Intelligent monitoring method for fatigue life of engineering machinery bearing structure

Abstract

The invention belongs to the technical field of mechanical structure health monitoring and life prediction, and particularly relates to an engineering machinery bearing structure fatigue life intelligent monitoring method, which comprises the steps of 1, building a high-fidelity geometric model, 2, building a finite element model, 3, building a finite element simulation data set, 4, building a neural network agent model, and 5, building a neural network model. 5, working parameters and loading history are obtained; and 6, a visual operation and maintenance platform is integrated. According to the intelligent monitoring method for the fatigue life of the engineering machinery bearing structure, the evolution process of fatigue damage of the structure can be intelligently deduced based on the working parameters and the loading history of the structure, dynamic calculation of the residual life and real-time early warning of the service risk are achieved, calculation resources are saved, the fatigue life evaluation efficiency is improved, and the engineering machinery bearing structure fatigue life monitoring method is suitable for popularization and application. The problems of dependence on off-line simulation, response lag, discontinuous evaluation and the like in traditional structure life management are solved, the intelligent operation and maintenance level of the engineering machinery is effectively improved, and safe and efficient operation of the engineering machinery is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mechanical structure health monitoring and life prediction, and in particular to an intelligent monitoring method for fatigue life of an engineering machinery load-bearing structure. BACKGROUND

[0002] Engineering machinery key load-bearing components are prone to fatigue damage and potential fracture failure under the action of high-frequency start-stop operation and large-amplitude cyclic loading. Such fatigue damage is usually cumulative, hidden and sudden. Once the structure fails, it may cause equipment failure and even safety accidents, resulting in economic losses and operational risks. Therefore, fatigue damage assessment and life prediction of in-service engineering machinery key load-bearing components are key engineering requirements to ensure operational safety and equipment reliability.

[0003] Traditional structure life management mainly relies on regular maintenance and manual experience judgment. Existing monitoring methods generally have limited monitoring range, unstable data quality, high system cost, and few involve measuring point selection in public solutions. The general method still relies on experience rules or statistical indicators. For example, CN103557890A and CN109524139B are based on experience selection of key positions and do not involve structural mechanics analysis or inverse model optimization, which may lead to monitoring blind spots and difficulty in reflecting the damage evolution process. At the same time, existing methods lack effective damage characterization, visualization analysis and safety state determination process for monitoring data, making it difficult to support accurate life prediction and operation and maintenance decisions. Finite element simulation technology has high calculation accuracy, but it has long computation time and strong dependence on hardware resources, which cannot achieve on-site rapid evaluation. For engineering service components, the working conditions are often complex and the load is variable, and a single simulation cannot cover all possibilities. If each evaluation relies on re-simulation, the efficiency is extremely low, and the evaluation results lack timeliness. The existence of the above problems limits the improvement of the life management level of engineering machinery key load-bearing structures, and it is difficult to meet the operational requirements of equipment reliability and safety in various operating environments.

[0004] Although existing mainstream solutions have the framework of a digital twin visualization platform, most of them still stay at the level of dynamic display of real-time operation data, and lack an intelligent decision-making core based on damage theory and failure mode. For example, in CN119276728A disclosed on the Chinese patent website, the device performance is inferred through machine learning, but due to the lack of physical failure model support, it is difficult to ensure the accuracy and reliability of the results. For example, CN113190886A needs to call the CAE solver multiple times for DOE sampling, which is time-consuming and dependent on specific software environment, and lacks real-time performance and deployment flexibility. As can be seen, the functions of such platforms are limited to varying degrees, making it difficult to support efficient, accurate and intelligent life management requirements of key load-bearing structures. SUMMARY

[0005] Based on the existing technical problems, the application provides an engineering machinery load-bearing structure fatigue life intelligent monitoring method, which comprehensively uses finite element analysis, deep learning modeling, neural network deployment and visual human-computer interaction technology, integrates fatigue damage theory, and thus constructs an integrated solution with digital perception, intelligent deduction, autonomous decision and data closed loop capability, perfects the modeling process of digital twin, and provides a new solution for intelligent operation and maintenance of key load-bearing structures of engineering machinery.

[0006] The engineering machinery load-bearing structure fatigue life intelligent monitoring method provided by the application comprises the following steps:

[0007] Step two, establish a finite element model;

[0008] Step three, construct a finite element simulation dataset;

[0009] Step four, construct a neural network proxy model;

[0010] Step five, obtain working parameters and loading history;

[0011] Step six, integrate a visual operation and maintenance platform.

[0012] Preferably, in the step one, the geometric model is constructed by a three-dimensional modeling software, and then the geometric model is exported into a corresponding file format according to the requirements.

[0013] According to the above technical scheme, the relevant information of each component of the lifting appliance is obtained according to the design data, and then a three-dimensional model of the lifting appliance in the engineering machinery is constructed by using a three-dimensional modeling software, so as to ensure sufficient detail restoration degree and ensure that the model structure is consistent with the actual situation, and an original model capable of truly reflecting the mechanical behavior of the lifting appliance is obtained.

[0014] Preferably, the geometric model is imported into the finite element simulation software in the step two according to the file type requirements, and the geometric model is simplified.

[0015] According to the above technical scheme, the electrical equipment, hydraulic pipeline and accessory structure arranged in the geometric model of the lifting appliance are removed; the smaller size welds and components in the main beam and telescopic beam are modeled as a whole; the larger size welds between the web plates of the main beam and between the main beam and the plate are modeled as a separate component; the structures such as small threaded holes are ignored; and the small chamfers and round corners in the lifting appliance structure are modeled according to right angles.

[0016] Preferably, after the geometric model is simplified, the material properties are defined, then the structure in the geometric model is meshed and mesh independence verification is carried out, and finally the relevant boundary conditions are set.

[0017] Through the above technical solutions, the main force-bearing components in the lifting appliance structure are high-strength steel materials, the buffer pad material between the main beam and the telescopic beam is nylon 66, and the remaining components are all ordinary carbon structural steel. The material parameters of the corresponding materials are given, including Poisson's ratio, elastic modulus, density, yield limit, and tensile strength.

[0018] Preferably, in step three, the integral step is set according to the working state and working time, the appropriate damping form is selected according to the structural characteristics and analysis requirements, the transient dynamics analysis of the full cycle process of the geometric model is performed, and the load spectrum is obtained.

[0019] Preferably, the load spectrum is analyzed by Ansys workbench and nCode DesignLife fatigue analysis, and the fatigue life under the corresponding load is obtained.

[0020] The maximum load and the minimum load that may occur in the actual working condition of the geometric model are added with a certain redundancy to obtain the upper limit load and the lower limit load of the geometric model, and the simulation data set is input .

[0021] A sufficient number of data points are extracted between the set upper limit load and the lower limit load by using the hyper-Latin cube sampling, and the fatigue life corresponding to each data point is obtained through simulation analysis, and these input and output data are recorded in the form of a table to form a load-life mapping data set of finite element simulation , wherein, .

[0022] Through the above technical solutions, there are three different working modes of the lifting appliance due to different sizes of the containers in the lifting work. For each working mode, the maximum and minimum loads that may occur in the actual working condition are identified and added with a certain redundancy as the upper limit and lower limit of the simulation data set input, points are taken within the input range by hyper-Latin cube sampling, the corresponding fatigue life is obtained through transient dynamics and Ansys workbench and nCode DesignLife fatigue analysis, and three load-life mapping data sets of finite element simulation are constructed.

[0023] Preferably, the neural network architecture in step four is established by Python based on the load-life mapping data set constructed in step three. The training set and the test set are divided in proportion after normalization, the model is trained, verified and parameterized, and a load-life proxy model is constructed.

[0024] Through the above technical solution, appropriate optimizer and loss function are selected in the training process, and the size of hyperparameters such as learning rate, training round, batch size, etc. is set.

[0025] Preferably, in the training process, the loss function calculates the error between the predicted value of the neural network model and the load-life mapping data set The labels are back-propagated to update the parameters of the neural network model;

[0026] The neural network model adopts mean absolute error (MAE), mean square error (MSE), and coefficient of determination (R2) as evaluation indexes. As evaluation indexes, if the indexes do not meet the expectations, the size of hyperparameters is adjusted, when the accuracy of the neural network model meets the requirements, the neural network model is saved as a target file format, and the normalization information during training is manually saved in the checkpoint of the target file, so as to prevent prediction deviation in the inference stage caused by inconsistent normalization;

[0027] The model weight and structure in the target file are exported as a general model deployment format, and the normalization information in the checkpoint of the target file is recorded.

[0028] Through the above technical solution, the input specification during training can be correctly restored when the model is used for inference.

[0029] Preferably, the working parameters in step five include the load and cycle number of the geometric model;

[0030] The load history includes the whole process of load change experienced by the geometric model structure over time during the working process, and the cumulative load of the geometric model for each working cycle and the corresponding cycle number, wherein ;

[0031] Based on the finite element model established in step three, static analysis is carried out to calculate the dangerous points of the geometric model structure, a static inversion model of the neural network model is established, and the load borne by the geometric model structure is inverted through the stress at the dangerous points of the geometric model structure , forming a stress-load mapping data set of finite element simulation The neural network model architecture is established again through the stress-load mapping data set of finite element simulation The training of the static inversion model is completed;

[0032] When the static finite element inversion model and the actual dynamic load environment will no longer match, the sliding window is used to judge the stability of the monitored stress value: the stress fluctuation variance is calculated for the stress value in each window , and when the fluctuation variance is lower than the preset threshold value, it is determined that the geometric model structure is in the horizontal operation stage, at which time the stress mean value in the window is taken as the input, and the output is obtained by the statics inversion model as the load of the geometric model in the operation cycle , under the action of which, according to the stress change sequence recorded by the stress sensor, the cycle number is obtained by using the rain flow counting method, and the acquisition of the work parameters of a single operation cycle is completed;

[0033] Considering the maximum dynamic load of the geometric model operation cycle , the stage is accompanied by rapid stress rise, and the minimum dynamic load at the end of the stage , the stage is accompanied by rapid stress drop, so the transition point of the first-order derivative of stress from negative value to positive value is the boundary of the work cycle, and the first-order derivative of stress is obtained by real-time forward difference calculation on the stress sequence: , which is used to identify the work cycle boundary;

[0034] Or step five can input relevant information by the user through the human-computer interaction interface of the platform.

[0035] Through the above technical scheme, the minimum time span is set to avoid false touch identification, and the sliding window stability detection is combined to avoid false cycle identification. The loading history of each cycle is stored in the database for the platform to call, which is used for sling residual life calculation and visual display of equipment load condition. Also, the user can input relevant information through the human-computer interaction interface of the platform.

[0036] Preferably, the integrated visualization operation and maintenance platform in step six calculates the residual life and compares it with the life threshold to update the device state;

[0037] Through 3DMax, the geometric model in step one is converted in file format, the target file format is exported and imported into the main scene of Unity3D for visualization display;

[0038] At the same time, Unity3D can dynamically visualize the stress sequence collected in step five by means of XChart chart, and the background of Unity3D dynamically evaluates the residual life of the structure by using the Miner linear cumulative damage theory, wherein the formula of the Miner linear cumulative damage theory is:

[0039] ; ​

[0040] wherein, is the total fatigue damage, is the actual cycle number, is the allowable cycle number under the corresponding load;

[0041] damage for calculating the remaining life, wherein the remaining life is 1, it is considered that the geometric model is intact; the remaining life is 0, that is, failure occurs;

[0042] According to the calculated remaining life, the device state of the geometric model is updated in real time, and an alarm threshold is set, when the calculated remaining life does not drop to , the device state remains normal service, when it drops to , a fatigue failure risk warning is issued, and when the device is repaired or replaced, the remaining life is updated to 1, and the device state is updated to normal service.

[0043] Through the above technical scheme, the geometric model is imported into the main scene of Unity3D, C# scripts are written to realize the rotation and attitude reset of objects in the three-dimensional scene with the help of the Unity engine, and it is convenient for users to inspect the geometric model. At the same time, a human-computer interaction interface is set, and the current working parameters can be input by the user, including the lifting mode, the lifting load and the cycle number. The platform supports repeated input of working parameters, all input data will be managed by the background database as a load history record, which is convenient for proxy model calling and dynamic damage accumulation updating. The structure life is quickly predicted through the deep learning proxy model, the current remaining life is calculated through the Miner theory, and the real-time display and intelligent warning of the device state are realized in combination with the visualization module.

[0044] The beneficial effects in the present application are:

[0045] The application can intelligently deduce the evolution process of fatigue damage based on the service parameters and loading history of the structure, realize dynamic calculation of the remaining life and real-time warning of the service risk. Compared with traditional finite element simulation calculation, the neural network proxy model constructed in the application saves computing resources and improves the efficiency of fatigue life assessment, and has good real-time performance and practicality. The model is deployed in a standardized and universal format, which facilitates integration and cross-platform calling, and improves the flexibility and scalability of the system. The evaluation system effectively combines damage theory and failure mechanism by introducing finite element simulation calculation dataset and Miner criterion, providing theoretical support for the intelligent deduction process and ensuring the accuracy and reliability of the remaining life prediction results. Further combined with the visualization platform developed based on the Unity3D engine, the geometric structure and health status of the equipment can be intuitively displayed, the human-computer interaction interface is friendly, and the operability of the platform is enhanced. The method improves the problems of dependence on offline simulation, lagging response and discontinuous evaluation in traditional structure life management, effectively improves the intelligent operation and maintenance level of engineering machinery, and provides protection for safe and efficient operation. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 A schematic diagram of an engineering machinery load-bearing structure fatigue life intelligent monitoring method is provided.

[0047] Figure 2 A neural network architecture diagram of an engineering machinery load-bearing structure fatigue life intelligent monitoring method is provided.

[0048] Figure 3 A neural network training loss curve diagram of an engineering machinery load-bearing structure fatigue life intelligent monitoring method is provided.

[0049] Figure 4 A neural network prediction effect comparison diagram of an engineering machinery load-bearing structure fatigue life intelligent monitoring method is provided.

[0050] Figure 5 A hoist fatigue life prediction system client main page diagram of an engineering machinery load-bearing structure fatigue life intelligent monitoring method is provided. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all embodiments.

[0052] Reference Figures 1-5 An engineering machinery load-bearing structure fatigue life intelligent monitoring method, as shown in Figure 1 , includes the following steps: step one, establishing a high-fidelity geometric model.

[0053] In step one, the geometric model is constructed by three-dimensional modeling software, and then the geometric model is exported to the corresponding file format according to the requirements.

[0054] Step two, establish a finite element model.

[0055] The geometric model is imported into the finite element simulation software in step two according to the file type requirements, and the geometric model is simplified. After the geometric model is simplified, the material properties are defined, then the structure in the geometric model is meshed and the mesh independence verification is carried out, and finally the relevant boundary conditions are set. It is realized in the following way:

[0056] Based on the established geometric model, import the finite element simulation software to simplify the model, keep the key components that affect the mechanical properties of the lifting appliance, and reasonably simplify the secondary factors that have little effect. Common simplification measures include deleting or closing the structure details such as chamfer, bolt hole, nameplate groove, small fillet, etc. that do not participate in the main force; using equivalent bar element or simplified block to replace the pin shaft, pull rod and other components with regular shape and size much smaller than the overall structure; combining multiple welded components into a single entity to reduce contact surface processing; removing accessories such as wires and pipelines, etc. Then, define the material properties, mesh the structure and carry out mesh independence verification. Local encryption strategy can be used for key stress areas to improve calculation accuracy. Finally, set the relevant boundary conditions to prepare for subsequent simulation calculation.

[0057] In an embodiment, the STEP file exported in step one is imported into Ansys workbench for model simplification. The electrical equipment, hydraulic pipeline and its accessory structures arranged in the lifting appliance geometric model are removed; the smaller size welds and components in the main beam and telescopic beam are modeled as a whole; the larger size welds between the main beam webs and between the main beam and the plate are modeled as a separate component; the smaller size threaded holes and other structures are ignored; the small chamfers, fillets and other structures in the lifting appliance structure are modeled as right angles. The main load-bearing components in the lifting appliance structure are high-strength steel material, the buffer pad material between the main beam and the telescopic beam is nylon 66, and the rest of the components are ordinary carbon structural steel. The corresponding materials are given the material parameters of Poisson's ratio, elastic modulus, density, yield limit and tensile strength. Then, according to the actual situation of the lifting appliance, the contact conditions are generated, the fixed support constraints are set, and the mesh is divided. The mesh is mainly hexahedron, and the local complex structure is tetrahedral mesh. The mesh independence verification is carried out to confirm that the mesh division is reasonable.

[0058] Step three, build a finite element simulation dataset.

[0059] First, the integration step size is set according to the working state and working time, and an appropriate damping form is selected according to the structural characteristics and analysis requirements. Transient dynamic analysis is then performed on the entire cycle process of the geometric model to obtain the load spectrum.

[0060] Then, the load spectrum was analyzed using Ansys Workbench and nCode DesignLife in conjunction with fatigue analysis to obtain the corresponding loads. fatigue life .

[0061] The maximum load that the geometric model may encounter under actual working conditions. and minimum load Add a certain amount of redundancy The upper limit load of the geometric model is then obtained. and lower limit load Simulation dataset input The goal is to make the simulation data as close as possible to actual working conditions.

[0062] Finally, hyper-Latin cube sampling was used at the set upper limit load. and lower limit load Extract a sufficient number of data points and obtain the fatigue life for each data point through simulation analysis. These input and output data are recorded in tabular form to form a load-life mapping dataset for finite element simulation. ,in, .

[0063] In some real-world applications, lifting operations involve three different modes depending on the container size. For each mode, the maximum and minimum loads that may occur in the actual working conditions are identified, and a certain amount of redundancy is added as the upper and lower limits of the simulation dataset input. Points are selected within the input range using hyper-Latin cube sampling. The corresponding fatigue life is obtained through transient dynamics combined with Ansys Workbench and nCode DesignLife fatigue analysis, thus constructing three load-life mapping datasets for finite element simulations. .

[0064] Step 4: Construct a neural network proxy model; build the neural network architecture from Step 4 using Python, based on the load-lifetime mapping dataset constructed in Step 3. The model is normalized and divided into training and testing sets according to a certain ratio. It is then used for training, validation, and hyperparameter tuning. Specifically, during training, an appropriate optimizer and loss function are selected, and hyperparameters such as learning rate, epochs, and batch size are set. A load-lifetime surrogate model is then constructed.

[0065] During training, the loss function will calculate the predicted values ​​of the neural network model and the load-lifetime mapping dataset. The error between labels is backpropagated to update the parameters of the neural network model;

[0066] The neural network model is evaluated using mean absolute error (MAE), mean squared error (MSE), and coefficient of determination. As an evaluation metric, if the metric does not meet expectations, the hyperparameter size is adjusted. When the accuracy of the neural network model meets the requirements, the neural network model is saved as a target file format, and the normalization information during training is manually saved in the checkpoint of the target file to prevent prediction deviations in the inference stage caused by inconsistent normalization.

[0067] Export the model weights and structure from the target file into a general model deployment format, and record the normalization information in the checkpoint of the target file so that the input specifications during training can be correctly restored when using the model for inference.

[0068] In this embodiment, a four-layer ANN model is established, and the model architecture is as follows: Figure 2 As shown, the input layer is a one-dimensional load on a lifting device, the hidden layer has two layers, each with 64 neurons, and the output is the logarithm of the one-dimensional fatigue life. The activation function used is ReLU. Using the dataset constructed in step three, data preprocessing is performed on the finite element calculations. Logarithmic values ​​were taken and Max-Min normalization was performed. Load-life surrogate models were then trained under three different operating modes. The comparison between the training loss curve and prediction performance under the 20-foot container lifting operating mode is shown in the figure below. Figure 3 and Figure 4 As shown, the weights and structures of the three trained models are exported into a general model deployment format, and the normalization information is saved.

[0069] Step 5: Obtain working parameters and loading history; the working parameters in Step 5 include the load and cycle number of the geometric model;

[0070] When the geometric model is a lifting device, the working parameters refer to the lifting device load and cycle number, and the loading history refers to the entire process of load changes experienced by the lifting device structure over time during operation. This refers to the load on the geometric model for each working cycle. The cumulative total of the corresponding cycle periods, where, ;

[0071] Carrying out statics analysis based on the finite element model established in step three, calculating the dangerous points of the geometric model structure such as the spreader, establishing a statics inversion model of the neural network model, and inverting the load borne by the geometric model structure through the stress at the dangerous points of the geometric model structure Specifically, the maximum and minimum loads possibly occurring in actual working conditions are investigated, and certain redundancy is added as the upper and lower limits of the data set, so that the finite element simulation data cover the actual working conditions as much as possible.

[0072] Data points are extracted from the stress-strain mapping data set of the finite element simulation through statics simulation calculation. The calculated stress is taken as the input of the neural network data set, and the corresponding load is taken as the output. The data is recorded in table form to form the stress-load mapping data set of the finite element simulation The neural network model architecture is established again through the stress-load mapping data set of the finite element simulation The training of the statics inversion model is completed. Strain gauges or stress sensors are arranged at the dangerous points of the actual spreader to monitor the stress at the dangerous points online. Since the stress of the spreader changes obviously in stages within a single working cycle, the lifting and unloading stages are accompanied by the existence of dynamic load.

[0073] When the static finite element inversion model no longer matches the actual dynamic load environment, the working stage needs to be identified, and the sliding window is used to judge the stability of the monitored stress value: a fixed time step is taken as a window, the number of data points in each window is consistent, and the window is slid forward as the work progresses, and the stress value data points corresponding to the time step are updated constantly. The stress fluctuation variance of the stress value in each window is calculated When the fluctuation variance is lower than the preset threshold, it is determined that the geometric model structure is in a horizontal running stage, at this time, the stress mean value in the window is taken as the input, and the output is obtained through the statics inversion model as the load of the geometric model in this working cycle Under the action of the load of this cycle, the cycle number is obtained by using the rain flow counting method according to the stress change sequence recorded by the stress sensor, and the working parameters of a single working cycle are obtained.

[0074] In addition, considering that the maximum dynamic load of the geometric model working cycle is accompanied by a rapid rise in stress, and the minimum dynamic load , get the first derivative of stress, which is used to identify the working cycle boundary. At the same time, set the minimum time span to avoid identifying false touch, combined with the aforementioned sliding window steady-state detection to avoid false cycle identification. The loading history of each cycle is stored in the database for platform to call, which is used for the remaining life calculation of the sling.

[0075] Or step five can be input by the user through the man-machine interface of the platform.

[0076] Step six, integrate the visual operation and maintenance platform. The integrated visual operation and maintenance platform in step six calculates the remaining life and compares it with the life threshold to update the device state.

[0077] The geometric model in step one is converted to STEP file format using 3DMax, exported to FBX format and imported into the main scene of Unity3D for visual display;

[0078] At the same time, the stress sequence collected in step five can be dynamically visualized by Unity3D with the help of XChart chart, so as to realize the intuitive presentation of the stress change process, which is convenient for users to understand the overall stress state of the structure. In addition, combined with the load-life surrogate model constructed in step four and the working parameters and loading history obtained in step five, the Unity3D background uses the Miner linear cumulative damage theory to dynamically evaluate the remaining life of the structure, wherein the formula of the Miner linear cumulative damage theory is:

[0079] ;

[0080] In the formula, is the total fatigue damage, is the actual cycle number, is the allowable cycle number under the corresponding load;

[0081] Damage is used to calculate the remaining life, wherein the remaining life is 1, it is considered that the geometric model is intact; the remaining life is 0, that is, failure occurs;

[0082] According to the calculated remaining life, the device state of the geometric model is updated in real time, and the alarm threshold is set When the calculated remaining life does not drop to , the device state remains normal service, when it drops to , the fatigue failure risk warning is issued, and when the device is repaired or replaced, the remaining life is updated to 1, and the device state is updated to normal service.

[0083] In the embodiment, the system supports online and offline local running modes. In the online mode, the Flask framework is used as a backend service platform, and the front end is jointly built by the Unity engine and the Web front end based on vue.js to realize the multi-terminal collaborative human-computer interaction interface. The front end of the system is used to receive the device working parameters input by the user, and transmit the parameters to the backend server for processing. The backend executes prediction through the deployed neural network model and estimates the remaining life through the Miner criterion, and feeds back the calculation results to the front end for visual presentation. Any front end can independently complete the input of working parameters, and the backend uniformly processes and pushes the results after receiving the data. At the same time, the consistency of the front and back end data and the synchronization of the data between the front ends are realized through the MySQL relational database set in the system.

[0084] In addition, the Unity front end supports interactive viewing of three-dimensional models, and the vue.js front end presents the device life status through the web interface, directly displays the predicted remaining life in the form of a graphical blood bar, and enhances the flexibility and visualization performance of the system. The dual-front-end architecture improves the operability and deployment adaptability of the system, and is suitable for various industrial device intelligent operation and maintenance scenarios.

[0085] In offline mode, only Unity client local prediction is supported, and the client main interface is as shown in Figure 5 The geometric model is imported into the main scene of Unity3D, and C# scripts are written to realize the rotation and attitude reset of objects in the three-dimensional scene with the help of the Unity engine, so as to facilitate the user to view the geometric model. The client has a human-computer interaction interface, which can input the current working parameters by the user, including lifting mode, lifting load and cycle number. The platform supports repeated input of working parameters, and all input data will be managed by the background database as load history records, which facilitates the proxy model to call and dynamically update damage accumulation. The structure life is quickly predicted through the deep learning proxy model, and the current remaining life is calculated through the Miner theory, and the device state is realized in real time through the visualization module. The real-time display and intelligent early warning of the device state is realized.

[0086] The system as a whole forms a closed-loop process from data input, life prediction, state update to interactive feedback, effectively improving the health management efficiency and safety and reliability of the spreader structure.

[0087] The application can intelligently deduce the evolution process of fatigue damage based on the service parameters and loading history of the structure, realize dynamic calculation of the remaining life and real-time warning of the service risk. Compared with traditional finite element simulation calculation, the neural network proxy model constructed in the application saves the calculation resources and improves the efficiency of fatigue life evaluation, and has good real-time performance and practicality. The model is deployed in a standardized and universal format, which is convenient for integration and cross-platform calling, and improves the flexibility and scalability of the system. The evaluation system effectively combines damage theory and failure mechanism by introducing finite element simulation calculation data set and Miner criterion, provides theoretical support for the intelligent deduction process, and guarantees the accuracy and reliability of the remaining life prediction results. Further combined with the visualization platform developed based on the Unity3D engine, the geometry and health state of the equipment can be intuitively displayed, the human-computer interaction interface is friendly, and the operability of the platform is enhanced. The method improves the problems of dependence on offline simulation, lagging response and discontinuous evaluation in traditional structure life management, effectively improves the intelligent operation and maintenance level of the engineering machinery, and provides protection for the safe and efficient operation of the engineering machinery.

[0088] The above is only the preferred specific embodiment of the application, but the protection scope of the application is not limited thereto, any person skilled in the art can make equivalent replacement or change according to the technical scheme and the inventive concept of the application within the technical range disclosed by the application, which should be covered in the protection scope of the application.

Claims

1. An intelligent monitoring method for fatigue life of a construction machinery load-bearing structure, characterized by comprising the following steps: Step 1, establishing a high-fidelity geometric model; Step 2, establishing a finite element model; Step 3, constructing a finite element simulation dataset; Step 4, constructing a neural network proxy model; Step 5, obtaining working parameters and loading history; and Step 6, integrating a visual operation and maintenance platform. In Step 1, the geometric model is constructed by a three-dimensional modeling software, and then the geometric model is exported into a corresponding file format according to requirements. The geometric model is imported into a finite element simulation software in Step 2 according to file type requirements, and the geometric model is simplified. After the geometric model is simplified, material properties are defined, the structure in the geometric model is meshed, mesh independence is verified, and relevant boundary conditions are set. In Step 3, integral steps are set according to working states and working time, appropriate damping forms are selected according to structural characteristics and analysis requirements, transient dynamics analysis is performed on the full cycle process of the geometric model, and a load spectrum is obtained. The model weight and structure in the target file are exported into a general model deployment format, and the normalization information in the checkpoint of the target file is recorded. The working parameters in Step 5 include the load and cycle number of the geometric model.

2. The method of claim 1, wherein: Alternatively, Step 5 can input relevant information by a user through a man-machine interface of a platform.

3. The method of claim 2, wherein: In Step 6, the integrated visual operation and maintenance platform calculates the remaining life and compares it with a life threshold, and updates the device state.

4. The method of claim 3, wherein: The geometric model in Step 1 is converted into a target file format by 3DMax, and the target file format is imported into a main scene of Unity3D for visual display.

5. A method of intelligent monitoring of fatigue life of a construction machine load bearing structure according to claim 4, characterized in that: Meanwhile, XChart charts of Unity3D can dynamically display the stress sequence collected in Step 5, and the background of Unity3D dynamically evaluates the remaining life of the structure by using a Miner linear cumulative damage theory, wherein the formula of the Miner linear cumulative damage theory is:

6. A method of intelligent monitoring of fatigue life of a construction machine load bearing structure according to claim 5, characterized in that: The load spectrum is analyzed by Ansys workbench and nCode DesignLife, and the fatigue life under the corresponding load is obtained ;​ the maximum load and minimum load that can occur in the actual operating conditions of the geometry model with a certain redundancy are obtained and the simulation data set input ; Super Latin cube sampling was used at the set upper limit load. and lower limit load Extract a sufficient number of data points and obtain the fatigue life for each data point through simulation analysis. These input and output data are recorded in tabular form to form a load-life mapping dataset for finite element simulation. ,in, .

7. A method of intelligent monitoring of fatigue life of a construction machine load bearing structure according to claim 6, characterized in that: establishing the neural network architecture in the fourth step by Python, based on the load-life mapping dataset constructed in the third step Normalization processing is performed, and the training set and the test set are divided in a certain proportion. The model is trained, verified, and parameter-adjusted to construct a load-life proxy model.

8. A method of intelligent monitoring of fatigue life of a construction machine load bearing structure according to claim 7, characterized in that: During training, a loss function will calculate the error between the neural network model's predictions and the load-life mapping dataset error between the labels and backpropagates to update the parameters of the neural network model; The neural network model adopts mean absolute error (MAE), mean square error (MSE) and coefficient of determination (R2) during evaluation As an evaluation index, if the index does not reach the expectation, the size of the hyperparameter is adjusted, when the accuracy of the neural network model reaches the requirement, the neural network model is saved as a target file format and the normalization information during training is manually saved in the checkpoint of the target file. ​ 9. A method of intelligent monitoring of fatigue life of a construction machine load bearing structure according to claim 8, characterized in that: ​ The load history includes the full history of the load experienced by the geometry model structure over time during the course of the work, for each work cycle of the geometry model with the accumulation of the corresponding cycle number, wherein ; Based on the finite element model established in step three, static analysis is carried out to calculate the dangerous points of the geometric model structure, and a static inversion model of the neural network model is established, and the load borne by the geometric model structure is inverted through the stress at the dangerous point of the geometric model structure , forming a stress-load mapping data set of finite element simulation , and the neural network model architecture is re-established through the stress-load mapping data set of finite element simulation The training of the static inversion model is completed; When the static finite element inversion model and the actual dynamic load environment will no longer match, the sliding window is used to monitor the stress value to determine stability: the stress fluctuation variance is calculated for the stress value in each window When the fluctuation variance is lower than the preset threshold value, it is determined that the geometric model structure is in the horizontal operation stage, at which time the stress mean value in the window is taken as the input, the output is obtained by the statics inversion model, and is taken as the load of the geometric model of the operation cycle Under the load of the cycle, according to the stress change sequence recorded by the stress sensor, the cycle number is obtained by using the rain flow counting method, and the acquisition of the single operation cycle work parameter is completed; Considering the maximum dynamic load of the geometry model working cycle The minimum dynamic load at the end of the phase with rapid stress rise The phase is accompanied by rapid stress drop, so the transition point of the first derivative of stress from negative to positive value is the working cycle boundary. Real-time forward difference calculation is performed on the stress sequence: , get the first derivative of stress, used to identify the working cycle boundary; ​ 10. A method of intelligent monitoring of fatigue life of a construction machine load bearing structure according to claim 9, characterized in that: ​ ​ ​ ； wherein is the total fatigue damage, is the actual number of cycles, is the number of cycles allowed under the corresponding load; damage for estimating the remaining life, wherein the remaining life is 1, the geometric model is considered intact; the remaining life is 0, i.e. failure occurs; According to the calculated residual life, the device state of the geometric model is updated in real time, and an alarm threshold is set When the calculated residual life does not decrease to , the device state remains normal service, when it decreases to , a fatigue failure risk warning is issued, and after the device is repaired or replaced, the residual life is updated to 1 and the device state is updated to normal service.

Images