A method and system for controlling direct injection of urea for denitrification based on batch phase change transport.

By dividing the flue into computational units and recording different phase change states of urea, the method of direct injection urea denitrification control is promoted by predicting flue gas conditions, which solves the problem of insufficient material balance framework in the existing technology and realizes accurate ammonia supply coordination and system safety in the multi-cycle prediction time domain.

CN123102145APending Publication Date: 2026-11-17HUANENG NANJING THERMAL POWER CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611525553.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-29
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies fail to differentiate between different phase change states of injected urea within the same material balance framework, and do not manage the remaining state of historical injection batches and the expected arrival of planned injection batches in multiple pending cycles separately. This makes it difficult to accurately coordinate the ammonia supply contribution of new and old batches in the multi-cycle prediction time domain, and lacks explicit maintenance of the conservation relationship between materials that have exceeded the limit and the remaining materials in transit when operating conditions change.

Method used

A boiler urea direct injection selective catalytic reduction denitrification control method is adopted. By dividing the flue into interconnected computing units, the state of unconverted urea, decomposable precursors and generated ammonia is recorded and maintained. The predicted state is advanced by using predicted flue gas conditions to form the expected arrival amount of the injected batch. The remaining state is updated at the end of the control cycle to ensure material balance and constraint solution.

Benefits of technology

It achieves material balance when operating conditions change, improves the time sharing capability and control coordination of the ammonia supply sequence, ensures system safety and control continuity, and avoids the defects of material balance constraints and state attribution tracking.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123102145A_ABST
    Figure CN123102145A_ABST
Patent Text Reader

Abstract

This invention discloses a urea direct injection denitrification control method and system based on batch phase change transport status, relating to flue gas denitrification in coal-fired boilers. The method establishes the injection batch based on the measured urea quantity at the nozzle. In the flue gas calculation unit, it maintains the equivalent of unconverted urea, the equivalent of decomposable precursors, and the mass of generated ammonia. Simultaneously, it advances phase change conversion and unit transport step-by-step, accumulating the effective ammonia exceeding the catalyst inlet and the equivalent that has not crossed the boundary in ammonia form. The sum of the total remaining state of the batch and the accumulated amount of exceeding the boundary is maintained as the theoretical ammonia equivalent. A predicted copy is generated from the formal state to calculate the expected arrival amount of historical batches, generating a multi-cycle candidate urea flow rate and atomized gas setpoint sequence. The candidate and historical batch arrival amounts are superimposed and constrained for solution. Only the first cycle instruction is executed, and the formal state is updated at the end of the cycle based on the measured nozzle throughput and flue gas conditions. This method manages the injection facts, remaining state, and planned injection quantity separately, and can safely exit when there are hard constraint conflicts.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of flue gas denitrification and process control technology for coal-fired boilers, and in particular to a urea direct injection denitrification control method and system based on batch phase change transport state. Background Technology

[0002] The effective reducing agent supply of a urea direct injection denitrification system depends on the actual mass of urea passing through the nozzle, atomization conditions, and the temperature and velocity of the flue gas along the path. Urea and its decomposition products that have been injected into the flue but have not yet reached the catalyst inlet will continue to be affected by subsequent changes in operating conditions during transport. Batches injected into the same injection area at different times may be at different phase transition stages and reach the target area at the catalyst inlet along different paths.

[0003] In existing urea direct injection denitrification control technologies, predictive control methods have been used to estimate downstream ammonia arrivals and apply constraints based on historical control actions, such as dynamic matrix control and total ammonia injection optimization based on predictive models. However, for urea direct injection processes within flue gas ducts, the remaining conversion and transport states of injected materials change in real time with operating conditions, and the actual injection volume is irreversible once injected. Existing technologies do not differentiate between tracking different phase change states of injected urea (e.g., unconverted urea, decomposable precursors, and ammonia generated) within the same material balance framework, nor do they separately manage and superimpose the remaining states of historical injection batches with the expected arrival volumes of planned injection batches for multiple pending cycles. This makes it difficult to accurately coordinate the ammonia supply contributions of new and old batches in the multi-cycle prediction time domain, and lacks explicit maintenance of the conservation relationship between materials that have exceeded limits and remaining materials in transit when operating conditions change. Summary of the Invention

[0004] The purpose of this invention is to provide a urea direct injection denitrification control method and system based on batch phase change transport status, in order to solve the following technical problems existing in the prior art: when the actual urea injection volume is irreversible and the remaining conversion and transport process continues to change with the operating conditions, the prior art does not distinguish and track the different phase change states of the injected urea under the same material balance framework, nor does it manage and superimpose the remaining state of the historical injection batches and the expected arrival volume of the planned injection batches in multiple pending cycles, which makes it difficult to accurately coordinate the ammonia supply contribution of the new and old batches in the multi-cycle prediction time domain, and lacks the technical problem of explicit maintenance of the conservation relationship between the material that has exceeded the limit and the remaining material in transit when the operating conditions change.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: The present invention provides a boiler urea direct injection selective catalytic reduction denitrification control method, applied to a denitrification system having at least two independently adjustable urea injection zones and at least two catalyst inlet target zones, executed by a controller according to a control cycle, characterized in that it includes: S1. Divide the flue from the reference section of each nozzle to the reference surface of the catalyst inlet into interconnected calculation units, and obtain the flue gas temperature and transport velocity of each calculation unit, as well as the actual flow rate of urea solution, urea mass fraction and atomized gas state of each injection area. S2. At the beginning of the control cycle, read the formal batch status up to the end of the previous control cycle. For each injected batch, associate and save the batch identifier, injection area, injection time, and theoretical ammonia equivalent calculated from the actual mass of pure urea passing through the nozzle reference section. In the calculation unit, record the unconverted urea state, decomposable precursor state, and ammonia generated state measured by theoretical ammonia equivalent, respectively. S3. The single-step advancement of the formal batch state and the predicted state shall follow the following conservation rules: Based on the flue gas temperature, transport rate and batch atomization formation conditions of each calculation unit, the transformation between the above states and the transport between calculation units shall be advanced within the same physical time substep. The effective ammonia mass that crosses the catalyst inlet reference surface in the form of ammonia and the urea equivalent that does not cross the reference surface in the form of ammonia shall be accumulated and recorded. The sum of the total amount of the remaining state of the batch and the accumulated amount that has exceeded the limit shall be kept as the theoretical ammonia equivalent of the batch, and the accumulated excess portion shall be removed from the remaining state. S4. The predicted state is copied from the formal batch state, and the predicted flue gas conditions are used to advance the predicted state to form the expected effective ammonia arrival amount of the injected batch in each target area and each predicted time slot. The prediction advancement does not change the formal batch state or the accumulated amount that has exceeded the limit. S5. Based on the predicted nitrogen oxide removal mass and ammonia nitrogen equivalent relationship, generate the effective ammonia demand of each target area in each predicted time slot. Construct a control sequence with multiple candidate urea flow rates and candidate atomizing gas setpoints including the current control cycle and at least one subsequent control cycle. Predict the phase change transport state of each candidate batch. Superimpose the arrival amount generated by each candidate batch according to its planned injection time with the expected arrival amount of the injected batch to obtain a control sequence that satisfies the constraints of injection branch flow rate, atomizing gas working range and pyrolysis conditions. S6. Only the current control cycle setting value of the control sequence is sent to the actuators of each injection area, and the unexecuted candidate batch is not written into the formal batch status; S7. At the end of the current control cycle, advance the original batch and the newly injected batch of the cycle according to the measured flue gas conditions and actual nozzle throughput of the cycle, update the cumulative amount of out-of-bounds and the remaining status, and then submit the updated results as the official batch status of the next control cycle.

[0006] This invention establishes injection batches based on the actual urea mass passing through the nozzle reference section and saves the non-overwriteable injection facts (step S2). Within the flue gas calculation unit, it maintains the unconverted urea equivalent, decomposable precursor equivalent, and generated ammonia mass for each batch. It updates the remaining state and accumulates out-of-bounds amounts using phase change and transport propulsion within the same physical time substep (step S3), maintaining material balance between the total injection volume, remaining materials, and out-of-bounds materials. At the start of the control cycle, the predicted arrival contribution of historical batches is calculated using a copy of the formal state (step S4). Candidate urea and atomizing gas sequences are set for the current and subsequent control cycles. The arrival contributions of each candidate batch at different injection times are superimposed with the arrival contributions of historical batches. After solving for sequences that satisfy equipment constraints, only the first cycle command is executed (steps S5 and S6). After the cycle ends, the formal state is updated based on the actual nozzle throughput and measured flue gas conditions (step S7). The predicted state copy and unexecuted candidate quantities are not submitted to the formal state, thus allowing for separate management of historical injection facts, remaining physical states, and planned injection volumes.

[0007] Optionally, multiple sampling intervals within the control cycle are treated as injection sub-batches. The pure urea mass is determined by integrating the product of the actual mass flow rate of the urea solution and the urea mass fraction within each sampling interval, and then multiplied by 34 / 60 to obtain the theoretical ammonia equivalent. When there is a variable tube storage between the flowmeter and the nozzle reference section, the actual flow rate through the nozzle is calculated based on the actual flowmeter measurement and the change in tube storage. The execution model for sending commands to the nozzle and the phase change transport model from the nozzle to the catalyst inlet are set separately.

[0008] Furthermore, the transfer fraction from the unconverted urea state to the decomposable precursor state and the transfer fraction from the decomposable precursor state to the ammonia-generated state are determined by the non-negative calibration rate and substep duration related to flue gas temperature, moisture content, droplet size, and residence state, respectively, with each transfer fraction ranging from 0 to 1. The transport fraction between computational units is determined based on the representative transport velocity of the corresponding phase, the computational unit length, and the substep duration. The substep duration is selected such that the output fraction of each unit is no greater than 1, and the sum of the diversion fractions of each target unit and the target region is 1.

[0009] Furthermore, for each batch, the re-releaseable wall retention state is recorded in terms of theoretical ammonia equivalent. The urea equivalent captured by the wall in the unit is transferred from the corresponding flow state to the wall retention state. The wall release amount re-enters the decomposable precursor state of the same batch. The wall retention amount that has not yet been released is retained in the total remaining state of the batch and is not cancelled as an out-of-bounds amount.

[0010] Furthermore, the connectivity between the computing units and the outlet relationships leading to the catalyst inlet target region are determined by the calibrated flow field model, tracer response, or partitioned disturbance response. When operating conditions change, the remaining material state of each batch is advanced based on the updated unit flue gas state and diversion share, while retaining the injection facts and the accumulated amount exceeding the limit. Injection batches from at least one injection zone can enter two target regions via different paths and achieve effective ammonia arrival in two different predicted time slots.

[0011] Furthermore, the controller is configured with a control time domain and a prediction time domain. Outside the control time domain but within the prediction time domain, candidate control sequences are supplemented according to a preset end-of-pipe hold strategy or end-of-pipe control strategy. The deviation between the historical batch arrival quantity and the sum of all candidate batch arrival quantities relative to the effective ammonia demand is calculated, along with the penalty for changes in adjacent candidate control quantities. Candidate materials remaining in the calculation unit or wall at the end of the prediction are retained as the prediction end-of-pipe residual state and used for end-of-pipe constraints or end-of-pipe penalties.

[0012] Furthermore, when obtaining intermediate measurement results that can be attributed to a specific batch, the mass is redistributed among the remaining material states while keeping the theoretical ammonia equivalent injected into that batch and the cumulative amount exceeding the limit unchanged. After correction, each remaining state is non-negative and their sum remains unchanged. The total outlet measurement that cannot be attributed to a specific batch is used for bounded correction of common model parameters, demand deviations, or state uncertainties. This correction does not reintroduce the exceeded mass into the remaining states.

[0013] Furthermore, the injection branch's stop-injection and stable-injection states are set separately, ensuring the urea flow rate is zero or between the minimum stable flow rate and the allowable upper limit. When a safe and feasible control sequence exists but the effective ammonia demand cannot be fully met, a safe and feasible sequence is output and the demand gap is recorded. When safety constraints conflict with ordinary rate-of-change constraints, a stop-injection or protected transition permitted by the equipment is executed according to the pre-set interlock priority, a fault flag is issued, and the actual flow rate is verified. Urea increments are prohibited when atomizing gas is insufficient; after urea reduction, atomizing gas is maintained according to preset purging conditions.

[0014] This invention also provides a boiler urea direct injection selective catalytic reduction denitrification control system, including at least two independently adjustable urea injection zones, urea flow measurement and adjustment units for each zone, atomized gas measurement and adjustment units, urea mass fraction acquisition units, flue gas temperature and velocity acquisition units, nitrogen oxide acquisition units, a selective catalytic reduction reactor with at least two catalyst inlet target zones, and a controller communicatively connected to each unit. The controller's memory includes a fact storage area for storing the actual flow rate and injection time of the injection batch, a formal state storage area for storing the unconverted urea equivalent, decomposable precursor equivalent, generated ammonia mass, and batch excess accumulation in the flue gas calculation unit, and a predictive state storage area independent of the formal state storage area. The controller's processor is configured to generate multiple candidate control sequences for execution cycles according to the above control method, output only the current cycle setpoint, update the batch phase change transport state and excess accumulation with the actual injection flow rate and measured flue gas conditions at the end of the cycle, cancel the excess state components, and update the formal state and relative time slot reference for the next cycle.

[0015] Furthermore, the controller also includes a quality flag and a status recovery storage area. During periods when actual flow measurement is invalid, it saves the injection volume range determined based on redundant measurements or the physical boundaries of the actuator, and includes any newly added material within this range in the status uncertainty. If the preset allowable duration is exceeded, a stop-spray or protected transition command is output to the corresponding branch. When the cumulative flow is restored and verifiable, the batches during this period are checked once. Compared with the prior art, the present invention has the following beneficial effects: 1. By establishing injection batches based on the actual urea mass passing through the nozzle reference section and preserving the non-overwriteable injection facts, the equivalent of unconverted urea, the equivalent of decomposable precursors, and the mass of ammonia generated are maintained within the flue gas calculation unit. Phase change conversion and inter-unit transport are advanced within the same physical time substep. The effective ammonia that crosses the catalyst inlet and the equivalent that does not cross the boundary in the form of ammonia are accumulated, maintaining the material balance between the total injection volume, the remaining material, and the material that has crossed the boundary. This ensures that the new and old batches can still maintain independent material ownership when the operating conditions change, and the material that has crossed the boundary will not reappear due to model updates. This overcomes the shortcomings of existing technologies that lack explicit material balance constraints and state ownership tracking.

[0016] 2. By replicating the predicted state from the formal batch state, the predicted flue gas conditions are used to advance the predicted state to form the historical batch expected arrival amount. At the same time, a control sequence is formed by multiple candidate urea flow rates and candidate atomizing gas setpoints, including the current control cycle and at least one subsequent control cycle. The candidate batch arrival amounts and historical batch arrival amounts are superimposed and the constraints are solved. Only the first cycle command is executed. Candidate batches that are not executed are not written into the formal state. This allows the historical injection facts, remaining physical state and planned injection amount to be managed separately. Multiple planning cycles share the predicted time domain demand, avoiding the concentration of the entire time domain demand in the current batch. This improves the time sharing capability and control coordination of the ammonia supply sequence.

[0017] 3. By setting the stop-spray state and stable spray state of the injection branch separately, the conflict between demand gap and hard constraint is distinguished. When the safety constraint and the normal rate of change constraint conflict, the stop-spray or protected transition allowed by the equipment is executed according to the pre-set interlock priority. When the atomizing gas is insufficient, the increase of urea is prohibited. After the urea is reduced, the atomizing gas is maintained according to the preset purging conditions. This allows for safe exit when there is a hard constraint conflict, without ignoring the safety boundary, and ensures the system safety and control continuity under extreme conditions. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the control system structure and signal connection provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram illustrating the relationship between injection facts, formal material status, predicted copies, and candidate sequence data in an embodiment of this application.

[0020] Figure 3 This is a schematic diagram illustrating phase transition, wall retention, transport, and boundary overflow cumulative conservation propulsion provided for embodiments of this application.

[0021] Figure 4 A control cycle flowchart provided for embodiments of this application.

[0022] Figure reference numerals: 100, Urea supply and concentration acquisition unit; 200, Zoned injection branch; 210, Urea flow measurement and adjustment unit; 220, Atomizing gas measurement and adjustment unit; 230, Spray gun and nozzle reference section; 250, Flue calculation unit; 300, SCR reactor; 310, Catalyst inlet target area; 320, Nitrogen oxides acquisition unit; 330, Ammonia slip acquisition unit; 400, Controller; 410, Injection fact storage area; 420, Formal material status storage area; 430, Predicted status copy; 440, Demand and candidate sequence calculation module; 450, Constraint allocation module; 460, Safety execution module. Detailed Implementation

[0023] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the following embodiments are used to illustrate the technical solutions of the present application, and are not intended to limit the scope of protection of the present application.

[0024] This invention provides a boiler urea direct injection selective catalytic reduction denitrification control method, applied to a denitrification system having at least two independently adjustable urea injection zones and at least two catalyst inlet target zones, executed by a controller according to a control cycle, characterized by comprising: S1. Divide the flue from the reference section of each nozzle to the reference surface of the catalyst inlet into interconnected calculation units, and obtain the flue gas temperature and transport velocity of each calculation unit, as well as the actual flow rate of urea solution, urea mass fraction and atomized gas state of each injection area. S2. At the beginning of the control cycle, read the formal batch status up to the end of the previous control cycle. For each injected batch, associate and save the batch identifier, injection area, injection time, and theoretical ammonia equivalent calculated from the actual mass of pure urea passing through the nozzle reference section. In the calculation unit, record the unconverted urea state, decomposable precursor state, and ammonia generated state measured by theoretical ammonia equivalent, respectively. S3. The single-step advancement of the formal batch state and the predicted state shall follow the following conservation rules: Based on the flue gas temperature, transport rate and batch atomization formation conditions of each calculation unit, the transformation between the above states and the transport between calculation units shall be advanced within the same physical time substep. The effective ammonia mass that crosses the catalyst inlet reference surface in the form of ammonia and the urea equivalent that does not cross the reference surface in the form of ammonia shall be accumulated and recorded. The sum of the total amount of the remaining state of the batch and the accumulated amount that has exceeded the limit shall be kept as the theoretical ammonia equivalent of the batch, and the accumulated excess portion shall be removed from the remaining state. S4. The predicted state is copied from the formal batch state, and the predicted flue gas conditions are used to advance the predicted state to form the expected effective ammonia arrival amount of the injected batch in each target area and each predicted time slot. The prediction advancement does not change the formal batch state or the accumulated amount that has exceeded the limit. S5. Based on the predicted nitrogen oxide removal mass and ammonia nitrogen equivalent relationship, generate the effective ammonia demand of each target area in each predicted time slot. Construct a control sequence with multiple candidate urea flow rates and candidate atomizing gas setpoints including the current control cycle and at least one subsequent control cycle. Predict the phase change transport state of each candidate batch. Superimpose the arrival amount generated by each candidate batch according to its planned injection time with the expected arrival amount of the injected batch to obtain a control sequence that satisfies the constraints of injection branch flow rate, atomizing gas working range and pyrolysis conditions. S6. Only the current control cycle setting value of the control sequence is sent to the actuators of each injection area, and the unexecuted candidate batch is not written into the formal batch status; S7. At the end of the current control cycle, advance the original batch and the newly injected batch of the cycle according to the measured flue gas conditions and actual nozzle throughput of the cycle, update the cumulative amount of out-of-bounds and the remaining status, and then submit the updated results as the official batch status of the next control cycle.

[0025] This invention establishes injection batches based on the actual urea mass passing through the nozzle reference section and saves the non-overwriteable injection facts (step S2). Within the flue gas calculation unit, it maintains the unconverted urea equivalent, decomposable precursor equivalent, and generated ammonia mass for each batch. It updates the remaining state and accumulates out-of-bounds amounts using phase change and transport propulsion within the same physical time substep (step S3), maintaining material balance between the total injection volume, remaining materials, and out-of-bounds materials. At the start of the control cycle, the predicted arrival contribution of historical batches is calculated using a copy of the formal state (step S4). Candidate urea and atomizing gas sequences are set for the current and subsequent control cycles. The arrival contributions of each candidate batch at different injection times are superimposed with the arrival contributions of historical batches. After solving for sequences that satisfy equipment constraints, only the first cycle command is executed (steps S5 and S6). After the cycle ends, the formal state is updated based on the actual nozzle throughput and measured flue gas conditions (step S7). The predicted state copy and unexecuted candidate quantities are not submitted to the formal state, thus allowing for separate management of historical injection facts, remaining physical states, and planned injection volumes.

[0026] In some embodiments, a nozzle reference section is set at each spray gun outlet, and an outlet reference surface is set at the catalyst inlet. Urea solution flow rate is expressed in kilograms per hour, urea mass fraction is dimensionless, time is expressed in seconds, U, V, W, and unconverted boundary records are expressed in kilograms of theoretical ammonia equivalent, and A and effective ammonia boundary records are expressed in kilograms of ammonia. i represents the injection area, j represents the catalyst inlet target area, b represents the injected sub-batch, and c and d represent flue gas calculation units.

[0027] Control period k corresponds to [t] k ,t k+1 ), duration Δ seconds, prediction time slot r corresponds to [t] k+ ,t k+ +1 Each cycle is divided into several computational substeps of length δ, with the sampling time aligned with the substep boundaries. The formal state at the beginning of the cycle only includes the state executed up to step t. k The injection record; the injection volume to be executed in cycle k is first predicted as a candidate and is not considered as injected material.

[0028] It should be noted that by uniformly adopting the theoretical ammonia equivalent as the unit of measurement, material balance calculations can be directly performed between different phase transition states, avoiding conversion errors caused by inconsistent units. Through the aforementioned time division, executed injections and planned injections are clearly separated on the time axis, ensuring a clear boundary between the formal state and the predicted state in the time dimension.

[0029] In some embodiments, at the end of cycle k, the formal state at the beginning of the cycle is re-advanced according to the measured flue gas conditions, and the measured injection volume from the nozzle is added to the corresponding unit within its occurrence interval according to the sampling interval, forming each sub-batch. Piecewise constant flow rate or linear interpolation integration is used within the sampling interval; when the substep is sufficiently small, midpoint injection can be used as an approximation. The injection interval and integration rules are simultaneously stored in the batch record. This advancement can be completed online step by step, but the verified formal state is submitted only once at the cycle boundary. By submitting the verified formal state only at the cycle boundary, the atomicity and traceability of the state update are guaranteed, and the misuse of intermediate states during the online advancement process is avoided.

[0030] In some embodiments, if the flow meter is located near the spray gun and the pipe flow from the flow meter to the nozzle is negligible, the nozzle throughput is determined by flow meter integration. If there is filling / draining, accumulation, or leakage, the pipe flow change is estimated based on the pipe volume, pressure, and cumulative upstream and downstream flow, and the nozzle throughput is obtained by mass balance. When reliable conversion is not possible, the channel enters a state of metering uncertainty, and the valve position command is not directly treated as the actual injection volume. The actuator response is only used to predict when the command will generate nozzle flow, and the arrival delay after the nozzle is added repeatedly.

[0031] In some embodiments, the theoretical ammonia equivalent Pb of a sub-batch is calculated using the following formula:

[0032] In the formula P b Let be the theoretical ammonia equivalent for the sub-batch, s be the integration interval within that batch, and q and c be the actual average urea solution flow rate and the concentration used for integrating the product of the flow rate and c, respectively, within that interval. The integral of the product of q and c over concentration changes is not replaced by the product of two independent arithmetic means. The periodic average flow rate is equal to the actual urea solution mass for that period multiplied by 3600 / Δ. The theoretical ammonia conversion is derived from the mass relationship between the complete hydrolysis of one mole of urea and the production of two moles of ammonia.

[0033] It should be noted that by converting the actual mass of pure urea passing through the nozzle reference section into the theoretical ammonia equivalent, the injection batch is established, clarifying the physical measurement benchmark and conversion relationship of the injection volume, thus providing a clear constraint on the total material volume of each batch. When variable pipe storage exists, the actual flow rate through the nozzle is calculated based on the flow meter's actual measurement and the change in pipe storage. Furthermore, the execution model for commands to the nozzle and the phase change transport model from the nozzle to the catalyst inlet are set separately, distinguishing the time delay between the actuator response and the subsequent phase change transport process, thus avoiding directly treating the valve position command as the actual injection volume.

[0034] In some embodiments, such as Figure 1 As shown, the urea supply unit 100 supplies urea to each branch 200, and each branch includes a mass flow meter, a regulating valve, and a spray gun 230. The atomizing gas unit 220 includes a flow meter or differential pressure transmitter, a regulating valve, and a purge port. Each branch can independently receive urea and atomizing gas setpoints. The system obtains zoned flue gas temperature, flue gas velocity, urea concentration, actual nozzle throughput, atomizing gas volume or spray gun differential pressure, as well as inlet and outlet NOx and outlet ammonia escape.

[0035] The connectivity, length, cross-sectional area, and outlet target region 310 of the flue gas calculation unit 250 are set according to the flue gas geometry and the calibrated flow field model. Unit temperature and velocity are obtained from corresponding measuring points or reconstructed from sparse measuring points and flow field mapping, and are accompanied by quality indicators. Urea concentration can be obtained from the preparation system, temperature-compensated density measurement, or laboratory results. Nitrogen oxide concentration is converted to mass flow rate using the concentration and flue gas volumetric flow rate under the same standard conditions and dry / wet basis; if flue gas mass flow rate is used, the volumetric flow rate is first converted according to the same basis density. The emission oxygen concentration needs to be restored to the corresponding actual oxygen basis before material balance calculation.

[0036] Data is timestamped, and analyzer signals are saved separately for sampling and analysis delays. Communication interruptions, exceeding measurement ranges, and signals that remain unchanged under sufficient process excitation are marked as abnormal; under normal steady-state conditions, a fault is not directly diagnosed simply because the signal remains unchanged. Boiler load, coal quantity, air quantity, and oxygen quantity models can supplement inlet NOx prediction, but the model's applicability and prediction error should be recorded.

[0037] It should be noted that, through the aforementioned hardware configuration, the controller 400 can acquire signals such as zoned flue gas temperature, flue gas velocity, urea concentration, actual flow rate through the nozzle, atomized gas volume or spray gun pressure difference, as well as inlet and outlet NOx and outlet ammonia escape signals, providing the necessary input data for the conservation propulsion of batch material states and multi-cycle predictive control. The connectivity of the flue gas calculation unit 250 is set according to the verified flow field model, ensuring the physical rationality of the transport propulsion.

[0038] In some embodiments, such as Figure 2 and Figure 3As shown, each batch stores injection facts that cannot be overridden by subsequent candidate instructions, including identifier, region, injection interval, actual pure urea mass, theoretical ammonia equivalent P, and atomization conditions. In each unit c, the formal state stores U, V, and A respectively: U is the theoretical ammonia equivalent corresponding to urea in its liquid carrier before reaching a decomposable state; V is the theoretical ammonia equivalent of the precursor that has met the decomposition conditions but has not yet generated ammonia; and A is the mass of ammonia already generated. U and V are lumped states used for control; V is not limited to a single chemical intermediate, nor is the evaporation mass of water included in P.

[0039] In some embodiments, W is added in cases where wall trapping and re-release are present, representing the theoretical ammonia equivalent of urea and precursors that may still be released from the wall. Wall retention is the stock of uncompleted transport and cannot be directly written off as an irreversible loss. For restricted conditions where no further wall retention is possible, W can be zero; if the wall condition cannot be reliably estimated, a non-negative upper bound is recorded and the system operates in a safe, restricted manner.

[0040] In some embodiments, the total remaining batch state Sb is calculated using the following formula: ; ; C represents the cumulative estimated amount of effective ammonia exceeding the boundary, derived from measured boundary conditions within the already occurred interval and assigned to each batch; it is not equivalent to the independent measured value for each batch. Z represents the equivalent amount of ammonia that did not cross the catalyst inlet reference surface and the non-recoverable amount that was separately confirmed to be discharged from the control boundary. When unconverted material crosses this reference surface, Z is recorded only once and is no longer included in the upstream transit status. The downstream conversion of this material belongs to the process after the reference surface and should be handled in the outlet feedback and risk verification; it cannot be simultaneously counted as upstream effective ammonia.

[0041] Each formal advancement only increases the new out-of-bounds amount of this substep to C or Z; the accumulated portion remains monotonic and does not return to U, V, A, or W. The total state quantity satisfies the above balance within the tolerance ε; floating-point errors are distributed according to a preset ratio within non-negative states; if the error exceeds the tolerance, submission is rejected and a model anomaly flag is set, without generating materials to replenish them out of thin air. Batch remaining quantities that are non-zero are retained; when storage resources are limited, batches with the same state position, particle size class, and model parameters can be merged, and the mass allocation and cumulative records of each original batch can be saved.

[0042] It should be noted that by decomposing the physicochemical state of the injected material into four traceable lumped state variables (U, V, A, and W), and increasing the new out-of-bounds amount to C or Z after each formal advancement, while ensuring that the accumulated portion does not return to the remaining state, the sum of the total remaining state of the batch and the accumulated out-of-bounds amount always equals the theoretical ammonia equivalent of that batch. This material balance constraint ensures the conservation of ammonia equivalent throughout the entire transportation and conversion process, avoiding ammonia equivalent drift caused by numerical calculation errors, and providing a reliable material baseline for subsequent predictive control and measurement correction.

[0043] In some embodiments, the characteristic particle size during droplet formation is mapped by the spray gun calibration to Φ(q / a,Δp,c). u Obtain and retain the particle size category or spray gun operating point. Phase change rate κ e κ d The values ​​are obtained from a nonnegative calibration table of temperature, humidity, particle size, and phase transition stage. The input boundaries, interpolation rules, and error limits of the table should be preserved; out-of-bounds points should not be extrapolated indefinitely. The rates described are parameters of the lumped control model, including the equivalent effects of evaporation and urea decomposition processes, and do not consider simple first-order reactions as the entire urea chemical mechanism. Evaporation fraction f e and decomposition share f d Calculate using the following formula: ; ; ; ; In the above formula, the superscript 'a' indicates the state after the phase transition operation is completed. In some embodiments, in the implementation with W, W is first transferred from W to V according to the release fraction f_w while W is reduced synchronously, and then the above phase transition formula is used. Subsequently, h represents the equivalent amount captured from U or V to the same cell wall, where h is the product of the capture fraction and the corresponding post-phase transition state, with the capture fraction between 0 and 1; h is zero for A. All h is added equally to the released W, and the flow state after deducting h is used as the transport input: ; The superscript tr indicates the pre-transport state after release, phase change, and trapping, with X representing U, V, or A. All subsequent internal transport and boundary outputs use the tr state of X, no longer using the a state before trapping, thus avoiding the simultaneous retention of trapped material in both the flowing state and W. h is zero when there is no wall trapping.

[0044] In some embodiments, for a transportable phase p, the output share g is determined by its representative velocity v, cell length L, and substep duration. Droplet and gas phase velocities can be given separately; W does not directly transport with the mainstream. The internal destination share t and the exit share o leading to the target region are set according to cell connectivity, and their sum is 1. If the velocities are reversed, the connectivity direction is updated; if a reliable connectivity relationship cannot be established, model anomaly handling is initiated.

[0045] ; ; ; In the formula, X represents U, V, or A, respectively. The residual portion of the internal source unit is not added repeatedly in the summation. The mass of A output through the export share is added to the corresponding target region C; the equivalent of U and V output through the export share is added to Z. Closed units without internal exports or boundary exports take g as zero.

[0046] In some embodiments, phase transition, trapping, and transport are approximated by operator splitting within a physical time substep of the same length δ. After completing all sub-operations, the physical time only increases by δ, without adding the evaporation time, decomposition time, and complete path L / v. δ is reduced until the predicted arrival difference is below a set tolerance after the substep is halved; when encountering an excessively large CFL output share, the substep is automatically subdivided, without truncating g and missing transport. Cell partitioning coarsening or different sub-operation sequences should be checked for convergence using the same method.

[0047] When operating conditions change, the existing U, V, A, and W are advanced with new local temperatures, velocities, and diversion shares. The injection facts and the accumulated out-of-bounds quantities remain unchanged, but the remaining arrival distribution can change. Therefore, this method preserves historical material attribution and does not require the use of inaccurate complete arrival time kernels. Residual state corrections caused by new measurements are recorded in independent versions; measurements that can be attributed to a batch are only redistributed among conserved residual states and cannot regenerate unarrived materials from the accumulated C or Z.

[0048] It should be noted that by completing all sub-operations of phase change, capture, and transport within the same physical time sub-step, the physical time only increases by δ, achieving temporal consistency between phase change and transport, and avoiding the errors caused by simply adding the evaporation time, decomposition time, and complete path transport time. The phase change fraction is determined by the non-negative calibration rate and sub-step duration related to flue gas temperature, moisture content, droplet size, and residence state. The transport fraction is determined based on the representative transport velocity of the corresponding phase state, the computational cell length, and the sub-step duration, enabling state progression to adapt to fluctuations in flue gas temperature and velocity caused by changes in boiler load. Wall capture and re-release are treated as reversible processes, avoiding the direct cancellation of wall-retained material as irreversible loss.

[0049] In some embodiments, at tk After reading the official status, a read-only starting point and a prediction copy are established. The copy is advanced based on the predicted flue gas conditions. In each time slot, the newly added effective ammonia export volume of the historical batch is recorded and summed according to the target area to obtain B. C and Z in the prediction calculation are only copy variables and do not modify the official cumulative records. Materials still within the cell or wall at the end of the prediction period are retained as the prediction end state, and the observed partial arrival volume is not normalized to the whole batch.

[0050] ; In some embodiments, the control time domain contains N_c cycles, where N_c is at least 2; the prediction time domain contains N_p cycles. The decision variable is q. i (k+s|k) and a i (k+s|k), where s is from 0 to N_c 1. When the prediction time domain is longer than the control time domain, subsequent candidate inputs are generated according to a preset end-hold or end-feedback strategy. Each candidate batch is independently added to the prediction unit according to its planned injection interval, and its exit contribution is calculated using the same phase change transport rule.

[0051] ; ΔC represents the effective ammonia mass of the candidate batch in the s-th planned injection cycle reaching the target region j in the r-th predicted time slot; this term is zero when r is earlier than s. The candidate coefficients depend on the corresponding urea flow rate and atomizing gas operating point, and the particle size, phase change, and arrival process are recalculated for each candidate. B only includes executed batches, and Y only includes planned batches; the two do not overlap.

[0052] In some embodiments, when the device and model are suitable for linearization, the arrival responses of each planned batch can be shifted according to the injection time to form a prediction matrix and quadratic programming can be used; when using a nonlinear model, finite candidate set enumeration or successive linearization can be used, and feasibility can be verified on the original physical model. If convergence is not achieved, only safe and feasible solutions that have been re-verified in the current cycle are allowed to be used; feasible solutions from the previous cycle cannot be directly regarded as still feasible when constraints have changed.

[0053] It should be noted that by replicating the predicted state from the formal batch state, and using the predicted flue gas conditions to advance the predicted state, the historical batch's expected arrival amount B is formed. Predictive advancement does not change the formal batch state or the accumulated amount that has exceeded the limit, thus achieving physical isolation between predictive calculations and formal records. B only includes executed batches, and Y only includes planned batches; the two do not overlap, allowing for separate management of historical injection facts and planned injection amounts. The control time domain N_c is at least 2, and the prediction time domain N_p is not less than N_c. Multiple planning cycles share the prediction time domain requirements, avoiding the concentration of the entire time domain requirement in the current batch.

[0054] In some embodiments, the difference between the predicted inlet NOx mass flow rate and the target outlet NOx mass flow rate is converted into the required effective ammonia flow rate. K is the equivalent coefficient of ammonia mass to NOx mass, set according to the NOx metering benchmark and the target ammonia-nitrogen molar ratio; for example, when NOx is expressed as the equivalent mass of NO2 and each mole of NOx targets the consumption of one mole of NH3, K is 17 / 46, which cannot be mixed with 17 / 30 under the NO mass benchmark. Zonal demand is allocated according to the local NOx mass share β, where β is non-negative and sums to 1 over all target zones.

[0055] ; ; R describes the demand not yet covered by historical batches, and E describes the potential overcapacity caused by historical batches. During allocation, the original values ​​of B and D are retained, and the deviation of B plus Y is used to evaluate all candidate plans. The entire prediction time domain must not be fitted using only the current candidate batch, nor should the sum of R from each time slot be used as the injection target for the current cycle.

[0056] In some embodiments, the objective function J is constructed as follows: ; ; When s is zero, the previous control variable is the most recent reliable setpoint currently being executed; otherwise, the previous candidate value in the sequence is used. Weights are non-negative and include necessary dimensional scaling. J_end is used to suppress execution states where too much unfinished material remains at the end or cannot continue. It can be set based on the square of the excess amount of the remaining equivalent at the end relative to the allowable steady-state inventory, as well as the deviation of the end control variable. The prediction time domain should cover the main transport response; the wall tail remains in the end state and is included in the risk boundary, and cannot be deleted or pre-allocated for the sake of a limited time domain.

[0057] In some embodiments, hard constraints include a stable injection range, atomized gas range, allowable gas-liquid ratio or particle size range, minimum pyrolysis conditions for the prediction unit, and actuator limitations. i In the on / off state, z i When q is 0 i z is zero i When q is 1 i Between the minimum stable flow rate and the upper limit; when spraying stops, purge air may be retained as required by the equipment, and the gas-liquid ratio condition applies only to q. i A jetting state greater than zero is applied. The priority of normal rate of change constraints and emergency interlocks is predetermined by the equipment's permissible actions.

[0058] Spatial coverage is limited by the flue unit and outlet connectivity, and demand cannot be transferred to branches that cannot cover the target area. If necessary, a spatiotemporal excess upper bound is set on the sum of B and Y; if B alone exceeds this bound, an irreversible excess is recorded, and new inputs that would continue to increase the slot are restricted. A feasible solution cannot be created by reducing B or relaxing the record. Demand insufficiency can be represented by an explicit non-negative gap variable and is not considered evidence that hard physical constraints have been met.

[0059] It should be noted that by converting the difference between the predicted inlet NOx mass flow rate and the target outlet NOx mass flow rate into effective ammonia demand using the ammonia nitrogen equivalent coefficient K, and then allocating it to each target region according to the local NOx mass share β, the effective ammonia demand of each target region in each prediction time slot can be accurately generated. The objective function J simultaneously includes a mass deviation term, a control variable change penalty term, and a terminal residual state penalty term, balancing demand tracking accuracy, control stability, and prediction completeness while satisfying hard constraints. The prediction time domain covers the main transport response, and the wall long tail is still retained in the terminal state and included in the risk boundary, ensuring the completeness of the prediction.

[0060] In some embodiments, such as Figure 4 As shown, when the safe and feasible domain is not empty but demand cannot be fully tracked, a safe and feasible sequence is selected and the projected demand gap is output. Simultaneously, NOx deviation risk and coordination signals are given, without claiming that the original emission target can be maintained after reducing the ammonia supply target. When the safe and feasible domain is empty, reducing D is not used as a means to resolve hard conflicts. Instead, a shutdown, purging, or protected transition permitted by the equipment is executed according to the preset interlocking strategy, and a cause code and actual flow confirmation are output.

[0061] In some embodiments, before increasing urea, the atomized gas is brought to the allowable range of the corresponding operating point and confirmed; after decreasing urea, the atomized gas is maintained until the spray gun purging conditions are met. When ammonia escape reaches a preset threshold, the total positive increment is frozen, new inputs to relevant branches are tightened, and the entire sequence is re-verified; hysteresis and duration are resumed to prevent repeated switching. After the hard interlock is executed, the actual residual flow through the nozzle is still written into the batch according to the actual measurement, and the valve closing command cannot be used as the basis for immediate zero flow.

[0062] In some embodiments, when the flow signal is temporarily invalid, redundant cumulative quantity or quality balance estimation is preferred, and the injection quantity range and source are recorded. When reliable estimation cannot continue, a safe exit command is issued to the branch. During the waiting period for confirmation, the possible addition of material is still reserved according to the upper limit of the physical flow. When the injection quantity is determined to be an interval, each credible boundary scenario maintains its own non-negative conservation state. The oversupply safety check takes the credible high arrival scenario, and the supply shortage check takes the credible low arrival scenario. The maximum values ​​of different scenarios are not combined into a fictitious material state.

[0063] In some embodiments, after redundant or cumulative flow is restored, the missing measurement intervals are checked once, the versions before and after the check are saved, and duplicate entries into the same interval are prohibited. During inlet or outlet analyzer maintenance, the corresponding integral correction is frozen, and historical batch states are retained. When the controller restarts, it recovers from the persistent state and cumulative flow logs; if recovery is not possible and there may be undischarged materials, it is not immediately put into automatic operation with an empty ledger; instead, restricted operation and state reconstruction are implemented first, and wall stagnation requires separate confirmation.

[0064] It should be noted that by distinguishing between demand gaps and hard constraint conflicts, and when the safe feasible domain is empty, the system executes spray stoppage or protected transition according to preset interlock priorities, without automatically ignoring safety boundaries, thus ensuring system safety under extreme operating conditions. A multi-layered equipment protection mechanism is provided, including prohibiting urea increments when atomizing gas is insufficient, maintaining atomizing gas according to preset purging conditions after urea reduction, and freezing the total positive increment feedback when ammonia escape exceeds the threshold. The interval estimation and recovery verification mechanism for short-term invalid flow signals ensures that the possible range information of injected material is not lost during measurement interruptions, and traceability verification is possible after recovery, guaranteeing the continuity of system control under abnormal conditions.

[0065] This invention also provides a boiler urea direct injection selective catalytic reduction denitrification control system, including at least two independently adjustable urea injection zones, urea flow measurement and adjustment units for each zone, atomized gas measurement and adjustment units, urea mass fraction acquisition units, flue gas temperature and velocity acquisition units, nitrogen oxide acquisition units, a selective catalytic reduction reactor with at least two catalyst inlet target zones, and a controller communicatively connected to each unit. The controller's memory includes a fact storage area for storing the actual flow rate and injection time of the injection batch, a formal state storage area for storing the unconverted urea equivalent, decomposable precursor equivalent, generated ammonia mass, and batch excess accumulation in the flue gas calculation unit, and a predictive state storage area independent of the formal state storage area. The controller's processor is configured to generate multiple candidate control sequences for execution cycles according to the above control method, output only the current cycle setpoint, update the batch phase change transport state and excess accumulation with the actual injection flow rate and measured flue gas conditions at the end of the cycle, cancel the excess state components, and update the formal state and relative time slot reference for the next cycle.

[0066] In some embodiments, the controller further includes a quality flag and a status recovery storage area. During periods when actual flow measurement is invalid, it stores the injection volume range determined based on redundant measurements or the physical boundaries of the actuator, and includes any newly added material within this range in the status uncertainty. If the preset allowable duration is exceeded, a stop-spray or protected transition command is output to the corresponding branch. When the accumulated flow is restored and verifiable, a one-time verification of the batches during the aforementioned period is performed. Reproducible computational examples The following values ​​are examples constructed to verify the calculation rules and do not represent field tests, actual spray gun calibration, or performance guarantees. The given shares, grids, and requirements are sufficient to reproduce the corresponding calculations; different examples verify material conditions, optimization sequences, and safe exits respectively, and do not treat the idealized arrival kernel as a real flue full-load model.

[0067] Example 1: State-based propagation under changing operating conditions In some embodiments, the theoretical ammonia equivalent P for the same batch is set to 0.100 kg, with two adjacent flue units, unit 2 directly connected to a target area. Initially, unit 1 has U=0.080, V=0.020, and A=0, while unit 2 has zero, W=0, and C=Z=0. δ is set to 1 second, and the evaporation and decomposition fractions in the first sub-step are both 0.5, meaning both rates are ln2 per second; the output fraction of both units is 0.5, and the internal or outlet destination is unique. Phase change occurs before transport, and physical time advances by only 1 second.

[0068]

[0069] All material quantities in the table are expressed in kilograms of ammonia or kilograms of theoretical ammonia equivalent.

[0070] In the first substep, the last two units all have U=0.020, V=0.015, and A=0.015, with no material overflowing. In the second substep, the simulated flue gas conditions change, and the evaporation fraction decreases to 0.25, i.e., κ. e = The decomposition rate is 0.75 seconds, the decomposition share remains at 0.5, and the output share increases to 1. After the phase change, each unit has U=0.015, V=0.010, and A=0.025. After transport, unit 1 is empty, unit 2 retains the above state, the newly added effective ammonia over-limit amount is 0.025, and the newly added unconverted over-limit equivalent is 0.025.

[0071]

[0072] If the second substep uses the full amount from the first substep, then the predicted effective ammonia overrun is 0.013750 kg. The difference between the two results is used to illustrate how the remaining state is affected by subsequent conditions; the update does not change P and does not regenerate material from the overrun record. This example is not used to extrapolate the numerical relationship between actual temperature and rate.

[0073] Example 2: Multi-period allocation of two regions In some embodiments, to individually verify the control sequence, the ideal limit of completed phase transition and total effective conversion of 1 is taken, with each batch of effective ammonia reaching half in the injection cycle and half in the next cycle. The first row of the spatial share matrix H is 0.75, 0.25, and the second row is 0.25, 0.75. The cycle Δ is 10 seconds, and the urea mass fraction is 0.50. Using x... iq represents the effective ammonia kilograms of a new batch in one cycle. i =21600x i / 17 kg per hour.

[0074] For each independent candidate period in each region, x takes the set {0, 0.020, 0.040, 0.060}, with a total of two independent candidate periods s=0 and s=1. Enumerate 4 4 =256 sequences. The prediction time slots r are 0, 1, and 2; the planned input required for slot 2 is maintained according to x(2)=x(1). Assume that when x is greater than zero, a=q / 2, and the spraying is handled according to the purging mode when the spraying stops; in this example, all candidate operating points are assumed to be within the calibrated operating domain. The urea range is zero or 25.411765 to 76.235294 kg / h, the upper limit of the ordinary single cycle change is 76.235294 kg / h, and the excess allowable amount is 0.005 kg / slot per zone.

[0075] The weight of each slot's quality deviation is set to 1, and the action penalty is 0.01 times the sum of squares of adjacent x changes in units of x variable. The predicted end only retains half of the input, and J_end is zero when the total amount is no greater than 0.060 kg; otherwise, the sequence is excluded. In this example, the arrival amounts of the three slots are: B(0) = 0.5H·x_prev, B(1) = B(2) = 0; Y(0) = 0.5H·x(0), Y(1) = 0.5H·[x(0) + x(1)], Y(2) = H·x(1). The last equation includes the contribution of slot 2 retaining the input, and the other half that has not yet arrived is still the predicted end state.

[0076] Let the previous actual batch x_prev = (0.040, 0.020), and the requirements of the three predicted tanks are all (0.035, 0.025). Enumeration yields x(0) = x(1) = (0.040, 0.020), with both deviation and action penalty being zero. Its current urea setting is (50.823529, 25.411765) kg / h, and the atomizing gas setting is (25.411765, 12.705882) kg / h; these values ​​are converted from the assumed gas-liquid relationship in this example and are not claimed to be from the actual spray gun calibration table.

[0077] Time slot history B, two-zone candidate Y, two-zone demand D, two zones

[0078] The arrival and demand figures in the table are in kilograms of ammonia, and the numbers within the area are arranged in the order of target area 1, target area 2, etc.

[0079] In cycle k, only x(0) is executed. If the actual execution equals the above value, then at the end of the cycle, half of the remaining historical batch and the first half of the current actual batch will both go out of bounds; the formal state will only retain the other half of the current actual batch. In the next cycle, B(0) will still be (0.017500, 0.012500), and the remaining slots will be zero. Candidate x(1) and its extension will never enter the formal state, and will be replanned in the next cycle without duplicate entries.

[0080] Let the current slot demand remain (0.035, 0.025), and the latter two slot demands become (0.025, 0.035), with other conditions unchanged. One of the optimal sequences is enumerated as x(0) = x(1) = (0.020, 0.040); another equally optimal sequence is x(0) = (0.040, 0.020) and x(1) = (0.020, 0.040). In this example, the first priority to reach the same objective function value is (0.030, 0.030), and the latter two slots are (0.025, 0.035). Currently, there is a 0.005 kg shortfall in target area 1 and a 0.005 kg excess in target area 2, which is within the allowable excess range in this example; the objective function is 0.000058. The controller outputs the shortfall of the selected sequence, without interpreting it as a necessary shortfall common to all optimal sequences. After actual execution, if the demand for all three slots in the next cycle is (0.025, 0.035), the sequence will be maintained and the deviation will return to zero.

[0081] To implement error checking, let the demand per slot in a single zone be 1, and the time share be (0.5, 0.5). If only the current batch x participates in the two-slot fitting, the optimal recursion is x=2. 0.5x_prev, its stable value is 4 / 3. Including the current and next batches, in the steady state of x_prev=1, x(0)=x(1)=1 makes both slots reach 1. This comparison is used to detect program errors that omit planned inputs.

[0082] Example 3: Hard Constraint Conflicts and Safe Exit Assuming the previous actual flow rate was 80 kg / h and the normal decrease limit was 10 kg / h, the normal constraint requires a minimum flow rate of 70 kg / h in this cycle. If the safety allowable upper limit becomes 50 kg / h, the feasible domain of the normal constraint is empty. Reducing demand D does not change this conclusion. In this example, it is assumed that the equipment allows emergency stop spraying to take precedence over the normal decrease limit. The controller issues a stop spraying order and maintains the specified purge gas, waiting for confirmation of the actual flow rate decrease. The actual throughput during valve closure is still recorded as an integral. If the equipment does not allow immediate stop spraying, its preset protected transition is executed, and constraints that cannot be simultaneously met are reported, without automatically ignoring safety boundaries.

[0083] It should be noted that the above calculation examples respectively tested three typical scenarios: state progression under changing operating conditions, multi-cycle allocation of two regions, and hard constraint conflict and safe exit. Example 1 verified the conservation progression algorithm and material balance check, showing that the remaining state is updated by subsequent conditions without changing P, and materials are not regenerated from the out-of-bounds records. Example 2 verified the superposition of multi-cycle candidate sequences and constraint solving, showing that candidate x(1) and its maintenance extension never enter the formal state, and are re-planned in the next cycle without duplicate entries. Example 3 verified the safety interlocking mechanism, showing that when there is a hard constraint conflict, safe exit is executed according to the preset interlocking priority, and the safety boundary is not ignored.

[0084] In the above embodiments, the spray gun working domain at least provides the mapping nodes and errors between urea concentration, gas-liquid ratio, pressure difference, and particle size; the phase change model at least provides κ e κ d The mapping, interpolation rules, and effective boundaries with temperature, humidity, particle size, and material state should be defined; the transport model should at least provide unit length, velocity, phase slip, and flow splitting relationships. Equipment matching can be performed after pre-calculation using a publicly available mechanistic model, or bench and field measurement calibration can be used. Before putting the system into control, arrival distribution, material balance, and sampling step size errors should be checked under independent operating conditions.

[0085] To reproduce a complete device implementation beyond the example, the controller configuration file stores all operating domains, sampling periods, substep subdivision limits, prediction and control time domains, end-point strategies, action weights, overload limits, fault hold-up times, purge times, and interlock priorities. When device parameters are missing, they are not extrapolated from example figures to the actual setpoints.

[0086] In some embodiments, the material state can adopt more particle size classes or finer urea reaction components, and be converted according to a unified theoretical ammonia equivalent; the flue gas calculation unit can adopt a segmented network or a three-dimensional finite volume unit. As long as the actual injection facts of each batch, the non-negative conservation of remaining materials, the distinction between the formal state and the predicted copy, the superposition of the arrival amounts of multiple candidate batches, and the actual write-back after the first step are retained, the specific discrete structure and constraint solver can be replaced.

[0087] It should be noted that the above parameter acquisition methods specify the minimum parameter requirements for the spray gun working domain, phase change model, and transport model, and clarify that these parameters can be obtained through pre-calculation using publicly available mechanistic models or through bench and field measurement calibration. Alternative implementation methods explicitly state that as long as the actual injection facts of each batch, the non-negative conservation propagation of remaining materials, the distinction between the formal state and the predicted copy, the superposition of arrival amounts of multiple candidate batches, and the actual write-back after the first step are retained, the specific discrete structure and constraint solver can be replaced, providing support for a broad protection scope.

[0088] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0089] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0090] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0091] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0092] In this invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0093] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A boiler urea direct injection selective catalytic reduction De-NOx control method, applied to a De-NOx system with at least two independently adjustable urea injection zones and at least two catalyst inlet target zones, executed by a controller according to a control period, characterized in that, include: S1. Divide the flue from the reference section of each nozzle to the reference surface of the catalyst inlet into interconnected calculation units, and obtain the flue gas temperature and transport velocity of each calculation unit, as well as the actual flow rate of urea solution, urea mass fraction and atomized gas state of each injection area. S2. At the beginning of the control cycle, read the formal batch status up to the end of the previous control cycle. For each injected batch, associate and save the batch identifier, injection area, injection time, and theoretical ammonia equivalent calculated from the actual mass of pure urea passing through the nozzle reference section. In the calculation unit, record the unconverted urea state, decomposable precursor state, and ammonia generated state measured by theoretical ammonia equivalent, respectively. S3. The single-step advancement of the formal batch state and the predicted state shall follow the following conservation rules: Based on the flue gas temperature, transport rate and batch atomization formation conditions of each calculation unit, the transformation between the above states and the transport between calculation units shall be advanced within the same physical time substep. The effective ammonia mass that crosses the catalyst inlet reference surface in the form of ammonia and the urea equivalent that does not cross the reference surface in the form of ammonia shall be accumulated and recorded. The sum of the total amount of the remaining state of the batch and the accumulated amount that has exceeded the limit shall be kept as the theoretical ammonia equivalent of the batch, and the accumulated excess portion shall be removed from the remaining state. S4. The predicted state is copied from the formal batch state, and the predicted flue gas conditions are used to advance the predicted state to form the expected effective ammonia arrival amount of the injected batch in each target area and each predicted time slot. The prediction advancement does not change the formal batch state or the accumulated amount that has exceeded the limit. S5. Based on the predicted nitrogen oxide removal mass and ammonia nitrogen equivalent relationship, generate the effective ammonia demand of each target area in each predicted time slot. Construct a control sequence with multiple candidate urea flow rates and candidate atomizing gas setpoints including the current control cycle and at least one subsequent control cycle. Predict the phase change transport state of each candidate batch. Superimpose the arrival amount generated by each candidate batch according to its planned injection time with the expected arrival amount of the injected batch to obtain a control sequence that satisfies the constraints of injection branch flow rate, atomizing gas working range and pyrolysis conditions. S6. Only the current control cycle setting value of the control sequence is sent to the actuators of each injection area, and the unexecuted candidate batch is not written into the formal batch status; S7. At the end of the current control cycle, advance the original batch and the newly injected batch of the cycle according to the measured flue gas conditions and actual nozzle throughput of the cycle, update the cumulative amount of out-of-bounds and the remaining status, and then submit the updated results as the official batch status of the next control cycle.

2. The control method according to claim 1, characterized by, In step S2, multiple sampling intervals within the control cycle are treated as injection sub-batches. The mass of pure urea is determined by the integral of the product of the actual mass flow rate of the urea solution and the mass fraction of urea within each sampling interval, and then multiplied by 34 / 60 to obtain the theoretical ammonia equivalent. When there is a variable tube storage between the flow meter and the nozzle reference section, the actual flow rate through the nozzle is calculated based on the actual flow meter measurement and the change in tube storage. The execution model for the instruction to the nozzle and the phase change transport model from the nozzle to the catalyst inlet are set respectively.

3. The control method according to claim 1, characterized by, In step S3, the transfer fraction from the unconverted urea state to the decomposable precursor state and the transfer fraction from the decomposable precursor state to the ammonia-generated state are determined by the non-negative calibration rate and substep duration related to flue gas temperature, moisture content, droplet size, and residence state, respectively, with each transfer fraction ranging from 0 to 1. The transport fraction between computational units is determined based on the representative transport rate of the corresponding phase, the computational unit length, and the substep duration. The substep duration is selected such that the output fraction of each unit is no greater than 1, and the sum of the diversion fractions of each target unit and the target region is 1.

4. The control method according to claim 3, characterized by, In step S3, the re-released wall retention state, measured in theoretical ammonia equivalent, is also recorded for each batch. The urea equivalent captured by the wall in the unit is transferred from the corresponding flow state to the wall retention state. The wall release amount re-enters the decomposable precursor state of the same batch. The wall retention amount that has not yet been released is retained in the total remaining state of the batch and is not cancelled as an out-of-bounds amount.

5. The control method according to claim 1, characterized by, In step S1, the connectivity between the computing units and the outlet relationship leading to the catalyst inlet target area are determined by the calibrated flow field model, tracer response, or partitioned disturbance response. When the operating conditions change, the remaining material state of each batch is advanced according to the updated unit flue gas state and diversion share, while retaining its injection fact and the accumulated amount that has exceeded the limit; the injection batch of at least one injection area can enter the two target areas through different paths and form an effective ammonia arrival amount in two different predicted time slots.

6. The control method according to claim 1, characterized by In step S5, the controller sets the control time domain and the prediction time domain respectively. Outside the control time domain but within the prediction time domain, the candidate control sequence is supplemented according to the preset end-hold strategy or end-control strategy. The deviation between the arrival amount of historical batches and the sum of the arrival amounts of all candidate batches relative to the effective ammonia demand and the change penalty of adjacent candidate control quantities are calculated. The candidate materials that remain in the calculation unit or wall at the end of the prediction are retained as the prediction end residual state and used for end constraints or end penalties.

7. The control method according to claim 1, characterized by In step S7, when obtaining intermediate measurement results that can be attributed to a specific batch, the mass is redistributed among the remaining material states while keeping the theoretical ammonia equivalent injected into the batch and the cumulative amount that has exceeded the limit unchanged. After correction, each remaining state is non-negative and its sum remains unchanged. Total export measurements that cannot be attributed to a specific batch are used for bounded correction of common model parameters, demand deviations, or state uncertainties, and the correction does not reintroduce out-of-bounds mass into the remaining state.

8. The control method according to claim 1, characterized by, In step S5, the stop-spray state and stable spray state of the injection branch are set respectively, so that the urea flow rate is zero or between the minimum stable flow rate and the allowable upper limit; when a safe and feasible control sequence exists but the effective ammonia demand cannot be fully met, a safe and feasible sequence is output and the demand gap is recorded; when the safety constraint and the normal rate of change constraint conflict, the stop-spray or protected transition allowed by the equipment is executed according to the pre-set interlock priority, a fault flag is issued and the actual flow rate is checked; when the atomizing gas is insufficient, the urea increment is prohibited, and after the urea is reduced, the atomizing gas is maintained according to the preset purging conditions.

9. A boiler urea direct injection selective catalytic reduction denitration control system comprising at least two independently adjustable urea injection zones, a urea flow measurement and adjustment unit for each zone, an atomizing gas measurement and adjustment unit, a urea mass fraction acquisition unit, a flue gas temperature and velocity acquisition unit, a nitrogen oxide acquisition unit, a selective catalytic reduction reactor having at least two catalyst inlet target zones, and a controller in communication with each unit, characterized in that, The controller's memory includes a fact storage area for storing the actual flow rate and injection time of the nozzle according to the injection batch, a formal state storage area for storing the unconverted urea equivalent, decomposable precursor equivalent, generated ammonia mass, and batch excess cumulative amount in the flue gas calculation unit, and a predictive state storage area independent of the formal state storage area; the controller's processor is configured to generate multiple candidate control sequences for execution cycles according to the control method of any one of claims 1 to 8, output only the current cycle setpoint, update the batch phase change transport state and excess cumulative amount with the actual nozzle flow rate and measured flue gas conditions at the end of the cycle, cancel the excess state components, and update the formal state and relative time slot reference for the next cycle.

10. The control system of claim 9, wherein, The controller also includes a quality flag and a status recovery storage area. During periods when the actual flow measurement is invalid, it saves the injection volume range determined based on redundant measurement or the physical boundary of the actuator, and includes any new materials that may be added within this range in the status uncertainty. When the preset allowable duration is exceeded, it outputs a stop spray or protection transition command to the corresponding branch. When the cumulative flow is restored and verifiable, it performs a one-time verification of the batches during the period, retains the verified version, and prohibits the repeated writing of the injection volume of the same sampling interval.