Flue gas deacidification feed-forward and feedback collaborative control system based on salt concentration feedback

Abstract

The application relates to the technical field of industrial control, and particularly discloses a flue gas deacidification discharge feed-forward-feedback collaborative control system based on salt concentration feedback, which collects the circulating liquid salt concentration of a wet deacidification tower, the opening degree of a water supplement valve, and the flue gas flow and temperature at the inlet of a quenching tower; reconstructs the salinity signal through untraceable Kalman filtering, filters out the false peak introduced by water supplement dilution; takes the opening degree of the water supplement valve as input excitation, takes the reconstructed salinity and the outlet temperature of the quenching tower as response, performs gain matrix analysis based on a relative gain array to obtain a coupling coefficient, generates a leading feed-forward compensation amount through asymmetric phase compensation and superimposes the leading feed-forward compensation amount on the frequency setting value of a reuse pump; discriminates the interference type according to the change rate of the reconstructed salinity, freezes the feedback integration when the change rate exceeds the limit and widens the proportional band, dominates the water spraying increment by the feed-forward instruction, restores the integration through a slope after the change rate falls back, and outputs a final collaborative instruction; and the application realizes stable control of the temperature, liquid level and salinity under water supplement disturbance and zero-emission continuous operation.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, specifically to a feedforward-feedback coordinated control system for flue gas desulfurization emissions based on salt concentration feedback. Background Technology

[0002] In the process of treating flue gas from hazardous waste incineration, the wet deacidification tower uses sodium carbonate as a dealkali removal agent to absorb acidic components in the flue gas. After the circulating liquid absorbs the acidic gas, the salt concentration continuously increases. In order to achieve zero discharge of wastewater, the high-salt circulating liquid discharged from the deacidification tower is used as the spray water for the quench tower. The high-temperature flue gas heat is used to evaporate the water, and the salt is discharged with the ash.

[0003] In this process, the water spray volume of the quench tower must meet two control requirements simultaneously: the water spray volume must match the heat load of the inlet flue gas in order to control the outlet temperature of the quench tower within the safe operating range of the bag filter. The spray volume needs to be matched with the salt concentration of the circulating liquid in the deacidification tower to ensure a balance between the system's salt discharge and salt production.

[0004] The two requirements are achieved through frequency adjustment of the reuse pump and water supply valve adjustment of the deacidification tower, respectively, and both share the same high-salt wastewater flow path.

[0005] In the process of reusing high-salt wastewater from flue gas desulfurization in hazardous waste incineration, when the water supply valve of the desulfurization tower is opened at a high flow rate due to liquid level control, a large amount of low-salt process water is injected into the circulation system in a short period of time, causing the online salinity feedback signal to drop sharply due to a non-real salinity deficit. The existing single-loop proportional-integral-derivative controller cannot distinguish between this dilution disturbance and the real salinity deficit, resulting in the frequency of the reuse pump being incorrectly reduced in the opposite direction. This misregulation also causes the outlet temperature of the quench tower to exceed the limit due to the sudden reduction in water spray volume, the bag filter to overheat and trip, and the liquid level of the circulation pool to rise and overflow due to the reduction in reuse volume, forcing emergency discharge. This results in a multi-variable coupled uncontrollable problem where the closed-loop circulation of zero wastewater discharge is forced to be interrupted during the water supply disturbance. Summary of the Invention

[0006] The purpose of this invention is to provide a feedforward-feedback coordinated control system for flue gas desulfurization emissions based on salt concentration feedback, so as to solve the problems mentioned above.

[0007] The objective of this invention can be achieved through the following technical solutions: The feedforward-feedback coordinated control system for flue gas desulfurization emissions based on salt concentration feedback includes: The data acquisition module collects the salt concentration measurement value of the circulating liquid in the wet deacidification tower, the opening signal of the water supply valve, and the flow rate and temperature data of the flue gas at the inlet of the quench tower. The salinity signal reconstruction module constructs a multi-dimensional state vector containing instantaneous salinity value, liquid level change rate and cumulative influence coefficient of water replenishment based on the collected salt concentration measurement value. It uses the water replenishment valve opening signal for auxiliary correction and dynamically reconstructs the salinity signal through unscented Kalman filtering to filter out false concentration spike noise introduced by water replenishment dilution and outputs the reconstructed salinity value. The coupling characteristic analysis module, based on the water supply valve opening signal and flue gas flow and temperature data, uses the reconstructed salinity value as feedback input, applies step excitation to each steady-state condition to obtain the open-loop gain sequence, performs determinant discrimination and reversible correction on the square matrix to obtain the reverse action coefficient, compares the reverse action coefficient with the open-loop gain to solve the relative gain matrix, and obtains the coupling coefficient matrix of each condition after piecewise linearization. The feedforward compensation generation module takes the water supply valve opening signal as a measurable disturbance and allocates the feedforward allocation coefficients of the reuse pump frequency channel and the salt inhibitor injection channel according to the coupling coefficient matrix. The reuse pump frequency uses advanced phase compensation and the salt inhibitor injection channel uses phase compensation to form a dual-channel asymmetric phase compensation network. It generates a feedforward compensation amount that leads the lag response of the quench tower temperature. The feedforward compensation amount is superimposed on the reuse pump frequency set value and outputs a feedforward adjustment command. The collaborative control output module determines the type of disturbance based on the rate of change of the reconstructed salinity value per unit time. When the rate of change exceeds the limit, the feedback integral action is frozen and the proportional band is widened. The feedforward adjustment command dominates the water injection increment during the disturbance period. After the rate of change falls back, the integral action is restored in a ramp manner, and the final collaborative control command for the reuse pump and salt inhibitor injection rate is output.

[0008] As a further aspect of the present invention: the output reconstructed salinity value specifically includes: Using the measured salt concentration as the observation input, a multidimensional state vector containing instantaneous salinity value, liquid level change rate and cumulative influence coefficient of water replenishment is constructed, and the initial state estimate is output. A set of sampling points surrounding the initial state estimate is generated based on a deterministic sampling strategy. Each sampling point is passed through a nonlinear state transition equation to obtain the predicted state distribution and output the predicted state and its covariance. Using the water supply valve opening signal as an auxiliary correction input, the salinity component in the predicted state is updated and the covariance is corrected based on the deviation between the predicted state and the measured salt concentration. The false concentration spike noise introduced by the water supply dilution is filtered out, and the reconstructed salinity value is output.

[0009] As a further aspect of the present invention: the output predicted state and its covariance specifically include: Using each sampling point in the sampling point set as input, the system maps each point through a nonlinear state transition relationship to obtain the predicted salinity component corresponding to each sampling point, and outputs the set of predicted salinity components for each sampling point. The predicted salinity component set of each sampling point is reorganized by mean value, and the estimated mean and variance of the salinity component in the predicted state are calculated. The preliminary predicted distribution and its covariance are output. Symmetric compression correction is applied to the variance of the salinity component estimation to eliminate the non-Gaussian skewness of the distribution introduced by nonlinear mapping during the transmission process, and the corrected prediction state and its covariance are output.

[0010] As a further aspect of the present invention: the output process of the coupling coefficient matrix is ​​as follows: Using the water supply valve opening signal as the input excitation and the reconstructed salinity value and quench tower outlet temperature as the response output, the open-loop gain value of the water supply channel to the salinity channel and temperature channel under each steady-state condition is obtained one by one, and the open-loop gain sequence under each condition is output. Perform matrix inversion on the open-loop gain sequence under each operating condition to obtain the reverse action coefficient of the water supply valve opening change on each response channel. Then compare the reverse action coefficient with the open-loop gain sequence to obtain the relative gain value under each operating condition and output the relative gain matrix under each operating condition. The relative gain matrix under different steady-state conditions is piecewise linearized and rearranged. The interaction strength between the water replenishment loop and the salt balance loop in each operating condition interval is extracted, and the coupling coefficient matrix under each operating condition is output.

[0011] As a further aspect of the present invention: the output process of the reverse action coefficient is as follows: Using the open-loop gain sequence under each operating condition as input, construct a square matrix with the gain transfer relationship between channels as elements; The determinant value of the square matrix is ​​judged. When the determinant value is less than the preset threshold, an equal amount of adjustment value is added to each element of the main diagonal and the determination is repeated until the determinant value falls into the invertible interval, and an invertible square matrix is ​​output. Perform column-by-column principal component elimination and back substitution operations on the reversible matrix to obtain the reverse action coefficients corresponding to each column. Combine the reverse action coefficients of each column in sequence to output the reverse action coefficients of the water supply valve opening change on each response channel.

[0012] As a further aspect of the present invention: the output feedforward adjustment command specifically includes: Using the coupling coefficient matrix as input, the feedforward allocation coefficients for the frequency channel of the reuse pump and the salt inhibitor injection channel are determined based on the change amplitude and change rate of the current water supply valve opening signal, and the feedforward allocation coefficients for the two channels are output. Using the water supply valve opening signal as the input excitation, different lead time constants and lag time constants are assigned according to the phase compensation direction of the two channels. After the excitation signal is processed by differential compensation and inertial delay of the corresponding time constants, the phase pre-adjustment of the two channels is output. The phase pre-adjustment of the two channels is multiplied by the corresponding feedforward allocation coefficient, and the product is then added to the set value of each channel to output the feedforward adjustment command for the reuse pump frequency and the feedforward adjustment command for the salt inhibitor dosing rate.

[0013] As a further aspect of the present invention: the feedforward allocation coefficients for the two output channels specifically include: The reciprocal of the diagonal element values ​​in the coupling coefficient matrix is ​​used as the basic allocation weight of the two channels. The two basic allocation weights are normalized and used as the initial allocation coefficients. The initial allocation coefficients are then output. The difference between the current value of the reuse pump frequency and the upper limit of the frequency is used as the adjustable margin of the reuse pump channel, and the difference between the current value of the salt inhibitor injection rate and the upper limit of the injection rate is used as the adjustable margin of the salt inhibitor channel. The initial allocation coefficient is corrected by the ratio of the adjustable margins of the two channels, and the corrected allocation coefficient is output. The corrected allocation coefficient is compared with the preset maximum capacity ratio of a single channel. When the corrected allocation coefficient exceeds the maximum capacity ratio of the corresponding channel, the allocation coefficient of the channel is reduced to the maximum capacity ratio, and the allocation coefficient of the other channel is supplemented accordingly. The final feedforward allocation coefficient of the two channels is then output.

[0014] As a further aspect of the present invention: the output process of the final cooperative control command is as follows: The rate of change of reconstructed salinity value per unit time is used as input. The rate of change is compared with the preset rate of change limit boundary. When the rate of change exceeds the limit boundary, it is identified as water replenishment and dilution type interference, and the interference type flag is output. Using the interference type flag as the trigger condition, the integral accumulation value in the feedback control is frozen, and the proportional band in the feedback control is widened by a preset step size. The feedback control state after freezing the integral and widening the proportional band is output. The condition for recovery is that the rate of change falls back to within the limit boundary. The integral accumulation value is gradually thawed and restored according to the preset order. After each order is restored, it is re-evaluated whether the rate of change exceeds the limit again, until the integral accumulation value is fully restored. Finally, the final coordinated control command of the reuse pump frequency setting value and the salt inhibitor injection rate setting value is output.

[0015] The beneficial effects of this invention are: (1) This invention introduces the water supply valve opening signal as a measurable disturbance into the feedforward channel, and generates a feedforward compensation amount that leads the lag response of the quench tower temperature by an asymmetric phase compensation network. This allows the frequency of the reuse pump to be pre-adjusted according to the water supply disturbance amplitude before the actual fluctuation of the quench tower outlet temperature occurs, shortening the time when the quench tower outlet temperature deviates from the set value after the water supply valve is activated, reducing the temperature overshoot amplitude, reducing the trigger frequency of the bag filter tripping due to over-temperature, and keeping the temperature control process stable during the water supply disturbance period, thus reducing the frequent intervention of the operator in the temperature circuit.

[0016] (2) The present invention dynamically reconstructs the salt concentration measurement value through unscented Kalman filtering, and filters out the false spike noise caused by water replenishment and dilution before it enters the feedback controller. At the same time, the change rate discrimination logic automatically freezes the feedback integral action during water replenishment disturbance and the feedforward command dominates the control output, so that the feedback controller does not generate reverse adjustment action when the salinity measurement value drops falsely. This avoids the rise in the circulating pool liquid level and emergency discharge caused by the incorrect reduction of the spray volume due to misjudgment of the reuse pump frequency. This enables the zero discharge process to maintain a closed-loop operation state continuously during water replenishment regulation, and the achievement of the zero discharge target is not interrupted by liquid level regulation intervention. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a system block diagram of the present invention; Figure 2 This is a flowchart of the process for outputting the reconstructed salinity value in this invention; Figure 3 This is a flowchart of the output process of the coupling coefficient matrix in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, this invention is a feedforward-feedback coordinated control system for flue gas desulfurization emissions based on salt concentration feedback, comprising: The data acquisition module collects the salt concentration measurement value of the circulating liquid in the wet deacidification tower, the opening signal of the water supply valve, and the flow rate and temperature data of the flue gas at the inlet of the quench tower. The salinity signal reconstruction module uses a dynamic salinity signal reconstruction algorithm constructed by unscented Kalman filtering to filter out false concentration spike noise introduced by water replenishment and dilution, and outputs the reconstructed salinity value based on the collected salt concentration measurement value. The coupling characteristic analysis module, based on the water supply valve opening signal and flue gas flow and temperature data, and with the reconstructed salinity value as feedback input, performs cross-channel dynamic gain matrix analysis based on relative gain array, quantitatively identifies the interaction strength between the water supply loop and the salt balance loop under various operating conditions, and outputs the coupling coefficient matrix under various operating conditions. The feedforward compensation generation module takes the water supply valve opening signal as a measurable disturbance and generates a feedforward compensation amount that leads the quench tower temperature lag response through an asymmetric phase compensation network based on the coupling coefficient matrix. The feedforward compensation amount is superimposed on the reuse pump frequency setpoint and outputs a feedforward adjustment command. The collaborative control output module determines the type of disturbance based on the rate of change of the reconstructed salinity value per unit time. When the rate of change exceeds the limit, the feedback integral action is frozen and the proportional band is widened. The feedforward adjustment command dominates the water injection increment during the disturbance period. After the rate of change falls back, the integral action is restored in a ramp manner, and the final collaborative control command for the reuse pump and salt inhibitor injection rate is output.

[0021] The data acquisition module collects measurements of the salt concentration of the circulating liquid in the wet desulfurization tower, the opening signal of the water supply valve, and the flow rate and temperature data of the flue gas at the inlet of the quench tower, specifically including: Collection of circulating solution salt concentration measurements: An online conductivity meter is installed on the outlet pipeline of the circulating pump in the wet deacidification tower. The range of the conductivity meter is set to 0 to 500 mSiemens per centimeter, and the accuracy is 1% of the full scale. Its electrode is inserted into the main pipeline of the circulating liquid to measure the total dissolved salt concentration of the circulating liquid in a continuous contact manner. The analog current signal output by the conductivity meter is transmitted to the analog input card of the distributed control system via a shielded cable. The distributed control system performs analog-to-digital conversion on the signal at a sampling rate of once per second, and converts the converted digital quantity into a salt concentration measurement value, which is used in subsequent control calculations in grams per liter.

[0022] Acquisition of water supply valve opening signal: An electric regulating valve is installed on the water supply pipeline of the deacidification tower. The electric regulating valve is equipped with a valve position feedback transmitter. The valve position feedback transmitter detects the stroke position of the valve stem in real time and converts it into a DC current signal proportional to the valve opening. After being distributed by a signal isolator, the DC current signal is sent to the analog input card of the distributed control system. The distributed control system collects the signal at a sampling rate of once per second. After calibration and conversion, the water supply valve opening signal is obtained and expressed as a percentage, where 0% corresponds to the valve being fully closed and 100% corresponds to the valve being fully open.

[0023] Data collection of flue gas flow rate and temperature at the quench tower inlet: An averaging pitot tube flowmeter is installed on the straight section of the flue gas duct at the quench tower inlet. The total pressure tap and static pressure tap of the flowmeter are connected to the high-pressure side and low-pressure side of the differential pressure transmitter, respectively. The differential pressure transmitter converts the differential pressure signal into a current signal and sends it to the distributed control system. The distributed control system obtains the flue gas flow measurement value in standard cubic meters per hour based on the pre-input flue gas cross-sectional area and flue gas density parameters and the conversion relationship between differential pressure and flow rate. A sheathed thermocouple is installed on the flue gas wall adjacent to the averaging pitot tube flowmeter. The measuring end of the thermocouple extends into the central area of ​​the flue gas duct. The thermoelectric potential signal generated by the thermocouple is converted into a current signal by a temperature transmitter and sent to the distributed control system. The distributed control system collects the signal at a sampling rate of once per second and obtains the flue gas temperature measurement value in degrees Celsius after conversion using a thermocouple calibration table.

[0024] Please see Figure 2 As shown, in the salinity signal reconstruction module, based on the collected salt concentration measurements, a dynamic salinity signal reconstruction algorithm constructed using unscented Kalman filtering filters out spurious concentration spike noise introduced by water replenishment and dilution, and outputs the reconstructed salinity value, specifically including: Using the real-time salt concentration measurement value collected by the distributed control system as the observation input, a 3-dimensional state vector is constructed. The first dimension of the 3-dimensional state vector is the instantaneous salinity value at the current moment, with the unit set to grams per liter. The second dimension is the rate of change of the deacidification tower liquid level per unit time, with the unit set to meters per second. This rate of change is obtained by the distributed control system through least squares linear fitting based on the historical data of the liquid level gauge over the past 60 seconds. The third dimension is the cumulative influence coefficient of the makeup water, which is initially set to 0. Its physical meaning represents the degree of cumulative dilution effect of the makeup water valve opening action on the salinity of the circulating liquid over the past 300 seconds. The current salt concentration measurement value is assigned to the first dimension of the state vector, the liquid level change rate obtained by fitting at the current moment is assigned to the second dimension of the state vector, and the third dimension is kept at 0, thus forming the initial state estimate.

[0025] A set of sampling points surrounding the initial state estimate is generated based on a deterministic sampling strategy. The number of sampling points in this set is equal to twice the dimension of the state vector plus 1, i.e., 7 sampling points. The first sampling point is the initial state estimate itself, and the remaining 6 sampling points are generated by offsetting the initial state estimate by a preset distance along each dimension in the positive and negative directions, respectively. The preset distance is equal to the product of the square root of the diagonal element of the covariance matrix of the corresponding dimension of the state vector and a preset distribution ratio, which is the square root of the dimension of the state vector, i.e., 1.73. Each of the 7 sampling points is mapped one by one through a nonlinear state transition relation. The input of the nonlinear state transition relation is the instantaneous salinity value, the rate of change of liquid level, and the cumulative influence coefficient of water replenishment in the sampling point, and the output is the predicted salinity component. The predicted salinity components are calculated as follows: the instantaneous salinity value at the sampling point is added to the salinity dilution increment obtained by multiplying the liquid level change rate by the unit time interval, and then the salinity reduction obtained by multiplying the preset evaporation salt removal coefficient by the cumulative influence coefficient of water replenishment is subtracted. The preset evaporation salt removal coefficient is 0.3, and the unit time interval is 1 second. After the above mapping, the predicted salinity components corresponding to the 7 sampling points are obtained, which constitute the predicted salinity component set.

[0026] The predicted salinity component set is recombined by mean normalization. The specific process of mean normalization is as follows: the values ​​of the 7 predicted salinity components are added together and then divided by 7 to obtain the estimated mean of the salinity components in the predicted state; the sum of the squares of the differences between each of the 7 predicted salinity components and the mean is calculated and then divided by 6 to obtain the estimated variance of the salinity components; the estimated mean of the salinity components is used as the initial predicted state output, and the estimated variance of the salinity components is used as the initial predicted covariance output.

[0027] A symmetric compression correction is performed on the estimated variance of the salinity component. The specific process of this symmetric compression correction is as follows: the cross covariance represented by each off-diagonal element in the preliminary prediction covariance matrix is ​​used as the basis for determining the compression direction. For directions with a cross covariance greater than 0, the corresponding covariance element in the preliminary prediction covariance matrix is ​​multiplied by the first compression coefficient. For directions with a cross covariance less than 0, the corresponding covariance element in the preliminary prediction covariance matrix is ​​multiplied by the second compression coefficient. The first compression coefficient is the reciprocal of the sum of the number of sampling points and the square of three times the preset distribution ratio, and is calculated to be 0.25. The second compression coefficient is the reciprocal of the difference between the number of sampling points and the square of three times the preset distribution ratio, and is calculated to be 0.28. Finally, the corrected prediction state and its covariance are obtained. The variance components of each dimension of the corrected prediction covariance are all smaller than the variance components of the corresponding dimension in the preliminary prediction covariance, thereby eliminating the non-Gaussian skewness of the distribution introduced by nonlinear mapping during the transmission process.

[0028] The opening signal of the water supply valve is used as an auxiliary correction input. The specific function of this auxiliary correction input is as follows: Based on the deviation between the salinity component in the corrected predicted state and the current period's salt concentration measurement value, a salinity component deviation is generated. This deviation is then allocated to the salinity component in the corrected predicted state with a preset correction weight. The preset correction weight is determined by the ratio of the corrected prediction covariance to the measurement noise covariance. The measurement noise covariance is taken as the square of the error corresponding to the factory accuracy of the online conductivity meter, that is, 1% multiplied by 500 millisiemens per centimeter of full scale, which is calculated to be 25. The salinity component after the correction weight allocation is output as the reconstructed salinity value for the current period. At the same time, the corrected prediction covariance is updated according to a preset attenuation coefficient and retained to the next sampling period. The preset attenuation coefficient is 0.98. During the reconstruction process, the low-salt process water injected when the water supply valve is opened causes a false spike in the conductivity meter reading, resulting in a rapid drop. The rate of change of the salt concentration corresponding to this false spike is much greater than that under normal operating conditions. This leads to an imbalance in the physical constraint relationship between the instantaneous salinity change, the liquid level change rate, and the cumulative influence coefficient of the water supply in the state vector. After the mapping and symmetric compression correction of the nonlinear state transition relationship, the predicted salinity component corresponding to this false spike deviates from the true salinity distribution. In the weight allocation correction step, its contribution to the reconstructed salinity value is reduced, thereby achieving the effect of filtering out the false concentration spike noise introduced by the water supply dilution.

[0029] Please see Figure 3 As shown, in the coupling characteristic analysis module, based on the water supply valve opening signal and flue gas flow and temperature data, and using the reconstructed salinity value as feedback input, a cross-channel dynamic gain matrix analysis based on a relative gain array is performed to quantitatively identify the interaction strength between the water supply loop and the salt balance loop under various operating conditions, and output the coupling coefficient matrix under each operating condition, specifically including: Using the water supply valve opening signal as the input excitation and the reconstructed salinity value and quench tower outlet temperature as the response outputs, under various steady-state operating conditions where the incinerator maintains stable operation, the distributed control system applies a step change to the water supply valve. The step amplitude is set to 5% of the valve's full opening stroke. The changes in each response output before and after the step application are recorded when the new steady state is reached. Each response output change is divided by the change in the water supply valve opening to obtain the open-loop gain value of the water supply channel to the salinity channel and temperature channel under that operating condition. This process is repeated for multiple steady-state operating points of the desulfurization tower's operating load, where each steady-state operating point includes at least 70%, 80%, 90%, 100%, and 110% of the design load, to obtain the open-loop gain sequence for each operating condition. The open-loop gain value for each operating condition is calculated according to the following formula: ; in, Indicates the first The open-loop gain of each response output channel under the current steady-state condition. The value is 1 or 2, corresponding to the salinity channel and the temperature channel, respectively. This represents the total number of times the step test is repeated under the current operating conditions, with a value of 5. Indicates the first The confidence weighting coefficient for the second step test is determined by the initial fluctuation amplitude of the output variable at the start of the test. When the initial fluctuation amplitude is less than 1% of the steady-state value... When the value is 1.0, and the initial fluctuation amplitude is between 1% and 3% of the steady-state value... The value is 0.5; Indicates the first The transition time taken for the response output to reach 90% of the new steady-state value from the moment the step is applied in the sub-step test is expressed in seconds. Indicates all under the current operating conditions The maximum transition time in this experiment, in seconds; Indicates the first In the second step test, the first The steady-state change of each response output channel before and after a step is applied; Indicates the first The step change in the opening of the water supply valve during the secondary step test is expressed as a percentage.

[0030] Using the open-loop gain sequence under various operating conditions as input, a square matrix is ​​constructed with the gain transfer relationship between channels as elements. The number of rows and columns of this square matrix is ​​equal to the number of response output channels plus one, i.e., a third-order square matrix. The first two rows and two columns are filled with the open-loop gain values ​​between each channel, the third row and third column are filled with the value 1, and the remaining off-diagonal positions are filled with the value 0. The determinant value of this square matrix is ​​judged. The determinant value of the square matrix is ​​calculated. When the absolute value of the determinant value is less than the preset threshold of 0.001, the square matrix is ​​judged to be ill-conditioned. At this time, an equal adjustment value of 0.0001 is added to each element of the main diagonal of the square matrix and the determinant value is recalculated. This addition and judgment process is repeated until the absolute value of the determinant value is greater than or equal to 0.001, and an invertible square matrix is ​​output.

[0031] The reversible matrix is ​​subjected to column-by-column principal component elimination and back substitution operations. Each column is eliminated independently. The elements of the column in each row below the row containing the principal component are reduced to zero through row transformation. Then, the back substitution is performed layer by layer from the last row upwards to solve for the reverse action coefficients corresponding to each column. The reverse action coefficients of each column are combined in turn to output the reverse action coefficients of the water supply valve opening change on each response channel. The principal component selection rule in the principal component elimination process is as follows: the element with the largest absolute value in the current column is selected as the principal component. If the absolute value of the principal component is less than the preset threshold of 0.001, the column is marked as a small principal component column and skipped. Independent elimination is performed on the column after the other columns have been processed.

[0032] The relative gain matrix is ​​solved by comparing the reverse action coefficient with the open-loop gain sequence. Specifically, the open-loop gain value of each channel is multiplied by its corresponding reverse action coefficient to obtain the element values ​​of the relative gain matrix for each operating condition. The elements of the relative gain matrix are calculated according to the following formula: ; in, Represents the th in the relative gain matrix Line 1 The element values ​​of the column, This indicates the output channel number, with a value of 1 or 2. This indicates the input channel number, with a value of 1. This indicates the total number of input channels, with a value of 2. Indicates the first The input channel for the first... Open-loop gain of each output channel; Represents the first in the reverse action coefficient matrix Line 1 The element values ​​of the column; Indicates the first The input channel for the first... Open-loop gain of each output channel, For summation index variables; Represents the first in the reverse action coefficient matrix Line 1 The element values ​​of the column; the calculated values ​​of each Arranged by row and column position, output the relative gain matrix for each operating condition.

[0033] The multiple relative gain matrices obtained under different steady-state conditions are piecewise linearized. The specific process of piecewise linearization is as follows: with the load as the horizontal axis, the range of 70% to 110% of the design load is divided into four equally spaced load intervals, with the width of each interval being 10% of the design load. Within each load interval, the arithmetic mean of the element values ​​at the same position in the relative gain matrix corresponding to all test load points within the interval is taken as the average interaction strength of the channel pair within the interval. The average interaction strength of each channel pair is arranged according to the row and column positions of the original matrix to form the coupling coefficient matrix corresponding to the load interval. The coupling coefficient matrix under each load interval is then output.

[0034] In the feedforward compensation generation module, the water supply valve opening signal is used as a measurable disturbance. Based on the coupling coefficient matrix, a feedforward compensation amount that leads the quench tower temperature lag response is generated through an asymmetric phase compensation network. The feedforward compensation amount is superimposed on the reuse pump frequency setpoint, and a feedforward adjustment command is output, specifically including: The element value at the diagonal position in the coupling coefficient matrix is ​​used as the basic allocation weight of the two channels. The diagonal element value represents the self-effect intensity of the water replenishment disturbance on the reuse pump frequency channel and the salt inhibitor injection channel in the current operating range. The reciprocal of the self-effect intensity is taken as the basic allocation weight of the channel. The larger the reciprocal value, the higher the response requirement of the channel to the water replenishment disturbance. The basic allocation weights of the two channels are added to obtain the total weight. Then, the basic allocation weight of each channel is divided by the total weight to make the sum of the allocation coefficients of the two channels equal to 1. The normalized result is used as the initial allocation coefficient of each channel and output.

[0035] The difference between the current value of the reuse pump frequency and the upper limit of the frequency is used as the adjustable margin of the reuse pump channel. The larger the difference, the more upward adjustment space the reuse pump frequency channel currently has. The difference between the current value of the salt inhibitor filling rate and the upper limit of the filling rate is used as the adjustable margin of the salt inhibitor channel. The larger the difference, the more upward adjustment space the salt inhibitor filling channel currently has. The adjustable margins of the two channels are divided by the sum of the adjustable margins of the two channels to obtain the margin ratio of each channel. When the initial allocation coefficient of a channel is greater than the margin ratio of that channel, it indicates that the allocation demand of that channel exceeds its physical execution capacity. At this time, the initial allocation coefficient of that channel is adjusted to be equal to the margin ratio, and the allocation coefficient of the other channel is adjusted to 1 minus the adjusted allocation coefficient of that channel. When the initial allocation coefficients of both channels do not exceed their respective margin ratios, the initial allocation coefficients remain unchanged. The allocation coefficients of the two channels after the above adjustment margin comparison and adjustment are used as the corrected allocation coefficients output.

[0036] The corrected allocation coefficient is compared with the preset maximum single-channel capacity ratio for each channel. The preset maximum single-channel capacity ratio is 0.7, which means that the feedforward compensation ratio undertaken by a single channel under any operating condition shall not exceed 70% of the total compensation. When the corrected allocation coefficient exceeds 0.7, the allocation coefficient of that channel is reduced to 0.7, and the allocation coefficient of the other channel is supplemented to 0.3. When the corrected allocation coefficients of both channels do not exceed 0.7, the corrected allocation coefficients remain unchanged. The allocation coefficients of the two channels finally determined after the above reduction and supplementation are output as the final feedforward allocation coefficients.

[0037] Using the water supply valve opening signal as the input excitation, different lead time constants and lag time constants are assigned to the two channels according to their phase compensation directions. The phase compensation direction of the reuse pump frequency channel is set to lead compensation, and its lead time constant is 0.3 times the lag time of the quench tower temperature response. The lag time of the quench tower temperature response is obtained through on-site step response testing and is set to 15 seconds. Therefore, the lead time constant of the reuse pump frequency channel is 4.5 seconds, and the lag time constant is 0 seconds. The phase compensation direction of the salt inhibitor injection channel is set to lag compensation, and its lead time constant is 0 seconds. The lag time constant is 1.2 times the duration of the fully open action of the water supply valve, which is set to 10 seconds. Therefore, the lag time constant of the salt inhibitor injection channel is 12 seconds. After the water supply valve opening signal is processed by the differential compensation and inertial delay of the corresponding time constants, the phase pre-adjustment of each of the two channels is obtained.

[0038] The phase pre-adjustment of the two channels is multiplied by the corresponding final feedforward allocation coefficient to obtain the feedforward compensation amount of each channel; the feedforward compensation amount is superimposed on the current set value of the reuse pump frequency to obtain the reuse pump frequency feedforward adjustment command and output to the reuse pump frequency inverter; the feedforward compensation amount is superimposed on the current set value of the salt inhibitor filling rate to obtain the salt inhibitor filling rate feedforward adjustment command and output to the salt inhibitor metering pump.

[0039] In the collaborative control output module, the disturbance type is determined based on the rate of change of the reconstructed salinity value per unit time. When the rate of change exceeds the limit, the feedback integral action is frozen and the proportional band is widened. The feedforward adjustment command dominates the water injection increment during the disturbance period. After the rate of change falls back, the integral action is restored in a ramp manner, and the final collaborative control command for the reuse pump and salt inhibitor injection rate is output, specifically including: The rate of change of reconstructed salinity value per unit time is used as input. The rate of change is calculated as follows: subtract the reconstructed salinity value obtained in the previous sampling period from the reconstructed salinity value obtained in the current sampling period, and then divide by the sampling period, which is 1 second, to obtain the rate of change at the current moment, in grams per liter per second. The absolute value of this rate of change is then compared with a preset rate of change limit boundary, which is 0.5 grams per liter per second. This value is based on the statistical upper limit of the rate of change of salt concentration under normal operating conditions. When the absolute value of the rate of change exceeds 0.5 grams per liter per second, it is identified as a water replenishment and dilution type interference, and the interference type flag is set to 1. Otherwise, the interference type flag is set to 0.

[0040] Using the interference type flag as the trigger condition, when the interference type flag is 1, the feedback integral action freeze operation is executed: the integral accumulation value in the feedback controller at the current moment is read, the integral accumulation value is stored in a temporary register, and the integral accumulation value is cleared to zero and the integral accumulation operation is stopped, so that the controller will no longer accumulate the deviation integrally in subsequent control cycles; while freezing the integral accumulation value, the proportional band in the feedback controller is widened from the current value to 1.5 times the current value. This proportional band widening operation is only executed once in a single interference event, and will not be repeated until the interference event ends. The feedback control state after freezing the integral and widening the proportional band is output.

[0041] The recovery condition is set at the absolute value of the reconstructed salinity value falling back to within a preset limit boundary of 0.5 g / L / s. When this condition is met, the gradual recovery of the integral action is initiated. The gradual recovery is executed step by step according to a preset order, with a preset order of 5 orders. Each order recovers one-fifth of the integral accumulation value stored in the temporary register. After each order is recovered, the absolute value of the reconstructed salinity value at the current moment is recalculated and compared with 0.5 g / L / s again. If the rate of change exceeds the limit boundary again, the recovery is immediately paused and the currently recovered integral accumulation value is frozen. After the rate of change falls back to within the limit boundary, the recovery of the subsequent orders continues until all 5 orders are completed and the integral accumulation value is completely restored to the original value stored in the temporary register. After the integral accumulation value is completely restored, the proportional band of the feedback controller is gradually pulled back to the original value according to the step size before widening. Finally, the reuse pump frequency setting value and the salt inhibitor dosing rate setting value are output as the final coordinated control command to their respective actuators.

[0042] The working principle of this invention is as follows: The salt concentration of the circulating liquid in the wet desulfurization tower, the opening signal of the water supply valve, and the flow rate and temperature data of the flue gas at the inlet of the quench tower are collected. Using the salt concentration measurement as the observation input, the salinity signal is reconstructed through an unscented Kalman filter to filter out false concentration spike noise introduced by water supply dilution, and the reconstructed salinity value is output. Using the water supply valve opening signal as the input excitation, and the reconstructed salinity value and the quench tower outlet temperature as the response output, cross-channel dynamic gain matrix analysis based on a relative gain array is performed to obtain the coupling coefficient matrix under each operating condition. Based on the coupling coefficient matrix, a feedforward compensation amount that leads the temperature lag response is generated through asymmetric phase compensation and superimposed on the reuse pump frequency setpoint, outputting a feedforward adjustment command. The type of disturbance is determined based on the rate of change of the reconstructed salinity value. When the rate of change exceeds the limit, the feedback integral action is frozen and the proportional band is widened. The feedforward adjustment command dominates the water spray increment during the disturbance period. After the rate of change falls back, the integral action is restored in a ramp manner, and the final coordinated control command is output. This invention achieves stable control of temperature, liquid level and salinity under water replenishment disturbance and zero wastewater discharge.

[0043] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A feedforward-feedback coordinated control system for flue gas desulfurization emissions based on salt concentration feedback, characterized in that, include: The data acquisition module collects the salt concentration measurement value of the circulating liquid in the wet deacidification tower, the opening signal of the water supply valve, and the flow rate and temperature data of the flue gas at the inlet of the quench tower. The salinity signal reconstruction module constructs a multi-dimensional state vector containing instantaneous salinity value, liquid level change rate and cumulative influence coefficient of water replenishment based on the collected salt concentration measurement value. It uses the water replenishment valve opening signal for auxiliary correction and dynamically reconstructs the salinity signal through unscented Kalman filtering to filter out false concentration spike noise introduced by water replenishment dilution and outputs the reconstructed salinity value. The coupling characteristic analysis module, based on the water supply valve opening signal and flue gas flow and temperature data, uses the reconstructed salinity value as feedback input, applies step excitation to each steady-state condition to obtain the open-loop gain sequence, performs determinant discrimination and reversible correction on the square matrix to obtain the reverse action coefficient, compares the reverse action coefficient with the open-loop gain to solve the relative gain matrix, and obtains the coupling coefficient matrix of each condition after piecewise linearization. The feedforward compensation generation module takes the water supply valve opening signal as a measurable disturbance and allocates the feedforward allocation coefficients of the reuse pump frequency channel and the salt inhibitor injection channel according to the coupling coefficient matrix. The reuse pump frequency uses advanced phase compensation and the salt inhibitor injection channel uses phase compensation to form a dual-channel asymmetric phase compensation network. It generates a feedforward compensation amount that leads the lag response of the quench tower temperature. The feedforward compensation amount is superimposed on the reuse pump frequency set value and outputs a feedforward adjustment command. The collaborative control output module determines the type of disturbance based on the rate of change of the reconstructed salinity value per unit time. When the rate of change exceeds the limit, the feedback integral action is frozen and the proportional band is widened. The feedforward adjustment command dominates the water injection increment during the disturbance period. After the rate of change falls back, the integral action is restored in a ramp manner, and the final collaborative control command for the reuse pump and salt inhibitor injection rate is output.

2. The flue gas desulfurization emission feedforward-feedback coordinated control system based on salt concentration feedback according to claim 1, characterized in that, The output reconstructed salinity value specifically includes: Using the measured salt concentration as the observation input, a multidimensional state vector containing instantaneous salinity value, liquid level change rate and cumulative influence coefficient of water replenishment is constructed, and the initial state estimate is output. A set of sampling points surrounding the initial state estimate is generated based on a deterministic sampling strategy. Each sampling point is passed through a nonlinear state transition equation to obtain the predicted state distribution and output the predicted state and its covariance. Using the water supply valve opening signal as an auxiliary correction input, the salinity component in the predicted state is updated and the covariance is corrected based on the deviation between the predicted state and the measured salt concentration. The false concentration spike noise introduced by the water supply dilution is filtered out, and the reconstructed salinity value is output.

3. The flue gas desulfurization emission feedforward-feedback coordinated control system based on salt concentration feedback according to claim 2, characterized in that, The output predicted state and its covariance specifically include: Using each sampling point in the sampling point set as input, the system maps each point through a nonlinear state transition relationship to obtain the predicted salinity component corresponding to each sampling point, and outputs the set of predicted salinity components for each sampling point. The predicted salinity component set of each sampling point is reorganized by mean value, and the estimated mean and variance of the salinity component in the predicted state are calculated. The preliminary predicted distribution and its covariance are output. Symmetric compression correction is applied to the variance of the salinity component estimation to eliminate the non-Gaussian skewness of the distribution introduced by nonlinear mapping during the transmission process, and the corrected prediction state and its covariance are output.

4. The feedforward-feedback coordinated control system for flue gas desulfurization emissions based on salt concentration feedback according to claim 1, characterized in that, The output process of the coupling coefficient matrix is ​​as follows: Using the water supply valve opening signal as the input excitation and the reconstructed salinity value and quench tower outlet temperature as the response output, the open-loop gain value of the water supply channel to the salinity channel and temperature channel under each steady-state condition is obtained one by one, and the open-loop gain sequence under each condition is output. Perform matrix inversion on the open-loop gain sequence under each operating condition to obtain the reverse action coefficient of the water supply valve opening change on each response channel. Then compare the reverse action coefficient with the open-loop gain sequence to obtain the relative gain value under each operating condition and output the relative gain matrix under each operating condition. The relative gain matrix under different steady-state conditions is piecewise linearized and rearranged. The interaction strength between the water replenishment loop and the salt balance loop in each operating condition interval is extracted, and the coupling coefficient matrix under each operating condition is output.

5. The flue gas desulfurization emission feedforward-feedback coordinated control system based on salt concentration feedback according to claim 4, characterized in that, The output process of the reverse action coefficient is as follows: Using the open-loop gain sequence under each operating condition as input, construct a square matrix with the gain transfer relationship between channels as elements; The determinant value of the square matrix is ​​judged. When the determinant value is less than the preset threshold, an equal amount of adjustment value is added to each element of the main diagonal and the determination is repeated until the determinant value falls into the invertible interval, and an invertible square matrix is ​​output. Perform column-by-column principal component elimination and back substitution operations on the reversible matrix to obtain the reverse action coefficients corresponding to each column. Combine the reverse action coefficients of each column in sequence to output the reverse action coefficients of the water supply valve opening change on each response channel.

6. The flue gas desulfurization emission feedforward-feedback coordinated control system based on salt concentration feedback according to claim 1, characterized in that, The output feedforward adjustment command specifically includes: Using the coupling coefficient matrix as input, the feedforward allocation coefficients for the frequency channel of the reuse pump and the salt inhibitor injection channel are determined based on the change amplitude and change rate of the current water supply valve opening signal, and the feedforward allocation coefficients for the two channels are output. Using the water supply valve opening signal as the input excitation, different lead time constants and lag time constants are assigned according to the phase compensation direction of the two channels. After the excitation signal is processed by differential compensation and inertial delay of the corresponding time constants, the phase pre-adjustment of the two channels is output. The phase pre-adjustment of the two channels is multiplied by the corresponding feedforward allocation coefficient, and the product is then added to the set value of each channel to output the feedforward adjustment command for the reuse pump frequency and the feedforward adjustment command for the salt inhibitor dosing rate.

7. The flue gas desulfurization emission feedforward-feedback coordinated control system based on salt concentration feedback according to claim 6, characterized in that, The feedforward allocation coefficients for the two output channels specifically include: The reciprocal of the diagonal element values ​​in the coupling coefficient matrix is ​​used as the basic allocation weight of the two channels. The two basic allocation weights are normalized and used as the initial allocation coefficients. The initial allocation coefficients are then output. The difference between the current value of the reuse pump frequency and the upper limit of the frequency is used as the adjustable margin of the reuse pump channel, and the difference between the current value of the salt inhibitor injection rate and the upper limit of the injection rate is used as the adjustable margin of the salt inhibitor channel. The initial allocation coefficient is corrected by the ratio of the adjustable margins of the two channels, and the corrected allocation coefficient is output. The corrected allocation coefficient is compared with the preset maximum capacity ratio of a single channel. When the corrected allocation coefficient exceeds the maximum capacity ratio of the corresponding channel, the allocation coefficient of the channel is reduced to the maximum capacity ratio, and the allocation coefficient of the other channel is supplemented accordingly. The final feedforward allocation coefficient of the two channels is then output.

8. The flue gas desulfurization emission feedforward-feedback coordinated control system based on salt concentration feedback according to claim 1, characterized in that, The output process of the final coordinated control command is as follows: The rate of change of reconstructed salinity value per unit time is used as input. The rate of change is compared with the preset rate of change limit boundary. When the rate of change exceeds the limit boundary, it is identified as water replenishment and dilution type interference, and the interference type flag is output. Using the interference type flag as the trigger condition, the integral accumulation value in the feedback control is frozen, and the proportional band in the feedback control is widened by a preset step size. The feedback control state after freezing the integral and widening the proportional band is output. The condition for recovery is that the rate of change falls back to within the limit boundary. The integral accumulation value is gradually thawed and restored according to the preset order. After each order is restored, it is re-evaluated whether the rate of change exceeds the limit again, until the integral accumulation value is fully restored. Finally, the final coordinated control command of the reuse pump frequency setting value and the salt inhibitor injection rate setting value is output.

Images