A digital-twin-model-based hoisting machinery operation and maintenance management system and method

Abstract

The present disclosure provides a kind of hoisting machinery operation and maintenance management system and method based on digital twin model, system includes: physical layer, for the running data of key structure of hoisting machinery is collected by sensor;Data layer is used for pre-processing and storage to running data;Model layer is used for constructing and managing the digital twin model of hoisting machinery, and based on fault diagnosis model and life prediction model respectively, analyze running data, output fault diagnosis result and life prediction result;Interaction layer is used for hierarchical early warning and maintenance strategy output according to fault diagnosis result and life prediction result, carry out fault display on digital twin model.The present disclosure is accurate to extract and analyze the fault characteristics in running data by running data acquisition and cooperate fault diagnosis model, improve fault diagnosis accuracy;Combined with life prediction result, hierarchical early warning and dynamic optimization of maintenance strategy are carried out, realize fault display based on digital twin model, reduce maintenance difficulty.

Description

Technical Field

[0001] This disclosure relates to the field of crane operation and maintenance technology, and in particular to a crane operation and maintenance management system and method based on a digital twin model. Background Technology

[0002] As core special equipment, the operational safety and reliability of lifting machinery are directly related to the safety of life and property and the efficiency of production and operation. Traditional operation and maintenance of lifting machinery mainly adopts the "periodic maintenance" and "reactive maintenance" models. However, with the development of lifting machinery towards larger, more complex, and higher-load directions, the traditional operation and maintenance model faces multiple prominent problems: Under the limitations of passive operation and maintenance, "periodic maintenance" is prone to over-maintenance or under-maintenance, while "reactive maintenance" results in long unplanned downtime, which can easily lead to serious safety accidents and production losses. Some existing crane condition monitoring systems have limited functions and can only realize simple over-limit alarms. Some operation and maintenance methods use sensor data collection combined with simple algorithms for fault diagnosis, but they have the following shortcomings: First, the data collection dimension is single, resulting in low diagnostic and prediction accuracy; second, the data processing depth is insufficient, lacking the ability to perform big data correlation analysis, making it difficult to uncover the potential fault patterns behind the data; third, it is impossible to visualize faults, and maintenance personnel cannot know the actual fault location in a timely manner, affecting operation and maintenance efficiency. Summary of the Invention

[0003] The purpose of this disclosure is to provide a crane machinery operation and maintenance management system and method based on a digital twin model to solve the problems existing in the prior art.

[0004] The embodiments of this disclosure adopt the following technical solution: a crane machinery operation and maintenance management system based on a digital twin model, comprising: a physical layer for collecting operational data of key structures of the crane machinery through sensors; a data layer for preprocessing and storing the operational data; a model layer for constructing and managing a digital twin model of the crane machinery, and analyzing the operational data based on a fault diagnosis model and a life prediction model, respectively, and outputting fault diagnosis results and life prediction results; and an interaction layer for outputting graded early warning and maintenance strategies based on the fault diagnosis results and the life prediction results, and displaying faults on the digital twin model.

[0005] In some embodiments, the sensors include at least: vibration sensors, strain sensors, temperature sensors, motor current sensors, oil sensors, and environmental sensors; the key structures include at least: boom, braking system, hoisting mechanism, traveling mechanism, and hydraulic system.

[0006] In some embodiments, the data layer includes: a preprocessing unit, configured to perform at least one of outlier removal, noise reduction, time alignment, and compression processing on the running data; a data transmission unit, configured to transmit the preprocessed running data to the model layer via any one of a 5G network, a local area network, or an industrial Ethernet; and a data storage unit, configured to store the running data via a distributed time-series database.

[0007] In some embodiments, the model layer includes: a 3D model management unit, configured to construct a digital mapping of the geometric shape of the lifting machinery to form a 3D geometric model of the lifting machinery; a digital twin model management unit, configured to construct a digital twin model of the lifting machinery based on the 3D geometric model, the physical properties and behavioral characteristics of the lifting machinery, and to perform real-time status synchronization of the digital twin model based on the operating data; a fault diagnosis unit, configured to establish and train a fault diagnosis model, the fault diagnosis model being configured to output fault diagnosis results based on the operating data, the fault diagnosis results including at least fault type, fault location and severity; and a life prediction unit, configured to establish and train a life prediction model, the life prediction model being configured to output life prediction results based on the operating data.

[0008] In some embodiments, the fault diagnosis model is established and trained based on the following steps: establishing a convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network; acquiring historical fault data and historical benchmark data during normal operation of the crane machinery, constructing a multi-source training dataset covering the full state of the crane machinery, and preprocessing the multi-source training dataset; dividing the preprocessed multi-source training dataset into a training set and a test set; setting a maximum number of iterations, training the convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network based on the training set until the number of iterations meets the maximum number of iterations; evaluating the performance of the trained convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network based on the test set, and using the evaluated convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network as the fault diagnosis model.

[0009] In some embodiments, the interaction layer includes: an early warning unit, configured to determine an early warning level and a response process based on the fault diagnosis results; a strategy generation unit, configured to output a maintenance strategy based on the fault diagnosis results, the early warning level, the life prediction results, and the current operating conditions; and an interactive display unit, configured to display the fault on the digital twin model based on the fault diagnosis results, the early warning level, the response process, and the maintenance strategy.

[0010] In some embodiments, the early warning unit is specifically used to: determine an early warning level based on a preset early warning classification rule, according to the fault type, the fault location, and the severity, wherein the early warning levels include: Level 1, Level 2, Level 3, and Level 4; when the early warning level is Level 1, the response flow output by the early warning unit is to display the fault based on the digital twin model; when the early warning level is Level 2, the response flow output by the early warning unit is to provide text warnings and push maintenance strategies based on the response flow of Level 1; when the early warning level is Level 3, the response flow output by the early warning unit is to provide audible and visual alarms based on the response flow of Level 2; and when the early warning level is Level 4, the response flow output by the early warning unit is to trigger a safe shutdown based on the response flow of Level 3.

[0011] In some embodiments, the strategy generation module is specifically used to: construct and manage a preset maintenance strategy database; search the preset maintenance strategy database for maintenance strategies corresponding to the fault diagnosis results, the early warning level, the life prediction results, and the current operating conditions, and output them.

[0012] In some embodiments, the interaction layer further includes: a remote maintenance unit for remote collaborative consultations between equipment manufacturers, users, and third-party maintenance providers, and for storing remote collaborative consultation records based on blockchain.

[0013] This disclosure also provides a method for operation and maintenance management of lifting machinery based on a digital twin model, comprising: deploying sensors on the key structures of the lifting machinery to collect operational data of the key structures; preprocessing and storing the operational data; constructing and managing a digital twin model of the lifting machinery, and analyzing the operational data based on a fault diagnosis model and a life prediction model, respectively, and outputting fault diagnosis results and life prediction results; outputting graded early warning and maintenance strategies based on the fault diagnosis results and the life prediction results, and displaying faults on the digital twin model.

[0014] The beneficial effects of this disclosure are as follows: by collecting operational data of key structures of lifting machinery, and combining it with a fault diagnosis model to accurately extract and analyze fault features in the operational data, the accuracy of fault diagnosis is improved; at the same time, by combining life prediction results to perform graded early warning and dynamic optimization of maintenance strategies, and realizing fault display based on a digital twin model, the visualization and virtual-real linkage control are completed to improve operation and maintenance efficiency and reduce maintenance difficulty. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the structure of the lifting machinery operation and maintenance management system based on a digital twin model in the first embodiment of this disclosure; Figure 2 This is a schematic diagram of the data layer structure in the first embodiment of this disclosure; Figure 3 This is a schematic diagram of the model layer structure in the first embodiment of this disclosure; Figure 4 This is the model structure of the fault diagnosis model in the first embodiment of this disclosure; Figure 5 This is a schematic diagram of the interaction layer structure in the first embodiment of this disclosure; Figure 6 This is a flowchart of the crane operation and maintenance management method based on a digital twin model in the second embodiment of this disclosure. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification 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 this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0018] To address the problems existing in the prior art, the first embodiment of this disclosure provides a crane machinery operation and maintenance management system based on a digital twin model, the structural diagram of which is shown below. Figure 1 As shown, it can be mainly configured at the construction site of lifting machinery and its control center, facilitating on-site operation and maintenance personnel to view and operate it. For example... Figure 1As shown, the system is mainly constructed from the following layers: physical layer 10, data layer 20, model layer 30, and interaction layer 40, forming a four-layer closed-loop architecture of "physical-data-model-interaction". It realizes the full-process coverage of lifting machinery from physical perception to intelligent operation and maintenance from bottom to top. Among them, physical layer 10 is used to collect the operation data of key structures of lifting machinery through sensors; data layer 20 is used to preprocess and store the operation data; model layer 30 is used to build and manage the digital twin model of lifting machinery, and analyze the operation data based on the fault diagnosis model and the life prediction model, respectively, and output the fault diagnosis results and life prediction results; interaction layer 40 is used to output graded early warning and maintenance strategies based on the fault diagnosis results and life prediction results, and display the fault on the digital twin model.

[0019] Specifically, physical layer 10 is the sensing source for intelligent operation and maintenance of the system in this embodiment. It directly connects to the actual lifting machinery and its supporting hardware through various sensors, responsible for comprehensively and accurately collecting all dimensions of equipment operating status information, providing raw data support for subsequent fault diagnosis and lifespan prediction. The lifting machinery in this embodiment can be any type of actually operating crane, including but not limited to bridge cranes and tower cranes; the sensors include at least vibration sensors, strain sensors, temperature sensors, motor current sensors, hydraulic sensors, environmental sensors, and other sensors of various types; the key structures include at least the boom, braking system, hoisting mechanism, traveling mechanism, hydraulic system, and other core components that are prone to failure or are essential for achieving the lifting function during actual crane operation.

[0020] It is important to note that when selecting key structures, choosing sensors, and placing them on cranes, the actual functions and working environment of the crane should be considered for targeted deployment. For example, for the crane boom, vibration sensors and strain sensors should be prioritized to collect vibration signals and stress-strain data during lifting and luffing processes, monitoring boom fatigue damage and structural deformation. Temperature sensors should also be deployed to monitor temperature changes in bearings and pins at boom connections in real time. For the braking system, temperature and vibration sensors should be placed at the brake disc and brake shoes to collect temperature rise and vibration frequency during braking, assessing the wear level and braking performance. Motor current sensors should be used to monitor current changes in the brake motor, assisting in determining braking effectiveness. The data acquisition frequency and range of all sensors should be set according to the actual performance of the machinery, ensuring all sensors have industrial-grade protection and support electromagnetic interference resistance to guarantee the stability and reliability of data acquisition in complex industrial environments.

[0021] In some embodiments, the physical layer 10 may further include a gateway device and a terminal device, wherein the gateway device serves as a communication hub between the physical layer and the data layer, and is responsible for collecting data from various sensors and realizing preliminary data forwarding at the edge; the terminal device can serve as an operating terminal for on-site maintenance personnel, and is used to facilitate maintenance personnel to view the basic status of the equipment and perform emergency operations.

[0022] Data layer 20 is the bridge connecting physical layer 10 and model layer 30. It is responsible for data preprocessing, transmission and storage, ensuring the accuracy and reliability of subsequent fault analysis and life prediction. Figure 2 The diagram shows the structure of the data layer 20, which mainly includes: a preprocessing unit 21, used to perform at least one of the following processing on the running data: outlier removal, noise reduction, time alignment, and compression; a data transmission unit 22, used to transmit the preprocessed running data to the model layer via any one of 5G network, local area network, or industrial Ethernet; and a data storage unit 23 used to store the running data via a distributed time-series database.

[0023] In this embodiment, the preprocessing unit 21 can remove outlier data using the isolated forest algorithm and denoise the signal using wavelet transform. Since the sampling triggering mechanisms of different sensors may differ, resulting in inconsistent data timestamps, the preprocessing unit 21 then aligns the time axes of multi-source sensor data based on dynamic time warping. This facilitates subsequent models capturing features between data at the same timestamp. Finally, adaptive coding is used to compress the data, reducing transmission and storage costs. In practical implementation, the preprocessing unit 21 can also use the 3σ principle to identify outlier data and perform secondary verification using the Grubbs test. For confirmed outliers, linear interpolation of adjacent data is used to complete the data, preventing single-point anomalies from affecting the overall analysis results. During denoising, high-frequency noise in vibration and strain signals can be removed using wavelet transform, while stable signals such as temperature and current can be directly denoised using the moving average method.

[0024] To adapt to the network environment requirements of different application scenarios, the data transmission unit 22 supports multiple data transmission methods such as 5G network transmission, local area network transmission, and industrial Ethernet transmission. When actually building the operation and maintenance system, the data transmission method can be selected independently based on the working environment of the crane and the surrounding network conditions; this embodiment does not impose specific restrictions. The actual communication protocol can adopt the OPCUA communication protocol, ensuring cross-device and cross-level data compatibility through unified data interaction specifications. The data storage unit 23 uses a distributed time-series database as the core storage engine, supporting horizontal expansion of storage nodes to achieve the storage of massive amounts of high-concurrency time-series data. Combined with the characteristics of the crane's operating data, it performs time-stamp-based association storage of operating data during operation, subsequent fault diagnosis results and life prediction results output by model analysis, maintenance strategies, and other data. It can also synchronously store the raw operating data collected by various sensors for easy and rapid retrieval.

[0025] Figure 3 The diagram shows the structure of the model layer 30 in this embodiment, which mainly includes a 3D model management unit 31, a digital twin model management unit 32, a fault diagnosis unit 33, and a lifespan prediction unit 34. The model layer 30 in this embodiment is key to achieving intelligent and predictive operation and maintenance, completing fault diagnosis, lifespan prediction, and virtual-physical linkage based on digital twins and big data algorithms.

[0026] Specifically, the 3D model management unit 31 is used to construct a digital mapping of the geometric shape of the lifting machinery, forming a 3D geometric model of the lifting machinery. This 3D geometric model is the basic visual carrier of the digital twin. In actual implementation, a laser 3D scanner can be used to perform a full-size scan of the lifting machinery to obtain 3D point cloud data of each component of the equipment. At the same time, the design drawings of the lifting machinery are used as a supplement and calibration basis for the point cloud data. Then, reverse engineering software is used to preprocess the point cloud data (denoising, simplification, alignment), and then surface reconstruction and solid modeling are performed to generate a 3D geometric model containing all key structures such as the boom, braking system, and hoisting mechanism. When the actual mechanical structure is repaired or modified, the corresponding 3D geometric model should also be updated in a timely manner to ensure the consistency between the model and the physical equipment.

[0027] Based on the three-dimensional geometric model, the digital twin model management unit 32 constructs a digital twin model of the lifting machinery by combining its physical attributes and behavioral characteristics. Simultaneously, it synchronizes the state of the digital twin model in real time using real-time operational data. Specifically, the physical attributes of the lifting machinery include, but are not limited to, the material parameters, structural parameters, and performance parameters of each structural component. The behavioral characteristics include, but are not limited to, the lifting trajectory, braking response characteristics, and load transfer patterns of each structural component. Combined with the geometric constraints imposed on each structural component by the three-dimensional geometric model, a digital twin model of the lifting machinery is formed. This digital twin model can be used for dynamic simulations such as boom stress distribution simulation and fault condition reproduction. The real-time collected operational data drives the dynamic updates of the digital twin model, achieving state synchronization between the "physical equipment and the virtual model."

[0028] The fault diagnosis unit 33 is used to establish and train a fault diagnosis model. The fault diagnosis model outputs fault diagnosis results based on operational data. The fault diagnosis results include at least the fault type, fault location, and severity. In this embodiment, the fault diagnosis model is a convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network (CNN-BiLSTM-Attention). Its model structure is as follows: Figure 4As shown, the system mainly includes an input layer, a CNN layer, a BiLSTM layer, an Attention layer, and a fully connected (FC) layer. The input layer receives multi-source, time-aligned running data. The CNN layer mainly consists of convolutional layers, pooling layers, and fully connected layers. The convolutional layers extract local features from the input data by setting convolutional kernels of different sizes. The pooling layers discard some network parameters according to predetermined rules, preserving the original features while reducing computational complexity and thus reducing the risk of overfitting. The fully connected layers integrate and reorganize the extracted local features to form global features. The BiLSTM layer can process both forward and backward information of the sequence. The forward LSTM is responsible for processing the input sequence from front to back to capture the forward context information. The backward LSTM processes the input sequence in reverse order, capturing backward contextual information to fully utilize the bidirectional dependencies in the sequence. The Attention layer (attention mechanism) is a strategy employed by the model when allocating information. By dynamically adjusting weights, it guides the model to select from various feature information, focusing on strongly correlated features and assigning them higher priority. The fully connected (FC) layer consists of 1-2 layers of fully connected neural networks, mapping the weighted feature vector output from the Attention layer to the task target space, and outputting fault diagnosis results based on the task type. In this embodiment, the fault diagnosis model utilizes CNN to extract local fault features from multi-source operational data (such as abnormal frequencies in vibration signals), captures the temporal dependencies of the data (such as the time trend of fault development) using BiLSTM, strengthens the weights of key fault features with the Attention mechanism, and finally outputs the fault diagnosis results.

[0029] In some embodiments, the fault diagnosis model is established and trained based on the following steps: First, a convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network is established; then, historical fault data of the lifting machinery is summarized, and benchmark data of normal equipment operation is supplemented to construct a multi-source training dataset covering the full state of the equipment. The multi-source training dataset is preprocessed, and this preprocessing should be consistent with the preprocessing performed on the operating data by the preprocessing unit 21 to ensure the consistency between the training data and the input data during actual fault diagnosis; then, the preprocessed multi-source training dataset is divided into a training set and a test set, with a ratio of 8:2; the training set is input into the CNN-BiLSTM-Attention model for training, and a maximum number of iterations is set. If the maximum number of iterations is not reached, the model continues to be trained; if the maximum number of iterations is reached, the training is terminated; for the trained model, the test set is used as input data to evaluate the model performance, verify the model's generalization ability, and ensure that the operational accuracy requirements are met. Finally, the evaluation-based convolutional neural network model based on an attention mechanism combined with a bidirectional long short-term memory neural network is used as the fault diagnosis model.

[0030] In this embodiment, the life prediction unit 23 is used to establish and train a life prediction model. The life prediction model is used to output life prediction results based on operating data. It should be noted that the life prediction model in this embodiment can be used to predict the overall operating life of the crane machinery, or it can be used to predict the life of one or more key components of the crane machinery. Alternatively, multiple life prediction models can be set up simultaneously to achieve life prediction for multiple subjects. This embodiment does not impose specific limitations. The life prediction model can be implemented based on a Long Short-Term Memory (LSTM) network. This model can effectively capture long-term dependencies in time-series data and is suitable for nonlinear prediction of the life decay process of crane machinery or key structures. For the life prediction model, its training data mainly selects key features related to life decay as model inputs, such as cumulative operating time, cumulative load, maximum stress and strain values, cumulative temperature exceeding the standard duration, and oil deterioration degree. Historical life data of key structures of the crane machinery (such as the fatigue life of the boom, the wear life of the brake pads, etc.) are collected. Combined with the corresponding operating data, a life prediction training dataset is constructed. A training process similar to that of the fault diagnosis model is adopted. The trained model can output the remaining service life of the key structure as the life prediction result.

[0031] Figure 5 The diagram shows the structure of the interaction layer 40 in this embodiment, which mainly includes: an early warning unit 41, used to determine the early warning level and response process based on the fault diagnosis results; a strategy generation unit 42, used to output maintenance strategies based on the fault diagnosis results, early warning level, life prediction results and current operating conditions; and an interactive display unit 43, used to display faults on the digital twin model based on the fault diagnosis results, early warning level, response process and maintenance strategies.

[0032] Specifically, the early warning unit 41 determines the early warning level based on preset early warning classification rules, according to the fault type, fault location, and severity, and triggers the corresponding response process. This response process is used to instruct the interactive display unit 43 on the method and effect of displaying the fault. The preset early warning classification rules used in this embodiment to determine the early warning level are constructed by combining the current type of lifting machinery, working scenario, task requirements, and expert experience. They are constructed by pre-configuring the correspondence between different fault types, different fault locations, and different severity levels and early warning levels. When the early warning unit 41 outputs the actual early warning level, it only needs to combine the pre-configured corresponding rules to determine the corresponding early warning level under the current fault type, fault location, and severity.

[0033] In this embodiment, the warning levels are divided into four levels, and the classification criteria are as follows: (1) Level 1 warning (minor warning): The fault type is minor wear, minor leakage and other faults that do not affect the normal operation of the equipment. The severity is minor. The corresponding response process is to display the fault based on the digital twin model. (2) Level II warning (general warning): The fault type is moderate wear, local abnormal noise, etc., which may affect the operating efficiency of the equipment but do not endanger safety. If it is not dealt with for a long time, the risk will be amplified. The severity is moderate. The corresponding response process is to push text warning and maintenance strategy based on the response process of Level I warning. (3) Level 3 warning (serious warning): The critical components of the equipment have failed, which has affected the safety performance. Continued operation may lead to the expansion of the failure. The severity is severe. The corresponding response process is to issue an audible and visual alarm based on the response process of Level 2 warning. (4) Level 4 warning (emergency warning): The core components of the equipment are faulty or there are major safety hazards. A safety accident may occur at any time. The severity is extremely serious. The corresponding response process is to trigger a safe shutdown based on the response process of Level 3 warning.

[0034] The strategy generation module 42 is used to output maintenance strategies. It builds and manages a preset maintenance strategy database, which collects maintenance plans corresponding to different fault types, warning levels, remaining lifespans, and operating conditions of the lifting machinery. Each maintenance plan includes maintenance content (such as replacing parts, tightening bolts, and adding lubricant), maintenance cycle, required tools and spare parts, and maintenance safety precautions. The database is constructed using a multi-dimensional index of "fault type-warning level-remaining lifespan-operating condition". During actual matching, a fuzzy matching algorithm can be used to search the database for the most matching maintenance strategy based on real-time fault diagnosis results, warning levels, lifespan prediction results, and current operating conditions, facilitating rapid on-site maintenance by maintenance personnel. In some embodiments, the execution effect of the maintenance strategy (such as whether the fault has been eliminated and the equipment's operating time after maintenance) can be periodically fed back to the database. Reinforcement learning algorithms can then be used to iteratively optimize the maintenance strategy, improving its relevance and effectiveness.

[0035] The interactive display unit 43 uses a digital twin model to visualize faults, allowing maintenance personnel to intuitively understand the equipment status. For example, different colored highlights on the digital twin model distinguish fault locations, and clicking on a highlighted area displays detailed fault diagnosis results (fault type, severity, and occurrence time). Simultaneously, key operating parameters such as temperature, vibration, and stress are displayed in real-time on the corresponding components of the digital twin model, and can be presented as numerical trend graphs. Furthermore, the output maintenance strategy is presented step-by-step on the digital twin model interface, with model animations demonstrating key maintenance steps to help maintenance personnel quickly understand and execute maintenance tasks. In some embodiments, for level three and level four warnings, the interactive display unit 43 can push the corresponding fault content and maintenance strategy to the maintenance personnel's handheld terminal in real time, and promptly remind maintenance personnel to handle the issue through audible and visual alarms.

[0036] In some embodiments, the interaction layer 40 can also be used as a remote maintenance unit to provide a remote collaborative consultation platform for equipment manufacturers, users, and third-party maintenance organizations. It supports remote consultation, technical support, and decision-making communication among multiple parties, including equipment manufacturers, maintenance teams, and enterprise management departments, based on the same platform. All maintenance activity records, including fault reports, diagnostic processes, repair operations, replacement parts, and cost consumption, are encrypted and stored through blockchain. Based on its decentralized, tamper-proof, and traceable characteristics, the authenticity, integrity, and security of maintenance records are ensured, providing a solid guarantee for full-process traceability, responsibility identification, experience accumulation, and reuse, and significantly improving the standardization and management level of the maintenance process.

[0037] This embodiment has the following beneficial effects: (1) Comprehensive data acquisition dimensions: Multimodal monitoring sensors are used to cover all dimensions of equipment structure, operation, environment and location data, solving the problem of single data acquisition in existing technologies; (2) High fault diagnosis accuracy: Combining big data mining and CNN-BiLSTM-Attention model, fault features are accurately extracted, improving the accuracy of fault diagnosis; (3) Achieve predictive maintenance: Based on the remaining life prediction, the equipment life can be predicted in advance, and the maintenance strategy generation can achieve dynamic optimization, changing passive maintenance to proactive predictive maintenance; (4) Strong virtual-real linkage capability: The digital twin model supports dynamic updates and simulation analysis, and AR visualization and virtual-real linkage control improve operation and maintenance efficiency and reduce maintenance difficulty; (5) Collaborative operation and maintenance and traceability: Remote collaborative units enable multi-party collaboration, and blockchain storage ensures that operation and maintenance records are traceable, thereby improving the standardization of operation and maintenance processes.

[0038] Based on the same inventive concept, the second embodiment of this disclosure provides a method for operation and maintenance management of lifting machinery based on a digital twin model, the flowchart of which is shown below. Figure 6 As shown, it mainly includes: S1, Sensors are placed on the key structures of the lifting machinery to collect the operating data of the key structures; S2, preprocesses and stores the running data; S3 constructs and manages digital twin models of lifting machinery, and analyzes operational data based on fault diagnosis models and life prediction models, respectively, and outputs fault diagnosis results and life prediction results. S4 outputs graded early warnings and maintenance strategies based on fault diagnosis results and life prediction results, and displays the faults on the digital twin model.

[0039] The specific implementation process of the crane operation and maintenance management method in this embodiment has been described in detail in the first embodiment of this disclosure, and will not be repeated here.

[0040] This embodiment integrates multi-source data acquisition and preprocessing, feature extraction and fault identification based on deep learning, dynamic updating and simulation analysis of digital twin models, remaining life prediction and maintenance decision generation, hierarchical early warning and virtual-real linkage response, which can realize real-time perception, trend prediction and intelligent decision-making of the operating status of lifting machinery, forming a fully closed-loop intelligent operation and maintenance system of "perception-analysis-decision-execution".

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit them. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A crane machinery operation and maintenance management system based on a digital twin model, characterized in that, include: The physical layer is used to collect operational data of key structures of lifting machinery through sensors; The data layer is used for preprocessing and storing the operational data; The model layer is used to build and manage the digital twin model of the lifting machinery, and analyze the operating data based on the fault diagnosis model and the life prediction model, respectively, and output the fault diagnosis results and life prediction results. The interaction layer is used to output graded early warning and maintenance strategies based on the fault diagnosis results and the life prediction results, and to display the fault on the digital twin model.

2. The crane machinery operation and maintenance management system according to claim 1, characterized in that, The sensors include at least: vibration sensors, strain sensors, temperature sensors, motor current sensors, oil sensors, and environmental sensors; The key structure includes at least: boom, braking system, hoisting mechanism, traveling mechanism, and hydraulic system.

3. The crane machinery operation and maintenance management system according to claim 1, characterized in that, The data layer includes: The preprocessing unit is used to perform at least one of the following processes on the running data: outlier removal, noise reduction, time alignment, and compression. The data transmission unit is used to transmit the preprocessed running data to the model layer via any one of a 5G network, a local area network, or an industrial Ethernet. A data storage unit is used to store the running data through a distributed time-series database.

4. The crane operation and maintenance management system according to any one of claims 1 to 3, characterized in that, The model layer includes: The three-dimensional model management unit is used to construct the geometric shape digital mapping of the lifting machinery and form a three-dimensional geometric model of the lifting machinery. The digital twin model management unit is used to construct a digital twin model of the lifting machinery based on the three-dimensional geometric model, the physical properties and behavioral characteristics of the lifting machinery, and to synchronize the status of the digital twin model in real time based on the operating data. A fault diagnosis unit is used to establish and train a fault diagnosis model. The fault diagnosis model is used to output fault diagnosis results based on the operating data. The fault diagnosis results include at least the fault type, fault location, and severity. The lifetime prediction unit is used to establish and train a lifetime prediction model, which is used to output lifetime prediction results based on the operating data.

5. The crane machinery operation and maintenance management system according to claim 4, characterized in that, The fault diagnosis model is established and trained based on the following steps: Establish a convolutional neural network model based on attention mechanism combined with a bidirectional long short-term memory neural network model; Historical fault data and historical baseline data during normal operation of the lifting machinery are obtained, a multi-source training dataset covering the full state of the lifting machinery is constructed, and the multi-source training dataset is preprocessed. The preprocessed multi-source training dataset is divided into a training set and a test set; Set a maximum number of iterations, and train the attention-based convolutional neural network combined with the bidirectional long short-term memory neural network model based on the training set until the number of iterations meets the maximum number of iterations; The performance of the trained attention-based convolutional neural network combined with bidirectional long short-term memory neural network model is evaluated based on the test set, and the evaluation-qualified attention-based convolutional neural network combined with bidirectional long short-term memory neural network model is used as a fault diagnosis model.

6. The crane machinery operation and maintenance management system according to claim 4, characterized in that, The interaction layer includes: The early warning unit is used to determine the early warning level and response procedure based on the fault diagnosis results; The strategy generation unit is used to output a maintenance strategy based on the fault diagnosis result, the early warning level, the life prediction result, and the current operating condition. An interactive display unit is used to display faults on the digital twin model based on the fault diagnosis results, the early warning level, the response process, and the maintenance strategy.

7. The crane machinery operation and maintenance management system according to claim 6, characterized in that, The early warning unit is specifically used for: Based on preset early warning classification rules, the early warning level is determined according to the fault type, the fault location, and the severity. The early warning levels include: Level 1, Level 2, Level 3, and Level 4. When the warning level is Level 1, the response process output by the warning unit is to display the fault based on the digital twin model; When the warning level is Level II, the response process output by the warning unit is to provide text warnings and push maintenance strategies based on the response process of Level I warnings. When the warning level is Level 3, the response procedure output by the warning unit is to perform an audible and visual alarm based on the response procedure of Level 2 warning. When the warning level is level four, the response process output by the warning unit is to trigger a safe shutdown based on the response process of level three warning.

8. The crane machinery operation and maintenance management system according to claim 7, characterized in that, The strategy generation module is specifically used for: Build and manage a database of pre-defined maintenance strategies; The system searches the preset maintenance strategy database for maintenance strategies that correspond to the fault diagnosis results, the early warning level, the life prediction results, and the current operating conditions, and then outputs these strategies.

9. The crane machinery operation and maintenance management system according to claim 6, characterized in that, The interaction layer also includes a remote maintenance unit, which is used for remote collaborative consultations between equipment manufacturers, users and third-party maintenance providers, and stores remote collaborative consultation records based on blockchain.

10. A method for operation and maintenance management of lifting machinery based on a digital twin model, characterized in that, include: Sensors are placed on the key structures of the lifting machinery to collect operational data of the key structures; The operational data is preprocessed and stored; A digital twin model of the lifting machinery is constructed and managed, and the operating data is analyzed based on the fault diagnosis model and the life prediction model, respectively, and the fault diagnosis results and life prediction results are output. Based on the fault diagnosis results and the life prediction results, graded early warning and maintenance strategies are output, and the faults are displayed on the digital twin model.

Images