Digital-twin-based rapid reliability early warning method and system for unmanned mine car

Abstract

The application discloses a kind of based on digital twinning unmanned mine car reliability rapid early warning method, system, it is related to unmanned mine car technical field, including the following steps: the digital twin simulation model of unmanned mine car is constructed;Based on the digital twin simulation model, the digital twin reliability model that the reliability state change of each module of the unmanned mine car is characterized is constructed;Based on the measured data of the unmanned mine car, the consistency of the digital twin simulation model and the digital twin reliability model is judged, and after judging consistent, reliability state database is constructed;Based on the reliability state database, digital twin reliability rapid early warning model is obtained by training, and the early warning model is used to identify and output the current reliability state of the unmanned mine car according to real-time acquisition vehicle-mounted sensor observation data, to realize early warning.The reliability state of each subsystem and module of unmanned mine car is rapidly and accurately warned.

Description

Technical Field

[0001] This invention relates to the field of unmanned mining truck technology, and in particular to a method and system for rapid early warning of the reliability of unmanned mining trucks based on digital twins. Background Technology

[0002] As a crucial component of smart mines, the reliability and operational status of unmanned mining trucks directly impact the safety, efficiency, and economy of mine production. Currently, reliability management of unmanned mining trucks primarily relies on two types of technologies: first, planned maintenance based on fixed cycles, involving regular shutdowns for inspection and upkeep of the vehicle's structure and key components; and second, real-time performance monitoring based on onboard sensors, using sensing devices such as LiDAR, cameras, and millimeter-wave radar to collect vehicle operating data and conduct preliminary analysis and alerts regarding the vehicle's status.

[0003] However, existing technologies still have the following obvious shortcomings: First, traditional periodic maintenance requires mining trucks to be shut down periodically, which is a complex and time-consuming process that relies on a large number of experienced professional maintenance personnel. This not only affects the efficiency of continuous mining operations but also brings high labor costs.

[0004] Secondly, existing real-time monitoring methods are mostly limited to single sensor data or simple threshold alarms, making it difficult to comprehensively and accurately assess and warn of the reliability status of the mining truck as a whole and its subsystems. Especially when the mining truck operates in harsh and dynamic mining environments, sensor data is easily interfered with, and the analysis results from a single data source are often unreliable and inaccurate, making it difficult to achieve early detection and precise location of faults.

[0005] In addition, existing methods generally lack the ability to efficiently integrate and simulate multi-source data and system behavior, and cannot build a reliability model in the digital space that is synchronized with the physical mining truck in real time and mapped to the whole state, resulting in delayed early warning and lack of foresight in maintenance decisions.

[0006] Therefore, how to achieve real-time, accurate, and rapid early warning of the reliability status of unmanned mining trucks without stopping the machine or requiring much intervention has become an urgent technical problem to be solved in the intelligent operation and maintenance of mines.

[0007] In recent years, digital twin technology has provided new ideas for the full life cycle management of equipment by constructing virtual mappings of physical entities. However, in the field of reliability early warning for unmanned mining trucks, there is still a lack of a complete methodology that can effectively integrate multi-source sensor data, realize dynamic model updates, and support rapid status identification and early warning. Summary of the Invention

[0008] In a first aspect, to address the aforementioned technical problems, this invention provides a rapid early warning method for the reliability of unmanned mining trucks based on digital twins, comprising the following steps: Constructing a digital twin simulation model of an unmanned mining truck; Based on the digital twin simulation model, a digital twin reliability model is constructed to characterize the reliability status changes of each module of the unmanned mining truck. Based on the measured data of the unmanned mining truck, the consistency of the digital twin simulation model and the digital twin reliability model is judged, and after the judgment is consistent, a reliability status database is constructed. Based on the aforementioned reliability status database, a digital twin reliability rapid early warning model is trained. The early warning model is used to identify and output the current reliability status of the unmanned mining truck based on real-time acquired vehicle sensor observation data, so as to achieve early warning.

[0009] Furthermore, the construction of the digital twin simulation model of the driverless mining truck specifically includes: Multiple sensors are deployed at key locations of the driverless mining truck to collect operational data; Based on the collected data, a three-dimensional environmental model of the mining scene and a three-dimensional geometric model of the unmanned mining vehicle are constructed. Build a virtual sensor model that matches the parameters of the real sensor. Establish a vehicle dynamics model for the unmanned mining truck and integrate an autonomous driving control algorithm; By integrating multi-source data through a data bus, a real-time synchronization mechanism is established between the digital twin simulation model and the physical mining truck.

[0010] Furthermore, a digital twin reliability model is constructed, specifically including: The unmanned mining truck is modeled in blocks according to the hierarchical structure of overall system-subsystem-module; Define a reliability level for each module and calculate a reliability degradation coefficient based on the module's initial reliability and current reliability to form a reliability degradation vector; The reliability degradation vector is mapped to an unreliability level vector to characterize the unreliability state of each module using categorical data.

[0011] Furthermore, the construction of the reliability status database specifically includes: Simulations were performed on the unmanned mining vehicle under various stochastic reliability conditions to obtain simulation data. The simulation data is augmented using a generative discriminative model to generate augmented training data. Based on the simulation data and the augmented training data, a reliability status database for training the early warning model is constructed.

[0012] Furthermore, the digital twin reliability rapid early warning model is a neural network model, which is trained through the reliability state database and is used to realize the mapping from the sensor observation data space to the reliability state space.

[0013] A second aspect of the present invention provides a rapid early warning system for the reliability of unmanned mining trucks based on digital twins, using the method described above, comprising: The simulation modeling module is configured to build a digital twin simulation model of the driverless mining truck; The reliability modeling module is configured to construct a digital twin reliability model that characterizes the reliability state changes of each module of the unmanned mining truck based on the digital twin simulation model. The consistency judgment and database construction module is configured to perform consistency judgment on the digital twin simulation model and the digital twin reliability model based on the measured data of the unmanned mining truck, and construct a reliability status database after the judgment is consistent. The early warning model construction and deployment module is configured to train a digital twin reliability rapid early warning model based on the reliability status database. The early warning model is used to identify and output the current reliability status of the unmanned mining truck based on the real-time acquired vehicle sensor observation data to achieve early warning.

[0014] Furthermore, the simulation modeling module is specifically used for: Acquire operational data from various sensors deployed in key parts of the mining truck; Based on the operational data, a three-dimensional environmental model of the mining scene and a three-dimensional geometric model of the unmanned mining truck are constructed. Build a virtual sensor model that matches the parameters of the real sensor. Establish a vehicle dynamics model for the unmanned mining truck and integrate an autonomous driving control algorithm; By integrating multi-source data through a data bus, a real-time synchronization mechanism is established between the digital twin simulation model and the physical mining truck.

[0015] Furthermore, the reliability modeling module is specifically used for: The unmanned mining truck is modeled in blocks according to the hierarchical structure of overall system-subsystem-module; Define a reliability level for each module and calculate a reliability degradation coefficient based on the module's initial reliability and current reliability to form a reliability degradation vector; The reliability degradation vector is mapped to an unreliability level vector to characterize the unreliability state of each module using categorical data.

[0016] Furthermore, the consistency judgment and database construction module is specifically used for: Simulations were performed on the unmanned mining vehicle under various stochastic reliability conditions to obtain simulation data. The simulation data is augmented using a generative discriminative model to generate augmented training data. Based on the simulation data and the augmented training data, a reliability status database for training the early warning model is constructed.

[0017] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: This invention constructs and integrates a digital twin simulation model and a reliability model, verifies consistency using measured data, and expands the training samples using generative discriminant techniques, ultimately training a real-time early warning model based on a neural network. This effectively solves the problems of complex processes, reliance on manual labor, poor real-time performance, and insufficient recognition accuracy under small sample conditions in traditional mine truck reliability monitoring. It achieves rapid and accurate early warning of the reliability status of various subsystems and modules of unmanned mine trucks. In practical mine applications, it can shorten fault diagnosis time by more than 40%, improving the operational safety and maintenance efficiency of mine trucks. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the construction process of the rapid early warning method for the reliability of unmanned mining trucks based on digital twins disclosed in this invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] This invention aims to provide a rapid reliability early warning method for unmanned mining trucks based on digital twins. Its core lies in constructing a digital twin that interacts in real time with the physical mining truck and evolves dynamically, achieving rapid identification and early warning of reliability status through a data-driven approach. The following describes... Figure 1 The construction process shown below elaborates on each step of the method of this invention: S1. Construct a digital twin simulation model of the unmanned mining truck.

[0022] Those skilled in the art will understand that this step aims to construct a high-fidelity virtual digital model (i.e., a digital twin simulation model) outside of the physical mining truck, a model capable of accurately simulating the structure, behavior, and response of the mining truck in a real mining environment. Implementation specifically includes the following details: S11. Define the application scenarios and performance indicators of the simulation model. For typical tasks of unmanned mining trucks in mine transportation, such as heavy-load uphill driving, curve driving, and operation in adverse weather conditions, define the key performance indicators that need to be simulated, including but not limited to vehicle vibration spectrum, stress and strain of key components, positioning accuracy, and control response delay.

[0023] S12. Data Acquisition and Processing. Multiple sensors are deployed in key components of the mining truck (such as the frame, suspension, powertrain, and steering mechanism). These include a triaxial accelerometer for vibration monitoring, a GNSS positioning module for high-precision trajectory tracking, a load sensor for ore weight detection, and LiDAR, cameras, and millimeter-wave radar for environmental perception. Sensor data is collected in real-time via an onboard Internet of Things (IoT) gateway and transmitted wirelessly to the cloud or edge server. Raw data is stored in a time-series database, and linear interpolation or spline interpolation is used to address missing data. Wavelet transform and other algorithms are employed for signal noise reduction to ensure data quality.

[0024] S13. Perform 3D modeling of the environment and vehicles. Use a drone equipped with an oblique photography camera to scan the mining area, generating high-precision 3D terrain point cloud data. Import this data into terrain tools such as Unity World Builder or Unreal Engine to construct a virtual mine scene including elements such as roads, ramps, and material yards, and simulate dynamic weather (such as sandstorms, rain, and snow). Simultaneously, use computer-aided design software to create a detailed 3D geometric model of the mining truck, and integrate finite element analysis methods to define the material properties (such as elastic modulus and density) and static stress distribution of each part of the vehicle body, providing a foundation for subsequent physical simulations.

[0025] S14. Build the sensor model. In a simulation platform (such as CarSim, Prescan, or a self-developed platform based on a game engine), configure virtual sensors for the virtual mining truck with parameters consistent with the actual model. For example, configure the LiDAR point cloud generation algorithm to simulate its scanning beam, angular resolution, and ranging error; build lens distortion and noise models for the camera; simulate the multipath effect and clutter environment of millimeter-wave radar. Ensure that the data format output by the virtual sensors is consistent with that of the physical sensors to achieve data-level benchmarking.

[0026] S15. Establish a dynamics and control system model. Based on multibody dynamics theory, use multibody dynamics simulation software such as ADAMS / Car or Simulink / Simscape to establish a dynamic model of the mining truck, including subsystems such as suspension, tires, steering, and transmission. This model can simulate complex mechanical behaviors such as the vibration response of the mining truck on rough roads and the slip ratio between the tires and the ground. Simultaneously, the actual autonomous driving algorithm module used by the mining truck (such as A / D-based systems) will be implemented. The algorithm's global path planning and reinforcement learning-based local obstacle avoidance module are integrated into the simulation environment to form a closed-loop test system of "environment-vehicle-controller".

[0027] S16. Achieve multi-source data integration and real-time synchronization. Design and deploy a unified data bus, such as using a Robot Operating System (ROS) or Apache Kafka message queue, as the data hub of the digital twin. This data bus is responsible for receiving real-time sensor data streams, control command status, and virtual environment parameters from the physical mining truck, and driving the digital twin simulation model to perform synchronous calculations and rendering. Establish a millisecond-level synchronization mechanism between the digital twin and the physical mining truck through a high-precision time synchronization protocol (such as PTP) to ensure that the two remain consistent in the time dimension.

[0028] S17. Configure and test simulation scenarios. Construct a library of various typical and extreme test scenarios, such as "starting with full load on a ramp" and "driving in a sandstorm with visibility less than 5 meters". Use test automation tools (such as TestBuilder scripts) to generalize parameters such as vehicle speed, load, road friction coefficient, and sensor failure modes, and execute Monte Carlo simulations. Verify the robustness and reliability of the entire digital twin simulation model under different working conditions through a large number of random experiments.

[0029] S18. Model Validation and Iterative Optimization. In the offline phase, the running trajectory and vibration data of the digital twin simulation are compared with the actual running data of the physical mining trucks during the same period. Indicators such as the root mean square error (RMSE) are calculated, and the dynamic model parameters (such as damping coefficient and tire stiffness) are repeatedly calibrated until the error is below a preset threshold. In the online application phase, based on the predictive capabilities of the digital twin, the scheduling strategy and energy consumption model of the mining trucks are dynamically optimized to achieve continuous self-evolution of the model.

[0030] S19. Deploy a visualization platform and support intelligent decision-making. Integrate the validated high-fidelity digital twin simulation model into a 3D visualization control platform driven by Unity or Unreal Engine. This platform can display the status, trajectory, key parameters, and early warning information of the digital twin of each vehicle in the mining truck cluster in real time, supporting managers to conduct remote status monitoring and emergency intervention.

[0031] S2. Based on the digital twin simulation model, construct a digital twin reliability model to characterize the reliability status changes of each module of the unmanned mining truck.

[0032] Those skilled in the art will understand that this step, based on the simulation model, constructs a model focused on characterizing and quantifying the evolution of the reliability state of mining trucks.

[0033] First, the unmanned mining truck is modeled using a block-based approach. A hierarchical decomposition method of "overall system - subsystem - module" is employed to model the mining truck. Let the overall system of the unmanned mining truck be A, which consists of NA subsystems, represented as... Each subsystem ( (Subsystem number) can be further decomposed into The basic functional modules are represented as follows: ,in, Representing the The first subsystem Each module. This structure clearly defines the granularity of reliability analysis.

[0034] Secondly, establish a reliability space for the digital twin. Define a reliability level for each module. This is used to quantify the probability of its normal operation or performance retention level. Factors causing reliability degradation include hardware wear and tear, software anomalies, communication delays, environmental interference, and human error. These factors ultimately manifest as a decrease in module reliability. A reliability degradation vector is defined. To comprehensively characterize the reliability status of the entire system: ,in, Indicates the first The reliability degradation factor of each module is calculated using the following formula: .here, This is the initial (health) reliability of the module. It is the first The current reliability of each module, The larger the value, the less reliable the module is.

[0035] To facilitate rapid classification and decision-making, the continuous reliability degradation coefficient is discretized into a finite number of unreliability levels. An unreliability level vector is defined. ,in Indicates the first The unreliability level of each module, , This represents the most severe level of unreliability. This can be determined by analyzing historical maintenance records. and The mapping relationship between them (e.g., setting threshold intervals) is used to transform continuous reliability assessments into explicit category labels.

[0036] S3. Based on the measured data of the unmanned mining truck, the consistency judgment of the digital twin simulation model and the digital twin reliability model is carried out, and after the judgment is consistent, a reliability status database is constructed.

[0037] After the initial construction of the digital twin simulation model and reliability model, this solution must compare and verify them with the physical entity. Measured data collected by the physical sensors deployed in S1 during the actual operation of the mining truck (such as vibration spectrum under specific working conditions, GPS trajectory, and control response time) are input into the digital twin model, and the same simulation scenario is run. Key indicators of the simulation output and measured data are compared, and consistency metrics (such as correlation coefficient and energy error in a specific frequency band) are calculated. If the consistency is lower than a preset threshold (e.g., trajectory RMSE greater than 0.5 meters), the model is considered to have low consistency, and the process must return to S1 to correct and iterate the parameters of the simulation model (such as dynamic parameters and sensor noise model). If the consistency meets the requirements, the process proceeds to the next step.

[0038] S4. Based on the reliability status database, a digital twin reliability rapid early warning model is trained. The early warning model is used to identify and output the current reliability status of the unmanned mining truck based on the real-time acquired vehicle sensor observation data to achieve early warning.

[0039] In this scheme, a large sample database M covering various possible reliability states is needed to train the subsequent rapid early warning model. Traditional methods involve analyzing the reliability degradation vectors of mining trucks under a large number of randomly generated conditions. Simulations are performed on the corresponding states, but this is extremely computationally expensive when the model is complex.

[0040] This invention proposes a database augmentation method based on a generative discriminative model. First, the reliability state space (i.e., different...) is... A relatively small number of random samples are taken from the vector. For each sampled reliability state, the corresponding fault or performance degradation mode (e.g., simulating a decrease in sensor accuracy or a reduction in suspension stiffness) is injected into the digital twin simulation model, and the simulation is run to obtain the corresponding virtual sensor observation data sequence under that state. This yields an initial, small-scale paired dataset of "simulation state-observation data".

[0041] Subsequently, generative discriminative models such as Generative Adversarial Networks (GANs) or Variational Autoencoders (VAEs) are introduced. This generative model is trained on the initial dataset, enabling it to learn how to handle a given reliability state label. This generates corresponding, realistic virtual sensor observation data. The distribution pattern is observed. After training, this generative model can efficiently synthesize a large number of diverse observation data samples for any specified or randomly generated reliability state. By evaluating the credibility of the generated data (e.g., using a discriminator network for filtering), the high-quality synthetic data is merged with the initial simulation data, thereby constructing a large-scale, high-quality augmented reliability state database M. This effectively solves the problem of insufficient model training data under small sample conditions.

[0042] S5, Digital Twin Reliability Rapid Early Warning Model Construction and Deployment The goal of this step is to build an early warning model that can work in real time. An early warning model is essentially a mapping function. .in, It is real-time observation data from vehicle-mounted sensors. The value space of , i.e. M represents the state space of the reliability state database constructed by S4. This model is based on the current time... The actual sensor observation vector obtained Quickly identify the most likely reliability state of the mining truck (i.e., the corresponding...) (Vector or specific fault module identifier).

[0043] In a preferred embodiment, the early warning model T is implemented using a deep neural network, such as a combination of a convolutional neural network (CNN) and a long short-term memory network (LSTM). The CNN is used to extract spatial features (such as images and point clouds) from multi-sensor data, while the LSTM is used to capture the sequence dependencies of sensor data over time. The neural network is trained under supervision using a large-scale augmented reliability state database M constructed using S4 to observe the data. As input, the corresponding reliability status label As a training objective, the network learns to identify implicit reliability degradation patterns from complex, multidimensional observation data.

[0044] After training, the model is deployed to the onboard computing unit or edge server of the mining truck and integrated with the digital twin. In actual operation, the early warning model receives sensor data streams from the physical mining truck in real time and instantly outputs the identification result of the current reliability status and the early warning level. This result is used to update the current state of the digital twin reliability model, ensuring that the digital twin always maintains a healthy state consistent with the physical entity. On the other hand, early warning information (such as "Right front suspension module reliability level 3, inspection recommended") is pushed to the central control visualization platform in real time to guide maintenance personnel to implement predictive maintenance, thereby intervening before or when a fault occurs and avoiding unplanned downtime.

[0045] Through the above five core steps, this invention constructs a complete technical closed loop from physical perception, virtual modeling, data augmentation to intelligent early warning, realizing rapid and accurate early warning of the reliability status of unmanned mining trucks, and significantly improving the intelligent level of operation and maintenance and the safety and economy of mining equipment.

[0046] Furthermore, based on the above method, this invention also provides a rapid reliability early warning system for unmanned mining trucks based on digital twins, including a simulation modeling module, a reliability modeling module, a consistency judgment and database construction module, and an early warning model construction and deployment module. The simulation modeling module is used to construct a digital twin simulation model of the unmanned mining truck; the reliability modeling module is used to construct a digital twin reliability model representing the reliability state changes of each module of the unmanned mining truck based on the digital twin simulation model; the consistency judgment and database construction module is used to perform consistency judgment on the digital twin simulation model and the digital twin reliability model based on the measured data of the unmanned mining truck, and construct a reliability state database after the judgment is consistent; the early warning model construction and deployment module is used to train a rapid reliability early warning model based on the reliability state database. The early warning model is used to identify and output the current reliability state of the unmanned mining truck based on real-time acquired onboard sensor observation data to achieve early warning.

[0047] In a further proposed solution, the simulation modeling module is specifically used to: acquire operational data from various sensors deployed at key locations on the mining truck; construct a 3D environmental model of the mining scene and a 3D geometric model of the unmanned mining truck based on the operational data; build a virtual sensor model with parameters consistent with the real sensors; establish a vehicle dynamics model of the unmanned mining truck and integrate autonomous driving control algorithms; and integrate multi-source data through a data bus to establish a real-time synchronization mechanism between the digital twin simulation model and the physical mining truck.

[0048] The reliability modeling module is specifically used to: block-based model the unmanned mining truck according to the hierarchical structure of the overall system-subsystem-module; define the reliability of each module, and calculate the reliability degradation coefficient based on the initial reliability and current reliability of the module to form a reliability degradation vector; map the reliability degradation vector to an unreliability level vector to use categorical data to characterize the unreliability state of each module.

[0049] The consistency judgment and database construction module is specifically used for: simulating unmanned mining trucks under various random reliability states to obtain simulation data; applying a generative discriminant model to augment the simulation data to generate augmented training data; and constructing a reliability state database for training the early warning model based on the simulation data and the augmented training data.

[0050] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A rapid early warning method for the reliability of unmanned mining trucks based on digital twins, characterized in that, Includes the following steps: Constructing a digital twin simulation model of an unmanned mining truck; Based on the digital twin simulation model, a digital twin reliability model is constructed to characterize the reliability status changes of each module of the unmanned mining truck. Based on the measured data of the unmanned mining truck, the consistency of the digital twin simulation model and the digital twin reliability model is judged, and after the judgment is consistent, a reliability status database is constructed. Based on the aforementioned reliability status database, a digital twin reliability rapid early warning model is trained. The early warning model is used to identify and output the current reliability status of the unmanned mining truck based on real-time acquired vehicle sensor observation data, so as to achieve early warning.

2. The rapid early warning method for the reliability of unmanned mining trucks based on digital twins according to claim 1, characterized in that, The construction of the digital twin simulation model for the driverless mining truck specifically includes: Multiple sensors are deployed at key locations of the driverless mining truck to collect operational data; Based on the collected data, a three-dimensional environmental model of the mining scene and a three-dimensional geometric model of the unmanned mining vehicle are constructed. Build a virtual sensor model that matches the parameters of the real sensor. Establish a vehicle dynamics model for the unmanned mining truck and integrate an autonomous driving control algorithm; By integrating multi-source data through a data bus, a real-time synchronization mechanism is established between the digital twin simulation model and the physical mining truck.

3. The rapid early warning method for the reliability of unmanned mining trucks based on digital twins according to claim 1, characterized in that, Constructing a digital twin reliability model specifically includes: The unmanned mining truck is modeled in blocks according to the hierarchical structure of overall system-subsystem-module; Define a reliability level for each module and calculate a reliability degradation coefficient based on the module's initial reliability and current reliability to form a reliability degradation vector; The reliability degradation vector is mapped to an unreliability level vector to characterize the unreliability state of each module using categorical data.

4. The rapid early warning method for the reliability of unmanned mining trucks based on digital twins according to claim 1, characterized in that, The construction of the reliability status database specifically includes: Simulations were performed on the unmanned mining vehicle under various stochastic reliability conditions to obtain simulation data. The simulation data is augmented using a generative discriminative model to generate augmented training data. Based on the simulation data and the augmented training data, a reliability status database for training the early warning model is constructed.

5. The rapid early warning method for the reliability of unmanned mining trucks based on digital twins according to claim 1, characterized in that, The digital twin reliability rapid early warning model is a neural network model, which is trained through the reliability state database and is used to realize the mapping from the sensor observation data space to the reliability state space.

6. A rapid reliability early warning system for unmanned mining trucks based on digital twins, applying the method described in any one of claims 1-5, characterized in that, include: The simulation modeling module is configured to build a digital twin simulation model of the driverless mining truck; The reliability modeling module is configured to construct a digital twin reliability model that characterizes the reliability state changes of each module of the unmanned mining truck based on the digital twin simulation model. The consistency judgment and database construction module is configured to perform consistency judgment on the digital twin simulation model and the digital twin reliability model based on the measured data of the unmanned mining truck, and construct a reliability status database after the judgment is consistent. The early warning model construction and deployment module is configured to train a digital twin reliability rapid early warning model based on the reliability status database. The early warning model is used to identify and output the current reliability status of the unmanned mining truck based on the real-time acquired vehicle sensor observation data to achieve early warning.

7. The rapid early warning system for the reliability of unmanned mining trucks based on digital twins according to claim 6, characterized in that, The simulation modeling module is specifically used for: Acquire operational data from various sensors deployed in key parts of the mining truck; Based on the operational data, a three-dimensional environmental model of the mining scene and a three-dimensional geometric model of the unmanned mining truck are constructed. Build a virtual sensor model that matches the parameters of the real sensor. Establish a vehicle dynamics model for the unmanned mining truck and integrate an autonomous driving control algorithm; By integrating multi-source data through a data bus, a real-time synchronization mechanism is established between the digital twin simulation model and the physical mining truck.

8. The rapid early warning system for the reliability of unmanned mining trucks based on digital twins according to claim 6, characterized in that, The reliability modeling module is specifically used for: The unmanned mining truck is modeled in blocks according to the hierarchical structure of overall system-subsystem-module; Define a reliability level for each module and calculate a reliability degradation coefficient based on the module's initial reliability and current reliability to form a reliability degradation vector; The reliability degradation vector is mapped to an unreliability level vector to characterize the unreliability state of each module using categorical data.

9. The rapid early warning system for the reliability of unmanned mining trucks based on digital twins according to claim 6, characterized in that, The consistency judgment and database construction module is specifically used for: Simulations were performed on the unmanned mining vehicle under various stochastic reliability conditions to obtain simulation data. The simulation data is augmented using a generative discriminative model to generate augmented training data. Based on the simulation data and the augmented training data, a reliability status database for training the early warning model is constructed.

Images