Method for constructing drift sensitive index of steady-state operating point based on alignment data

Abstract

The invention provides a steady-state operation point drift sensitive index construction method based on alignment data, and belongs to the technical field of metallurgical process control, and the method comprises the steps: firstly, calculating the total causal time difference based on the strip steel operation speed and the furnace area length, carrying out the space-time alignment of the plate temperature and each furnace area variable, and screening a steady-state reference area to construct a baseline statistic; then, on the basis of the aligned data and baseline statistics, a standardized residual error, a drift rate and a temperature-energy consistency index are calculated on line; and finally, carrying out smoothing processing on the indexes, carrying out fusion calculation on the comprehensive drift intensity of each furnace area and the plate temperature, when the comprehensive drift intensity exceeds a preset threshold value, judging that steady-state operation point drift occurs, and identifying a leading furnace area and a drift direction. Through multi-dimensional index fusion, the problems that in the prior art, slow drifting is not sensitive, and a drifting source cannot be accurately positioned are effectively solved, early and accurate detection of steady-state operation point drifting is achieved, and reliable guarantee is provided for stable operation of the production process.

Description

Technical Field

[0001] This invention relates to the field of continuous annealing heating process control technology for cold-rolled strip steel, and in particular to a method for constructing a steady-state operating point drift sensitivity index based on alignment data. Background Technology

[0002] In the continuous annealing heating process of cold-rolled strip steel, even after data alignment and baseline construction are completed, there is still a general lack of index methods on site that can sensitively detect the slow drift of the "steady-state operating point". Although existing technologies may have variable trend monitoring (such as furnace temperature deviation, plate temperature deviation) or feedback control functions, these methods mainly focus on the rapid response to instantaneous deviations or explicit disturbances, and have not specifically constructed sensitive indexes for the specific condition of "slow drift of the steady-state operating point".

[0003] Due to the complex operating conditions of the cold rolling continuous annealing heating process, characterized by strong coupling of multiple variables, large hysteresis, large inertia, and slow response, it is difficult to effectively identify this slow, cumulative drift by relying solely on residual deviation analysis. When the system slowly deviates from its original optimal steady-state operating point after long-term operation due to factors such as gradual changes in equipment performance and environmental fluctuations, conventional methods often present the illusion that "control is still normal in the short term" because the control system re-establishes a local equilibrium near the "new drift operating point." This makes it difficult for operators to detect in time, ultimately leading to gradual changes in product quality and even a large number of hidden defects. Summary of the Invention

[0004] Given the technical defects and drawbacks of existing technologies, there is an urgent need for a method that can construct additional sensitive indicators (such as standardized residuals, drift rate, and temperature-energy consistency indicators) to detect drift of steady-state operating points earlier and more accurately, thereby improving the monitoring capabilities of production operation systems.

[0005] This invention provides a method for constructing a steady-state operating point drift sensitivity index based on aligned data to overcome or at least partially solve the above problems. The specific solution is as follows:

[0006] As a first aspect of the present invention, a method for constructing a steady-state operating point drift sensitivity index based on aligned data is provided, comprising the following steps:

[0007] S1. Calculate the total causal time difference between the plate temperature and the heating conditions of each furnace zone. Based on the total causal time difference, align the strip exit plate temperature data with the heating section data of each furnace zone in time and space to obtain an aligned variable sequence. Based on the aligned variable sequence, select the steady-state reference zone by setting filtering conditions. On the steady-state reference zone, calculate the standard deviation of the furnace temperature and the standard deviation of the exit plate temperature of each furnace zone as baseline statistics.

[0008] S2. Based on the aligned data and the baseline statistics, calculate a set of drift-sensitive indices online. The drift-sensitive indices include the standardized residuals calculated based on the baseline statistics, the drift rate calculated based on sliding window linear fitting, and the temperature-energy consistency index characterizing the relationship between gas flow rate and furnace temperature response.

[0009] In some embodiments, in step S1, the total causal time difference Calculated using the following formula:

[0010]

[0011] in The running time of the strip steel in furnace zone A. , Let a be the length of furnace zone a. For the strip running speed, This is the thermal response time difference (i.e., the time difference when the cross-correlation estimate between the plate temperature and the corresponding furnace temperature is at its maximum).

[0012] The total causal time difference Used to compare strip tapping plate temperature data with corresponding furnace heating section data variables. Alignment is performed to obtain an aligned variable sequence under a unified time grid. The aligned variable sequence includes the aligned furnace temperature, gas flow rate, combustion air flow rate, and outlet plate temperature of each zone.

[0013] In some embodiments, the data variables of each furnace zone heating section include the furnace temperature of each zone, the gas flow rate of each zone, and the combustion air flow rate of each zone. In step S1, the selection of the steady-state reference zone must meet the following conditions:

[0014] The variation range of the furnace temperature setpoint in each zone shall not exceed the first threshold. ,Right now:

[0015]

[0016] Gas flow rate changes in each district shall not exceed the second threshold. ,Right now:

[0017]

[0018] Air flow changes in each district should not exceed the third threshold. ,Right now:

[0019]

[0020] The standard deviation of the outlet plate temperature does not exceed the fourth threshold. ,Right now:

[0021]

[0022] in, , , The screening criteria are set as needed according to the requirements of the steel plant / process, and are located in a sliding window. Within the process, the alignment variable sequence is evaluated, and when all conditions are met, the window center is moved. Included in steady-state reference region

[0023] in, This indicates the furnace temperature setpoint for furnace zone a after alignment. This indicates the gas flow rate in furnace zone a after alignment. This indicates the combustion air flow rate in furnace zone a after alignment. For Centered on, window length is Sliding window, and For window The determination formula indicates that, within the sliding window, the absolute value of the difference between any two times the furnace temperature setpoint, gas flow rate, and air flow rate of each zone does not exceed the corresponding threshold.

[0024] In some embodiments, in step S2, the standardized residuals include standardized residuals of furnace temperature in each zone and standardized residuals of plate temperature, and their calculation formulas are as follows:

[0025] Standardized residuals of furnace temperature in each zone:

[0026]

[0027] Plate temperature standardized residual:

[0028]

[0029] in, for The actual furnace temperature of furnace zone A after time alignment. for The furnace temperature setpoint for furnace zone a at time a. Let A be the standard deviation of furnace temperature in furnace zone a; for The outlet plate temperature after real-time alignment Set the board temperature value. This represents the standard deviation of the export plate temperature.

[0030] In some embodiments, in step S2, the drift rate includes the furnace temperature drift rate and the plate temperature drift rate for each zone, and their calculation formulas are as follows:

[0031] Furnace temperature drift rate in each zone:

[0032]

[0033] Plate temperature drift rate:

[0034]

[0035] in, For Centered on, window length is Sliding window, This represents the mean of the time series within the window. The mean value of the furnace temperature sequence in furnace zone a within the window. The mean of the plate temperature sequence within the window. In order to be in The actual furnace temperature of furnace zone A after time alignment. In order to be in The temperature of the outlet plate after alignment.

[0036] In some embodiments, in step S2, the formula for calculating the temperature-energy consistency index is:

[0037]

[0038] in, For Centered on, window length is Sliding window, for Gas flow rate in furnace zone A after time alignment Let be the mean of the gas flow rate sequence in furnace zone a within the window. for Furnace temperature in furnace zone A after time alignment The mean value of the furnace temperature sequence in furnace zone a within the window. To prevent extremely small constants with a denominator of zero.

[0039] In some embodiments, the method further includes S3: smoothing the drift-sensitive index, calculating the comprehensive drift intensity of each furnace zone and plate temperature, and when the comprehensive drift intensity exceeds a preset threshold, determining that a steady-state operating point drift has occurred, and determining the dominant furnace zone and drift direction.

[0040] In some embodiments, the smoothing of the drift-sensitive index employs an exponentially weighted moving average method, specifically including:

[0041] Smoothed furnace temperature standardized residuals for each zone:

[0042]

[0043] Smoothed furnace temperature drift rates for each zone:

[0044]

[0045] Smoothed plate temperature normalized residual:

[0046]

[0047] Smoothed plate temperature drift rate:

[0048]

[0049] in, for Standardized residual of furnace temperature in furnace zone at time a for Furnace temperature drift rate at time a for Standardized residual of plate temperature at any time for Time plate temperature drift rate, The EWMA smoothing factor for temperature. is the EWMA smoothing coefficient for the temperature slope, and ;

[0050] Calculate the overall drift intensity of each furnace zone The formula is:

[0051]

[0052] in, for Furnace temperature deviation score for furnace zone a at time a for Furnace temperature drift rate score for furnace zone a at time a for Temperature-energy consistency score for furnace zone at time a. The weights are non-negative and satisfy the following conditions: .

[0053] In some embodiments, the furnace temperature deviation score for each zone Furnace temperature drift rate score for each zone Plate temperature deviation score Plate temperature drift rate score and temperature-energy consistency score The calculation formulas are as follows:

[0054]

[0055]

[0056] in, for Standardized residuals of furnace temperature in furnace zone a after time smoothing for The furnace temperature drift rate of furnace zone a after time smoothing for Reference value for furnace temperature drift rate at time a. for Standardized residual temperature after time-smoothing for The temperature drift rate of the board after smoothing is as follows: This is a reference value for board temperature drift rate. for Temperature-energy consistency index of furnace area at time a This is the reference value for temperature-energy consistency in furnace zone A. Let be the temperature-energy consistency dispersion of furnace zone a.

[0057] In some embodiments, the specific method for determining the dominant furnace zone and drift direction in step S3 is as follows:

[0058] Main furnace area Drift intensity in all furnace zones The largest furnace area, namely:

[0059]

[0060] Calculate the dominant furnace area Corresponding furnace temperature drift rate after smoothing If the smoothed furnace temperature drift rate corresponding to the dominant furnace area If the drift direction is determined to be heating; if If the value is less than 0, the drift direction is determined to be cooling.

[0061] The present invention has the following beneficial effects:

[0062] This invention first eliminates the interference of strip operation lag on data correlation through spatiotemporal alignment, ensuring accurate mapping of the causal relationship between furnace temperature and plate temperature. Simultaneously, it combines multi-dimensional indicators (residual, drift rate, energy consistency) to avoid misjudgment based on a single indicator and enhance sensitivity to slight drift. Second, it suppresses the impact of short-term fluctuations during production and highlights long-term drift trends by employing sliding window fitting and smoothing. Baseline statistics are dynamically updated based on a steady-state reference area to adapt to changes in operating conditions at different production stages. Finally, it accurately identifies the dominant drift source by integrating drift intensity information from multiple furnace zones. The determination of drift direction (heating / cooling) provides a direct basis for adjusting heating strategies, avoiding energy waste or quality defects. Attached Figure Description

[0063] Figure 1This invention provides a method for constructing a steady-state operating point drift sensitivity index based on aligned data. Detailed Implementation

[0064] To enable those skilled in the art to better understand the technical solutions of the present invention, exemplary embodiments of the present invention are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0065] Where there is no conflict, the various embodiments of the present invention and the features thereof may be combined with each other.

[0066] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0067] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0068] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0069] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information all comply with relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution follows relevant national laws and regulations (e.g., the "Information Security Technology - Personal Information Security Specification"). For example: appropriate measures are taken for personal information access control; restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; and explicit identity targeting is eliminated when using personal information to avoid precisely locating a specific individual.

[0070] To address at least one of the technical problems existing in the aforementioned related technologies, the present invention provides a method for constructing a steady-state operating point drift sensitivity index based on aligned data. Figure 1 A flowchart illustrating a method for constructing a steady-state operating point drift sensitivity index based on aligned data, provided in an embodiment of the present invention, includes the following steps:

[0071] S1. Data Alignment and Baseline Construction: Based on the strip running speed and the length of each furnace zone, calculate the total causal time difference between the strip temperature and the heating process of each furnace zone; according to the total causal time difference, align the strip exit plate temperature data with the heating section data variables of the corresponding furnace zone in time and space to form an aligned variable sequence under a unified time grid.

[0072] A steady-state reference region is selected from the aligned variable sequence, and the standard deviations of the furnace temperature and the outlet plate temperature for each furnace zone are calculated within the steady-state reference region as baseline statistics.

[0073] S2. Online calculation of drift sensitivity indicators: Based on the aligned variable sequence and the baseline statistics, the following drift sensitivity indicators are calculated in real time:

[0074] Standardized residuals: Based on the baseline statistics, calculate the standardized residuals of furnace temperature and outlet plate temperature for each furnace zone;

[0075] Drift rate: Based on linear fitting of sliding window, the drift rate of furnace temperature and outlet plate temperature in each furnace zone is calculated;

[0076] Temperature-energy consistency index: By analyzing the response relationship between gas flow rate and furnace temperature, an index characterizing the consistency between energy input and temperature output is constructed;

[0077] S3. Drift determination and dominant factor identification: The drift-sensitive index is smoothed, and the index data of each furnace zone and plate temperature are integrated to calculate the comprehensive drift intensity. When the comprehensive drift intensity exceeds the preset threshold, it is determined that a steady-state operating point drift has occurred, and the furnace zone and drift direction that dominate the drift are identified.

[0078] This invention first eliminates the interference of strip operation lag on data correlation through spatiotemporal alignment, ensuring accurate mapping of the causal relationship between furnace temperature and plate temperature. Simultaneously, it combines multi-dimensional indicators (residual, drift rate, energy consistency) to avoid misjudgment based on a single indicator and enhance sensitivity to slight drift. Second, it suppresses the impact of short-term fluctuations during production and highlights long-term drift trends by employing sliding window fitting and smoothing. Baseline statistics are dynamically updated based on a steady-state reference area to adapt to changes in operating conditions at different production stages. Finally, it accurately identifies the dominant drift source by integrating drift intensity information from multiple furnace zones. The determination of drift direction (heating / cooling) provides a direct basis for adjusting heating strategies, avoiding energy waste or quality defects.

[0079] Example Description

[0080] Assuming a continuous annealing furnace in a cold rolling mill contains four heating zones (A1~A4), the strip running speed is 2m / s, and the sampling frequency is 1Hz, after adopting the scheme of this invention:

[0081] S1 Implementation Details:

[0082] Calculate the total causal time difference: For example, the length of furnace zone A1 is 10m, the strip running time is 5s, the heat transfer time difference is 3s, and the total time difference is 8s;

[0083] Align data: Align the furnace temperature of furnace zone A1 at time t with the outlet plate temperature at t+8s;

[0084] Baseline construction: Calculate the standard deviation of furnace temperature (±0.5℃) and the standard deviation of outlet plate temperature (±1.2℃) for A1~A4 in steady-state production data (e.g., 1-hour steady-state production data).

[0085] S2 index calculation:

[0086] Standardized residual: The deviation between the real-time furnace temperature and the set value divided by the baseline standard deviation; for example, a deviation exceeding ±2 is considered abnormal.

[0087] Drift rate: The furnace temperature trend is fitted by a sliding window (10 min). For example, a slope > 0.1℃ / min is considered drift.

[0088] Temperature-energy consistency: An alarm is triggered when the correlation coefficient between gas flow and furnace temperature is lower than a preset value (e.g., 0.8).

[0089] S3 Drift Judgment

[0090] The overall drift intensity is combined with A1~A4 and plate temperature indicators, and the weighting is distributed as residual (40%), drift rate (30%), and consistency (30%).

[0091] The threshold is set to twice the standard deviation of the historical normal data average. When the threshold is exceeded, an alarm is triggered and the dominant furnace area is located (such as A2 temperature rise drift).

[0092] In some embodiments, S1 specifically includes the following steps:

[0093] S101. Data Acquisition: Collect the actual values ​​of strip steel type, strip steel speed, length of each furnace zone, data variables of the heating section of each furnace zone, and strip steel exit plate temperature; wherein, the data variables of the heating section of each furnace zone include furnace temperature of each zone, gas flow rate of each zone, and combustion air flow rate of each zone;

[0094] S102. Data Alignment: Based on the strip speed, the length of each furnace zone, and the thermal response time difference, calculate the residence time of the strip in each furnace zone; using the residence time, align the strip exit plate temperature data with the heating section data variables of each furnace zone according to a unified time grid to obtain an aligned variable sequence; the aligned variable sequence includes the aligned furnace temperature of each zone, the gas flow rate of each zone, the combustion air flow rate of each zone, and the aligned exit plate temperature;

[0095] S103. Steady-state reference region selection: Based on the alignment variable sequence, a steady-state reference region is selected by setting selection criteria; the selection criteria include:

[0096] The variation range of the furnace temperature setpoint in each zone shall not exceed the first threshold.

[0097] The change in gas flow rate in each zone does not exceed the second threshold.

[0098] Airflow variations in each district should not exceed the third threshold.

[0099] The standard deviation of the export plate temperature does not exceed the fourth threshold.

[0100] S104. Baseline statistics calculation: On the steady-state reference zone selected in S103, calculate the standard deviation of furnace temperature and the standard deviation of outlet plate temperature for each furnace zone, as baseline statistics characterizing process stability.

[0101] This scheme establishes a high-precision steady-state operating benchmark for the continuous annealing heating process of cold-rolled strip steel through a systematic data preprocessing and screening method, specifically reflected in the following three aspects:

[0102] Improving the accuracy of data correlation: By comprehensively considering the strip running speed, furnace zone length, and thermal response time difference, the residence time of the strip in each furnace zone is accurately calculated. This step solves the problem of data asynchrony caused by the movement of the strip in the furnace and the delay in heat transfer. It accurately aligns the heating variables (such as furnace temperature and gas flow rate) in each zone of the furnace with the final outlet plate temperature data in the time dimension, providing a reliable data foundation for subsequent steady-state determination and model building.

[0103] Achieving accurate identification of steady-state operating conditions: The steady-state reference zone is defined by using multi-dimensional screening conditions. It not only examines the fluctuation range of key control variables (furnace temperature setpoint, gas flow rate, and air flow rate), but also combines the stability of the final product quality index (outlet plate temperature). This comprehensive judgment method can effectively exclude data generated by non-steady-state operating conditions such as production fluctuations and coil changes, ensuring that the selected data segments truly represent the process in a stable and balanced state.

[0104] Establish reliable statistical benchmarks: Within the selected steady-state reference zone, calculate the standard deviation of furnace temperature and outlet plate temperature for each furnace zone as baseline statistics for the process. These statistics not only reflect the normal fluctuation level of the process under steady-state conditions, but more importantly, they provide crucial reference benchmarks for subsequent steady-state operating point drift detection. By comparing the differences between the current data and these baseline statistics, it is possible to quantitatively determine whether a slow, imperceptible drift has occurred in the process, thereby achieving early warning.

[0105] Example Description

[0106] Suppose a steel plant is producing a batch of the same type of cold-rolled strip steel and needs to monitor its heating process in a continuous annealing furnace. Using the solution of this invention:

[0107] S101. Data Acquisition: When the system starts running, it first collects the current strip type (e.g., galvanized steel), running speed (e.g., 100 m / min), length of each furnace zone (e.g., length of preheating zone, heating zone, soaking zone), and real-time measurements of furnace temperature, gas flow rate, air flow rate, and outlet plate temperature of each furnace zone.

[0108] S102. Data Alignment: Based on the strip speed and furnace length, the system calculates the total time required for the strip to travel from entering the preheating zone to leaving the soaking zone. Simultaneously, considering the delay in heat transfer, the system uses an algorithm to estimate the time difference between when the strip is heated to a point in the furnace and when its temperature is measured by the outlet detector. These two times are added together to obtain the total residence time of the strip in the furnace. Then, the system shifts the outlet plate temperature data forward by this total residence time, aligning it with the furnace heating variables at the corresponding time on the time axis. For example, the outlet plate temperature measured at 2:00 PM is actually determined by the furnace heating status at 1:50 PM; therefore, the plate temperature data at 2:00 PM is correlated with the furnace temperature, gas flow rate, and other data at 1:50 PM.

[0109] S103. Steady-State Reference Zone Screening: The system sets screening criteria, such as requiring the furnace temperature setpoint to change by no more than 5°C within 10 minutes, the gas flow rate to change by no more than 2 cubic meters per hour, and the standard deviation of the outlet plate temperature to not exceed 3°C. The system uses a sliding time window (e.g., 30 minutes) to traverse the aligned data. When all data within a certain time window simultaneously meets all the above conditions, the system marks that window as the "steady-state reference zone".

[0110] S104. Baseline Statistics Calculation: The system extracts all data segments marked as "steady-state reference area". On these data segments, the standard deviations of furnace temperature in the preheating zone, heating zone, and soaking zone, as well as the standard deviation of outlet plate temperature, are calculated respectively. For example, the calculated standard deviations are 2.1℃ for the preheating zone, 1.8℃ for the heating zone, 1.5℃ for the soaking zone, and 2.5℃ for the outlet plate temperature. These values ​​constitute the steady-state baseline statistics under the current production conditions, representing the fluctuation level of the process under normal steady-state conditions.

[0111] In some embodiments, in step S102, the residence time of the strip in each furnace zone is calculated. Use the following formula:

[0112]

[0113] in, , Let a be the length of furnace zone a. For the strip running speed, This is the thermal response time difference, which is the time difference when the cross-correlation estimate between the plate temperature and the corresponding furnace temperature is at its maximum.

[0114] The dwell time Used to compare strip tapping plate temperature data with corresponding furnace heating section data variables. Alignment yields a sequence of alignment variables under a unified time grid, where... A time index representing the k-th sampling time.

[0115] Traditional alignment methods typically only consider the mechanical transport time of the strip within the furnace (i.e., convection time), neglecting the time required for heat to transfer from the furnace to the strip surface and heat it up (i.e., thermal response time). This approach calculates the cross-correlation between the strip temperature and the furnace temperature to accurately estimate this thermal response time difference, and adds it to the convection time to obtain a more physically accurate "total causal time difference." This ensures that the outlet strip temperature data is precisely matched in time with the furnace heating state that actually affects it.

[0116] Since the accuracy of data alignment directly determines the accuracy of subsequent steady-state determination and model building, this solution provides a higher-quality data foundation for subsequent steps through more precise alignment. Whether it's selecting the steady-state reference region or calculating baseline statistics, both rely on the correct temporal relationships between variables. This solution reduces analytical errors caused by data misalignment through physical correction, thereby making the judgment results of the entire monitoring system more reliable and trustworthy.

[0117] Example Description

[0118] Continuing from the previous example, the system is executing step S102, which is data alignment, specifically including:

[0119] Calculate the convection time: For the preheating zone, its length is 50 meters, and the strip running speed is 100 meters / minute (i.e., 1.67 meters / second). According to the formula... The calculated convection time of the strip passing through the preheating zone is approximately 50 / 1.67 ≈ 30 seconds.

[0120] Calculating the thermal response time difference: Analysis of historical data revealed that when the furnace temperature in the preheating zone changes, the change in the outlet plate temperature occurs with a delay. Calculation of the cross-correlation function between the furnace temperature sequence and the plate temperature sequence showed that the cross-correlation value reaches its maximum when the plate temperature sequence lags the furnace temperature sequence by approximately 15 seconds. Therefore, the system determines the thermal response time difference in the preheating zone. It lasts for 15 seconds.

[0121] Calculate the total stay time: according to the formula The total dwell time in the preheating zone is 30 seconds + 15 seconds = 45 seconds.

[0122] Data alignment is performed by shifting the outlet plate temperature data forward by 45 seconds. For example, the outlet plate temperature measured at 2:00:00 PM corresponds to the preheating zone heating status data at 1:59:15 PM. In this way, the system accurately correlates the outlet plate temperature with the heating variables of each furnace zone, laying a solid foundation for subsequent steady-state determination and drift detection.

[0123] In some embodiments, the filtering conditions in step S103 include:

[0124] The variation range of the furnace temperature setpoint in each zone shall not exceed the first threshold. ;

[0125] Gas flow rate changes in each district shall not exceed the second threshold. ;

[0126] Air flow changes in each district should not exceed the third threshold. ;

[0127] The standard deviation of the outlet plate temperature does not exceed the fourth threshold. ;

[0128] in, , , , The screening criteria are set as needed according to the requirements of the steel plant / process, and are located in a sliding window. Within the process, the alignment variable sequence is evaluated, and when all conditions are met, the window center is moved. Included in steady-state reference region Furthermore, it is possible to require that a series of consecutive adjacent steady-state sets constitute a longer steady-state reference region.

[0129] Traditional steady-state assessments may only focus on the stability of a single variable (such as the outlet plate temperature). This approach comprehensively assesses the process from three dimensions: input, process, and output. It examines the stability of the control system's input commands (furnace temperature setpoint), the execution of energy inputs (whether the gas and air flow rates are stable), and the stability of the final product's quality indicators (outlet plate temperature). This multi-dimensional assessment can more comprehensively reflect whether the process is in true steady-state equilibrium.

[0130] By setting thresholds, this solution effectively filters out non-steady-state data caused by minor equipment fluctuations, measurement noise, or transient disturbances. For example, even if the outlet plate temperature fluctuates slightly within a short period, as long as its standard deviation is within the allowable range and other control variables remain stable, this period can still be considered steady-state. This improves the system's tolerance to normal fluctuations and avoids frequent false alarms due to oversensitivity. By employing a sliding window mechanism, the system can dynamically and continuously identify and update the steady-state reference area as the production process progresses. This ensures that the data used to establish the baseline is always the most stable and representative data under the current production conditions, providing a real-time and accurate reference benchmark for subsequent drift detection.

[0131] Example Description

[0132] Continuing the previous example, the system is executing step S103, namely, steady-state reference region selection, which specifically includes...

[0133] Setting thresholds: For example, based on process requirements, the system sets the following screening thresholds:

[0134] Furnace temperature setpoint variation range 5℃

[0135] Gas flow rate change 2 cubic meters / hour

[0136] airflow changes 3 cubic meters / hour

[0137] Standard deviation of export plate temperature 3℃

[0138] Sliding window determination: For example, the system uses a sliding window of 30 minutes to move from the beginning of the aligned data sequence.

[0139] When the window is moved to the time period of 1:30 PM to 2:00 PM, the system checks the data within the window:

[0140] The difference between the maximum and minimum values ​​of the furnace temperature setpoint within the window is 3℃, which is less than the threshold of 5℃.

[0141] The difference between the maximum and minimum gas flow rates within the window is 1.5 cubic meters per hour, which is less than the threshold of 2 cubic meters per hour.

[0142] The difference between the maximum and minimum airflow within the window is 2.5 cubic meters per hour, which is less than the threshold of 3 cubic meters per hour.

[0143] The standard deviation of the outlet plate temperature within the window is 2.1℃, which is less than the threshold of 3℃.

[0144] Since all conditions are met, the system will include the window center time (1:45 PM) in the steady-state reference region.

[0145] Window movement: As the window continues to slide forward, the system continues to make judgments. In this way, the system can accurately filter out those data segments that are truly in a stable and balanced state from continuous production data, providing a high-quality data source for subsequent calculation of baseline statistics.

[0146] In some embodiments, the formulas for determining the variation range of the furnace temperature setpoint in each zone, the variation of the gas flow rate in each zone, and the variation of the air flow rate in each zone are as follows:

[0147]

[0148] in, Indicates in Temperature setpoint after furnace zone alignment at time a. Indicates in Temperature setpoint after furnace zone alignment at time a. Indicates in The gas flow rate after furnace zone alignment at time a Indicates in The gas flow rate after furnace zone alignment at time a Indicates in The combustion air flow rate after the furnace zone is aligned at time a. Indicates in The combustion air flow rate after the furnace zone is aligned at time a. For Centered on, window length is Sliding window, and For window Any two moments within the window, i.e. The difference (i.e. fluctuation) between any two moments within the sliding window must not exceed the preset threshold; the determination formula indicates that within the sliding window, the absolute value of the difference between any two moments of the furnace temperature setpoint, gas flow rate, and air flow rate in each zone does not exceed the corresponding threshold.

[0149] In some embodiments, the formula for determining the standard deviation of the outlet plate temperature is:

[0150]

[0151] in, For Centered on, window length is Sliding window, for The time series of outlet plate temperature after time alignment, where l represents the outlet position; the determination formula indicates that within the sliding window, the standard deviation of the outlet plate temperature does not exceed the fourth threshold. ;

[0152] In some embodiments, in step S104, the standard deviation of furnace temperature in each furnace zone is calculated. The formula is:

[0153]

[0154] in, The steady-state reference region selected in step S103. for The time series of furnace temperatures in furnace zone 'a' after time alignment, where 'a' represents the furnace zone number. For time; the formula is expressed in the steady-state reference region. Within the furnace, the standard deviation of the furnace temperature in each zone is used as the baseline statistic;

[0155] Calculate the standard deviation of the outlet plate temperature The formula is:

[0156]

[0157] in, The steady-state reference region selected in step S103. for The time series of outlet plate temperature after time alignment, where l represents the outlet position; the formula is expressed in the steady-state reference region. The standard deviation of the outlet plate temperature is used as the baseline statistic.

[0158] In the above embodiments, by calculating and outputting the standard deviation of key variables, the selected steady-state data are transformed into a quantifiable statistical benchmark with clear physical meaning. Specifically:

[0159] The core output of this solution is the standard deviation of the furnace temperature and the outlet plate temperature in each furnace zone. These standard deviations precisely quantify the normal fluctuation level of the process under ideal steady-state conditions. For example, a furnace temperature with a standard deviation of 2°C means that, under steady-state conditions, its normal fluctuation range is approximately within ±2°C. This provides a clear and comparable "benchmark" for subsequent drift detection.

[0160] By comparing current data with these baseline statistics, process status can be transformed from raw, dimensional physical quantities (such as temperature and flow rate) into unitless, standardized indicators. For example, by calculating the deviation of the current furnace temperature from the set value and then dividing it by the standard deviation of that furnace temperature, a "standardized residual" can be obtained. This residual directly reflects the ratio of the current fluctuation to the historical steady-state fluctuation, allowing data from different furnace areas, different times, and different steel grades to be compared and judged under the same standard, greatly enhancing the versatility and accuracy of the monitoring system.

[0161] Example Description

[0162] Continuing from the previous example, the system has completed step S103, selecting a series of steady-state reference regions. For example, if these areas cover data from multiple time periods such as 1:30 PM to 2:00 PM and 2:15 PM to 2:45 PM, then:

[0163] Calculate the standard deviation of furnace temperature: The system extracts the standard deviation in the steady-state reference region. The system collects furnace temperature data for all preheating zones. Assuming there are 1000 data points, the system calculates the standard deviation of these 1000 data points, obtaining σ(Tfa) = 2.1℃. Similarly, the standard deviation of the furnace temperature in the heating zone is calculated to be 1.8℃, and the standard deviation of the furnace temperature in the soaking zone is calculated to be 1.5℃.

[0164] Calculate the standard deviation of the outlet plate temperature: The system extracts the standard deviation in the steady-state reference region. Calculate the standard deviation of all outlet plate temperature data within the range to obtain... =2.5℃.

[0165] Establishing baseline statistics: At this point, the system has established a steady-state baseline for the current production conditions. These baseline statistics (standard deviation of furnace temperature in the preheating zone: 2.1℃, heating zone: 1.8℃, soaking zone: 1.5℃, and outlet plate temperature: 2.5℃) are saved as a reference for subsequent real-time monitoring. If, during subsequent production, the temperature fluctuation of a furnace significantly exceeds its corresponding baseline standard deviation, the system can issue an early warning, indicating a possible anomaly or drift in that area.

[0166] In some embodiments, the method further includes S105, which includes, based on the steady-state reference region Based on the aligned variable sequence, a prediction model for the strip tapping temperature is established. The prediction model is as follows:

[0167]

[0168] Where f is a prediction function obtained through training by a machine learning algorithm. for The outlet plate temperature after real-time alignment for Furnace temperature in furnace zone A after time alignment for Gas flow rate in furnace zone A after time alignment for The combustion air flow rate of furnace zone a after time alignment, where a represents the furnace zone number and l represents the outlet position; the prediction model is used to predict the outlet plate temperature based on the furnace temperature, gas flow rate and combustion air flow rate of each zone.

[0169] In the above embodiments, by establishing a plate temperature prediction model based on machine learning, the results of the aforementioned data alignment and steady-state screening are transformed into an intelligent tool with predictive capabilities, specifically reflected in the following two aspects:

[0170] Achieving Soft Measurement and Prediction of Strip Temperature: In the continuous cold rolling process, the exit strip temperature is the final product quality indicator, but its measurement often suffers from lag or blind spots. This solution utilizes high-quality data established on a steady-state reference zone to train a machine learning model. This model can accurately predict the temperature of the strip about to exit the furnace based on the real-time heating status (furnace temperature, fuel gas, air) of each zone within the furnace. This provides operators with a "virtual instrument" that allows them to anticipate product quality and achieve proactive control.

[0171] Providing a basis for process optimization and drift diagnosis: This predictive model is not only a predictive tool, but also a "digital twin" of the process mechanism. When there is a significant deviation between the actual measured plate temperature and the model's predicted plate temperature, this deviation itself reveals potential anomalies or drift in the process. By analyzing the pattern of the deviation, the source of the drift can be diagnosed, for example, whether it is a decrease in the heat transfer efficiency of a certain furnace area or a problem with the combustion system, thus providing operators with more specific clues for corrective action.

[0172] Example Description

[0173] Continuing with the previous example, the system has completed steps S101 to S104, obtained high-quality steady-state reference region data, and calculated baseline statistics. S105 includes:

[0174] Model training: The system will train the steady-state reference region. All data within the system is used as the training set. The input features (X) include: preheating zone furnace temperature, preheating zone gas flow rate, preheating zone air flow rate, heating zone furnace temperature, heating zone gas flow rate, heating zone air flow rate, soaking zone furnace temperature, soaking zone gas flow rate, and soaking zone air flow rate. The output label (Y) is the corresponding outlet plate temperature. The system selects a machine learning algorithm (e.g., random forest or neural network) for training to obtain a prediction function f.

[0175] Model application: During the production process, the system collects heating variable data of each furnace zone in real time and aligns them using the method in step S102. Then, the aligned data is input into the trained prediction model f, and the model outputs a predicted outlet plate temperature value.

[0176] Deviation analysis: The system compares the plate temperature value predicted by the model with the actual measured plate temperature value. If the deviation is within the normal range, the process is operating normally. If the deviation continues to increase, it indicates that the steady-state operating point may have drifted, requiring operators to pay attention and check the relevant equipment or parameters.

[0177] In some embodiments, the standardized residuals include standardized residuals of furnace temperature in each zone and standardized residuals of plate temperature, and their calculation formulas are as follows:

[0178] Standardized residuals of furnace temperature in each zone:

[0179]

[0180] Plate temperature standardized residual:

[0181]

[0182] in, The actual furnace temperature of furnace zone a after alignment. The setpoint for furnace temperature in furnace zone A. Let A be the standard deviation of furnace temperature in furnace zone a; To align the outlet plate temperature, Set the board temperature value. The standard deviation of the export plate temperature;

[0183] The drift rate includes the furnace temperature drift rate and plate temperature drift rate for each zone, and their calculation formulas are as follows:

[0184] Furnace temperature drift rate in each zone:

[0185]

[0186] Plate temperature drift rate:

[0187]

[0188] in, This represents the mean of the time series within the window. The mean value of the furnace temperature sequence in furnace zone a within the window. The mean of the plate temperature sequence within the window. In order to be in The actual furnace temperature of furnace zone A after time alignment. In order to be in The temperature of the outlet plate after alignment.

[0189] The formula for calculating the temperature-energy consistency index is as follows:

[0190] ,

[0191] in, For Centered on, window length is Sliding window, , The gas flow rate in furnace zone A after alignment. Let be the mean of the gas flow rate sequence in furnace zone a within the window. The furnace temperature of furnace zone a after alignment, The mean value of the furnace temperature sequence in furnace zone a within the window. To prevent extremely small constants with a denominator of zero.

[0192] In the above embodiments, by dividing the deviation between the actual value and the set value by the standard deviation of the steady-state reference zone, the deviations of different furnace zones and different dimensions are unified into dimensionless standardized residuals. This allows for direct comparison of the drift degree of different furnace zones, avoiding misjudgments caused by different furnace zone set temperatures or different fluctuation ranges, and significantly improving the sensitivity of drift detection. Through linear fitting of a sliding window, instantaneous temperature fluctuations are transformed into trend-based drift rates. This method can effectively filter out random noise in the production process, capture slow, cumulative drift trends, and solve the problem of traditional methods being insensitive to slow drift. By analyzing the response relationship between gas flow and furnace temperature, an index characterizing the matching degree between energy input and temperature output is constructed. This index can effectively distinguish between "normal adjustment" and "abnormal drift." For example, when the gas flow increases but the furnace temperature response is insufficient, it may indicate air leakage in the furnace or a decrease in heat transfer efficiency, thus providing key clues for diagnosing the cause of drift.

[0193] The three indicators described above depict the state of the steady-state operating point from three dimensions: "current deviation," "trend of change," and "response relationship." By comprehensively evaluating these indicators, the system can detect potential drift risks earlier, before obvious product quality problems occur, and preliminarily locate the furnace area where the problem may occur, providing operators with timely corrective guidance.

[0194] Example Description

[0195] A continuous annealing furnace in a cold rolling mill contains four heating zones (A1~A4), with a strip running speed of 2 m / s and a sampling frequency of 1 Hz. The current production specification is non-oriented silicon steel (NGO), and the target exit plate temperature is 850℃.

[0196] Implementation process after adopting the solution of this invention:

[0197] Calculate the standardized residuals:

[0198] The system first calculates the standard deviation of the furnace temperature in furnace zone A1 from the steady-state reference area (such as stable production data from the past hour), for example, 0.5℃, and the standard deviation of the outlet plate temperature, for example, 1.2℃.

[0199] During real-time calculation, if the actual furnace temperature in furnace zone A1 is 802℃ and the set value is 800℃, then its standardized residual is (802-800) / 0.5 = 4. This indicates that the furnace temperature in this zone has deviated from the set value by 4 standard deviations, which is a significant anomaly.

[0200] Calculate the drift rate:

[0201] The system selects a preset time (e.g., 10 minutes) sliding window (e.g., 600 data points) to perform linear fitting on the furnace temperature of furnace zone A1. If the slope obtained by fitting is 0.05℃ / min, it indicates that the furnace temperature of the furnace zone is slowly rising at a rate of 0.05℃ per minute, and there is an obvious trend of temperature rise drift.

[0202] Calculate temperature-energy consistency

[0203] Within the same 10-minute sliding window, the system analyzes the response relationship between gas flow and furnace temperature in furnace zone A1.

[0204] If the calculated temperature-energy consistency index is 0.5, which is far below the reference value in the steady-state reference zone (such as 0.9), it indicates that the furnace temperature response is very sluggish when the gas flow rate increases. This suggests that there may be equipment problems such as furnace leakage, burner blockage, or decreased heat transfer efficiency, leading to a mismatch between energy input and temperature output.

[0205] Through the calculation of the above three indicators, the system not only detected a significant temperature drift in the A1 furnace area, but also preliminarily determined through the temperature-energy consistency index that the drift may be related to equipment performance degradation rather than a simple control deviation, providing on-site maintenance personnel with a precise direction for fault diagnosis.

[0206] In some embodiments, the method further includes calculating reference estimates for each index in the steady-state reference region, specifically including:

[0207] Furnace temperature drift rate reference value:

[0208]

[0209] Reference value for board temperature drift rate:

[0210]

[0211] Temperature-Energy Consistency Reference Values:

[0212]

[0213] Temperature-energy uniformity dispersion:

[0214]

[0215] Where P95 represents the 95th percentile, median represents the median, and MAD represents the absolute deviation from the median.

[0216] In the above embodiments, by extracting statistical characteristics (such as 95th percentile, median, and dispersion) of indicators such as drift rate and temperature-energy consistency from the historical steady-state reference area, a dynamic and adaptive judgment benchmark is established for subsequent drift detection. This benchmark reflects the inherent fluctuation level of the production system under normal and stable operating conditions, making drift judgment more consistent with actual working conditions and avoiding misjudgments that may be caused by using fixed thresholds. By using the 95th percentile as the reference value for the drift rate, the extreme fluctuations that occasionally occur during steady-state operation can be effectively eliminated, ensuring that the detection system has tolerance for fluctuations within the normal range, while maintaining high sensitivity to abnormal drifts that exceed the normal range. By using the median and median absolute deviation (MAD) as the reference values ​​for temperature-energy consistency, the influence of outliers can be resisted, and the response relationship between gas and furnace temperature under normal conditions can be estimated more robustly, thereby more accurately identifying abnormal changes in the response relationship.

[0217] By comparing the current indicators with reference estimates (such as furnace temperature deviation scores for each zone) Furnace temperature drift rate score for each zone Plate temperature deviation score Plate temperature drift rate score and temperature-energy consistency score This allows the degree of drift to be quantified into a relative value (score), enabling operators to intuitively understand the severity of the current drift relative to historical normal levels, thus providing a quantitative basis for decision-making.

[0218] Example Description

[0219] Suppose that in the A3 furnace area of ​​a continuous annealing furnace in a cold rolling mill, the system has already filtered out the steady-state reference area data from the past 24 hours.

[0220] Implementation process after adopting the solution of this invention

[0221] Calculate the reference value for drift rate

[0222] The system extracts all furnace temperature drift rate data from furnace zone A3 within the steady-state reference region, calculates the 95th percentile of its absolute value, and obtains... =0.02℃ / min. This means that during normal steady-state operation, the temperature drift rate of the A3 furnace is less than 0.02℃ / min for 95% of the time.

[0223] Calculate temperature-energy consistency reference values

[0224] The system extracts all temperature-energy consistency index data of furnace area A3 within the steady-state reference region, calculates the median, and obtains... =0.9. This means that under normal conditions, for every unit increase in gas flow rate in the A3 furnace area, the furnace temperature rises by an average of 0.9℃.

[0225] At the same time, the median absolute deviation (MAD) of this indicator is calculated to obtain... =0.05. This reflects the fluctuation range of this response relationship under normal conditions.

[0226] Through the above calculations, the system establishes a dynamic judgment benchmark for the A3 furnace area. When the drift rate of the A3 furnace area exceeds 0.02℃ / min, or the temperature-energy consistency index deviates significantly from 0.9 (considering a fluctuation range of 0.05), the system can determine that the furnace area has experienced steady-state operating point drift. This benchmark, established through self-learning based on historical data, is more scientific and accurate than a fixed threshold.

[0227] In some embodiments, the smoothing of the drift-sensitive index employs an exponentially weighted moving average method, specifically including:

[0228] Smoothed furnace temperature standardized residuals for each zone:

[0229]

[0230] Smoothed furnace temperature drift rates for each zone:

[0231]

[0232] Smoothed plate temperature normalized residual:

[0233]

[0234] Smoothed plate temperature drift rate:

[0235]

[0236] in, For the standardized residuals of furnace temperature in each zone, For the furnace temperature drift rate in each zone, For the plate temperature standardized residual, For plate temperature drift rate, The EWMA smoothing factor for temperature. The EWMA smoothing coefficient for the temperature slope. and Parameters can be adjusted as needed. In this invention, considering the slow change of the time-steady-state operating point, the parameters are typically set as follows: ;

[0237] Calculate the overall drift intensity of each furnace zone The formula is:

[0238]

[0239] in, The furnace temperature deviation of furnace zone A is scored. The furnace temperature drift rate of furnace zone a is scored. The temperature-energy consistency score for furnace zone A is given. The weights are non-negative and satisfy the following conditions: .

[0240] In the above embodiments, the exponentially weighted moving average (EWMA) method is used to smooth the original drift-sensitive index. This method can effectively filter out short-term fluctuations caused by sensor noise and instantaneous disturbances, and highlight long-term drift trends caused by gradual changes in equipment performance and environmental fluctuations. By setting different smoothing coefficients (e.g., the smoothing coefficient for temperature slope is smaller than that for temperature residuals), it can more sensitively capture slowly changing drift trends while maintaining the ability to respond to steady-state deviations, thereby improving the anti-interference capability and stability of the detection system. By weighted fusion of deviation scores, drift rate scores, and temperature-energy consistency scores, a comprehensive drift intensity index is constructed. This fusion method overcomes the limitations that may exist with single indicators (e.g., the residual index is not sensitive to slow drift, while the drift rate index is sensitive to instantaneous fluctuations), and can more comprehensively and accurately reflect the overall drift state of the steady-state operating point. By adjusting the weighting coefficients, it can flexibly adapt to the different requirements of drift sensitivity for different furnace areas or different processes, making the detection results more targeted and reliable.

[0241] The overall drift intensity is a quantified value that can be directly compared with a preset threshold to automatically determine the drift of the steady-state operating point. This provides a clear and executable decision-making basis for subsequent automatic correction control or operation guidance, reducing reliance on human experience and the risk of misjudgment.

[0242] Example Description

[0243] Suppose that in the A2 furnace area of ​​a continuous annealing furnace in a cold rolling mill, the system has calculated the drift-sensitive index at the current moment in real time, but the data fluctuates to some extent.

[0244] Implementation process after adopting the solution of this invention

[0245] Indicator smoothing:

[0246] The system received the original furnace temperature drift rate index for furnace zone A2, which fluctuated drastically in a short period of time (e.g., -0.02, 0.05, -0.01, 0.03 ℃ / min).

[0247] The system uses the EWMA method for smoothing and sets the smoothing coefficient. ,like =0.1. After smoothing, a stable furnace temperature drift rate of 0.01 ℃ / min was obtained, which effectively filtered out noise and clearly showed that there was a slight heating trend in the furnace area.

[0248] The comprehensive drift intensity system obtains various scores after smoothing in the A2 furnace area, such as furnace temperature deviation score = 0.8, furnace temperature drift rate score = 1.2, and temperature-energy consistency score = 1.5.

[0249] The system is based on preset weights (assuming) Calculate the overall drift intensity:

[0250] = 0.4 * 0.8 + 0.3 * 1.2 + 0.3 * 1.5 = 1.13

[0251] The overall drift intensity (1.13) exceeded the preset drift threshold (e.g., 1.0), and the system determined that a steady-state operating point drift had occurred in the A2 furnace area.

[0252] Through smoothing and comprehensive scoring, the system not only eliminated the interference of instantaneous fluctuations, but also accurately determined that there was a significant steady-state operating point drift in the A2 furnace area by integrating multi-dimensional information, providing a reliable basis for subsequent alarm and correction operations.

[0253] In some embodiments, the furnace temperature deviation score for each zone Furnace temperature drift rate score for each zone Plate temperature deviation score Plate temperature drift rate score and temperature-energy consistency score The calculation formulas are as follows:

[0254]

[0255] Furnace temperature deviation score for each zone Indicates in At that time, the degree to which the overall furnace temperature in the furnace area deviates from the set value is directly taken as the smoothed and standardized residual. The absolute value is used as the deviation score when When this occurs, it indicates that the furnace temperature in that area is basically operating along the set value; when When this occurs, it indicates that the furnace temperature in that area has deviated from one standard deviation, and so on.

[0256]

[0257] Furnace temperature drift rate score for each zone Used to characterize the strength of the furnace temperature change trend in this furnace area (e.g., slow rise, slow fall, or trend towards stability), when When this occurs, it indicates that the furnace temperature change trend in that area has approached the normal maximum level (the boundary of steady-state operation). This indicates that the furnace temperature change trend in this area has exceeded the steady-state range, and there is a possibility of steady-state operating point drift.

[0258]

[0259] Plate temperature deviation rating This indicates the severity of the current deviation of the strip temperature from the set value;

[0260]

[0261] Plate temperature drift rating Used to indicate whether there is a continuous upward / downward trend in plate temperature along the length (or time) direction, and the strength of such fluctuations;

[0262]

[0263] Temperature-Energy Consistency Score This rating indicates whether the "temperature response" of each furnace zone matches the "energy input / ratio". This rating is mainly used to help determine whether the drift is "caused by normal energy regulation" or "abnormal (heat dissipation, air leakage, measurement deviation, etc.)".

[0264] in, The normalized residual of furnace temperature in furnace zone a after smoothing. The smoothed furnace temperature drift rate of furnace zone a. This is the reference value for the furnace temperature drift rate in furnace zone a. For the smoothed plate temperature standardized residual, The smoothed board temperature drift rate, This is a reference value for board temperature drift rate. The temperature-energy consistency index for furnace zone A. This is the reference value for temperature-energy consistency in furnace zone A. Let denot be the temperature-energy consistency dispersion of furnace zone a.

[0265] In the above embodiments, by dividing the smoothed residuals, drift rate, and temperature-energy consistency index by their reference values ​​calculated in the steady-state reference region (such as the P95 quantile, median, and dispersion), all indicators are uniformly converted into dimensionless scores. This normalization process eliminates the differences in the dimensions and orders of magnitude of different indicators, enabling the fair and effective weighted fusion of information from the three different dimensions of deviation, trend, and consistency, providing a standardized data foundation for subsequent comprehensive judgment.

[0266] The scoring mechanism transforms raw physical quantities (such as temperature and slope) into relative strengths compared to historical steady-state operating conditions. For example, a drift rate score greater than 1 indicates that the current trend has exceeded the maximum fluctuation level of historical steady-state operation. This provides an objective and quantitative basis for determining "abnormal drift," reducing the subjectivity of human experience-based judgment and improving the reliability of detection results. Since the score is a reference value calculated based on historical steady-state data, the system can automatically adapt to the normal fluctuation range under different steel grades and production loads. When operating conditions change, the system automatically updates the reference value, thereby maintaining the accuracy of drift detection and avoiding false alarms or missed alarms caused by improper fixed threshold settings.

[0267] In some embodiments, the specific method for determining the dominant furnace zone and drift direction in step S3 is as follows:

[0268] Main furnace area Drift intensity in all furnace zones The largest furnace area, namely:

[0269]

[0270] Calculate the dominant furnace area Corresponding furnace temperature drift rate after smoothing If the smoothed furnace temperature drift rate corresponding to the dominant furnace area If >0, then the drift direction is determined to be heating; if If the value is less than 0, the drift direction is determined to be cooling.

[0271] In some embodiments, the method further includes calculating the overall temperature drift intensity of the plate. :

[0272]

[0273] in, For plate temperature deviation scoring, Scoring for plate temperature drift rate, The weight coefficients are non-negative and satisfy the following conditions: .

[0274] In the above embodiments, an overall drift intensity index of plate temperature is constructed by weighted and fused together the two dimensions of "deviation" and "trend" of plate temperature. ; through non-negative weighting coefficients and It allows for flexible adjustment of the importance of "deviation" and "trend" in drift judgment based on specific steel grades, specifications, or quality requirements; and satisfies the requirements through weighting coefficients. The constraints ensured the overall drift strength. The calculation results have normalization characteristics, avoiding misjudgments caused by sudden changes in a single indicator.

[0275] In some embodiments, the method further includes, in At any given time, calculate the drift intensity of all furnace zones. The furnace zone with the highest drift intensity is called the dominant furnace zone. Constantly dominating the furnace area record

[0276]

[0277] In some embodiments, based on the drift intensity of each furnace zone and the overall drift intensity of the plate temperature, it is determined whether steady-state operating point drift has occurred, and when drift is determined to have occurred, the dominant furnace zone and drift direction are located, specifically including...

[0278] Two thresholds are set: furnace temperature drift threshold. and plate temperature drift threshold .

[0279] like This indicates that the temperature fluctuations in all furnace areas are within an acceptable range, and it is determined that no significant steady-state operating point drift was detected at the current moment.

[0280] like This indicates that at the current moment, a drift in the steady-state operating point of the furnace temperature has been detected, and this is determined by the furnace temperature drift rate corresponding to the dominant furnace area. The sign of the value determines whether the temperature drift in the furnace area is increasing or decreasing.

[0281] Similarly, determine the overall strength of the plate temperature drift:

[0282] like This indicates that the fluctuations in plate temperature are within an acceptable range, and it is determined that no obvious steady-state operating point drift was detected at the current moment.

[0283] like If so, it is determined that: at the current moment, a steady-state operating point drift has occurred, and this is reflected in the furnace temperature drift rate corresponding to the dominant furnace area. The sign of the value determines whether the drift direction is heating up or cooling down;

[0284] Among them, the furnace temperature drift rate of the dominant furnace area The sign of the value determines whether the drift direction is increasing or decreasing, specifically including:

[0285] like If so, the drift direction is determined to be an increase in temperature;

[0286] like If so, the drift direction will be determined as cooling.

[0287] In the above embodiments, by comparing the overall drift intensity of all furnace zones, the furnace zone with the highest drift intensity is identified as the "dominant furnace zone." This mechanism simplifies the complex multivariate drift problem into a clear fault source localization problem. By analyzing the sign of the smoothed furnace temperature drift rate corresponding to the dominant furnace zone, the drift direction is clearly determined as "heating up" or "cooling down," which is crucial for subsequent corrective actions. For example, if the drift is determined to be "heating up," it may be necessary to reduce the gas supply to that furnace zone; if it is determined to be "cooling down," it may be necessary to increase the gas supply. This directional judgment provides operators with clear and actionable guidance for corrective actions, avoiding control deterioration due to incorrect directional judgment.

[0288] Example Description

[0289] Assume that a continuous annealing furnace in a cold rolling mill contains 4 heating zones (A1~A4), and the system has calculated the comprehensive drift intensity of each furnace zone in real time.

[0290] Implementation process after adopting the solution of this invention

[0291] Determine the dominant furnace area

[0292] The system obtains the overall drift intensity of each furnace zone at the current moment: assuming .

[0293] According to the formula The system identified the A2 furnace area as having the highest drift intensity (1.5), and therefore determined that the A2 furnace area is the current dominant drift furnace area.

[0294] Determine the drift direction

[0295] The system obtains the smoothed furnace temperature drift rate corresponding to furnace zone A2. = -0.02 ℃ / min.

[0296] According to the judgment rules: if >0 indicates a temperature increase, while <0 indicates a temperature decrease.

[0297] Since -0.02 < 0, the system determines that the drift direction of furnace zone A2 is "cooling".

[0298] The system's final output result: "Steady-state operating point drift detected, dominant furnace zone A2, drift direction is cooling." This result clearly informs the operator that the problem mainly lies in furnace zone A2, and that this furnace zone is slowly cooling down. The operator should immediately check for any abnormalities in the burner, gas supply, or control system of furnace zone A2, and consider appropriately increasing the gas flow setpoint for this furnace zone.

[0299] Based on the same inventive concept, this invention also provides an online detection system for steady-state operating point drift in the continuous annealing heating process of cold-rolled strip steel, comprising:

[0300] The data alignment and baseline construction module is configured to calculate the total causal time difference between the strip temperature and the heating status of each furnace zone based on the strip running speed and the length of each furnace zone; according to the total causal time difference, the strip exit plate temperature data and the heating section variables of the corresponding furnace zone are spatiotemporally aligned to form an aligned variable sequence under a unified time grid; a steady-state reference zone is selected from the aligned variable sequence, and the standard deviation of the furnace temperature and the standard deviation of the exit plate temperature of each furnace zone are calculated within the steady-state reference zone as baseline statistics;

[0301] The drift sensitivity index calculation module is configured to calculate a set of drift sensitivity indices in real time based on the aligned variable sequence and the baseline statistics. The drift sensitivity indices include the standardized residuals calculated based on the baseline statistics, the drift rate calculated based on the sliding window linear fitting, and the temperature-energy consistency index characterizing the relationship between gas flow and furnace temperature response.

[0302] The drift determination and dominant factor identification module is configured to smooth the drift-sensitive index, integrate the index data of each furnace zone and plate temperature, and calculate the comprehensive drift intensity; when the comprehensive drift intensity exceeds a preset threshold, it is determined that a steady-state operating point drift has occurred, and the furnace zone and drift direction that dominate the drift are identified.

[0303] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. A method for constructing a steady-state operating point drift sensitivity index based on aligned data, characterized in that, Includes the following steps: S1. Calculate the total causal time difference between the plate temperature and the heating status of each furnace zone. Based on the total causal time difference, align the strip exit plate temperature data with the heating section data of each furnace zone in time and space to obtain an aligned variable sequence. Based on the aligned variable sequence, filter out the steady-state reference zone by setting filtering conditions. On the steady-state reference zone, calculate the standard deviation of the furnace temperature and the standard deviation of the exit plate temperature of each furnace zone as baseline statistics. S2. Based on the aligned data and the baseline statistics, calculate a set of drift-sensitive indices online. The drift-sensitive indices include the standardized residuals calculated based on the baseline statistics, the drift rate calculated based on sliding window linear fitting, and the temperature-energy consistency index characterizing the relationship between gas flow rate and furnace temperature response.

2. The method according to claim 1, characterized in that, In step S1, the total causal time difference Calculated using the following formula: ； in The running time of the strip steel in furnace zone A. , Let a be the length of furnace zone a. For the strip running speed, This is the thermal response time difference (i.e., the time difference when the cross-correlation estimate between the plate temperature and the corresponding furnace temperature is at its maximum). The total causal time difference Used to compare strip tapping plate temperature data with corresponding furnace heating section data variables. Alignment is performed to obtain an aligned variable sequence under a unified time grid. The aligned variable sequence includes the aligned furnace temperature, gas flow rate, combustion air flow rate, and outlet plate temperature of each zone.

3. The method according to claim 1, characterized in that, The data variables for each furnace zone heating section include furnace temperature, gas flow rate, and combustion air flow rate. In step S1, the steady-state reference zone must meet the following conditions: The variation range of the furnace temperature setpoint in each zone shall not exceed the first threshold. ,Right now: ； Gas flow rate changes in each district shall not exceed the second threshold. ,Right now: ； Airflow variation in each district does not exceed the third threshold. ,Right now: ； The standard deviation of the outlet plate temperature does not exceed the fourth threshold. ,Right now: ； in, , , , The screening criteria are set as needed according to the requirements of the steel plant / process, and are located in a sliding window. Within the process, the alignment variable sequence is evaluated, and when all conditions are met, the window center is moved. Included in steady-state reference region ; in, This indicates the aligned furnace temperature setpoint for furnace zone a. This indicates the gas flow rate in furnace zone a after alignment. This indicates the combustion air flow rate in furnace zone a after alignment. For Centered on, window length is Sliding window, and For window The determination formula indicates that, within the sliding window, the absolute value of the difference between any two times the furnace temperature setpoint, gas flow rate, and air flow rate of each zone does not exceed the corresponding threshold.

4. The method according to claim 1, characterized in that, In step S2, the standardized residuals include the standardized residuals of furnace temperature in each zone and the standardized residuals of plate temperature, and their calculation formulas are as follows: Standardized residuals of furnace temperature in each zone: ； Plate temperature standardized residual: ； in, for The actual furnace temperature of furnace zone A after time alignment. for The furnace temperature setpoint for furnace zone a at time a. Let A be the standard deviation of furnace temperature in furnace zone a; for The outlet plate temperature after real-time alignment Set the board temperature value. This represents the standard deviation of the export plate temperature.

5. The method according to claim 1, characterized in that, In step S2, the drift rate includes the furnace temperature drift rate and the plate temperature drift rate for each zone, and their calculation formulas are as follows: Furnace temperature drift rate in each zone: ； Plate temperature drift rate: ； in, For Centered on, window length is Sliding window, This represents the mean of the time series within the window. The mean value of the furnace temperature sequence in furnace zone a within the window. The mean of the plate temperature sequence within the window. In order to be in The actual furnace temperature of furnace zone A after time alignment. In order to be in The temperature of the outlet plate after alignment.

6. The method according to claim 1, characterized in that, In step S2, the formula for calculating the temperature-energy consistency index is: ； in, For Centered on, window length is Sliding window, for Gas flow rate in furnace zone A after time alignment Let be the mean of the gas flow rate sequence in furnace zone a within the window. for Furnace temperature in furnace zone A after time alignment The mean value of the furnace temperature sequence in furnace zone a within the window. To prevent extremely small constants with a denominator of zero.

7. The method according to claim 1, characterized in that, The method further includes S3: smoothing the drift-sensitive index, calculating the comprehensive drift intensity of each furnace zone and plate temperature, and when the comprehensive drift intensity exceeds a preset threshold, determining that a steady-state operating point drift has occurred, and determining the dominant furnace zone and drift direction.

8. The method according to claim 1, characterized in that, The smoothing process for the drift-sensitive index employs an exponentially weighted moving average method, specifically including: Smoothed furnace temperature standardized residuals for each zone: ； Smoothed furnace temperature drift rates for each zone: ； Smoothed plate temperature normalized residual: ； Smoothed plate temperature drift rate: ； in, for Standardized residual of furnace temperature in furnace zone at time a for Furnace temperature drift rate at time a for Standardized residual of plate temperature at any time for Time plate temperature drift rate, The EWMA smoothing factor for temperature. is the EWMA smoothing coefficient for the temperature slope, and ≤ ; Calculate the overall drift intensity of each furnace zone The formula is: ； in, for Furnace temperature deviation score for furnace zone a at time a for Furnace temperature drift rate score for furnace zone a at time a for Temperature-energy consistency score for furnace zone at time a. The weights are non-negative and satisfy the following conditions: + + =1.

9. The method according to claim 8, characterized in that, The furnace temperature deviation score for each zone Furnace temperature drift rate score for each zone Plate temperature deviation score Plate temperature drift rate score and temperature-energy consistency score The calculation formulas are as follows: ； ； ； ； ； in, for Standardized residuals of furnace temperature in furnace zone a after time smoothing for The furnace temperature drift rate of furnace zone a after time smoothing for Reference value for furnace temperature drift rate at time a. for Standardized residual temperature after time-smoothing for The temperature drift rate of the board after smoothing is as follows: This is a reference value for board temperature drift rate. for Temperature-energy consistency index of furnace area at time a This is the reference value for temperature-energy consistency in furnace zone A. Let denot be the temperature-energy consistency dispersion of furnace zone a.

10. The method according to claim 8, characterized in that, In step S3, the specific method for determining the dominant furnace zone and drift direction is as follows: Main furnace area Drift intensity in all furnace zones The largest furnace area, namely: ； Calculate the dominant furnace area Corresponding furnace temperature drift rate after smoothing If the smoothed furnace temperature drift rate corresponding to the dominant furnace area If >0, then the drift direction is determined to be heating; if If the value is less than 0, the drift direction is determined to be cooling.

Images