A global load spectrum real-time identification method for a mining excavator working device

Abstract

This invention belongs to the field of load testing technology for mining engineering equipment, specifically relating to a real-time identification method for the full-domain load spectrum of a mining excavator's working device. The method includes the following steps: constructing a rigid-flexible coupled dynamic model of the mining excavator; establishing flexible finite element models for the boom and stick; solving the rigid-flexible coupled nonlinear dynamic coupling control equations; real-time acquisition of multi-source measured data; inputting the multi-source measured data into the model for real-time simulation to obtain full-domain simulated stress data; deploying strain sensors at key high-stress areas of the excavator to collect measured strain data; comparing the simulated and measured strain characteristics, dynamically optimizing the model until the error meets accuracy requirements; and generating a real-time full-domain load spectrum based on the optimized and calibrated model. Through rigid-flexible coupled modeling, real-time multi-source data driving, and dynamic calibration, the method achieves accurate capture of the full-domain load of the working device, overcoming the limitations of existing methods and significantly improving the real-time performance and accuracy of the load spectrum.

Description

Technical Field

[0001] This invention belongs to the field of load testing technology for mining engineering equipment, and specifically relates to a method for real-time identification of the full-domain load spectrum of a mining excavator working device. Background Technology

[0002] Large mining excavators are key equipment in mining operations. Their working devices, such as the boom and stick, which directly participate in excavation operations, are subjected to alternating loads, impact loads, and complex and variable environmental loads over long periods, such as changes in material properties, slope undulations of the working face, and wind interference. The accuracy of the load spectrum directly affects structural fatigue life assessment, safety condition monitoring, and maintenance strategy formulation. Therefore, how to efficiently and accurately identify the full-range load spectrum of the working device in real time has always been a core technical challenge that urgently needs to be overcome in the field of mining engineering equipment testing.

[0003] Currently, there are three methods for identifying the load spectrum of mining excavators: simulation method, actual measurement method, and combined simulation and actual measurement method. However, each of these methods has some problems in actual engineering applications, which restricts the accuracy of the load.

[0004] Simulation methods employ rigid body modeling, neglecting the influence of structural strain and deformation on load transfer. Furthermore, the pre-defined material properties often deviate from reality, leading to errors in load calculations. Patent CN113865899A proposes a load spectrum monitoring method based on a model observer. This method uses rigid body dynamics modeling as its core framework, monitoring information such as hydraulic cylinder piston displacement, pressure, and connecting rod angular displacement. It then constructs an observer model by combining the kinematic and dynamic models of the hydraulic system and working device. However, this method fails to address the load transfer mismatch between rigid body modeling and actual structural dynamic deformation, and it also suffers from the problem of pre-defined materials not matching the dynamic characteristics of actual mine materials, such as hardness and friction coefficient.

[0005] The experimental method focuses on strain testing of the working device during the excavation stroke, but the measurement points are limited and the coverage is incomplete, making it impossible to obtain full-range load data for the working device. Patent CN119738145A proposes a simplified test method for equivalent load of the excavator boom stick, which calculates the load through strain gauge measurement data. However, the sensors are only placed in some high-stress areas, resulting in incomplete load data coverage and defects such as local clarity but overall fracture.

[0006] The combined simulation and measurement method takes the integration of simulation and measurement as its core idea, attempting to balance the full coverage of simulation with the realism of measurement. Patent CN114330058A proposes a combined method of dynamic model + discrete element model + strain measurement. By using the dynamic model and discrete element model, the load spectrum is obtained through joint simulation and strain analysis is performed. Then, actual strain is obtained by attaching strain gauges to the physical model, and the model parameters are adjusted accordingly to optimize the load spectrum. However, the patent does not establish a real-time data interaction and dynamic fusion mechanism between simulation and measurement, and only performs data comparison and model correction through offline static methods.

[0007] In summary, the bottleneck of existing technologies lies in the inability to simultaneously ensure the comprehensiveness and real-time performance of the load spectrum while maintaining recognition accuracy. In other words, there is a lack of a technical approach for accurately and in real-time acquiring the comprehensive load spectrum. The single simulation method, lacking real-time interaction with on-site measured data, is essentially an offline calculation, unable to achieve real-time load recognition and its accuracy is also affected by the model. While the single measurement method can directly acquire data from on-site sensors to ensure real-time performance, it is limited by the complex physical installation space of the excavator and the range of measuring points, making it impossible to acquire the comprehensive load spectrum. The current combined simulation and measurement method uses offline correction of measured data, which improves comprehensive coverage and accuracy to some extent, but the correction process is offline, losing the original real-time recognition advantage of the measurement method. Therefore, there is a clear lack of a technical solution that treats simulation and measurement as equal dual core data sources and can adaptively adjust the fusion strategy according to the working conditions. This gap in the concept of comprehensive coverage, real-time response, and high-precision output has become a core obstacle restricting the intelligent and precise development of load spectrum recognition technology for mining excavators. Summary of the Invention

[0008] In order to overcome the shortcomings of the prior art, this invention proposes a real-time identification method for the full-domain load spectrum of a mining excavator's working device.

[0009] This invention is achieved using the following technical solution:

[0010] A method for real-time identification of the full-domain load spectrum for the working device of a mining excavator includes the following steps:

[0011] S1: Construct a rigid-flexible coupled dynamic model of a mining excavator using simulation modeling software; for at least two structural components of the excavator's working device, namely the boom and stick, establish a flexible finite element model and solve it using the rigid-flexible coupled nonlinear dynamic coupled control equation; simplify the modeling of the non-working device that provides support and movement functions for the working device.

[0012] S2: Collect multi-source measured data during the excavator's operation in real time and perform preprocessing;

[0013] S3: The preprocessed multi-source measured data is used as a driving signal and input to the rigid-flexible coupling dynamic model for real-time simulation calculation to obtain the simulation stress data of the entire working device.

[0014] S4: By deploying strain sensors at key high-stress areas on the excavator body, the measured strain data at the corresponding locations are collected in real time.

[0015] S5: Compare the strain information in the simulated stress data with the measured strain data in all dimensions, and dynamically optimize and calibrate the rigid-flexible coupling dynamic model based on the comparison results until the error between the simulated stress data and the measured strain data meets the preset accuracy requirements.

[0016] S6: Based on the optimized and calibrated rigid-flexible coupling dynamic model, a full-domain load fusion model is constructed. The model is adaptively determined according to the type of measurement point and the working condition. It integrates two types of data sources: with sensors, the actual measurement is the core, and without sensors or in case of faults, the simulation fully covers the data. The model is then accurately fused by combining composite error correction terms.

[0017] S7: Based on the global load fusion model, the global load spectrum of the excavator working device is identified and output in real time.

[0018] Furthermore, the excavator is a rack and pinion push-press excavator; the working device includes at least a boom, stick, bucket, push-press mechanism and lifting mechanism; the non-working device includes at least a tracked chassis, slewing platform, lifting support and counterweight.

[0019] Furthermore, the rigid-flexible coupled nonlinear dynamic coupling control equation is as follows:

[0020] ;

[0021] The rigid-flexible coupled nonlinear dynamic coupled control equations are applied to the two structural components, the boom and the stick; for each of the boom and the stick, the equations in... , , A, E, I, and F represent the axial dynamic displacement, lateral dynamic displacement, material density, cross-sectional area, elastic modulus, moment of inertia, and dynamic distributed load of the structural component, respectively.

[0022] Further, in step S1, the boom and stick are divided into tetrahedral solid unit meshes with a basic mesh size of 50mm; and the mesh is densified at the connection between the stick and the bucket, around the boom hinge hole, and in the weld area, with the mesh size reduced to 20mm.

[0023] Furthermore, the multi-source measured data includes motion parameter data, environmental effect data, and material characteristic parameters matched based on a dynamic database of mine material characteristics;

[0024] The motion parameter data is collected in real time through automated software and the excavator's electronic control system. The collected motion parameters include at least the pushing speed, lifting speed, and slewing angular velocity. Based on the collected motion parameters, the pushing displacement, lifting displacement, and slewing angular displacement are obtained through integral calculation, and the displacements are mapped to the boundary conditions of the rigid-flexible coupled dynamic model through coordinate system transformation.

[0025] Furthermore, the dynamic database of mine material characteristics stores physical parameters of various mine materials and their corresponding visual appearance features, and can match and call the corresponding physical parameters based on the material appearance features collected in real time, and can iteratively update the stored parameters according to the operation feedback data.

[0026] Furthermore, the environmental impact data includes excavation resistance data obtained through multi-sensor fusion and material gravity data obtained through a pressure sensor array.

[0027] Furthermore, the acquisition of the excavation resistance data specifically includes:

[0028] The tension of the wire rope in the hoisting mechanism is collected by a wire rope tension sensor. and its angle with the horizontal direction ;

[0029] The contact pressure between the bucket teeth and the material is collected by strain gauge pressure sensors mounted on the bucket teeth. , i=1,2,3,4,5;

[0030] The pin torque T is collected by a pin sensor installed at the boom-bucket hinge pin, and then calculated using the formula... Converted to force at the hinge r is the radius of the pin shaft;

[0031] Confidence weighting based on wire rope tension sensor Confidence weights of strain gauge pressure sensors Confidence weights of pin-shaft sensors The weighted fusion algorithm is used to calculate the excavation resistance. The calculation formula is:

[0032] .

[0033] Furthermore, in step S2, the multi-source measured data is preprocessed using an adaptive Kalman filter algorithm.

[0034] Furthermore, the dynamic optimization and calibration of the rigid-flexible coupling dynamic model based on the comparison results specifically includes:

[0035] Based on the simulation stress data obtained in step S3, extract the simulation strain sequence. (t=1,2,...,n, where n is the number of data samples), and a measured strain sequence is formed based on the measured strain data obtained in step S4. ;

[0036] Extract and compare simulated strain sequences and measured strain sequence The following are its core characteristics: peak strain, trough strain, mean strain, strain amplitude, and strain trend coefficient;

[0037] The overall deviation between simulation and measured data is quantified using the Multi-Feature Dynamic Weighted Comprehensive Error (SMWFE). The calculation formula for SMWFE is as follows:

[0038] SMWFE= ;

[0039] in, For the normalization error of each individual feature, The corresponding weights for each individual feature. The normalized error of the strain trend coefficient, This is the trend deviation penalty coefficient. This is the trend error threshold;

[0040] The parameters of the rigid-flexible coupling dynamic model are iteratively optimized until the SMWFE and the peak error and valley error meet the preset accuracy threshold.

[0041] This invention provides a real-time full-domain load spectrum identification method for the working device of a mining excavator, which has the following advantages compared with the prior art:

[0042] 1. Overcoming the limitations of rigid body modeling, a rigid-flexible coupled dynamic model combined with multi-feature dynamic calibration is adopted to solve the problem of mismatch between simulation and actual structural deformation and load transfer, and significantly improve the accuracy of load simulation.

[0043] 2. Integrate multi-source measured data and full-domain simulation data, and cover sensorless areas and fault scenarios through scene adaptive fusion logic to achieve full-domain load capture of the working device without blind spots.

[0044] 3. Relying on real-time data interaction and dynamic error correction mechanisms, it adapts to complex and variable scenarios such as material characteristics and impact conditions, overcoming the static limitations of traditional methods and ensuring the real-time performance, continuity, and high reliability of the load spectrum. (See attached diagram for details.)

[0045] Figure 1 This is a flowchart illustrating a real-time identification method for the full-domain load spectrum of a mining excavator's working device, provided by an embodiment of the present invention.

[0046] Figure 2 This is a schematic diagram of the layout of the excavator environmental action data acquisition device provided in an embodiment of the present invention.

[0047] Figure 3 This is a schematic diagram of the strain gauge bonding position of the excavator working device provided in an embodiment of the present invention.

[0048] Figure 4 This is a logic diagram of the global load fusion model provided in the embodiments of the present invention.

[0049] Figure 5 The following is a schematic diagram of the full-domain load spectrum of the working device of the mining excavator provided in the embodiments of the present invention: (a) measured load time history diagram; (b) simulated load time history diagram; (c) fused output load time history diagram.

[0050] In the picture: 1. Crane boom; 2. Stick; 3. Bucket; 4. Wire rope; 5. Head sheave. Detailed Implementation

[0051] The core innovation of this invention addresses the technological gap in existing technologies that separate simulation and field measurements and fail to integrate the core approach to obtaining load spectra. It proposes a unique closed-loop concept combining real-time linkage of dual data sources with scene adaptive matching. Instead of treating simulation and field measurements as isolated data sources, it constructs an integrated system that combines rigid-flexible coupling simulation for full-domain coverage, multi-source field measurements for a true benchmark, dynamic calibration to correct model biases, and scene adaptive fusion and weight allocation.

[0052] The core logic of this closed-loop system is as follows: Simulation of a rigid-flexible coupled model is driven by real-time multi-source data; the model is dynamically calibrated using measured strain data; and finally, the fusion weights of the two types of data are adaptively adjusted according to the working conditions to output a load spectrum with full coverage, real-time response, and high reliability. Multi-source measurements provide a realistic benchmark for the closed loop; the rigid-flexible coupled model provides full coverage; dynamic calibration addresses the deviation between the model and reality; and scenario fusion adapts to the dynamic needs of complex working conditions. These four elements form a logical chain of input, calibration, fusion, and output, none of which can be omitted, ultimately achieving a full-domain, real-time, and high-precision load spectrum.

[0053] This invention employs a multi-method fusion strategy because existing technologies cannot meet the accuracy, coverage, and real-time precision identification requirements of the full-domain load spectrum for mining excavator working devices. While single simulation methods can cover the entire domain, they are prone to calculation distortion due to their inability to capture sudden changes in material hardness and nonlinear structural deformation in real time. Single measurement methods, while accurate, cannot be applied to areas where strain gauges cannot be attached, such as enclosed cavities and hinge points, leading to overall fragmentation of the load spectrum. Traditional offline calibration methods cannot meet the requirements for real-time identification. Mining operations are complex and variable, necessitating a technical solution that can dynamically adapt to different scenarios, balance accuracy and coverage, and respond to changes in operating conditions in real time. Therefore, this invention organically integrates various technical modules according to the logic of simulation providing full-domain coverage, measurement providing a real benchmark, calibration improving model accuracy, and fusion adapting to different operating conditions, providing a real-time identification method for the full-domain load spectrum of mining excavator working devices.

[0054] The innovation of this invention lies in breaking through the limitations of existing technologies that isolate simulation and actual measurement or only perform offline static correction. For the first time, it treats both as equal real-time data sources, constructing a closed loop that drives simulation through multi-source actual measurement, outputs full-domain data from simulation, dynamically calibrates the actual measurement model, and fuses the output load spectrum. This achieves seamless data linkage, solving the pain point of existing technologies where accuracy and coverage cannot be simultaneously achieved. Furthermore, it establishes a two-dimensional adaptation logic based on sensor status and operating conditions. For typical mining conditions, it automatically adjusts the fusion weights of simulation and actual measurement data, avoiding the problem of fixed fusion strategies in existing technologies that cannot adapt to complex operating conditions, thus achieving accurate adaptation in dynamic scenarios.

[0055] This invention is achieved using the following technical solution:

[0056] A method for real-time identification of the full-domain load spectrum for the working device of a mining excavator includes the following steps:

[0057] S1: Construct a rigid-flexible coupled dynamic model of a mining excavator using simulation modeling software; for at least two structural components of the excavator's working device, namely the boom and stick, establish a flexible finite element model and solve it using the rigid-flexible coupled nonlinear dynamic coupling control equation; simplify the modeling of the non-working device that provides support and movement functions for the working device.

[0058] The excavator is a rack and pinion push-press excavator; the working device includes at least a boom, stick, bucket, push-press mechanism and lifting mechanism; the non-working device includes at least a tracked chassis, slewing platform, lifting support and counterweight.

[0059] The rigid-flexible coupling nonlinear dynamic coupling control equation is as follows:

[0060] ;

[0061] The rigid-flexible coupled nonlinear dynamic coupled control equations are applied to the two structural components, the boom and the stick; for each of the boom and the stick, the equations in... , , A, E, I, and F represent the axial dynamic displacement, lateral dynamic displacement, material density, cross-sectional area, elastic modulus, moment of inertia, and dynamic distributed load of the structural component, respectively.

[0062] After completing the geometric modeling, in order to numerically solve the aforementioned rigid-flexible coupled nonlinear dynamic coupling control equations, it is necessary to mesh the model and define the flexible body. The boom and stick are meshed using tetrahedral solid elements, with a basic mesh size of 50mm; and the mesh is refined at the connection between the stick and bucket, around the boom hinge hole, and in the weld area, reducing the mesh size to 20mm.

[0063] S2: Collect multi-source measured data during the excavator's operation in real time, and preprocess the multi-source measured data using an adaptive Kalman filter algorithm.

[0064] The multi-source measured data includes motion parameter data, environmental effect data, and material characteristic parameters matched based on a dynamic database of mine material characteristics.

[0065] The motion parameter data is collected in real time through automated software and the excavator's electronic control system. The collected motion parameters include at least the pushing speed, lifting speed, and slewing angular velocity. Based on the collected motion parameters, the pushing displacement, lifting displacement, and slewing angular displacement are obtained through integral calculation, and the displacements are mapped to the boundary conditions of the rigid-flexible coupled dynamic model through coordinate system transformation.

[0066] The dynamic database of mine material characteristics stores physical parameters of various mine materials and their corresponding visual appearance features. It can match and call the corresponding physical parameters based on the material appearance features collected in real time, and can iteratively update the stored parameters according to the operation feedback data.

[0067] The environmental impact data includes excavation resistance data obtained through multi-sensor fusion and material gravity data obtained through a pressure sensor array.

[0068] The acquisition of the excavation resistance data specifically includes:

[0069] The tension of the wire rope in the hoisting mechanism is collected by a wire rope tension sensor. and its angle with the horizontal direction ;

[0070] The contact pressure between the bucket teeth and the material is collected by strain gauge pressure sensors mounted on the bucket teeth. , i=1,2,3,4,5;

[0071] The pin torque T is collected by a pin sensor installed at the boom-bucket hinge pin, and then calculated using the formula... Converted to force at the hinge r is the radius of the pin shaft;

[0072] Confidence weighting based on wire rope tension sensor Confidence weights of strain gauge pressure sensors Confidence weights of pin-shaft sensors The weighted fusion algorithm is used to calculate the excavation resistance. The calculation formula is:

[0073] .

[0074] S3: The preprocessed multi-source measured data is used as a driving signal and input to the rigid-flexible coupling dynamic model for real-time simulation calculation to obtain the simulation stress data of the entire working device.

[0075] S4: By deploying strain sensors at key high-stress areas on the excavator body, the measured strain data at the corresponding locations are collected in real time.

[0076] S5: Compare the strain information in the simulated stress data with the measured strain data in all dimensions, and dynamically optimize and calibrate the rigid-flexible coupling dynamic model based on the comparison results until the error between the simulated stress data and the measured strain data meets the preset accuracy requirements.

[0077] The dynamic optimization and calibration of the rigid-flexible coupling dynamic model based on the comparison results specifically includes:

[0078] Based on the simulation stress data obtained in step S3, extract the simulation strain sequence. (t=1,2,...,n, where n is the number of data samples), and a measured strain sequence is formed based on the measured strain data obtained in step S4. ;

[0079] Extract and compare simulated strain sequences and measured strain sequence The following are its core characteristics: peak strain, trough strain, mean strain, strain amplitude, and strain trend coefficient;

[0080] The overall deviation between simulation and measured data is quantified using the Multi-Feature Dynamic Weighted Comprehensive Error (SMWFE). The calculation formula for SMWFE is as follows:

[0081] SMWFE= ;

[0082] in, For the normalization error of each individual feature, The corresponding weights for each individual feature. The normalized error of the strain trend coefficient, This is the trend deviation penalty coefficient. This is the trend error threshold;

[0083] The parameters of the rigid-flexible coupling dynamic model are iteratively optimized until the SMWFE and the peak error and valley error meet the preset accuracy threshold.

[0084] S6: Based on the optimized and calibrated rigid-flexible coupling dynamic model, construct a global load fusion model.

[0085] The core objective of the global load fusion model is to achieve real-time, continuous, and highly reliable output of the global load across the excavator's working device by accurately integrating two types of core data sources. The decision-making logic automatically adjusts the weight ratio of electrical drive simulation and measured strain data for different operating environments.

[0086] The core of the model lies in its adaptive adjustment of the fusion weights based on the actual working conditions. Specific working scenarios include:

[0087] Scenario 1: During normal, stable operation with uniform material distribution and no severe vibration, the linearity of physical sensors is optimal. Fusion verification can identify logical errors caused by temperature drift or loosening of the sensors. In contrast, simulation methods often fail to recognize changes in such real-world environments due to fixed model parameters. Therefore, in this scenario, measured strain data is used as the core, with electrical drive simulation data serving as redundant verification items.

[0088] Scenario 2: Severe impact conditions during hard rock excavation, jamming, and collision with isolated boulders. In this instant of impact, physical sensors cannot fully record the rapid stress escalation process due to mechanical lag. However, electrical signals respond extremely quickly, allowing the transient load process to be reconstructed and the transient stress gradient to be completed by driving a simulation model. Finally, the peak deviation can be calibrated using measured data.

[0089] Scenario 3: Harsh mining environment leads to wire breaks or spalling, resulting in intermittent sensor failures. To ensure the continuity of the load spectrum when hardware is damaged, a virtual sensor mechanism is automatically activated, using measured feedback from adjacent measuring points and simulation models for reconstruction.

[0090] Scenario 4: In structural monitoring blind spots such as inside enclosed cavities, hinged gaps, and narrow welds, sensors cannot be attached due to physical installation limitations. In these areas, the load identification task is directly undertaken by the electrical drive simulation data after real-time dynamic calibration.

[0091] Meanwhile, the model introduces a composite error correction term. This correction term is not a fixed value, but is dynamically adjusted in real time based on the work cycle and working condition. It can effectively compensate for systematic errors under different working conditions and different data sources, further improve the accuracy of the fused load data, and ensure that the final output full-domain load spectrum not only conforms to the authenticity of the measured data, but also has full-domain coverage of the simulation data.

[0092]

[0093] The final load at position p at time t. It is an adaptive weighting factor.

[0094] S7: Based on the global load fusion model, the global load spectrum of the excavator working device is identified and output in real time.

[0095] The present invention will be further explained and described below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only some embodiments, 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.

[0096] Example 1

[0097] This embodiment provides a real-time identification method for the full-domain load spectrum of a mining excavator's working device. The aim is to obtain high-precision dynamic load data across the entire working device through real-time interaction and dynamic calibration between a rigid-flexible coupling simulation model and field measured data. The core of this method lies in data-driven simulation, online comparison and optimization, and full-domain fusion reconstruction. Its implementation process is as follows: Figure 1 As shown, the specific steps include:

[0098] S1: Create a three-dimensional geometric model of the mining excavator using SolidWorks and import it into a multibody dynamics simulation platform to build a rigid-flexible coupled dynamic model.

[0099] This embodiment takes a rack and pinion push-type excavator as an example, such as... Figures 2-3As shown, its working device includes a lifting boom 1, a stick 2 hinged to the lifting boom 1 and driven by a pushing mechanism, a bucket 3 hinged to the front end of the stick 2, and a lifting mechanism for lifting the bucket 3; wherein, the pushing mechanism includes a pushing motor, a reducer, and a pushing pinion meshing with the rack of the stick 2; the lifting mechanism includes a lifting motor, a reducer, a drum, and a wire rope 4, the wire rope 4 being guided by a sheave 5 installed at the top of the lifting boom 1 and then connected to the top lifting point of the bucket 3; its non-working device (i.e., the components that provide support and movement for the working device) mainly includes a tracked chassis, a slewing platform, an A-frame, and a counterweight, wherein the A-frame is fixed on the slewing platform and is used to support the rear hinge point of the lifting boom 1.

[0100] During modeling, the working device is precisely modeled, achieving an accuracy of over 95% of the actual dimensions, while retaining detailed features of key connecting structures such as hinge holes and pins. Non-working devices are reasonably simplified without affecting the overall dynamic characteristics to reduce the computational load on the model.

[0101] For the two core load-bearing components, boom 1 and stick 2, a flexible finite element model is established, and a rigid-flexible coupled nonlinear dynamic coupling control equation is used to solve the problem, replacing the conventional linear finite element solution method, to achieve precise coupling between structural dynamic deformation and load transfer. The equations are applied to boom 1 and stick 2 respectively, and for each component, the equation form is as follows:

[0102] ;

[0103] in, This indicates the axial dynamic displacement of the component (boom 1 or stick 2); This indicates the lateral dynamic displacement of the component; The material density of the component is 7850 kg / m³ in this embodiment; A is the cross-sectional area of ​​the component; E is the elastic modulus of the component, which is 210 GPa in this embodiment; I is the moment of inertia of the component; F is the dynamic distributed load of the component, which integrates excavation resistance, inertial force, driving force and environmental load.

[0104] After completing the geometric modeling, in order to numerically solve the aforementioned rigid-flexible coupled nonlinear dynamic coupling control equations, it is necessary to mesh the model and define the flexible body. Tetrahedral solid elements are used for meshing, with a basic mesh size of 50mm selected to balance computational efficiency and accuracy. For key stress concentration areas such as the connection between the boom 2 and bucket 3, the periphery of the hinge hole of the boom 1, and welds, the mesh is refined, with the mesh size reduced to 20mm to accurately capture local high stress distributions.

[0105] In addition, constraints need to be added to the model to simulate real connections: rotational constraints are added between boom 1 and slewing platform, between stick 2 and boom 1, and between bucket 3 and stick 2. All constraints are given a friction coefficient to simulate the frictional resistance of actual articulated joints, thus improving the realism of the model simulation.

[0106] S2: Real-time acquisition of multi-source measured data under normal operation of the excavator, including motion parameter data, environmental effect data, and material characteristic parameters matched based on the dynamic database of mine material characteristics, and preprocessing of the raw data.

[0107] (1) Construction and matching of material property database

[0108] A dynamic database of mine material properties is constructed, linking and storing key physical parameters (such as hardness, coefficient of friction, and density) of various mine materials with their corresponding visual appearance characteristics (such as color and particle size). Parameter values ​​are obtained through statistical fitting of laboratory tests and field measurement data, with an initial data accuracy of ±3%. During operation, material appearance characteristics are collected using lidar and matched against the database, automatically retrieving the corresponding basic physical parameters of the material. Simultaneously, the deviation between the measured material characteristics and the basic parameters for each operation is recorded, and the database parameters are continuously updated through iterative optimization to improve matching accuracy.

[0109] (2) Acquisition and processing of motion parameter data

[0110] Motion parameter data is acquired through direct communication between the Siemens TIA V16 automation software and the excavator's PLC electronic control system using the Profinet real-time Ethernet protocol, with the acquisition frequency set to 10Hz. The pushing speed is captured synchronously. Thrust acceleration Speed ​​increase , increase acceleration angular velocity angular acceleration These are the raw parameters. The acquisition process employs a dual-buffered frame synchronization mechanism to remove data with missing frames or failed verification, ensuring data integrity.

[0111] Based on the collected velocity and acceleration data, the cumulative displacement is calculated using the trapezoidal integral method, and a drift correction term is introduced to eliminate integration errors. The core displacement is calculated as follows:

[0112] Thrust displacement :

[0113] ;

[0114] in, τ represents the initial pushing displacement; τ is the integration time variable. It is the fundamental displacement of the velocity integral; This is the acceleration drift correction factor, set to 0.02; The sampling time interval (corresponding to a 10Hz sampling frequency) =0.1s).

[0115] Increase displacement :

[0116] ;

[0117] in, This is the initial lift displacement; It is the angle between the wire rope 4 and the vertical direction, used to correct the horizontal velocity component.

[0118] angular displacement :

[0119] ;

[0120] in, The initial rotation angle, It is an angular acceleration compensation term that improves the dynamic response accuracy of angular displacement calculation.

[0121] The calculated thrust displacement , increase displacement angular displacement The boundary conditions of the rigid-flexible coupled dynamic model are mapped using the coordinate system transformation matrix: pushing displacement. As the boundary condition for axial translation at the hinge joint between boom 2 and lifting arm 1; lifting displacement Mapped to the vertical translation boundary conditions of the crane boom head; rotational angular displacement As the rotational boundary condition of the rotating platform. This is determined by a lidar (installation location see...). Figure 2 (At point a) real-time scanning of the three-dimensional coordinates of the end of the boom 2 and the edge of the bucket 3 to extract the actual displacement. Calculate displacement deviation ,when Model boundary condition correction is triggered when the value is > 5mm.

[0122] (3) Environmental impact data collection

[0123] Environmental impact data is accurately collected through sensor arrays and multi-source data fusion algorithms, mainly including excavation resistance data and material gravity data.

[0124] Obtaining excavation resistance data: A three-source collaborative solution of tension sensor + pressure sensor + pin sensor is adopted.

[0125] By using the wire rope 4 at a position close to the bucket 3 (see...)Figure 2 An S-type wire rope tension sensor (range 0-500kN, accuracy ±0.1%FS) is installed at point b to collect the wire rope tension. and its angle with the horizontal direction .

[0126] Through the tips of the 5 main bucket teeth (see Figure 2 A strain gauge pressure sensor (range 0-300kN, accuracy ±0.2%FS) is embedded at point c to collect the contact pressure between the bucket teeth and the material. (i=1-5), and use mean filtering with a window size of 5 to eliminate impulse noise.

[0127] By means of the boom-bucket hinge pin (see Figure 2 A pin-type torque sensor (range 0-100kN·m, accuracy ±0.15%FS) is installed at point d to collect the pin torque T, and the torque is measured using the formula... (r is the radius of the pin) Converted to force at the hinge .

[0128] Based on sensor accuracy and historical error statistics, determine the confidence weight of each sensor: =0.4 (tension sensor) =0.3 (pressure sensor) =0.3 (pin sensor). The final excavation resistance is calculated using a weighted fusion algorithm. :

[0129]

[0130] Material gravity data acquisition: On the inner wall of bucket 3 (see...) Figure 2 Eight array-type pressure sensors are evenly arranged at point e to collect the contact pressure between the material and the inner wall of the bucket 3 in real time. Simultaneously, the bucket tilt angle is recorded through the electronic control system. Calculate the total pressure of the material on bucket 3. = ( (This refers to the sensitive area of ​​a single sensor). The material's gravity G is calculated using the following formula:

[0131] G= ;

[0132] in, , This refers to the sliding friction between the material and the inner wall of the bucket 3; The friction coefficient between the material and the inner wall of the bucket 3 is determined by the type of material identified by the lidar and retrieved from the database. The normal pressure of the material on bucket 3 is obtained by decomposing data from an array of pressure sensors.

[0133] (4) Data preprocessing

[0134] An adaptive Kalman filter algorithm was used to preprocess the aforementioned multi-source measured data. This algorithm can dynamically adjust the process noise covariance Q and the observation noise covariance R (for example, Q=10 under steady-state conditions). -4 R=10 -3 Under impact conditions, Q=10 -3 R=10 -2 This allows for rapid filtering of electrical noise and vibration interference, with a single frame processing time of ≤0.02ms. After preprocessing, the multi-source measured data are uniformly converted into CSV format, and the geometric mapping of parameters such as excavation resistance and slope angle is completed to generate simulation-recognizable drive signals.

[0135] S3: Real-time simulation calculation based on driving signals

[0136] Through the API interface of the dynamic simulation software, the preprocessed drive signal is input in real time at a frequency of 10Hz into the rigid-flexible coupled dynamic model constructed in step S1, with the solution step size set to 0.1s, to achieve real-time linkage simulation between the model and the on-site working conditions. The simulation calculation extracts stress-time data of all finite element nodes of boom 1 and stick 2 in real time, covering unmeasurable points such as hinge holes and welds that are traditionally difficult to install sensors on, thus initially constructing a simulation stress dataset of the entire working device.

[0137] S4: Acquisition of measured strain data

[0138] In the critical high-stress areas of the excavator physical prototype (see...) Figure 3 Specific positions of the boom 1 and stick 2 are shown in the diagram: Position f is the lower end of the boom 1 connected to the wire rope 4, bearing combined tensile and compressive stress; position g is the lower part of the main boom of boom 1 near the sheave 5, bearing bending stress; positions h and p are the symmetrical upper and lower areas above and below the hinge point connecting stick 2 and bucket 3, key positions bearing bending moment and shear force; position q is the hinge point area connecting boom 1 and slewing platform, key stress position bearing maximum bending moment and support reaction force. Strain gauges are attached to these positions to collect measured strain data in real time during operation, forming a measured strain sequence. (t=1,2,...,n, where n is the number of data samples), providing a direct basis for subsequent model calibration.

[0139] S5: Model Dynamic Optimization and Calibration

[0140] The edge computing module extracts strain information from the simulated stress data in S3 to form a simulated strain sequence. (t=1,2,...,n, where n is the number of data samples) and the measured strain sequence acquired by S4 Perform full-dimensional feature comparison and dynamic optimization calibration.

[0141] First, the following core features of the two strain sequences are extracted simultaneously:

[0142] Peak strain: , ;

[0143] Strain valley value: , ;

[0144] Mean strain: , ;

[0145] Strain amplitude: , ;

[0146] Strain trend coefficient: obtained through linear regression By fitting the strain sequence and using the slope k as a trend coefficient, we obtain... and .

[0147] Calculate the normalization error of each individual feature:

[0148] ;

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] in, , , , , The corresponding engineering limit value is used to normalize the error to the [0,1] interval.

[0154] Then, the Multi-Feature Dynamic Weighted Comprehensive Error (SMWFE) is used to quantify the overall deviation between the simulation and measured data. Based on the load characteristics of the mining excavator, fixed weights are set for each feature error: , , , , The formula for calculating SMWFE is:

[0155] SMWFE= ;

[0156] in, The trend deviation penalty coefficient is set to 1.2. The trend error threshold is set to 0.04. The added penalty term aims to strengthen trend consistency.

[0157] Finally, based on SMWFE and key error indicators, the parameters of the rigid-flexible coupling dynamic model were iteratively adjusted (such as local correction of material properties and fine-tuning of connection stiffness) for dynamic optimization and calibration. The optimization process continued until the preset accuracy requirements were met: SMWFE ≤ 3.5%, and peak error... ≤ 2.5%, Valley error ≤ 2.5%.

[0158] S6: Construct a global load fusion model

[0159] Based on the high-precision rigid-flexible coupling dynamic model optimized and calibrated in step S5, a global load fusion model is constructed. The core logic is scene adaptive data integration + dynamic error correction, which realizes the accurate fusion of the authenticity of measured data and the global simulation data.

[0160] The model takes two types of data as its core inputs: first, the measured strain data collected in step S4 covering key high-stress measurement points; and second, the electrical + simulation binding data generated in real time in step S3, which can cover the entire working device. The model integrates and outputs real-time, continuous, and highly reliable full-domain load data, which not only solves the problem of incomplete coverage of measured points but also avoids the distortion defects of single simulation data.

[0161] like Figure 4 As shown, the model automatically adjusts the data integration strategy by monitoring the status of measuring points (whether sensors are present or faulty) and operating conditions (such as stable and impact conditions) in real time. The specific adaptation logic is as follows:

[0162] When there are sensors and the operating conditions are stable, load calculation is performed based on measured strain data. Electrical and simulation data are bound together for redundancy verification to avoid logical errors from a single sensor and ensure data authenticity.

[0163] When a sensor is present but encounters impact conditions such as excavating hard rock or a stuck blade, the simulation model is driven by the rapid response characteristics of electrical data to complete the transient stress gradient. Then, the peak deviation is calibrated by actual measurement data, taking into account both dynamic response speed and peak accuracy.

[0164] In areas without sensor coverage, such as enclosed cavities and hinged gaps, the load calculation task is directly undertaken by electrical + simulation data. The reliability of the data is ensured by a dynamically calibrated model, achieving full coverage without blind spots.

[0165] When a sensor fails but the operating conditions are stable, the virtual sensor mechanism is automatically activated. Electrical and simulation data are used to replace the faulty sensor data, and the measured data from adjacent normal measurement points are called to calibrate the model. This avoids load spectrum breakage caused by single-point failure and ensures the continuity of output data.

[0166] The model employs a non-fixed composite error correction term, which is adjusted based on key information such as average strain deviation, operating cycle, and peak deviation from two types of data sources. Different correction weights are applied according to the type of operating condition. Under stable operating conditions, the focus is on correcting the average strain deviation and peak deviation, while under impact operating conditions, the emphasis is on optimizing the compensation effect of peak deviation, effectively compensating for system errors in different scenarios.

[0167] S7: Real-time identification and output of global load spectrum

[0168] Based on the global load fusion model in S6, the system identifies and outputs a dynamic load spectrum covering the entire area of ​​boom 1 and stick 2 in real time. (See schematic diagram below.) Figure 5 This load spectrum includes both measured correlation data of strain gauge bonding points and high-confidence simulation data of unmeasurable points such as closed cavities and hinged gaps.

[0169] The output load spectrum data format supports direct import from various mainstream engineering software (such as fatigue analysis software and finite element analysis software), and can be used for accurate fatigue life calculation and lightweight structural optimization design. Simultaneously, the system automatically generates a complete load spectrum report, including data acquisition condition descriptions, model optimization and calibration process records, load spectrum accuracy verification results, and key load statistics (such as maximum stress, average stress, and load cycle count), providing full-process technical support for engineering applications.

[0170] Through the implementation of the above seven steps, this invention achieves a leap from offline disconnect to online fusion of simulation and measured data, solving the problems of insufficient load spectrum accuracy and poor dynamic adaptability caused by traditional methods due to model distortion, incomplete measurement point coverage, and lack of real-time fusion. It provides a reliable data foundation for the precise design, safe operation and maintenance, and full life cycle cost management of mining excavators.

[0171] Example 2

[0172] This embodiment verifies the necessity of iterative optimization of the model for improving the accuracy of load calculation. Within the framework of the method described in Embodiment 1, the same mining excavator was tested under typical operating conditions. The degree of agreement between the simulation calculation results and measured strain data was compared using the original rigid-flexible coupling dynamic model, the model after the first iteration optimization, and the model after the second iteration optimization. Evaluation indicators included multi-feature dynamic weighted composite error (SMWFE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determination (R²). The comparison results are shown in Table 1.

[0173] Table 1 Comparison of Iterative Optimization Accuracy of Simulation Models

[0174]

[0175] As shown in Table 1, after two dynamic iterative optimization calibrations, the multi-feature dynamic weighted comprehensive error (SMWFE) of the model significantly decreased from 8.5% to 2.9%, and the error indices of each component were also greatly improved, with R² approaching 1. This fully demonstrates that the dynamic optimization calibration step in this invention is crucial and effective in improving the accuracy of the simulation model to meet the requirements of engineering applications (such as SMWFE ≤ 3.5% set in Example 1).

[0176] Example 3

[0177] This embodiment is used to illustrate the superiority of the method of the present invention. Three typical load spectrum compilation schemes in the prior art are selected as comparative examples and compared with the method of the present invention (denoted as the dynamic identification method) under the same mining excavator test conditions. The three core performance indicators are mainly evaluated: load data coverage, stress deviation rate at key points, and the matching degree between the load spectrum and fatigue life analysis. The comparison results are shown in Table 2.

[0178] Table 2 Comparison of Core Performance Indicators of Different Load Spectrum Compilation Schemes

[0179]

[0180] Analysis of Table 2 shows that:

[0181] Data coverage: The method of this invention has the highest coverage (95%) because it uses simulation calculations to cover the entire area and calibrates with measured data to ensure its reliability, which is much higher than the measured method that relies on only a limited number of measurement points (60%).

[0182] Key point deviation rate: Thanks to the accuracy of the rigid-flexible coupling dynamic model and real-time dynamic calibration, the stress calculation deviation rate at key points (2.1%) is comparable to that of the high-precision measured method (2.4%), and is significantly better than other simulation methods.

[0183] Fatigue life matching degree: This is the ultimate manifestation of the value of load spectrum engineering. The high-precision, full-coverage load spectrum output by the method of this invention results in a matching degree of up to 92% between the fatigue life analysis based on it and the actual life, and its overall performance is significantly better than all comparative examples.

[0184] In summary, the real-time identification method for the full-domain load spectrum of the mining excavator working device provided by this invention effectively overcomes the shortcomings of the prior art by integrating rigid-flexible coupling modeling, real-time multi-source data driving and online dynamic calibration, and has achieved significant improvements in coverage, accuracy and engineering practicality.

[0185] 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 method for real-time identification of the full-domain load spectrum for the working device of a mining excavator, characterized in that: Includes the following steps: S1: Construct a rigid-flexible coupled dynamic model of a mining excavator using simulation modeling software; for at least two structural components of the excavator's working device, namely the boom and stick, establish a flexible finite element model and solve it using the rigid-flexible coupled nonlinear dynamic coupled control equation; simplify the modeling of the non-working device that provides support and movement functions for the working device. S2: Collect multi-source measured data during the excavator's operation in real time and perform preprocessing; S3: The preprocessed multi-source measured data is used as a driving signal and input to the rigid-flexible coupling dynamic model for real-time simulation calculation to obtain the simulation stress data of the entire working device. S4: By deploying strain sensors at key high-stress areas on the excavator body, the measured strain data at the corresponding locations are collected in real time. S5: Compare the strain information in the simulated stress data with the measured strain data in all dimensions, and dynamically optimize and calibrate the rigid-flexible coupling dynamic model based on the comparison results until the error between the simulated stress data and the measured strain data meets the preset accuracy requirements. S6: Based on the optimized and calibrated rigid-flexible coupling dynamic model, a full-domain load fusion model is constructed. The model is adaptively determined according to the type of measurement point and the working condition. It integrates two types of data sources: with sensors, the actual measurement is the core, and without sensors or in case of faults, the simulation fully covers the data. The model is then accurately fused by combining composite error correction terms. S7: Based on the global load fusion model, the global load spectrum of the excavator working device is identified and output in real time.

2. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 1, characterized in that: The excavator is a rack and pinion push-press excavator; the working device includes at least a boom, stick, bucket, push-press mechanism and lifting mechanism; the non-working device includes at least a tracked chassis, slewing platform, lifting support and counterweight.

3. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 2, characterized in that: The rigid-flexible coupling nonlinear dynamic coupling control equation is as follows: ； The rigid-flexible coupled nonlinear dynamic coupled control equations are applied to the two structural components, the boom and the stick; for each of the boom and the stick, the equations in... , , A, E, I, and F represent the axial dynamic displacement, lateral dynamic displacement, material density, cross-sectional area, elastic modulus, moment of inertia, and dynamic distributed load of the structural component, respectively.

4. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 3, characterized in that: In step S1, the boom and stick are divided into tetrahedral solid unit meshes with a basic mesh size of 50mm; and the mesh is densified at the connection between the stick and the bucket, around the boom hinge hole, and in the weld area, with the mesh size reduced to 20mm.

5. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 2, characterized in that: The multi-source measured data includes motion parameter data, environmental effect data, and material characteristic parameters matched based on a dynamic database of mine material characteristics; The motion parameter data is collected in real time through automated software and the excavator's electronic control system. The collected motion parameters include at least the pushing speed, lifting speed, and slewing angular velocity. Based on the collected motion parameters, the pushing displacement, lifting displacement, and slewing angular displacement are obtained through integral calculation, and the displacements are mapped to the boundary conditions of the rigid-flexible coupled dynamic model through coordinate system transformation.

6. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 5, characterized in that: The dynamic database of mine material characteristics stores physical parameters of various mine materials and their corresponding visual appearance features. It can match and call the corresponding physical parameters based on the material appearance features collected in real time, and can iteratively update the stored parameters according to the operation feedback data.

7. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 5, characterized in that: The environmental impact data includes excavation resistance data obtained through multi-sensor fusion and material gravity data obtained through a pressure sensor array.

8. The method for real-time identification of the full-domain load spectrum for a mining excavator working device according to claim 6, characterized in that: The acquisition of the excavation resistance data specifically includes: The tension of the wire rope in the hoisting mechanism is collected by a wire rope tension sensor. and its angle with the horizontal direction ; The contact pressure between the bucket teeth and the material is collected by strain gauge pressure sensors mounted on the bucket teeth. , i=1,2,3,4,5; The pin torque T is collected by a pin sensor installed at the boom-bucket hinge pin, and then calculated using the formula... Converted to force at the hinge r is the radius of the pin shaft; Confidence weighting based on wire rope tension sensor Confidence weights of strain gauge pressure sensors Confidence weights of pin-shaft sensors The weighted fusion algorithm is used to calculate the excavation resistance. The calculation formula is: 。 9. The method for real-time identification of the full-domain load spectrum for the working device of a mining excavator according to claim 1, characterized in that: In step S2, the adaptive Kalman filter algorithm is used to preprocess the multi-source measured data.

10. A method for real-time identification of the full-domain load spectrum for a mining excavator working device according to claim 1, characterized in that: The dynamic optimization and calibration of the rigid-flexible coupling dynamic model based on the comparison results specifically includes: Based on the simulation stress data obtained in step S3, extract the simulation strain sequence. (t=1,2,...,n, where n is the number of data samples), and a measured strain sequence is formed based on the measured strain data obtained in step S4. ; Extract and compare simulated strain sequences and measured strain sequence The following are its core characteristics: peak strain, trough strain, mean strain, strain amplitude, and strain trend coefficient; The overall deviation between simulation and measured data is quantified using the Multi-Feature Dynamic Weighted Comprehensive Error (SMWFE). The calculation formula for SMWFE is as follows: SMWFE= ; in, For the normalization error of each individual feature, The corresponding weights for each individual feature. The normalized error of the strain trend coefficient, This is the trend deviation penalty coefficient. This is the trend error threshold; The parameters of the rigid-flexible coupling dynamic model are iteratively optimized until the SMWFE and the peak error and valley error meet the preset accuracy threshold.

Images