Denitration Intelligent Control Method and System Based on PID and Inverted Transformer

Abstract

The present invention discloses a denitration intelligent control method and system based on PID and inverse Transformer. The method includes: collecting the operation parameter sequence of the SCR denitration system in real time; swapping the input time series and feature dimension in the Transformer model to form an inverse Transformer model, obtaining the predicted outlet NOx concentration, calculating the adjustment value of the ammonia injection amount according to the deviation between it and the target outlet NOx concentration, and adjusting the opening degree of the ammonia injection valve in real time, and online adjusting the inverse Transformer model based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration. The present invention is applied to the field of denitration control, and can more accurately predict the change trend of NOx concentration in the future period of time, so as to adjust the ammonia injection amount in advance, effectively overcoming the lag defect of the original PID controller.

Description

Technical Field

[0001] The present invention relates to the technical field of denitration control, and specifically to a denitration intelligent control method and system based on PID and inverse Transformer. Background Art

[0002] Selective Catalytic Reduction (SCR) denitration technology is the mainstream solution for reducing nitrogen oxide (NOx) emissions in industrial flue gas treatment. Its principle is that under the action of a catalyst (such as V2O5-WO3 / TiO2), ammonia (NH3) is injected into the flue as a reducing agent to convert NOx into harmless N2 and H2O. However, in actual operation, the precise control of ammonia injection volume faces many problems. For example, from ammonia injection to NOx concentration detection, it needs to go through links such as mixing, catalytic reaction, and sensor response, and the total lag time reaches 30 - 60 seconds. Traditional feedback control is prone to overshoot or oscillation due to the lag effect.

[0003] Traditional PID controllers that rely on error integral and differential feedback are prone to integral saturation and phase delay in lag systems. Although Long Short-Term Memory (LSTM) and Recurrent Neural Network (RNN) can capture temporal dependencies, their gradient vanishing / exploding problems lead to the accumulation of prediction errors for ultra-long sequences (>60 seconds). In addition, although the self-attention mechanism of the Transformer architecture can model global dependencies, fixed positional encoding is difficult to adapt to the dynamic temporal offsets of the SCR system, and multi-head attention is prone to ignoring local key features (such as the transient response of mutation working conditions) in long sequences. Summary of the Invention

[0004] Aiming at the problem that it is difficult to accurately regulate the ammonia injection volume in the existing SCR denitration process, the present invention provides a denitration intelligent control method and system based on PID and inverse Transformer, which can more accurately predict the change trend of NOx concentration in the next period of time, so as to adjust the ammonia injection volume in advance, thus effectively overcoming the lag defect of the original PID controller.

[0005] To achieve the above object, the present invention provides a denitration intelligent control method based on PID and inverse Transformer, including the following steps:

[0006] Step 1, collect the operation parameter sequence of the SCR denitration system in real time and preprocess it;

[0007] Step 2: Swap the time series and feature dimensions of the input in the Transformer model to form an inverted Transformer model, and input the sequence of operating parameters into the inverted Transformer model to obtain the predicted outlet NOx concentration within the prediction time window;

[0008] Step 3: Based on the deviation between the predicted outlet NOx concentration and the target outlet NOx concentration, calculate the adjustment value of the ammonia injection amount based on a PID controller;

[0009] Step 4: Based on the adjustment value of the ammonia injection amount, adjust the opening degree of the ammonia injection valve in real time, collect the actual outlet NOx concentration, and perform online adjustment on the inverted Transformer model and the PID controller based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration.

[0010] In one embodiment, the sequence of operating parameters has multiple feature data sequences, including but not limited to the inlet NOx concentration, the outlet NOx concentration, the ammonia slip rate, the reactor temperature, and the opening degree of the ammonia injection valve.

[0011] In one embodiment, in Step 1, the preprocessing includes operating on each of the feature data sequences:

[0012] First, calculate the mean μ and standard deviation σ of the feature data in the feature data sequence, and mark the feature data that satisfies as outliers;

[0013] Interpolate and replace the outliers in the feature data sequence, and perform normalization processing on the interpolated feature data sequence, which is:

[0014] ;

[0015] where, is the feature data corresponding to the feature data after normalization processing.

[0016] In one embodiment, the loss function of the inverted Transformer model in the training stage is:

[0017] ;

[0018] where, is the weight coefficient in the th round of iteration during the training process of the inverted Transformer model;

[0019] Represents the Huber loss, specifically:

[0020] ;

[0021] ;

[0022] where, is the number of samples in a training batch of the reverse Transformer model, is the -th sample's corresponding Huber loss in the batch, is the predicted value of the outlet NOx concentration for the -th sample in the batch, is the true value of the outlet NOx concentration for the -th sample in the batch, is a hyperparameter;

[0023] Represents the MSE loss, specifically:

[0024] ;

[0025] ;

[0026] where, is the MSE loss corresponding to the -th sample in the batch.

[0027] In one embodiment, the weight coefficient is specifically:

[0028] ;

[0029] where, , are the initial weight and the final weight respectively, and , is the decay coefficient, is the number of training epochs.

[0030] In one embodiment, in step 3, the ammonia injection amount adjustment value is specifically:

[0031] ;

[0032] ;

[0033] where, is the ammonia injection amount adjustment value, is the deviation at the current moment, is the target outlet NOx concentration, To improve the predicted outlet NOx concentration for the next T seconds output by the Transformer model, For the deviation accumulation from the initial moment to the current moment, For the adopted interval, , , They are the proportional gain, integral gain, and derivative gain respectively.

[0034] In one embodiment, in step 4, the specific operation of adjusting the ammonia injection valve opening in real time based on the ammonia injection amount adjustment value is as follows:

[0035] If , then increase the opening of the ammonia injection valve, and the adjustment amount is , where is the maximum valve opening, is the adjustment coefficient;

[0036] If , then decrease the opening of the ammonia injection valve, and the adjustment amount is ;

[0037] If , then keep the current opening of the ammonia injection valve unchanged.

[0038] In one embodiment, in step 4, the online adjustment of the inverse Transformer model and the PID controller specifically includes:

[0039] Step 401, set a monitoring time window, and calculate the mean absolute error between the actual outlet NOx concentration and the predicted outlet NOx concentration within each monitoring time window;

[0040] Step 402, determine whether the mean absolute error of N consecutive monitoring time windows exceeds the set threshold:

[0041] If so, proceed to step 403;

[0042] Otherwise, wait to calculate the mean absolute error of the next monitoring time window and then perform step 402 again;

[0043] Step 403, after updating the learning rate of the inverse Transformer model, calculate the mean absolute error of the next monitoring time window, and determine whether it is less than the set threshold:

[0044] If so, save the learning rate of the inverse Transformer model and then proceed to step 404;

[0045] Otherwise, perform step 403 again;

[0046] Step 404: Dynamically adjust the proportional gain and integral time of the PID controller based on the mean absolute error of the current monitoring time window and the mean absolute error of the previous monitoring time window.

[0047] In one embodiment, step 404 is specifically as follows:

[0048] ;

[0049] ;

[0050] Wherein, and are the adjusted proportional gain and integral time, and are the proportional gain and integral time before adjustment, is the adjustment coefficient, is the mean absolute error of the current monitoring time window, is the mean absolute error of the previous monitoring time window.

[0051] To achieve the above object, the present invention also provides a denitration intelligent control system based on PID and inverse Transformer, which performs denitration intelligent control by using the above method. The denitration intelligent control system includes:

[0052] A data acquisition unit for real-time collecting the operation parameter sequence of the SCR denitration system and preprocessing it;

[0053] A data prediction unit for swapping the time series and feature dimensions of the input in the Transformer model to form an inverse Transformer model, and inputting the operation parameter sequence into the inverse Transformer model to obtain the predicted outlet NOx concentration within the prediction time window;

[0054] A PID calculation unit for calculating the ammonia injection amount adjustment value based on the PID controller according to the deviation between the predicted outlet NOx concentration and the target outlet NOx concentration;

[0055] An online adjustment unit for adjusting the opening degree of the ammonia injection valve in real time according to the ammonia injection amount adjustment value, collecting the actual outlet NOx concentration, and performing online adjustment on the inverse Transformer model and the PID controller based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration.

[0056] Compared with the prior art, the present invention has the following beneficial technical effects:

[0057] In the present invention, by swapping the input time series and the feature dimension in the Transformer model and transferring the self-attention mechanism from the time dimension to the feature dimension, the feature information in the long sequence can be better captured, and thus the change trend of NOx concentration in the future for a period of time can be predicted more accurately. Accordingly, the PID controller is coordinated to adjust the ammonia injection amount in advance, thereby effectively overcoming the lag defect of the original PID controller. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the structures shown in these drawings without creative efforts.

[0059] Figure 1 It is a flowchart of the denitration intelligent control method based on PID and inverted Transformer in Embodiment 1 of the present invention;

[0060] Figure 2 It is a flowchart for online adjustment of the inverted Transformer model and the PID controller in Embodiment 1 of the present invention;

[0061] Figure 3 It is a structural block diagram of the denitration intelligent control based on PID and inverted Transformer in Embodiment 2 of the present invention;

[0062] Figure 4 It is a structural block diagram of the terminal device in Embodiment 3 of the present invention.

[0063] The realization, functional characteristics, and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0065] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0066] Example 1

[0067] As Figure 1 shown, a denitration intelligent control method based on PID and inverse Transformer disclosed in this embodiment mainly includes the following steps:

[0068] Step 1: Real-time collect the operation parameter sequence of the SCR denitration system and preprocess it;

[0069] Step 2: Swap the time series and feature dimension of the input in the Transformer model to form an inverse Transformer model, and input the operation parameter sequence into the inverse Transformer model to obtain the predicted outlet NOx concentration within the prediction time window, where NOx refers to nitrogen oxides such as nitric oxide, nitrogen dioxide, and nitrous oxide;

[0070] Step 3: Calculate the ammonia injection amount adjustment value based on the PID controller according to the deviation between the predicted outlet NOx concentration and the target outlet NOx concentration;

[0071] Step 4: Based on the ammonia injection amount adjustment value, adjust the opening of the ammonia injection valve in real time, collect the actual outlet NOx concentration, and perform online adjustment on the inverse Transformer model and the PID controller based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration.

[0072] In this embodiment, the operation parameter sequence has multiple characteristic data sequences, including inlet NOx concentration, outlet NOx concentration, ammonia slip rate, reactor temperature, ammonia injection valve opening, etc., and the sampling frequency can be set to 1 second, 2 seconds, etc. It should be noted that in the specific application process, the characteristic data sequence is not limited to including the aforementioned characteristic data, and can also include other characteristic data such as the amount of coal input into the furnace, component, and flue gas volume.

[0073] In the specific implementation process of Step 1, the preprocessing of the operation parameter sequence includes removing outliers, supplementing missing points, and normalizing each characteristic data sequence in the operation parameter sequence. The specific implementation process is as follows:

[0074] First, use box plots, Z-score methods or other methods to identify outliers in the characteristic data sequence and eliminate extreme values caused by noise or sensor errors. For example, calculate the mean μ and standard deviation σ of the characteristic data in the characteristic data sequence, and mark the characteristic data in the characteristic data sequence that satisfies as outliers;

[0075] Then, perform interpolation replacement on the outliers in the characteristic data sequence to ensure the continuity of the characteristic data sequence and avoid misjudging the periodic law of the model due to data loss;​

[0076] Finally, normalize the interpolated characteristic data sequence to improve the convergence speed of the data during optimization processes such as gradient descent. The normalization process is as follows:

[0077] ;

[0078] where is the characteristic data The corresponding characteristic data after normalization processing.

[0079] In the specific implementation process of step 2, in this embodiment, by swapping the input time series and the feature dimension in the Transformer model, an inverted Transformer model is formed, enabling the Transformer to transfer the self-attention mechanism from the time dimension to the feature dimension, so as to better capture the feature information in the long sequence, and then more accurately predict the change trend of NOx concentration in the future for a period of time.

[0080] Specifically, the input structure of the original Transformer model is (B, T, D), where B is the batch size, T is the number of time steps, and D is the feature dimension. In this embodiment, the input structure of the inverted Transformer model is (B, D, T), that is, the input time series and the feature dimension are swapped. For example, in the SCR denitration system, the operating parameter sequences are respectively selected as valve opening, ammonia concentration, inlet NOx concentration, outlet NOx concentration, and reactor temperature. The original Transformer model analyzes the dependence relationship of the time dimension T through the self-attention mechanism (such as the correlation of NOx concentration at previous and subsequent moments), but the interaction modeling of the feature dimension D (such as the relationship between multiple parameters such as valve opening and temperature) is insufficient. In the SCR denitration system, the outlet NOx concentration is affected by the cross-feature synergy of multiple operating parameters (for example, ammonia concentration and reactor temperature jointly determine the reaction efficiency). Relying only on time attention may ignore the complex coupling relationship between features. In the inverted Transformer model in this embodiment, the focus of the self-attention mechanism is transferred from the time dimension T to the feature dimension D, and each feature is regarded as an independent sequence. The model focuses on analyzing the global association between different features (such as how the valve opening affects ammonia concentration and temperature), rather than the time fluctuation of a single feature, so as to more accurately predict the change trend of NOx concentration in the future for a period of time.

[0081] In the specific application process, the construction process of the inverted Transformer model is as follows: First, use the tensor transpose operation to convert the original input (B, T, D) to (B, D, T); Second, retain the encoder-decoder structure of the Transformer, but adjust the multi-head attention mechanism to adapt to the inverted input dimension.

[0082] In this embodiment, the inverse Transformer model can be either pre-trained offline first and then trained online, or directly trained online. For example, offline training uses historical data to train the model, and online training can be carried out in the form of updating sliding window data. In this embodiment, the loss function in the training stage of the inverse Transformer model is jointly constructed by combining the Huber loss and the MSE loss , specifically:

[0083] ;

[0084] Among them, is the weight coefficient in the -th round of iteration in the training process of the inverse Transformer model;

[0085] represents the Huber loss, specifically:

[0086] ;

[0087] ;

[0088] Among them, is the number of samples included in a training batch of the inverse Transformer model, is the Huber loss corresponding to the -th sample in the batch, is the predicted value of the outlet NOx concentration corresponding to the -th sample in the batch, is the true value of the outlet NOx concentration corresponding to the -th sample in the batch, is a hyperparameter used to control the transition point of the loss function from quadratic to linear, for example, it can be set to;

[0089] represents the MSE loss, specifically:

[0090] ;

[0091] ;

[0092] Among them, is the MSE loss corresponding to the -th sample in the batch.

[0093] The Huber loss has strong robustness to outliers (such as sensor noise and sudden changes in operating conditions). When the absolute value of the prediction error exceeds the hyperparameter When the loss function degenerates into linear growth, it can effectively avoid abnormal samples from dominating the gradient update. The MSE loss is sensitive to small deviations in normal data and can enhance the fitting accuracy of the model for steady-state operating conditions through the square term. Therefore, in this embodiment, the loss function in the training stage of the inversion Transformer model is constructed by combining the Huber loss and the MSE loss , which can effectively balance the robustness and accuracy of the outlet NOx concentration.

[0094] As a preferred embodiment, this embodiment also discloses a method for dynamically adjusting the weight coefficient in the loss function. The specific implementation process is as follows:

[0095] ;

[0096] Among them, , are the initial weight and the final weight respectively, and , is the attenuation coefficient used to control the transition smoothness. For example, it can be set to , is the number of training epochs. In the specific application process, it can be set to , , that is, at the beginning of training, a high weight is given to the Huber Loss to suppress the interference of outliers. As the training progresses, it gradually transitions to being dominated by the MSE to improve the fitting accuracy of the model for conventional data.

[0097] In this embodiment, the process of the PID controller calculating the ammonia injection amount adjustment value is as follows:

[0098] ;

[0099] ;

[0100] Among them, is the ammonia injection amount adjustment value, is the deviation at the current moment, is the target outlet NOx concentration, is the predicted outlet NOx concentration of the future T seconds output by the improved Transformer model, is the cumulative deviation from the initial moment to the current moment, is the adopted interval, , , are the proportional gain, integral gain, and derivative gain respectively.

[0101] After calculating the ammonia injection amount adjustment value , the opening of the ammonia injection valve can be adjusted in real time based on the ammonia injection amount adjustment value. The specific implementation process is as follows:

[0102] If , increase the opening of the ammonia injection valve, and the adjustment amount is , where is the maximum opening of the valve, is the adjustment coefficient. For example, it can be set to ;

[0103] If , decrease the opening of the ammonia injection valve, and the adjustment amount is ;

[0104] If , keep the current opening of the ammonia injection valve unchanged.

[0105] In the specific implementation process of step 4, in this embodiment, the inversion Transformer model and the PID controller are adjusted online through the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration, so as to form a continuously optimized closed-loop process. Refer to Figure 2 , the specific implementation process of adjusting the inversion Transformer model and the PID controller online includes the following steps:

[0106] Step 401, set the monitoring time window, and calculate the mean absolute error between the actual outlet NOx concentration and the predicted outlet NOx concentration within each monitoring time window, where the monitoring time window can be set to 10 min to 30 min;

[0107] Step 402, determine whether the mean absolute error of N consecutive (for example, 3) monitoring time windows exceeds the set threshold:

[0108] If so, go to step 403;

[0109] Otherwise, wait to calculate the mean absolute error of the next monitoring time window and then go back to step 402;

[0110] Step 403, after updating the learning rate of the inversion Transformer model (for example, reducing the learning rate to 10% of the original value, etc.), calculate the mean absolute error of the next monitoring time window, and determine whether it is less than the set threshold:

[0111] If so, save the learning rate of the inversion Transformer model and then go to step 404;

[0112] Otherwise, go back to step 403;

[0113] Step 404: Dynamically adjust the proportional gain and integral time of the PID controller based on the mean absolute error of the current monitoring time window and that of the previous monitoring time window, so that the PID controller always matches the current working condition. For example, when the model prediction error decreases, increasing the proportional gain can speed up the response speed, and shortening the integral time can enhance the integral effect and accelerate the elimination of the steady-state error. Specifically:

[0114] ;

[0115] ;

[0116] where, and are the adjusted proportional gain and integral time, and are the proportional gain and integral time before adjustment, is the mean absolute error of the current monitoring time window, is the mean absolute error of the previous monitoring time window, is an adjustment coefficient to prevent parameter mutation, and can be set to, for example, .

[0117] In the specific implementation process of step 403, it is not limited to adjusting by updating the learning rate. It is also possible to choose to update the form of the running parameter sequence time window. For example, on the basis of the existing running parameter sequence time window, add 2 hours or shorten it by 1 hour, etc. It is also possible to freeze the embedding layer of the improved Transformer model and only train the self-attention layer and the fully connected layer, or reset the weight coefficients of the loss function, etc.

[0118] It should be noted that although the steps in this embodiment Figure 1 and Figure 2 are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, Figure 1 and Figure 2 at least a part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these sub-steps or stages is not necessarily sequential either, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0119] Embodiment 2

[0120] Based on the denitration intelligent control method based on PID and inverse Transformer in Embodiment 1, this embodiment discloses a denitration intelligent control system based on PID and inverse Transformer, referring to Figure 3 , the denitration intelligent control system includes a data acquisition unit, a data prediction unit, a PID calculation unit and an online adjustment unit. Specifically:

[0121] The data acquisition unit is used to collect the operation parameter sequence of the SCR denitration system in real time and preprocess it;

[0122] The data prediction unit is used to swap the input time series and feature dimensions in the Transformer model to form an inverse Transformer model, and input the operation parameter sequence into the inverse Transformer model to obtain the predicted outlet NOx concentration within the prediction time window;

[0123] The PID calculation unit is used to calculate the adjustment value of the ammonia injection amount based on the PID controller according to the deviation between the predicted outlet NOx concentration and the target outlet NOx concentration;

[0124] The online adjustment unit is used to adjust the opening of the ammonia injection valve in real time according to the adjustment value of the ammonia injection amount, collect the actual outlet NOx concentration, and perform online adjustment on the inverse Transformer model and the PID controller based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration.

[0125] In this embodiment, the specific working processes and working principles of the installation relationship calculation unit, the transmission relationship calculation unit and the alignment unit are the same as those of the method in Embodiment 1, so they will not be elaborated in this embodiment. Each unit module can be implemented in whole or in part by software, hardware and their combination. Each unit module can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each unit module above.

[0126] Embodiment 3

[0127] As Figure 4 shown, this embodiment discloses a terminal device, including a transmitter, a receiver, a memory and a processor. Among them, the transmitter is used to send instructions and data, the receiver is used to receive instructions and data, the memory is used to store computer execution instructions, and the processor is used to execute the computer execution instructions stored in the memory to implement the method in Embodiment 1 above.

[0128] It should be noted that the above-mentioned memory can be either independent or integrated with the processor. When the memory is set independently, the terminal device further includes a bus for connecting the memory and the processor.

[0129] Embodiment 4

[0130] This embodiment discloses a computer-readable storage medium, in which computer-executable instructions are stored. When the processor executes the computer-executable instructions, the method in Embodiment 1 above is implemented.

[0131] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0132] The above are only the preferred embodiments of the present invention, and do not limit the protection scope of the present invention accordingly. Any equivalent structural transformation made by using the description and drawings of the present invention under the inventive concept of the present invention, or direct / indirect application in other related technical fields, is included in the protection scope of the present invention.

Claims

1. A denitrification intelligent control method based on PID and inverse Transformer, characterized in that: The steps include: Step 1: collect the operating parameter sequence of the SCR denitration system in real time and pre-process it; Step 2, swapping the time series and feature dimensions input into the Transformer model to form an inverse Transformer model, and inputting the operating parameter sequence into the inverse Transformer model to obtain a predicted outlet NOx concentration within a prediction time window; Step 3, calculating an ammonia injection amount adjustment value based on a PID controller according to a deviation between the predicted outlet NOx concentration and the target outlet NOx concentration; Step 4, adjusting the opening of the ammonia injection valve in real time based on the ammonia injection amount adjustment value, collecting the actual outlet NOx concentration, and adjusting the inverse Transformer model and the PID controller online based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration; The loss function of the inverted Transformer model during the training phase for: ； in, To reverse the Transformer model training process The weight coefficient of the round iteration; Represents the Huber loss, specifically: ； ； in, The number of samples contained in a training batch for the inverted Transformer model, For the batch The Huber loss corresponding to samples is: For the batch The predicted value of outlet NOx concentration corresponding to each sample is: For the batch The actual value of outlet NOx concentration corresponding to the sample, is a hyperparameter; Represents the MSE loss, specifically: ； ； in, For the batch The MSE loss corresponding to each sample.

2. The denitrification intelligent control method based on PID and inverse Transformer according to claim 1 is characterized in that: The operating parameter sequence has a variety of characteristic data sequences, including but not limited to inlet NOx concentration, outlet NOx concentration, ammonia escape rate, reactor temperature, and ammonia injection valve opening.

3. The denitrification intelligent control method based on PID and inverse Transformer according to claim 2 is characterized in that: In step 1, the preprocessing includes performing the following operations on each of the feature data sequences: First, calculate the mean μ and standard deviation σ of the feature data in the feature data sequence, and Characteristic data Mark as an outlier point; The abnormal points in the feature data sequence are interpolated and replaced, and the interpolated feature data sequence is normalized to: ； in, For feature data The corresponding feature data after normalization.

4. The denitrification intelligent control method based on PID and inverse transformer according to claim 1, 2 or 3, characterized in that: The weight coefficient Specifically: ； in, , are the initial weight and final value weight respectively, and , is the attenuation coefficient, is the number of training rounds.

5. The denitrification intelligent control method based on PID and inverse transformer according to claim 1, 2 or 3, characterized in that: In step 3, the ammonia injection amount adjustment value is specifically: ； ； in, is the ammonia injection amount adjustment value, is the deviation at the current moment, is the target outlet NOx concentration, To improve the predicted outlet NOx concentration T seconds into the future output by the Transformer model, is the accumulated deviation from the initial moment to the current moment, To use the interval, , , They are proportional gain, integral gain and differential gain respectively.

6. The denitrification intelligent control method based on PID and inverse transformer according to claim 5 is characterized in that: In step 4, the real-time adjustment of the ammonia injection valve opening based on the ammonia injection amount adjustment value is specifically as follows: like , then increase the opening of the ammonia injection valve, and the adjustment amount is ,in, is the maximum valve opening, is the adjustment coefficient; like , then reduce the opening of the ammonia injection valve, the adjustment amount is ; like , the current opening of the ammonia injection valve remains unchanged.

7. The denitrification intelligent control method based on PID and inverse transformer according to claim 1, 2 or 3, characterized in that: In step 4, online adjustment of the inverse Transformer model and the PID controller specifically includes: Step 401, setting a monitoring time window, and calculating the mean absolute error between the actual outlet NOx concentration and the predicted outlet NOx concentration within each monitoring time window; Step 402, determine whether the mean absolute error of N consecutive monitoring time windows exceeds a set threshold: If yes, proceed to step 403; Otherwise, wait for calculating the mean absolute error of the next monitoring time window and then proceed to step 402 again; Step 403, after updating the learning rate of the inverse Transformer model, calculate the mean absolute error of the next monitoring time window and determine whether it is less than a set threshold: If yes, after saving the learning rate of the inverted Transformer model, proceed to step 404; Otherwise, proceed to step 403 again; Step 404: dynamically adjust the proportional gain and the integral time of the PID controller based on the mean absolute error of the current monitoring time window and the mean absolute error of the previous monitoring time window.

8. The intelligent denitrification control method based on PID and inverse transformer according to claim 7 is characterized in that: Step 404 is specifically as follows: ； ； in, , is the adjusted proportional gain and integral time, , is the proportional gain and integral time before adjustment, is the adjustment factor, is the mean absolute error of the current monitoring time window, is the mean absolute error of the previous monitoring time window.

9. A denitrification intelligent control system based on PID and inverse Transformer, characterized in that: The method according to any one of claims 1 to 8 is used to perform intelligent denitration control, wherein the intelligent denitration control system comprises: The data acquisition unit is used to collect the operating parameter sequence of the SCR denitration system in real time and pre-process it; A data prediction unit, used for swapping the time series and feature dimensions inputted in the Transformer model to form an inverse Transformer model, and inputting the operating parameter sequence into the inverse Transformer model to obtain a predicted outlet NOx concentration within a prediction time window; A PID calculation unit, configured to calculate an ammonia injection amount adjustment value based on a PID controller according to a deviation between the predicted outlet NOx concentration and the target outlet NOx concentration; The online adjustment unit is used to adjust the opening of the ammonia injection valve in real time according to the ammonia injection amount adjustment value, collect the actual outlet NOx concentration, and make online adjustments to the inverse Transformer model and the PID controller based on the deviation between the actual outlet NOx concentration and the predicted outlet NOx concentration.

Images