Dynamic collaborative optimal selection control method for denitration of garbage incinerator

Abstract

The embodiment of the invention discloses a dynamic collaborative optimal selection control method for denitration of a garbage incinerator, and relates to the technical field of garbage incineration treatment. The method comprises the steps that operation parameters, collected in real time, of the garbage incinerator are obtained and preprocessed; based on the mechanism of waste incineration and denitration reaction and in combination with historical operation data, a dynamic model of the denitration process of the waste incinerator is established, and the denitration process is described by adopting an autoregressive moving average model; according to the preprocessed operation parameters and the established dynamic model, the working conditions of the garbage incinerator are recognized and classified in real time; multivariable model predictive control is adopted to make a cooperative control strategy; and outputting the formulated cooperative control strategy to execution mechanisms of the denitration system and the incineration system. According to the method, the denitration efficiency can be improved, nitrogen oxide emission can be accurately controlled, and the ammonia escape rate and the denitration cost can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of waste incineration treatment, and particularly to a dynamic collaborative optimization control method for denitrification of a waste incinerator. Background Art

[0002] With the acceleration of the urbanization process, the problem of waste treatment has gradually become the focus of social attention; waste incineration, as an effective waste treatment method, has been widely used around the world; waste incineration can not only reduce the volume of waste and relieve the landfill pressure, but also generate heat energy to supply the urban heating system; however, during the waste incineration process, the problems of combustion efficiency and pollution emissions have always been the key challenges in the technical and environmental protection fields; parameters such as temperature, pressure, flue gas composition, and oxygen content in the incinerator have an important impact on the efficiency and emission level of the combustion process; if these parameters cannot be monitored in real time and adjusted precisely, it may lead to low combustion efficiency and excessive pollutant emissions, affecting environmental protection and energy utilization.

[0003] The invention patent application CN107890770A discloses an SNCR acoustic wave temperature measurement and zonal injection system, and its technical key points include:

[0004] Acoustic wave temperature measurement: Measuring the furnace cross-section temperature through an acoustic wave sensor, constructing a two-dimensional temperature distribution map, and controlling the injection of the spray gun in zones.

[0005] Temperature window matching: Dynamically selecting the spray gun to be put into / withdrawn according to the optimal reaction temperature range of the reducing agent (such as ammonia water / urea) (for example, 850 - 950 °C for ammonia water).

[0006] This invention patent application has the following disadvantages: It is only applicable to the SNCR process and depends on the accuracy of the acoustic wave sensor, and the high-temperature and high-dust environment may affect the reliability of temperature measurement.

[0007] The invention patent CN113578006B discloses a SCR denitrification control method based on control strategy optimization, and its technical key points include:

[0008] Cascade PID control: The main controller stabilizes the outlet NOx concentration, the secondary controller optimizes the ammonia injection amount, and combines with the feedforward effect to improve the dynamic response.

[0009] Variable parameter design: Adjusting the PID parameters according to the change rate of the inlet NOx concentration, and anti-interference design to cope with data distortion during the maintenance of the CEMS instrument.

[0010] This invention patent has the following disadvantages: (1) It depends on a linearized model and has limited adaptability to extreme working conditions (such as drastic fluctuations in waste composition); (2) The feedforward parameters need to be adjusted manually based on experience, and the degree of intelligence is relatively low.

[0011] The invention patent application CN119532742A discloses a combustion management system for a waste incinerator, and its technical key points include:

[0012] Data analysis application of the ARMA model: The autoregressive moving average model (ARMA) is used for the analysis of the periodic changes in the combustion data of the waste incinerator. By establishing a time series model of historical combustion parameters (such as temperature, oxygen content), the periodic fluctuation law of the combustion process is identified. The core is to optimize the model order through the AIC / BIC criterion to capture the trend and seasonal characteristics of the combustion data.

[0013] Modular design of the combustion management system: The system includes three modules: data acquisition, ARMA analysis, and combustion parameter adjustment. The output of the ARMA model is used to guide the adjustment of parameters such as the damper opening of the burner and the feeding speed, realizing the preliminary optimization of the combustion efficiency.

[0014] This invention patent application has the following disadvantages: (1) Lack of a closed-loop control mechanism. The ARMA model is only used as an offline data analysis tool and does not form a closed loop with the real-time control strategy - the model analysis results cannot automatically trigger control instructions, resulting in control lag; (2) Single application scenario. The technology only optimizes the stability of the combustion system itself and does not involve the coordinated control of the denitration system, and cannot solve the coupling problem of nitrogen oxide generation and denitration efficiency during the incineration process; (3) Problem of model staticization. The parameters of the ARMA model are fixed after training and do not consider the changes in system characteristics caused by factors such as fluctuations in waste composition and equipment aging, and the model mismatch is likely to occur during long-term operation. Summary of the Invention

[0015] In view of this, the embodiments of the present invention provide a dynamic collaborative optimization control method for waste incinerator denitration to improve the denitration efficiency, accurately control nitrogen oxide emissions, and reduce the ammonia slip rate and denitration cost.

[0016] A dynamic collaborative optimization control method for waste incinerator denitration includes:

[0017] Step S101: Obtain the operation parameters of the waste incinerator collected in real time and perform preprocessing;

[0018] Step S102: Based on the mechanism of waste incineration and denitration reactions, combined with historical operation data, establish a dynamic model of the waste incinerator denitration process, where the autoregressive moving average model is used to describe the denitration process;

[0019] Step S103: According to the preprocessed operation parameters and the established dynamic model, perform real-time identification and classification of the operating conditions of the waste incinerator;

[0020] Step S104: Use multivariable model predictive control to formulate a collaborative control strategy;

[0021] Step S105: Output the formulated collaborative control strategy to the actuators of the denitration system and the incineration system.

[0022] Preferably, in the step S101, the operating parameters include the waste feed rate, waste composition, incinerator temperature, flue gas flow rate, nitrogen oxide concentration in the flue gas, and / or ammonia slip rate.

[0023] Preferably, in the step S102, the discrete-time model of the autoregressive moving average model is:

[0024] ;

[0025] where y(t) is the system output at time t; is the autoregressive coefficient; is the moving average coefficient; is a white noise sequence; p and q are the orders of autoregression and moving average respectively, determined by the information criterion method.

[0026] Preferably, in the step S102, the method for establishing and optimizing the dynamic model includes:

[0027] Step A1: Determine the input variables and output variables in combination with the denitration reaction mechanism of waste incineration;

[0028] Step A2: Collect at least 1000 groups of operating data under different working conditions;

[0029] Step A3: Construct a discrete-time model , and initialize the autoregressive order p and the moving average order q;

[0030] Step A4: Screen the optimal p / q through the AIC / BIC information criterion;

[0031] Step A5: Use historical data to train the model parameters , , and establish a dynamic mapping relationship between the input variables and the output variables;

[0032] Step A6: Evaluate the model accuracy through the root mean square error and the mean absolute error. If the accuracy meets the standard, solidify the model. If the accuracy does not meet the standard, return to step A4.

[0033] Preferably, in the step S103, the working conditions of the waste incinerator include stable working conditions, variable load working conditions, and waste composition fluctuation working conditions.

[0034] Preferably, in the step S103, the working condition characteristics are described by constructing a membership function. Taking the load change rate and the temperature fluctuation rate as an example, the membership function of the stable working condition is defined as:

[0035] ;

[0036] Among them, is the load change rate threshold, is the temperature volatility threshold; when the calculated is greater than or equal to the first preset threshold, it is determined that the waste incinerator is in a stable operating condition; when the calculated is less than the first preset threshold, the load change rate is greater than the second preset threshold, and the temperature volatility is less than or equal to the third preset threshold, it is determined that the waste incinerator is in a variable load operating condition; when the calculated is less than the first preset threshold, the load change rate is less than or equal to the second preset threshold, and the temperature volatility is greater than the third preset threshold, it is determined that the waste incinerator is in a waste composition fluctuation operating condition.

[0037] Preferably, in the step S104, the optimization objective function of the multivariable model predictive control is:

[0038] ;

[0039] Among them, measures the error between the predicted system output y(k|t) at time t for time k and the target output vector y r , and weights the errors of different output variables through the weight matrix Q; is used to constrain the magnitude of the control input u(k|t), and the weight matrix R determines the penalty degree for the change of the control input; further constrains the deviation between the system output and the target output at the end of the prediction horizon Np, and the weight matrix F ensures the long-term performance of the system within the entire prediction horizon;

[0040] The constraint conditions are:

[0041] ;

[0042] Among them, A, B, and E are the linearized system matrices, which respectively reflect the influences of the system state, control input, and external disturbance on the system state at the next moment; x(k|t) represents the prediction of the system state at time k at time t, u(k|t) is the control input at the corresponding time, and d(k|t) represents the external disturbance; u min and u max are respectively the lower and upper limit vectors of the control input; y NOx,max and y NOx,min are respectively the maximum allowable value and the minimum value of the nitrogen oxide emission concentration.

[0043] Preferably, in the step S104, the recursive least squares method is used to update the model parameters, and the parameter update law is as follows:

[0044] ;

[0045] where is the forgetting factor, is the regression vector, is the model parameter vector estimated in real time, and P(t) is the covariance matrix.

[0046] Preferably, the step S105 is further as follows:

[0047] Output the formulated cooperative control strategy to the actuators of the denitration system and the incineration system. At the same time, monitor the control effect in real time, compare the actual operation data with the control target, calculate the deviation, and use the PID feedback control algorithm to adjust and optimize the control strategy online according to the deviation magnitude and change trend to form a closed-loop control.

[0048] Preferably, the step S105 includes:

[0049] Step B1: Send the parameters generated by the cooperative control strategy to the actuator;

[0050] Step B2: Adjust the reductant injection amount and the incinerator operation parameters;

[0051] Step B3: Real-time collect the outlet NOx concentration, ammonia slip rate, and actual reductant consumption through sensors;

[0052] Step B4: Compare the actual value with the target value, and calculate the NOx concentration deviation and ammonia slip rate deviation;

[0053] Step B5: Preset a deviation threshold. If it does not exceed the deviation threshold, keep the existing control parameters and the process ends. If it exceeds, go to Step B6;

[0054] Step B6: According to the deviation magnitude and change trend, adaptively adjust the PID controller parameters, finely adjust the cooperative control strategy, and go to Step B1.

[0055] The denitration dynamic cooperative optimization control method for the waste incinerator of the present invention has the following beneficial effects:

[0056] 1. Strong dynamic adaptability: It can perceive the changes in the working conditions during the waste incineration process in real time. By establishing a dynamic model and identifying and classifying the working conditions, it can adjust the control strategy in time to adapt to complex working conditions such as fluctuations in waste composition and load changes, and improve the robustness of the denitration system.

[0057] 2. High denitration efficiency: Through the coordinated control of the denitration system and the incineration system and multi-variable coordinated control, precise regulation of the denitration process is achieved, which can effectively improve the nitrogen oxide removal rate and ensure stable compliance of nitrogen oxide emissions.

[0058] 3. Reduce ammonia slip rate: Precise control of the injection amount of the denitration agent and reaction conditions reduces the overuse of the denitration agent, lowers the ammonia slip rate, and reduces the harm to the environment and subsequent equipment.

[0059] 4. Cost savings: Optimize the denitration agent and energy consumption, taking into account both environmental protection and economic benefits. The usage amount of the denitration agent is optimized, while the operating efficiency of the incineration system is improved, reducing energy consumption and denitration costs, and having good economic and environmental benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] 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 following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0061] Figure 1 It is a schematic flow chart of the dynamic collaborative optimization control method for denitration of the waste incinerator of the present invention;

[0062] Figure 2 It is the overall schematic diagram of the dynamic collaborative optimization control method for denitration of the waste incinerator of the present invention;

[0063] Figure 3 It is the flow chart for establishing and optimizing the dynamic model in the present invention;

[0064] Figure 4 It is the flow chart for implementing the control strategy and feedback adjustment in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] The following will describe the embodiments of the present invention in detail with reference to the drawings.

[0066] It should be clear that 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 based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0067] The embodiments of the present invention provide a dynamic collaborative optimization control method for denitration of a waste incinerator, as Figure 1-2 shown, including:

[0068] Step S101: Obtain the operating parameters of the waste incinerator collected in real time and perform preprocessing;

[0069] This step corresponds to Figure 2 "Data acquisition and processing" in

[0070] Specifically, sensors can be used to collect the operating parameters of the waste incinerator in real time, including the waste feed rate, waste composition (such as the content of elements like carbon, hydrogen, oxygen, nitrogen, etc.), incinerator temperature (including furnace temperature, flue gas temperature, etc.), flue gas flow rate, nitrogen oxide concentration in the flue gas, ammonia slip rate, etc. Preprocess the collected data, including filtering, noise reduction, normalization, etc., to remove interference and outliers in the data and improve the accuracy and reliability of the data.

[0071] Step S102: Based on the mechanisms of waste incineration and denitrification reactions, combined with historical operating data, establish a dynamic model for the denitrification process of the waste incinerator. Among them, an autoregressive moving average model is used to describe the denitrification process;

[0072] This step corresponds to Figure 2 "Establishing a dynamic model" in

[0073] This model uses waste composition, incineration temperature, flue gas flow rate, etc. as input variables, and nitrogen oxide removal rate, ammonia slip rate, etc. as output variables, and can reflect the dynamic relationship between various variables in the denitrification process.

[0074] This solution uses an autoregressive moving average model (ARMA) to describe the denitrification process, and its discrete-time model is as follows:

[0075] ;

[0076] where y(t) is the system output at time t, such as the NOx concentration after denitrification;

[0077] are autoregressive coefficients, used to describe the influence of the system's historical output on the current output;

[0078] are moving average coefficients, reflecting the effect of past white noise on the current output;

[0079] is a white noise sequence, representing unpredictable random interference;

[0080] p and q are the orders of autoregression and moving average respectively, which need to be determined according to the characteristics of the actual data. To accurately adapt to the complex characteristics of the denitration process of the waste incinerator, when determining the orders of p and q, we used the information criterion method, such as the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Through the analysis and calculation of a large amount of actual operation data, with the goal of minimizing the AIC or BIC value, the optimal combination of p and q is selected to ensure that the ARMA model can accurately capture the dynamic change law in the denitration process and achieve an accurate description of the time-varying characteristics of the denitration process.

[0081] Different from the use of ARMA only for data analysis in the patent application CN119532742A, in this case, through the combination of mechanism analysis and data-driven, the ARMA model is first used for the dynamic modeling of the denitration process, establishing a dynamic mapping relationship between the input variables (such as waste composition, incineration temperature, flue gas flow rate, etc.) and the output variables (such as nitrogen oxide removal rate, ammonia slip rate, etc.), and solving the problem of modeling the time-varying non-linear characteristics of the denitration reaction.

[0082] As an optional embodiment, as Figure 3 shown, in the step S102, the method for establishing and optimizing the dynamic model includes:

[0083] Step A1 (mechanism analysis): Combining the mechanism of waste incineration denitration reaction (such as ammonia nitrogen oxidation-reduction reaction), determine the input variables (feed rate, temperature, reductant dosage) and the output variables (NOx removal rate, ammonia slip rate).

[0084] Step A2 (collection of historical data): Collect at least 1000 groups of operation data under different working conditions, including scenarios such as stable load, variable load, and composition fluctuation.

[0085] Step A3 (initialization of ARMA model): Construct a discrete-time model , and initialize the autoregressive order p and the moving average order q.

[0086] Step A4 (order optimization): Screen the optimal p / q through the AIC / BIC information criterion (such as p = 2, q = 1 in a certain case) to minimize the model prediction error.

[0087] Step A5 (model training): Use historical data to train the model parameters , , and establish a dynamic mapping relationship between the input variables and the output variables.

[0088] Step A6 (error verification): Evaluate the model accuracy through the root mean square error (RMSE) and the mean absolute error (MAE). For example, before implementation, RMSE = 30mg / Nm³, and after optimization, it drops to 10mg / Nm³.

[0089] Step A7 (Model solidification): After the error meets the standard, embed the model into the control system for real-time prediction and control strategy generation; if the error does not meet the standard, return to Step A4, that is, re-screen the optimal p / q order through the AIC / BIC criterion, adjust the autoregressive order p and the moving average order q, and then execute Steps A5 - A6 again until the accuracy meets the standard and the model is solidified.

[0090] Step S103: According to the preprocessed operating parameters and the established dynamic model, identify and classify the working conditions of the waste incinerator in real time;

[0091] This step corresponds to Figure 2 "Working condition identification and classification" in

[0092] As an optional embodiment, in Step S103, the working conditions of the waste incinerator include different types such as stable working conditions, variable load working conditions, and waste composition fluctuation working conditions, etc. For different types of working conditions, corresponding control strategies are formulated to achieve dynamic switching of control strategies.

[0093] In the method of the present invention, constructing the membership function plays a key role in accurately describing the working condition characteristics. Taking the load change rate and the temperature volatility these two important parameters that can significantly reflect the operating state of the waste incinerator as an example, the membership function of the stable working condition is elaborated in detail. The membership function of the stable working condition is defined as:

[0094] ;

[0095] Among them, is the load change rate threshold, is the temperature volatility threshold, which is determined by analyzing historical data. The historical data covers the load changes and temperature fluctuations of the waste incinerator at different operating stages, different waste types, and different environmental conditions. After screening, sorting, and statistical calculation of these data, reasonable and values are finally determined to ensure that the membership function can accurately reflect the actual working conditions and solve the problem of automatic switching of control strategies under complex working conditions (there is no multi-dimensional working condition classification in the prior art).

[0096] In practical applications, when the calculated , the system will determine that the waste incinerator is in a stable working condition. This determination means that the load and temperature changes of the incinerator are in a relatively stable state, and at this time, the denitration control can be carried out according to the strategy under the conventional stable working condition. On the contrary, when (the load change rate and the temperature volatility is the variable load working condition; the load change rate and the temperature volatility is the fluctuation condition of the waste composition), the system will immediately trigger the variable condition control logic to cope with the instability of the incinerator operation state, ensure that the denitrification process can adapt to the condition changes, and continue to operate efficiently.

[0097] This function first couples the load and temperature fluctuations for analysis. Compared with several existing technologies mentioned above, it can more accurately identify complex conditions (such as the coupling scenario of load mutation and temperature anomaly), providing a key basis for the dynamic switching of control strategies.

[0098] Step S104: Use Model Predictive Control (MPC) to formulate a coordinated control strategy;

[0099] This step corresponds to Figure 2 "formulating a coordinated control strategy" in

[0100] Coordinated control of the denitrification system and the incineration system:

[0101] The fluctuations of parameters such as the combustion temperature and feed rate of the incineration system directly affect the denitrification effect, and it is necessary to achieve dynamic matching of the two systems through coordinated control. To achieve the coordinated control between the two, this method is constructed on the advanced framework of multi-variable model predictive control. The framework's rolling optimization algorithm dynamically and precisely adjusts the control input.

[0102] Taking the actual operation scenario as an example, in the incineration system, when the waste feed rate changes due to factors such as waste source and quality, the temperature in the combustion chamber fluctuates up and down. At the same time, the flue gas generation volume also shows corresponding increases and decreases. These changes will further be transmitted to the denitrification system, directly affecting the generation volume of nitrogen oxides (NOx). With the help of this coordinated control strategy, the system can capture the real-time state changes of the incineration system and adjust the key control parameters of the denitrification system in advance. It can not only ensure that the NOx emissions meet the environmental protection standards but also enable the incineration system to continuously maintain an efficient and stable operation state, achieving a win-win situation for environmental protection and production benefits.

[0103] Multi-variable coordinated control:

[0104] The present invention uses multi-variable model predictive control to coordinate the relationship between multiple variables. Within the prediction time domain Np, it is committed to solving the rolling optimization problem. The core of this problem is to reasonably adjust the control input to make the system output as close as possible to the expected target, while ensuring that the change of the control input is within a reasonable range to avoid over-adjustment leading to system instability or increased energy consumption. Its optimization objective function can be expressed as:

[0105] ;

[0106] This objective function comprehensively considers the deviation between the system output and the target output within the next Np time steps starting from the current time t, as well as the magnitude of the control input. Among them, measures the error between the predicted system output y(k|t) at time t and the target output vector y r . The error of different output variables is weighted by the weight matrix Q to highlight the output indicators that need to be focused on. For example, in the waste incineration scenario, if the control precision of the nitrogen oxide emission concentration is crucial, larger weights can be assigned to the elements related to nitrogen oxide emission in the weight matrix Q, so that the system gives priority to reducing the error of this indicator during the optimization process. is used to constrain the magnitude of the control input u(k|t). The weight matrix R determines the degree of punishment for the change of the control input, avoiding the instability of the system or additional energy consumption caused by too large a control input.

[0107] In the actual operation of the equipment, too large a control input may cause the equipment components to bear too high pressure or current, not only shortening the service life of the equipment, but also causing unnecessary energy waste. By setting an appropriate weight matrix R, the amplitude of the control input can be effectively restricted, ensuring that the system operates within a safe and energy-saving range.

[0108] And further constrains the deviation between the system output and the target output at the end point Np of the prediction horizon. The weight matrix F ensures the long-term performance of the system within the entire prediction horizon and enhances the closed-loop stability. It is like setting a clear target boundary for the long-term operation of the system, preventing the system from gradually deviating from the expected trajectory during operation, and ensuring that the system can stably output compliant results for a long time.

[0109] To ensure the feasibility and practical physical meaning of the optimization problem, this optimization process needs to satisfy a series of strict constraint conditions, as follows:

[0110] ;

[0111] Among them, the state transition equation describes the evolution law of the system state over time.

[0112] A, B, and E are the linearized system matrices, which respectively reflect the influences of the system state, control input, and external disturbance on the system state at the next moment.

[0113] $x(k|t)$ represents the prediction of the system state at time $k$ at time $t$, $u(k|t)$ is the control input at the corresponding time, and $d(k|t)$ represents the external disturbance. For example, in a waste incinerator, changes in the external environmental temperature, sudden changes in the waste composition, etc. can all be regarded as the external disturbance $d(k|t)$. Through this equation, the system can accurately predict the state at the next moment based on the current state, control input, and external disturbance, providing a solid data basis for subsequent control decisions.

[0114] Control input constraint $u$ min $\underline{u}\leq u(k|t)\leq\overline{u}$ max ensures that the control input is always within the physically realizable range, preventing equipment damage or abnormal operation caused by too large or too small control input. Here, $\underline{u}$ min and $\overline{u}$ max are the lower and upper bound vectors of the control input respectively, and each element corresponds to the value range of a control variable. Taking the denitration agent injection pump as an example, there is a safe operating range for its working flow rate, and the corresponding flow rate values are the lower bound $\underline{u}$ min and the upper bound $\overline{u}$ max of the control input variable. Any control instruction outside this range may cause equipment failure or ineffective execution of the denitration task.

[0115] For the key index $y$ NOx in the system output (such as the nitrogen oxide emission concentration in the denitration process of a waste incinerator), strict upper and lower bound constraints are also set .

[0116] $\overline{y}$ NOx,max and $\underline{y}$ NOx,min are the maximum and minimum allowable values of the nitrogen oxide emission concentration respectively. This constraint ensures that the system always meets relevant requirements such as environmental protection during operation, and controls the nitrogen oxide emission within a reasonable range. Environmental protection regulations have clear and strict standards for the nitrogen oxide emission of waste incineration plants. Once the emission concentration exceeds the upper bound $\overline{y}$ NOx,max , there will be a risk of high fines or even production suspension for rectification; if the emission concentration is much lower than the lower bound $\underline{y}$ NOx,min for a long time, although it meets the environmental protection requirements, it may mean that the denitration process consumes excessive resources and affects production efficiency.

[0117] During the actual operation process, the parameters of the system model may change due to various factors (such as equipment aging, environmental changes, etc.), thus affecting the control effect. To address this issue, this method uses the recursive least squares (RLS) method for online model parameter update to track the changes in system characteristics in real time to ensure the effectiveness and adaptability of the control algorithm. The parameter update law is as follows:

[0118] ;

[0119] Among them, is the forgetting factor, and its value range is usually between [0, 1], which is used to adjust the forgetting speed of historical data. A smaller value means that the algorithm pays more attention to recent data and responds faster to system changes, but may be more sensitive to noise. For example, when the area where the waste incinerator is located suddenly encounters extreme weather and the equipment operating environment changes drastically, a smaller can enable the algorithm to quickly capture these changes and adjust the model parameters in a timely manner to adapt to the new working conditions. However, if there is more noise interference in the system measurement data, a smaller value may cause large fluctuations in parameter updates due to the influence of noise. A larger value makes the algorithm more dependent on historical data, can smooth the influence of noise to a certain extent, but has relatively weak tracking ability for system dynamic changes. During the stage when the equipment operation is relatively stable and the environmental changes are small, a larger value helps to utilize the effective information in historical data and improve the stability and accuracy of parameter estimation. By reasonably selecting the forgetting factor , a balance can be achieved between tracking system changes and suppressing noise.

[0120] is the regression vector, which contains historical input and output information related to the system output, and uses this information to construct the relationship between model parameters and system output. In the waste incineration system, may cover data such as waste feed rate, combustion temperature, denitration agent injection rate, and corresponding nitrogen oxide emission concentration in the past period. Through the combination of regression vectors, it provides a rich information basis for the algorithm to accurately estimate model parameters.

[0121] T represents the matrix transpose operator (Transpose), which is used to convert the regression vector from a row vector to a column vector to ensure the dimensional consistency of matrix multiplication.

[0122] is the model parameter vector estimated in real time. As new data is continuously input, the algorithm uses a recursive formula to continuously update this parameter vector so that it can accurately reflect the current characteristics of the system. For example, as the waste incinerator operates for a long time, the refractory materials in the furnace gradually wear, the combustion efficiency changes, and the model parameters will also change accordingly. Through the recursive least squares method, can track these changes in a timely manner and ensure that the model always matches the actual system.

[0123] P(t) is the covariance matrix, which is used to measure the uncertainty of parameter estimation. Each time the parameters are updated, the covariance matrix is also adjusted accordingly to reflect the change in the accuracy of parameter estimation with the accumulation of data. When the system is just started and the amount of data is small, the covariance matrix P(t) is large, indicating a high uncertainty in parameter estimation; as the running time increases and data accumulates continuously, P(t) gradually decreases and the parameter estimation accuracy improves continuously. Through the above parameter update process of the recursive least squares method, the system can adapt to the changes of model parameters in real time, thus continuously maintaining good control performance.

[0124] In this step, for different working conditions, a multi-variable model predictive control (MPC) is used to construct an optimization objective function, and rolling optimization is carried out in combination with equipment constraints and environmental protection requirements, and the model parameters are updated online through the RLS algorithm.

[0125] Different from the MPC in the patent application CN118836453A which is only used for the control of a single system (combustion-supporting air), the present invention extends the MPC algorithm to the coordinated control of the denitration and incineration dual systems for the first time. By constructing an optimization objective function including cross-system constraints (such as the incineration temperature affecting the denitration reaction efficiency), the dynamic matching of the dual systems is realized, and the problem of adjustment lag between systems in the prior art is solved.

[0126] The present invention realizes the online adaptive update of the ARMA model parameters for the first time (different from several existing technologies mentioned above), through the forgetting factor dynamically adjusts the weights of historical data, enabling the model to track in real time the changes in system characteristics caused by fluctuations in garbage composition, equipment aging, etc., filling the problem of the decline in control accuracy caused by model mismatch in the prior art.

[0127] Step S105: Output the formulated coordinated control strategy to the actuators of the denitration system and the incineration system.

[0128] This step corresponds to Figure 2 "Implementation of control strategy and feedback adjustment" in

[0129] Preferably, this step is further: Output the formulated coordinated control strategy to the actuators of the denitration system and the incineration system, such as the denitration agent injection pump, the incinerator burner, etc., to realize the real-time control of the denitration process and the incineration process. At the same time, monitor the control effect in real time, compare the actual operation data with the control target, calculate the deviation, and use the PID feedback control algorithm to adjust and optimize the control strategy online according to the magnitude and change trend of the deviation, forming a closed-loop control to ensure that the denitration system is always in the best operating state.

[0130] In this way, a complete closed-loop of "dynamic modeling - operating condition identification - coordinated control - real-time optimization" is formed, which is different from the single control logic of the existing technologies (such as the one-way control in patent application CN119532742A and the lack of parameter self-adaptation in patent application CN118836453A). Through the coordination of PID feedback and RLS update, the real-time self-adaptation / self-optimization of the model and control strategy is realized, ensuring long-term stable operation under complex operating conditions.

[0131] Preferably, as Figure 4 shown, the step S105 includes:

[0132] Step B1 (control instruction output): Send the parameters generated by the coordinated control strategy (such as ammonia injection amount, combustion temperature set value) to the actuators (injection pump, burner).

[0133] Step B2 (actuator action): Adjust the reductant injection amount (such as dynamically adjusting from 500 L / h to 650 L / h) and the operating parameters of the incinerator (such as the furnace temperature of 850 °C ± 50 °C).

[0134] Step B3 (real-time monitoring): Real-time collect data such as the outlet NOx concentration (target ≤ 100 mg / Nm³), ammonia slip rate (target ≤ 7.8 ppm), and actual reductant consumption through sensors.

[0135] Step B4 (deviation calculation): Compare the actual value with the target value, and calculate the NOx concentration deviation (such as actual , deviation of 5 mg / Nm³) and ammonia slip rate deviation.

[0136] Step B5 (deviation determination): Preset a deviation threshold (such as NOx concentration fluctuation > 6.5% or ammonia slip rate > 1.5 ppm). If the threshold is exceeded, trigger feedback adjustment.

[0137] Step B6 (PID parameter adjustment): According to the deviation magnitude and change trend, adaptively adjust the PID controller parameters (such as Kp = 0.8 under stable operating conditions, Kp = 1.2 under variable operating conditions), fine-tune the control strategy, and go back to step B1.

[0138] Step B7 (maintain the current strategy): When the deviation is within the allowable range, keep the existing control parameters to avoid system fluctuations caused by excessive adjustment, and the process ends.

[0139] Implementation case

[0140] 1) Data collection and processing

[0141] Taking a 750 t / d domestic waste incineration power plant in Zhejiang as an example, the following core parameters are collected in real time:

[0142] Garbage feed rate: 750t / d (fluctuation ±5% at stable load);

[0143] Inlet NOx concentration: 300mg / Nm³ (standard state, 11% O2), the daily average value of the outlet index is ≤100mg / Nm³, and the current process detection value is 150mg / Nm³;

[0144] Reducing agent: 20% concentration ammonia water, initial injection volume 500L / h (corresponding to a NOx removal rate of about 50%);

[0145] Incineration temperature: furnace center temperature 850±50℃, flue gas temperature 300±30℃;

[0146] Flue gas flow rate: 120,000Nm³ / h (standard state).

[0147] 2) Building a dynamic model

[0148] ARMA model parameter calculation:

[0149] The outlet NOx concentration is taken as the output variable y(t), and the input variables include the feed volume L(t), the inlet NOx concentration CNOx,in(t), and the ammonia injection volume u(t). The optimal order p=2, q=1 is determined through historical data training, and the model expression is: .

[0150] Model Validation:

[0151] Prediction error before implementation: RMSE=30mg / Nm³, MAE=25mg / Nm³;

[0152] After implementation, through RLS online update, RMSE dropped to 10mg / Nm³, MAE=8mg / Nm³, meeting the forecast accuracy requirement of export index ≤100mg / Nm³.

[0153] 3) Working condition identification and classification Determination of stable working condition:

[0154] Load change rate ,when And the temperature fluctuation rate When , the membership function is calculated as: , it is determined to be a stable operating condition and the conventional control strategy is triggered; if the feed rate fluctuation is greater than 5% (such as an instantaneous increase to 800t / d), then , start the variable load control logic.

[0155] 4) Collaborative control strategy formulation and multivariable optimization calculation:

[0156] The objective function is to optimize the ammonia injection amount u(t) with the outlet NOx ≤ 100 mg / Nm³ as the constraint. p= 10, the weight matrix Q = diag[10, 1] (NOx weight takes precedence), R = diag[0.1] (restricts the fluctuation of the injection volume).

[0157] Constraint conditions: State transition equation: (simplified model);

[0158] Control input: 200 L / h ≤ u(k) ≤ 800 L / h (safety range of the ammonia injection pump);

[0159] Output constraint: y NOx,out ≤ 100 mg / Nm³.

[0160] RLS parameter update: Forgetting factor , when the inlet NOx concentration suddenly rises to 350 mg / Nm³, the algorithm completes the parameter update within 5 time steps (about 30 seconds), and the injection volume is dynamically adjusted from 500 L / h to 650 L / h to ensure that the outlet concentration is stable at 95 ± 5 mg / Nm³.

[0161] The data improvement before and after implementing the method of the present invention is compared as shown in Table 1:

[0162] Table 1

[0163] Index Before implementation After implementation Improvement rate Denitration efficiency 52.1% 66.7% 28% Ammonia escape rate 11.3 ppm ≤7.8 ppm 31% Reductant dosage 520 L / h 425 L / h 18.3% Load response time 10 minutes 3 minutes 70% Compliance stability Daily average value fluctuation ±18.3% Daily average value fluctuation ±6.7% Improved by 2.7 times

[0164] In the method of the present invention, technologies such as ARMA, MPC, and RLS are not simply superimposed, but achieve deep collaboration through the operating condition identification module - ARMA provides accurate prediction for MPC, and the output of MPC drives RLS to update the model parameters online, forming a "prediction - control - optimization" closed loop. This systematic innovation is not involved in the several existing technologies mentioned above, effectively solving the problems of control lag and parameter mismatch under complex operating conditions.

[0165] The method of the present invention realizes the coordinated control of the denitration and incineration systems for the first time, synchronously adjusts the incineration temperature and the dosage of the denitration agent through the multi - variable optimization objective function (for example, when the nitrogen content in the garbage suddenly increases, automatically coordinates to increase the incineration temperature to reduce NOx generation, and at the same time optimizes the injection volume of the reducing agent), which is different from the adjustment logic of the existing patent CN118836453A that only targets a single combustion air system.

[0166] In summary, the dynamic collaborative optimization control method for denitrification of the waste incinerator in the present invention aims at the problem of controlling the emission of nitrogen oxides (NOx) during the waste incineration process. By integrating technologies such as dynamic modeling, real-time operating condition identification, multivariable collaborative control, and autoregressive moving average model (ARMA), it realizes the collaborative optimization of the denitrification system and the incineration system, and is applicable to efficient denitrification control under complex operating conditions such as fluctuating waste composition and changing load, belonging to the cross-field of industrial process control and environmental pollution treatment. The method of the present invention can dynamically adjust the denitrification control strategy according to the real-time changes of the operating conditions during the waste incineration process, realize the collaborative optimization of the denitrification system and the incineration system, improve the denitrification efficiency, accurately control the emission of nitrogen oxides, and reduce the ammonia slip rate and denitrification cost. This method also has the following beneficial effects:

[0167] 1. Strong dynamic adaptability: It can perceive the changes in the operating conditions during the waste incineration process in real time. By establishing a dynamic model and classifying the operating conditions, it can adjust the control strategy in a timely manner to adapt to complex operating conditions such as fluctuating waste composition and changing load, and improve the robustness of the denitrification system.

[0168] 2. High denitrification efficiency: Through the collaborative control of the denitrification system and the incineration system and multivariable coordinated control, it realizes the precise regulation of the denitrification process, can effectively improve the nitrogen oxide removal rate, and ensure the stable compliance of nitrogen oxide emissions.

[0169] 3. Reducing the ammonia slip rate: By precisely controlling the injection amount of the denitrification agent and the reaction conditions, it reduces the excessive use of the denitrification agent, reduces the ammonia slip rate, and reduces the harm to the environment and subsequent equipment.

[0170] 4. Cost savings: Optimize the denitrification agent and energy consumption, taking into account both environmental protection and economic benefits. It optimizes the usage amount of the denitrification agent, improves the operating efficiency of the incineration system at the same time, reduces the energy consumption and denitrification cost, and has good economic and environmental benefits.

[0171] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A dynamic collaborative optimization control method for denitrification of a waste incinerator, characterized in that, Including: Step S101: Obtain the operating parameters of the waste incinerator collected in real time and perform preprocessing; Step S102: Based on the mechanisms of waste incineration and denitrification reactions, combined with historical operating data, establish a dynamic model for the denitrification process of the waste incinerator. Among them, an autoregressive moving average model is used to describe the denitrification process; Step S103: According to the preprocessed operating parameters and the established dynamic model, perform real-time identification and classification of the operating conditions of the waste incinerator; Step S104: Adopt multivariable model predictive control to formulate a collaborative control strategy; Step S105: Output the formulated collaborative control strategy to the actuators of the denitrification system and the incineration system.

2. The denitration dynamic collaborative optimization control method for the waste incinerator according to claim 1, characterized in that, In the step S101, the operating parameters include the waste feeding amount, waste composition, incinerator temperature, flue gas flow rate, nitrogen oxide concentration in the flue gas, and / or ammonia slip rate.

3. The denitration dynamic collaborative optimization control method for a waste incinerator according to claim 1, characterized in that In the step S102, the discrete-time model of the autoregressive moving average model is: ； where y(t) is the system output at time t; are autoregressive coefficients; are moving average coefficients; is a white noise sequence; p and q are the orders of autoregression and moving average respectively, determined by the information criterion method.

4. The denitration dynamic collaborative optimization control method for the waste incinerator according to claim 3, wherein, In the step S102, the methods for establishing and optimizing the dynamic model include: Step A1: Determine the input variables and output variables in combination with the waste incineration denitrification reaction mechanism; Step A2: Collect at least 1000 groups of operating data under different operating conditions; Step A3: Construct a discrete-time model , initialize the autoregressive order p and the moving average order q; Step A4: Screen the optimal p / q through the AIC / BIC information criterion; Step A5: Train the model parameters using historical data , to establish a dynamic mapping relationship between the input variables and the output variables; Step A6: Evaluate the model accuracy through the root mean square error and the mean absolute error. If the accuracy meets the standard, solidify the model. If the accuracy does not meet the standard, return to step A4.

5. The denitration dynamic collaborative optimization control method of the waste incinerator according to claim 1, characterized in that, In the step S103, the operating conditions of the waste incinerator include stable operating conditions, variable load operating conditions, and waste composition fluctuation operating conditions.

6. The denitration dynamic collaborative optimization control method for a waste incinerator according to claim 5, characterized in that, In the step S103, the membership function is constructed to describe the working condition characteristics, taking the load change rate and the temperature volatility as examples. The membership function of the stable working condition is defined as: ； Among them, is the load change rate threshold, is the temperature volatility threshold; when the calculated is greater than or equal to the first preset threshold, it is determined that the waste incinerator is in a stable operating condition; when the calculated is less than the first preset threshold, the load change rate is greater than the second preset threshold, and the temperature volatility is less than or equal to the third preset threshold, it is determined that the waste incinerator is in a variable load operating condition; when the calculated is less than the first preset threshold, the load change rate is less than or equal to the second preset threshold, and the temperature volatility is greater than the third preset threshold, it is determined that the waste incinerator is in a waste composition fluctuation operating condition.

7. The denitration dynamic collaborative optimization control method for a waste incinerator according to claim 1, characterized in that In the step S104, the optimization objective function of the multivariable model predictive control is: ； Among them, Measure the error between the predicted system output y(k|t) at time k at time t and the target output vector y r And weight the errors of different output variables through the weight matrix Q; Used to constrain the magnitude of the control input u(k|t), and the weight matrix R determines the degree of penalty for changes in the control input; Further constrain the deviation between the system output and the target output at the end Np of the prediction horizon, and the weight matrix F ensures the long-term performance of the system within the entire prediction horizon; The constraint conditions are: ； Among them, A, B, and E are the linearized system matrices, which respectively reflect the influences of the system state, control input, and external disturbance on the system state at the next moment; x(k|t) represents the prediction of the system state at time k at time t, u(k|t) is the control input at the corresponding time, and d(k|t) represents the external disturbance; u min and u max are respectively the lower and upper limit vectors of the control input; y NOx,max and y NOx,min are respectively the maximum allowable value and minimum value of the nitrogen oxide emission concentration.

8. The denitration dynamic collaborative optimization control method for a waste incinerator according to claim 7, characterized in that, In the step S104, the recursive least squares method is used to update the model parameters, and the parameter update law is: ； wherein, is the forgetting factor, is the regression vector, is the vector of model parameters estimated in real time, and P(t) is the covariance matrix.

9. The denitration dynamic collaborative optimization control method for a waste incinerator according to any one of claims 1-8, characterized in that, The step S105 is further: Output the formulated collaborative control strategy to the actuators of the denitrification system and the incineration system. At the same time, monitor the control effect in real time, compare the actual operating data with the control target, calculate the deviation, and according to the magnitude and change trend of the deviation, use the PID feedback control algorithm to perform online adjustment and optimization of the control strategy to form a closed-loop control.

10. The dynamic collaborative optimization control method for denitrification of a waste incinerator according to claim 9, characterized in that The step S105 includes: Step B1: Send the parameters generated by the collaborative control strategy to the actuator; Step B2: Adjust the reductant injection amount and the operating parameters of the incinerator; Step B3: Collect the outlet NOx concentration, ammonia slip rate, and actual reductant consumption in real time through sensors; Step B4: Compare the actual value with the target value and calculate the NOx concentration deviation and the ammonia slip rate deviation; Step B5: Preset a deviation threshold. If it does not exceed the deviation threshold, maintain the existing control parameters and the process ends. If it exceeds, go to step B6; Step B6: According to the magnitude and change trend of the deviation, adaptively adjust the parameters of the PID controller, finely adjust the collaborative control strategy, and go to step B1.

Images