Model-driven SCR denitrification and ammonia injection targeted control loop

Abstract

The present invention relates to a model-driven SCR denitration ammonia injection targeted control loop, which includes an SCR steady-state model, a dynamic prediction model, a targeted control model, an SCR inlet NO x Concentration soft measurement model and SCR ammonia slip soft measurement model, SCR denitrification system outlet NO x After the concentration is added to the output value of the dynamic prediction model, it is used as the input signal of the SCR steady-state model. The SCR steady-state model gives a reference instruction for the targeted control model to adjust according to the direction and size of the input signal and the deviation degree in the previous and next cycles. At the same time, the reference instruction is also used as the dynamic prediction model, SCR inlet NO x The signal input value of the concentration soft measurement model and the SCR ammonia slip soft measurement model, the SCR inlet NO x The signal output of the concentration soft measurement model and the SCR ammonia slip soft measurement model is fed back to the dynamic prediction model. The present invention can effectively reduce the amount of ammonia injection and improve the operation economy of the unit.

Description

Technical Field

[0001] The invention relates to the technical field of ammonia injection regulation of an SCR denitration system of a coal-fired unit, and in particular to a model-driven SCR denitration ammonia injection targeted control loop. Background Art

[0002] At present, there are two main ways to regulate the ammonia injection in the SCR denitrification system of coal-fired units in my country: one is the constant mole ratio control method, and the other is the fixed outlet NO x Set value control mode (Constant Outlet NO x Control), where:

[0003] Fixed ammonia nitrogen molar ratio method: This control method is widely used and is a typical control strategy. It mainly determines the fixed ammonia nitrogen molar ratio based on the catalyst denitrification capacity and denitrification efficiency, thereby removing NO in the flue gas. x This control strategy is a single-loop control system. The set value can be adjusted according to the unit operation status. The control parameters are easy to debug and set. The control principle diagram is as follows: Figure 1 shown. Figure 1 The SCR reactor uses imported NO x The concentration is multiplied by the flue gas flow rate to obtain NO x The instantaneous content is multiplied by the ammonia nitrogen molar ratio coefficient K to obtain the required basic amount of ammonia spray, which is then compared with the outlet NO fed back. x The required ammonia injection amount is corrected by comparing the content. After the PID control controller calculates, it gives a signal to adjust the opening of the ammonia injection regulating valve to control the actual required ammonia injection amount. The ammonia nitrogen molar ratio coefficient K can be set manually. 1 (x), F 2 (x), F 3 (x), F 4 (x) are function generators, which calculate the flue gas flow and NO according to the boiler load. X The purpose of this is to ensure that the destocking system continues to operate when the flue gas analyzer (CEMS) fails.

[0004] The disadvantages of this control strategy are: ① Since the boiler combustion process is dynamic, the nitrogen oxides in the flue gas will change in real time, and it is very easy to cause excessive denitrification after fixing the ammonia-nitrogen molar ratio; ② Excessive ammonia injection will not be able to accurately control the escape amount of ammonia; ③ Increase operating costs and secondary environmental pollution; ④ Only consider the denitrification efficiency, at the inlet NO x When the concentration is relatively low, it is easy to over-spray ammonia, which increases operating costs.

[0005] Fixed outlet NO x Set value control mode: outlet NO x The fixed value control is to keep the outlet NO x The content is constant and the reactor NO is controlled according to the ambient air quality standard. x It is easier to monitor the denitrification efficiency than to control the fixed value, and it also reduces the consumption of ammonia. x Compared with the above-mentioned fixed molar ratio control method, the fixed value control method is basically the same in the control loop, and the main difference is that the molar ratio is a variable value, and the ammonia nitrogen molar ratio is related to the reactor and the reactor SCR outlet NO x The value corresponds to the boiler load. x The control principle diagram of fixed value control of ammonia injection amount Figure 2 This control method also belongs to a single-loop control system, which only considers the outlet NO x The concentration change at the inlet NO x When the concentration is too high, the outlet concentration setting value is too low, which will exceed the denitrification capacity of the catalyst, easily leading to excessive ammonia injection and increasing operating costs.

[0006] The above control methods all need to be based on the export NO x The required ammonia volume is obtained by measuring the concentration and denitration efficiency. The injection volume of ammonia is adjusted by the on-site regulating valve. Too little ammonia injection will result in too low denitration efficiency, while too much ammonia injection will easily increase the ammonia escape rate, causing dust accumulation and corrosion in the tail flue, which is also a waste of ammonia supply from an economic point of view. x The content and inlet flue gas volume are crucial parameters for calculating ammonia consumption. The former can be detected in real time by the denitrification CEMS analyzer, while the inlet flue gas volume is limited by the size of the flue and the unevenness of the gas flow in the flue, making it difficult to measure accurately. This is also the key point in controlling ammonia consumption. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention provides a model-driven SCR denitration ammonia injection targeted control loop to achieve a stable SCR denitration control system outlet NO x The function of the unit operating parameters such as concentration can effectively reduce the amount of ammonia injection, improve the unit's operating economy, reduce the workload of operators, and improve the unit's economic operation level, while strictly meeting national environmental protection requirements.

[0008] The technical solution adopted by the present invention to solve the technical problem is: the model-based SCR denitrification ammonia injection targeted control loop includes an SCR steady-state model, a dynamic prediction model, a targeted control model, an SCR inlet NO x Concentration soft-sensing model and SCR ammonia slip soft-sensing model, where:

[0009] SCR denitration system outlet NO x The concentration is added to the output value of the dynamic prediction model and used as the input signal of the SCR steady-state model. The SCR steady-state model gives a reference instruction for the targeted control model to adjust the ammonia demand according to the direction and size of the input signal and the deviation degree in the previous and next cycles. At the same time, the reference instruction is also used as the dynamic prediction model, SCR inlet NO x The signal input value of the concentration soft measurement model and the SCR ammonia slip soft measurement model, the SCR inlet NO x The signal outputs of the concentration soft-sensing model and the SCR ammonia slip soft-sensing model are fed back to the dynamic prediction model.

[0010] In a preferred embodiment of the model-driven SCR denitration ammonia injection targeted control loop provided by the present invention, the SCR inlet NO x The concentration soft-sensing model is based on the output value of the SCR steady-state model, combined with the unit load and boiler load, SOFA valve opening, and dynamic changes in primary and secondary air volume, and uses multivariate process detection to select the factors that affect the inlet NO of the SCR denitrification reactor. x The dominant factor of NO content in the reactor inlet was established by using the least squares support vector machine. x The prediction model of the content.

[0011] In a preferred embodiment of the model-driven SCR denitration ammonia injection targeted control loop provided by the present invention, the SCR ammonia slip soft measurement model is based on the measured reactor inlet and outlet NO x The theoretical required ammonia injection amount is calculated based on the content and flue gas volume, the actual ammonia injection amount is compared with the theoretical required ammonia injection amount, and the final ammonia escape amount is obtained by averaging over a certain period of time.

[0012] In a preferred embodiment of the model-driven SCR denitration ammonia injection targeted control loop provided by the present invention, the targeted control model is a directional model implemented based on the DCS system control software, through the inlet NO at a certain moment x The concentration, ammonia consumption and ammonia escape can quantitatively give the required ammonia demand.

[0013] In a preferred embodiment of the model-driven SCR denitration ammonia injection targeted control loop provided by the present invention, the dynamic prediction model includes an inner control loop and an outer control loop, wherein:

[0014] The inner control loop is composed of the injection ammonia flow rate change transfer function And PI controller, where: ammonia injection flow rate change transfer function It is the transfer function of the ammonia injection flow rate change when the ammonia injection regulating valve changes its opening;

[0015] The outer control loop is composed of SCR outlet NO x Concentration change transfer function (G SCR (s)), MD-TDOF-PID controller and load weighter, where: SCR outlet NO x Concentration change transfer function (G SCR (s)) is the transfer function established by the change of NOx concentration at the reactor outlet during the change of ammonia injection flow rate; the load weighter is based on the actual load Pe in this load segment and can be defined as [Pe1, Pe2]. The weights of the actual load at Pe1 and Pe2 are defined as x1 and x2, then (Note: abs is the absolute value function).

[0016] In a preferred embodiment of the model-driven SCR denitrification and ammonia injection targeted control loop provided by the present invention, the control loop of the MD-TDOF-PID controller includes a filter, a main controller and a feedback compensator connected in sequence.

[0017] In a preferred embodiment of the model-driven SCR denitration ammonia injection targeted control loop provided by the present invention, the main controller includes a system gain (K c ), IMC (Q(s)) and ideal model (G i (s)), wherein the two-degree-of-freedom parameters λ and α in the IMC are used to adjust the signal tracking and anti-interference performance of the controller respectively, and the function (C(s)) of the main controller is obtained according to the simulation of the actual operation data on site:

[0018]

[0019] Where: K is the object gain, T is the time parameter, and L is the lag time.

[0020] In a preferred embodiment of the model-driven SCR denitration and ammonia injection targeted control loop provided by the present invention, the feedback compensator forms a closed-loop control loop by adding a negative feedback link to the controlled object.

[0021] Compared with the prior art, the beneficial effects of the model-driven SCR denitrification ammonia injection targeted control loop provided by the present invention are: the present invention uses the MD-TDOF-PID control system predictive control model to control the SCR system outlet NO x The method of concentration was used to design and configure the complete logic of the predictive control model, and it was successfully applied to actual projects to achieve a stable SCR denitrification control system outlet NO xThe function of the unit operating parameters such as concentration, under the premise of strictly meeting the national environmental protection requirements, effectively reduces the amount of ammonia injection, improves the operation economy of the unit, reduces the workload of operators, and improves the economic operation level of the unit. It has the following advantages:

[0022] 1) Use dynamic soft measurement technology to effectively solve NO x The hysteresis problem of the measurement signal is solved, which improves the accuracy of the ammonia injection under variable load conditions;

[0023] 2) To overcome the long delay and hysteresis characteristics of the SCR denitrification system, a multi-model adaptive control scheme based on model drive was applied, which effectively improved the dynamic quality of the closed-loop control system and increased the NO x Control accuracy;

[0024] 3) It overcomes the disadvantages of the SCR denitration system CEMS device not refreshing data and not tracking control during purging and failure, and adopts online reverse deduction logic to solve problems such as poor control quality during purging;

[0025] 4) The input and output data of the SCR denitrification system and the chimney inlet data were introduced to construct a soft measurement model of ammonia slip, reduce the ammonia injection flow rate, and improve the economic level of the unit.

[0026] The present invention can identify online parameter models such as unit load, secondary air volume, boiler load, SOFA secondary air door opening, grinding group primary air volume, etc., and at the same time use the online identification model to make real-time corrections to the adaptive data model parameters to achieve precise control of the ammonia injection amount. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0028] Figure 1 It is the SAMA control logic diagram of SCR ammonia injection amount with fixed ammonia nitrogen molar ratio in traditional control mode;

[0029] Figure 2 The traditional control method uses fixed outlet NO x Concentration of SCR ammonia injection amount SAMA control logic diagram;

[0030] Figure 3 It is a structural block diagram of a model-driven SCR denitration and ammonia injection targeted control loop provided by the present invention;

[0031] Figure 4 yes Figure 3 A structural diagram of the dynamic prediction model provided;

[0032] Figure 5 yes Figure 4 The structural block diagram of the MD-TDOF-PID controller provided;

[0033] Figure 6 yes Figure 3 The SAMA control logic diagram of SCR denitrification and ammonia injection based on model driven is provided. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] According to the reaction mechanism and operation data of SCR denitrification, the effects of SCR inlet flue gas parameters and ammonia slip at SCR denitrification outlet on denitrification efficiency are studied, and the optimal ammonia / nitrogen molar ratio is determined according to the control requirements of SCR outlet NOx; and the accurate measurement of flue gas flow is one of the key parameters for calculating the amount of ammonia injection. Due to the inaccuracy of the flue gas flow measurement equipment, the present invention performs online soft measurement of flue gas flow based on relevant operation data to improve the accuracy of flue gas flow data; in addition, due to NO x The measurement of parameters generally has a large hysteresis characteristic. In order to ensure that the dynamic process of the flue gas at the SCR inlet meets the requirements of the optimal ammonia / nitrogen molar ratio, the NO x Dynamic estimation of parameters to obtain real-time NO in flue gas x concentration; on the above basis, the theoretically required ammonia injection amount is calculated as the feedforward control signal of the ammonia injection control system.

[0036] SCR outlet NO x Affected by various factors, in order to overcome the influence of various uncertain factors, ensure that SCR outlet NO x To operate near the given value, this project adopts a model-driven targeted control strategy, as follows:

[0037] Firstly, the ammonia injection amount disturbance test was carried out under different load conditions, and the ammonia injection amount and SCR outlet NO under different load conditions were identified by genetic algorithm. x The transfer function model between

[0038] Then, the model-driven PID controller is designed for each load condition;

[0039] Finally, a closed-loop adaptive control strategy for NOx at the outlet of the SCR reactor is implemented through a multi-model weighted control method.

[0040] The output of the controller serves as the feedback control signal of the ammonia injection control system. The feedforward control signal and the feedback control signal together constitute the real-time set value of the ammonia injection amount. The actual value of the ammonia injection amount is required to quickly track the set value to ensure the optimal ammonia / nitrogen molar ratio. Ultimately, an optimized control strategy for the SCR denitrification process is formed, and the effectiveness of the control strategy is verified on the Matlab platform. Finally, the optimized control strategy is converted into DCS control logic and applied to the SCR denitrification system to ensure its safe and stable operation.

[0041] The control technology scheme of the present invention is as follows:

[0042] See also Figure 3 The model-driven SCR denitrification ammonia injection targeted control loop includes an SCR steady-state model, a dynamic prediction model, a targeted control model, an SCR inlet NO x Concentration soft-sensing model and SCR ammonia slip soft-sensing model, where:

[0043] SCR denitration system outlet NO x The concentration is added to the output value of the dynamic prediction model and used as the input signal of the SCR steady-state model. The SCR steady-state model gives a reference instruction for the targeted control model to adjust the ammonia demand according to the direction and size of the input signal and the deviation degree in the previous and next cycles. At the same time, the reference instruction is also used as the dynamic prediction model, SCR inlet NO x The signal input value of the concentration soft measurement model and the SCR ammonia slip soft measurement model, the SCR inlet NO x The signal outputs of the concentration soft-sensing model and the SCR ammonia slip soft-sensing model are fed back to the dynamic prediction model.

[0044] SCR denitration reactor inlet NO x The dynamic evaluation quality of NO is extremely important for the SCR control effect. x The measurement of NO content has a large delay, so the SCR denitrification system inlet NO x The dynamic estimation of NO content becomes an important factor affecting the regulation quality evaluation of the denitrification control system. Therefore, according to the operating data of supercritical units under different combustion conditions, principal component analysis and multivariate process monitoring are used to select the factors affecting the NO content at the inlet of the SCR denitrification reactor. x The dominant factor of the content is: xThe concentration soft-sensing model is based on the output value of the SCR steady-state model, combined with the unit load and boiler load, SOFA valve opening, and dynamic changes in primary and secondary air volume, and uses multivariate process detection to select the factors that affect the inlet NO of the SCR denitrification reactor. x The dominant factor of NO content in the reactor inlet was established by using the least squares support vector machine. x The prediction model of the content.

[0045] In the process of principal component analysis and multivariate process monitoring, it is first necessary to find several auxiliary variables that are closely related to the measured variable and easy to obtain, and then establish a relevant model based on these auxiliary variables to estimate the target variable. x The dominant factor of the content. Using the dominant element and the reactor inlet NO after T seconds x The content data were used as training data for the least squares support vector machine prediction model.

[0046] According to the commissioning experience and the reaction mechanism of the unit's denitrification reaction zone, the initial auxiliary variables are initially selected to establish the model: total air volume, total coal volume, air-to-coal ratio, air preheater secondary air temperature, unit load, secondary air volume, ratio of burnout damper opening to the total opening of all dampers, weighted sum of instantaneous coal volume of each mill, weighted sum of primary air volume of each mill, and air-to-coal ratio of each mill A, B, C, D, and E, a total of 15 auxiliary variables. (Note: Considering the impact of the instantaneous coal volume and primary air volume of each mill on the NO at the reactor inlet, the initial coal volume of each mill is significantly increased, and the initial air volume of each mill is significantly increased. x The weighted sum of the instantaneous coal volume of each mill = 2*A instantaneous coal volume + B instantaneous coal volume + C instantaneous coal volume + D instantaneous coal volume + 4*E instantaneous coal volume; the weighted sum of the primary air volume of each mill = 0.3*A mill primary air volume + 0.1*(B mill primary air volume + C mill primary air volume + D mill primary air volume) + 5*E mill primary air volume) In order to be able to obtain the current reactor inlet NO through the estimation model x The actual content of NO is not the NO content of 40 to 60 seconds before the reactor inlet shown by the current DCS. x When selecting model training data, the reactor inlet NO x The content data should be the data of the auxiliary variable after 40 to 60 seconds.

[0047] Since the current ammonia escape measurement instrument is not representative and is easily affected by temperature and other factors, especially in the process of mixing ammonia and flue gas, the diffusion of the chemical reaction products affects the measurement process and has a large inertia on the measurement results, resulting in inaccurate instantaneous ammonia escape measurement. Generally, the ammonia escape value measured in the denitrification control area is relatively small, and its magnitude is below ppm, which makes its measurement more difficult. The current measurement methods are mainly based on laser in-situ method, laser spectrum analysis method, and chemiluminescence analysis method. Its measurement methods have limitations to varying degrees, low sensitivity, sample gas is greatly affected by field conditions, and has large delays. Based on this, the SCR ammonia escape soft measurement model of the present invention is based on the measured reactor inlet and outlet NO x The theoretical required ammonia injection amount is calculated based on the content and flue gas volume, the actual ammonia injection amount is compared with the theoretical required ammonia injection amount, and the final ammonia escape amount is obtained by averaging over a certain period of time.

[0048] The online soft measurement method of ammonia escape is adopted to compensate for the measurement limitations of the device. The specific treatment plan is as follows:

[0049] It is known that the NO in the dry flue gas at the reactor inlet standard condition with an oxygen content of 6% x Content The actual oxygen content of NO x The content is calculated by the following formula:

[0050]

[0051] In the denitration system, the denitration reduction reaction occurs at the furnace outlet (for SCR), and the actual nitrogen oxide component in the flue gas is still NO, which accounts for about 95% of the total volume. 2 It accounts for about 5% of the total volume. When calculating the reducing agent consumption, the boiler NO x Emission boundary conditions (i.e. furnace NO x Emissions) are based on NO 2 Mass concentration (mg / Nm 3 ) is taken as the benchmark, so the NO and NO of the flue gas at the furnace outlet can be calculated according to the following formula 2 content:

[0052]

[0053] Where: NO in flue gas x Concentration, (mg / Nm 3 , dry flue gas under standard conditions and actual oxygen content);

[0054] C NO is the NO concentration (mg / Nm3 );

[0055] NO is the standard state and actual dry flue gas oxygen content. 2 Concentration (mg / Nm 3 );

[0056] 1.34 is the volume fraction (ppmv) of NO converted to mass concentration (mg / Nm 3 ) coefficient;

[0057] 2.05 = NO 2 The volume fraction (ppmv) was converted into mass concentration (mg / Nm 3 ) coefficient.

[0058] Since the flue gas measurement point is located at the inlet of the denitrification reactor, and the oxygen content of the flue gas at the inlet and outlet of the denitrification reactor is different, it is only necessary to measure the NO x The content value is converted to the oxygen content of the flue gas at the reactor inlet.

[0059]

[0060] In the formula, NO at the reactor outlet x The content is converted to the value under the oxygen content of the inlet flue gas;

[0061] is the NO at the reactor outlet under actual oxygen content x content;

[0062] is the oxygen content of flue gas at the reactor inlet;

[0063] is the oxygen content of the flue gas at the reactor outlet. The above formulas respectively give the NO content at the reactor outlet (converted to the value of the inlet flue gas oxygen content) and NO 2 The contents (converted to the values ​​of inlet flue gas oxygen content) are as follows:

[0064]

[0065] In the SCR process, NO and NO 2 The reaction of other nitrogen oxides with ammonia is very complicated, but the main reactions can be expressed in the following two chemical equations:

[0066] 4NO+4NH 3 +O 2 →4N 2 +6H 2 O

[0067] 2NO2 +4NH 3 +O 2 →3N 2 +6H 2 O

[0068] According to the chemical reaction equation, 1 mole of NO and 1 mole of NH 3 Reaction, 1 mol of NO 2 With 2 mol NH 3 Reaction, so the consumption of ammonia when the reaction is complete can be calculated according to the following formula:

[0069]

[0070] Where: W is NH 3 Consumption (kg / h); V q is the flue gas flow rate at the reactor inlet (Nm 3 / h, dry flue gas under actual oxygen content); C NO is the NO concentration in the flue gas at the reactor inlet (mg / Nm 3 , dry flue gas at actual oxygen content); NO in the flue gas entering the reactor 2 Concentration (mg / Nm 3 , dry flue gas under actual oxygen content).

[0071] From the above formula, we can get:

[0072]

[0073] Where: 17 is NH 3 Molecular mass, 30 is the molecular mass of NO, 46 is the molecular mass of NO 2 Molecular mass.

[0074] Specific implementation:

[0075] Average smoke volume The time is 30s; (the k in the brackets represents the current time, and ki represents the instantaneous value before the i sampling time, the same below); 7200: 1 hour has 7200s, and 1s is a sampling period.

[0076] Average amount of ammonia injected into the SCR reactor inlet within 30 seconds:

[0077]

[0078] 10 3 The unit conversion factor for converting mg to g.

[0079] Theoretically, the average amount of ammonia sprayed within 30 seconds is:

[0080]

[0081] Then the amount of ammonia that did not participate in the reaction within 30 seconds is: W zj -W zlj , g;

[0082] The ammonia slip during this time period is:

[0083] 22.4 Under standard conditions (STP) [0℃ (273K), 1.01×10^5Pa], the volume occupied by 1 mole of any ideal gas is approximately 22.4 liters. This volume is called the molar volume of the gas, and the unit is L / mol (liter / mole).

[0084] The targeted control model is a directional model based on the DCS system control software. Its main principle is: through the entrance NO at a certain moment x The concentration, ammonia consumption, and ammonia escape volume can quantitatively give the required ammonia demand to prevent the SCR denitration system from measuring NO x The uneven concentration results in incomplete reaction or excessive ammonia injection.

[0085] The ultimate purpose of the present invention is to control the NO x The concentration is within the set value range, and the ammonia injection amount of the SCR denitrification system is consistent with the outlet NO x The online measurement of concentration changes is a long-delay link, and its characteristics are also affected by parameters such as random group load. Therefore, according to the disturbance test data of ammonia injection at different loads, the model identification of the software object is carried out online, and the dynamic characteristic transfer function of the SCR denitrification system under three typical load conditions is obtained, as shown in Table 1:

[0086] Table 1: Dynamic transfer function of SCR denitrification system

[0087]

[0088]

[0089] is the transfer function of ammonia injection flow rate change;

[0090] G SCR (s) is SCR outlet NO x Concentration change transfer function.

[0091] According to the dynamic transfer function of the SCR denitrification system in Table 1, under the disturbance of the opening of the ammonia injection regulating valve, the outlet NO xThe content has a large delay, while the amount of ammonia injection has an obvious leading effect. Therefore, the present invention adopts cascade control as the basic structure of the SCR denitration control system, the outer loop adopts MD-TDOF-PID control, and the inner loop adopts PI control. Therefore, the dynamic prediction model includes an inner loop control loop and an outer loop control loop, such as Figure 4 shown.

[0092] The inner control loop is composed of the injection ammonia flow rate change transfer function And PI controller, It is the transfer function of the ammonia injection control valve in the process of changing the opening, corresponding to the change of the ammonia injection flow rate. In different load sections, the transfer function has different differences, but the difference part has certain regularity. In the actual application process, since the inner loop changes little with the group load, a fixed gain PI controller is often used in the control process.

[0093] The outer control loop is composed of SCR outlet NO x Concentration change transfer function (G SCR (s)), MD-TDOF-PID controller and load weighter, G SCR (s) is the transfer function established by the change of NOx concentration at the reactor outlet caused by the change of ammonia injection flow rate. The load weighter is based on the actual load Pe in this load segment, which can be defined as [Pe1, Pe2]. The weight of the actual load at Pe1 and Pe2 is defined as x1 and x2, then (Note: abs is the absolute value function), the output value of the load weighter is V, which is Ui is the output value of the MD-TDOF-PID controller under different working conditions. It can be seen that the weighted controller output value V has a smooth transition effect, so that it has numerical output within the full load range, avoiding jump fluctuations caused by data switching under different working conditions.

[0094] Figure 4 In: y is the output value of the dynamic prediction model, which is NO x The concentration setting value is Figure 3 Operate with the feedback signal.

[0095] From Table 1, we can see that the controlled object of the inner loop is The random group load changes very little, so the inner loop uses a fixed gain PI controller to meet the requirements. On the basis of the inner loop parameters being tuned, it is equivalent to a gain link K p , and G SCR (s) are taken as the controlled objects of the system to design the MD-TDOF-PID controller.

[0096] MD-TDOF-PID has excellent control performance, can improve or stabilize the characteristics of the controlled object, quickly track the changes in the set value, and has strong interference suppression capabilities. It has good robustness for this kind of SCR denitrification control system with large delay and large inertia.

[0097] The structural block diagram of the MD-TDOF-PID control system is as follows: Figure 5 The control loop of the MD-TDOF-PID controller includes a filter, a main controller and a feedback compensator connected in sequence. In the figure, r is the outlet NO of the SCR denitration system. x The concentration is filtered to generate data with adjustable response speed, thus producing a smooth engineering quantity for subsequent controllers.

[0098] The feedback compensator forms a closed-loop control loop by adding a negative feedback link to the controlled object, that is, the feedback compensator adds a PD feedback link F(s) to the controlled object P(s), so that the unstable object can be stabilized, interference factors are eliminated, and the dynamic characteristics of the object are improved and stabilized. After being compensated by the feedback compensator, P(s) generates a relative controlled object G(s), which is approximately described by the first-order inertial time-lag object in engineering, that is:

[0099]

[0100] Where: K is the object gain, T is the time constant, and L is the lag time.

[0101] The main controller includes the system gain (K c ), IMC device (Q(s)) and ideal model (Gi(s)), wherein the two-degree-of-freedom parameters λ and α in the IMC device are used to adjust the signal tracking and anti-interference capabilities of the controller, respectively.

[0102] The ideal model is set according to the generalized object G(s). According to the internal model control principle, the parameters in the control system are set as:

[0103] K c =1 / K,T c =T,L c =L

[0104] According to the simulation of the actual operation data on site, the function of the main controller (C(s)) is obtained:

[0105]

[0106] The closed-loop transfer function of the system is simplified to:

[0107]

[0108] The filter is mainly used to adjust the response speed of the system output. Through this link, the system input function is:

[0109]

[0110] The system disturbance transfer function is:

[0111]

[0112] when When G r (s)=1,G d (s) = 0, so according to the parameter settings of the above system, the control system can track all input signals and suppress interference well.

[0113] The inner loop is tuned according to the attenuation curve method (attenuation rate ψ = 0.95), and the parameters of the PI controller are: δ = 0.47, T i =0.702. Under the same parameter setting, the unit step response of the inner loop under the three working conditions can reach a steady state of 1 at t = 3s, meeting the requirements of the fast follow-up system of the outer loop. On the basis of the inner loop being tuned, it is equivalent to a gain of 1, and the outer loop object G SCR (s) together constitute the controlled object P(s) in the control system. On this basis, the MD-TDOF-PID controller is designed and its parameters are tuned.

[0114] The design methods of various parameters in this controller are as follows:

[0115] The controlled object P(s) is expanded into the form of a denominator polynomial through Taylor formula, that is:

[0116]

[0117] Feedback Compensator:

[0118]

[0119] When the controlled object is in an ideal state of stability and no overshoot, it can be expressed as:

[0120]

[0121] Where: σ is the response time, β 2 and β 3 is the state factor of the response, β 2 =0.42, β 3 =0.091.

[0122] After P(s) is compensated by the PD feedback F(s), the obtained generalized controlled object G(s) (represented by the parameters in P(s) and F(s)) is model matched with its ideal state formula, and the calculation formula for the parameters in F(s) is obtained as follows:

[0123]

[0124] The generalized controlled object approximated by first-order inertia plus time delay is expanded according to Taylor's formula at s = 0, that is:

[0125]

[0126] By matching the above formula with the ideal state formula, the calculation formula of the parameters in the main controller C(s) is obtained as follows:

[0127]

[0128] The calculation formulas for the two degree of freedom parameters λ and α in Q(s) are:

[0129]

[0130] Where: u lim is the maximum value allowed by the manipulated variable u, and ω is the frequency of the interference.

[0131] According to the above formula, the values ​​of the feedback compensator F(s), the approximate generalized object G(s), and the parameters λ and α under the three typical working conditions in Table 1 are obtained, as shown in Table 2 below:

[0132] Table 2 Parameter configuration of MD-TDOF-PID control system

[0133]

[0134]

[0135] Depend on Figure 5 It can be seen that the operating variable u that controls the opening of the ammonia injection regulating valve is the controller output u under three typical working conditions. i The weighted sum of is:

[0136]

[0137] Where: w i is the output weight of the i-th controller, and its value is directly related to the switching quality of the system.

[0138] As can be seen from Table 2, the controlled object of the SCR denitrification system varies greatly with the load, so the weighted device uses the load for scheduling. In actual operation, the unit load P can only be in any two similar load intervals, such as [P 1 , P2 ], so only the output weights of the two controllers close to the current load are not zero. 1 , P 2 The corresponding weight w 1 、w 2 The calculation method is:

[0139]

[0140] The output u of the global controller adopts the weighted output of each local controller, which can effectively overcome the oscillation that occurs during the changing working conditions and make the control output transition smoothly.

[0141] Finally, the SAMA control logic diagram of the model-driven SCR denitration and ammonia injection of the present invention is as follows: Figure 6 shown.

[0142] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for constructing a model-driven SCR denitration ammonia injection targeted control loop, characterized in that: Including SCR steady-state model, dynamic prediction model, targeted control model, SCR inlet NO x Concentration soft-sensing model and SCR ammonia slip soft-sensing model, where: The dynamic prediction model includes an inner control loop and an outer control loop, wherein the inner control loop is composed of an injection ammonia flow rate change transfer function and PI controller; the outer control loop is composed of SCR outlet NO x Concentration Change Transfer Function , MD-TDOF-PID controller and load weighter; The SCR inlet NO x The concentration soft-sensing model is based on the output value of the SCR steady-state model, combined with the unit load and boiler load, SOFA valve opening, and dynamic changes in primary and secondary air volume, and uses multivariate process detection to select the factors that affect the inlet NO of the SCR denitrification reactor. x The dominant factor of NO content in the reactor inlet was established by using the least squares support vector machine. x Prediction model of content; SCR denitration system outlet NO x The concentration is added to the output value of the dynamic prediction model and used as the input signal of the SCR steady-state model. The SCR steady-state model gives a reference instruction for the targeted control model to adjust the ammonia demand according to the direction and size of the input signal and the deviation degree in the previous and next cycles. At the same time, the reference instruction is also used as the dynamic prediction model, SCR inlet NO x The signal input value of the concentration soft measurement model and the SCR ammonia slip soft measurement model, the SCR inlet NO x The signal outputs of the concentration soft-sensing model and the SCR ammonia slip soft-sensing model are fed back to the dynamic prediction model.

2. The method for constructing a model-driven SCR denitration ammonia injection targeted control loop according to claim 1, characterized in that: The SCR ammonia slip soft measurement model is based on the measured reactor inlet and outlet NO x The theoretical required ammonia injection amount is calculated based on the content and flue gas volume, the actual ammonia injection amount is compared with the theoretical required ammonia injection amount, and the final ammonia escape amount is obtained by averaging over a certain period of time.

3. The method for constructing a model-driven SCR denitration ammonia injection targeted control loop according to claim 1, characterized in that: The targeted control model is a directional model based on the DCS system control software. x The concentration, ammonia consumption and ammonia escape can quantitatively give the required ammonia demand.

4. The method for constructing a model-driven SCR denitration ammonia injection targeted control loop according to claim 1, characterized in that: The ammonia injection flow rate change transfer function It is the transfer function of the ammonia injection flow rate change when the ammonia injection regulating valve changes its opening; The SCR outlet NO x Concentration Change Transfer Function It is the transfer function established by the change of NOx concentration at the reactor outlet caused by the change of ammonia injection flow rate.

5. The method for constructing a model-driven SCR denitration ammonia injection targeted control loop according to claim 1, characterized in that: The control loop of the MD-TDOF-PID controller includes a filter, a main controller and a feedback compensator connected in sequence.

6. The method for constructing a model-driven SCR denitration ammonia injection targeted control loop according to claim 5, characterized in that: The main controller includes a system gain 、IMC and ideal model , where the two-degree-of-freedom parameters λ and α in the IMC are used to adjust the signal tracking and anti-interference performance of the controller respectively, and the function of the main controller is obtained according to the simulation of the actual operation data on site. : Where: K is the object gain, T is the time parameter, and L is the lag time.

7. The method for constructing a model-driven SCR denitration ammonia injection targeted control loop according to claim 6, characterized in that: The feedback compensator forms a closed-loop control circuit by adding a negative feedback link to the controlled object.

Images