Corrosion information management system for refining and chemical field equipment

Abstract

The present application belongs to the technical field of digital twinning and database structure, and particularly relates to a device corrosion information management system in the refining field. The system comprises a data access unit, a digital twinning construction unit, a warning unit and a flow unit. The data access unit generates fused time series data sets by interpolating and aligning continuous monitoring signals and discrete detection data, and binds process and test characteristics according to spatial topological relations to form a multi-dimensional feature vector. The twinning construction unit extracts two-dimensional logical relations and three-dimensional physical coordinates to establish a correlation mapping matrix and generate a three-dimensional digital twin. The warning unit outputs a warning condition by matching a mechanism model calculation and a case library. The flow unit generates a structured work order and generates model adjustment parameters to update model weights according to the difference value of the operation data after disposal. The system presents a physical location with a working condition parameter combination in a three-dimensional space, and corrects corrosion grade evaluation and life prediction parameters according to actual disposal data.

Description

Technical Field

[0001] This invention belongs to the field of digital twin and database structure technology, specifically relating to a corrosion information management system for equipment in the refining and chemical industry. Background Technology

[0002] Existing conventional solutions for equipment corrosion management in the refining and chemical industry establish independent relational databases to store continuous thickness measurement data collected by online monitoring systems, discrete thickness measurement data entered offline by ultrasonic testing, process operating parameters recorded by distributed control systems, and laboratory analysis data. In terms of equipment spatial display, the conventional solution stores and displays two-dimensional piping and instrumentation diagrams separately from empty equipment views built using 3D modeling software. The two-dimensional diagrams are used to express the logical relationships between equipment nodes and pipeline connections, while the 3D model is used to view the spatial coordinates and geometric dimensions of pipe sections. Regarding risk identification, the conventional solution sets independent fixed thresholds for various monitoring data; when a single data point exceeds the corresponding threshold, an alarm is triggered. For hazard handling, after reviewing the alarm records, managers manually fill in the investigation tasks in the office system, assign responsibilities, and track the execution results.

[0003] Based on the specific implementation of the aforementioned existing technologies, the continuously collected online monitoring data and the non-continuously entered offline detection data are separated on the time axis. The logical topological relationship in the two-dimensional drawings and the physical spatial coordinates in the three-dimensional model are independent of each other. As a result, after receiving a threshold alarm for a single data point, the system cannot integrate and correlate the discrete time series data, process parameters, and equipment physical spatial coordinates. Consequently, it is impossible to accurately trace the specific physical location of the corrosion anomaly and the corresponding combination of operating parameters in three-dimensional space. This makes the subsequent generated disposal tasks lack accurate spatial positioning basis and operating condition targeting conditions. Summary of the Invention

[0004] The purpose of this invention is to provide a corrosion information management system for equipment in the refining and chemical industry, which can solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A corrosion information management system for equipment in the refining and chemical industry includes a multi-source heterogeneous data access unit, a digital twin construction unit, an intelligent analysis and early warning unit, and a closed-loop task flow unit. The multi-source heterogeneous data access unit acquires online monitoring data, offline detection data, process operating parameters, and laboratory analysis data of the refining and chemical equipment. The digital twin construction unit integrates equipment maps, piping and instrumentation diagrams, and equipment aerial views to construct a three-dimensional digital twin of the refining and chemical equipment. The intelligent analysis and early warning unit performs fusion analysis on the data acquired by the multi-source heterogeneous data access unit, combining it with a pre-set corrosion mechanism model and case library to output corrosion level assessment results and remaining life prediction results. The closed-loop task flow unit automatically generates a structured work order containing investigation, detection, and maintenance nodes upon receiving the abnormal assessment results output by the intelligent analysis and early warning unit. After obtaining the feedback data from the maintenance execution, it inputs the post-maintenance equipment operating data into the intelligent analysis and early warning unit for effect verification and model parameter updates. The three-dimensional digital twin receives the corrosion level assessment results and performs spatial positioning and dynamic mapping.

[0007] Preferably, the multi-source heterogeneous data access unit includes a time-series alignment subunit and a feature extraction subunit. The time-series alignment subunit is used to perform time-axis interpolation alignment on the continuous time-series signals contained in the online monitoring data and the non-continuous timestamp discrete data contained in the offline detection data, based on a set time window, to generate a fused time-series dataset. The feature extraction subunit is used to extract static features of medium flow rate, temperature, and pressure from the process operating parameters, extract dynamic features of corrosive medium concentration and dew point temperature from the laboratory analysis data, and bind the static features and the dynamic features to the corresponding time nodes of the fused time-series dataset according to the equipment spatial topology relationship, forming a multi-dimensional feature vector with spatial topology identifier.

[0008] Preferably, the digital twin construction unit includes a primitive analysis subunit and a topology mapping subunit. The primitive analysis subunit is used to perform primitive recognition on the piping instrumentation diagram, extract the logical relationships between equipment nodes, pipeline connections, and valves, perform three-dimensional geometric reconstruction on the equipment empty view, and extract the spatial coordinates of pipe segments and the geometric dimensions of pipe fittings. The topology mapping subunit is used to bind the equipment nodes and pipeline connections extracted by primitive recognition with the spatial coordinates of pipe segments extracted by the three-dimensional geometric reconstruction, establish an association mapping matrix containing two-dimensional logical topological relationships and three-dimensional physical spatial coordinates, and project the logical relationships between equipment nodes, pipeline connections, and valves onto the device map based on the association mapping matrix to generate the three-dimensional digital twin.

[0009] Preferably, the intelligent analysis and early warning unit includes a mechanism matching subunit and an association early warning subunit. The mechanism matching subunit is equipped with multiple types of corrosion mechanism models, which are used to input the data obtained by the multi-source heterogeneous data access unit into each corrosion mechanism model for forward calculation and output the corrosion rate fitting curve corresponding to each corrosion mechanism model. The association early warning subunit is used to obtain the deviation value between the corrosion rate fitting curve and the preset benchmark corrosion rate curve. When the deviation value exceeds the set deviation threshold, the unit retrieves the historical case set in the case library that matches the current equipment type and process parameters, and extracts the characteristic operating condition parameters in the historical case set as the association early warning condition output.

[0010] Preferably, the closed-loop task flow unit includes a work order structured subunit and a state machine flow rotor unit; the work order structured subunit is used to parse the equipment identifier, risk level, and corrosion type fields in the anomaly assessment result, match the preset task handling template, and fill in to generate a structured work order containing investigation task nodes, detection task nodes, and maintenance task nodes; the state machine flow rotor unit is used to define multiple discrete states of the structured work order, and when the execution result data of the current task node is received, the structured work order is transferred from the current discrete state to the discrete state corresponding to the next task node, until the discrete state flow corresponding to all task nodes is completed, and an archive identifier is generated.

[0011] Preferably, the closed-loop task flow unit further includes a verification feedback subunit; the verification feedback subunit is used to, upon receiving the archive identifier, retrieve the processed equipment operation data of the equipment identifier within a preset verification period, calculate the actual corrosion rate index corresponding to the processed equipment operation data; calculate the difference between the actual corrosion rate index and the predicted corrosion rate index output by the intelligent analysis and early warning unit when generating the structural chemical order, and generate model adjustment parameters based on the difference calculation result; input the model adjustment parameters to the intelligent analysis and early warning unit to update the weight coefficient of the corrosion mechanism model and the matching priority of the case library.

[0012] Preferably, the feature extraction subunit further includes a waveform denoising module and a defect identification module; the waveform denoising module is used to perform wavelet packet decomposition on the ultrasonic thickness measurement waveform signal in the online monitoring data, remove frequency band components below a set energy threshold, reconstruct the signal, and output a denoised waveform signal; the defect identification module is used to extract the time-domain envelope and frequency-domain dominant frequency offset of the denoised waveform signal, input the time-domain envelope and the frequency-domain dominant frequency offset into a pre-trained anomaly classifier, output the defect type label corresponding to the discrete data with discontinuous timestamps, and append the defect type label to the multidimensional feature vector.

[0013] Preferably, the topology mapping subunit further includes a hierarchical sectioning module and a dynamic rendering module; the hierarchical sectioning module is used to generate a sectioning plane along the normal vector direction of the pipe section of the three-dimensional digital twin in response to the received external sectioning command, and extract the cross-sectional contour line where the sectioning plane intersects with the geometric dimensions of the pipe fitting; the dynamic rendering module is used to obtain the corrosion level assessment result output by the intelligent analysis and early warning unit, determine the rendering color of the cross-sectional contour line according to the preset level and color mapping table, fill the area surrounded by the cross-sectional contour line with the rendering color, and superimpose the cross-sectional contour line and the rendering color on the three-dimensional digital twin.

[0014] Preferably, the associated early warning subunit further includes a multi-parameter coupling analysis module; the multi-parameter coupling analysis module is used to obtain the rate of change of heat transfer coefficient and the change of medium flow rate in the process operating parameters when the equipment type is a heat exchanger, and to obtain the tube-side pressure difference signal in the online monitoring data; input the rate of change of heat transfer coefficient, the change of medium flow rate and the tube-side pressure difference signal into a preset scaling and scouring correlation matrix for matrix multiplication operation, and output the scaling tendency index and the scouring corrosion tendency index; when the scaling tendency index or the scouring corrosion tendency index exceeds the corresponding set boundary, the associated early warning condition is triggered to output.

[0015] Preferably, the structured work order subunit further includes an external corrosion analysis module and a special ledger matching module; the external corrosion analysis module is used to extract the equipment external surface temperature distribution data and insulation layer damage area data from the anomaly assessment results when the corrosion type field is corrosion under the insulation layer; the special ledger matching module is used to retrieve the external corrosion engineering ledger based on the equipment identifier, obtain the insulation material type and last replacement time of the corresponding pipeline, and combine the insulation material type, the last replacement time, the equipment external surface temperature distribution data and the insulation layer damage area data to fill the investigation task node, generating a structured work order containing a special inspection instruction for the insulation layer.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0017] 1. The time-series alignment subunit uses a time window as a reference to interpolate and align continuous time-series signals and discrete data with discontinuous timestamps to generate a fused time-series dataset. The primitive analysis subunit extracts the logical relationships of equipment nodes in the two-dimensional drawings and the physical space coordinates in the three-dimensional empty view. The topology mapping subunit binds the two to establish an association mapping matrix and projects it onto the equipment map to generate a three-dimensional digital twin. The feature extraction subunit binds the static features of process operation parameters and the dynamic features of laboratory analysis data to the corresponding time nodes of the fused time-series dataset according to the equipment spatial topology relationship. The multi-dimensional feature vectors are unified under the association mapping matrix. The three-dimensional digital twin receives the corrosion level assessment results for spatial positioning and dynamic mapping, presenting a specific physical location with a combination of operating parameters in three-dimensional space.

[0018] 2. The mechanism matching subunit inputs the fused data into a multi-type corrosion mechanism model for forward calculation and outputs a corrosion rate fitting curve. The correlation early warning subunit retrieves the case library and outputs correlation early warning conditions when the deviation between the fitting curve and the benchmark curve exceeds a threshold. The work order structuring subunit parses the fields in the anomaly assessment results, matches them with the task handling template, and fills in the template to generate a structured work order containing investigation, detection, and maintenance nodes. The state machine flow rotor unit defines discrete states and controls state transitions. After receiving the archiving identifier, the verification feedback subunit retrieves the equipment operation data within the verification period to calculate the actual corrosion rate index, calculates the difference between the actual corrosion rate index and the predicted corrosion rate index to generate model adjustment parameters, inputs the model adjustment parameters to update the weight coefficients of the corrosion mechanism model and the matching priority of the case library, and corrects the output parameters of the corrosion level assessment results and remaining life prediction results based on the actual handling data. Attached Figure Description

[0019] Figure 1 A flowchart illustrating the overall workflow of the equipment corrosion information management system in the refining and chemical industry provided in this embodiment of the invention;

[0020] Figure 2 This is a flowchart of multi-source heterogeneous data temporal alignment and feature extraction provided in an embodiment of the present invention;

[0021] Figure 3 A flowchart for the construction and visualization of a three-dimensional digital twin provided in this embodiment of the invention;

[0022] Figure 4 A flowchart for intelligent corrosion risk analysis and correlation early warning provided in this embodiment of the invention;

[0023] Figure 5 This is a flowchart of the structural chemical unit generation and state machine transition provided in an embodiment of the present invention;

[0024] Figure 6 The flowchart for verification feedback and model self-optimization provided in the embodiments of the present invention. Detailed Implementation

[0025] refer to Figure 1 In one embodiment, the corrosion information management system for equipment in the refining and chemical industry includes a multi-source heterogeneous data access unit, a digital twin construction unit, an intelligent analysis and early warning unit, and a closed-loop task flow unit. In this embodiment, the multi-source heterogeneous data access unit connects to multiple existing business systems and monitoring terminals of the refining and chemical plant through standardized interfaces and protocols to acquire corrosion-related data across all dimensions of the refining and chemical equipment, specifically including online monitoring data, offline detection data, process operating parameters, and laboratory analysis data. The online monitoring data originates from distributed online monitoring terminals deployed at key parts of the refining and chemical equipment, covering continuous time-series data from fixed-point ultrasonic thickness measurement, online corrosion potential monitoring data, and distributed monitoring data of the equipment's outer wall temperature. Real-time data access is achieved through the OPCUA and ModbusTCP protocols, with the data update frequency consistent with the terminal sampling frequency. The offline detection data originates from the results of periodic offline inspections performed by on-site inspection personnel, covering discrete data from manual ultrasonic thickness measurement, magnetic particle testing results, penetrant testing results, and infrared thermal imaging detection data. Batch data entry and access are completed through a structured form interface. The data update cycle is matched with the on-site testing operation cycle; the process operation parameters come from the distributed control system and programmable logic controller system of the refining and chemical plant, covering the temperature, pressure, flow rate, valve opening, pump operating frequency, and pressure difference between the tube side and shell side of the heat exchanger. The data is synchronized through an industrial real-time database interface; the laboratory analysis data comes from the laboratory information management system of the refining and chemical enterprise, covering the sulfur content, chloride ion content, pH value, water content, hydrogen sulfide content, crude oil acid value, and the calculated dew point temperature of the medium. The laboratory analysis results are automatically synchronized through a standardized data interface.

[0026] In this embodiment, the digital twin construction unit is used to integrate the plant map, piping and instrumentation diagram, and equipment aerial view to construct a three-dimensional digital twin of the refining and chemical equipment. The plant map is a geographic information file showing the overall layout of the refining and chemical plant, including the boundary coordinates of each production unit, the planar positioning information of equipment and pipe racks, and the plant's coordinate reference system. The piping and instrumentation diagram is a two-dimensional process drawing file conforming to petrochemical industry drawing standards, including equipment tag numbers, pipe numbers, valves, logical connections of instrument components, and process control logic. The equipment aerial view is a three-dimensional model file containing the geometric information of equipment and pipe sections, including the direction of pipe centerlines, the spatial location of pipe fittings, and the geometric dimensions of pipes and equipment. The digital twin construction unit first completes the format parsing and coordinate system unification of the three types of drawing files, unifying the planar projection coordinates of the plant map, the relative coordinates of the piping and instrumentation diagram, and the local model coordinates of the equipment aerial view into the same plant world coordinate system through a coordinate transformation algorithm. Then, it completes the fusion mapping of two-dimensional logical information and three-dimensional physical information, ultimately generating a three-dimensional digital twin with complete process logic attributes, three-dimensional spatial coordinates, and overall plant planar positioning information.

[0027] In this embodiment, the intelligent analysis and early warning unit is used to fuse and analyze the data acquired by the multi-source heterogeneous data access unit. Combining a pre-set corrosion mechanism model and a case library, it outputs corrosion level assessment results and remaining service life prediction results. The pre-set corrosion mechanism model is built upon the fundamental theories of corrosion electrochemistry, fluid mechanics, and heat transfer, covering common corrosion types in the refining and chemical industry. Each model corresponds to specific input parameters and output results. The case library is a pre-constructed dataset of corrosion cases in the refining and chemical industry. Each case entry includes equipment type, equipment material, process operating parameters, corrosion type, corrosion rate, hazard mitigation plan, and post-treatment effect tracking data. The intelligent analysis and early warning unit first performs standardized preprocessing on the accessed multi-source data, completing outlier removal, missing value filling, and data normalization. Then, it inputs the preprocessed data into the corresponding type of corrosion mechanism model for calculation, obtaining the corrosion rate calculation results under the corresponding operating conditions. Combining the equipment's design wall thickness and historical wall thickness detection data, it completes the corrosion level assessment and remaining service life prediction. The corrosion level assessment is completed according to the preset grading rules, which are consistent with the industry standards for corrosion management of equipment in the petrochemical industry. The remaining life prediction is calculated based on the remaining allowable wall thickness of the equipment and the calculated corrosion rate.

[0028] In this embodiment, the closed-loop task flow unit automatically generates a structural work order containing investigation, detection, and repair nodes upon receiving the anomaly assessment result from the intelligent analysis and early warning unit. After obtaining the feedback data on the handling execution, it inputs the equipment operation data after handling into the intelligent analysis and early warning unit for effect verification and model parameter updates. The anomaly assessment result is an assessment result where the corrosion level exceeds a preset safety threshold or the remaining lifespan is lower than a preset minimum service period. Upon receiving this result, the closed-loop task flow unit automatically parses the unique equipment identifier, corrosion location, corrosion type, and risk level information contained in the result, matches it with a preset standardized task handling template, and automatically fills in and generates a structural work order. The structural work order includes investigation task nodes, detection task nodes, and repair task nodes. Each node corresponds to a clear task content, execution standard, responsible position, completion deadline, and acceptance requirements. During the structural work order flow process, the closed-loop task flow unit receives the execution feedback data of each task node in real time, including on-site investigation records, detection reports, repair construction records, and acceptance confirmation documents. After all task nodes are completed and pass acceptance, the work order is archived. After the work order is archived, the closed-loop task flow unit automatically retrieves the post-treatment operation data of the equipment within the preset verification period, including online monitoring data, process operation parameters, and newly added offline detection data. This data is then input into the intelligent analysis and early warning unit to verify the treatment effect. At the same time, based on the actual post-treatment operation data, adjustment instructions for model parameters are generated to update and optimize the matching rules between the corrosion mechanism model and the case library.

[0029] In this embodiment, a 3D digital twin is used to receive corrosion level assessment results and perform spatial positioning and dynamic mapping. Each device, pipe segment, and fitting element in the 3D digital twin has a unique spatial topology identifier, which corresponds to the device identifier in the multi-source data and corrosion assessment results. After receiving the corrosion level assessment results, the 3D digital twin locates the physical location of the device and pipe segment corresponding to the corrosion anomaly in the 3D scene based on the correspondence between the device identifier and the spatial topology identifier. At the same time, the assessment results of corrosion level, corrosion rate, remaining wall thickness, and remaining life are dynamically overlaid and displayed at the corresponding positions in the 3D scene in the form of color mapping, numerical labels, and highlighting, realizing the accurate positioning and intuitive presentation of corrosion anomaly information in 3D physical space.

[0030] Table 1. Specifications for Multi-Source Heterogeneous Data Access for Refining and Chemical Equipment

[0031]

[0032] This table clarifies the sources, core parameter ranges, access methods, and update rules for various types of access data, providing a standardized basis for the standardized access and unified processing of multi-source heterogeneous data, and ensuring that data from different sources and in different formats can be uniformly received and processed by the system.

[0033] This embodiment achieves unified access to all dimensions of corrosion-related data for refining and chemical equipment through a multi-source heterogeneous data access unit, realizes the fusion modeling of two-dimensional process logic and three-dimensional physical space through a digital twin construction unit, realizes intelligent assessment of corrosion risk based on corrosion mechanism and historical cases through an intelligent analysis and early warning unit, realizes closed-loop management of the entire process of corrosion hazard handling and model self-optimization through a closed-loop task flow unit, and realizes spatial positioning and visualization mapping of corrosion information through a three-dimensional digital twin. It solves the problems of data time dimension separation and the independence of two-dimensional logic and three-dimensional spatial information in the prior art, and realizes unified management of corrosion information of refining and chemical equipment throughout the entire process and in all dimensions.

[0034] refer to Figure 2 In another embodiment, the multi-source heterogeneous data access unit includes a time-series alignment subunit and a feature extraction subunit. The time-series alignment subunit is used to perform time-axis interpolation alignment on the continuous time-series signals contained in the online monitoring data and the discrete non-continuous timestamp data contained in the offline detection data, based on a set time window, to generate a fused time-series dataset. The set time window is a configurable fixed time step, determined based on the process fluctuation characteristics of the refining equipment and the sampling characteristics of the monitoring data. In this embodiment, the set time window is 1 hour, meaning a unified target time node is generated on the time axis at 1-hour intervals. The online monitoring data consists of continuous time-series signals with equal time intervals, whose sampling interval is less than the set time window. The time-series alignment subunit first downsamples the continuous time-series signals, and then calculates the statistical characteristic values ​​of the continuous signals within each time window, including the mean, maximum, minimum, and standard deviation, to generate continuous time-series feature data corresponding to the target time node. Offline detection data is non-continuous discrete data, with corresponding data values ​​existing only at the time point when the detection operation is completed. The time interval between adjacent data points is much larger than the set time window. The time alignment sub-unit uses a linear interpolation algorithm to perform interpolation processing on the time axis of the discrete offline detection data, generating interpolated data corresponding to the target time node.

[0035] The calculation process for linear interpolation is achieved through the following formula:

[0036]

[0037] in, For the interpolation target time node, and To detect two adjacent timestamps in discrete data offline, and satisfy the following conditions: , and for and The offline detection data value corresponding to the time. For the target time node The corresponding interpolation calculation results.

[0038] The temporal alignment subunit concatenates the processed continuous temporal feature data with the interpolated discrete data according to a unified target time node. Each target time node corresponds to a complete set of online monitoring and offline detection data, generating a unified fused temporal dataset with a unified time axis.

[0039] In this embodiment, the feature extraction subunit extracts static features of medium flow rate, temperature, and pressure from process operating parameters, and dynamic features of corrosive medium concentration and dew point temperature from laboratory analysis data. The static and dynamic features are then bound to the corresponding time nodes in the fused time-series dataset according to the equipment spatial topology, forming a multi-dimensional feature vector with a spatial topology identifier. The static features are statistical characteristics of process operating parameters within a defined time window, including the mean, maximum, minimum, and fluctuation amplitude of medium flow rate, temperature, and pressure within the time window, used to characterize the stable process conditions of the equipment within the corresponding time window. The dynamic features are the change characteristics between two adjacent laboratory analysis data points, including the rate of change of corrosive medium concentration and the trend of dew point temperature change, used to characterize the dynamic changes in the corrosive properties of the medium. After extracting the static and dynamic features, the feature extraction subunit binds the extracted features to the data of the corresponding equipment and time node in the fused time-series dataset based on the unique identifiers of the equipment and pipe segments. Each time node's feature data corresponds to a unique equipment spatial topology identifier, which characterizes the physical location of the equipment or pipe segment corresponding to that set of data in three-dimensional space.

[0040] The construction of multidimensional feature vectors is achieved through the following formula:

[0041]

[0042] in, For the target time node The corresponding multidimensional feature vector, to for The eigenvalues ​​of the static and dynamic features corresponding to each time step. The total dimension of the features This is a spatial topology identifier for the device, a unique string encoding used to identify the physical spatial location of the device or pipe segment corresponding to this feature vector.

[0043] In this embodiment, the feature extraction subunit further includes a waveform denoising module and a defect identification module. The waveform denoising module performs wavelet packet decomposition on the ultrasonic thickness measurement waveform signal in the online monitoring data, removes frequency band components below a set energy threshold, reconstructs the signal, and outputs a denoised waveform signal. The ultrasonic thickness measurement waveform signal is the original ultrasonic echo time-domain signal collected by the online monitoring terminal. The signal contains noise components caused by on-site electromagnetic interference and medium flow vibration. The waveform denoising module first performs multi-level wavelet packet decomposition on the original waveform signal, decomposing the original signal into wavelet packet coefficient components of different frequency bands. The wavelet packet decomposition process is implemented by the following formula:

[0044]

[0045] in, The scale for wavelet packet decomposition. The translation coefficient is... This represents the frequency band number of the wavelet packet. For wavelet packet basis functions, For the first Decomposition scale, first Wavelet packet coefficients corresponding to the frequency band.

[0046] After wavelet packet decomposition, the waveform denoising module calculates the signal energy of each frequency band component, compares the energy value of each frequency band with the set energy threshold, removes frequency band components with energy values ​​lower than the set energy threshold, retains high-frequency and low-frequency components containing effective ultrasonic echo information, and then reconstructs the retained wavelet packet coefficients to generate a denoised waveform signal after removing noise components.

[0047] Table 2 Comparison of noise reduction effects of ultrasonic thickness measurement waveforms at different resolution scales

[0048]

[0049] This table compares the waveform noise reduction effects under different wavelet packet decomposition scales, providing a quantitative basis for the selection of decomposition scale. In this embodiment, the decomposition scale is selected as 4, which ensures the correlation between the reconstructed signal and the original effective signal while ensuring the signal-to-noise ratio improvement effect, and avoids the loss of effective information.

[0050] In this embodiment, the defect identification module extracts the time-domain envelope and frequency-domain dominant frequency offset of the denoised waveform signal. It inputs these parameters into a pre-trained anomaly classifier, outputs defect type labels corresponding to the discrete data with discontinuous timestamps, and appends these labels to the multi-dimensional feature vector. The time-domain envelope is extracted using a Hilbert transform, and the calculation process is implemented using the following formula:

[0051] in, The time-domain envelope of the noise-reduced waveform signal. This is the ultrasonic signal after noise reduction. For Hilbert transform operators, It is the imaginary unit.

[0052] The extraction process of the frequency domain dominant frequency offset is as follows: A fast Fourier transform is performed on the denoised waveform signal to obtain the signal's frequency domain distribution curve. The frequency corresponding to the peak value in the frequency domain distribution curve is extracted as the signal's dominant frequency. The difference between this dominant frequency and the dominant frequency of a preset defect-free standard waveform is calculated to obtain the frequency domain dominant frequency offset. The anomaly classifier is a classification model trained using supervised learning. The training samples are a historical ultrasonic thickness measurement waveform dataset with defect type labels, including defect-free, uniform corrosion, pitting, cracks, and wall thickness exceeding tolerance. The defect identification module inputs the extracted temporal envelope features and the frequency domain dominant frequency offset into the anomaly classifier. The model outputs the corresponding defect type label, which is then appended to the multi-dimensional feature vector of the corresponding time node and device, enriching the dimension and information content of the feature vector.

[0053] This embodiment achieves unified alignment of continuous online monitoring data and non-continuous offline detection data on the time axis through a time alignment subunit, solving the problem of time dimension fragmentation of data from different sources and generating a unified fused time-series dataset. A feature extraction subunit extracts features from process parameters and laboratory data and binds feature data to spatial topology identifiers, generating multi-dimensional feature vectors with spatial location information. A waveform noise reduction module removes noise from ultrasonic thickness measurement waveform signals, improving the quality of the effective signal. A defect identification module enables automatic identification of defect types based on waveform features, providing more comprehensive feature input for subsequent corrosion risk assessment.

[0054] refer to Figure 3In another embodiment, the digital twin construction unit includes a primitive analysis subunit and a topology mapping subunit. The primitive analysis subunit is used to perform primitive recognition on the piping and instrumentation diagram (PID), extract the logical relationships between equipment nodes, pipe connections, and valves, perform 3D geometric reconstruction of the equipment empty view, and extract the spatial coordinates of pipe segments and the geometric dimensions of pipe fittings. The PID is a DWG format CAD drawing file. The primitive analysis subunit first parses the graphic database of the drawing file, extracts all graphic entities contained in the drawing, including block references, lines, arcs, polylines, and text annotation entities. Then, based on a pre-trained primitive classification model, it classifies and identifies the extracted block reference entities. The classification categories include equipment and instrument nodes such as containers, heat exchangers, pumps, valves, flow meters, and pressure gauges. At the same time, it extracts the lines and polylines connected to the equipment nodes, identifies them as pipe connections, and extracts the text annotation information of the pipe number, pipe diameter, and medium type corresponding to the pipe connections. After identifying the equipment nodes and pipeline connections, the element parsing subunit extracts the connection relationship between valves and pipelines, constructs the logical correspondence between valve opening / closing status and pipeline on / off status, and completes the full extraction of two-dimensional logical topology information in the pipeline instrumentation diagram.

[0055] The equipment's empty view is a 3D model file in STEP format. The element analysis subunit first parses the geometric topology data of the 3D model file, extracting all geometric entities contained in the model, including cylinders, cones, toroids, and extruded bodies. Based on the topological connection relationships of these geometric entities, it identifies pipe segments, elbows, tees, reducers, flanges, and other pipe fitting entities, completing the 3D geometric reconstruction. For the identified pipe segment entities, the element analysis subunit extracts the 3D coordinates of the start point, end point, and intermediate nodes of the pipe segment's centerline as the pipe segment's spatial coordinates. For the identified various pipe fitting entities, it extracts parameters such as the outer diameter, inner diameter, wall thickness, radius of curvature, and branch dimensions as the pipe fitting's geometric dimensions. Simultaneously, it extracts the attribute information corresponding to each pipe fitting entity, including equipment tag number, pipe number, and material type, completing the full extraction of 3D physical information.

[0056] In this embodiment, the topology mapping subunit is used to bind the equipment nodes and pipeline connections extracted by primitive identification to the spatial coordinates of the pipe segments extracted by 3D geometric reconstruction, establishing an association mapping matrix that includes 2D logical topological relationships and 3D physical spatial coordinates. Based on the association mapping matrix, the logical relationships of equipment nodes, pipeline connections, and valves are projected onto the device map to generate a 3D digital twin. The binding of equipment nodes and pipe segment entities is based on the unique identifiers of equipment tag numbers and pipe numbers. That is, the equipment tag numbers extracted from the piping instrumentation diagram must completely match the equipment attribute tag numbers extracted from the 3D empty view, and the same applies to pipe numbers. Based on this matching relationship, the corresponding binding of 2D logical elements and 3D physical elements is completed.

[0057] The correlation mapping matrix is ​​constructed using the following formula:

[0058]

[0059] in, For the correlation mapping matrix, This represents the total number of equipment nodes and pipe segments. For the first Each element is a two-dimensional logically unique identifier in the piping and instrumentation diagram. For the first The three-dimensional physical unique identifier of each element in the three-dimensional empty view. For the first The three-dimensional spatial coordinates of each element in the unified factory area world coordinate system.

[0060] The association mapping matrix establishes the correspondence between two-dimensional logical identifiers, three-dimensional physical identifiers, and spatial coordinates, realizing the association and binding of two-dimensional logical topological information and three-dimensional physical spatial information. Based on this association mapping matrix, the topology mapping sub-unit projects the equipment nodes, pipeline connections, and valve logical relationships extracted from the piping and instrumentation diagram onto the device map that has undergone coordinate transformation. It overlays two-dimensional process logic information onto the three-dimensional physical space model and the plant's planar positioning information, generating a three-dimensional digital twin that simultaneously contains two-dimensional process logic attributes, three-dimensional physical space coordinates, and plant-wide planar positioning information. Each element in the three-dimensional digital twin can be bidirectionally queried and mapped between two-dimensional logical information and three-dimensional spatial information through the association mapping matrix.

[0061] Table 3. Mapping Relationship between Piping and Instrumentation Flow Diagram Elements and Geometric Elements in 3D Empty View

[0062]

[0063] This table clarifies the correspondence between two-dimensional logical primitives and three-dimensional physical elements, providing a data foundation for the construction of the association mapping matrix, ensuring the accurate binding of two-dimensional process logic information and three-dimensional physical space information, and avoiding mapping misalignment problems.

[0064] In this embodiment, the topology mapping subunit further includes a hierarchical sectioning module and a dynamic rendering module. The hierarchical sectioning module, in response to a received external sectioning command, generates a sectioning plane along the normal vector direction of the pipe section in the 3D digital twin, and extracts the cross-sectional contour line where the sectioning plane intersects with the pipe's geometric dimensions. The external sectioning command is an operation command sent by the user through an interactive terminal, containing the unique identifier of the pipe to be sectioned and the sectioning position coordinates. After receiving the sectioning command, the hierarchical sectioning module first extracts the centerline data of the corresponding pipe from the 3D digital twin based on the unique pipe identifier, calculates the tangent vector of the pipe centerline at the sectioning position coordinates, uses the perpendicular vector of this tangent vector as the normal vector of the pipe section, and generates a sectioning plane based on the normal vector and the sectioning position coordinates. The spatial equation of the sectioning plane is realized through the following formula:

[0065] in, Let cosine be the direction cosine of the normal vector of the pipe section. Let be the coordinates of a point on the centerline of the pipe at the cutting location. This equation is the spatial plane equation of the cutting plane in a unified world coordinate system.

[0066] After generating the cutting plane, the hierarchical cutting module calculates the intersection relationship between the cutting plane and the geometric entity of the pipe fitting, and extracts the intersection line as the cross-sectional contour line. The cross-sectional contour line includes the intersection contour of the outer circle and inner hole of the pipe, as well as the intersection contour of pipe fittings such as elbows and tees, and fully presents the wall thickness and cross-sectional shape information of the pipe at the cutting position.

[0067] In this embodiment, the dynamic rendering module acquires the corrosion level assessment results output by the intelligent analysis and early warning unit, determines the rendering color of the cross-sectional outline based on a preset level and color mapping table, fills the area enclosed by the cross-sectional outline with the rendering color, and overlays and displays the cross-sectional outline and rendering color on the 3D digital twin. The preset level and color mapping table is a pre-configured correspondence between corrosion levels and RGB color values, with different corrosion levels corresponding to different rendering colors. After receiving the corrosion level assessment results, the dynamic rendering module locates the corresponding pipe and section position in the 3D digital twin based on the device identifier and associated mapping matrix. According to the corrosion level at that position, it matches the corresponding rendering color from the mapping table and fills the pipe wall thickness area enclosed by the cross-sectional outline with the color. At the same time, it overlays and displays numerical labels for corrosion rate, remaining wall thickness, and remaining life at the cross-sectional position, visually presenting the corrosion situation at the section position in the 3D digital twin.

[0068] This embodiment achieves automatic identification of elements and extraction of two-dimensional logical topology information in piping instrumentation diagrams through the element parsing subunit, as well as three-dimensional geometric reconstruction and physical information extraction from equipment empty views; establishes an association mapping matrix between two-dimensional logic and three-dimensional space through the topology mapping subunit, realizing deep fusion and unified projection of the two types of information, generating a complete three-dimensional digital twin; realizes arbitrary position sectioning of the three-dimensional digital twin through the hierarchical sectioning module, allowing for intuitive viewing of the cross-sectional information inside the pipeline; and realizes the visualization mapping of corrosion level assessment results on the three-dimensional cross-section through the dynamic rendering module, providing support for intuitive viewing and analysis of corrosion information.

[0069] refer to Figure 4 In another embodiment, the intelligent analysis and early warning unit includes a mechanism matching subunit and an association early warning subunit. The mechanism matching subunit contains multiple types of corrosion mechanism models, used to input data obtained from the multi-source heterogeneous data access unit into each corrosion mechanism model for forward calculation, outputting a corrosion rate fitting curve corresponding to each corrosion mechanism model. The multiple types of corrosion mechanism models include a uniform corrosion mechanism model, a pitting corrosion mechanism model, an erosion corrosion mechanism model, a stress corrosion cracking mechanism model, and a corrosion mechanism model under the insulation layer. Each model is constructed based on the corresponding basic corrosion theory and has a clearly defined range of input parameters and output results. The mechanism matching subunit first matches the corresponding corrosion mechanism model based on the equipment type, medium type, and material type. Then, it inputs the process parameters, medium test data, and material property data obtained and preprocessed by the multi-source heterogeneous data access unit into the matched corrosion mechanism model for forward calculation, obtaining the calculated corrosion rate value under the corresponding operating condition. Finally, based on the calculated corrosion rate values ​​at continuous time nodes, it fits and generates a continuous curve of corrosion rate change over time, i.e., the corrosion rate fitting curve.

[0070] The forward calculation of the uniform corrosion mechanism model is achieved through the following formula:

[0071]

[0072] in, The corrosion rate is uniform, and the unit is mm / a; The unit conversion constant has a value of 3270. The corrosion current density is expressed in A / cm² and is calculated using a polarization curve model based on the medium temperature, corrosive medium concentration, and material type. This represents the molar mass of the metal, expressed in g / mol. This represents the number of electrons gained or lost in the metal corrosion reaction. This is the Faraday constant, with a value of 96485 C / mol; This represents the density of the metal, expressed in g / cm³.

[0073] Table 4. Correspondence between input parameters and output results of typical corrosion mechanism models.

[0074]

[0075] This table clarifies the input parameters, output results, and applicable scenarios for various corrosion mechanism models, providing a standard for model matching and forward calculation of mechanism matching sub-units, and ensuring the adaptability of corrosion rate calculation results to actual equipment operating conditions.

[0076] In this embodiment, the associated early warning subunit is used to obtain the deviation value between the corrosion rate fitting curve and the preset benchmark corrosion rate curve. When the deviation value exceeds the set deviation threshold, it retrieves a historical case set from the case library that matches the current equipment type and process parameters, and extracts the characteristic operating condition parameters from the historical case set as the associated early warning condition output. The preset benchmark corrosion rate curve is a standard curve of the corrosion rate changing over time under the design conditions and normal operating conditions of the equipment, which is pre-constructed based on equipment design documents and material corrosion manual data. The deviation value is calculated using the following formula:

[0077] in, Time node The corresponding corrosion rate deviation value, Time node The corresponding fitted corrosion rate value, Time node The corresponding baseline corrosion rate value.

[0078] The associated early warning subunit compares the calculated deviation value with a set deviation threshold. When the deviation value exceeds the set threshold, a case library retrieval operation is triggered. The retrieval process involves: extracting the equipment type, material model, and process condition feature vector of the current equipment; calculating the cosine similarity between this feature vector and the feature vectors of each historical case in the case library; and filtering out cases with similarity higher than a preset similarity threshold to form a historical case set. Then, characteristic operating condition parameters leading to corrosion anomalies are extracted from the historical case set, including the medium temperature threshold, the upper limit of chloride ion concentration, and the critical value of medium flow rate. These parameters are output as associated early warning conditions. When subsequent operating condition parameters reach these warning conditions, a corrosion risk warning is triggered in advance.

[0079] In this embodiment, the associated early warning subunit also includes a multi-parameter coupling analysis module. This module, when the equipment type is a heat exchanger, acquires the rate of change of the heat transfer coefficient and the change in medium flow velocity from the process operating parameters, and acquires the tube-side pressure difference signal from the online monitoring data. It inputs the rate of change of the heat transfer coefficient, the change in medium flow velocity, and the tube-side pressure difference signal into a preset scaling and erosion correlation matrix for matrix multiplication, outputting a scaling tendency index and an erosion corrosion tendency index. When the scaling tendency index or the erosion corrosion tendency index exceeds the corresponding set boundary, an associated early warning condition is triggered. The rate of change of the heat transfer coefficient is the relative change rate between the currently calculated heat exchanger heat transfer coefficient and the design heat transfer coefficient; the change in medium flow velocity is the difference between the current tube-side medium flow velocity and the normal operating flow velocity; and the tube-side pressure difference signal is the pressure difference between the inlet and outlet of the heat exchanger tubes. The scaling and erosion correlation matrix is ​​a coupling coefficient matrix pre-constructed based on heat exchanger heat transfer and fluid mechanics theories and historical operating data, used to quantify the coupled influence of multiple parameters on scaling and erosion corrosion.

[0080] Matrix operations are performed using the following formula:

[0081]

[0082] in, This is a scaling tendency index. To assess the scouring corrosion tendency index, For the scale flushing correlation matrix, The rate of change of the heat transfer coefficient, This represents the change in medium flow velocity. This is the tube-side differential pressure signal.

[0083] The multi-parameter coupling analysis module compares the calculated scaling tendency index and scouring corrosion tendency index with their corresponding set boundary values. When either index exceeds the corresponding set boundary, the associated early warning condition is triggered, thus enabling early warning of the risks of scaling and scouring corrosion in the heat exchanger.

[0084] refer to Figure 5In this embodiment, the closed-loop task flow unit includes a work order structuring subunit and a state machine flow rotor unit. The work order structuring subunit is used to parse the equipment identifier, risk level, and corrosion type fields in the anomaly assessment results, match them with a preset task handling template, and fill in and generate a structured work order containing investigation task nodes, detection task nodes, and maintenance task nodes. The anomaly assessment results are a structured dataset containing fields such as unique equipment identifier, risk level, corrosion type, corrosion location, remaining wall thickness, and remaining lifespan. The work order structuring subunit first parses and extracts these fields, and based on the extracted risk level and corrosion type, matches the corresponding standardized task handling template from a preset template library. The template library pre-stores handling templates corresponding to different risk levels and different corrosion types. Each template contains standardized investigation, detection, and maintenance task content, as well as corresponding execution standards and acceptance requirements. The work order structuring subunit fills the parsed equipment identifier, corrosion location, and corrosion parameters into the matched template, automatically generating a structured work order. The structured work order contains sequentially flowing investigation task nodes, detection task nodes, and maintenance task nodes, with each node corresponding to complete task execution and acceptance requirements.

[0085] In this embodiment, the work order structured subunit also includes an external corrosion analysis module and a special ledger matching module. The external corrosion analysis module is used to extract the equipment's external surface temperature distribution data and insulation layer damage area data from the anomaly assessment results when the corrosion type field is "corrosion under the insulation layer." The special ledger matching module is used to retrieve the external corrosion engineering ledger based on the equipment identifier, obtain the type of insulation material and the last replacement time for the corresponding pipeline, and combine the insulation material type, last replacement time, equipment external surface temperature distribution data, and insulation layer damage area data to fill the investigation task node, generating a structured work order containing a special inspection instruction for the insulation layer. The external corrosion engineering ledger is a pre-built structured database containing insulation layer information, external corrosion layer information, construction records, and replacement history data for all equipment and pipelines with insulation layers within the plant area. When the corrosion type is corrosion under the insulation layer, the special ledger matching module retrieves the ledger based on the equipment identifier to obtain the corresponding historical information of the insulation layer and anti-corrosion layer. The external anti-corrosion analysis module extracts the temperature distribution and damage area data from the anomaly assessment results and fills them into the investigation task node together to generate targeted task instructions for special inspection of the insulation layer, damage repair, and anti-corrosion layer detection, thereby improving the targeting and executability of the work order.

[0086] In this embodiment, the state machine flow rotor unit is used to define multiple discrete states of the structured work order. When the execution result data of the current task node is received, the structured work order is transferred from the current discrete state to the discrete state corresponding to the next task node, until the discrete state flow corresponding to all task nodes is completed, and an archiving identifier is generated. The state machine flow rotor unit predefines the discrete states of the entire life cycle of the structured work order, including six discrete states: pending allocation, pending execution, in execution, pending acceptance, accepted, and archived. It also defines the allowed operations and state transition rules for each state. After the structured work order is generated, the initial state is pending allocation. After the responsible person is assigned the work order, the state transitions to pending execution. After the task executor receives the work order, the state transitions to in execution. After the executor uploads the execution result data, the state transitions to pending acceptance. After the acceptor completes the acceptance and confirms the pass, the state transitions to accepted. When all task nodes have completed the acceptance, the work order state transitions to archived, and a unique archiving identifier is generated, completing the entire process flow of the work order.

[0087] The state transition process is achieved through the following formula:

[0088]

[0089] in, This represents the current discrete state of the work order. Input data for the execution result of the current task node. The state transition function is a predefined state transition function, constructed based on the state transition rules. This represents the next discrete state after the transition.

[0090] refer to Figure 6 In this embodiment, the closed-loop task flow unit further includes a verification feedback subunit. Upon receiving the archiving identifier, the verification feedback subunit retrieves the equipment operation data after treatment within a preset verification period, calculates the actual corrosion rate index corresponding to the treated equipment operation data, calculates the difference between the actual corrosion rate index and the predicted corrosion rate index output by the intelligent analysis and early warning unit when generating the structural work order, and generates model adjustment parameters based on the difference calculation result. The model adjustment parameters are then input to the intelligent analysis and early warning unit to update the weight coefficients of the corrosion mechanism model and the matching priority of the case library. The preset verification period is a pre-set operation data collection period after the work order is archived, used to collect the actual operation data of the equipment after the hazard treatment is completed, including online monitoring wall thickness data, process operation parameters, newly added offline detection data, and laboratory analysis data. The actual corrosion rate index is calculated based on the initial wall thickness detection value after treatment, the wall thickness detection value after the verification period ends, and the time interval between the two detection nodes.

[0091] The difference in prediction error is calculated using the following formula:

[0092]

[0093] in, The error value for corrosion rate prediction. To verify the actual corrosion rate index calculated within the cycle, This refers to the predicted corrosion rate index output by the intelligent analysis and early warning unit when generating structural chemical data sheets.

[0094] The verification feedback subunit generates corresponding model adjustment parameters based on the calculated prediction error value. When the prediction error is positive, the weight coefficients of relevant input parameters in the corrosion mechanism model are increased, and the matching priority of similar cases in the case library is also increased. When the prediction error is negative, the weight coefficients of relevant parameters and the matching priority are decreased. The generated model adjustment parameters are input to the intelligent analysis and early warning unit to automatically update the weight coefficients of the corrosion mechanism model and the matching priority of the case library, realizing the self-correction and optimization of the model and improving the accuracy of subsequent corrosion rate prediction and risk assessment.

[0095] This embodiment achieves forward calculation of corrosion rate based on corrosion mechanism through the mechanism matching subunit, generating a continuous corrosion rate fitting curve; it achieves monitoring of corrosion rate deviation and matching retrieval of historical cases through the correlation early warning subunit, outputting targeted correlation early warning conditions; it achieves multi-parameter coupling analysis and early warning of heat exchanger scaling and scouring corrosion risks through the multi-parameter coupling analysis module; it achieves automatic generation of structured work orders and customization of special work orders for corrosion under the insulation layer through the work order structuring subunit; it achieves state definition and automated flow of the entire life cycle of work orders through the state machine flow rotor unit; and it achieves verification of treatment effect and automatic updating and optimization of corrosion mechanism model and case library through the verification feedback subunit, forming a closed-loop management of the entire process of corrosion risk assessment, hidden danger treatment, effect verification, and model optimization.

Claims

1. A corrosion information management system for equipment in the refining and chemical industry, characterized in that, It includes a multi-source heterogeneous data access unit, a digital twin construction unit, an intelligent analysis and early warning unit, and a closed-loop task flow unit; The multi-source heterogeneous data access unit is used to acquire online monitoring data, offline detection data, process operating parameters, and laboratory analysis data of refining and chemical equipment. The digital twin building unit is used to integrate the equipment map, pipeline instrumentation diagram and equipment empty view to build a three-dimensional digital twin of the refining and chemical equipment; The intelligent analysis and early warning unit is used to perform fusion analysis on the data obtained by the multi-source heterogeneous data access unit, and output corrosion level assessment results and remaining life prediction results by combining the preset corrosion mechanism model and case library. The closed-loop task flow unit is used to automatically generate a structural work order containing investigation, detection and maintenance nodes when it receives the abnormal assessment result output by the intelligent analysis and early warning unit. After obtaining the handling execution feedback data, it inputs the equipment operation data after handling into the intelligent analysis and early warning unit for effect verification and model parameter update. The three-dimensional digital twin is used to receive the corrosion level assessment result and perform spatial positioning and dynamic mapping.

2. The equipment corrosion information management system for the refining and chemical industry according to claim 1, characterized in that, The multi-source heterogeneous data access unit includes a time-series alignment subunit and a feature extraction subunit; The time-series alignment subunit is used to perform time-axis interpolation alignment of the continuous time-series signals contained in the online monitoring data and the non-continuous timestamp discrete data contained in the offline detection data, based on a set time window, to generate a fused time-series dataset. The feature extraction subunit is used to extract static features of medium flow rate, temperature and pressure from the process operating parameters, extract dynamic features of corrosive medium concentration and dew point temperature from the test analysis data, and bind the static features and the dynamic features to the corresponding time nodes of the fused time series dataset according to the spatial topology relationship of the equipment, forming a multidimensional feature vector with spatial topology identifier.

3. The equipment corrosion information management system for the refining and chemical industry according to claim 1, characterized in that, The digital twin construction unit includes a primitive analysis subunit and a topology mapping subunit; The primitive analysis subunit is used to perform primitive recognition on the piping and instrumentation diagram, extract the logical relationships between equipment nodes, pipeline connections and valves, perform three-dimensional geometric reconstruction on the empty view of the equipment, and extract the spatial coordinates of pipe segments and the geometric dimensions of pipe fittings. The topology mapping subunit is used to bind the equipment nodes and pipeline connections extracted by the primitive identification with the spatial coordinates of the pipe segments extracted by the three-dimensional geometric reconstruction, establish an association mapping matrix containing two-dimensional logical topological relationships and three-dimensional physical spatial coordinates, and project the logical relationships of the equipment nodes, pipeline connections and valves onto the device map based on the association mapping matrix to generate the three-dimensional digital twin.

4. The equipment corrosion information management system for the refining and chemical industry according to claim 1, characterized in that, The intelligent analysis and early warning unit includes a mechanism matching subunit and a correlation early warning subunit; The mechanism matching subunit is equipped with multiple types of corrosion mechanism models, which are used to input the data obtained by the multi-source heterogeneous data access unit into each corrosion mechanism model for forward calculation and output the corrosion rate fitting curve corresponding to each corrosion mechanism model. The associated early warning subunit is used to obtain the deviation value between the corrosion rate fitting curve and the preset benchmark corrosion rate curve. When the deviation value exceeds the set deviation threshold, it retrieves the historical case set in the case library that matches the current equipment type and process parameters, and extracts the characteristic operating condition parameters in the historical case set as the associated early warning condition output.

5. The equipment corrosion information management system for the refining and chemical industry according to claim 1, characterized in that, The closed-loop task flow unit includes a work order structured sub-unit and a state machine flow rotor unit. The work order structured sub-unit is used to parse the equipment identifier, risk level and corrosion type fields in the anomaly assessment result, match the preset task handling template, and fill in and generate a structured work order containing investigation task nodes, detection task nodes and maintenance task nodes. The state machine flow rotor unit is used to define multiple discrete states of the structured work order. When the execution result data of the current task node is received, the structured work order is transferred from the current discrete state to the discrete state corresponding to the next task node, until the discrete state flow corresponding to all task nodes is completed, and an archive identifier is generated.

6. The equipment corrosion information management system for the refining and chemical industry according to claim 5, characterized in that, The closed-loop task flow unit also includes a verification feedback subunit; The verification feedback subunit is used to retrieve the processed equipment operation data of the equipment identifier within a preset verification period after receiving the archive identifier, and calculate the actual corrosion rate index corresponding to the processed equipment operation data. The difference between the actual corrosion rate index and the predicted corrosion rate index output by the intelligent analysis and early warning unit when generating the structural chemical product is calculated, and model adjustment parameters are generated based on the difference calculation results. The model adjustment parameters are input into the intelligent analysis and early warning unit to update the weight coefficients of the corrosion mechanism model and the matching priority of the case library.

7. The equipment corrosion information management system for the refining and chemical industry according to claim 2, characterized in that, The feature extraction subunit also includes a waveform noise reduction module and a defect identification module; The waveform noise reduction processing module is used to perform wavelet packet decomposition on the ultrasonic thickness measurement waveform signal in the online monitoring data, remove frequency band components below the set energy threshold, reconstruct the signal, and output the noise-reduced waveform signal. The defect identification module is used to extract the time-domain envelope and frequency-domain main frequency offset of the denoised waveform signal, input the time-domain envelope and the frequency-domain main frequency offset into a pre-trained anomaly classifier, output the defect type label corresponding to the discrete data of the non-continuous timestamps, and append the defect type label to the multi-dimensional feature vector.

8. The equipment corrosion information management system for the refining and chemical industry according to claim 3, characterized in that, The topology mapping subunit also includes a hierarchical sectioning module and a dynamic rendering module; The hierarchical sectioning module is used to generate a sectioning plane along the normal vector direction of the pipe section of the three-dimensional digital twin in response to the received external sectioning command, and extract the cross-sectional contour line where the sectioning plane intersects with the geometric dimensions of the pipe fitting; The dynamic rendering module is used to obtain the corrosion level assessment result output by the intelligent analysis and early warning unit, determine the rendering color of the cross-sectional outline according to the preset level and color mapping table, fill the area surrounded by the cross-sectional outline with the rendering color, and overlay the cross-sectional outline and the rendering color on the three-dimensional digital twin.

9. The equipment corrosion information management system for the refining and chemical industry according to claim 4, characterized in that, The associated early warning subunit also includes a multi-parameter coupling analysis module; The multi-parameter coupling analysis module is used to obtain the rate of change of heat transfer coefficient and the change of medium flow velocity in the process operating parameters when the equipment type is a heat exchanger, and to obtain the tube-side pressure difference signal in the online monitoring data. The heat transfer coefficient change rate, the medium flow rate change, and the tube pressure difference signal are input into a preset scaling and scouring correlation matrix for matrix multiplication, and the scaling tendency index and scouring corrosion tendency index are output. When the scaling tendency index or the scouring corrosion tendency index exceeds the corresponding set boundary, the associated early warning condition is triggered and output.

10. The equipment corrosion information management system for the refining and chemical industry according to claim 5, characterized in that, The work order structured sub-unit also includes an external corrosion analysis module and a special ledger matching module; The external corrosion analysis module is used to extract the equipment external surface temperature distribution data and insulation layer damage area data from the anomaly assessment results when the corrosion type field is corrosion under the insulation layer. The special ledger matching module is used to retrieve the external anti-corrosion engineering ledger based on the equipment identifier, obtain the type of insulation material and the last replacement time of the corresponding pipeline, and combine the type of insulation material, the last replacement time, the temperature distribution data of the outer surface of the equipment and the data of the damaged area of ​​the insulation layer to fill the investigation task node, and generate a structural engineering order containing a special inspection instruction for the insulation layer.

Images