Machine learning-based blast furnace injection control system
By using a decision tree model based on machine learning and real-time parameter monitoring, the pulverized coal injection control of the blast furnace is optimized, solving the problems of poor control accuracy and high equipment operating costs in the existing technology, and realizing efficient and economical blast furnace smelting.
Patent Information
- Application Number
- CN202410192244.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-21
- Publication Date
- 2025-11-21
- Estimated Expiration
- 2044-02-21
AI Technical Summary
Existing pulverized coal injection control systems for blast furnaces suffer from poor control accuracy, low pulverized coal injection efficiency or high adjustment frequency, and high equipment operating costs, thus failing to improve smelting efficiency and economy.
A decision tree model is constructed using a machine learning-based decision unit. Real-time parameters of decision features are collected by a monitoring unit. The decision tree model is then used to determine whether the injection parameters of the blast furnace equipment need to be adjusted. Feedback cycles and influence ratio ranges are set, and adjustment commands are generated to optimize injection control.
It improves the accuracy and efficiency of blast furnace injection control, reduces equipment operating costs, ensures that the equipment is always in optimal operating condition, and enhances smelting efficiency and economy.
Smart Images
Figure CN118092169B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of blast furnace injection control technology, and in particular to a blast furnace injection control system based on machine learning. Background Technology
[0002] To ensure the smooth and stable flow of furnace charge in blast furnace production, the pulverized coal injection system must guarantee both the stability of continuous heating to the blast furnace and the complete combustion of the pulverized coal entering the furnace, thereby improving the economic efficiency and environmental friendliness of the steelmaking process. Currently, pulverized coal injection is mainly carried out through the pulverized coal injection control system.
[0003] However, in practice, it has been found that there are two main types of pulverized coal injection control. One is to formulate a pulverized coal supply plan and then not make any adjustments, which has the problems of poor control accuracy and low pulverized coal injection efficiency. The second is to establish a control model and make rapid adjustments based on the real-time changes of the monitoring values. However, this has the problems of high adjustment frequency, and even slight changes in the monitoring values will trigger adjustment commands, which increases the operating cost of the blast furnace equipment and cannot improve the blast furnace smelting efficiency and economy. Summary of the Invention
[0004] The purpose of this application is to provide a blast furnace injection control system based on machine learning to solve the above-mentioned technical problems, thereby improving the efficiency and economy of blast furnace smelting.
[0005] In some embodiments of this application, a decision-making unit is added. The decision-making unit constructs a decision tree model and multiple decision features. The monitoring unit collects real-time parameters of each decision feature to generate the fluctuation amount of the decision feature. Based on the decision tree model, it is determined whether the injection parameters of the blast furnace equipment need to be adjusted, thereby reducing the number of adjustments, improving adjustment efficiency and accuracy, avoiding the problem of frequent adjustments to the operating parameters of the equipment due to small fluctuations in the monitored values, and reducing the operating cost of the equipment.
[0006] In some embodiments of this application, by setting the feedback cycle duration of each decision feature, the operation of the blast furnace equipment is monitored in real time, and various fault risks are warned in a timely manner. At the same time, the blast furnace equipment can be quickly adjusted so that the equipment is always in the optimal operating state, thereby improving the blast furnace smelting efficiency and economy.
[0007] In some embodiments of this application, a blast furnace injection control system based on machine learning is provided, comprising:
[0008] The decision-making unit establishes a decision tree model based on the blast furnace equipment parameters and multiple decision features of the decision tree model;
[0009] The monitoring unit includes multiple monitoring sub-modules, which are used to collect real-time parameters of various decision features;
[0010] The central control unit determines whether to generate an adjustment command based on the real-time parameters of each decision feature and the decision tree model.
[0011] The central control unit includes:
[0012] The first processing module is used to establish a control model and correct the injection parameters of the blast furnace equipment based on the real-time parameters of the decision characteristics and the control model.
[0013] In some embodiments of this application, the decision-making unit includes:
[0014] The first decision module is used to establish a decision feature sequence A, A = (a1, a2, ..., an), based on the historical parameters of the blast furnace equipment, where n is the number of decision features and ai is the i-th decision feature;
[0015] The second decision module is used to set the influence weight value of each decision feature and establish an influence weight value sequence B, B = (b1, b2, ..., bn), where bi is the influence weight value of the i-th decision feature;
[0016] The third decision module is used to set the decision level of all decision features and to build a decision tree model.
[0017] In some embodiments of this application, the third decision module is further configured to:
[0018] The first influence weight range (B1, B2), the second influence weight range (B2, B3), and the third influence weight range (B3, B4) are preset.
[0019] If bi is within the preset first influence weight value range, the third decision module sets the i-th decision feature ai as a first-level decision feature;
[0020] If bi is within the preset second influence weight value range, the third decision module sets the i-th decision feature ai as a secondary decision feature;
[0021] If bi is within the preset third influence weight range, the third decision module sets the i-th decision feature ai as a level-three decision feature.
[0022] In some embodiments of this application, the monitoring unit further includes:
[0023] The first working module is used to set the monitoring and evaluation values of each decision feature and establish a monitoring and evaluation value sequence C, C = (c1, c2, ..., cn), where ci is the monitoring and evaluation value of the i-th decision feature;
[0024] The second working module is used to set the feedback cycle duration for each decision feature based on the monitoring and evaluation value sequence C.
[0025] In some embodiments of this application, setting the feedback period for each decision feature includes:
[0026] The first monitoring and evaluation value range (C1, C2), the second monitoring and evaluation value range (C2, C3), and the third monitoring and evaluation value range (C3, C4) are preset.
[0027] If ci is within the preset first monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset first feedback cycle duration T1, i.e., t = T1;
[0028] If ci is within the preset second monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset second feedback cycle duration T2, i.e., t = T2;
[0029] If ci is within the preset third monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset third feedback cycle duration T3, i.e., t = T3; and T1 > T2 > T3.
[0030] In some embodiments of this application, the central control unit includes:
[0031] The second processing module is used to generate the fluctuation amount of each decision feature based on the real-time parameters of each decision feature.
[0032] The second processing module is also used to establish a fluctuation evaluation value sequence D, D = (d1, d2, ..., dn), where di is the fluctuation evaluation value of the i-th decision feature;
[0033] The third processing module is used to generate a decision evaluation value f based on the fluctuation evaluation series. When the decision evaluation value f is greater than the preset first fluctuation evaluation value threshold F1, an adjustment instruction is generated.
[0034] In some embodiments of this application, generating the decision evaluation value f includes:
[0035] Generate a decision weight coefficient sequence E based on the influence proportion value sequence B, E = (e1, e2, ..., en), where ei is the decision weight coefficient of the i-th decision feature;
[0036] Generate decision evaluation value f,
[0037] In some embodiments of this application, generating adjustment instructions includes:
[0038] Obtain real-time parameters of all decision features at the current time point;
[0039] The injection parameters for the blast furnace equipment are generated based on real-time parameters and the control model.
[0040] In some embodiments of this application, the establishment of the fluctuation evaluation value series includes:
[0041] Based on the decision feature sequence A, select the target decision features sequentially;
[0042] The feedback time nodes of the target decision features are generated based on the feedback cycle duration of the target decision features;
[0043] Based on the feedback time point, obtain the real-time parameters of the target decision characteristics and generate the real-time fluctuation of the target decision characteristics;
[0044] Establish a mapping table between the fluctuation amount and fluctuation evaluation value of the target decision characteristics, and generate the fluctuation evaluation value of the target decision characteristics at the current feedback time node;
[0045] The fluctuation evaluation value of the target decision characteristics is updated cyclically according to the preset feedback time nodes.
[0046] In some embodiments of this application, the central control unit further includes:
[0047] The first correction module is used to set a correction coefficient m based on the fluctuation evaluation value of the target decision characteristics at the current feedback time node, and to correct the duration of the next feedback cycle based on the correction coefficient m.
[0048] The first fluctuation evaluation value range (D1, D2), the second fluctuation evaluation value range (D2, D3), and the third fluctuation evaluation value range (D3, D4) are preset.
[0049] If the fluctuation evaluation value at the current feedback time point is within the preset first fluctuation evaluation value range, the correction coefficient m is set to the preset first correction coefficient m1, that is, m = m1;
[0050] If the fluctuation evaluation value at the current feedback time point is within the preset second fluctuation evaluation value range, the correction coefficient m is set to the preset second correction coefficient m2, that is, m = m2;
[0051] If the fluctuation evaluation value at the current feedback time point is within the preset third fluctuation evaluation value range, the correction coefficient m is set to the preset third correction coefficient m3, that is, m = m3; and m3 < m2 < m1 < 1.
[0052] Compared with existing technologies, the blast furnace injection control system based on machine learning proposed in this application has the following advantages:
[0053] A decision-making unit is added, which constructs a decision tree model and multiple decision features. The monitoring unit collects real-time parameters of each decision feature to generate the fluctuation of the decision feature. Based on the decision tree model, it is determined whether the injection parameters of the blast furnace equipment need to be adjusted, thereby reducing the number of adjustments, improving adjustment efficiency and accuracy, avoiding the problem of frequent adjustments to the equipment's operating parameters due to small fluctuations in the monitored values, and reducing the operating cost of the equipment.
[0054] By setting the feedback cycle duration for each decision feature, the operation of the blast furnace equipment can be monitored in real time, and various fault risks can be warned in a timely manner. At the same time, the rapid adjustment of the blast furnace equipment can be realized, so that the equipment is always in the optimal operating state, thereby improving the blast furnace smelting efficiency and economy. Attached Figure Description
[0055] Figure 1 This is a schematic diagram of a blast furnace injection control system based on machine learning in a preferred embodiment of this application. Detailed Implementation
[0056] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.
[0057] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.
[0058] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.
[0059] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.
[0060] like Figure 1 As shown, a preferred embodiment of the blast furnace injection control system based on machine learning in this application includes:
[0061] The decision-making unit establishes a decision tree model based on the blast furnace equipment parameters and multiple decision features of the decision tree model;
[0062] The monitoring unit includes multiple monitoring sub-modules, which are used to collect real-time parameters of various decision features;
[0063] The central control unit determines whether to generate adjustment instructions based on the real-time parameters of each decision feature and the decision tree model.
[0064] The central control unit includes:
[0065] The first processing module is used to establish a control model and correct the injection parameters of the blast furnace equipment based on the real-time parameters of the decision characteristics and the control model.
[0066] Specifically, the first decision module is used to establish a decision feature sequence A, A = (a1, a2, ..., an), based on the historical parameters of the blast furnace equipment, where n is the number of decision features and ai is the i-th decision feature;
[0067] The second decision module is used to set the influence weight value of each decision feature and establish an influence weight value sequence B, B = (b1, b2, ..., bn), where bi is the influence weight value of the i-th decision feature;
[0068] The third decision module is used to set the decision level of all decision features and to build a decision tree model.
[0069] Specifically, the third decision module is also used for:
[0070] The first influence weight range (B1, B2), the second influence weight range (B2, B3), and the third influence weight range (B3, B4) are preset.
[0071] If bi is within the preset first influence weight value range, the third decision module sets the i-th decision feature ai as the first-level decision feature;
[0072] If bi is within the preset second influence weight value range, the third decision module sets the i-th decision feature ai as a secondary decision feature;
[0073] If bi is within the preset third influence weight range, the third decision module sets the i-th decision feature ai as a third-level decision feature.
[0074] Specifically, the decision characteristics include various operating parameters and external parameters of the blast furnace equipment, including but not limited to the diameter of pulverized coal, the temperature and humidity of pulverized coal, the ambient temperature, the blast furnace temperature, and the wind speed. Based on the type of each operating parameter, historical data, and the impact of fluctuations on the blast furnace equipment, a corresponding influence weight value is generated. The larger the influence weight value, the greater the impact of the corresponding decision characteristics on the overall operation of the blast furnace equipment when fluctuations occur.
[0075] Specifically, based on the importance of decision characteristics, third-level decision characteristics are higher than second-level decision characteristics, which are higher than first-level decision characteristics.
[0076] Specifically, the range of influence weight values can be set based on historical data.
[0077] Specifically, a decision tree model is established based on all decision characteristics. The changes in the values of each decision characteristic are used to determine whether to adjust the real-time injection parameters of the blast furnace equipment. This ensures the operating efficiency of the blast furnace equipment.
[0078] In a preferred embodiment of this application, the monitoring unit further includes:
[0079] The first working module is used to set the monitoring and evaluation values of each decision feature and establish a monitoring and evaluation value sequence C, C = (c1, c2, ..., cn), where ci is the monitoring and evaluation value of the i-th decision feature;
[0080] The second working module is used to set the feedback cycle duration for each decision feature based on the monitoring and evaluation value sequence C.
[0081] Specifically, when setting the feedback period for each decision feature, the following are included:
[0082] The first monitoring and evaluation value range (C1, C2), the second monitoring and evaluation value range (C2, C3), and the third monitoring and evaluation value range (C3, C4) are preset.
[0083] If ci is within the preset first monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset first feedback cycle duration T1, i.e., t = T1;
[0084] If ci is within the preset second monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset second feedback cycle duration T2, i.e., t = T2;
[0085] If ci is within the preset third monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset third feedback cycle duration T3, i.e., t = T3; and T1 > T2 > T3.
[0086] Specifically, the greater the weight of the decision feature, the higher the corresponding monitoring and evaluation value; the higher the monitoring and evaluation value, the shorter the duration of a single feedback monitoring cycle. Multiple feedback time nodes are set based on the feedback monitoring cycle duration. When a feedback time node is reached, the real-time parameters of the corresponding decision feature are obtained.
[0087] Specifically, in the above embodiments, by establishing a monitoring and evaluation value range, the corresponding feedback cycle duration is dynamically adjusted according to the monitoring and evaluation values of different decision characteristics, thereby improving the monitoring efficiency of blast furnace equipment, providing timely warnings of various fault risks, and enabling rapid adjustment of blast furnace equipment so that the equipment is always in the optimal operating state, thereby improving the efficiency and economy of blast furnace smelting.
[0088] In a preferred embodiment of this application, the central control unit includes:
[0089] The second processing module is used to generate the fluctuation amount of each decision feature based on the real-time parameters of each decision feature.
[0090] The second processing module is also used to establish a fluctuation evaluation value sequence D, D = (d1, d2, ..., dn), where di is the fluctuation evaluation value of the i-th decision feature;
[0091] The third processing module is used to generate a decision evaluation value f based on the fluctuation evaluation series. When the decision evaluation value f is greater than the preset first fluctuation evaluation value threshold F1, an adjustment instruction is generated.
[0092] Specifically, generating the decision evaluation value f includes:
[0093] Generate a decision weight coefficient sequence E based on the influence proportion value sequence B, E = (e1, e2, ..., en), where ei is the decision weight coefficient of the i-th decision feature;
[0094] Generate decision evaluation value f,
[0095] Specifically, the larger the influence ratio, the higher the corresponding decision weight coefficient. By increasing the decision weight coefficient, the decision evaluation value becomes more accurate. At the same time, the first fluctuation evaluation value threshold F1 is set based on historical operating data. When the preset conditions are met, the adjustment command will be generated, thereby reducing the number of adjustments, improving adjustment efficiency and accuracy, avoiding the problem of frequent adjustments to the equipment's operating parameters due to small fluctuations in the monitored values, and reducing the equipment's operating costs.
[0096] Specifically, when generating adjustment instructions, the following are included:
[0097] Obtain real-time parameters of all decision features at the current time point;
[0098] The injection parameters for the blast furnace equipment are generated based on real-time parameters and the control model.
[0099] Specifically, since the feedback time points of each decision feature are not the same, when the fluctuation evaluation value of each decision feature changes, the decision evaluation value f will be updated. When the decision evaluation value f is greater than the preset first fluctuation evaluation value threshold F1, the real-time parameters of all decision features are obtained and the optimal injection parameters are regenerated.
[0100] In a preferred embodiment of this application, establishing the fluctuation evaluation value series includes:
[0101] Based on the decision feature sequence A, select the target decision features sequentially;
[0102] The feedback time nodes of the target decision features are generated based on the feedback cycle duration of the target decision features;
[0103] Based on the feedback time point, obtain the real-time parameters of the target decision characteristics and generate the real-time fluctuation of the target decision characteristics;
[0104] Establish a mapping table between the fluctuation amount and fluctuation evaluation value of the target decision characteristics, and generate the fluctuation evaluation value of the target decision characteristics at the current feedback time node;
[0105] The fluctuation evaluation value of the target decision characteristics is updated cyclically according to the preset feedback time nodes.
[0106] Specifically, real-time fluctuation refers to the difference between the real-time parameters of the target decision characteristics at the current feedback time point and the previous feedback time point.
[0107] Specifically, the central control unit also includes:
[0108] The first correction module is used to set a correction coefficient m based on the fluctuation evaluation value of the target decision characteristics at the current feedback time node, and to correct the duration of the next feedback cycle based on the correction coefficient m.
[0109] The first fluctuation evaluation value range (D1, D2), the second fluctuation evaluation value range (D2, D3), and the third fluctuation evaluation value range (D3, D4) are preset.
[0110] If the fluctuation evaluation value at the current feedback time point is within the preset first fluctuation evaluation value range, the correction coefficient m is set to the preset first correction coefficient m1, that is, m = m1;
[0111] If the fluctuation evaluation value at the current feedback time point is within the preset second fluctuation evaluation value range, the correction coefficient m is set to the preset second correction coefficient m2, that is, m = m2;
[0112] If the fluctuation evaluation value at the current feedback time point is within the preset third fluctuation evaluation value range, the correction coefficient m is set to the preset third correction coefficient m3, that is, m = m3; and m3 < m2 < m1 < 1.
[0113] Specifically, the corrected feedback cycle duration ti = m * Ti, where Ti is T1, T2, and T3.
[0114] Specifically, by setting correction coefficients and dynamically adjusting the feedback cycle duration, the monitoring efficiency for various decision-making characteristics can be improved. This allows for timely early warning of various fault risks and enables rapid adjustment of the blast furnace equipment.
[0115] According to the first concept of this application, a decision-making unit is added. The decision-making unit constructs a decision tree model and multiple decision features. The monitoring unit collects real-time parameters of each decision feature to generate the fluctuation amount of the decision feature. Based on the decision tree model, it is determined whether the injection parameters of the blast furnace equipment need to be adjusted, thereby reducing the number of adjustments, improving adjustment efficiency and accuracy, avoiding the problem of frequent adjustments to the operating parameters of the equipment due to small fluctuations in the monitored values, and reducing the operating cost of the equipment.
[0116] According to the second concept of this application, by setting the feedback cycle duration of each decision feature, the operation of the blast furnace equipment can be monitored in real time, and various fault risks can be warned in a timely manner. At the same time, the rapid adjustment of the blast furnace equipment can be realized, so that the equipment is always in the optimal operating state, thereby improving the blast furnace smelting efficiency and economy.
[0117] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.
Claims
1. A blast furnace injection control system based on machine learning, characterized in that, include: The decision-making unit establishes a decision tree model based on the blast furnace equipment parameters; The monitoring unit includes multiple monitoring sub-modules, which are used to collect real-time parameters of various decision features; The central control unit determines whether to generate an adjustment command based on the real-time parameters of each decision feature and the decision tree model. The central control unit includes: The first processing module is used to establish a control model and correct the blast furnace equipment injection parameters based on the real-time parameters of the decision characteristics and the control model. The central control unit includes: The second processing module is used to generate the fluctuation amount of each decision feature based on the real-time parameters of each decision feature. The second processing module is also used to establish a fluctuation evaluation value sequence D, D=(d1,d2…dn), where di is the fluctuation evaluation value of the i-th decision feature; The third processing module is used to generate a decision evaluation value f based on the fluctuation evaluation series. When the decision evaluation value f is greater than the preset first fluctuation evaluation value threshold F1, an adjustment instruction is generated.
2. The blast furnace injection control system based on machine learning as described in claim 1, characterized in that, The decision-making unit includes: The first decision module is used to establish a decision feature sequence A, A=(a1,a2…an) based on the blast furnace equipment parameters, where n is the number of decision features and ai is the i-th decision feature; The second decision module is used to set the influence weight value of each decision feature and establish an influence weight value sequence B, B=(b1,b2…bn), where bi is the influence weight value of the i-th decision feature; The third decision module is used to set the decision level of all decision features and to build a decision tree model.
3. The blast furnace injection control system based on machine learning as described in claim 2, characterized in that, The third decision module is also used for: The first influence weight range (B1, B2), the second influence weight range (B2, B3), and the third influence weight range (B3, B4) are preset. If bi is within the preset first influence weight value range, the third decision module sets the i-th decision feature ai as a first-level decision feature; If bi is within the preset second influence weight value range, the third decision module sets the i-th decision feature ai as a secondary decision feature; If bi is within the preset third influence weight range, the third decision module sets the i-th decision feature ai as a level-three decision feature.
4. The blast furnace injection control system based on machine learning as described in claim 3, characterized in that, The monitoring unit also includes: The first working module is used to set the monitoring and evaluation values of each decision feature and establish a monitoring and evaluation value sequence C, C=(c1,c2…cn), where ci is the monitoring and evaluation value of the i-th decision feature; The second working module is used to set the feedback cycle duration for each decision feature based on the monitoring and evaluation value sequence C.
5. The blast furnace injection control system based on machine learning as described in claim 4, characterized in that, Setting the feedback period for each decision feature includes: The first monitoring and evaluation value range (C1, C2), the second monitoring and evaluation value range (C2, C3), and the third monitoring and evaluation value range (C3, C4) are preset. If ci is within the preset first monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset first feedback cycle duration T1, i.e., t=T1; If ci is within the preset second monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset second feedback cycle duration T2, i.e., t=T2; If ci is within the preset third monitoring and evaluation value range, the feedback cycle duration ti of the i-th decision feature is set to the preset third feedback cycle duration T3, i.e., t=T3; and T1>T2>T3.
6. The blast furnace injection control system based on machine learning as described in claim 5, characterized in that, When generating the decision evaluation value f, the following are included: Generate a decision weight coefficient sequence E, E=(e1,e2…en) based on the influence proportion value sequence B, where ei is the decision weight coefficient of the i-th decision feature; Generate decision evaluation value f, f= .
7. The blast furnace injection control system based on machine learning as described in claim 6, characterized in that, When generating adjustment instructions, the following are included: Obtain real-time parameters of all decision features at the current time point; The injection parameters for the blast furnace equipment are generated based on real-time parameters and the control model.
8. The blast furnace injection control system based on machine learning as described in claim 7, characterized in that, When establishing the fluctuation evaluation value series, the following should be included: Based on the decision feature sequence A, select the target decision features sequentially; The feedback time nodes of the target decision features are generated based on the feedback cycle duration of the target decision features; Based on the feedback time point, obtain the real-time parameters of the target decision characteristics and generate the real-time fluctuation of the target decision characteristics; Establish a mapping table between the fluctuation amount and fluctuation evaluation value of the target decision characteristics, and generate the fluctuation evaluation value of the target decision characteristics at the current feedback time node; The fluctuation evaluation value of the target decision characteristics is updated cyclically according to the preset feedback time nodes.
9. The blast furnace injection control system based on machine learning as described in claim 8, characterized in that, The central control unit also includes: The first correction module is used to set a correction coefficient m based on the fluctuation evaluation value of the target decision characteristics at the current feedback time node, and to correct the duration of the next feedback cycle based on the correction coefficient m. The first fluctuation evaluation value range is preset to (D1, D2), the second fluctuation evaluation value range is (D2, D3), and the third fluctuation evaluation value range is (D3, D4). If the fluctuation evaluation value at the current feedback time point is within the preset first fluctuation evaluation value range, the correction coefficient m is set to the preset first correction coefficient m1, that is, m=m1; If the fluctuation evaluation value at the current feedback time point is within the preset second fluctuation evaluation value range, the correction coefficient m is set to the preset second correction coefficient m2, i.e., m = m2; If the fluctuation evaluation value at the current feedback time point falls within the preset third fluctuation evaluation value range, the correction coefficient m is set to the preset third correction coefficient m3, i.e., m = m3; and m3 <m2<m1<1。
Citation Information
Patent Citations
Waste heat valve control optimization method based on fusion driving
CN115681597A
Power generation control method based on wind power SCADA system
CN117220411A