A sintering machine full life cycle operation and maintenance method and system based on digital twinning and a medium

Abstract

The application discloses a sintering machine full life cycle operation and maintenance method and system based on digital twinning, relates to the technical field of sintering, and comprises the following steps: acquiring single-dimension standard data streams corresponding to each monitoring dimension of a physical entity of a sintering machine; mapping each single-dimension standard data stream to form a single-twin body state data stream; generating sub-component health quantification data streams of each sub-component, generating an initial system diagnosis result data stream of a preliminary diagnosis conclusion of the whole machine based on the sub-component health quantification data streams of each component; generating an intelligent decision control instruction data stream with a target identifier; performing maintenance operations in response to the intelligent decision control instruction data stream, and updating each single-twin body state data stream and dynamic attributes of the digital twin body. The method is based on a complete closed-loop idea of perception, calculation, decision, execution and feedback, and realizes intelligent operation and maintenance of the sintering machine trolley from commissioning, operation, maintenance to scrapping of the full life cycle.

Description

Technical Field

[0001] This invention relates to the field of sintering technology, and in particular to a method, system and medium for the full life cycle operation and maintenance of sintering machines based on digital twins. Background Technology

[0002] Sintering machines are key equipment in steel production, consisting of a mobile cluster of hundreds of trolleys. Currently, the management of sintering machine trolleys and other equipment generally suffers from the following pain points: First, information is fragmented and lacks a unified view during the sintering process: equipment status (ballast plates, grate bars), process parameters (wind box temperature, negative pressure), and maintenance execution (lubrication, replacement) data belong to different systems, forming data silos and making it impossible to perform correlation analysis based on actual production stages; Second, management is extensive and lacks granularity: traditional management is based on the entire machine, failing to provide refined tracking and health management of individual trolleys throughout their entire lifecycle from inventory, online operation, offline maintenance, to scrapping; Third, maintenance is passive and decision-making is lagging: existing technologies mainly rely on fixed-cycle preventative and reactive maintenance, failing to perform predictive maintenance based on real-time equipment health status, leading to frequent unplanned downtime, high maintenance costs, and shortened equipment lifespan; Fourth, there is a disconnect between process and equipment: process operation does not consider equipment health status, while equipment degradation is often caused by process anomalies, making collaborative optimization impossible.

[0003] Patent application number 2021112212538 discloses a fault identification and analysis method, system, and medium for sintering machine trolley side panels, which uses image processing to achieve fault identification and analysis of the trolley side panels; patent application number 2022101532467 discloses a diagnostic system, method, and storage medium for sintering machine trolley grate bars, which discloses fault identification and analysis of sintering machine trolley grate bars; patent application number 2021112212542 discloses a fault detection method, system, and medium for sintering machine trolley wheels, which discloses fault identification and analysis of sintering machine trolley wheels. The existing technologies utilize local detection devices to perform local fault detection at each stage, lacking a technical concept that can organically integrate all elements to achieve intelligent operation and maintenance of the sintering machine throughout its entire lifecycle.

[0004] Therefore, there is an urgent need for a new method to achieve intelligent diagnosis and optimization decision-making for the sintering machine as a whole, and to realize real-time, all-element intelligent operation and maintenance of the sintering machine throughout its entire life cycle. Summary of the Invention

[0005] The main objective of this invention is to provide a digital twin-based method, system, and medium for the full lifecycle operation and maintenance of sintering machines. This aims to solve the technical problem that existing technologies, which rely solely on local detection devices for local fault detection at various stages, cannot achieve intelligent diagnosis, optimization decision-making, and intelligent operation and maintenance of the sintering machine as a whole throughout its entire lifecycle.

[0006] To achieve the above objectives, this invention provides a digital twin-based full lifecycle operation and maintenance method for sintering machines, enabling intelligent operation and maintenance throughout the entire lifecycle of the sintering machine. The digital twin is a comprehensive virtual entity with a Physical Information Neural Network (PINN) and a physical mechanism model as its core, and a geometric model as its form carrier. This comprehensive virtual entity is configured with inherent and dynamic attributes. The inherent attributes include unique identification codes for each sintering trolley and its components. The method includes the following steps: S1. Obtain the single-dimensional standard data stream corresponding to each monitoring dimension of the physical entity of the sintering machine. Each single-dimensional standard data stream in the single-dimensional standard data stream is marked with identity information, timestamp information and physical location identifier. S2. Based on the single-dimensional standard data stream and the inherent attribute information, perform identity binding, spatiotemporal alignment and dynamic attribute mapping in sequence to map each of the single-dimensional standard data streams into a single twin state data stream; S3. The digital twin calls the physical information neural network (PINN) and the physical mechanism model of the kernel, performs single-dimensional state judgment calculation based on the state data stream of each single twin, generates sub-component health quantification data stream of each sub-component, and generates initial system diagnosis result data stream of preliminary diagnosis conclusion of the whole machine based on the sub-component health quantification data stream of each component. S4. Based on the quantified health data stream of the sub-component and the constraint verification results of the process characteristics of each cycle stage in the entire life cycle, a sub-component maintenance instruction data stream with target identifiers is generated; a multi-dimensional correlation analysis is performed on the quantified health data stream of the sub-component and the initial system diagnosis result data stream to obtain optimized whole system diagnosis results; then, combined with data coupling correlation, whole machine operating conditions and the cycle stage process characteristics of the current life cycle stage, a whole machine maintenance instruction data stream with target identifiers is generated; the sub-component maintenance instruction data stream and the whole machine maintenance instruction data stream are combined to form an intelligent decision control instruction data stream, and the target identifier corresponds one-to-one with the unique identity code of each vehicle and component in the digital twin; S5. Respond to the intelligent decision control command data stream to perform maintenance operations, collect the maintenance operation completion status data stream corresponding to the execution of the intelligent decision control command data stream, map the maintenance operation completion status data stream to the preset dynamic label field corresponding to the digital twin, and update the status data stream of each individual twin and the dynamic attributes of the digital twin; the preset dynamic label field is a structured data unit in the digital twin used to store the real-time status and operation and maintenance records of each component.

[0007] Furthermore, step S1 specifically includes: Collect multi-source heterogeneous raw data streams corresponding to multiple monitoring dimensions of the physical entity of the sintering machine, and mark each single dimension of the multi-source heterogeneous raw data stream with identity information, timestamp information and physical location identifier respectively; Anomaly removal and normalization are performed on the single-dimensional data stream in the multi-source heterogeneous original data stream to obtain a single-dimensional standard data stream.

[0008] Furthermore, image information of the sideboard of the physical trolley is acquired at a first preset position point, and the deformation amount of the sideboard is acquired based on the image information of the sideboard. The grate image information of the physical trolley is obtained at the second preset position point, and the grate deformation amount and blockage area ratio are obtained based on the grate image information; At the third preset position point, obtain the wheel image information of the physical trolley, and obtain the wheel deformation and wheel axle clearance based on the wheel image information; At the fourth preset location point, the wind box status information corresponding to the physical trolley during sintering is obtained. The wind box status information includes at least the wind box negative pressure and the wind box temperature. The deformation of the guardrail, the deformation of the grate bars, the proportion of the blocked area, the deformation of the wheel, the wheel axle clearance, the negative pressure of the air box, and the temperature of the air box are all single-dimensional data and are respectively marked with identity information, timestamp information and physical location identifier.

[0009] Furthermore, the method for generating the sub-component health metric data stream in step S3 includes: The image information of the guardrail, the image information of the grate, and the image information of the wheel are processed to obtain dimensionless quantified values ​​of guardrail health, grate health, and wheel health. Furthermore, the entire life cycle includes the trolley commissioning break-in period, the trolley normal operation period, the trolley maintenance and repair period, and the trolley scrapping assessment period. The preset stage operation and maintenance decision rules are predetermined based on the equipment status characteristics, operation and maintenance priority characteristics, and process requirement characteristics in the process characteristics of the cycle stage. The preset phase operation and maintenance decision rules specifically include: During the break-in period of the trolley, periodic inspection instructions are generated based on the health quantification values ​​of each sub-component. During the normal operation of the trolley, process parameter adjustment and maintenance instructions are generated based on the optimized whole system diagnostic results and the health status of each sub-component. During the maintenance and repair period of the trolley, maintenance strategy instructions are generated based on the abnormal component identification and the degree of abnormality. During the trolley's scrap assessment period, scrap assessment and component replacement instructions are generated based on the development trend of the health quantification values ​​of each sub-component.

[0010] Furthermore, the maintenance operation completion status data stream includes maintenance operation type, operation time, operation parameters, and operation execution result. After the maintenance operation completion status data stream is mapped to a preset dynamic label field, the digital twin updates the preset dynamic label field of the corresponding component.

[0011] This invention also provides a sintering machine full lifecycle operation and maintenance system based on digital twins, including a perception layer, a network transmission layer, a digital twin application layer, and an execution layer for communication connection; The sensing layer is used to collect multi-source heterogeneous monitoring data of the physical entity of the sintering machine, mark it with timestamp information, and then output it to the network transmission layer. The sensing layer includes an identity recognition unit, a physical state acquisition unit, and a physical process parameter acquisition unit. The network transport layer is used to realize bidirectional data transmission of multi-source heterogeneous monitoring data and control commands; The digital twin application layer is the intelligent decision-making center. The digital twin application layer has a built-in sintering machine digital twin with a physical information neural network (PINN) and physical mechanism model as the intelligent core, a geometric model as the form carrier, and real-time data fusion. The digital twin is a fusion of data intelligence and physical intelligence. The digital twin application layer is used to realize the virtual-real mapping of physical entities and multi-dimensional intelligent decision-making, and output control commands to the network transmission layer. The execution layer is used to receive control commands and execute corresponding sintering machine operation and maintenance operations, and to send the operation completion status data back to the network transmission layer.

[0012] The present invention also provides a digital twin-based sintering machine full lifecycle operation and maintenance medium, which stores a computer program. When the computer program is executed, it implements the above-mentioned digital twin-based sintering machine full lifecycle operation and maintenance method.

[0013] Compared with existing technologies, the sintering machine full lifecycle operation and maintenance method based on digital twin provided by this invention has the following beneficial effects: This invention provides a full lifecycle operation and maintenance method for sintering machines based on digital twins. It is based on constructing a digital twin with a Physical Information Neural Network (PINN) and a physical mechanism model as its core, and a geometric model as its morphological carrier. After acquiring multiple single-dimensional standard data streams in the physical entity space, including marker identity information, timestamp information, and physical location identifiers, a full-link data stream processing mechanism is sequentially executed, involving identity binding, spatiotemporal alignment, and dynamic attribute mapping. This achieves real-time and accurate synchronization of the physical entity state corresponding to each single-dimensional standard data stream with the comprehensive virtual entity. Then, it generates quantified health data streams for each sub-component, based on… The system diagnostic result data stream generates a preliminary diagnostic conclusion for the entire machine based on the quantified health data stream of each component. Next, a multi-dimensional correlation analysis is performed on the quantified health data stream of each component and the system diagnostic result data stream to obtain an optimized overall system diagnostic result. Using this optimized overall system diagnostic result, and based on preset stage operation and maintenance decision rules corresponding to the process characteristics of each cycle stage throughout the entire lifecycle, an intelligent decision control command data stream with target identifiers is generated. Finally, maintenance operations are performed in response to the intelligent decision control command data stream, and the maintenance execution results are back-mapped to update the dynamic attributes of the digital twin. This invention's solution is based on a complete closed-loop approach of perception, calculation, decision-making, execution, and feedback. It not only achieves intelligent operation and maintenance of the sintering machine trolley throughout its entire lifecycle from commissioning, operation, maintenance to scrapping, but also improves equipment status identification accuracy, optimizes operation and maintenance strategies, and reduces production losses caused by air leakage and abnormal operating conditions by combining the coupling correlation of the health status of each component, the overall machine operating conditions, and the process characteristics of the current lifecycle stage. This significantly improves the stability of sintering production and the efficiency of equipment management. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating the full lifecycle operation and maintenance method for sintering machines based on digital twins in one embodiment of the present invention.

[0016] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0020] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0021] Please refer to the appendix. Figure 1 This invention provides a digital twin-based method for the full lifecycle operation and maintenance of sintering machines, enabling intelligent operation and maintenance throughout the entire lifecycle of the sintering machine. The digital twin is a comprehensive virtual entity with a physical information neural network (PINN) and a physical mechanism model as its core and a geometric model as its form carrier. The comprehensive virtual entity is configured with inherent attributes and dynamic attributes. The inherent attributes include unique identification codes for each sintering trolley and its components. The method includes the following steps: To achieve intelligent operation and maintenance of the sintering machine throughout its entire lifecycle, the digital twin is a comprehensive virtual entity with a Physical Information Neural Network (PINN) and a physical mechanism model as its core and a geometric model as its form carrier. This comprehensive virtual entity is configured with inherent and dynamic attributes. The inherent attributes include a unique identification code for each sintering trolley and its components, and include the following steps: S1. Obtain the single-dimensional standard data stream corresponding to each monitoring dimension of the physical entity of the sintering machine. Each single-dimensional standard data stream in the single-dimensional standard data stream is marked with identity information, timestamp information and physical location identifier. S2. Based on the single-dimensional standard data stream and the inherent attribute information, identity binding, spatiotemporal alignment, and dynamic attribute mapping are performed sequentially to map each of the single-dimensional standard data streams into a single twin state data stream. The identity binding is achieved by matching the identity identifier information in the single-dimensional standard data stream with the unique identity code in the digital twin to achieve a unique match between the single-dimensional standard data stream and each component of the digital twin. The spatiotemporal alignment is achieved by unifying the timestamp information of the single-dimensional standard data stream to a reference time axis and converting the physical location identifier into the virtual space coordinates of the digital twin. The dynamic attribute mapping is achieved by injecting the single-dimensional standard data stream into the preset dynamic label fields of each component of the corresponding digital twin according to the type classification. S3. The digital twin calls the physical information neural network (PINN) and the physical mechanism model (physical constraints) of the kernel, performs single-dimensional state judgment calculation based on the state data stream of each single twin, generates the sub-component health quantification data stream (dimensionless value) of each sub-component, and generates the initial system diagnosis result data stream of the preliminary diagnosis conclusion of the whole machine based on the sub-component health quantification data stream of each component. S4. Generate a sub-component maintenance instruction data stream with target identifier based on the sub-component health quantification data stream and the constraint verification results of the process characteristics of each cycle stage in the entire life cycle; A multi-dimensional correlation analysis is performed on the sub-component health quantification data stream and the initial system diagnostic result data stream to obtain optimized whole system diagnostic results. Then, the whole system maintenance instruction data stream with target identifiers is generated by combining the data coupling correlation, the whole machine operating conditions and the cycle stage process characteristics of the current life cycle stage. The sub-component maintenance instruction data stream and the whole machine maintenance instruction data stream are combined to form an intelligent decision control instruction data stream, and the target identifier corresponds one-to-one with the unique identity code of each vehicle and component in the digital twin; The multi-dimensional correlation analysis described in this invention can specifically employ a multivariate regression model or a graph neural network model trained based on historical operational data. Taking the coupling between the deformation of the guardrail and the negative pressure of the bellows as an example, the deformation of the guardrail, the grate blockage rate, and the trolley position code are used as inputs. Through a pre-trained deep neural network, the predicted negative pressure deviation of each bellows is output and compared with the measured value to determine the source of air leakage. This neural network model adopts a three-layer fully connected structure, with the loss function being the mean squared error, and is trained using at least six months of historical data. This embodiment is not limited to this; other machine learning models that can achieve feature coupling can also be applied.

[0022] S5. Respond to the intelligent decision control command data stream to perform maintenance operations, collect the maintenance operation completion status data stream corresponding to the execution of the intelligent decision control command data stream, map the maintenance operation completion status data stream to the preset dynamic label field corresponding to the digital twin, and update the status data stream of each individual twin and the dynamic attributes of the digital twin.

[0023] The present invention provides a digital twin-based full lifecycle operation and maintenance method for sintering machines. Based on the construction of a digital twin with a Physical Information Neural Network (PINN) and a physical mechanism model as its core and a geometric model as its morphological carrier, the method acquires multiple single-dimensional standard data streams in the physical entity space, including identifiers, timestamps, and physical location identifiers. Then, it sequentially executes a full-link data stream processing mechanism involving identity binding, spatiotemporal alignment, and dynamic attribute mapping. This achieves real-time and accurate synchronization of the physical entity state corresponding to each single-dimensional standard data stream with the comprehensive virtual entity. Next, it generates sub-component health quantification data streams for each sub-component, based on the health of each component... The system diagnostic result data stream generates a preliminary diagnostic conclusion for the entire machine based on the quantified health data stream of the sub-components. Next, a multi-dimensional correlation analysis is performed on the quantified health data stream of the sub-components and the system diagnostic result data stream to obtain an optimized overall system diagnostic result. Using the optimized overall system diagnostic result, and based on preset stage operation and maintenance decision rules corresponding to the process characteristics of each cycle stage throughout the entire lifecycle, an intelligent decision control command data stream with target identifiers is generated. Finally, maintenance operations are performed in response to the intelligent decision control command data stream, and the maintenance execution results are back-mapped to update the dynamic attributes of the digital twin. This invention's solution is based on a complete closed-loop approach of perception, calculation, decision-making, execution, and feedback. It not only realizes intelligent operation and maintenance of the sintering machine trolley throughout its entire lifecycle from commissioning, operation, maintenance to scrapping, but also improves equipment status identification accuracy, optimizes operation and maintenance strategies, and reduces production losses caused by air leakage and abnormal operating conditions by combining the coupling correlation of the health status of each component, the overall machine operating conditions, and the process characteristics of the current lifecycle stage. This significantly improves the stability of sintering production and the efficiency of equipment management.

[0024] Understandably, the digital twin-based sintering machine lifecycle operation and maintenance method is applied to the digital twin-based sintering machine lifecycle operation and maintenance system. This system includes a perception layer, a network transmission layer, a digital twin application layer, and an execution layer for communication connections. The perception layer collects multi-source heterogeneous monitoring data of the sintering machine's physical entity, marks it with timestamp information, and outputs it to the network transmission layer. The perception layer includes an identity recognition unit, a physical state acquisition unit, and a physical process parameter acquisition unit. The network transmission layer enables bidirectional data exchange between multi-source heterogeneous monitoring data and control commands. The digital twin application layer serves as the intelligent decision-making center. It incorporates a sintering machine digital twin with a Physical Information Neural Network (PINN) and physical mechanism model as its intelligent kernel, a geometric model as its morphological carrier, and real-time data fusion. This digital twin is a fusion of data intelligence and physical intelligence. The digital twin application layer enables virtual-real mapping of the physical entity and multi-dimensional intelligent decision-making, and outputs control commands to the network transmission layer.

[0025] In an optional embodiment of the present invention, the sensors of the sensing layer are deployed at the sintering machine site to collect omnidirectional physical data of the physical entities. The identification unit uses an RFID reader / writer group and / or an industrial-grade machine vision camera arranged along the track to read the RFID electronic license plate installed on the side of each trolley or identify the unique physical features of the trolley, assigning a unique identification code to each physical trolley and accurately recording its time and position at each key point. This time and position are finally converted into a reference time axis and virtual space coordinates. The physical state acquisition unit uses a high-resolution industrial camera and edge computing. The equipment, deployed at specific locations on the sintering machine track (such as behind the head star wheel and before the tail curve), is used to collect images of the trolley side panels to identify their deformation and cracking states, and to collect images of the trolley grate bars to identify their blockage, breakage, or warping states. The physical process parameter acquisition unit establishes a data communication interface with the sintering machine's main control system (DCS / PLC) to acquire comprehensive sintering process parameters in real time, including but not limited to: exhaust gas temperature, negative pressure, oxygen content, and valve opening of each air box; main flue temperature, negative pressure, and oxygen content; sintering machine speed and material layer thickness; and temperature, oxygen content, and flow rate of the multi-zone air hoods during hot air sintering. The sensing layer also includes auxiliary sensing units such as an infrared thermometer for non-contact detection of the trolley wheel axle bearing temperature and a vibration sensor for detecting vibration states.

[0026] In an optional embodiment of the present invention, the execution layer is used to receive control commands from the digital twin application layer and perform specific operation and maintenance operations. In the physical entity of the sintering machine, the intelligent grease injection machine is deployed on the trolley circulation line and can automatically move to the designated wheel axle position of the trolley according to the command to perform quantitative and precise grease injection. The online replacement device includes automated equipment such as a robot arm, lifting, traction and alignment for replacing faulty trolleys without stopping the sintering machine, so as to realize the automatic replacement of faulty trolleys. The human-machine interaction terminal is set up at the inspection post or the central control room to receive system alarm information and maintenance work orders, and to provide decision support for operators.

[0027] In an optional embodiment of the present invention, the network transmission layer includes network devices such as industrial switches and wireless APs to form a reliable data transmission backbone network, which is used to upload multi-source heterogeneous data collected by the perception layer to the digital twin application layer, and to downlink control commands from the application layer to each device in the execution layer.

[0028] In an optional embodiment of the present invention, the main functions of the digital twin application layer include: constructing three-dimensional models of components such as sintering machines, trolleys, side panels, and grate bars based on BIM or high-precision three-dimensional scanning, and finally generating digital twins for visualization and spatial positioning; defining the structure and relationships of all twin data (identity, location, status, process parameters); encapsulating equipment kinematics, fault evolution laws, and business logic (such as the correlation model between side panel deformation and air leakage rate); and achieving virtual-real synchronization through data fusion and mapping engines. When performing specific virtual-real synchronization mapping: identity binding is performed, creating a unique, life-long twin sub-object in the virtual space for each physical trolley (via its RFID), with the twin sub-object having a unique identification code; spatiotemporal alignment is performed, using the trolley motion model and track geometry model to calculate the precise coordinates of any trolley in the virtual space at any given moment, aligning all timestamped data to a unified time reference; state injection is performed, dynamically injecting the aligned data (such as a 5mm deformation of the side panel and a 15kPa negative pressure in the lower bellows) into the corresponding attributes of the virtual twin, realizing the synchronization and linkage between the physical entity and the virtual model.

[0029] In an optional embodiment of the present invention, the sub-component health quantification data stream is used to assess the health status of each sub-component, including the health of the sideboard, the health of the grate, and the health of the wheel axles. The overall health of the trolley is calculated by fusing multiple indicators for key components, and the preliminary diagnostic conclusion for the entire machine can include the overall health of the trolley. Specifically, the general formula for calculating the health of the sideboard is: ,in To measure the actual deformation, This represents the maximum allowable deformation. If the calculated result is negative, it is set to 0. The health status of other components can be defined in a similar manner, such as the health status of the grate bars. wheel and axle health Etc. This embodiment is not limited to this.

[0030] During actual deployment, RFID readers and industrial cameras are installed along the sintering machine track at the head star wheel, tail curve, and key intermediate points. Intelligent grease injection machines and online replacement devices are installed at appropriate locations on the trolley circulation line. All field equipment is connected via industrial Ethernet and wireless network. For data integration, secure communication is established with the sintering machine's main control DCS / PLC system through an industrial gateway to obtain comprehensive process parameters such as bellows temperature, negative pressure, oxygen content, and valve opening in real time. For model building, a three-dimensional geometric model of the sintering machine is constructed using BIM technology on the central server, defining each trolley, sideboard, and grate as an independently addressable and updateable twin. Simultaneously, a trolley kinematic model and equipment health assessment algorithm model are built-in. A data fusion engine is developed, and data interface programs are written to integrate the sensing layer data in real time. When the system is running, for example, when a trolley numbered A-101 passes a detection point, its identity is identified, the sideboard image is collected and analyzed, showing a deformation of 4mm, and the negative pressure of the bellows below it is 14kPa. After being spatiotemporally aligned, this data is injected into the corresponding attributes of the A-101 vehicle twin in the virtual model. The health assessment engine then calculates the health of its side panels. If the health is below the threshold, an early warning work order is generated, and it may be suggested to fine-tune the relevant air box valves. The entire process is visible in real time in the 3D interface.

[0031] Further, step S1 specifically includes: collecting multi-source heterogeneous raw data streams corresponding to multiple monitoring dimensions of the physical entity of the sintering machine; marking each single-dimensional data stream of the multi-source heterogeneous raw data stream with identity information, timestamp information and physical location identifier; and performing anomaly removal and normalization processing on the single-dimensional data stream in the multi-source heterogeneous raw data stream to obtain a single-dimensional standard data stream.

[0032] Understandably, the monitoring dimensions include the monitoring characteristics of different physical entities, such as guardrails, grates, wheel axle bearings, and bellows. In specific implementation, according to the attribute classification rules of the digital twin, the identity and location data in the standardized monitoring data stream are mapped to the identity attribute field of the digital twin, the component status, process parameters, and sensor data are mapped to the real-time status attribute field of the digital twin, and the time and location identifiers are mapped to the spatiotemporal attribute field of the digital twin. The data in each field is updated in real time, forming a twin status data stream that is synchronized with the physical entity status at the millisecond level.

[0033] In one specific implementation, multi-source heterogeneous raw data streams corresponding to multiple monitoring dimensions of the sintering machine physical entity are collected. Each single-dimensional data point in the multi-source heterogeneous raw data stream is labeled with identity information, timestamp information, and physical location identifier. The labeled multi-source heterogeneous raw data stream is then output, achieving preliminary classification and identification of the input data. The input data includes physical state data and process parameter data of each component of the sintering machine physical entity, acquired through sensing layer sensors and image acquisition devices. The single-dimensional data streams in the multi-source heterogeneous raw data streams undergo anomaly removal and normalization processing. A fusion processing strategy of 3σ criterion anomaly removal and normalization is adopted to map the data to the [0, 1] interval, ensuring the comparability of data from different dimensions. Finally, a single-dimensional standard data stream is obtained, which is quantified into a score from 0 to 100 through a linear transformation method multiplied by 100. A higher score indicates a better component operating status.

[0034] Furthermore, image information of the sideboard of the physical trolley is acquired at a first preset position point, and the deformation amount of the sideboard is acquired based on the image information of the sideboard. The grate image information of the physical trolley is obtained at the second preset position point, and the grate deformation amount and blockage area ratio are obtained based on the grate image information; At the third preset position point, obtain the wheel image information of the physical trolley, and obtain the wheel deformation and wheel axle clearance based on the wheel image information; At the fourth preset location point, the wind box status information corresponding to the physical trolley during sintering is obtained. The wind box status information includes at least the wind box negative pressure and the wind box temperature. The deformation of the guardrail, the deformation of the grate bars, the proportion of the blocked area, the deformation of the wheel, the wheel axle clearance, the negative pressure of the air box, and the temperature of the air box are all single-dimensional data and are respectively marked with identity information, timestamp information and physical location identifier.

[0035] Specifically, based on the image information of the guardrail, an image edge detection algorithm is used to obtain the deformation amount of the guardrail; based on the image information of the grate, an image segmentation algorithm is used to obtain the deformation amount of the grate and the proportion of the blocked area; based on the image information of the wheel, a contour extraction algorithm is used to obtain the deformation amount of the wheel and the wheel axle clearance.

[0036] Furthermore, the method for generating the sub-component health quantification data stream in step S3 includes: processing the sideboard image information, the grate image information, and the wheel image information to obtain dimensionless sideboard health quantification values, grate health quantification values, and wheel health quantification values.

[0037] Furthermore, the entire life cycle includes the trolley commissioning break-in period, the trolley normal operation period, the trolley maintenance and repair period, and the trolley scrapping assessment period. The preset stage operation and maintenance decision rules are predetermined based on the equipment status characteristics, operation and maintenance priority characteristics, and process requirement characteristics in the process characteristics of the cycle stage. The preset phased operation and maintenance decision rules specifically include: during the trolley's break-in period, generating periodic inspection instructions based on the health quantification values ​​of each sub-component; during the trolley's normal operation period, generating process parameter adjustment and maintenance instructions based on the optimized overall system diagnostic results and the health status of each sub-component; during the trolley's maintenance and repair period, generating maintenance strategy instructions based on abnormal component identification and abnormality level; and during the trolley's scrap assessment period, generating scrap assessment and component replacement instructions based on the development trend of the health quantification values ​​of each sub-component.

[0038] More preferably, a sub-component maintenance instruction data stream with target identifiers is generated based on the sub-component health quantification data stream and the constraint verification results of the process characteristics at each stage of the entire life cycle. Specifically, this includes: pre-setting health thresholds and constraint verification standards for each sub-component at different stages of the entire life cycle, wherein the constraint verification standards are pre-set based on the equipment status characteristics, maintenance priority characteristics, and process requirement characteristics in the process characteristics of each stage; comparing the sub-component health quantification value of each sub-component with the health threshold of the corresponding stage, and determining the sub-component health level based on the constraint verification results; generating corresponding sub-component maintenance instructions for sub-components with different health levels, wherein the sub-component maintenance instructions contain target identifiers, wherein the target identifiers correspond one-to-one with the unique identification code of the corresponding component in the digital twin, and summing all sub-component maintenance instructions to form a sub-component maintenance instruction data stream.

[0039] Furthermore, the maintenance operation completion status data stream includes maintenance operation type, operation time, operation parameters, and operation execution result. After the maintenance operation completion status data stream is mapped to a preset dynamic label field, the digital twin updates the preset dynamic label field of the corresponding component.

[0040] Preferably, this invention provides a digital twin-based full lifecycle operation and maintenance method for sintering machines, applicable to the intelligent operation and maintenance of sintering machine trolleys from commissioning, operation, maintenance to scrapping. The digital twin is a comprehensive virtual entity with a physical information neural network (PINN) and physical mechanism model as its core and a geometric model as its form carrier. The comprehensive virtual entity is configured with inherent attributes and dynamic attributes. The inherent attributes include the unique identification code of each sintering trolley and its components (e.g., the unique code of trolley A-101 is SJ-T-2026001, its side panel component code is SJ-T-2026001-LB01, and its wheel axle component code is SJ-T-2026001-LZ01). The preset dynamic label fields are pre-configured based on the monitoring dimensions, health calculations, and constraint verification requirements of each component of the sintering machine, and include at least a component identity field, a monitoring data field, a health field, a constraint verification field, and an operation and maintenance record field. Each preset dynamic label field corresponds to a unique data type and data format (e.g., the deformation monitoring data field of the side panel is numerical). The digital twin's Physical Information Neural Network (PINN) is constructed using PINN (Physical Information Neural Network). The physical mechanism model is based on the pre-set physical operating laws and design parameters of the sintering machine and its components. Health metrics are all converted to a linear transformation from a dimensionless value of 0 to 100, resulting in a score of 0-100. Based on statistical analysis of historical fault data, a score ≥80 is considered normal (no significant deterioration), 60 ≤ score < 80 is considered sub-healthy (recommended for monitoring), and a score < 60 is considered abnormal (requiring intervention, such as shutdown for maintenance or replacement). The health weights of each component are preset as follows: panel weight. =0.4), grate weight =0.3), wheel and axle weight =0.3), and the sum of the weights is 1. This threshold can be dynamically adjusted according to the specific production line conditions.

[0041] It should be noted that the preset dynamic tag field is the core structured carrier for data interaction, status synchronization, and operation and maintenance decision-making between the digital twin and the physical entity of the sintering machine. Its core features are pre-configuration, scenario adaptation, and dynamic updates. It is not a fixed static field. In specific use, it can be deeply integrated with different operation and maintenance scenarios and data flows of each step throughout the entire life cycle to achieve scenario-based adaptation. Specifically, the preset dynamic tag field is for data matching of the digital twin, solving the problems of messy sintering machine data, inability to correspond with the virtual twin, and lack of basis for operation and maintenance. When integrating, the preset dynamic tag field is pre-configured based on the monitoring dimensions, health calculation requirements, and constraint verification requirements of each component of the sintering machine. It provides a unified data storage and association interface for all flowing data (single-dimensional standard data flow, sub-component health, diagnostic results, maintenance records, etc.), ensuring that the data is identifiable, associative, and traceable. At the same time, it can flexibly accept data from each step according to the scenario requirements of different stages of the entire life cycle, realizing the binding of physical entity data, digital twin data, and operation and maintenance operations. It is the key link connecting the physical entity and the virtual twin. Its core functions are reflected in four aspects: unified data collection to avoid chaos, with each field corresponding to a type of data, enabling data classification and storage, laying the foundation for orderly data management in digital twins; achieving virtual-physical correspondence, bridging the gap between physical and virtual systems, allowing each set of data from the physical trolley to accurately correspond to the corresponding component in the virtual twin, achieving real-time synchronization of their states; supporting precise decision-making, integrating scattered data to provide clear basis for operation and maintenance decisions, avoiding reliance on experience; and adapting to the entire lifecycle, allowing for flexible adjustments and additions based on the needs of different lifecycle stages, enabling full lifecycle data traceability, and adapting to the key points of operation and maintenance at each stage.

[0042] In the scenario implementation example of the break-in period for a new sintering trolley (A-101, unique code SJ-T-2026001), the new trolley A-101 (unique code SJ-T-2026001) is a newly commissioned sintering trolley, in the break-in period of its entire life cycle. The process characteristics of this stage are: equipment components are in the break-in phase, with no significant wear; the core maintenance requirements are regular inspections and parameter calibration to promptly identify initial installation deviations and potential hazards, avoiding component damage during the break-in process. The preset maintenance decision rules for this stage are high-frequency inspections + parameter calibration + light maintenance. The specific implementation steps are as follows: Step 1: The sensing layer collects multi-source heterogeneous raw data streams from the vehicle (A-101, code SJ-T-2026001), specifically including: collecting side panel image information at the first preset location point, collecting grate image information at the second preset location point, collecting wheel and axle image information at the third preset location point, and collecting air box negative pressure and air box temperature data at the fourth preset location point; all the above raw data are labeled with identification tags (side panel: SJ-T-2026001-LB01, grate: SJ-T-202600). 1-BT01, axle: SJ-T-2026001-LZ01, bellows: SJ-T-2026001-FX01), timestamp (UTC time) and physical location identifier (on-site installation coordinates); the 3σ criterion is used to remove abnormal data caused by sensor failure (such as invalid data where the bellows negative pressure is abnormally 0), and all data are mapped to the [0,1] interval through the min-max normalization method. After data completion, a single-dimensional standard data stream is generated to ensure that the data is standardized and identifiable.

[0043] Step 2: Based on the single-dimensional standard data stream output by S1, and combined with the unique identification codes of the trolley A-101 and each component in the digital twin (corresponding one-to-one with the identification identifiers in the data), identity binding, spatiotemporal alignment, and dynamic attribute mapping are performed sequentially: A hash matching algorithm is used to match the identification identifiers of each single-dimensional data with the unique codes of the corresponding components in the digital twin (e.g., SJ-T-2026001-LB01 matches the digital twin's side panel code, and SJ-T-2026001-LZ01 matches the digital twin's wheel axle code), thus achieving unique binding between data and components; the timestamps of all data are unified to the UTC reference time axis, and the physical location identifiers are converted into three-dimensional virtual coordinates of the digital twin (error ≤ 50mm), achieving spatiotemporal synchronization; the standardized data is injected into the dynamic label fields of the corresponding components according to type (e.g., side panel deformation is injected into the side panel monitoring-deformation field), and finally mapped to form a single twin state data stream, achieving accurate correspondence between physical data and the virtual model.

[0044] Step 3: The digital twin invokes the Physical Information Neural Network (PINN Neural Network Constraint) and the physical mechanism model. Using the single twin's state data stream as input, the physical mechanism model first verifies the data's rationality (e.g., the maximum allowable deformation of the side panel is preset to 5mm; verifying whether the measured deformation is within a reasonable range). Then, the Physical Information Neural Network (PINN) mines implicit data correlations (e.g., the correlation between wheel axle temperature and vibration), and calculates the health of each sub-component based on a preset health formula. Among them, the health of the parapet: the measured deformation was 0.8mm, and the maximum allowable deformation was 5mm. =0.84, converted to 84 points (normal); Among them, the health of the grates: the proportion of visually recognized blocked area is 2%. =0.98, converted to 98 points (normal); Among them, the wheel and axle health status is: measured temperature 65℃ (threshold ≤ 80℃), vibration 1.2mm / s (threshold ≤ 2.5mm / s), substituted into the composite function. =f(65,1.2)=0.92, which translates to 92 points (normal). The overall health score of the trolley was calculated using a weighted average: sideboard 84 points, grate 98 points, and wheel axle 92 points. With weights of 0.4, 0.3, and 0.3 respectively, the overall health score was calculated as 0.4×84 + 0.3×98 + 0.3×92 = 33.6 + 29.4 + 27.6 = 90.6 points (normal). Based on the above calculation results, a sub-component health measurement data stream (84 points, 98 points, 92 points) and an initial system diagnostic result data stream were generated. The diagnostic conclusion is that all components of trolley A-101 are operating normally and are in the normal running-in state during the break-in period, with no abnormal hidden dangers.

[0045] S4. Based on the sub-component health quantification data stream (all scores are normal) and the process characteristic constraint verification results during the commissioning break-in period, generate a sub-component maintenance instruction data stream: all components are in normal condition, generate inspection instructions, including target identifiers (corresponding to the codes of each component: side panel SJ-T-2026001-LB01, grate bar SJ-T-2026001-BT01, wheel axle SJ-T-2026001-LZ01, air box SJ-T-2026001-FX01), inspection items (side panel deformation, grate bar blockage, wheel axle temperature). Vibration and bellows parameters); combined with the initial system diagnostic results (no abnormalities), overall machine operating conditions (run-in period trolley running speed 1.0m / min) and break-in period process characteristics, generate a whole machine maintenance instruction data stream: calibrate bellows negative pressure and temperature parameters daily, check the installation tightness of side panels and grate bars weekly, and lubricate and maintain the wheel axles monthly; integrate the sub-component maintenance instructions with the whole machine maintenance instructions according to the target identifier to form an intelligent decision control instruction data stream, where the target identifiers all correspond to the unique code SJ-T-2026001 of trolley A-101 and the identifiers of each component.

[0046] Step 5: The execution layer receives the intelligent decision control command data stream and performs inspection, parameter calibration, and lubrication maintenance operations through the operation terminal. After the operation is completed, the maintenance operation completion status data stream is collected (including operation type, operation time, operation parameters, and execution results: no abnormalities in inspection, qualified parameter calibration, and lubrication in place). This data stream is mapped to the corresponding dynamic tag field of the digital twin, and the status data stream of the single twin and the dynamic attributes of the digital twin are updated to ensure that the digital twin is consistent with the status of the physical trolley A-101, forming a closed loop.

[0047] In the scenario implementation for the maintenance and overhaul period of the old trolley (A-056, unique code SJ-T-2023056), the old trolley A-056 (unique code SJ-T-2023056) has been in use for 3 years and is in the full life cycle maintenance and overhaul period. The process characteristics of this stage are: the equipment components have a certain degree of wear, and some components may be in abnormal condition. The core maintenance requirements are to identify abnormal components, repair and replace damaged components, and check the status of related components to prevent the fault from escalating. The preset maintenance decision rules for this stage are: anomaly location - fault repair - related detection - acceptance calibration. The specific implementation steps are as follows: Step 1: The sensing layer collects multi-source heterogeneous raw data streams of the vehicle (A-056, code SJ-T-2023056), specifically including: sideboard image information (first preset position point), grate image information (second preset position point), wheel axle temperature and vibration data (third preset position point), and air box negative pressure and temperature data (fourth preset position point); the above raw data are respectively labeled with identification marks (sideboard: SJ-T-2023056-LB01, grate: SJ-T-2023056-BT01, ...). Wheel axle: SJ-T-2023056-LZ01, bellows: SJ-T-2023056-FX01, timestamp (UTC time) and physical location identifier; abnormal wheel axle vibration data are removed using the 3σ criterion (e.g., a vibration value of 5.8 mm / s, exceeding the reasonable range of 3.0 mm / s, is removed), and the data is mapped to the [0,1] interval through min-max normalization processing, and the two missing sets of bellows temperature data are filled in (using linear interpolation method), finally generating a single-dimensional standard data stream.

[0048] Step two: Based on the single-dimensional standard data stream and the inherent attributes of the trolley A-056 in the digital twin (unique code SJ-T-2023056 and codes of each component), perform identity binding, spatiotemporal alignment, and dynamic attribute mapping: match each data identity identifier with the corresponding component code of the digital twin through hash matching to avoid data confusion; unify the timestamp to the UTC reference time axis, and convert the physical location identifier into virtual coordinates of the digital twin; inject standardized data into the dynamic label field of the corresponding component to generate a single twin state data stream, and realize real-time synchronization between physical data and virtual model.

[0049] Step 3: The digital twin invokes the Physical Information Neural Network (PINN) and physical mechanism model, using the single twin's state data stream as input, to perform single-dimensional state determination calculations. The health metrics of each sub-component are quantified as follows: Among them, the health of the parapet: the measured deformation was 3.2mm, and the maximum allowable deformation was 5mm. =0.36, converted to 36 points (abnormal), physical mechanism model verification shows that the deformation exceeds the sub-health threshold (60 points), physical information neural network (PINN) found that the deformation is showing a continuous upward trend, which indicates that the railing is severely deformed and needs to be repaired immediately; Among them, the health of the grates: visual recognition shows a blockage area ratio of 35%. =0.65, converted to 65 points (sub-health), indicating severe blockage of the baffle, which needs to be cleaned; Among them, the wheel and axle health status is as follows: measured temperature 88℃ (threshold ≤ 80℃), vibration 3.2mm / s (threshold ≤ 2.5mm / s), substituted into the composite function. =0.28, converted to 28 points (abnormal), judged as excessive wheel axle temperature and vibration, with potential wear risk, requiring inspection; Meanwhile, the negative pressure of the bellows was detected to be 13.5 kPa (reasonable range 8-12 kPa) and the temperature was 138℃ (reasonable range 100-130℃). The dimensionless value was calculated to be 0.32, which was converted to 32 points (abnormal). This indicates that attention should be paid to abnormal process parameters, which may be related to air leakage caused by deformation of the baffle or blockage of the grate.

[0050] The overall health score of the trolley is 38.85 (abnormal). Generate a sub-component health quantification data stream (36 points, 65 points, 28 points) and an initial system diagnostic result data stream. The diagnostic conclusion is that the A-056 trolley has severe side panel deformation, abnormal wheel axle, and sub-healthy grate bars. At the same time, the excessive air box parameters indicate that there may be a related air leakage problem.

[0051] Step 4: Based on the quantitative data stream of sub-component health (abnormal and sub-health scores) and the constraint verification results of the process characteristics during the maintenance and repair period, generate the sub-component maintenance instruction data stream: For the side panel (36 points), generate instructions to replace the side panel and calibrate the installation position after replacement (target identifier SJ-T-2023056-LB01); for the grate bar (65 points), generate instructions to thoroughly clean the grate bar blockage and check the grate bar deformation (target identifier SJ-T-2023056-BT01); for the wheel axle (28 points), generate instructions to disassemble and repair the wheel axle, replace the worn bearing, and calibrate the wheel axle clearance (target identifier SJ-T-2023056-LZ01); all instructions include operation priority (wheel axle and grate bar are first-level priority and executed immediately; side panel is second-level priority and executed during maintenance intervals) and operation time limit.

[0052] By combining the sub-component health measurement data stream, the initial system diagnostic results (machine abnormality), the machine's operating conditions (maintenance trolley out of service), and the process characteristics during the maintenance and repair period, a machine maintenance instruction data stream is generated: Trolley A-056 is shut down for maintenance, first performing priority maintenance on the axles, side panels, and air boxes, then cleaning the grate bars. After the maintenance is completed, the health of all components is retested, and all monitoring sensors are calibrated to ensure that the equipment returns to normal operation. The sub-component maintenance instructions and the machine maintenance instructions are matched and integrated according to the target identifier (trolley and component codes) to form an intelligent decision control instruction data stream, clarifying the operation process and division of responsibilities.

[0053] Step 5: The execution layer receives the intelligent decision control command data stream, and the automated operation and maintenance equipment and operation and maintenance personnel jointly perform maintenance operations: First, clean the grate bar blockage, check for grate bar detachment, deformation and damage (if there is no detachment or damage, no replacement is required), disassemble the wheel axle; if the wheel is detached, it needs to be replaced in time, then replace the worn bearing, and calibrate the wheel axle clearance; finally, replace the deformed side panel, calibrate the installation position; check for air box leaks and perform sealing repairs. After the operation is completed, the maintenance operation completion status data stream is collected (including maintenance type, operation time, replacement part model, operation parameters, execution results: wheel axle temperature 62℃, vibration 1.1mm / s, sideboard deformation 0.5mm, air box negative pressure 11kPa, temperature 120℃, grate blockage area ratio 5%). This data stream is mapped to the corresponding dynamic label field of the digital twin, and the status data stream of the single twin is updated (such as wheel axle health updated to 86 points, sideboard health updated to 88 points) and the dynamic attributes of the digital twin to ensure that the virtual model is consistent with the physical trolley A-056 after maintenance.

[0054] This invention also provides a digital twin-based full lifecycle operation and maintenance system for sintering machines. It includes the perception layer for communication connectivity, the network transmission layer, the digital twin application layer, and the execution layer; The sensing layer is used to collect multi-source heterogeneous monitoring data of the physical entity of the sintering machine, mark it with timestamp information, and then output it to the network transmission layer. The sensing layer includes an identity recognition unit, a physical state acquisition unit, and a physical process parameter acquisition unit. The network transport layer is used to realize bidirectional data transmission of multi-source heterogeneous monitoring data and control commands; The digital twin application layer is the intelligent decision-making center. The digital twin application layer has a built-in sintering machine digital twin with a physical information neural network (PINN) and physical mechanism model as the intelligent core, a geometric model as the form carrier, and real-time data fusion. The digital twin is a fusion of data intelligence and physical intelligence. The digital twin application layer is used to realize the virtual-real mapping of physical entities and multi-dimensional intelligent decision-making, and output control commands to the network transmission layer. The execution layer is used to receive control commands and execute corresponding sintering machine operation and maintenance operations, and to send the operation completion status data back to the network transmission layer.

[0055] The present invention also provides a digital twin-based sintering machine full lifecycle operation and maintenance medium, which stores a computer program. When the computer program is executed, it implements the above-mentioned digital twin-based sintering machine full lifecycle operation and maintenance method.

[0056] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for the full lifecycle operation and maintenance of a sintering machine based on digital twins, characterized in that, To achieve intelligent operation and maintenance of the sintering machine throughout its entire lifecycle, the digital twin is a comprehensive virtual entity with a physical information neural network and a physical mechanism model as its core and a geometric model as its form carrier. This comprehensive virtual entity is configured with inherent attributes and dynamic attributes. The inherent attributes include unique identification codes for each sintering trolley and its components, and include the following steps: S1. Obtain the single-dimensional standard data stream corresponding to each monitoring dimension of the physical entity of the sintering machine. Each single-dimensional standard data stream in the single-dimensional standard data stream is marked with identity information, timestamp information and physical location identifier. S2. Based on the single-dimensional standard data stream and the inherent attribute information, perform identity binding, spatiotemporal alignment and dynamic attribute mapping in sequence to map each of the single-dimensional standard data streams into a single twin state data stream; S3. The digital twin calls the physical information neural network and the physical mechanism model of the kernel, performs single-dimensional state judgment calculation based on the state data stream of each single twin, generates sub-component health measurement data stream of each sub-component, and generates initial system diagnosis result data stream of preliminary diagnosis conclusion of the whole machine based on the sub-component health measurement data stream of each component. S4. Generate a sub-component maintenance instruction data stream with target identifier based on the sub-component health quantification data stream and the constraint verification results of the process characteristics of each cycle stage in the entire life cycle; A multi-dimensional correlation analysis is performed on the sub-component health quantification data stream and the initial system diagnostic result data stream to obtain optimized whole system diagnostic results. Then, combined with data coupling correlation, whole machine operating conditions and the cycle stage process characteristics of the current life cycle stage, a whole machine maintenance instruction data stream with target identifier is generated. The sub-component maintenance instruction data stream and the whole machine maintenance instruction data stream are combined to form an intelligent decision control instruction data stream, and the target identifier corresponds one-to-one with the unique identity code of each vehicle and component in the digital twin; S5. Respond to the intelligent decision control command data stream to perform maintenance operations, collect the maintenance operation completion status data stream corresponding to the execution of the intelligent decision control command data stream, map the maintenance operation completion status data stream to the preset dynamic label field corresponding to the digital twin, and update the status data stream of each individual twin and the dynamic attributes of the digital twin.

2. The sintering machine full lifecycle operation and maintenance method based on digital twin as described in claim 1, characterized in that, Step S1 specifically includes: Collect multi-source heterogeneous raw data streams corresponding to multiple monitoring dimensions of the physical entity of the sintering machine, and mark each single dimension of the multi-source heterogeneous raw data stream with identity information, timestamp information and physical location identifier respectively; Anomaly removal and normalization are performed on the single-dimensional data stream in the multi-source heterogeneous original data stream to obtain a single-dimensional standard data stream.

3. The sintering machine full lifecycle operation and maintenance method based on digital twin as described in claim 2, characterized in that, Obtain the image information of the sideboard of the physical trolley at a first preset position point, and obtain the deformation amount of the sideboard based on the image information of the sideboard; The grate image information of the physical trolley is obtained at the second preset position point, and the grate deformation amount and blockage area ratio are obtained based on the grate image information; At the third preset position point, obtain the wheel image information of the physical trolley, and obtain the wheel deformation and wheel axle clearance based on the wheel image information; At the fourth preset location point, the wind box status information corresponding to the physical trolley during sintering is obtained. The wind box status information includes at least the wind box negative pressure and the wind box temperature. The deformation of the guardrail, the deformation of the grate bars, the proportion of the blocked area, the deformation of the wheel, the wheel axle clearance, the negative pressure of the air box, and the temperature of the air box are all single-dimensional data and are respectively marked with identity information, timestamp information and physical location identifier.

4. The sintering machine full lifecycle operation and maintenance method based on digital twin as described in claim 3, characterized in that, The generation method of the sub-component health metric data stream in step S3 includes: The image information of the guardrail, the image information of the grate, and the image information of the wheel are processed to obtain dimensionless quantified values ​​of guardrail health, grate health, and wheel health.

5. The sintering machine full lifecycle operation and maintenance method based on digital twin as described in claim 1, characterized in that, The entire life cycle includes the trolley commissioning and break-in period, the trolley normal operation period, the trolley maintenance and repair period, and the trolley scrapping assessment period. The preset stage operation and maintenance decision rules are predetermined based on the equipment status characteristics, operation and maintenance priority characteristics, and process requirement characteristics in the process characteristics of the cycle stage. The preset phase operation and maintenance decision rules specifically include: During the break-in period of the trolley, periodic inspection instructions are generated based on the health quantification values ​​of each sub-component. During the normal operation of the trolley, process parameter adjustment and maintenance instructions are generated based on the optimized whole system diagnostic results and the health status of each sub-component. During the maintenance and repair period of the trolley, maintenance strategy instructions are generated based on the abnormal component identification and the degree of abnormality. During the trolley's scrap assessment period, scrap assessment and component replacement instructions are generated based on the development trend of the health quantification values ​​of each sub-component.

6. The sintering machine full lifecycle operation and maintenance method based on digital twin as described in claim 1, characterized in that, The maintenance operation completion status data stream includes maintenance operation type, operation time, operation parameters, and operation execution result. After the maintenance operation completion status data stream is mapped to a preset dynamic label field, the digital twin updates the preset dynamic label field of the corresponding component.

7. A sintering machine full lifecycle operation and maintenance system based on digital twins, characterized in that, It includes the perception layer for communication connectivity, the network transmission layer, the digital twin application layer, and the execution layer; The sensing layer is used to collect multi-source heterogeneous monitoring data of the physical entity of the sintering machine, mark it with timestamp information, and then output it to the network transmission layer. The sensing layer includes an identity recognition unit, a physical state acquisition unit, and a physical process parameter acquisition unit. The network transport layer is used to realize bidirectional data transmission of multi-source heterogeneous monitoring data and control commands; The digital twin application layer is the intelligent decision-making center. The digital twin application layer has a built-in sintering machine digital twin with physical information neural network and physical mechanism model as intelligent kernel, geometric model as form carrier and real-time data fusion. The digital twin is a fusion of data intelligence and physical intelligence. The digital twin application layer is used to realize the virtual-real mapping of physical entities and multi-dimensional intelligent decision-making, and output control commands to the network transmission layer. The execution layer is used to receive control commands and execute corresponding sintering machine operation and maintenance operations, and to send the operation completion status data back to the network transmission layer.

8. A sintering machine lifecycle operation and maintenance medium based on digital twins, characterized in that, The system contains a computer program that, when executed, implements the digital twin-based full lifecycle operation and maintenance method for sintering machines as described in any one of claims 1 to 7.

Images