TWI824333B - Method of control system controlling nitrogen oxides and ammonia water in selective non-catalytic reduction system - Google Patents

Abstract

A method of a control system controlling emission of nitrogen oxides by computing a proper rate of spaying ammonia water includes a controller of the control system processing a plurality of datasets to generate a plurality of updated datasets, each updated dataset including a rate of spaying ammonia water and a plurality of updated variable values, generating a set of regression coefficients according to a plurality of average nitrogen oxide concentrations and the plurality of updated datasets, and the processor generating a target rate of spaying ammonia water according to a target average nitrogen oxide concentration, the set of regression coefficients and an updated dataset. The target rate of spaying ammonia water is the proper rate of spaying ammonia water.

Description

Optimal ammonia and nitrogen oxide control method based on selective non-catalytic denitrification reduction system

The invention relates to cement production, particularly an optimal ammonia water and nitrogen oxide control method based on a selective non-catalytic denitrification reduction system.

The high-temperature environment of cement production will produce nitrogen oxide emissions. In order to avoid the impact of nitrogen oxides on the environment and comply with government regulations, a selective non-catalytic reduction (SNCR) system is currently used, which sprays ammonia water. Suppresses nitrogen oxides. However, too much ammonia will cause the unreacted ammonia to be discharged and become ammonia escape, causing secondary pollution and accelerating the corrosion of machinery and equipment.

Embodiments of the present invention provide a control system that controls nitrogen oxide emissions by calculating the optimal ammonia spray volume. The control system includes a processor. The processor processes a plurality of sets of data sets to generate a plurality of sets of update data sets. Each set of update data The set contains an ammonia spray volume and a plurality of update variable values, and then a plurality of average nitrogen oxide concentrations and a plurality of data sets are used to generate a set of regression coefficients, and the processor generates a set of regression coefficients based on the target average nitrogen oxide concentration, the set of regression coefficients and the complex number Each updated variable value generates the target ammonia spray volume, which is the current optimal ammonia spray volume.

Figure 1 is a block diagram of a control system 1 in an embodiment of the present invention. Control system 1 can be used in cement plants to accurately produce the minimum ammonia spray volume under given nitrogen oxide emission standards, thereby controlling nitrogen oxide emissions while reducing ammonia escape in cement plants and reducing corrosion of cement production line equipment.

The control system 1 includes sensors 101 to 10N and a processor 12, where N is a positive integer greater than 1. The sensors 101 to 10N can be disposed on a plurality of production line equipment in the cement plant, and are coupled to the processor 12 through wired or wireless connections. At least one sensor can be installed on each production line equipment. The processor 12 can be installed in a central control system, and the central control system can be installed in a cement plant or a remote computer room. Multiple production line equipment may include raw material scales, preheating towers, calcining furnaces, smoke chambers, chimneys, decomposition furnaces, rotary kilns, grate coolers, high temperature fans, cooling fans, kiln head hoods, kiln head exhaust fans and other products. line equipment. The sensors 101 to 10N may be temperature sensors, pressure sensors, concentration sensors, flow meters or other types of sensors. In some embodiments, the placement locations and types of sensors 101 to 10N can be obtained from expert opinions. The sensors 101 to 10N can obtain the ammonia water spray volume and the variable values of the plurality of variables in a timed or irregular manner, and transmit the ammonia water spray volume and the variable values of the plurality of variables to the processor 12 for storage in the database in chronological order. It is a plurality of data sets. Each data set corresponds to a time point and contains ammonia spray volume and a plurality of variable values. Table 1 shows the variables that can be obtained by sensors 101 to 10N, as well as the unit, maximum value, and minimum value of each variable.

Table 1 variables unit maximum value minimum value Calcining furnace temperature ℃ 1538 488 smoke chamber oxygen concentration % 25 0 Chimney oxygen concentration % 14 0 Head coal flow ton/h 18 0 kiln head pressure pa 0 -200 Kiln head temperature ℃ 1204 0 Raw feed amount ton/h 510 0 tailing coal flow ton/h twenty three 0 Nitrogen oxide concentration at kiln tail mg/m ³ 962 0 clinker temperature ℃ 1132 0 Smoke chamber temperature ℃ 1347 0 Three air temperatures ℃ 1037 0 Ammonia spray volume ton/h 1999 117 Chimney NOx concentration mg/m ³ 397 0 Carbon monoxide concentration at the outlet of the preheating tower ppm 20000 0

The ammonia water injection amount can be the ammonia water injection amount in Table 1, and the plural variables can include the calcining furnace temperature, smoke chamber oxygen concentration, chimney oxygen concentration, coal flow rate, kiln head pressure, kiln head temperature, and raw meal in Table 1 Feeding volume, tail coal flow, kiln tail nitrogen oxide concentration, clinker temperature, smoke chamber temperature, tertiary air temperature, chimney nitrogen oxide concentration and preheating tower outlet carbon monoxide concentration, but are not limited to these.

Since the concentration of nitrogen oxides can be affected by the working conditions of the cement kiln and the amount of ammonia sprayed, the processor 12 can use the average nitrogen oxide concentration as a dependent variable, and use the ammonia sprayed amount and the set of corresponding variables as corresponding variables to perform regression analysis and establish regression model, and then calculate the recommended ammonia spray volume based on the regression model to achieve the target average nitrogen oxide concentration. For example, the target average nitrogen oxide concentration in the next 5 minutes can be 280 mg/m ³ .

Figure 2 is a flow chart of a method 200 for controlling nitrogen oxide emissions by the control system 1 . The method 200 includes steps S202 to S206 for performing regression analysis to predict the average nitrogen oxide concentration. Any reasonable technical changes or step adjustments fall within the scope disclosed by the present invention. The details of steps S202 to S206 are as follows:

Step S202: Sensors 101 to 10N obtain the ammonia spray volume and multiple variable values;

Step S203: The processor 12 processes a plurality of sets of data sets to generate a plurality of sets of update data sets, each set of update data sets includes ammonia spray volume and a plurality of update variable values;

Step S204: The processor 12 generates a set of regression coefficients based on the plurality of average nitrogen oxide concentrations and the sets of updated data sets;

Step S206: The processor 12 generates the target ammonia spray volume based on the target average nitrogen oxide concentration, the set of regression coefficients and a set of updated data sets.

In step S202, the sensors 101 to 10N of each cement plant regularly acquire the ammonia spray volume and a plurality of variable values of a plurality of variables. The processor 12 converts the ammonia spray volume and the variable values into the same time unit (for example, the time unit is per minute), and/or performs data processing (for example, calculating the average nitrogen oxide concentration of the chimney nitrogen oxide concentration), and then The converted ammonia spray volume, a plurality of converted variable values and a plurality of data processed variable values are sent to the database to store a plurality of average nitrogen oxide concentrations and a plurality of data sets. The average nitrogen oxide concentration may be the average nitrogen oxide concentration within a predetermined time, such as the average nitrogen oxide concentration for 5 minutes. Each data set contains the ammonia spray volume at a specific time and the values of these variables.

In step S203, the processor 12 obtains the sets of data sets within the predetermined period, performs extreme value pre-processing and/or data smoothing pre-processing on the plurality of corresponding strain values of the sets of data sets, and based on the processed These set data sets filter variables to produce complex set update data sets. The predetermined time period can be set according to requirements. For example, the predetermined time period can be 10 minutes, and each data set can contain 10 variable values. The extreme value pre-processing can detect outliers corresponding to the variable values in the data sets, and update the outliers to values within a plurality of preset ranges of the corresponding variables, such as within the preset ranges of the corresponding variables. Extreme values. The data smoothing pre-processing can replace the corresponding strain values that suddenly increase or decrease with the average value of the corresponding strain values. In some embodiments, the processor 12 may first perform extreme value pre-processing, and then perform data smoothing pre-processing to generate updated variable values. Each update data set contains ammonia spray volume and multiple update variable values.

In some embodiments, when performing extreme value pre-processing, for each variable in the data sets, the processor 12 will determine the corresponding variable values in the data sets that exceed the preset range. Replaced by the extreme value of the default range. The processor 12 generates updated variable values based on the latest values of the corresponding variable values. For example, the preset range of the chimney nitrogen oxide concentration is 0 mg/m ³ to 397 mg/m ³ . If the variable value of the chimney nitrogen oxide concentration is less than 0, the processor 12 will update the variable value of the chimney nitrogen oxide concentration to 0; if the variable value of the chimney nitrogen oxide concentration is greater than 397, the processor 12 will update the chimney nitrogen oxide concentration. The variable value for substance concentration is updated to 397. Table 2 shows a plurality of corresponding strain values for the nitrogen oxide concentration in the chimney in the embodiment, and each corresponding strain value is separated by 1 second.

Table 2 time NOx Updated NOx 2020-4-14 00:50 -20 0 2020-4-14 00:51 -10 0 2020-4-14 00:52 0 0 2020-4-14 00:53 5 5 2020-4-14 00:54 10 10 2020-4-14 00:55 18 18 2020-4-14 00:56 58 58 2020-4-14 00:57 143 143 2020-4-14 00:58 181 181 2020-4-14 00:59 82 82

As shown in Table 2, at time 2020-04-14 00:50 and time 2020-04-14 00:51, since the corresponding strain values of the nitrogen oxide concentration in the chimney are -20 and -10 respectively, the processor 12 will The corresponding strain value of the chimney nitrogen oxide concentration is updated to 0. Between time 2020-04-14 00:51 and time 2020-04-14 00:59, since the variable value of the chimney nitrogen oxide concentration is within the preset range, the processor 12 does not need to update the chimney nitrogen oxide Concentration variable value.

In other embodiments, when performing extreme value preprocessing, for each variable in the data sets, if the difference between two consecutive corresponding strain values in the corresponding strain values in the data sets is greater than The processor 12 will replace the value after the two consecutive corresponding strain values with the value before the two consecutive corresponding strain values; if the difference between the two consecutive corresponding strain values in the data set is less than the predetermined value, the processor 12 will not Updated value. The processor 12 generates updated variable values based on the latest values of the corresponding variable values. If the difference between two consecutive corresponding strain values is too large, the corresponding strain value measured by the sensor may be abnormal, so the newer later value is replaced by the older previous value to ensure the accuracy of the regression coefficient. Table 3 shows a plurality of corresponding variable values of the chimney oxygen concentration in the embodiment, and each variable value is separated by 1 second. The predetermined value can be 2%. When the difference between two consecutive corresponding strain values of the chimney oxygen concentration is greater than or equal to 2%, the chimney oxygen concentration is greater than 17%, or the chimney oxygen concentration is less than 4%, the chimney oxygen concentration may be regarded as abnormal, and the processor 12 will The value after the two consecutive corresponding values is replaced by the value before the two consecutive corresponding values. Otherwise, the chimney oxygen concentration can be regarded as normal, and the processor 12 will not update the subsequent value.

Form 3 time O2 The updated O2 2020-05-20 23:40 5.0 5.0 2020-05-20 23:41 5.0 5.0 2020-05-20 23:42 5.0 5.0 2020-05-20 23:43 5.0 5.0 2020-05-20 23:44 5.0 5.0 2020-05-20 23:45 5.0 5.0 2020-05-20 23:46 5.0 5.0 2020-05-20 23:47 5.0 5.0 2020-05-20 23:48 13.0 5.0 2020-05-20 23:49 19.0 5.0 2020-05-20 23:50 17.0 5.0 2020-05-20 23:51 15.0 5.0 2020-05-20 23:52 13.0 5.0 2020-05-20 23:53 11.0 5.0 2020-05-20 23:54 10.0 5.0 2020-05-20 23:55 9.0 5.0 2020-05-20 23:56 8.0 5.0 2020-05-20 23:57 8.0 5.0 2020-05-20 23:58 7.0 5.0 2020-05-20 23:59 7.0 5.0

As shown in Table 3, between the time 2020-05-20 23:40 and the time 2020-05-20 23:47, due to the difference between the two continuous variable values (5%, 5%) of the chimney oxygen concentration (0%) is less than 2% and the variable value (5%) of the stack oxygen concentration is between 2% and 17%, so the processor 12 will not update the value; at time 2020-05-20 23:48, due to The difference (8%) between the two continuous variable values of the chimney oxygen concentration (5%, 13%) is greater than 2%, so the processor 12 updates the later value of the two continuous variable values of the chimney oxygen concentration to the previous value (5% ); At time 2020-05-20 23:49, since the variable value of the chimney oxygen concentration (19%) is greater than 17%, the processor 12 updates the subsequent value of the two consecutive variable values of the chimney oxygen concentration to the previous value (5% ); At time 2020-05-20 23:50, since the variable value (17%) of the chimney oxygen concentration is equal to 17%, the processor 12 updates the subsequent value of the two consecutive variable values of the chimney oxygen concentration to the previous value (5% ); between time 2020-05-20 23:51 and time 2020-05-20 23:59, since the difference between the two continuous variable values of the chimney oxygen concentration is greater than or equal to 2%, the processor 12 will The subsequent value of the two continuous variables of the chimney oxygen concentration is updated to the previous value (5%).

In some embodiments, when performing data smoothing pre-processing, for each variable in the data sets, if the latest value among the corresponding variable values in the data sets is substantially equal to the predetermined value, then processing The processor 12 replaces the latest value with the average of the remaining values of the corresponding variable values; if the latest value in the set of data sets is greater than the predetermined value, the processor 12 does not change the latest value. The processor 12 generates updated variable values based on the latest values of the data sets. For example, the corresponding variable values of the data sets may include 6 variable values (V(t-5), V(t-4), V(t-3), V(t -2), V(t-1), V(t),), each variable value is separated by 1 minute, and the predetermined value can be 0. At time t, if the latest value is 0, the processor 12 replaces the latest value V(t) with the average of the remaining values (V(t-1) to V(t-5)) of the corresponding variable values, As shown in formula (2):

V(t)=(V(t-1)+ V(t-2)+…+V(t-5))/5 Formula (2)

In other embodiments, when performing data smoothing pre-processing, for each variable in the sets of data sets, the processor 12 converts the latest values of the corresponding variable values in the sets of data sets to the corresponding values. The average value of the variable value is replaced, and the average value can be a 5-minute moving average, as shown in formula (3):

V(t)=(V(t)+ V(t-2)+…+V(t-4))/5 Formula (3)

Among them, V(t) is the latest value, and V(t-4)) is the variable value 4 minutes ago. The processor 12 generates updated variable values based on the latest values of the corresponding variable values. Table 4 shows a plurality of corresponding strain values for the ammonia water injection amount in the embodiment, and each corresponding strain value is separated by 1 minute.

Form 4 Time(t) NH3 NH3_M5 Updated NH3 2020-16-18 23:50 988.203 - 988.203 2020-16-18 23:51 994.297 - 994.297 2020-16-18 23:52 1021.953 - 1021.953 2020-16-18 23:53 1061.953 - 1061.953 2020-16-18 23:54 1100.234 1033.3280 1033.3280 2020-16-18 23:55 1137.422 1063.1718 1063.1718 2020-16-18 23:56 1178.047 1099.9218 1099.9218 2020-16-18 23:57 1215.938 1138.7188 1138.7188 2020-16-18 23:58 1253.281 1176.9844 1176.9844 2020-16-18 23:59 1293.672 1215.6720 1215.6720

As shown in Table 4, at time 2020-16-18 23:54, processor 12 calculates the moving average NH3_M5 from time 2020-16-18 23:50 to time 2020-16-18 23:54/5 to 1033.328ton /h(=(988.203+994.297+1021.953+1061.953+1100.234)/5), and from time 2020-16-18 23:55 to time 2020-16-18 23:59, processor 12 calculates its 5-point movement respectively Average NH3_M5. From time 2020-16-18 23:50 to time 2020-16-18 23:53, since there is no 5-minute moving average NH3_M5, the processor 12 uses the ammonia spray volume as the updated ammonia spray volume; at time 2020-16 From -18 23:54 to 2020-16-18 23:59, the processor 12 uses the moving average NH3_M5 as the updated ammonia injection volume.

The processor 12 may use forward selection, backward selection or stepwise selection to select a specific variable from a plurality of variables based on the processed data sets. A set of corresponding variables of the cement plant to generate these sets of updated data sets. For example, the processor 12 may use backward selection to eliminate statistically insignificant corresponding variables one by one from a plurality of variables until a set of statistically significant corresponding variables remains. Regression models built using this set of corresponding variables will improve accuracy. Compared with the regression model using all corresponding variables, the regression model using this set of corresponding variables can increase the R-square from 0.94 to 0.95 and reduce the mean absolute percentage error (MAPE) from 3.46%. to 3.22%, and the mean absolute error (MAE) decreased from 62 to 61.8. All three indicators show that the accuracy of the regression model has increased.

In step S204, the processor 12 obtains the average nitrogen oxide concentrations of a specific cement plant and the updated data sets from the database to perform regression analysis on the specific cement plant to generate a set of regression coefficients. This set of regression coefficients is used to establish a regression model of average nitrogen oxide concentration. For example, equation (1) shows a regression model for average nitrogen oxide concentration:

Average NOx=β0+β1*X1+β2*X2+β3*X3+β4*X4+β5*X5+β6*X6+β7*X7+β8*X8+β9*X9+β10*X10 Formula (1)

Among them: average NOx is the average nitrogen oxide concentration;

β0 is the offset value;

X1 series is the head coal flow rate;

β1 is the coefficient of head coal flow rate;

X2 is the raw material feeding amount;

β2 is the coefficient of raw material feeding amount;

X3 series is tail coal flow;

β3 is the coefficient of tailing coal flow;

X4 is the chimney oxygen concentration;

β4 is the coefficient of oxygen concentration in the chimney;

X5 is the carbon monoxide concentration at the outlet of the preheating tower;

β5 is the coefficient of carbon monoxide concentration at the outlet of the preheating tower;

X6 series is the kiln head temperature;

β6 is the coefficient of kiln head temperature;

The X7 series has three air temperatures;

β7 is the coefficient of the third wind temperature;

X8 is the amount of ammonia sprayed one minute ago;

β8 is the coefficient of ammonia spray volume one minute ago;

X9 is the concentration of nitrogen oxides in the chimney;

β9 is the coefficient of nitrogen oxide concentration in the chimney;

X10 is the amount of ammonia sprayed; and

β10 is the coefficient of ammonia spray volume.

After establishing the regression model, in step S206, the processor 12 brings the target average nitrogen oxide concentration and the updated variable values of a plurality of corresponding variables (X1 to X9) of a set of updated data sets of the regression model into formula (1), that is, It is possible to estimate the target ammonia water injection volume x10. In some embodiments, the set of update data may be the latest set of update data. In other embodiments, the set of updated data sets may be a set of updated data sets every fixed time, such as every minute. Table 5 shows the updated variable values of the corresponding variables (X1 to X9) in the embodiment.

Form 5 regression coefficient coefficient Update variable value Head coal flow (X1) -5.0427 11.862 Raw material feeding amount (X2) -0.2199 380.464 Tail coal flow (X3) -1.7488 19.714 Chimney oxygen concentration (X4) -3.0445 8.333 Carbon monoxide concentration at the outlet of the preheating tower (X5) -4884.2369 0.003 Kiln head temperature (X6) -0.1774 1177.778 Three air temperatures (X7) -0.0698 975.043 Amount of ammonia sprayed one minute ago (X8) -0.9939 964 Chimney nitrogen oxide concentration (X9) -0.8403 300.498

In one example, the target average nitrogen oxide concentration in the next 5 minutes is 280, the coefficient β10 of the target ammonia spray volume is 1, and the offset value β0 is 308.4710. The processor 12 will use the formula (1) and the values in Table 5 The predicted value of the calculated target ammonia spray volume is 980.9118 (280 = -308.4710+β1*X1+β2*X2+β3*X3+β4*X4+β5*X5+β6*X6+β7*X7+β8*X8+β9* X9+β10*980.9118). In step S206, the ammonia water spraying device sprays ammonia water according to the predicted value of the target ammonia water spray volume (980.9118 ton/h) to control the emission of nitrogen oxides. Each cement plant can monitor the concentration of nitrogen oxides, and the central control system can display the real-time nitrogen oxide concentration of each cement plant, and indicate government standards and internal control standards for management. Similarly, each cement plant can monitor the ammonia slip concentration, and the central control system can present the real-time ammonia slip concentration of each cement plant. At the same time, each cement plant can measure the thickness of the equipment at the kiln tail chimney, kiln head chimney, and air duct entrance and exit of the electric bag composite dust collection equipment to monitor equipment corrosion. If the corrosion degree exceeds the system set standard, the central control system will warning.

Figure 3 shows the effect of control system 1 in controlling nitrogen oxide emissions, in which the horizontal axis represents time and the vertical axis represents nitrogen oxide concentration and ammonia injection volume. The simulation curve 40 represents the predicted value of the target ammonia water injection amount, the simulation curve 42 represents the actual value of the ammonia water injection amount, and the simulation curve 44 represents the actual value of the nitrogen oxide concentration. Figure 3 shows that the actual value of the ammonia spray volume is essentially equal to the predicted value of the target ammonia spray volume, and the nitrogen oxide concentration is controlled within 280mg/ ^m3 .

The control system 1 and method 200 are suitable for cement plants. Under given nitrogen oxide emission standards, they can accurately produce the minimum ammonia spray volume, thereby controlling the emission of nitrogen oxides while reducing ammonia escape in the cement plant and reducing the cost of cement production line equipment. of corrosion. The above are only preferred embodiments of the present invention, and all equivalent changes and modifications made in accordance with the patentable scope of the present invention shall fall within the scope of the present invention.

1:Control system 101 to 10N: Sensor 12: Processor 200:Method S202 to S206: steps 40: Predicted value of target ammonia spray volume 42: Actual value of ammonia spray volume 44: Actual value of nitrogen oxide concentration

Figure 1 is a block diagram of a control system in an embodiment of the present invention. Figure 2 is a flow chart of a method for controlling nitrogen oxide emissions of one of the control systems in Figure 1. Figure 3 shows the effectiveness of the control system in Figure 1 for controlling nitrogen oxide emissions.

200:Method

S202 to S206: steps

Claims (10)

A method for controlling nitrogen oxide emissions by a control system, the control system including a processor, the method including: The processor processes a plurality of sets of data sets to generate a plurality of sets of update data sets, each set of update data sets includes an ammonia spray volume and a plurality of update variable values; The processor generates a set of regression coefficients based on the plurality of average nitrogen oxide concentrations and the sets of updated data sets; and The processor generates a target ammonia spray volume based on a target average nitrogen oxide concentration, the set of regression coefficients and a set of updated data sets. As claimed in claim 1, the method further includes: The sensors obtain the ammonia spray volume and the variable values. The method as described in claim 1 or 2, wherein the variable values include a calcining furnace temperature, a smoke chamber oxygen concentration, a chimney oxygen concentration, a coal flow rate, a kiln head pressure, a kiln head temperature, a first feed material Feeding amount, coal flow rate at the end of the kiln, nitrogen oxide concentration at the kiln end, clinker temperature, smoke chamber temperature, air temperature and tertiary air temperature, and carbon oxide concentration at the outlet of the preheating tower. The method as described in claim 1, wherein the processor processes the data sets to generate the updated data sets including: the processor obtains the sets of data; and The processor processes a plurality of corresponding strain values of the sets of data to generate a processed complex set of data. The method as described in claim 4, wherein the processor's processing of the corresponding strain values of the sets of data sets includes: Corresponding strain values that exceed a preset range among the corresponding strain values are replaced with extreme values of the preset range. The method as described in claim 4, wherein the processor's processing of the corresponding strain values of the sets of data sets includes: When the difference between two consecutive corresponding strain values among the corresponding strain values is greater than a predetermined value, the processor replaces a later value of the two consecutive corresponding strain values with a previous value of the two consecutive corresponding strain values. The method as described in claim 4, wherein the processor's processing of the corresponding strain values of the sets of data sets includes: If a latest value among the corresponding strain values is substantially equal to a predetermined value, the processor replaces the latest value with an average of the remaining values of the corresponding strain values. The method as described in claim 4, wherein the processor's processing of the corresponding strain values of the sets of data sets includes: The processor replaces a latest value among the corresponding strain values with an average value of the corresponding strain values. The method described in claim 4 further includes: selecting a set of corresponding variables based on the processed sets of data to generate the sets of updated data. The method of claim 1 further includes the processor generating an average ammonia spray amount based on a plurality of ammonia spray amounts of the data sets.

Images