A dry quenching and discharging control method based on model predictive control

Abstract

The present application belongs to the technical field of dry quenching coke discharging, and particularly relates to a dry quenching coke discharging control method based on model predictive control. First, time is segmented according to the coke charging plan. After segmentation, the total coke charging amount is calculated according to the number of furnaces in each time interval and the coke charging amount of each furnace. The reference amplitude of each time interval is calculated according to the average coke discharging amount corresponding to the amplitude. Then, a mathematical model of the coke discharging amount and the material level is established, a corresponding model of the amplitude and the coke discharging amount is established, and the material level and the amplitude at each time point in the entire planning period are calculated according to the coke pushing plan to obtain the expected value. Finally, the actual amplitude used is calculated through the MPC algorithm. The present application aims to realize stable control of the material level by accurately controlling the coke discharging amplitude.

Description

Technical Field

[0001] This invention belongs to the field of dry quenching and coke removal technology, specifically relating to a dry quenching and coke removal control method based on model predictive control. Background Technology

[0002] Dry quenching (CDQ) technology is a widely used waste heat recovery method in the steel industry. It recovers the sensible heat of hot coke by using inert gas circulation. The steam produced without energy expenditure can be used for power generation or as a steam heat source for steel plants, making it a highly economical and energy-saving system. With the increasing severity of global energy shortages and the further improvement of environmental protection laws and regulations, dry quenching technology has become an important part of the development of steel enterprises due to its outstanding energy-saving and environmental protection effects.

[0003] Currently, most coking plants rely on manual control for dry quenching, requiring 24-hour monitoring. This high labor cost, coupled with the inherent tendency for workers to become complacent and prone to errors during prolonged monitoring, necessitates the introduction of automated control. However, typical automated controls generate instructions based on real-time conditions. Such control is not only inefficient but also lacks robustness, functioning only effectively during stable operation. In the event of unforeseen circumstances, it becomes difficult to handle.

[0004] During the operation of a dry quenching furnace, controlling the coke discharge amplitude to adjust the coke discharge speed, and thus the rate at which the material level descends within the furnace, is crucial. Traditional control methods often fail to effectively balance the dynamic equilibrium between coke charging and discharging, resulting in large material level fluctuations and low control precision. Summary of the Invention

[0005] In response to the shortcomings of existing technologies, the inventors have developed a model predictive control-based dry quenching coke discharge control method through long-term practice. The aim is to achieve stable control of the material level by precisely controlling the discharge amplitude.

[0006] The dry quenching and coke removal control method based on model predictive control of the present invention includes the following steps:

[0007] S1: Divide the time into segments according to the focus loading plan. Consider all continuous focus loading with a focus loading interval of m to n minutes as the first segment, all continuous focus loading with a focus loading interval of less than m minutes as the second segment, and all continuous focus loading with a focus loading interval of more than n minutes as the third segment. If the number of focus loadings in a certain time interval is less than i, merge it with the longer segment of the other two segments.

[0008] S2: After the segmentation is completed, the total coke loading is calculated based on the number of furnaces in each time interval and the amount of coke loaded in each furnace.

[0009] S3: Set the initial material level and desired material level, and the desired change in material level height. = Expected material level - Initial material level.

[0010] S4: Establish a mathematical model for coke discharge rate and material level.

[0011] S5: Establish a corresponding model for amplitude and coke discharge.

[0012] S6: Based on the coke pushing plan, calculate the expected values ​​of amplitude and expected material level height at each moment within the entire planning cycle.

[0013] S6: Calculate the actual amplitude and expected material level using the MPC algorithm model.

[0014] Furthermore, in step S1, m, n, and i are all natural numbers greater than zero, and m < n and i < 10.

[0015] Furthermore, in step S2, the total amount of coke loaded is expressed by the following formula:

[0016] Total coke volume = .

[0017] Furthermore, in step S4, the coke discharge rate-material level change model is defined as follows:

[0018] ;

[0019] Where pjl is the amount of coke discharged, and t is the duration. It is the density of coke, S g It refers to the bottom area of ​​the dry quenching furnace.

[0020] Furthermore, in step S5, the relationship between amplitude and coke discharge is expressed as follows:

[0021] ;

[0022] Where a and b are parameters, pjl is the amount of coke discharged, and X is the amplitude.

[0023] Furthermore, in step S6, the expected values ​​of the amplitude and the expected material level height at each moment within the entire planning cycle are calculated, using the following method:

[0024] The average amount of coke removed during duration t is calculated using the following formula:

[0025] Average coke discharge rate = ΔH × S g ×ρ / t;

[0026] The corresponding amplitude can be obtained from the average coke discharge rate. After a duration of t1 under this amplitude, the expected material level height is calculated using the following formula:

[0027] Expected material level height = (ΔH × t / t0) × t1;

[0028] Where t0 equals 1, and the unit is hours.

[0029] Furthermore, in step S7, the specific calculation method of the MPC algorithm model is as follows:

[0030] ;

[0031] ;

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] Among them, H( ) is the controlled variable, which in the scenario is the material level at the next moment;

[0037] It is the planned material level height for the next moment;

[0038] X is the amplitude;

[0039] This indicates the change in amplitude at the next moment;

[0040] This indicates the predicted change in material level at the next moment.

[0041] M represents the control step size, and P represents the prediction step size;

[0042] Q, R, and S are all weight matrices.

[0043] The beneficial effects of this invention are:

[0044] The control method of this invention can control the material level in the dry quenching furnace more stably and with stronger robustness. In case of unexpected situations, the possibility of the material level exceeding the upper and lower limits is smaller, and there is a larger reaction space. At the same time, it can predict the coke discharge of dry quenching, provide early warning, and allow for timely manual intervention. Attached Figure Description

[0045] Figure 1 This refers to the prediction of material level and amplitude without the involvement of the MPC algorithm of this invention.

[0046] Figure 2 The predicted material level and amplitude are obtained after the MPC algorithm of this invention is applied. Detailed Implementation

[0047] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. These embodiments are described in sufficient detail to enable those skilled in the art to understand and practice the invention. Logical, implementation, and other changes may be made to the embodiments without departing from the spirit and scope of the invention. Therefore, the following detailed description should not be construed as limiting, and the scope of the invention is defined solely by the claims.

[0048] To address the low efficiency of current dry quenching coke discharge control methods, this invention proposes a model predictive control-based coke discharge control method. During dry quenching furnace operation, the coke discharge speed can be adjusted by controlling the discharge amplitude, thereby changing the rate at which the material level descends within the furnace. Furthermore, considering that coke charging causes the material level to rise, adjusting the discharge amplitude allows the material level to fluctuate within a certain range, minimizing the number of adjustments required. The specific steps include the following:

[0049] S1: Divide the time into segments according to the coking plan. All consecutive coking pushes with a time interval between m and n minutes are considered the first segment. All consecutive coking pushes with a time interval less than m minutes are considered the second segment. All consecutive coking pushes with a time interval greater than n minutes are considered the third segment. If the number of coking pushes within a certain time interval is less than i, it is merged with the longer of the other two segments. In this embodiment, m, n, and i are all natural numbers greater than zero, and m < n and i < 10.

[0050] S2: After segmentation, calculate the total coke loading based on the number of furnaces in each time interval and the coke loading amount of each furnace, i.e., total coke loading amount = .

[0051] S3: Set the initial material level and desired material level, and the desired change in material level height. = Expected material level - Initial material level.

[0052] Traditional methods calculate the baseline amplitude at each time interval based on the average coke discharge rate corresponding to the amplitude, in order to achieve a theoretical balance between coke loading and discharging. However, in actual production, the situation is not always ideal; for example, there may be maintenance or periods when coke loading is stopped. Actual coke loading, discharging, and material level changes are also not fixed. Therefore, the coke discharge rate for the same amplitude is constantly changing.

[0053] Based on the above, the present invention sets an initial material level and an expected material level before and after each time interval, that is, it is expected that the material level will change from the initial material level to the expected material level after the end of this period.

[0054] S4: Establish a mathematical model for coke discharge rate and material level. Although the material level surface inside the furnace is not a plane, but rather composed of multiple cones, the actual detected material level should be a certain height from the apex to the base of the cone. However, in actual production, the material level is allowed to vary within a range to encompass this measurement error. Therefore, the coke discharge rate-material level variation model is defined as follows:

[0055] ;

[0056] Where pjl is the amount of coke discharged, and t is the duration. It is the density of coke, S g It refers to the bottom area of ​​the dry quenching furnace.

[0057] S5: Establish a correspondence model between amplitude and coke discharge rate. Theoretically, this is a time-series characteristic, approximately linear. That is, while the coke discharge rate is affected by amplitude, it is also related to factors such as the properties of the coke and the production environment, which change over time. In actual production, there is a relatively large tolerance for changes in material level; for example, in a certain coking plant, the material level can be between 23m and 33m. Therefore, the time-series characteristics can be ignored, and linear regression can be performed using data from the most recent period to obtain the relationship between amplitude and coke discharge rate, expressed as:

[0058] ;

[0059] Where a and b are parameters, pjl is the amount of coke discharged, and X is the amplitude.

[0060] S6: Based on the coke pushing plan, calculate the expected values ​​of amplitude and expected material level height at each moment within the entire planning cycle. The specific method is as follows:

[0061] The average amount of coke removed during duration t is calculated using the following formula:

[0062] Average coke discharge rate = ΔH × S g ×ρ / t;

[0063] The corresponding amplitude can be obtained from the average coke discharge rate. After a duration of t1 under this amplitude, the expected material level height is calculated using the following formula:

[0064] Expected material level height = (ΔH × t / t0) × t1;

[0065] Where t0 equals 1, and the unit is hours.

[0066] S7: Calculate the actual amplitude and expected material level using the MPC algorithm model. Since various errors exist during actual operation, the expected material level will not always be achieved using the corresponding amplitude. Therefore, we further refine the MPC algorithm to calculate the actual amplitude used.

[0067] In traditional MPC model predictive control, the loss function is:

[0068] ;

[0069] In dry quenching scenarios, the frequency and amplitude of the amplitude adjustment need to be as small as possible, while the material level needs to be as stable as possible. Therefore, the squared increment of the controlled variable is added to the loss function to achieve the objective. In summary, the scenario is set as a quadratic programming problem, and the actual amplitude and expected material level are calculated as follows.

[0070] ;

[0071] ;

[0072] ;

[0073] ;

[0074] ;

[0075] ;

[0076] In the above formula, H( ) is the controlled variable, which in the scenario is the material level at the next moment; The vector X(t) represents the planned material level height at the next moment, and is the control variable, which is the amplitude in the scenario. This indicates the change in amplitude at the next moment. This represents the predicted change in material level at the next moment; M represents the control step size, and P represents the prediction step size; Q, R, and S are all weight matrices. The problem can be solved using solvers such as HPIPM and OSQP.

[0077] See Figure 1 and Figure 2 In the graph, the yellow line represents the change in material level, and the blue line represents the change in amplitude. Figure 1 This is the prediction result without the involvement of the MPC algorithm of this invention. Figure 2 This is the prediction result obtained from the MPC algorithm of this invention when MPC is involved. Figure 1 and Figure 2 As can be seen, after the MPC algorithm of this invention is involved, the prediction of material level is more stable, and the frequency of amplitude change of the vibrating feeder is significantly reduced.

[0078] In actual production, appropriate prediction and control step lengths can be set according to the coking plan to achieve optimal control of the current material level and early warning of future conditions.

[0079] The present invention has been described in detail above. The above description is only a preferred embodiment of the present invention and should not be construed as limiting the scope of the present invention. All equivalent changes and modifications made in accordance with the scope of this application should still fall within the scope of the present invention.

Claims

1. A dry quenching coke removal control method based on model predictive control, characterized in that, Includes the following steps: S1: Divide the time into segments according to the coking plan. All continuous coking pushes with a coking time interval of m to n minutes are considered as the first segment. All continuous coking pushes with a coking time interval of less than m minutes are considered as the second segment. All continuous coking pushes with a coking time interval of more than n minutes are considered as the third segment. If the number of coking pushes in a certain time interval is less than i, merge it with the longer segment of the other two segments. S2: After the segmentation is completed, the total coke loading is calculated based on the number of furnaces in each time interval and the amount of coke loaded in each furnace; S3: Set the initial material level and desired material level, and the desired change in material level height. = Expected material level - Initial material level; S4: Establish a mathematical model for coke discharge rate and material level; the coke discharge rate-material level change model is defined as follows: ； Where pjl is the amount of coke discharged, and t is the duration. It is the density of coke, S g It is the bottom area of ​​the dry quenching furnace; S5: Establish a corresponding model for amplitude and coke discharge rate; the relationship between amplitude and coke discharge rate is expressed as: ； Where a and b are parameters, pjl is the amount of coke discharged, and X is the amplitude; S6: Based on the coke pushing plan, calculate the expected values ​​of amplitude and expected material level height at each moment within the entire planning cycle; S7: Calculate the actual amplitude and expected material level using the MPC algorithm model; The specific calculation method of the MPC algorithm model is as follows: ； ； ； ； ； ； Among them, H( ) is the controlled variable, which in the scenario is the material level at the next moment; It is the planned material level height for the next moment; X is the amplitude; This indicates the change in amplitude at the next moment; This indicates the predicted change in material level at the next moment. M represents the control step size, and P represents the prediction step size; Q, R, and S are all weight matrices.

2. The method according to claim 1, characterized in that, In step S1, m, n, and i are all natural numbers greater than zero, and m < n and i < 10.

3. The method according to claim 1, characterized in that, In step S2, the total amount of coke loaded is expressed by the following formula: Total coke volume = .

4. The method according to claim 1, characterized in that, In step S6, the expected values ​​of amplitude and material level height at each moment within the entire planning cycle are calculated. The specific method is as follows: The average amount of coke removed during duration t is calculated using the following formula: Average coke discharge rate = ΔH × S g ×ρ / t; The corresponding amplitude can be obtained from the average coke discharge rate. After a duration of t1 under this amplitude, the expected material level height is calculated using the following formula: Expected material level height = (ΔH × t / t0) × t1; Where t0 equals 1, and the unit is hours.

Images