Intelligent combustion control method and system for a steel rolling heating furnace

Abstract

This invention relates to the field of industrial automation and intelligent manufacturing technology, and discloses an intelligent combustion control method and system for a steel rolling heating furnace. The method includes acquiring flow rate, furnace temperature, environmental data, and response parameters; aligning the flow rate and calculating a weighted ratio based on the furnace temperature; filtering and mapping if the ratio exceeds the limit to obtain a proportional imbalance offset; extracting and aligning feature vectors based on the offset to obtain a fluctuation feature sequence; mapping the sequence to a frequency range to obtain an amplitude threshold, and if the threshold is exceeded, performing reverse mapping to obtain adjustment suggestions; combining the response parameters with time axis alignment and gain mapping to obtain an optimization scheme; and performing time axis offset processing and deviation calibration to obtain the final control sequence. This method can achieve real-time and accurate closed-loop control under complex dynamic disturbances.

Description

Technical Field

[0001] This invention relates to the field of industrial automation and intelligent manufacturing technology, and in particular to an intelligent combustion control method and system for steel rolling heating furnaces. Background Technology

[0002] Currently, in the wave of green transformation in modern industrial production and manufacturing, combustion control technology for steel rolling furnaces has become a core element in improving energy efficiency, reducing carbon emissions, and ensuring the quality of steel heat treatment. This technology optimizes the heat distribution within the furnace by precisely adjusting the air-fuel ratio and is widely used in industrial control systems for processes such as steel metallurgy, large-scale forging, and metal processing. It is a key technological support for achieving efficient production and energy conservation and emission reduction goals.

[0003] In existing technologies, a control method based on fixed parameters or human experience is typically employed. This involves collecting real-time fuel and air flow rates using flow sensors and employing basic proportional control algorithms (such as PID control) to maintain a preset air-fuel ratio. However, this method often falls short when dealing with complex dynamic conditions. Because the system relies heavily on static settings, it ignores dynamic deviations caused by equipment aging during combustion, changes in the characteristics of the flow meter medium, and the interference of environmental noise on sensor accuracy. Furthermore, existing technologies struggle to quantify the physical response delays of actuators (such as valves and fans), leading to a time misalignment between the control commands issued and the actual combustion state, resulting in furnace temperature fluctuations and even energy waste.

[0004] In summary, existing technologies suffer from poor real-time combustion control under complex dynamic disturbances and inaccurate fuel-air ratio matching. Summary of the Invention

[0005] This invention provides an intelligent combustion control method and system for steel rolling heating furnaces to solve the problems of poor real-time combustion control and inaccurate fuel-air ratio matching under complex dynamic disturbances.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides an intelligent combustion control method for a steel rolling heating furnace, comprising: Acquire fuel air flow data, furnace temperature distribution data, environmental impact data, and response parameters of actuators; The flow rate synchronization sequence is obtained by time series alignment of the fuel-air flow rate data, and spatial weighting calculation is performed based on the flow rate synchronization sequence and the furnace temperature distribution data to obtain the preliminary fuel-air ratio. If the initial fuel-air ratio exceeds the preset flow matching threshold, the environmental impact data is filtered to obtain a clean environmental noise value, and the clean environmental noise value is combined with the furnace temperature distribution data to perform correlation mapping to obtain the ratio imbalance offset. Abnormal fluctuation feature vectors are obtained by extracting abnormal fluctuation features based on the proportional imbalance offset, and time scale alignment is performed based on the historical records associated with the abnormal fluctuation feature vectors to obtain an aligned fluctuation feature sequence containing fluctuation amplitude. The amplitude threshold is obtained by frequency interval mapping based on the alignment fluctuation feature sequence. If the fluctuation amplitude of the alignment fluctuation feature sequence exceeds the amplitude threshold, the abnormal fluctuation mode is reverse mapped to obtain combustion parameter adjustment suggestions. Based on the combustion parameter adjustment suggestions and the response parameters of the execution components, a time axis alignment correlation dataset is obtained, and a gain mapping is performed based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme. The optimized combustion parameter adjustment scheme is used to perform variable time axis offset processing to obtain the sequence execution time axis, and the time deviation is calibrated according to the sequence execution time axis to obtain the final combustion control sequence.

[0007] Secondly, the present invention provides an intelligent combustion control system for a steel rolling heating furnace, comprising: The data acquisition module is used to acquire fuel air flow data, furnace temperature distribution data, environmental impact data, and response parameters of the actuators. The preliminary calculation module is used to align the fuel-air flow rate data with the time series to obtain a flow rate synchronization sequence, and to perform spatial weighted calculation based on the flow rate synchronization sequence and the furnace temperature distribution data to obtain a preliminary fuel-air ratio. The offset analysis module is used to filter the environmental impact data to obtain a clean environmental noise value if the initial fuel-air ratio exceeds a preset flow matching threshold, and then perform correlation mapping based on the clean environmental noise value and the furnace temperature distribution data to obtain the ratio imbalance offset. The feature alignment module is used to extract abnormal fluctuation features based on the proportional imbalance offset to obtain an abnormal fluctuation feature vector, and to perform time scale alignment based on the historical records associated with the abnormal fluctuation feature vector to obtain an aligned fluctuation feature sequence containing the fluctuation amplitude. The suggestion generation module is used to obtain an amplitude threshold by mapping the frequency range according to the aligned fluctuation feature sequence. If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, the abnormal fluctuation mode is reverse mapped to obtain a combustion parameter adjustment suggestion. The scheme optimization module is used to perform time axis alignment and correlation based on the combustion parameter adjustment suggestions and the response parameters of the execution components to obtain a time delay correlation deviation dataset, and to perform gain mapping based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme. The sequence construction module is used to perform variable time axis offset processing according to the optimized combustion parameter adjustment scheme to obtain the sequence execution time axis, and to perform time deviation calibration according to the sequence execution time axis to obtain the final combustion control sequence.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention significantly improves the initial accuracy of fuel-air ratio by multi-source data fusion and spatial mapping correction. First, this invention uses sensors to acquire time-stamped flow and temperature data, and employs a time-series alignment algorithm to handle data synchronization issues. Then, it extracts the spatial mapping relationship of heat distribution based on differences in furnace temperature distribution areas, generating corrected weights to weight the flow matrix. This approach, by refining the global ratio into a local balance calculation of the fusion space heat demand, effectively suppresses combustion imbalance caused by equipment operation deviations or uneven heat distribution, thereby improving the heating quality of steel and providing precise input for the dynamic optimization control of the entire furnace.

[0009] (2) This invention enhances the robustness of the control system under complex operating conditions through environmental noise quantification analysis and dynamic frequency determination. When the fuel-air ratio exceeds the threshold, the system extracts and filters environmental interference noise, constructs an association mapping matrix to quantify the specific offset of the ratio imbalance; simultaneously, based on the actual frequency of the aligned fluctuation characteristic sequence, it dynamically determines the preset fluctuation range to identify valid anomalies. This process can eliminate short-term abnormal fluctuations and identify real interference trends, enabling the system to adapt to various complex operating conditions such as high load or high noise, thereby effectively preventing potential combustion imbalance and ensuring the continuity and stability of the heating furnace operation.

[0010] (3) This invention achieves long-term stability of the combustion environment by integrating the response delay of the actuator with real-time deviation compensation. This invention obtains a time-delay correlation dataset by acquiring historical sensor deviation records and correlating the physical lag time of the actuator, and calculates the real-time deviation compensation coefficient accordingly. Subsequently, the delay compensation time is used to perform time-axis offset processing on the control sequence to construct the final combustion control sequence. This scheme solves the problem of misalignment between command issuance and actual response in the time dimension. By issuing commands in advance or fine-tuning the command intensity, system lag and state drift are offset, thereby improving the accuracy of control precision, effectively maintaining the stability of the combustion environment in the furnace, and continuously optimizing energy efficiency. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the intelligent combustion control method for steel rolling heating furnaces provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the intelligent combustion control system for a steel rolling heating furnace provided in the second embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] Reference Figure 1 The first embodiment of the present invention provides an intelligent combustion control method for a steel rolling heating furnace, comprising the following steps: S11, acquire fuel air flow data, furnace temperature distribution data, environmental impact data and response parameters of actuators; S12, the flow synchronization sequence is obtained by time series alignment based on the fuel-air flow data, and spatial weighting calculation is performed based on the flow synchronization sequence and the furnace temperature distribution data to obtain the preliminary fuel-air ratio. S13, if the initial ratio of fuel to air exceeds the preset flow matching threshold, then the environmental impact data is filtered to obtain a pure environmental noise value, and the pure environmental noise value is combined with the furnace temperature distribution data to perform correlation mapping to obtain the ratio imbalance offset. S14, extract abnormal fluctuation features based on the proportional imbalance offset to obtain an abnormal fluctuation feature vector, and perform time scale alignment based on the historical records associated with the abnormal fluctuation feature vector to obtain an aligned fluctuation feature sequence containing fluctuation amplitude. S15, frequency range mapping is performed based on the aligned fluctuation feature sequence to obtain an amplitude threshold. If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, reverse mapping of abnormal fluctuation mode is performed to obtain combustion parameter adjustment suggestions. S16, Based on the combustion parameter adjustment suggestion and the response parameters of the execution component, time axis alignment is performed to obtain a time delay correlation deviation dataset, and gain mapping is performed based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme; S17, according to the optimized combustion parameter adjustment scheme, the variable time axis offset processing is performed to obtain the sequence execution time axis, and the time deviation calibration is performed according to the sequence execution time axis to obtain the final combustion control sequence.

[0014] In step S11, fuel air flow data, furnace temperature distribution data, environmental impact data, and actuator response parameters are acquired, including: The instantaneous value of fuel flow and the dynamic change of air flow are obtained from the vortex flow meter, and the data are correlated and stored to obtain fuel-air flow data. Local furnace temperature values ​​are obtained from thermocouples, and the furnace temperature distribution data is obtained by calculating the regional average value based on the local furnace temperature values. Environmental interference signals are acquired from environmental sensing devices to obtain environmental impact data; The signal transmission period between the vortex flowmeter and the thermocouple is obtained, and the response parameters of the actuator are calculated by time accumulation based on the signal transmission period.

[0015] During the acquisition of fuel-air flow data, the system captures raw data streams at a fixed acquisition frequency of once per second using vortex flow meters installed on the gas pipeline and combustion air pipeline of the steel rolling furnace. The instantaneous fuel flow value refers to the volumetric flow rate recorded by the gas flow meter at a specific time cross-section, such as 500 cubic meters of gas per hour acquired in a certain second. The dynamic change in air flow refers to the numerical fluctuation sequence of the combustion air flow meter at continuous time points, such as 6,000 cubic meters of air per hour. Each acquired flow value is assigned a timestamp accurate to milliseconds, thus forming the basic unit in the flow synchronization sequence for subsequent combustion ratio calculation. The acquired instantaneous fuel flow value and the dynamic change in air flow are associated and stored according to the same timestamp to form fuel-air flow data. For acquiring furnace temperature distribution data, the system uses multiple thermocouples arranged in different furnace areas such as the preheating zone, heating zone, and soaking zone of the heating furnace to obtain local furnace temperature values. These local furnace temperature values ​​reflect the Celsius temperature at a specific measurement point at the current moment.

[0016] In practical operation, the system performs arithmetic averaging on the temperature values ​​reported by all thermocouples within a pre-defined specific area. This involves summing the temperature values ​​recorded by each thermocouple within the area and dividing the sum by the total number of thermocouples in that area; the quotient is the regional mean. By integrating multiple regional means, furnace temperature distribution data describing the spatial state of heat within the furnace is generated, which is used to assess the local combustion characteristics of different areas. When acquiring environmental impact data, the system captures interference noise values ​​in the furnace's operating environment using vibration sensors or electromagnetic monitoring equipment. This environmental impact data includes interference signals caused by mechanical vibrations or electromagnetic field fluctuations from nearby equipment; this feature is subsequently used to correct for proportional imbalance offsets. To acquire the response parameters of the actuators, the system first retrieves the signal transmission period of the vortex flowmeter and thermocouple in the data transmission path. The signal transmission period refers to the time span required for the sensor to convert the sensed physical signal into a digital signal and transmit it to the control center, for example, 0.2 seconds. Subsequently, the system combines the physical action time required for the actuator, such as a fuel valve or combustion fan, to generate mechanical action from receiving the command, and calculates the sum of the signal transmission period and the physical action time. The total duration obtained is determined as the response parameter of the actuator, which is used for delay compensation in the subsequent control sequence generation.

[0017] It is worth noting that the response parameters of the execution component include signal transmission delay and physical action delay of the execution mechanism. The time accumulation calculation based on the signal transmission period refers to adding delay components such as the physical action duration of the execution mechanism and the medium transmission duration based on the signal transmission period, so as to obtain the response parameters of the whole process.

[0018] It should be noted that the sampling frequency of the fuel air flow data needs to be set with reference to the general technical standards of industrial automation instruments. For example, referring to the relevant specifications of national standards for vortex flow meters used in industrial process control systems, it is usually set in the range of once per second to ten times per second to ensure the real-time capture of flow fluctuations. The regional division of the local furnace temperature values ​​is determined according to the mechanical structure specifications of the heating furnace. For example, the furnace body is divided into independent temperature control sections according to the distribution of the burners. For example, assuming that four thermocouples are installed in the heat spreader zone, and the temperatures recorded in a certain sampling period are 1090 degrees Celsius, 1110 degrees Celsius, 1105 degrees Celsius, and 1095 degrees Celsius respectively, then these four values ​​are added together to get 4400 degrees Celsius. Then, 4400 degrees Celsius is divided by four, and the resulting 1100 degrees Celsius is the local average furnace temperature of that area. The execution component response parameters described in this embodiment are essentially the total time delay in the control loop, including signal conversion delay, network transmission delay, and actuator dynamic response delay. Quantization of these parameters allows for feedforward adjustment of control commands along the time axis. Those skilled in the art should understand that interference noise in the environmental impact data can be extracted using median or mean filtering algorithms after high-frequency sampling to obtain clean environmental noise values ​​as a correction reference.

[0019] In step S12, a flow synchronization sequence is obtained by time-series alignment based on the fuel-air flow rate data, and a spatial weighted calculation is performed based on the flow synchronization sequence and the furnace temperature distribution data to obtain a preliminary fuel-air ratio, including: A flow synchronization sequence is obtained by performing data synchronization processing based on the fuel air flow data; Based on the furnace temperature distribution data, the spatial mapping relationship of the heat distribution pattern inside the furnace is extracted, and the contribution quantification is performed based on the spatial mapping relationship to obtain the heat distribution correction weight; The local combustion state characteristics are obtained by weighting the flow synchronization sequence using the heat distribution correction weight, and a preliminary fuel-air ratio is obtained by mapping the proportions based on the local combustion state characteristics.

[0020] The system first performs spatiotemporal calibration and feature synthesis on multi-dimensional data. The flow synchronization sequence is defined as a fuel and air flow data matrix aligned under a unified time reference, derived from data synchronization processing of the original flow data stream with millisecond-level timestamps obtained in step S11. In practice, due to differences in sensor response or transmission paths, there may be slight temporal misalignments between gas and air flow. The system uses the timestamp of the gas flow sequence as a reference and employs an interpolation algorithm to calculate the estimated value of the air flow sequence at the same time point, thus organizing the originally misaligned raw data into a standard flow matrix with strictly consistent timestamps for each row, thereby obtaining the flow synchronization sequence. The spatial mapping relationship of the furnace heat distribution pattern is defined as a correlation model describing the degree of influence of adjusting the flow in a specific area on the temperature of different temperature control zones throughout the furnace. The sensitivity of each preset temperature control zone to the adjustment command is defined as the ratio between the change in fuel flow in a specific combustion section and the change in temperature in the corresponding temperature control zone. The sensitivity is based on thermal test data from the empty furnace heating test and the full-load operation condition of the heating furnace. It is set according to the physical differences in furnace volume, burner rated power and furnace wall heat storage characteristics in different temperature control zones.

[0021] The system analyzes the real-time temperature of each temperature control segment in the furnace temperature distribution data to identify the uniformity of heat distribution in each region, and performs contribution quantification based on the sensitivity. Specifically, it first calculates the absolute value of the temperature deviation between the current real-time temperature and the target set temperature of each temperature control zone. This absolute value is multiplied by the sensitivity of the corresponding region to obtain a preliminary influence factor. Then, the preliminary influence factor of this region is divided by the sum of the preliminary influence factors of all regions in the entire furnace; the quotient is the priority coefficient value of that region. After normalizing the priority coefficient values ​​of each region, the system determines the resulting value as the heat distribution correction weight. The heat distribution correction weight is a vector containing the weights of each temperature control zone, and the sum of the weight values ​​of all regions must equal one. If, after normalization, the sum of the weight vectors is not equal to one due to rounding errors, the system will proportionally scale the weight values ​​to ensure that the sum of the weight vectors is strictly equal to one, thus ensuring that the total fuel flow remains constant in subsequent weighted calculations and avoiding the introduction of additional system performance errors.

[0022] The local combustion state characteristic is defined as the equivalent combustion parameters of a specific area after incorporating the spatial heat demand. It originates from a weighted calculation of the flow synchronization sequence using the heat distribution correction weight. Specifically, the system multiplies the heat distribution correction weight with the fuel flow rate and air flow rate values ​​in the flow synchronization sequence, and the resulting product is determined as the local combustion state characteristic. The preliminary fuel-air ratio is defined as a comprehensive air-fuel ratio parameter that considers both total balance and spatial heat distribution demand. It is obtained by performing a ratio mapping process on the local combustion state characteristic. Specifically, the calculated weighted fuel flow rate value is divided by the weighted air flow rate value, and the quotient is determined as the preliminary fuel-air ratio.

[0023] It is worth noting that the local combustion state characteristics represent the equivalent combustion intensity of each temperature control zone after considering the space heat demand. This characteristic is not only used to calculate the preliminary fuel-air ratio in this step, but also provides a regional analytical basis for the extraction of abnormal fluctuation characteristics in subsequent steps, enabling the system to locate the specific furnace area where the ratio imbalance occurs.

[0024] It should be noted that the selection of interpolation algorithms should refer to general principles of industrial data processing, such as linear interpolation or spline interpolation methods. For example, assuming that at a certain moment the gas flow rate is 500 cubic meters per hour and the air flow rate is 6,000 cubic meters per hour, if the furnace temperature distribution data identifies that the temperature deviation of the homogenizing zone is the highest and that this zone has a high sensitivity, causing its calculated priority coefficient value to be quantized to 0.6, while the preheating zone weight is quantized to 0.1, then the heat distribution correction weight generated by the system will significantly increase the flow calculation weight of the homogenizing zone. In this case, the weighted calculation of the standard flow matrix using the heat distribution correction weight is not a simple calculation of the global ratio of 1:12, but rather an adjustment of the gain for the specific area's demand using the correction weight. The resulting preliminary fuel-air ratio can guide the control system to make more precise spatial adjustments, effectively suppressing uneven temperature distribution within the furnace. It is worth noting that the sensitivity setting should be based on the thermodynamic characteristics specification of the heating furnace, and the mapping relationship of different combustion sections' contributions to the overall furnace thermal field should be determined through calibration tests. Those skilled in the art will understand that the calculation logic of the priority coefficient value is intended to prioritize compensation for the region with the largest thermal deviation and the most sensitive adjustment response, in order to achieve rapid thermal balance.

[0025] In step S13, if the initial fuel-air ratio exceeds a preset flow matching threshold, the environmental impact data is filtered to obtain a clean environmental noise value. Then, the clean environmental noise value is correlated with the furnace temperature distribution data to obtain a ratio imbalance offset, including: Based on the environmental impact data, a time series smoothing algorithm is used to remove high-frequency signals to obtain a clean environmental noise value; A correlation mapping matrix is ​​obtained by performing a multi-dimensional correlation analysis based on the pure environmental noise value and the furnace temperature distribution data. The deviation of the initial fuel-air ratio is calculated by using the correlation mapping matrix to obtain the ratio imbalance offset.

[0026] The system first makes a preliminary judgment on the stability of the combustion state. The preset flow matching threshold is defined as the maximum percentage limit that allows the fuel-air ratio to deviate from the theoretical optimal value. It is set based on the combustion efficiency simulation test under the rated operating conditions of the heating furnace. By collecting stable operating data from 30 consecutive production shifts, the 95th percentile of the air-fuel ratio fluctuation is calculated, that is, the preliminary fuel-air ratio value collected at the current moment is obtained. The difference between the preliminary fuel-air ratio value and the preset theoretical air-fuel ratio target value is calculated to obtain the absolute value of the ratio deviation. Then, the absolute value of the ratio deviation is divided by the preset theoretical air-fuel ratio target value, and the quotient is the air-fuel ratio fluctuation value.

[0027] The system compares the calculated air-fuel ratio fluctuation value with the 95th percentile. If the air-fuel ratio fluctuation value exceeds this percentile, it is determined to exceed the preset flow matching threshold, for example, set to ±5%. If the absolute value of the deviation between the preliminary fuel-air ratio generated in step S12 and the theoretical value exceeds this threshold, subsequent correction logic is triggered. Environmental impact data is defined as external interference signals acting on the flow sensor, originating from the mechanical and electromagnetic environment where the sensor is located. The system uses a time series smoothing algorithm to remove high-frequency signals based on the environmental impact data. This involves adding the original noise values ​​of the current sampling time and the nine consecutive sampling times preceding it to obtain the total noise, then dividing the total noise by ten, and the resulting arithmetic mean is determined as the pure environmental noise value. It should be noted that the preset flow matching threshold is not a fixed value and needs to be calibrated according to the rated heat load adjustment ratio of the boiler burner. The specific setting method is to collect the flow data of the boiler under rated operating conditions for 24 consecutive hours, calculate the standard deviation of the air-fuel ratio from the design value, and take three times the standard deviation as the flow matching threshold. If a different model of burner is used, the threshold value must be recalibrated according to its physical specifications to ensure a balance between control sensitivity and stability.

[0028] Subsequently, the system performs multi-dimensional correlation analysis based on the pure environmental noise values ​​and the furnace temperature distribution data to obtain a correlation mapping matrix. The correlation mapping matrix is ​​defined as a mathematical topological structure describing the contribution of environmental noise of different magnitudes to furnace temperature fluctuations in different temperature control zones. Its construction method involves using the pure environmental noise values ​​as row vectors and the temperature deviations of each temperature control zone in the furnace temperature distribution data as column vectors. The cross-correlation coefficients of the row and column elements are calculated to obtain the element values ​​characterizing the degree of interference, thus completing the matrix construction. The proportional imbalance offset is defined as the quantitative value of the actual deviation of the fuel-air ratio from the target value after removing interference factors. Specifically, the system performs deviation quantification calculation on the preliminary fuel-air ratio using the correlation mapping matrix. Specifically, the preliminary fuel-air ratio is used as an input vector, and multiplied with the correlation mapping matrix, i.e., the product of the input vector element and the corresponding element of the matrix is ​​calculated and summed. The resulting value is determined as the proportional imbalance offset.

[0029] It is worth noting that the cross-correlation coefficient uses the pure environmental noise numerical sequence as the independent variable and the temperature deviation sequence of each temperature control zone as the dependent variable. The Pearson correlation coefficient between the two is calculated over sixty consecutive sampling points. This coefficient reflects the degree of linear influence of environmental noise on the temperature fluctuation of each temperature control zone. The calculated coefficients are arranged in order of temperature control zone to form an association mapping matrix. This matrix is ​​recalculated every four hours to adapt to the slow time-varying characteristics of the heating furnace's operating status.

[0030] It should be noted that the window length of the time series smoothing algorithm needs to be set according to the typical frequency of electromagnetic interference at the site. Those skilled in the art can refer to the suggestions on anti-interference processing in the Industrial Automation Instrumentation Engineering Construction and Acceptance Specifications and optimize and adjust it within a range of five to twenty sampling periods. For example, if the current ambient noise value shows that the homogenization zone is subject to significant mechanical vibration interference, and analysis of the correlation mapping matrix reveals that this interference has caused the air flow meter reading to be falsely inflated by 3%, then when performing deviation quantification calculation, the system will subtract this false increment from the preliminary fuel-air ratio, and the difference obtained is the true ratio imbalance offset. It is worth noting that the setting of the correlation mapping matrix needs to be combined with the on-site calibration test of the heating furnace, by artificially applying a controlled interference source and recording the temperature feedback of each area, thereby determining the correlation coefficient in the matrix. In one implementation, if the initial fuel-air ratio does not exceed the preset flow matching threshold, the system determines that the current combustion is under control, does not trigger the filtering and offset calculation steps, directly skips steps S13 and S14, sets the ratio imbalance offset to zero, and maintains the current air-fuel ratio output in subsequent steps.

[0031] In step S14, abnormal fluctuation features are extracted based on the proportional imbalance offset to obtain an abnormal fluctuation feature vector, and time scale alignment is performed based on the historical records associated with the abnormal fluctuation feature vector to obtain an aligned fluctuation feature sequence containing fluctuation amplitude, including: Extract the pre-stored historical feature matrix; The corresponding temperature data sequence is retrieved based on the proportional imbalance offset, and the abnormal fluctuation feature vector is obtained by calculating the rate of change based on the temperature data sequence. The aligned fluctuation feature sequence is obtained by resampling the abnormal fluctuation feature vector and the historical record feature matrix.

[0032] The system quantifies combustion anomalies through deep correlation analysis of spatiotemporal data. The historical record feature matrix is ​​defined as a pre-stored dataset in the system's storage unit, encompassing the mapping relationship between proportional offsets and temperature fluctuations under multiple sets of typical historical abnormal operating conditions. It originates from extracting abnormal events from at least thirty operating cycles of the heating furnace and includes labeled stable operating condition features and known fault mode features. The system first extracts the historical record feature matrix from the storage unit as a comparison benchmark. For example, the historical record feature matrix is ​​stored in the system's relational database, and its structure includes a timestamp index, proportional deviation dimension, temperature rise rate dimension, and operating condition label dimension. Its construction involves collecting full sensor data from the heating furnace over the past sixty working days and manually identifying typical abnormal samples such as "gas pressure fluctuations" and "regulator valve obstruction." The system performs an incremental update monthly, storing newly occurring and verified abnormal sequences in the matrix after normalization.

[0033] Next, based on the proportional imbalance offset generated in step S13, the system automatically retrieves the corresponding temperature data sequence. This temperature data sequence is defined as the furnace temperature distribution data within five sampling periods before and after the proportional imbalance offset was generated. The abnormal fluctuation feature vector is defined as a multi-dimensional numerical vector describing the dynamic characteristics of the current combustion state imbalance. Its acquisition process involves the system calculating the difference in temperature values ​​between two adjacent sampling points for the retrieved temperature data sequence, dividing this difference by the sampling interval time, and obtaining the quotient as the temperature change rate. Subsequently, the temperature change rates of each temperature control zone are arranged in regional order, and the resulting vector is the abnormal fluctuation feature vector. The aligned fluctuation feature sequence, including fluctuation amplitude, is defined as a feature sequence that is completely aligned with historical abnormal patterns on the time axis and retains the current fluctuation intensity. In practice, since the frequency of the currently collected data may not be consistent with the storage frequency in the historical feature matrix, the system performs resampling processing on the historical feature matrix based on the timestamp of the abnormal fluctuation feature vector. That is, a linear resampling method is used to interpolate or extract the currently collected features or historical features according to a uniform least common divisor frequency, so that the sequence length and time interval of the two are completely matched. The final sequence is the aligned fluctuation feature sequence, and each element in the sequence contains the actual fluctuation amplitude value corresponding to the alignment time.

[0034] It should be noted that the range of temperature data sequences to be retrieved needs to be set according to the thermal inertia of the heating furnace. Typically, data from ten to thirty seconds before and after the triggering moment of the proportional imbalance offset is taken to ensure coverage of the complete dynamic response process. For example, if the proportional imbalance offset generated in the heat spreader at a certain moment is +0.3, the system automatically extracts the temperature data before and after that moment. Calculations show that the temperature change rate is five degrees Celsius per second, and the generated abnormal fluctuation feature vector will capture this rapid temperature rise trend. By resampling the historical record feature matrix, time offsets caused by sensor sampling jitter or communication delays can be eliminated, making subsequent feature matching more accurate. It is worth noting that the historical record feature matrix should be set based on the historical operating condition log of the heating furnace and should be regularly updated to ensure it contains the latest process operating parameters. Those skilled in the art should understand that if a corresponding timestamp or feature pattern cannot be matched from the storage unit under the current operating conditions, the system will use the current abnormal fluctuation feature vector as a benchmark and generate a temporary aligned fluctuation feature sequence through moving average processing for subsequent calculations.

[0035] In step S15, an amplitude threshold is obtained by frequency interval mapping based on the aligned fluctuation feature sequence. If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, an abnormal fluctuation mode reverse mapping is performed to obtain combustion parameter adjustment suggestions, including: The real-time fluctuation frequency is obtained by performing a fast Fourier transform on the aligned fluctuation feature sequence. The amplitude threshold is obtained by interval matching based on the real-time fluctuation frequency; If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, the fluctuation amplitude is input to a pre-built combustion experience knowledge base for reverse mapping to obtain combustion parameter adjustment suggestions.

[0036] During step S15, the system achieves accurate decision-making through frequency domain analysis and expert knowledge fusion. The real-time fluctuation frequency is defined as a characteristic parameter describing the oscillation period of the aligned fluctuation feature sequence per unit time, derived from the frequency domain transformation of the time-domain sequence. Specifically, the system performs a Fast Fourier Transform on the aligned fluctuation feature sequence, decomposing the time-domain signal into multiple sinusoidal components, extracting the dominant frequency component with the most concentrated energy distribution, and determining this dominant frequency as the real-time fluctuation frequency. The amplitude threshold is defined as the allowable deviation of the combustion state from its maximum physical extreme value at a specific fluctuation frequency.

[0037] Specifically, the system pre-constructs a frequency range table, which divides frequencies into a low-frequency steady-state region, a mid-frequency fluctuation region, and a high-frequency noise region. The frequency range division is based on the dynamic response characteristics of the furnace combustion system and the frequency distribution of signal interference. The 0-0.5Hz range is defined as the low-frequency steady-state region, reflecting long-period drift in combustion conditions; the 0.5-5Hz range is defined as the mid-frequency fluctuation region, reflecting system oscillations caused by fluctuations in induced draft fan speed or valve regulation lag; and the range above 5Hz is defined as the high-frequency noise region, primarily originating from sensor electronic noise or turbulent fluid flow in the pipeline. The preset amplitude limits for each frequency range are set according to general values ​​regarding the degree of combustion stability disruption at different frequencies. For example, the amplitude limit for the low-frequency steady-state region is set at 10%, the mid-frequency fluctuation region at 5%, and the high-frequency noise region at 8%. The system performs interval matching by comparing the calculated real-time fluctuation frequency with the frequency range table. If the real-time fluctuation frequency falls into a specific interval, the amplitude limit corresponding to that interval is determined as the amplitude threshold.

[0038] In one implementation, the frequency range is divided based on actual dynamic response test data of the furnace combustion system. By applying pulse excitation to the actuator, the furnace temperature response curve is collected, and the cutoff frequency of the system amplitude-frequency characteristic is determined through spectrum analysis. The upper limit of the low-frequency steady-state region corresponds to the corner frequency of the main inertial link of the system, the upper limit of the mid-frequency fluctuation region corresponds to the octave range of the actuator response bandwidth, and the lower limit of the high-frequency noise region is set according to the sensor sampling frequency and the typical frequency band of electromagnetic interference in the field. For a typical steel rolling furnace, the above frequency range can be calibrated based on the actual measured data in the field, rather than being fixed. Specifically, the fluctuation amplitude distribution under historical normal operating conditions in each frequency range is statistically analyzed, and the 95th percentile of the distribution is taken as the initial limit. Then, it is corrected in combination with the requirements for combustion stability in actual operation, so that the limit value can both filter out random disturbances and respond to real anomalies in a timely manner.

[0039] Subsequently, the system executes threshold exceedance judgment logic. If the fluctuation amplitude of the aligned fluctuation feature sequence does not exceed the amplitude threshold, the system determines that the current combustion fluctuation is within the permissible range of the system's adaptive adjustment, does not trigger deep correction logic, and directly outputs the current combustion parameter adjustment suggestion as a zero value, i.e., maintains the existing control command of the actuator unchanged. Conversely, if the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, abnormal fluctuation mode reverse mapping processing is performed. The fluctuation amplitude refers to the relative fluctuation intensity obtained by dividing the difference between the maximum and minimum values ​​in the aligned fluctuation feature sequence by the arithmetic mean of the sequence, expressed as a percentage. Specifically, the system combines the current real-time fluctuation frequency and fluctuation amplitude into a retrieval vector, performs similarity comparison in a pre-built combustion experience knowledge base, and matches the corresponding abnormal mode label. The abnormal mode label includes air-fuel ratio mismatch mode, load drastic change mode, or fuel pressure fluctuation mode. The combustion experience knowledge base is a nonlinear mapping model constructed based on historical expert operation records, combustion dynamics simulation models, and typical fault repair logic. Subsequently, the system retrieves a preset control compensation strategy based on the abnormal mode label, multiplies the fluctuation amplitude by the weighting coefficient corresponding to the mode, and the resulting product is determined as the parameter deviation that needs to be adjusted. The combustion parameter adjustment suggestion is defined as a set of quantitative intervention instructions for fuel flow rate, air flow rate, or the ratio of the two, which are derived from the parameter deviation calculated by the above-mentioned reverse mapping process, such as "increase the combustion air flow rate by 100 cubic meters per hour".

[0040] It is worth noting that the combustion experience knowledge base can collect historical operating data of the heating furnace over at least thirty operating cycles, including fuel flow rate, air flow rate, furnace temperature distribution, actuator commands and feedback values. Professional thermal engineers can label each abnormal fluctuation event to form an abnormal pattern tag. A nearest neighbor retrieval algorithm based on feature vectors is used to establish a multi-dimensional index with fluctuation frequency, fluctuation amplitude, and temperature change rate as retrieval dimensions. When real-time abnormal fluctuation features are input, the system calculates the Euclidean distance with each sample in the knowledge base, selects the three samples with the smallest distance, and performs a weighted average of their corresponding adjustment suggestions inversely proportional to the distance. The closer the sample is, the higher the weight, and the farther the sample is, the lower the weight. When fewer than three samples are retrieved, all retrieved samples are averaged equally to obtain the final combustion parameter adjustment suggestions. The weight coefficients are determined by the actual effect feedback of the corresponding adjustment measures in the historical records of each sample, and the weight coefficients are corrected quarterly based on newly added samples.

[0041] It should be noted that the specific implementation of the abnormal fluctuation mode reverse mapping described in this embodiment is to use the dominant frequency extracted by fast Fourier transform as the feature input, and retrieve the corresponding operating condition template (such as valve jamming, secondary air pressure fluctuation, etc.) in the experience knowledge base, thereby outputting a targeted flow regulation compensation value. This process aims to simulate the intuitive judgment of senior thermal engineers on combustion flame fluctuations and quantify it into specific command parameters.

[0042] It should be noted that the number of processing points for the Fast Fourier Transform (FFT) needs to be set according to the signal sampling frequency to ensure that the spectral resolution can effectively distinguish the aforementioned frequency ranges. For example, if the system detects a real-time fluctuation frequency of 1.2 Hz, falling into the mid-frequency fluctuation range, and the retrieved amplitude threshold is 5%, while the current fluctuation amplitude is 7%, then inverse mapping is triggered. The system identifies this frequency as belonging to a typical "air following delay caused by drastic load changes" mode and calculates the specific airflow compensation value based on the weighting coefficient of this mode. It is worth noting that the frequency range table and amplitude limits should be set based on the dynamic modal analysis report and historical operating condition log of the heating furnace. Those skilled in the art can periodically fine-tune the thresholds according to the aging of the equipment. Those skilled in the art should understand that if the fluctuation amplitude does not exceed the amplitude threshold, the system determines that the current fluctuation is a random disturbance within the allowable range, does not perform inverse mapping, and directly outputs a combustion parameter adjustment suggestion of zero.

[0043] In step S16, a time delay correlation deviation dataset is obtained by aligning the combustion parameter adjustment suggestions with the response parameters of the execution components over a time axis. Gain mapping is then performed based on this dataset to obtain an optimized combustion parameter adjustment scheme, including: The physical lag time is extracted based on the response parameters of the execution component, and the combustion parameter adjustment suggestions are time-axis shifted using the physical lag time to obtain a time delay correlation deviation dataset. The real-time coefficient for deviation compensation is calculated by performing amplification of control parameters based on the aforementioned time delay-related deviation dataset. The combustion parameter adjustment suggestions are corrected based on the deviation compensation real-time coefficient to obtain an optimized combustion parameter adjustment scheme.

[0044] During step S16, the system ensures precise time matching between control actions and furnace physical requirements through time delay compensation logic. Physical lag time is defined as the total time required for the actuator to complete the expected physical action from receiving the command, derived from the analysis of the actuator response parameters obtained in step S11. Specifically, the system extracts the actuator action delay value, medium transmission delay value, and sensor detection delay value from the actuator response parameters, and sums these three values ​​over time; the total time obtained is determined as the physical lag time. The time delay-related deviation dataset is defined as a reference dataset describing the expected deviation at a specific future moment after considering system delay. In specific operation, the system uses the physical lag time to perform time-axis shifting on the combustion parameter adjustment suggestions generated in step S15. That is, based on the physical lag time value, the execution time corresponding to the adjustment suggestion is advanced by the physical lag time, aligning the execution time of the adjustment suggestion with the physical moment when the sensor detects the deviation on the time axis, thereby generating the time delay-related deviation dataset. The deviation compensation real-time coefficient is defined as a gain factor used to dynamically adjust the control strength, which is derived from the proportional quantification of the degree of deviation under the influence of time delay.

[0045] Specifically, the system calculates the amplification ratio of control parameters based on the time-delay-related deviation dataset. Specifically, it divides the expected deviation value in the time-delay-related deviation dataset by a preset rated control benchmark value to obtain the deviation ratio. Then, it sums the deviation ratio with a given value to determine the real-time deviation compensation coefficient. If the expected deviation value is negative, it indicates that the current actual combustion state exceeds expectations, requiring a reduction in adjustment intensity. In this case, the calculated real-time deviation compensation coefficient is less than one. The system also multiplies this coefficient with each component of the adjustment suggestion to achieve reverse correction of the intensity. Its processing logic is symmetrical to that of the positive deviation case. The expected deviation value refers to the core adjustment quantity that needs compensation extracted from the time-delay-related deviation dataset, i.e., the adjustment value of the corresponding fuel flow rate or air flow rate in the combustion parameter adjustment suggestion generated in step S15. The preset rated control benchmark value is defined as the total fuel consumption required for the heating furnace to maintain rated power under standard full-load conditions. The setting is based on the rated hourly gas consumption on the nameplate or thermal performance design document of the heating furnace. A common setting method is to calculate it by dividing the total rated heat load of the furnace burner by the unit calorific value of the fuel, for example, setting it to 10,000 cubic meters per hour. The optimized combustion parameter adjustment scheme is defined as the final set of execution instructions after intensity correction. The system corrects the combustion parameter adjustment suggestion based on the deviation compensation real-time coefficient. The combustion parameter adjustment suggestion is not a single value, but a multi-dimensional instruction vector containing fuel flow adjustment components and air flow adjustment components. Specifically, the correction process involves multiplying the deviation compensation real-time coefficient by the fuel flow adjustment component value and the air flow adjustment component value in the combustion parameter adjustment suggestion, respectively, to obtain the corrected fuel flow adjustment value and the corrected air flow adjustment value, thus constituting the optimized combustion parameter adjustment scheme.

[0046] It should be noted that the setting of the physical lag time should be based on the communication protocol cycle of the heating furnace control system and the technical specifications of the actuator. Its typical range is usually between 0.5 and 5 seconds. Those skilled in the art should calibrate it according to the actual length of the physical pipeline and the valve response speed. For example, if the extracted physical lag time is two seconds, and the current combustion parameter adjustment suggestion includes increasing the gas flow rate by 100 cubic meters per hour and increasing the air flow rate by 1200 cubic meters per hour, and if the calculated deviation compensation real-time coefficient is 1.1, then the system will multiply 1.1 by 100 and 1200 respectively to obtain an optimized scheme that increases the gas flow rate by 110 cubic meters per hour and the air flow rate by 1320 cubic meters per hour. It is worth noting that the selection of the rated control reference value should be consistent with the flow meter unit. Those skilled in the art should understand that correcting the adjustment suggestion through the deviation compensation real-time coefficient aims to overcome the inertial effect of physical lag by increasing the adjustment gain when the system senses a large time axis misalignment, thereby achieving a faster recovery of the combustion state.

[0047] In step S17, variable time axis offset processing is performed according to the optimized combustion parameter adjustment scheme to obtain the sequence execution time axis, and time deviation calibration is performed according to the sequence execution time axis to obtain the final combustion control sequence, including: The adjustment action nodes are extracted from the optimized combustion parameter adjustment scheme, and the instruction issuance time advance of the adjustment action nodes is processed to obtain the sequence execution time axis. If the sequence execution timeline meets the preset timeline alignment threshold, then the action details are bound according to the sequence execution timeline to obtain the final combustion control sequence.

[0048] During step S17, the system counteracts the response inertia of the physical system through a time-feedforward strategy for the instructions. An adjustment action node is defined as the instruction trigger point in the optimized combustion parameter adjustment scheme, carrying the specific execution time and target opening amplitude. It originates from the structured analysis of the control instruction flow in the optimized combustion parameter adjustment scheme. The system identifies the adjustment target values ​​of each fuel valve and combustion fan in the scheme and locks the original trigger timestamp corresponding to each adjustment action. The sequence execution time axis is defined as the sequence of control instruction issuance times after advance correction. In specific operation, the system performs instruction issuance time advance processing on the adjustment action node, that is, retrieves the physical lag duration value determined in step S16, subtracts the physical lag duration value from the original trigger timestamp of the adjustment action node, and the difference is determined as the advance trigger time point of that action node. All advance trigger time points are arranged in chronological order to form the sequence execution time axis. The preset time axis alignment threshold is defined as the maximum error limit that allows the instruction issuance time to deviate from the minimum step size of the system control cycle. The setting is based on the operation scan cycle of the programmable logic controller of the heating furnace. The general setting method is to multiply the operation scan cycle by two as the threshold benchmark, and then fine-tune it according to the measured value of communication jitter of the control system so that the threshold is between 20 milliseconds and 100 milliseconds. For example, it can be set to 50 milliseconds. The system calculates the absolute value of the deviation by comparing the difference between each time point in the sequence execution time axis and the system real-time clock, and then executes the calibration judgment logic.

[0049] If the absolute value of the deviation is less than or equal to the preset time axis alignment threshold, the current command issuance time is determined to be within the high-precision synchronization range. The system then performs action detail binding based on the sequence execution time axis to obtain the final combustion control sequence. If the absolute value of the deviation is greater than the preset time axis alignment threshold, the system, while performing the command discard operation, sends a "maintain previous state value" locking signal to the actuator to prevent adjustment loss due to excessive time delay. Subsequently, the system temporarily doubles the sampling frequency for rapid resampling to shorten the generation cycle of the next control sequence. Action detail binding is defined as the process of converting logical adjustment commands into low-level execution code with an industrial protocol format. Specifically, based on the sequence execution time axis that meets the alignment requirements, the system encapsulates the corresponding fuel flow adjustment value and air flow adjustment value into the corresponding communication address frame, and associates the specific pulse width modulation signal or analog output level signal with the trigger point on the time axis one-to-one, thereby obtaining the final combustion control sequence. Finally, the system sends the final combustion control sequence to the actuator through the communication interface to achieve closed-loop stable adjustment of the combustion state.

[0050] It's important to note that the essence of command issuance time advance processing is to compensate for system feedback delay using feedforward control principles. The advance time must be strictly equal to the extracted physical lag duration. For example, if the physical lag duration is three seconds, and the optimized scheme requires increasing flow at the tenth second, the advance processing will correct the trigger point to the seventh second. If, due to system communication congestion, the calculated absolute value of the deviation reaches one hundred milliseconds when the actual time reaches the seventh second, exceeding the fifty-millisecond threshold, the system will abandon the control sequence. It's worth noting that the preset time axis alignment threshold is set to prevent severe command misalignment due to network jitter, ensuring the synchronization of combustion adjustment. When the absolute value of the deviation exceeds the threshold, discarding the command ensures system safety, preventing drastic flow fluctuations at incorrect times, and achieving a more stable adjustment effect through rapid iteration in the next cycle.

[0051] In summary, this invention achieves real-time and precise control and closed-loop optimization of combustion parameters under complex dynamic disturbances by combining spatiotemporal alignment and correction of multi-source sensor data with environmental noise quantification analysis and dynamic compensation mechanism for response delay of execution components, effectively maintaining the stability of the combustion environment inside the furnace.

[0052] Reference Figure 2 The second embodiment of the present invention provides an intelligent combustion control system for a steel rolling heating furnace, comprising: The data acquisition module is used to acquire fuel air flow data, furnace temperature distribution data, environmental impact data, and response parameters of the actuators. The preliminary calculation module is used to align the fuel-air flow rate data with the time series to obtain a flow rate synchronization sequence, and to perform spatial weighted calculation based on the flow rate synchronization sequence and the furnace temperature distribution data to obtain a preliminary fuel-air ratio. The offset analysis module is used to filter the environmental impact data to obtain a clean environmental noise value if the initial fuel-air ratio exceeds a preset flow matching threshold, and then perform correlation mapping based on the clean environmental noise value and the furnace temperature distribution data to obtain the ratio imbalance offset. The feature alignment module is used to extract abnormal fluctuation features based on the proportional imbalance offset to obtain an abnormal fluctuation feature vector, and to perform time scale alignment based on the historical records associated with the abnormal fluctuation feature vector to obtain an aligned fluctuation feature sequence containing the fluctuation amplitude. The suggestion generation module is used to obtain an amplitude threshold by mapping the frequency range according to the aligned fluctuation feature sequence. If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, the abnormal fluctuation mode is reverse mapped to obtain a combustion parameter adjustment suggestion. The scheme optimization module is used to perform time axis alignment and correlation based on the combustion parameter adjustment suggestions and the response parameters of the execution components to obtain a time delay correlation deviation dataset, and to perform gain mapping based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme. The sequence construction module is used to perform variable time axis offset processing according to the optimized combustion parameter adjustment scheme to obtain the sequence execution time axis, and to perform time deviation calibration according to the sequence execution time axis to obtain the final combustion control sequence.

[0053] It should be noted that the intelligent combustion control system for a steel rolling heating furnace provided in this embodiment of the invention is used to execute all the process steps of the intelligent combustion control method for a steel rolling heating furnace described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0054] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0055] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for intelligent combustion control of a steel rolling heating furnace, characterized in that, include: Acquire fuel air flow data, furnace temperature distribution data, environmental impact data, and response parameters of actuators; The flow rate synchronization sequence is obtained by time series alignment of the fuel-air flow rate data, and spatial weighting calculation is performed based on the flow rate synchronization sequence and the furnace temperature distribution data to obtain the preliminary fuel-air ratio. If the initial fuel-air ratio exceeds the preset flow matching threshold, the environmental impact data is filtered to obtain a clean environmental noise value, and the clean environmental noise value is combined with the furnace temperature distribution data to perform correlation mapping to obtain the ratio imbalance offset. Abnormal fluctuation feature vectors are obtained by extracting abnormal fluctuation features based on the proportional imbalance offset, and time scale alignment is performed based on the historical records associated with the abnormal fluctuation feature vectors to obtain an aligned fluctuation feature sequence containing fluctuation amplitude. The amplitude threshold is obtained by frequency interval mapping based on the alignment fluctuation feature sequence. If the fluctuation amplitude of the alignment fluctuation feature sequence exceeds the amplitude threshold, the abnormal fluctuation mode is reverse mapped to obtain combustion parameter adjustment suggestions. Based on the combustion parameter adjustment suggestions and the response parameters of the execution components, a time axis alignment correlation dataset is obtained, and a gain mapping is performed based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme. The optimized combustion parameter adjustment scheme is used to perform variable time axis offset processing to obtain the sequence execution time axis, and the time deviation is calibrated according to the sequence execution time axis to obtain the final combustion control sequence.

2. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The acquisition of fuel air flow data, furnace temperature distribution data, environmental impact data, and actuator response parameters includes: The instantaneous value of fuel flow and the dynamic change of air flow are obtained from the vortex flow meter, and the data are correlated and stored to obtain fuel-air flow data; Local furnace temperature values ​​are obtained from thermocouples, and the furnace temperature distribution data is obtained by calculating the regional average value based on the local furnace temperature values. Environmental interference signals are acquired from environmental sensing devices to obtain environmental impact data; The signal transmission period between the vortex flowmeter and the thermocouple is obtained, and the response parameters of the actuator are calculated by time accumulation based on the signal transmission period.

3. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The process of aligning the fuel-air flow rate data over time to obtain a flow rate synchronization sequence, and then performing spatial weighting calculations based on the flow rate synchronization sequence and the furnace temperature distribution data to obtain a preliminary fuel-air ratio, includes: A flow synchronization sequence is obtained by performing data synchronization processing based on the fuel air flow data; Based on the furnace temperature distribution data, the spatial mapping relationship of the heat distribution pattern inside the furnace is extracted, and the contribution quantification is performed based on the spatial mapping relationship to obtain the heat distribution correction weight; The local combustion state characteristics are obtained by weighting the flow synchronization sequence using the heat distribution correction weight, and a preliminary fuel-air ratio is obtained by mapping the proportions based on the local combustion state characteristics.

4. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, If the initial fuel-air ratio exceeds a preset flow matching threshold, the environmental impact data is filtered to obtain a clean environmental noise value. Then, the clean environmental noise value is correlated with the furnace temperature distribution data to obtain a ratio imbalance offset, including: Based on the environmental impact data, a time series smoothing algorithm is used to remove high-frequency signals to obtain a clean environmental noise value; A correlation mapping matrix is ​​obtained by performing multi-dimensional correlation analysis based on the pure environmental noise value and the furnace temperature distribution data. The deviation of the initial fuel-air ratio is calculated by using the correlation mapping matrix to obtain the ratio imbalance offset.

5. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The process of extracting abnormal fluctuation features based on the proportional imbalance offset to obtain an abnormal fluctuation feature vector, and aligning the abnormal fluctuation feature vector with historical records associated with it over time to obtain an aligned fluctuation feature sequence containing fluctuation amplitude, includes: Extract the pre-stored historical feature matrix; The corresponding temperature data sequence is retrieved based on the proportional imbalance offset, and the abnormal fluctuation feature vector is obtained by calculating the rate of change based on the temperature data sequence. The aligned fluctuation feature sequence is obtained by resampling the abnormal fluctuation feature vector and the historical record feature matrix.

6. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The step involves mapping the frequency range based on the aligned fluctuation feature sequence to obtain an amplitude threshold. If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, an abnormal fluctuation pattern reverse mapping is performed to obtain combustion parameter adjustment suggestions, including: The real-time fluctuation frequency is obtained by performing a fast Fourier transform on the aligned fluctuation feature sequence. The amplitude threshold is obtained by interval matching based on the real-time fluctuation frequency; If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, the fluctuation amplitude is input to a pre-built combustion experience knowledge base for reverse mapping to obtain combustion parameter adjustment suggestions.

7. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The process of aligning and correlating the combustion parameter adjustment suggestions with the response parameters of the actuators to obtain a time delay correlation deviation dataset, and performing gain mapping based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme, includes: The physical lag time is extracted based on the response parameters of the execution component, and the combustion parameter adjustment suggestions are time-axis shifted using the physical lag time to obtain a time delay correlation deviation dataset. The real-time coefficient for deviation compensation is calculated by performing amplification of control parameters based on the aforementioned time delay-related deviation dataset. The combustion parameter adjustment suggestions are corrected based on the deviation compensation real-time coefficient to obtain an optimized combustion parameter adjustment scheme.

8. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The process of obtaining the sequence execution time axis by performing variable time axis offset processing based on the optimized combustion parameter adjustment scheme, and obtaining the final combustion control sequence by performing time deviation calibration based on the sequence execution time axis, includes: The adjustment action nodes are extracted from the optimized combustion parameter adjustment scheme, and the instruction issuance time advance of the adjustment action nodes is processed to obtain the sequence execution time axis. If the sequence execution timeline meets the preset timeline alignment threshold, then the action details are bound according to the sequence execution timeline to obtain the final combustion control sequence.

9. The intelligent combustion control method for a steel rolling heating furnace according to claim 1, characterized in that, The response parameters of the actuator include the actuator action delay, the medium transmission delay, and the sensor detection delay.

10. An intelligent combustion control system for a steel rolling heating furnace, characterized in that, include: The data acquisition module is used to acquire fuel air flow data, furnace temperature distribution data, environmental impact data, and response parameters of the actuators. The preliminary calculation module is used to align the fuel-air flow rate data with the time series to obtain the flow rate synchronization sequence, and to perform spatial weighted calculation based on the flow rate synchronization sequence and the furnace temperature distribution data to obtain the preliminary fuel-air ratio. The offset analysis module is used to filter the environmental impact data to obtain a clean environmental noise value if the initial fuel-air ratio exceeds a preset flow matching threshold, and then perform correlation mapping based on the clean environmental noise value and the furnace temperature distribution data to obtain the ratio imbalance offset. The feature alignment module is used to extract abnormal fluctuation features based on the proportional imbalance offset to obtain an abnormal fluctuation feature vector, and to perform time scale alignment based on the historical records associated with the abnormal fluctuation feature vector to obtain an aligned fluctuation feature sequence containing the fluctuation amplitude. The suggestion generation module is used to obtain an amplitude threshold by performing frequency range mapping based on the aligned fluctuation feature sequence. If the fluctuation amplitude of the aligned fluctuation feature sequence exceeds the amplitude threshold, the abnormal fluctuation mode is reverse mapped to obtain a combustion parameter adjustment suggestion. The scheme optimization module is used to perform time axis alignment and correlation based on the combustion parameter adjustment suggestions and the response parameters of the execution components to obtain a time delay correlation deviation dataset, and to perform gain mapping based on the time delay correlation deviation dataset to obtain an optimized combustion parameter adjustment scheme. The sequence construction module is used to perform variable time axis offset processing according to the optimized combustion parameter adjustment scheme to obtain the sequence execution time axis, and to perform time deviation calibration according to the sequence execution time axis to obtain the final combustion control sequence.

Images