A coke tower bulging deformation real-time monitoring and hierarchical early warning system and method

Abstract

The application discloses a coke tower bulging deformation real-time monitoring and grading early warning system and method, relates to the field of optical fiber sensing technology and petrochemical equipment safety monitoring. The system comprises an optical fiber sensing module, a data analysis processing module and an early warning platform. The optical fiber sensing module is spirally arranged on the surface of the coke tower by optical fibers to measure the real-time phase change caused by the bulging deformation and temperature of the coke tower in an optical fiber sensing mode. The data analysis processing module establishes a deformation-time sequence model, determines the real-time strain by using a temperature-strain decoupling algorithm according to the real-time phase change, and then obtains the real-time deformation variable of the coke tower by using the deformation-time sequence model. The accelerating stage of the bulging deformation is identified by using a bidirectional long short-term memory network combined with a sliding window fitting mode. The early warning platform determines the crack risk grade by a risk grade matrix early warning rule according to the accumulated deformation variable and the fatigue damage cumulative value generated by the periodic operation of the coke tower. The application significantly improves the timeliness and reliability of the safety monitoring of the coke tower.

Description

Technical Field

[0001] This application relates to the field of fiber optic sensing technology and safety monitoring of petrochemical equipment, and in particular to a real-time monitoring and graded early warning system and method for bulging deformation of coke towers. Background Technology

[0002] Coke towers are subjected to cyclical thermal stress at 490℃ for extended periods, as well as the expansion pressure of internal coke, leading to bulging plastic deformation and even cracks in areas such as tower section welds and skirt supports. Existing coke tower deformation monitoring technologies have several shortcomings: contact sensors (such as strain gauges) require dense deployment, are prone to failure in high-temperature environments, and are difficult to cover the entire tower; non-contact optical measurements (such as laser ranging) are affected by coke adhesion and steam interference, making continuous monitoring impossible; and traditional fiber optic monitoring technologies (such as FBG fiber gratings) only support discrete point measurements and cannot capture overall deformation trends. Summary of the Invention

[0003] The purpose of this application is to provide a real-time monitoring and graded early warning system and method for the swelling and deformation of coke towers, which can significantly improve the timeliness and reliability of safety monitoring of coke towers.

[0004] To achieve the above objectives, this application provides the following solution: In the first aspect, this application provides a real-time monitoring and graded early warning system for the bulging and deformation of coke towers, including: an optical fiber sensing module, a data analysis and processing module, and an early warning platform; The fiber optic sensing module uses fiber optic cables spirally laid on the surface of the coke tower to measure the real-time phase change caused by the expansion and deformation of the coke tower, the real-time phase change caused by the temperature of the coke tower, and the temperature of the coke tower in a fiber optic sensing manner. The data analysis and processing module is used to establish a deformation-time series model based on the historical strain and deformation of the coke tower using the least squares method. The data analysis and processing module is also used to determine the real-time strain of the coke tower by using a temperature-strain decoupling algorithm based on the real-time phase changes generated during expansion and deformation and the real-time phase changes corresponding to the temperature of the coke tower; based on the real-time strain of the coke tower, it uses a deformation-time series model to obtain the real-time deformation of the coke tower; based on the real-time deformation of the coke tower over a preset time period, it uses a bidirectional long short-term memory network combined with a sliding window fitting method to identify the expansion and deformation acceleration stage; when the phase change of the expansion and deformation acceleration stage is greater than the phase fluctuation threshold, it triggers a crack initiation risk warning. The early warning platform is used to construct risk level matrix early warning rules; and based on the cumulative deformation and fatigue damage accumulated by the periodic operation of the coke tower, the crack risk level is determined through the risk level matrix early warning rules.

[0005] Secondly, this application provides a method for real-time monitoring and graded early warning of bulging deformation in coke towers, including: Based on the historical strain and deformation of the coke tower, a deformation-time series model is established using the least squares method; The system acquires the real-time phase changes caused by the expansion and deformation of the coke tower, the real-time phase changes caused by the temperature of the coke tower, and the temperature of the coke tower. Based on the real-time phase change generated during expansion and deformation and the real-time phase change corresponding to the temperature of the coke tower, a temperature-strain decoupling algorithm is used to determine the real-time strain of the coke tower. Based on the real-time strain of the coking tower, the real-time deformation of the coking tower is obtained using a deformation-time series model; Based on the real-time deformation of the coke tower over a preset time period, a bidirectional long short-term memory network combined with a sliding window fitting method is used to identify the accelerated expansion and deformation stage. When the phase change during the accelerated expansion and deformation phase exceeds the phase fluctuation threshold, a crack initiation risk warning is triggered. Construct risk level matrix early warning rules; Based on the cumulative deformation and fatigue damage accumulated during the periodic operation of the coking tower, the crack risk level is determined using the risk level matrix early warning rules.

[0006] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a real-time monitoring and graded early warning system and method for coke tower bulging deformation. Utilizing optical fibers as sensors, the system is spirally deployed along the coke tower. The fibers continuously monitor strain changes along its length, providing high spatial resolution data. Furthermore, because optical fibers transmit optical signals, they are unaffected by electromagnetic interference, ensuring data accuracy and improving the reliability of coke tower safety monitoring. A deformation-time series model is constructed to accurately identify the accelerated expansion deformation stage. Using a risk level relationship matrix and risk level matrix early warning rules, a graded response is achieved through an early warning platform, forming a closed-loop management system from trend monitoring to risk control. This significantly improves the timeliness and reliability of petrochemical equipment safety monitoring. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1A schematic diagram of the structure of a real-time monitoring and graded early warning system for the bulging and deformation of a coke tower provided in an embodiment of this application; Figure 2 This is a schematic diagram of the fiber optic cable layout provided in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the principle of the temperature-strain decoupling algorithm provided in the embodiments of this application; Figure 4 This is a flowchart illustrating a method for real-time monitoring and graded early warning of coke tower bulging deformation provided in an embodiment of this application. Detailed Implementation

[0009] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0010] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0011] In one exemplary embodiment, such as Figure 1 As shown, a real-time monitoring and graded early warning system for coke tower swelling and deformation is provided, comprising: an optical fiber sensing module, a data analysis and processing module, and an early warning platform. The optical fiber sensing module uses optical fibers spirally arranged on the surface of the coke tower to measure the real-time phase changes generated during expansion and deformation, the real-time phase changes caused by the coke tower temperature, and the coke tower temperature itself. The data analysis and processing module is used to establish a deformation-time series model based on the historical strain and deformation of the coke tower using the least squares method. The data analysis and processing module is also used to determine the real-time strain of the coke tower using a temperature-strain decoupling algorithm based on the real-time phase changes generated during expansion and deformation and the real-time phase changes corresponding to the coke tower temperature; to obtain the real-time deformation of the coke tower using the deformation-time series model based on the real-time strain; to identify the accelerated expansion and deformation stage using a bidirectional long short-term memory network combined with a sliding window fitting method based on the real-time deformation of the coke tower over a preset time period; and to trigger a crack initiation risk warning when the phase change during the accelerated expansion and deformation stage exceeds a phase fluctuation threshold. The early warning platform is used to construct risk level matrix early warning rules; and based on the cumulative deformation and fatigue damage accumulated by the periodic operation of the coke tower, the crack risk level is determined through the risk level matrix early warning rules.

[0012] This application utilizes a high-temperature resistant distributed fiber optic sensor network, phase-sensitive signal demodulation, and dynamic temperature compensation technology to achieve real-time location and quantitative analysis of tower deformation. This system can replace traditional point sensors and operates stably for extended periods in high-temperature and corrosive environments. Furthermore, it establishes a deformation accumulation model to predict structural failure risks, achieving closed-loop management from trend monitoring to risk control, providing a more cost-effective solution for structural health management of petrochemical plants.

[0013] As an optional implementation method, such as Figure 2 As shown, the fiber optic sensing module includes a distributed measurement fiber optic network and a phase-sensitive signal demodulation module. The distributed measurement fiber optic network comprises parallel-deployed measurement fibers and reference fibers, both spirally arranged on the surface of the coke tower. The measurement fibers generate changing optical signals when the coke tower expands and deforms; the reference fibers sense changes in ambient temperature and generate optical signals. The phase-sensitive signal demodulation module demodulates the optical signals from the measurement fibers to obtain the phase changes generated when the coke tower expands and deforms, and, based on the optical signals from the reference fibers, obtains the real-time phase changes caused by the coke tower temperature and the coke tower temperature itself.

[0014] For example, the measuring optical fiber uses high-temperature resistant metal-coated fiber (temperature resistance ≥500℃), spirally wound along the tower (50cm spacing), and fixed to the tower with high-temperature adhesive (500℃ resistant adhesive) to ensure synchronization with structural deformation and real-time sensing of minute strains caused by swelling or thermal stress. Parallel temperature optical fibers, also serving as reference fibers, are suspended without adhesive fixation, or only loosely fixed at certain points with adhesive. This allows for both temperature monitoring and temperature compensation to eliminate common-mode thermal noise, separating temperature from mechanical strain signals.

[0015] Phase-sensitive signal demodulation: A high-frequency pulsed optical signal is emitted into an optical fiber via a laser. As the light wave propagates through the fiber, it interacts with the strain on the tower surface. When the tower expands and deforms, the local strain in the fiber alters the phase characteristics of the optical signal, thereby modulating the phase difference of the backscattered Rayleigh light. A high-sensitivity detector captures these phase changes and converts them into an electrical signal using coherent detection technology. Using time-domain reflectometry (TD-RS) positioning technology, the location of the deformation, D = vt / 2, can be precisely determined based on the round-trip time t of the optical pulse. Here, D represents the distance from the emission point (i.e., the device end) to the deformation point in the fiber, and v is the speed of light in the fiber. Laser parameters: wavelength 1550 nm, pulse width 10 ns, repetition frequency 1 kHz; detector sampling rate 200 MSa / s, phase resolution 0.01 rad.

[0016] "By emitting high-frequency pulsed light signals into an optical fiber via a laser, the light waves interact with the strain on the tower surface as they propagate through the fiber. When the tower expands and deforms, the local strain in the optical fiber changes the phase characteristics of the light signal, thereby modulating the phase difference of the backscattered Rayleigh light." This is the working principle of a phase-sensitive optical time-domain reflectometer (Φ-OTDR). Therefore, this application provides a coke tower bulging deformation monitoring system based on Φ-OTDR.

[0017] As an optional implementation method, temperature interference suppression: combining the periodic temperature changes during the operation of the coke tower (such as cycling from a high temperature of 490℃ to a normal temperature), a dual-fiber reference method and a temperature-strain decoupling algorithm are used to eliminate the interference of thermal expansion on the phase signal, thereby obtaining the phase change caused only by mechanical strain. ).

[0018] Phase-strain mapping: based on the phase change caused by mechanical strain ( ) and fiber axial strain ( The linear relationship between local deformation and strain ( ) is used to quantify the local deformation of the tower. There is a linear relationship between them.

[0019] like Figure 3 As shown, the formula used by the temperature-strain decoupling algorithm to determine the real-time strain of the coke tower is: ; ; In the formula, In response, To measure the phase change of the optical fiber, For the phase change of the reference fiber, The strain phase sensitivity coefficient, The effective sensing length of the optical fiber. To detect the wavelength of light, The refractive index of the optical fiber core. is the optical elastic coefficient of the optical fiber material.

[0020] The range of values ​​for the parameters in the above formula is as follows: =0.26~0.29, =1.45~1.46.

[0021] The formula used to obtain the coke tower temperature via the reference optical fiber is as follows: , This is the temperature phase sensitivity coefficient. This represents the change in ambient temperature. , The temperature coefficient of the refractive index of the optical fiber. .

[0022] As an optional implementation, in establishing a deformation-time series model using the least squares method based on the historical strain and deformation of the coke tower, the data analysis and processing module includes: (1) The relationship between deformation and strain is established as follows: In the formula, for The deformation of the coke tower body at any given time; for Adaptability at all times; As the initial deformation baseline value, The strain-deformation conversion factor. The first-order time drift coefficient, This is the second-order time drift coefficient; For model error term, and It follows a normal distribution with a mean of 0; (2) Based on the historical strain and historical deformation of the coke tower, The optimization objective is to minimize the root mean square value. The least squares method is used to optimize the relationship between deformation and strain. , , , Perform iterative optimization to obtain , , , The value; (3) , , , By substituting the numerical values ​​into the relationship between deformation and strain, a deformation-time series model is obtained.

[0023] The above Related to the material properties of the coke tower body and the location of the measuring optical fiber, it must meet the following requirements: =1 / (E×S), where E is the elastic modulus of the coke tower material and S is the correlation coefficient of the moment of inertia of the tower section at the measurement point.

[0024] As an alternative implementation method, the coke tower deformation data is a non-stationary time series (affected by temperature cycling and material fatigue, with trend term + periodic term + random noise). Traditional LSTM is difficult to accurately capture the abrupt change characteristics of the "bulging deformation acceleration stage". Therefore, an improved bidirectional LSTM is used, combined with sliding window fitting to achieve trend recognition.

[0025] Network structure: The bidirectional long short-term memory network consists of an input layer, a hidden layer, and an output layer connected in sequence; the input layer has 100 neurons; the hidden layer consists of 2 layers, each with 128 neurons; the output layer has 1 neuron; the activation function used is the ReLU activation function.

[0026] The input layer has 100 neurons corresponding to 100 strain data points: matching the sampling frequency and data dimension of coke tower deformation monitoring, ensuring the integrity of input information and avoiding feature loss due to data truncation. Two hidden layers (128 neurons each): compared to a single-hidden-layer LSTM, this enhances the model's ability to extract features from complex non-stationary time-series data (trend term + periodic term + random noise), and can more accurately uncover the deep correlation between strain and deformation. The ReLU activation function replaces the traditional tanh / sigmoid: effectively alleviating the gradient vanishing problem and improving model training stability, especially suitable for the abrupt feature learning needs of the "accelerated deformation stage of bulging and convex deformation".

[0027] For example, the bidirectional long short-term memory network used in this application is an improved bidirectional LSTM. The improved bidirectional LSTM can also enhance the weight of mutation features by introducing an attention mechanism, thereby improving the recognition accuracy in the accelerated stage of bulging deformation.

[0028] As an optional implementation, the data analysis and processing module includes the following functions: (1) Data preprocessing Input data: Real-time deformation data of the coke tower ring body (unit: μm), sampling frequency 1 time / 10min, historical data sample size ≥365 days (covering 180 complete production cycles).

[0029] Denoising: A wavelet thresholding denoising algorithm (db4 wavelet basis, 3 decomposition layers) is used to remove random noise caused by equipment vibration and environmental interference.

[0030] Data standardization: Z-score standardization is used to convert deformed data into a standard sequence with a mean of 0 and a variance of 1, eliminating dimensional differences between different monitoring points. ; in, This is the average of historical data. The standard deviation is the historical data.

[0031] (2) Model training and trend recognition module Sliding window settings: Window size W = 7 days (covering 3 temperature cycles), step size S = 1 day, meaning that the deformation value for the next day is predicted daily based on the standardized deformation data of the previous 7 days. When the predicted deformation value for the next day exceeds the safety threshold, an alarm is issued promptly to prevent the equipment from operating beyond its limits. The window length for the sliding window fitting is determined by calibration using historical deformation abrupt change data, ranging from 5 to 30 time sampling points.

[0032] Acceleration phase determination: Define "deformation acceleration" When three consecutive sliding windows and When the month-on-month growth rate is ≥15%, it is judged as the "accelerated stage of bulging deformation". In the formula, For the first in the sliding window The deformation acceleration of the sun, For the first in the sliding window Day and the The daily deformation difference, For the first in the sliding window Day and the The daily deformation difference, The value is 1.

[0033] (3) Abnormal deformation fluctuation detection (crack initiation early warning sub-algorithm) The deformation sequence of the circumferential weld region was subjected to Fast Fourier Transform (FFT) to extract phase spectrum features: ; in: For frequency, For time series indexing.

[0034] In identifying the accelerated expansion and deformation stage based on the real-time deformation of the coking tower over a preset time period, using a bidirectional long short-term memory network combined with a sliding window fitting method, the execution logic of the data analysis and processing module includes: setting the size and step size of the sliding window; sequentially extracting data for each sliding window from the real-time deformation of the coking tower over the preset time period according to the size and step size of the sliding window; and applying a formula to the data of each sliding window. Obtain the deformation acceleration of each sliding window; when the deformation acceleration of three consecutive sliding windows satisfies and At that time, it was determined to be the stage of accelerated expansion and deformation of the coke tower; among which, The first sliding window The deformation acceleration of the sun.

[0035] As an optional implementation, the data analysis and processing module is further configured to: construct an original deformation sequence from the real-time deformation of the coking tower over a preset time period; perform a Fourier transform on the original deformation sequence to extract phase spectrum features; standardize the phase spectrum features and the original deformation sequence; concatenate the standardized phase spectrum features and the standardized original deformation sequence to obtain a fused feature vector; and input the fused feature vector into a bidirectional long short-term memory network to predict the deformation of the coking tower. The standardization process can employ a min-max standardization method to map the original deformation sequence data and the phase spectrum features to the same numerical range.

[0036] As can be seen, the phase spectrum features include phase angle distribution, phase spectrum peak position, and phase spectrum variance. The phase spectrum features are used to optimize the input feature set of the improved bidirectional LSTM. The fused feature vector is input into the input layer of the improved bidirectional LSTM to improve the model's accuracy in identifying the abrupt changes in the accelerated stage of bulging deformation in the circumferential weld area.

[0037] As an optional implementation, a normal phase fluctuation threshold is set. (Based on 3 times the phase standard deviation of 365 days of historical data, i.e.) When the phase deviation of any two consecutive sampling points At that time, a risk warning for crack initiation was triggered.

[0038] As an optional implementation method, the risk level matrix has three core input parameters, and the risk level determination rules can be quantitatively set and flexibly adjusted according to the actual object. The three core input parameters are shown in Table 1.

[0039] Table 1 Three Core Input Parameters

[0040] The cumulative fatigue damage value is calculated using Miner's linear cumulative damage theory: ; in For the first i Number of cycles at each stress level This represents the fatigue life corresponding to the stress level. The stress level is determined primarily through three methods: finite element analysis to model the structure and calculate the stress distribution in key areas; selecting appropriate stress amplitudes based on the SN curve of the reference material; and statistical analysis of historical stresses measured under actual working conditions using sensors. Determining the number of cycles involves two aspects: design life requirements and load spectrum compilation. When the design life is known, the allowable number of cycles at each stress level can be inferred using Miner's theory. When the design life is unknown, load spectrum compilation can be used, including obtaining the time history of actual or simulated working loads, simplifying variable amplitude loads to constant amplitude stress cycles using the rainflow counting method, statistically counting the number of cycles at each stress level, and statistically processing random loads to extract typical stress levels and cycle numbers.

[0041] Adjustment Mechanism: Basic Parameter Calibration: Before new equipment is put into use, initial parameters are set based on the material factory test report (SN curve) and design documents (number of temperature cycle withstand times); after every 100 temperature cycles, the parameters are corrected using the latest monitoring data. , , Equal threshold.

[0042] Industry adaptability coefficient: For different refinery coke tower specifications (diameter, wall thickness) and operating temperature (400℃~550℃), a correction coefficient k (0.8~1.2) is introduced to adjust the thresholds of δ and a (for example, when the wall thickness increases, k is taken as 1.1 and the threshold is appropriately relaxed).

[0043] As an optional implementation method, taking a coke tower in a certain factory as an example, with a diameter of 8 meters, a height of 25 meters, a wall thickness of 26 millimeters, and a material of chromium-molybdenum steel (such as 15CrMoR), the following early warning rules can be adopted.

[0044] The warning levels in the risk level matrix warning rules include: normal, minor, severe, and urgent. If δ < 100 μm or a < 5 μm / day 2 If N < 300 or D < 0.2, the warning level is judged as normal; where δ is the cumulative deformation, a is the deformation acceleration, N is the number of temperature cycles, and D is the cumulative fatigue damage value. If 100μm≤δ<200μm or 5μm / day 2 ≤a<15μm / day 2 If 300≤N<500 or 0.2≤D<0.4, the warning level is determined to be minor; If 200μm≤δ<350μm or 15μm / day 2 ≤a<30μm / day 2 If 500≤N<800 or 0.4≤D<0.7, the warning level is determined to be severe; If δ≥350μm or a≥30μm / day 2 If N≥800 or D≥0.7, the warning level is determined to be emergency.

[0045] The above early warning rules can be summarized in Table 2.

[0046] Table 2 Risk Level Matrix and Early Warning Rules

[0047] When implementing this method, the process by which the early warning platform determines the crack risk level includes: ① Obtain the real-time deformation of the coke tower during a preset time period from the data analysis and processing module, and obtain the synchronous coke tower temperature from the fiber optic sensing module. ② Determine the working cycle based on the temperature of the coke tower; ③ Determine the cumulative deformation based on the real-time deformation of the coke tower over a preset time period; ④ Based on the real-time deformation of the coke tower over a preset time period, the deformation acceleration is calculated using a bidirectional long short-term memory network combined with a sliding window fitting method; ⑤ Calculate the cumulative fatigue damage value based on the material's SN curve; ⑥ Based on the cumulative deformation, deformation acceleration, working cycles, and cumulative fatigue damage, the crack risk level is determined using the risk level matrix early warning rules.

[0048] Early warning system operation process: Real-time data input: Deformation data and real-time temperature data from 100 monitoring points; Work cycle determination: The work cycle is determined based on temperature, and the work cycle is also input into the system; Model calculation: The deformation value is predicted by the output of the improved bidirectional LSTM, and the anomalous phase fluctuation is calculated. Matrix matching: Substitute δ, a, N, and D into the risk level matrix to determine the warning level; Progressive response: Normal: Maintain regular inspections (once a month); Mild: Increase monitoring frequency (once every 4 hours) and record data trends; Serious: Shut down and inspect the circumferential weld area (within 72 hours) to assess the risk of cracking; Emergency: Stop the machine immediately and initiate the maintenance process (do not continue operation).

[0049] The system of this application includes: a distributed sensing fiber optic network, a phase-sensitive signal demodulation module, a data analysis and processing module, and an intelligent early warning platform.

[0050] 1. Distributed Fiber Optic Sensing Network: The "sensory nerves" of the deformation monitoring system, it utilizes optical fibers as sensors, deployed spirally along the monitored structure. The fiber optics can continuously monitor strain changes along its length, providing high spatial resolution data. Because optical fibers use optical signals for transmission, they are unaffected by electromagnetic interference, ensuring data accuracy.

[0051] 2. Phase-sensitive signal demodulation module: The system's "signal converter" is responsible for converting the optical signal transmitted in the optical fiber into a processable electrical signal. Through techniques such as interferometry, it extracts the phase change in the optical signal, enabling the detection of minute phase changes and thus achieving precise monitoring of minute strains.

[0052] 3. Data Analysis and Processing Module: The "brain" of the system, responsible for processing and analyzing the raw data obtained from the demodulation module, removing noise and outliers, improving data quality, using algorithms and models to conduct in-depth analysis of the data, and storing the processed data in the database for easy subsequent querying and analysis.

[0053] 4. Intelligent Early Warning Platform: The system's "decision center," responsible for real-time monitoring and early warning based on information provided by the data processing module. It displays real-time response data, allowing users to understand the status of the monitored object at any time; based on preset thresholds and algorithms, it promptly issues early warning signals when monitoring data is abnormal; and it provides data analysis reports and early warning history records to help users make informed decisions.

[0054] The technical benefits brought about by this application are mainly reflected in the following aspects: (1) Improved accuracy monitoring capabilities By using fiber optic sensing technology to achieve synchronous and high-precision measurement of deformation and temperature, the problem of data distortion in high-temperature and corrosive environments caused by traditional monitoring methods is solved, providing a reliable basis for equipment condition assessment.

[0055] (2) Intelligent trend prediction The constructed deformation-time series model can identify key trends such as accelerated bulging deformation in advance. By combining the number of temperature cycles and material fatigue curves, it can realize the dynamic prediction of equipment degradation process, transforming passive maintenance into proactive prevention.

[0056] (3) Closed-loop risk management The established four-level early warning matrix (normal / minor / serious / urgent) forms a complete chain of "monitoring-early warning-response". The intelligent platform automatically triggers the graded response mechanism, improving risk management efficiency by more than 40%.

[0057] (4) Extended equipment lifespan Early warning mechanisms based on material fatigue curves can detect potential risks 3-5 production cycles in advance, reducing unplanned equipment downtime by 60% and significantly extending the service life of key components.

[0058] (5) Scientific management decision-making The combination of historical data and real-time monitoring provides data support for optimizing equipment maintenance strategies, reducing maintenance costs by 30% while ensuring production continuity.

[0059] (6) Reduced monitoring costs The optical fiber can withstand a temperature range of -50℃ to 600℃ and has a lifespan of over 10 years; a single system can cover towers over 30 meters high, and the overall cost is reduced by 60% compared to traditional methods.

[0060] Based on the same inventive concept, this application also provides a method for real-time monitoring and graded early warning of coke tower bulging and deformation applied to the system described above. The solution provided by this method is similar to the solution described in the system above. Therefore, the specific limitations of one or more embodiments of the real-time monitoring and graded early warning method for coke tower bulging and deformation provided below can be found in the system limitations above, and will not be repeated here.

[0061] In one exemplary embodiment, such as Figure 4 As shown, a method for real-time monitoring and graded early warning of bulging deformation in a coke tower is provided, including the following steps 101 to 108.

[0062] Step 101: Based on the historical strain and deformation of the coke tower, establish a deformation-time series model using the least squares method.

[0063] Step 102: Obtain the real-time phase change caused by the expansion and deformation of the coke tower, the real-time phase change caused by the temperature of the coke tower, and the temperature of the coke tower.

[0064] Step 103: Based on the real-time phase change generated during expansion and deformation and the real-time phase change corresponding to the temperature of the coke tower, the temperature-strain decoupling algorithm is used to determine the real-time strain of the coke tower.

[0065] Step 104: Based on the real-time strain of the coking tower, obtain the real-time deformation of the coking tower using a deformation-time series model.

[0066] Step 105: Based on the real-time deformation of the coke tower over a preset time period, the accelerated expansion and deformation stage is identified by using a bidirectional long short-term memory network combined with a sliding window fitting method.

[0067] Step 106: When the phase change during the expansion deformation acceleration stage is greater than the phase fluctuation threshold, a crack initiation risk warning is triggered.

[0068] Step 107: Construct risk level matrix early warning rules.

[0069] Step 108: Based on the cumulative deformation and fatigue damage accumulated during the periodic operation of the coke tower, determine the crack risk level using the risk level matrix early warning rules.

[0070] This application discloses a real-time monitoring system and graded early warning method for bulging deformation of coke towers. It can achieve accurate measurement of deformation and temperature, and at the same time construct a deformation-time series model to accurately identify key trends such as the accelerated stage of bulging deformation. It establishes a risk level relationship matrix by combining the number of temperature cycles and material fatigue curves, and realizes graded response through an intelligent early warning platform, forming a closed-loop management from trend monitoring to risk control, thereby significantly improving the timeliness and reliability of safety monitoring of petrochemical equipment.

[0071] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0072] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A real-time monitoring and graded early warning system for the bulging and deformation of a coke tower, characterized in that, include: Fiber optic sensing module, data analysis and processing module, and early warning platform; The fiber optic sensing module uses fiber optic cables spirally laid on the surface of the coke tower to measure the real-time phase change caused by the expansion and deformation of the coke tower, the real-time phase change caused by the temperature of the coke tower, and the temperature of the coke tower in a fiber optic sensing manner. The data analysis and processing module is used to establish a deformation-time series model based on the historical strain and deformation of the coke tower using the least squares method. The data analysis and processing module is also used to determine the real-time strain of the coke tower by using a temperature-strain decoupling algorithm based on the real-time phase changes generated during expansion and deformation and the real-time phase changes corresponding to the temperature of the coke tower; based on the real-time strain of the coke tower, the module uses a deformation-time series model to obtain the real-time deformation of the coke tower; and based on the real-time deformation of the coke tower over a preset time period, the module uses a bidirectional long short-term memory network combined with a sliding window fitting method to identify the accelerated expansion and deformation stage. When the phase change during the accelerated expansion and deformation phase exceeds the phase fluctuation threshold, a crack initiation risk warning is triggered. The early warning platform is used to construct risk level matrix early warning rules; Based on the cumulative deformation and fatigue damage accumulated during the periodic operation of the coke tower, the crack risk level is determined using the risk level matrix early warning rules.

2. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, The fiber optic sensing module includes: a distributed measurement fiber optic network and a phase-sensitive signal demodulation module; The distributed measurement fiber optic network includes parallel-deployed measurement fibers and reference fibers, both of which are spirally arranged on the surface of the coke tower. The measurement fibers are used to generate changing optical signals when the coke tower expands and deforms; the reference fibers are used to sense changes in ambient temperature and generate optical signals. The phase-sensitive signal demodulation module is used to demodulate the optical signal of the measurement fiber to obtain the phase change generated when the coke tower expands and deforms, and to obtain the real-time phase change caused by the temperature of the coke tower and the temperature of the coke tower based on the optical signal of the reference fiber.

3. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, The formula used by the temperature-strain decoupling algorithm to determine the real-time strain of the coke tower is: ； ； In the formula, In response, To measure the phase change of the optical fiber, For the phase change of the reference fiber, The strain phase sensitivity coefficient, The effective sensing length of the optical fiber. To detect the wavelength of light, The refractive index of the optical fiber core. is the optical elastic coefficient of the optical fiber material.

4. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, In establishing a deformation-time series model using the least squares method based on the historical strain and deformation of the coking tower, the data analysis and processing module includes: The relationship between deformation and strain is established as follows: In the formula, for Deformation of the coke tower body at any given time; for Adaptability at all times; As the initial deformation baseline value, The strain-deformation conversion factor is... The first-order time drift coefficient, This is the second-order time drift coefficient; For model error term, and It follows a normal distribution with a mean of 0; Based on the historical strain and deformation of the coke tower, The optimization objective is to minimize the root mean square value. The least squares method is used to optimize the relationship between deformation and strain. , , , Perform iterative optimization to obtain , , , The value; Will , , , By substituting the numerical values ​​into the relationship between deformation and strain, a deformation-time series model is obtained.

5. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, The bidirectional long short-term memory network includes an input layer, a hidden layer, and an output layer connected in sequence. The input layer has 100 neurons; The hidden layer consists of two layers, each with 128 neurons; The output layer has one neuron; The ReLU activation function is used.

6. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, In identifying the accelerated expansion and deformation stage based on the real-time deformation of the coke tower over a preset time period, using a bidirectional long short-term memory network combined with a sliding window fitting method, the data analysis and processing module specifically includes: Set the size and step size of the sliding window; According to the size and step size of the sliding window, the data of each sliding window is extracted sequentially from the real-time deformation of the coke tower over a preset time period; For the data in each sliding window, use the formula The deformation acceleration of each sliding window is obtained; where, For the first in the sliding window The deformation acceleration of the sun, For the first in the sliding window Day and the The daily deformation difference, For the first in the sliding window Day and the The daily deformation difference, The value is 1; When the deformation acceleration of three consecutive sliding windows satisfies and At that time, it was determined to be the stage of accelerated expansion and deformation of the coke tower; among which, The first sliding window The deformation acceleration of the sun.

7. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, The data analysis and processing module is also used for: The real-time deformation of the coke tower over a preset time period is used to construct the original deformation sequence. Perform a Fourier transform on the original deformation sequence to extract phase spectrum features; The phase spectrum features and the original deformation sequence are standardized. By splicing the standardized phase spectrum features and the standardized original deformation sequence, a fused feature vector is obtained. The fused feature vector is input into a bidirectional long short-term memory network to predict the deformation of the coke tower.

8. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 1, characterized in that, The warning levels in the risk level matrix warning rules include: normal, minor, severe, and urgent. If δ < 100 μm or a < 5 μm / day 2 If N < 300 or D < 0.2, the warning level is judged as normal; where δ is the cumulative deformation, a is the deformation acceleration, N is the number of temperature cycles, and D is the cumulative fatigue damage value. If 100μm≤δ<200μm or 5μm / day 2 ≤a<15μm / day 2 If 300≤N<500 or 0.2≤D<0.4, the warning level is determined to be minor; If 200μm≤δ<350μm or 15μm / day 2 ≤a<30μm / day 2 If 500≤N<800 or 0.4≤D<0.7, the warning level is determined to be severe; If δ≥350μm or a≥30μm / day 2 If N≥800 or D≥0.7, the warning level is determined to be emergency.

9. The real-time monitoring and graded early warning system for coke tower bulging and deformation according to claim 8, characterized in that, In determining the crack risk level based on the cumulative deformation and fatigue damage accumulated during the periodic operation of the coking tower, and using a risk level matrix early warning rule, the early warning platform includes: The data analysis and processing module obtains the real-time deformation of the coke tower over a preset time period, and the fiber optic sensing module obtains the synchronous coke tower temperature. The working cycle is determined based on the temperature of the coke tower. The cumulative deformation is determined based on the real-time deformation of the coke tower over a preset time period; Based on the real-time deformation of the coke tower over a preset time period, the deformation acceleration is calculated by using a bidirectional long short-term memory network combined with a sliding window fitting method. Calculation of cumulative fatigue damage based on material SN curves; Based on the cumulative deformation, deformation acceleration, working cycles, and cumulative fatigue damage, the crack risk level is determined using the risk level matrix early warning rules.

10. A method for real-time monitoring and graded early warning of bulging deformation in a coke tower, characterized in that, include: Based on the historical strain and deformation of the coke tower, a deformation-time series model is established using the least squares method; The system acquires the real-time phase changes caused by the expansion and deformation of the coke tower, the real-time phase changes caused by the temperature of the coke tower, and the temperature of the coke tower. Based on the real-time phase change generated during expansion and deformation and the real-time phase change corresponding to the temperature of the coke tower, a temperature-strain decoupling algorithm is used to determine the real-time strain of the coke tower. Based on the real-time strain of the coking tower, the real-time deformation of the coking tower is obtained using a deformation-time series model; Based on the real-time deformation of the coke tower over a preset time period, a bidirectional long short-term memory network combined with a sliding window fitting method is used to identify the accelerated expansion and deformation stage. When the phase change during the accelerated expansion and deformation phase exceeds the phase fluctuation threshold, a crack initiation risk warning is triggered. Construct risk level matrix early warning rules; Based on the cumulative deformation and fatigue damage accumulated during the periodic operation of the coking tower, the crack risk level is determined using the risk level matrix early warning rules.

Images