by point, completes the classification marking and outputs the result data table as a reference for the correction execution operation. Through the above analysis process, the ammonia injection and concentration correction parameters can be linked and organized into structured records to generate ammonia injection control deviation ratio data.

[0039] S513: According to the ammonia injection control deviation ratio data, the ammonia injection amount and the liquid supply concentration at the corresponding point are corrected, the parameter change values ​​and correction directions before and after the adjustment are recorded, and the ammonia injection control adjustment record is established; According to the ammonia injection control offset ratio data, the current ammonia injection set value will be called in the point data recording module and combined with the adjustment ratio result to form a parameter update instruction. In actual operation, the corresponding point such as point C has an original ammonia injection volume of 4.2L / min, which needs to be increased by 1.4L / min. After adjustment, it is set to 5.6L / min, and the liquid supply concentration is adjusted from the original 31% to 33%; the above adjustment operations are simultaneously recorded in the control record table, and the field content covers the value before adjustment, the value after adjustment, the adjustment direction and the adjustment amplitude. For example, the adjustment direction of the ammonia injection volume at point C is "increase", the amplitude is 33.3%, and the concentration adjustment direction is also "increase", the amplitude is 6.45%; the record structure is used to construct cross-cycle comparison and ammonia injection control strategy traceability. By saving the point, cycle, parameter value and direction identification, the ammonia injection control action management and full process tracking are realized, and finally the ammonia injection control adjustment record is established.

[0040] A boiler denitration intelligent control system, comprising: The jump recognition module obtains the NOx concentration values at the detection point in the current and previous cycles, calculates the difference and compares it with the jump recognition threshold to screen for jump points and generate a concentration jump determination result; The response recognition module obtains the air flow velocity value and the flow velocity reference value according to the concentration jump determination result, calculates the disturbance deviation ratio, determines the response point, and generates an ammonia injection response area recognition result; The time delay determination module extracts the ammonia injection start time and the NOx concentration decrease time according to the ammonia injection response area recognition result, calculates and determines the response time delay value, and generates NOx response time delay data; The residual correlation module obtains the response lag value and the NOx residual value based on the NOx response time delay data, performs linear fitting, extracts the slope to judge the co-variation relationship, and generates residual lag correlation data; The regulation and adjustment module extracts the ammonia injection amount and the liquid supply concentration according to the residual lag correlation data, calculates the adjustment data in combination with the residual value and executes the correction action, and generates an ammonia injection regulation and adjustment record.

[0041] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as the technical solution content of the present invention is not departed from, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent control method for boiler denitration, characterized in that, It includes the following steps: S1: Obtain the current NOx concentration value at the NOx detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling period, calculate the concentration difference and compare it with the jump identification threshold to generate a jump concentration determination result; S2: According to the jump concentration determination result, combine the air flow velocity value and the standard flow rate reference value to calculate the disturbance deviation, judge whether the deviation value exceeds the reference, synchronously calculate the ratio of the concentration difference to the disturbance deviation, screen the area where the ratio exceeds the limit, and generate a spray ammonia response area identification result; S3: Based on the spray ammonia response area identification result, extract the spray ammonia start time and the time point when the NOx concentration first decreases, calculate the response delay value, and generate NOx response delay data; S4: According to the NOx response delay data, statistically calculate the difference between the response delay value and the theoretical response time as the lag value, collect the real-time NOx residual value at the corresponding point, perform a linear fit of the lag value and the real-time residual value and judge the co-directional change relationship to generate residual lag correlation data; S5: Determine the residual lag correlation data, extract the spray ammonia amount and the liquid supply concentration value in the positive relationship micro-region, combine the real-time NOx residual value to calculate the spray ammonia amount adjustment ratio and the concentration adjustment ratio, and perform a correction action to generate a spray ammonia regulation adjustment record.

2. The intelligent control method for boiler denitration according to claim 1, characterized in that, The jump concentration determination result includes the jump point position identifier, the jump amplitude level, and the jump trend direction. The spray ammonia response area identification result includes the high response area position, the response intensity level, and the response influence range. The NOx response delay data includes the response time difference sequence, the response start time set, and the response end time set. The residual lag correlation data includes the linear fit slope value, the change direction consistency index, and the lag and residual fit degree. The spray ammonia regulation adjustment record includes the spray ammonia adjustment ratio, the liquid supply concentration adjustment ratio, and the adjustment applicable area.

3. The intelligent control method for boiler denitration according to claim 1, characterized in that, The steps for obtaining the jump concentration determination result include: S111: Based on the current NOx concentration value at the NOx flue gas detection point in the boiler furnace and the NOx concentration value at the corresponding moment in the previous sampling period, calculate the difference between the two, and use the absolute difference between the current NOx concentration value and the NOx concentration value in the previous period for the difference calculation to obtain the concentration change difference; S112: According to the concentration change difference, compare it item by item with the jump identification threshold, identify the concentration change difference items greater than the jump identification threshold, and use the formula: ； Calculate the multi - cycle composite concentration fluctuation value , obtain the jump concentration judgment index value, where represents the NOx concentration value of the current cycle, represents the NOx concentration value of the previous cycle, represents the average value of the jump recognition threshold; S113: According to the magnitude relationship between the jump concentration judgment index value and the jump identification threshold, mark all the time points and the corresponding detection point numbers where the jump concentration judgment index value is greater than the jump identification threshold to generate a jump concentration determination result.

4. The intelligent control method for boiler denitration according to claim 1, characterized in that The steps for obtaining the spray ammonia response area identification result include: S211: Based on the jump concentration determination result, combine the air flow velocity value of each detection point and the corresponding standard flow rate reference value, calculate the difference between the two, take the absolute value to obtain the flow velocity disturbance amount value of each detection point, and calculate the disturbance degree through the ratio of the disturbance amount value to the reference value to obtain the disturbance deviation ratio; S212: According to the disturbance deviation ratio, and combine it with the concentration change difference of the corresponding detection point for joint processing, using the formula: ； Calculate the composite ratio of concentration perturbations , where represents the NOx concentration value at the i-th moment, represents the NOx concentration value at the previous moment, represents the air flow velocity value at the i-th point, represents the standard flow velocity reference value at the i-th point; S213: According to the concentration perturbation composite ratio, compare with the preset concentration perturbation identification threshold, screen all spatial regions where the ratio exceeds the threshold, record the position information and time information of the corresponding detection points, and establish the identification result of the ammonia injection response region.

5. The intelligent control method for boiler denitration according to claim 1, characterized in that, The steps for obtaining the NOx response time delay data include: S311: Based on the set of points in the identification result of the ammonia injection response region, collect the continuous time series values of the NOx concentration corresponding to each point after the ammonia injection action. According to the change trend of the NOx concentration value in the collection sequence of each point, judge the numerical fluctuation direction between each time series point, identify the moment when it first changes from a non-decreasing state to a decreasing state, and obtain the first NOx concentration decrease time point. S312: According to the first NOx concentration decrease time point, combine with the trigger time of the ammonia injection action at each point, calculate the time difference between the two respectively, record the time interval data corresponding to each point, and obtain the point response time delay value. S313: According to the point response time delay value, arrange the response times of each point after the ammonia injection action is triggered in the order of point number and time, form a cross-point response time set, and establish the NOx response time delay data.

6. The intelligent control method for boiler denitration according to claim 1, wherein The steps for obtaining the residual lag correlation data include: S411: According to the response time delay values of each point in the NOx response time delay data, combine with the current liquid supply condition, calculate the theoretical response time of the reaction at each point under the current liquid supply condition, and subtract the response time delay value of each point from the theoretical response time to obtain the response lag value of each point. S412: Combine the response lag values of each point, collect the difference between the NOx real-time concentration and the reference concentration before ammonia injection corresponding to each point after the end of the current cycle, construct a set of NOx real-time residual values corresponding to each point, and use the formula: ； Calculate the residual lag slope value , where represents the response lag value at the i-th point, represents the real-time residual value of NOx at the i-th point, represents the average value of the response lag values of all points, represents the average value of the real-time residual values of NOx of all points, represents the number of points; S413: According to the residual lag slope value, sort out the point slope value and the co-directionality judgment result, and output the correlation intensity between the lag and the residual of each point in time sequence to establish the residual lag correlation data.

7. The intelligent control method for boiler denitration according to claim 1, characterized in that, The steps for obtaining the ammonia injection regulation adjustment record include: S511: According to the residual lag correlation data, screen the point regions with positive slope values, identify the micro-regions with positive residual lag relationships within the current cycle, extract the ammonia injection amount data and the liquid supply concentration value of the points corresponding to the micro-regions, and obtain the ammonia injection and liquid supply data of the positive micro-regions. S512: Based on the ammonia injection and liquid supply data of the positive micro-regions, combine with the NOx real-time residual value of the corresponding point, calculate the offset degree of the current liquid supply parameter on the residual response, calculate the ammonia injection amount adjustment ratio and the liquid supply concentration adjustment ratio, and generate the ammonia injection regulation offset ratio data. S513: According to the ammonia injection regulation offset ratio data, perform the correction operations on the ammonia injection amount and the liquid supply concentration of the corresponding points, record the parameter change values and the correction directions before and after the adjustment, and establish the ammonia injection regulation adjustment record.

8. An intelligent control system for boiler denitration, characterized in that, The system for executing the method according to any one of claims 1-7 includes: The jump identification module obtains the NOx concentration values of the detection point in the current and the previous cycle, calculates the difference and compares it with the jump identification threshold, screens the jump points, and generates the concentration jump determination result. The response recognition module obtains the air flow velocity value and the flow velocity reference value according to the concentration jump determination result, calculates the disturbance deviation ratio, determines the response point, and generates the identification result of the ammonia injection response area; The time delay determination module extracts the ammonia injection start time and the NOx concentration decrease time according to the identification result of the ammonia injection response area, calculates and determines the response time delay value, and generates the NOx response time delay data; The residual correlation module obtains the response lag value and the NOx residual value based on the NOx response time delay data, performs linear fitting, extracts the slope to judge the co-variation relationship, and generates the residual lag correlation data; The regulation and adjustment module extracts the ammonia injection amount and the liquid supply concentration according to the residual lag correlation data, calculates the adjustment data in combination with the residual value and executes the correction action, and generates the ammonia injection regulation and adjustment record.

Images