Pressure balance control method for cooperative operation of quenching boiler and cracking furnace

Abstract

The invention relates to the technical field of chemical production control, and discloses a pressure balance control method for cooperative operation of a quenching boiler and a cracking furnace, and the method comprises the steps: collecting data in real time through pressure sensors disposed at an inlet and an outlet of the quenching boiler and a temperature sensor disposed at a reaction section of the cracking furnace, constructing a pressure state matrix, and pressure gradient change and flow path information are combined to predict pressure distribution characteristics. And analyzing pressure intensity distribution, flow delay characteristics and a change trend by using a cooperative control model, dynamically dividing subunits, calculating a pressure intensity gradient, and generating spatial and temporal distribution characteristics of pressure balance. Based on the characteristics, the system automatically adjusts the opening degree and flow of the valve and coolant distribution, and resource optimization configuration of the quenching boiler and the cracking furnace is achieved. The pressure control precision and the resource utilization rate are improved, stable operation of equipment is guaranteed, and the chemical production benefits are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical production control, and particularly to a pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace. Background Art

[0002] In the process of chemical production, the coordinated operation of a quench boiler and a cracking furnace is a key link, and its pressure balance control directly affects the safety, stability, and efficiency of production. With the continuous development of the chemical industry, the requirements for the automation and intelligence of the production process are getting higher and higher, and some deficiencies of traditional pressure balance control methods are gradually emerging.

[0003] Traditional control methods often control based on the data of a single pressure or temperature sensor, and it is difficult to comprehensively and accurately reflect the pressure state during the coordinated operation of the quench boiler and the cracking furnace. Due to the complex spatial relationship and mutual influence between multiple pressure sensors deployed at the inlet and outlet of the quench boiler and multiple temperature sensors in the reaction section of the cracking furnace, and the adjacent two pressure sensors are separated by a set distance, the acquisition and processing of single data cannot fully consider the distribution characteristics of pressure in space and time.

[0004] In actual operation, the pressure balance is affected by various factors, such as flow rate changes, pipeline resistance, equipment operating status, etc. Traditional methods are difficult to comprehensively analyze and predict these factors, resulting in insufficient timeliness and accuracy of pressure balance control. For example, when there are pressure fluctuations or abnormalities in the system, traditional methods may not be able to quickly determine the root cause and scope of influence of the problem, and thus cannot take effective control measures in a timely manner, which may lead to a decrease in production efficiency and even cause safety accidents.

[0005] In addition, with the continuous expansion of the scale of chemical production, the structures and operating parameters of the quench boiler and the cracking furnace are becoming more and more complex, and the requirements for pressure balance control are also getting higher and higher. Traditional control methods lack the ability to dynamically model and optimize the system, and it is difficult to meet the pressure balance control requirements under different working conditions. For example, under different production loads, the pressure distribution and change trend of the system may change greatly, and traditional methods cannot adjust the control strategy in a timely manner according to these changes, thus affecting the stability and efficiency of production.

[0006] In addition, the existing pressure balance control methods also have deficiencies in resource regulation. Traditional methods often adopt fixed regulation strategies and cannot dynamically allocate resources according to the real-time distribution characteristics of pressure balance, resulting in low resource utilization efficiency. For example, when there is a deviation in pressure balance, traditional methods may not be able to accurately allocate resources to the physical units that need to be regulated, thus affecting the recovery speed and effect of pressure balance. Summary of the Invention

[0007] The object of the present invention is to provide a pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: A pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace, the method comprising: Collect pressure data and temperature data during the coordinated operation of the quench boiler and the cracking furnace through the pressure sensor and the temperature sensor to generate a pressure state data set; Receive real-time data from multiple pressure sensors and multiple temperature sensors through the coordinated control server to construct a pressure state matrix; According to the pressure state data set and the real-time data of each sensor, determine the distribution characteristics of pressure balance. Among them, determining the distribution characteristics of pressure balance includes: processing the pressure state matrix, extracting pressure characteristics in combination with the pressure state data set, and predicting the pressure distribution based on the pressure gradient change and the flow path information. Output the spatio-temporal distribution characteristics of pressure balance through the coordinated control model, and update the pressure state data set according to the spatio-temporal distribution characteristics; According to the distribution characteristics, dynamically regulate the resources of the quench boiler and the cracking furnace.

[0009] Preferably, the determining the distribution characteristics of pressure balance includes: Process the pressure state matrix to extract the pressure intensity distribution, flow delay characteristics, and pressure change trend; Perform pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution and the flow delay characteristics, divide the quench boiler and the cracking furnace into multiple sub-units and mark the unit identification. According to the pressure intensity of the sub-units and the pressure state data set, perform association matching, and mark the unit identification in the pressure state data set; Calculate the pressure intensity gradient according to the position of the pressure sensor, predict the distribution of pressure balance according to the pressure intensity gradient and the pressure change trend, and calculate the pressure prediction information of each sub-unit; Construct a coordinated control model, use the pressure prediction information as the input parameter of the coordinated control model, perform spatial association modeling on the pressure prediction information through the coordinated control model, and output the spatio-temporal distribution characteristics of pressure balance; Update the pressure state data set according to the spatio-temporal distribution characteristics to obtain the distribution characteristics of pressure balance.

[0010] Preferably, the processing of the pressure state matrix includes: Normalize the pressure state matrix, intercept the pressure hot spot area in the matrix through a sliding window, filter the noise of the hot spot area, and calculate the pressure change trend through the differential decomposition algorithm; Calculate the spatial correlation characteristics of the pressure state matrix, calculate the interference intensity, pipeline stability coefficient and pressure blind area index between units according to the spatial correlation characteristics, construct a feature fusion network, and calculate the flow delay characteristics through the feature fusion network; Extract the time-domain features and frequency-domain features collected by each pressure sensor, calculate the pressure feature vector of the sensor according to the phase difference between the time-domain features and the frequency-domain features, perform feature matching on the pressure sensors at different positions according to the pressure feature vector, and calculate the pressure change trend.

[0011] Preferably, the pressure intensity modeling of the pressure state matrix includes: According to the pressure intensity distribution, extract pressure intensity sampling points in each frame of data, perform correlation mapping according to the sampling points and the flow delay characteristics, generate a pressure intensity map, spatially align the pressure intensity maps collected by multiple sensors, and calculate the pressure intensity distribution of the unit; Set an interference threshold value, locate the interference source according to the pressure intensity values of multiple frames of pressure state matrices, calculate the interference intensity difference. If the interference intensity difference is greater than or equal to the interference threshold value, it indicates that there is a pressure blind area in the unit. Perform flow model constraint compensation on the current unit, perform iterative correction on the pressure intensity distribution of the current unit according to the corresponding flow resistance loss model of the current unit, and calculate the intensity compensation value of the blind area pressure according to the correction result; Perform pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution, and perform time delay annotation on the pressure model of the unit through the flow delay characteristics.

[0012] Preferably, the calculation of the pressure intensity gradient according to the pressure sensor position includes: Extract pressure intensity change points according to multiple groups of pressure coverage data, map the change points to a unified coordinate framework according to the deployment positions of the sensors, and fit the change points through a spatial interpolation algorithm to generate a pressure field model of the unit; Perform equally spaced sampling along the flow path of the pressure field model, calculate the pressure attenuation rate, interference fluctuation index and pressure change slope of the path according to the sampling results, and calculate the pressure change parameters according to the pressure attenuation rate, interference fluctuation index and pressure change slope; According to the deployment parameters and acquisition accuracy of the pressure sensors, project the distribution characteristics of the pressure change in each frame of data onto the pressure field model, partition according to the number of sensors along the flow direction of the pressure field model, analyze the change law of the pressure change in the partition, and calculate the pressure distribution characteristics according to the change law; According to the pressure change parameter and the pressure distribution characteristic, calculate the pressure intensity gradient. The calculation process of the pressure intensity gradient includes: based on the coordinate position range from the first pressure sensor to the last pressure sensor, select spatial coordinate points in the sensor deployment direction, cumulatively calculate the product of the pressure field intensity characteristic weight value and the pressure distribution characteristic weight value within the spatial resolution range, and superimpose the influence value of the sensor acquisition frequency on the pressure intensity change rate.

[0013] Preferably, the calculating the pressure prediction information of each sub-unit includes: Taking the main flow path of the pressure field model as the reference line, using the peak position of the pressure change in each frame of data as the reference point, calculate the pressure offset, and draw the pressure distribution curve according to the coordinates; According to the pressure intensity gradient, correct the growth rate and direction in the pressure change trend; Starting from the nearest pressure distribution point, continue to draw the distribution curve according to the correction results of the growth rate and direction to generate the pressure distribution points in the next time period until the distribution points cover the entire target unit to generate the pressure prediction information.

[0014] Preferably, the constructing the collaborative control model includes: The input layer is used to organize the pressure prediction information into spatial distribution data and perform normalization processing; The feature fusion layer is used to extract the unit correlation features of the pressure by processing the spatial distribution data and construct the dependency relationship between physical units; The resource allocation layer is used to integrate the correlation relationship of pressure balance on spatial units and generate a resource regulation strategy.

[0015] Preferably, the obtaining the distribution characteristics of pressure balance includes: According to the spatio-temporal distribution characteristics of pressure balance output by the collaborative control model, correspond the identifier of the sub-unit with the spatio-temporal distribution characteristics; Reorganize the unit data in the pressure state dataset according to the spatio-temporal characteristics to generate a unit distribution map sorted by the pressure balance intensity; According to the reorganized unit distribution map, output the optimized pressure balance distribution characteristics.

[0016] Preferably, the dynamically regulating the resources of the quench boiler and the cracking furnace includes: Map the unit identifier one by one with the unit of the distribution characteristics of pressure balance; According to the spatio-temporal distribution characteristics of pressure balance, control the execution actions of resources, including valve opening adjustment, flow rate regulation and coolant distribution operations; According to the spatial distribution of pressure balance and the preset regulation strategy, dynamically allocate resources to the corresponding physical units.

[0017] Preferably, the execution actions for controlling resources include: When the pressure balance in the target unit reaches a preset intensity threshold, triggering an instruction to increase the opening of the adjacent valve; Dynamically combining available fluid resources according to a flow regulation strategy to generate a coolant distribution vector; Adjusting the pipeline parameters of the transmission node of the target unit based on the coolant distribution vector.

[0018] Compared with the prior art, the beneficial effects of the present invention are: By deploying multiple pressure sensors at the inlet and outlet of the quench boiler and multiple temperature sensors in the reaction section of the cracking furnace, it is possible to collect pressure data and temperature data during the operation of the system in real time and comprehensively, generate a pressure state data set, and provide an accurate and rich data basis for subsequent pressure balance control.

[0019] The collaborative control server receives real-time data from multiple sensors and constructs a pressure state matrix. By processing this matrix, key information such as pressure intensity distribution, flow delay characteristics, and pressure change trends can be extracted, thereby enabling a deeper understanding of the pressure state of the system. Operations such as standardizing the pressure state matrix, filtering noise, and differential decomposition effectively improve the quality and reliability of the data, ensuring the accuracy of subsequent analysis.

[0020] When determining the distribution characteristics of pressure balance, by performing pressure intensity modeling on the pressure state matrix, dividing the quench boiler and the cracking furnace into multiple sub-units and marking unit identifiers, refined management of the system is achieved. By associating and matching the pressure intensity of the sub-units with the pressure state data set, the pressure conditions of each sub-unit can be grasped more accurately. At the same time, by calculating the pressure intensity gradient and predicting the pressure balance distribution, a scientific basis is provided for the dynamic regulation of resources.

[0021] The constructed collaborative control model includes an input layer, a feature fusion layer, and a resource allocation layer, which can perform spatial correlation modeling on pressure prediction information and output the spatio-temporal distribution characteristics of pressure balance. This enables the system to comprehensively understand the distribution of pressure balance from both time and space dimensions, providing strong support for the precise allocation and dynamic adjustment of resources. Updating the pressure state data set according to the spatio-temporal distribution characteristics can continuously optimize the control strategy of the system and improve the accuracy and adaptability of control.

[0022] When dynamically regulating the resources of the quench boiler and the cracking furnace, mapping each unit with the unit identification to the distribution characteristics of pressure balance one by one can accurately control the execution actions such as valve opening adjustment, flow regulation, and coolant distribution according to the spatio-temporal distribution characteristics of pressure balance, and realize the dynamic distribution of resources. When the pressure balance reaches the preset intensity threshold at the target unit, an instruction to increase the opening of the adjacent valve is triggered, which can timely respond to the pressure change and ensure the stable operation of the system. Dynamically combining the available fluid resources according to the flow regulation strategy to generate a coolant distribution vector, and adjusting the pipeline parameters of the transmission node of the target unit based on this vector improves the utilization efficiency of resources and the operation efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is the working principle diagram of the pressure balance control method for the coordinated operation of the quench boiler and the cracking furnace described in the present invention; Figure 2 It is the flow chart for determining the distribution characteristics of pressure balance; Figure 3 It is the flow chart for processing the pressure state matrix; Figure 4 It is the flow chart for calculating the pressure intensity gradient. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] Please refer to Figures 1-4 , a pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to the present invention, which includes a plurality of pressure sensors deployed at the inlet and outlet of the quench boiler, a plurality of temperature sensors deployed in the reaction section of the cracking furnace, and a coordinated control server connecting these sensors, with a set distance between adjacent two pressure sensors. The specific implementation steps are as follows: Collect the pressure data and temperature data during the coordinated operation of the quench boiler and the cracking furnace through the pressure sensors and temperature sensors to generate a pressure state data set. Among them, the pressure sensors monitor the pressure changes at different positions of the quench boiler in real time, and the temperature sensors collect the temperature in the reaction section of the cracking furnace in real time.

[0026] The coordinated control server receives the real-time data of a plurality of pressure sensors and a plurality of temperature sensors and constructs a pressure state matrix. The server integrates and processes the received data and stores the real-time state information of pressure and temperature in the form of a matrix.

[0027] Based on the pressure state data set and the real-time data of each sensor, determine the distribution characteristics of pressure balance. This process includes processing the pressure state matrix, extracting pressure characteristics in combination with the pressure state data set, predicting the pressure distribution according to the pressure gradient change and flow path information, outputting the spatio-temporal distribution characteristics of pressure balance through the collaborative control model, and then updating the pressure state data set according to the spatio-temporal distribution characteristics.

[0028] According to the distribution characteristics, dynamically regulate the resources of the quench boiler and the cracking furnace. Based on the determined distribution characteristics of pressure balance, reasonably allocate and adjust the resources in the system to achieve pressure balance and stable operation of the system.

[0029] Example 1: When determining the distribution characteristics of pressure balance, it is necessary to process the pressure state matrix and extract the pressure intensity distribution, flow delay characteristics and pressure change trend. First, standardize the pressure state matrix to eliminate the influence of data with different dimensions and make the data comparable. Then, intercept the pressure hot spots in the matrix through a sliding window. These areas are often the positions where the pressure changes violently or are crucial for the system operation. Filter the noise of these areas and use a suitable filtering algorithm to remove the noise mixed in the data acquisition and transmission process to ensure the accuracy of the data. Then, analyze the processed data through the differential decomposition algorithm to calculate the pressure change trend and thus understand the law of pressure change over time.

[0030] When calculating the spatial correlation characteristics of the pressure state matrix, analyze the correlation between data at different positions in the matrix, and thus obtain the interference intensity between units, the pipeline stability coefficient and the pressure blind area index. Based on these characteristics, construct a feature fusion network. This network can comprehensively process various feature information, obtain the flow delay characteristics through the calculation of the network, and then master the delay situation when the fluid flows in the system.

[0031] Extract the time-domain characteristics and frequency-domain characteristics of the data collected by each pressure sensor. The time-domain characteristics can reflect the change of pressure in the time series, such as the fluctuation amplitude and period of pressure; the frequency-domain characteristics reveal the frequency components of pressure change and can help analyze the reasons for pressure fluctuation. Calculate the pressure feature vector of the sensor according to the phase difference between the time-domain characteristics and the frequency-domain characteristics. This vector can comprehensively characterize the pressure characteristics of the position where the sensor is located. Then, perform feature matching on the pressure sensors at different positions according to the pressure feature vector, analyze the pressure correlation between different sensors through the matching, and then calculate the pressure change trend.

[0032] Perform pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution and flow delay characteristics. Extract pressure intensity sampling points in each frame of data. These sampling points are key information points that can represent the pressure intensity in the frame of data. Associate and map the sampling points with the flow delay characteristics, combine the pressure intensity with the flow delay characteristics of the fluid, and generate a pressure intensity map. Align the pressure intensity maps collected by multiple sensors spatially to ensure the consistency and comparability of data from different sensors in space, thereby calculating the pressure intensity distribution of the unit and clearly understanding the pressure intensity in each unit.

[0033] Set an interference threshold value, which is used to determine whether there is a pressure blind area in the unit. Locate the interference source based on the pressure intensity values of multiple frames of the pressure state matrix to determine the source location of the interference. Calculate the interference intensity difference. If the difference is greater than or equal to the interference threshold value, it indicates that there is a pressure blind area in the unit. At this time, perform flow model constraint compensation on the current unit. According to the flow resistance loss model corresponding to the current unit, iteratively correct the pressure intensity distribution of the current unit. Through continuous iteration, make the pressure intensity distribution more in line with the actual situation. Calculate the intensity compensation value of the blind area pressure based on the correction result to make up for the impact of the pressure blind area on the system operation. Finally, perform pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution, establish a model that accurately describes the system pressure intensity, and perform time delay annotation on the pressure model of the unit through the flow delay characteristics so that the model can reflect the time delay characteristics of pressure changes.

[0034] When calculating the pressure intensity gradient according to the pressure sensor positions, first process multiple groups of pressure coverage data and extract the pressure intensity change points from them. These points are the positions where the pressure changes significantly and have important analysis value. Map the change points to a unified coordinate framework according to the deployment positions of the sensors so that the change points at different positions can be analyzed and processed in the same coordinate system. Fit the change points through a spatial interpolation algorithm to generate a pressure field model of the unit, which can intuitively represent the pressure distribution in the unit.

[0035] Perform equally spaced sampling along the flow path of the pressure field model to obtain the pressure data of each point on the path. Calculate the pressure attenuation rate, interference fluctuation index, and pressure change slope of the path according to the sampling results. These parameters can reflect the change characteristics of the pressure on the flow path. Calculate the pressure change parameters based on these parameters to lay a foundation for calculating the pressure intensity gradient. Project the distribution characteristics of the pressure change in each frame of data onto the pressure field model according to the deployment parameters and acquisition accuracy of the pressure sensors. Divide the pressure field model into zones along the flow direction according to the number of sensors, analyze the pressure changes in each zone, find the rules of pressure changes in the zone, and calculate the pressure distribution characteristics according to the rules to understand the pressure distribution in different regions.

[0036] Finally, calculate the pressure intensity gradient based on the pressure change parameters and the pressure distribution characteristics. The specific process is as follows: Based on the coordinate position range from the first pressure sensor to the last pressure sensor, select spatial coordinate points in the sensor deployment direction, cumulatively calculate the product of the pressure field strength characteristic weight value and the pressure distribution characteristic weight value within the spatial resolution range, and at the same time superimpose the influence value of the sensor acquisition frequency on the pressure intensity change rate, so as to obtain an accurate pressure intensity gradient.

[0037] When calculating the pressure prediction information of each sub-unit, use the main flow path of the pressure field model as the reference line, take the peak position of the pressure change in each frame of data as the reference point, calculate the pressure offset, and draw the pressure distribution curve according to the coordinates. According to the pressure intensity gradient, correct the growth rate and direction in the pressure change trend to make the prediction more in line with the actual situation. Starting from the nearest pressure distribution point, continue to draw the distribution curve according to the correction results of the growth rate and direction to generate the pressure distribution points in the next time period until the distribution points cover the entire target unit, and generate the pressure prediction information.

[0038] Construct a collaborative control model, which includes an input layer, a feature fusion layer, and a resource allocation layer. The input layer organizes the pressure prediction information into spatial distribution data and performs standardization processing to unify the data format for subsequent processing. The feature fusion layer extracts the unit correlation features of the pressure by processing the spatial distribution data, constructs the dependency relationship between physical units, so as to understand the mutual influence between units. The resource allocation layer integrates the correlation relationship of pressure balance on spatial units to generate a resource regulation strategy, providing guidance for the dynamic regulation of system resources.

[0039] According to the spatio-temporal distribution characteristics of pressure balance output by the collaborative control model, correspond the identification of the sub-units with the spatio-temporal distribution characteristics, so that the pressure balance spatio-temporal characteristics of each sub-unit can be accurately identified. Reorganize the unit data in the pressure state dataset according to the spatio-temporal characteristics to generate a unit distribution map sorted by the pressure balance intensity, intuitively showing the pressure balance of each unit. According to the reorganized unit distribution map, output the optimized pressure balance distribution characteristics.

[0040] Example 2: When processing the pressure state matrix, it is necessary to perform standardization processing on the matrix. Since the deployment positions of the pressure sensors at the inlet and outlet of the quench boiler are different, the dimensions of the collected data may be different. Through standardization processing, the influence of different dimensions on the data can be eliminated, making the data from sensors at different positions comparable and laying a unified data foundation for subsequent analysis. For example, map the real-time data of each pressure sensor to the same numerical interval to ensure that there will be no analysis deviation due to dimension problems in subsequent processing.

[0041] After the standardization process, the sliding window technique is used to intercept the pressure hot spots in the matrix. The size of the sliding window here needs to be determined according to the set distance between adjacent pressure sensors and the frequency of system pressure changes. Usually, a window size that can cover at least the data of two adjacent sensors is selected to ensure that the intercepted area contains sufficient pressure change information. These hot spots are generally locations with large pressure gradient changes, such as the inlet of the quench boiler and the elbow of the outlet pipeline. The pressure changes at these locations have a greater impact on the overall pressure balance of the system and need to be analyzed key points.

[0042] When filtering the noise of the intercepted hot spot area, a combination of median filtering and Kalman filtering is adopted. Median filtering can effectively remove impulse noise and avoid data anomalies caused by instantaneous interference of sensors; Kalman filtering can, based on the system state equation and observation equation, perform optimal estimation on the pressure data, further smooth the data, and reduce the influence of random noise.

[0043] When calculating the pressure change trend through the differential decomposition algorithm, the pressure data on the time series is decomposed into a trend term, a periodic term, and a random term. The trend term reflects the overall change direction of the pressure, the periodic term reflects the periodic fluctuation of the pressure over time, and the random term represents the unpredictable accidental fluctuation.

[0044] When calculating the spatial correlation characteristics of the pressure state matrix, the correlation coefficients of the data in different rows and columns of the matrix are analyzed. For example, for multiple pressure sensors deployed on the same pipeline cross-section, the correlation of the collected data should be relatively high. If the correlation decreases abnormally, it may indicate problems such as pipeline leakage or blockage near this cross-section. Based on these correlation characteristics, the interference intensity between units is calculated, and this intensity value reflects the mutual influence degree of the pressure changes of adjacent units; at the same time, the pipeline stability coefficient is calculated to evaluate the stability performance of the pipeline under pressure fluctuations; and the pressure blind area index is calculated to characterize the areas in the system where pressure monitoring may be insufficient.

[0045] When constructing the feature fusion network, a multi-layer perceptron structure is adopted. The spatial correlation characteristics, interference intensity, pipeline stability coefficient, and pressure blind area index, etc. are used as the input layer nodes. Through the weight matrix operation of the hidden layer, the non-linear fusion of features is realized. The output layer nodes of the network correspond to the flow delay characteristics. This characteristic continuously adjusts the network weights through the training data, so that the output flow delay characteristics can accurately reflect the actual flow delay of the fluid in the pipeline, providing key parameters for subsequent pressure intensity modeling.

[0046] For the data collected by each pressure sensor, time-domain features and frequency-domain features are extracted. Time-domain feature extraction includes calculating the mean, variance, peak value, rise time, etc. of the pressure data. These parameters can intuitively reflect the variation characteristics of pressure in the time dimension, such as the amplitude and frequency of pressure fluctuations. Frequency-domain feature extraction converts the time-domain signal into a frequency-domain signal through fast Fourier transform to obtain the main frequency components of pressure fluctuations. For example, if a certain frequency component in the frequency domain accounts for a significant proportion, it may indicate the existence of a periodic pressure disturbance source in the system, such as the operating frequency of a pump, etc.

[0047] When calculating the pressure feature vector of the sensor according to the phase difference between the time-domain feature and the frequency-domain feature, first determine the corresponding phases of each feature point (such as peak point, zero-crossing point) in the time-domain signal in the frequency domain, calculate the statistical quantities (such as mean, standard deviation) of these phase differences, and combine them with the key parameters of the time-domain feature and the frequency-domain feature into a multi-dimensional vector. This vector contains the time-domain, frequency-domain and phase information of the pressure change at the sensor location and can comprehensively characterize the pressure characteristics of this location.

[0048] When performing feature matching on pressure sensors at different positions, the cosine similarity algorithm is used to calculate the similarity of the pressure feature vectors of each sensor. Sensors with high similarity indicate that the pressure change characteristics at their locations are similar and can be classified into the same pressure feature region; sensors with low similarity may be in different pressure influence regions or there are abnormal situations. Through this matching analysis, the system can be divided into multiple pressure feature regions, providing a basis for regional division for subsequent pressure distribution prediction and resource regulation.

[0049] When calculating the pressure change trend, comprehensively consider the pressure change situation in the regions after feature matching of each sensor. For sensors in the same feature region, take the average value of their pressure change trends as the pressure change trend of this region; for different regions, analyze the differences and correlations of their trends to obtain an overall picture of the pressure change trend of the entire system.

[0050] Example 3: When performing pressure intensity modeling on the pressure state matrix, first, according to the pressure intensity distribution, pressure intensity sampling points need to be extracted from each frame of data. Due to the deployment of pressure sensors at different positions at the inlet, outlet and pipeline of the quench boiler, each frame of data will show pressure intensity values at different positions. By setting a sampling threshold, select the points with significant pressure intensity changes as sampling points, and these points can effectively characterize the pressure characteristics of this frame of data. For example, in the pressure sensor data at the inlet of the quench boiler, when the pressure intensity exceeds 10% of the upper and lower limits of the normal operating range, mark this point as a sampling point.

[0051] The extracted sampling points are associated and mapped with the flow delay features, which are obtained through the previous spatial correlation feature analysis of the pressure state matrix and the calculation of the feature fusion network, and reflect the time delay of the fluid flowing from one position to another in the pipeline. By establishing the mapping relationship between the pressure intensity value of the sampling point and the flow delay time at the corresponding position, a pressure intensity map is generated. The map uses the spatial position as the horizontal axis and the pressure intensity as the vertical axis, and at the same time uses the color depth in the map to represent the length of the flow delay time, so as to visually associate the two key parameters of pressure intensity and flow delay in the spatial dimension.

[0052] Spatially align the pressure intensity maps collected by multiple sensors. Since the deployment positions of the sensors are different, it is necessary to establish a unified spatial coordinate system based on the physical structure of the quench boiler and the cracking furnace. For example, taking the inlet of the reaction section of the cracking furnace as the coordinate origin, the material flow direction as the x-axis, and the directions perpendicular to the flow direction as the y-axis and z-axis, convert the pressure intensity maps of each sensor to this coordinate system to ensure the consistency and comparability of the data of different sensors in terms of spatial position, and then calculate the pressure intensity distribution of the unit to clarify the spatial distribution of the pressure intensity in each sub-unit.

[0053] When setting the interference threshold value, it is necessary to comprehensively consider the pressure fluctuation range during the normal operation of the system and the historical interference data. By statistically analyzing the pressure data during the normal operation of the system, calculate the mean and standard deviation of the pressure intensity. Usually, the interference threshold value is set as the mean plus 2 times the standard deviation, which is used as the benchmark for judging whether there is a pressure blind area in the unit. When locating the interference source based on the pressure intensity values of multiple frames of pressure state matrices, determine the approximate position of the interference source by comparing the change amplitude and time sequence of the pressure intensity at different positions.

[0054] When calculating the difference in interference intensity, for adjacent sub-units, obtain the change value of the pressure intensity within the same time period, and subtract the smaller change value from the larger change value to get the difference. If this difference is greater than or equal to the interference threshold value, it indicates that there may be a pressure blind area in this unit, that is, the pressure in this area fails to be effectively monitored or there is an abnormal pressure distribution. At this time, perform a flow model constraint compensation on the current unit. According to the flow resistance loss model corresponding to the current unit, this model describes the pressure loss caused by pipeline resistance, elbows, and valves when the fluid flows in this unit; based on the flow resistance loss model, perform iterative correction on the pressure intensity distribution of the current unit. First, calculate the theoretical pressure loss according to the initial pressure intensity distribution and the flow resistance loss model, and compare it with the actual measured pressure difference. If the difference is large, adjust the pressure intensity distribution parameters and recalculate until the error between the theoretical calculated value and the actual measured value is within the allowable range (usually set to not exceed 5%). According to the corrected pressure intensity distribution, calculate the intensity compensation value of the blind area pressure, and this compensation value is used to correct the pressure in the pressure blind area subsequently to ensure the accuracy of the system pressure balance analysis.

[0055] Perform pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution. Organize the pressure intensity values in each frame of data according to the spatial position and time sequence to form a model that can reflect the dynamic change of the system pressure intensity. At the same time, perform time delay annotation on the pressure model of the unit through the flow delay feature, and add the corresponding flow delay time label to each pressure intensity data point in the model, so that the model can not only reflect the spatial distribution of the pressure intensity, but also reflect the time delay characteristic of the pressure change, thus more accurately describing the pressure dynamic behavior of the system.

[0056] Example 4: When calculating the pressure intensity gradient according to the position of the pressure sensor, multiple sets of pressure coverage data need to be processed. These data come from pressure sensors deployed at different positions at the inlet, outlet, and pipeline of the quench boiler. For example, in a certain chemical production system, 2 pressure sensors are deployed at the inlet of the quench boiler, 3 pressure sensors are deployed at the outlet, and 1 pressure sensor is deployed every 5 meters on the pipeline. The adjacent two pressure sensors are 5 meters apart. When the system is running, these sensors will collect pressure data in real time to form multiple sets of pressure coverage data. Extract the pressure intensity change points from these data, that is, the positions where the pressure value changes significantly. For example, when the difference in the pressure value of a certain sensor between two adjacent sampling moments exceeds 15% of the normal fluctuation range, mark the pressure point at this moment as a change point.

[0057] According to the deployment locations of the sensors, map the change points to a unified coordinate framework. Taking the center of the reaction section of the cracking furnace as the coordinate origin, set the direction in which the material flows from the cracking furnace to the quench boiler as the positive x-axis direction, and the direction perpendicular to the ground and upward as the positive z-axis direction to establish a three-dimensional rectangular coordinate system. For example, the coordinates of the first pressure sensor at the inlet are (0, 0, 1.5), the second is (0, 0, 2.0), and the coordinates of the first sensor at the outlet are (10, 0, 1.8), etc. Map the positions of all pressure intensity change points to this coordinate system according to the coordinates of the corresponding sensors, so that the change points at different positions can be analyzed in the same coordinate system.

[0058] Fit the change points through a spatial interpolation algorithm to generate the pressure field model of the unit. Use the Kriging interpolation method, which takes into account the spatial correlation of the data. According to the coordinates and pressure values of the change points, calculate the pressure estimation values at any position within the entire unit space, thereby constructing a continuous pressure field model. For example, for the pipeline unit between the quench boiler and the cracking furnace, fit the pressure distribution surface within this pipeline through the interpolation method to visually display the pressure distribution within this unit, such as the change trend of the pressure distribution at the pipeline elbow.

[0059] Perform equally spaced sampling along the flow path of the pressure field model. Assume that the flow path is the material flow trajectory from the outlet of the reaction section of the cracking furnace to the inlet of the quench boiler. Select a sampling point every 0.5 meters to obtain the pressure data at this point. Calculate the pressure decay rate of the path, that is, the pressure reduction value per unit length. For example, in a 5-meter-long pipeline section, the inlet pressure is 1.2 MPa and the outlet pressure is 1.1 MPa, and the pressure decay rate is (1.2 - 1.1) / 5 = 0.02 MPa / m; calculate the interference fluctuation index. By analyzing the fluctuation amplitude and frequency of the pressure values at the sampling points, evaluate the degree of influence of the interference on the pressure. For example, when the pressure fluctuates frequently and with a large amplitude in a short period of time, the interference fluctuation index is relatively high; calculate the pressure change slope, that is, the change rate of pressure with distance, which reflects how fast the pressure changes along the flow path. For example, for a certain path where the pressure drops from 1.0 MPa to 0.8 MPa and the length is 4 meters, the slope is (0.8 - 1.0) / 4 = -0.05 MPa / m. Calculate the pressure change parameters based on these parameters. Weightedly sum the pressure decay rate, interference fluctuation index, and pressure change slope according to certain weights (such as taking 0.4, 0.3, and 0.3 respectively) to obtain a comprehensive pressure change parameter, which is used to characterize the pressure change characteristics of this flow path.

[0060] According to the deployment parameters and acquisition accuracy of the pressure sensors, for example, the deployment height of a certain type of pressure sensor is 1.5 meters from the ground, and the acquisition accuracy is ±0.01 MPa. Project the distribution characteristics of the pressure change in each frame of data onto the pressure field model. Each frame of data contains the pressure values of each sensor at the same moment. Map these pressure values to the corresponding positions in the pressure field model to form the pressure distribution projection at that moment. Divide the area along the flow direction of the pressure field model according to the number of sensors. If there are 10 sensors deployed along the flow direction, divide the flow path into 10 intervals, and each interval corresponds to the monitoring range of one sensor. Analyze the variation law of the pressure change within the divided area. For example, within a certain area, the pressure fluctuates periodically with time, the fluctuation period is 5 minutes, and the fluctuation amplitude is 0.05 MPa. Calculate the pressure distribution characteristics according to this law to determine the average level, fluctuation range and other characteristics of the pressure within this area.

[0061] Calculate the pressure intensity gradient according to the pressure change parameters and pressure distribution characteristics. Based on the coordinate position range from the first pressure sensor to the last pressure sensor, such as the coordinate of the first sensor is (0, 0, 1.5) and the last one is (15, 0, 1.7), select several spatial coordinate points in the sensor deployment direction (x-axis direction), for example, take a point every 1 meter. For each coordinate point, calculate the product of the pressure field intensity characteristic weight value and the pressure distribution characteristic weight value within the spatial resolution range (such as a spherical area with a radius of 0.5 meters centered on this point). The pressure field intensity characteristic weight value is determined according to the magnitude of the pressure gradient within this area, and the pressure distribution characteristic weight value is determined according to the stability of the pressure distribution within this area. At the same time, superimpose the influence value of the sensor acquisition frequency on the pressure intensity change rate. If the sensor acquisition frequency is 10 Hz, a higher acquisition frequency can reflect the pressure change more timely and has a greater influence value on the change rate, and vice versa. By accumulating these products and influence values, the pressure intensity gradient is obtained. This gradient reflects the change intensity and direction of the pressure in space. For example, in the x-axis direction, the pressure intensity gradient is 0.03 MPa / m, indicating that for every 1 meter forward along the x-axis, the pressure increases by an average of 0.03 MPa.

[0062] During the calculation process, it is necessary to fully consider the actual situation of sensor deployment, such as whether the positions of the sensors are uniform and whether the acquisition accuracies are consistent. For example, if the sensors are deployed densely in a certain area, the data collected can reflect the pressure change more precisely, and a higher weight should be given to this area when calculating the pressure intensity gradient; if the acquisition accuracy of a certain sensor is low, appropriate error correction should be performed on its data during the calculation. Through this calculation method based on the actual deployment parameters and data characteristics, ensure that the calculation result of the pressure intensity gradient can accurately reflect the actual pressure distribution of the system, providing a key basis for subsequent calculation of the pressure prediction information of the sub-unit, constructing a collaborative control model, and performing resource dynamic regulation.

[0063] Example 5: When dynamically regulating the resources of the quench boiler and the cracking furnace, it is first necessary to map each unit with the unit identification and the distribution characteristics of pressure balance one by one. Taking a chemical production system as an example, the system divides the quench boiler and the cracking furnace into 5 units according to the physical structure, namely the cracking furnace reaction section unit, the quench boiler inlet pipeline unit, the quench boiler heat exchange tube unit, the quench boiler outlet pipeline unit, and the gas-liquid separation unit. Each unit is assigned a unique identification, such as U1, U2, U3, U4, U5. Through the previously determined distribution characteristics of pressure balance, clarify the pressure balance state of each unit. For example, the pressure intensity distribution of the U2 unit at a certain moment shows a gradient characteristic of high at the inlet and low at the outlet. Associate this characteristic with the identification of the U2 unit to establish a clear corresponding relationship for subsequent precise regulation of the pressure situation of each unit.

[0064] According to the spatio-temporal distribution characteristics of pressure balance, control the execution actions of resources, including valve opening adjustment, flow rate regulation, and coolant distribution operations. When the pressure balance reaches the preset intensity threshold in the target unit, trigger the instruction to increase the opening of the adjacent valve. For example, if the pressure intensity of the U3 unit (quench boiler heat exchange tube unit) exceeds the preset threshold of 1.5 MPa, the system automatically judges that there may be a risk of excessive pressure in this unit. At this time, trigger the instruction to increase the opening of the valve V1 at the inlet of the U3 unit, and increase the valve opening from 50% to 60% to increase the fluid throughput and reduce the pressure of this unit.

[0065] Dynamically combine available fluid resources according to the flow rate regulation strategy to generate a coolant distribution vector. Assume that the available coolant resources in the system include circulating water and chilled water. The flow rate regulation range of circulating water is 100 - 300 m³ / h, and the flow rate regulation range of chilled water is 50 - 150 m³ / h. When the pressure balance distribution of the U4 unit (quench boiler outlet pipeline unit) shows a high temperature and large pressure fluctuation at the outlet, the system calculates according to the preset flow rate regulation strategy that it is necessary to combine 200 m³ / h of circulating water and 100 m³ / h of chilled water to generate a coolant distribution vector of [200, 100]. Based on this vector, adjust the pipeline parameters of the transmission node of the target unit, such as adjusting the valve opening of the corresponding pipeline, so that the circulating water and chilled water enter the U4 unit at the set flow rate to cool this unit to stabilize the pressure.

[0066] According to the spatial distribution of pressure balance and the preset regulation strategy, resources are dynamically allocated to the corresponding physical units. For example, through the analysis of the pressure field model, it is found that there is a pressure blind area in the middle region of Unit U1 (the reaction section unit of the cracking furnace), and the pressure in this area fails to be effectively transmitted to the sensor, resulting in insufficient pressure monitoring. According to the preset regulation strategy, it is necessary to increase the coolant supply to this area to balance the pressure. The system automatically allocates an additional 50 m³ / h of circulating water to the middle region of Unit U1 and injects it through the cooling water pipeline in this area to adjust the pressure distribution in this area and improve the situation of the pressure blind area.

[0067] During the adjustment of the valve opening, the adjustment accuracy and response time of the valve need to be considered. For example, the adjustment accuracy of Valve V1 is ±1%, and the response time is 5 seconds. After the system sends an instruction to increase the opening, it will monitor the actual change of the valve opening in real time to ensure that the valve operates accurately according to the instruction. If it is detected that the valve opening does not reach 60% after 5 seconds, the system will send a supplementary adjustment instruction until the valve opening reaches the target value.

[0068] When adjusting the flow rate, it is necessary to comprehensively consider the supply capacity of each fluid resource and the system pressure balance requirements. When the system needs to adjust the flow rate of multiple units simultaneously, such as both Unit U2 and Unit U3 need to increase the flow rate, the system will reasonably allocate the flow rate according to the pressure balance priority of each unit and the total supply of fluid resources. For example, the risk of pressure imbalance in Unit U2 is higher, and the priority is set to Level 1, while the priority of Unit U3 is set to Level 2, and the total circulating water supply is 400 m³ / h. Then, first meet the 200 m³ / h required by Unit U2, and then allocate the remaining 200 m³ / h to Unit U3.

[0069] In the operation of coolant distribution, the influence of the temperature and pressure of the coolant on the system needs to be considered. The temperature of the circulating water is 25°C, and the pressure is 0.4 MPa. The temperature of the chilled water is 5°C, and the pressure is 0.6 MPa. When distributing the coolant, it is necessary to ensure that the temperature and pressure of the coolant entering each unit meet the process requirements of this unit to avoid adverse effects on the system due to abnormal coolant parameters. For example, the upper limit of the coolant pressure allowed by the heat exchange tubes of Unit U3 is 0.5 MPa. When distributing the chilled water, it is necessary to reduce the chilled water pressure from 0.6 MPa to 0.45 MPa through a pressure regulating valve and then inject it into this unit.

[0070] During the process of dynamic resource regulation, the system will monitor the change of the pressure balance state of each unit in real time and adjust the regulation strategy according to the latest pressure distribution characteristics. For example, after increasing the valve opening and distributing the coolant to Unit U3, continuously monitor the change of the pressure intensity of this unit. If the pressure gradually drops to 1.2 MPa within 10 minutes and is within the normal range, maintain the current regulation state; if the pressure does not drop significantly or continues to rise, start the standby regulation plan, such as further increasing the valve opening or increasing the coolant supply.

[0071] It should be noted that in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0072] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace, which is applied to a chemical production control system. The system includes multiple pressure sensors deployed at the inlet and outlet of the quench boiler, multiple temperature sensors deployed in the reaction section of the cracking furnace, and a coordinated control server connecting the pressure sensors and the temperature sensors. The distance between two adjacent pressure sensors is a set distance. It is characterized in that, The method includes: Collecting pressure data and temperature data during the coordinated operation of the quench boiler and the cracking furnace through the pressure sensor and the temperature sensor, and generating a pressure state data set; Receiving real-time data of multiple pressure sensors and multiple temperature sensors through the coordinated control server, and constructing a pressure state matrix; Determining the distribution characteristics of pressure balance according to the pressure state data set and the real-time data of each sensor. Wherein, determining the distribution characteristics of pressure balance includes: processing the pressure state matrix, extracting pressure characteristics in combination with the pressure state data set, predicting the pressure distribution according to the pressure gradient change and the flow path information, outputting the spatio-temporal distribution characteristics of pressure balance through the coordinated control model, and updating the pressure state data set according to the spatio-temporal distribution characteristics; Dynamically regulating the resources of the quench boiler and the cracking furnace according to the distribution characteristics.

2. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 1, wherein, The determining the distribution characteristics of pressure balance includes: Processing the pressure state matrix, and extracting the pressure intensity distribution, the flow delay characteristics and the pressure change trend; Performing pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution and the flow delay characteristics, dividing the quench boiler and the cracking furnace into multiple sub-units and marking unit identifiers, and performing association matching according to the pressure intensity of the sub-units and the pressure state data set, and marking the unit identifiers in the pressure state data set; Calculating the pressure intensity gradient according to the position of the pressure sensor, predicting the distribution of pressure balance according to the pressure intensity gradient and the pressure change trend, and calculating the pressure prediction information of each sub-unit; Constructing a coordinated control model, using the pressure prediction information as the input parameter of the coordinated control model, performing spatial association modeling on the pressure prediction information through the coordinated control model, and outputting the spatio-temporal distribution characteristics of pressure balance; Updating the pressure state data set according to the spatio-temporal distribution characteristics, and obtaining the distribution characteristics of pressure balance.

3. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 2, characterized in that The processing the pressure state matrix includes: Normalizing the pressure state matrix, intercepting the pressure hot spot area in the matrix through a sliding window, filtering the noise of the hot spot area, and calculating the pressure change trend through a differential decomposition algorithm; Calculating the spatial correlation characteristics of the pressure state matrix, calculating the interference intensity, the pipeline stability coefficient and the pressure blind area index between units according to the spatial correlation characteristics, constructing a feature fusion network, and calculating the flow delay characteristics through the feature fusion network; Extracting the time domain characteristics and frequency domain characteristics collected by each pressure sensor, calculating the pressure feature vector of the sensor according to the phase difference between the time domain characteristics and the frequency domain characteristics, performing feature matching on the pressure sensors at different positions according to the pressure feature vector, and calculating the pressure change trend.

4. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 3, wherein The performing pressure intensity modeling on the pressure state matrix includes: Extracting pressure intensity sampling points in each frame of data according to the pressure intensity distribution, performing association mapping according to the sampling points and the flow delay characteristics, generating a pressure intensity map, spatially aligning the pressure intensity maps collected by multiple sensors, and calculating the pressure intensity distribution of the unit; Set the interference threshold value, locate the interference source according to the pressure intensity value of the multi-frame pressure state matrix, calculate the interference intensity difference. If the interference intensity difference is greater than or equal to the interference threshold value, it indicates that there is a pressure blind area in this unit. Perform flow model constraint compensation on the current unit, and according to the flow resistance loss model corresponding to the current unit, iteratively correct the pressure intensity distribution of the current unit, and calculate the intensity compensation value of the blind area pressure according to the correction result; Perform pressure intensity modeling on the pressure state matrix according to the pressure intensity distribution, and perform time delay annotation on the pressure model of the unit through the flow delay feature.

5. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 4, characterized in that The calculating the pressure intensity gradient according to the pressure sensor position includes: Extract the pressure intensity change points according to multiple groups of pressure coverage data, map the change points to a unified coordinate framework according to the deployment positions of the sensors, and fit the change points through a spatial interpolation algorithm to generate the pressure field model of the unit; Perform equally spaced sampling along the flow path of the pressure field model, calculate the pressure attenuation rate, interference fluctuation index and pressure change slope of the path according to the sampling results, and calculate the pressure change parameters according to the pressure attenuation rate, interference fluctuation index and pressure change slope; According to the deployment parameters and acquisition accuracy of the pressure sensors, project the distribution characteristics of the pressure change in each frame of data onto the pressure field model, partition along the flow direction of the pressure field model according to the number of sensors, analyze the change rules of the pressure change within the partition, and calculate the pressure distribution characteristics according to the change rules; Calculate the pressure intensity gradient according to the pressure change parameters and the pressure distribution characteristics. The calculation process of the pressure intensity gradient includes: based on the coordinate position range from the first pressure sensor to the last pressure sensor, select spatial coordinate points in the sensor deployment direction, cumulatively calculate the product of the pressure field strength characteristic weight value and the pressure distribution characteristic weight value within the spatial resolution range, and superimpose the influence value of the sensor acquisition frequency on the pressure intensity change rate.

6. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 5, characterized in that, The calculating the pressure prediction information of each sub-unit includes: Taking the main flow path of the pressure field model as the reference line, taking the peak position of the pressure change in each frame of data as the reference point, calculate the pressure offset, and draw the pressure distribution curve according to the coordinates; Correct the growth rate and direction in the pressure change trend according to the pressure intensity gradient; Starting from the nearest pressure distribution point, continue to draw the distribution curve according to the correction results of the growth rate and direction to generate the pressure distribution points in the next time period until the distribution points cover the entire target unit to generate the pressure prediction information.

7. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 2, wherein The constructing the cooperative control model includes: The input layer is used to organize the pressure prediction information into spatial distribution data and perform normalization processing; The feature fusion layer is used to extract the unit association features of the pressure by processing the spatial distribution data and construct the dependency relationship between physical units; The resource allocation layer is used to integrate the association relationship of the pressure balance on the spatial units to generate the resource regulation strategy.

8. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 2, characterized in that, The obtaining the distribution characteristics of the pressure balance includes: According to the spatio-temporal distribution characteristics of the pressure balance output by the cooperative control model, correspond the identifier of the sub-unit to the spatio-temporal distribution characteristics; Recombine the unit data in the pressure state dataset according to the spatio-temporal characteristics to generate a unit distribution map sorted by pressure balance intensity; Output the optimized pressure balance distribution characteristics according to the recombined unit distribution map.

9. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 1, characterized in that, The dynamic regulation of the resources of the quench boiler and the cracking furnace includes: One-to-one map the unit identifiers with the units of the distribution characteristics of the pressure balance; Control the execution actions of the resources according to the spatio-temporal distribution characteristics of the pressure balance, including valve opening adjustment, flow rate regulation and coolant distribution operations; Dynamically allocate the resources to the corresponding physical units according to the spatial distribution of the pressure balance and the preset regulation strategy.

10. The pressure balance control method for the coordinated operation of a quench boiler and a cracking furnace according to claim 9, characterized in that, The control of the execution actions of the resources includes: When the pressure balance reaches the preset intensity threshold in the target unit, trigger the instruction to increase the opening of the adjacent valve; Dynamically combine the available fluid resources according to the flow rate regulation strategy to generate a coolant distribution vector; Adjust the pipeline parameters of the transmission node of the target unit based on the coolant distribution vector.

Images