A method for detecting damage to a connector, the connector, and a meter box.

Abstract

This invention relates to the field of connector technology, providing a method for detecting damage to connectors, a connector, and a meter box. The detection method includes the following steps: using an ultrasonic phased array probe to perform a full-matrix capture scan on the physically tested connector to obtain raw internal ultrasonic data; based on the raw internal ultrasonic data, reconstructing a three-dimensional acoustic image of the connector using a full-focusing method, and calculating a first three-dimensional stress field based on the three-dimensional acoustic image; establishing a parametric three-dimensional finite element model of the connector, and running an explicit dynamic simulation on the parametric three-dimensional finite element model to calculate a second three-dimensional stress field; performing a voxel-to-voxel subtraction operation between the first and second three-dimensional stress fields to generate a three-dimensional stress difference field; and diagnosing whether there is damage inside the connector and locating the damage location based on the local stress difference peak value in the three-dimensional stress difference field. This solves the problem of testing stress-related damage inside connectors.

Description

Technical Field

[0001] This invention relates to the field of connector technology, and in particular to a method for detecting damage to connectors, a connector, and a meter box. Background Technology

[0002] As a key connection component in smart grids, meter box connectors are directly responsible for the electrical connection, current conduction, and distribution between the electricity metering device and external lines. Their reliability directly affects the accuracy of electricity metering and the stable operation of the power system. Currently, these connectors typically adopt a modular design, including a frame, detachable adapter plugs, self-locking springs, and other precision structures to achieve quick plugging and unplugging and reliable connection.

[0003] In existing technologies, modular meter box connectors face several technical challenges. Firstly, in terms of structural design, traditional connectors are complex to install and disassemble, easily leading to component damage. Furthermore, long-term use may result in loose connections, affecting the accuracy of meter readings and the stability of the power system. Secondly, considering that different usage scenarios require access to different external circuits, traditional connectors lack sufficient versatility, necessitating the selection of different connector specifications based on the type of external circuit, increasing maintenance costs and operational complexity.

[0004] During manufacturing and quality control, connectors undergo various physical tests, including mating life, mechanical shock, and short-circuit electrodynamics, to verify their performance and durability. These tests may cause damage such as residual stress, microcracks, plastic deformation, or contact fatigue within the connector. This damage is often invisible initially but gradually evolves, eventually leading to increased connection resistance, overheating, or even failure, resulting in measurement errors or connection malfunctions.

[0005] Currently, the inspection of connectors after physical testing mainly relies on traditional methods such as external visual inspection, electrical continuity testing, and dimensional measurement. These methods can only detect macroscopic, surface defects and are powerless against potential damage such as microcracks and material yielding caused by internal stress concentration. Although technologies such as industrial CT and X-ray imaging can detect internal geometric defects, they are difficult to quantitatively assess the mechanical state (such as residual stress and stress concentration), and cannot determine the severity of the damage or its impact on future performance.

[0006] In the field of nondestructive testing (NDT) technology, ultrasonic testing is widely used in industry due to its high sensitivity to anomalies in the internal structure of materials. Traditional ultrasonic testing methods (such as the pulse-echo method) qualitatively determine the presence, location, and size of defects by analyzing the propagation time, amplitude attenuation, and echo characteristics of sound waves in the material. However, for devices with complex structures, such as modular meter box connectors, containing multiple material interfaces and tiny moving parts, traditional ultrasonic methods face significant challenges: the complex geometry and multiple interfaces generate a large amount of stray echoes and acoustic noise, severely interfering with the identification and extraction of true defect signals.

[0007] More importantly, traditional qualitative or semi-quantitative non-destructive testing methods mainly focus on the discovery of geometric defects, while stress-related damage that may occur during physical testing (such as microplastic deformation and fatigue initiation caused by stress concentration) cannot be detected by simple acoustic imaging. Although ultrasonic stress measurement technology based on acoustoelasticity theory is an advanced quantitative non-destructive testing method and has been studied for stress measurement in fields such as welding and bearings, its application to the quality inspection of complex meter box connectors remains a gap. The successful application of this technology heavily relies on accurate acoustic wave propagation modeling of complex three-dimensional structures, precise description of different material interfaces and contact states, and data processing algorithms that accurately invert the measured acoustic signals (sound velocity, attenuation) into a three-dimensional stress field.

[0008] Therefore, there is an urgent need in this field for an innovative solution that not only improves the structural design of connectors themselves, enhancing their reliability and versatility, but also deeply integrates advanced non-destructive testing technology with connector design and manufacturing, enabling accurate quantitative assessment of the internal mechanical health of connectors after physical testing. This solution needs to be able to closely integrate with connector-specific structures (such as guide grooves, C-type springs, and wire unwinding mechanisms) and their dynamic operating states (such as clamping and disconnection), and provide the necessary physical basis and optimized conditions for ultrasonic testing, thereby achieving a quantitative assessment of stress damage within the connector after physical testing. Summary of the Invention

[0009] The purpose of this invention is to quantitatively assess the stress damage inside the connector after physical tests such as fatigue tests.

[0010] To address the above objectives, in one embodiment of the present invention, a (modular meter box) connector is provided, comprising:

[0011] The frame has multiple guide grooves formed inside;

[0012] The adapter plug is detachably disposed within the guide groove;

[0013] The adapter plug includes:

[0014] A current bar is located at the end of the adapter plug near the meter interface, and is used to establish an electrical connection with the meter interface;

[0015] A wiring hole is provided at the end of the adapter plug away from the meter interface for inserting external wires;

[0016] A clamping mechanism, located in the internal cavity of the adapter plug, is made of conductive material and connected to the current bar. It is used to clamp the external wire after it is inserted through the wiring hole, so as to fix it and make it conductive with the current bar.

[0017] The adapter plug has at least a portion of its outer shell made of a material with specific acoustic properties, the acoustic impedance of which matches the acoustic impedance of the internal structural material to optimize the transmission and reflection of ultrasonic waves, facilitating ultrasonic non-destructive testing.

[0018] Optionally, the clamping mechanism includes a self-locking spring;

[0019] The self-locking spring has a C-shaped structure and includes:

[0020] Top section,

[0021] The vertical connecting section has one end connected to the current bar via a connecting claw and fixed to the top of the internal cavity of the adapter plug.

[0022] Clamping section;

[0023] The self-locking spring is made of phosphor bronze, and its acoustic elastic coefficient has been calibrated so that the stress state it experiences can be calculated by measuring the change in the propagation speed of ultrasonic waves within it.

[0024] When an external wire is inserted and contacts the top segment, the clamping segment is configured to elastically deform to pop out and clamp the external wire into the internal cavity;

[0025] The modular meter box connector also includes a wire retraction mechanism that contacts the clamping section;

[0026] The wire ejection mechanism is configured to apply an external force to the clamping section, causing it to spring back and release the clamped external wire, thereby allowing the external wire to exit through the terminal hole;

[0027] The exposed surface of the unwinding button in the unwinding mechanism is made of an acoustic impedance material different from the surrounding housing or has a special geometric texture, so that it can be clearly identified in the ultrasound image and thus serve as a reference benchmark for verifying the operation status.

[0028] Optionally, the material with specific acoustic properties has an acoustic impedance value that gradually changes from the outer surface of the shell inward, smoothly transitioning from an acoustic impedance value Z_out close to that of the coupling agent to an acoustic impedance value Z_in close to that of the internal metal component, and |Z_out-Z_in|>30 MRayl.

[0029] In the optimized design of the connector, the specific implementation of the clamping mechanism further improves the operational reliability and detection accuracy. The clamping mechanism employs a self-locking spring with a C-shaped structure, including a top section, a vertical connecting section, and a clamping section. The vertical connecting section is fixed to the current bar via connecting claws, ensuring mechanical stability. The self-locking spring is made of phosphor bronze, a material that not only possesses excellent elastic conductivity but also has a precisely calibrated acoustoelastic coefficient; for example, the longitudinal wave acoustoelastic coefficient K11 is experimentally determined to be 2.5 × 10⁻⁶. -5 MPa -1 This allows for the calculation of the stress state experienced by the ultrasonic wave by measuring the change in the propagation velocity within the reed. When the external wire is inserted and contacts the top section, the clamping section elastically deforms and pops out, clamping the wire onto the current bar to form a reliable connection. This self-locking mechanism avoids the complexity of traditional methods such as bolts or welding, enabling rapid insertion and removal. Simultaneously, the connector is equipped with a wire release mechanism that contacts the clamping section. An external force is applied via the wire release button to cause the reed to spring back, releasing the wire. The exposed surface of the wire release button uses an acoustic impedance material or special geometric texture different from the surrounding housing, making it clearly visible in the ultrasonic image and serving as a reference for verifying the operational status. This solves the problem of missing positioning references during detection and improves image registration accuracy. Optionally, a number of pits or protrusions are precisely machined on a specific outer surface of the adapter housing. These pits or protrusions have specific geometric shapes and sizes, and their diameter D satisfies the following with respect to the center frequency f of ultrasonic testing: D = V_sound / (2f), where V_sound is the speed of sound in the material. These microstructure arrays together constitute an embedded acoustic lens.

[0030] Optionally, the acoustic impedance of the material with specific acoustic properties is controlled within the range of Z1 ± 0.5 MRayl, where Z1 is the average acoustic impedance of the material used for the current bar and the clamping mechanism in the internal cavity of the adapter plug.

[0031] Furthermore, the material with specific acoustic properties is composited in the housing of the adapter plug in a specific three-dimensional distribution manner. The distribution manner is as follows: acoustic reinforcement material strips arranged in a grid or array are embedded in the housing wall corresponding to the key components in the internal cavity; the acoustic impedance value Z2 of the acoustic reinforcement material strips satisfies |Z2- Z1| ≥ 3MRayl.

[0032] Material acoustic property optimization: It was clarified that the materials of key components (such as adapter plug housing and self-locking spring) must have specific and calibrated acoustic or acoustoelastic properties (such as acoustic impedance matching and known acoustoelastic coefficients). This provides a physical basis and calibration basis for ultrasound-based non-destructive testing methods (such as ultrasonic imaging and acoustoelastic stress measurement).

[0033] In some of the embodiments described above or below, the structural features were identified by adding structural features that could serve as detection reference points (such as the acoustic reflection reference point, the acoustic impedance difference marker, and the RFID tag). These features play a crucial role in subsequent ultrasonic scanning, image registration, feature recognition, and localization, improving the accuracy and repeatability of the detection. Regarding integrated sensing and signal extraction: the possibility of built-in sensors (temperature, strain) was considered, and a signal extraction interface was reserved, enabling the synchronous acquisition of real-time state data during the physical testing process.

[0034] The connector also features a standardized testing interface: a standardized testing window and maintenance channel are provided on the frame, and automatic application of the coupling agent is considered, simplifying the testing operation process, ensuring consistency of testing conditions, and making the testing method easier to implement and integrate. Simultaneously, the fixing and boundary conditions are also considered: a standardized fixture interface for the test platform is added to ensure consistency of the connector's installation boundary conditions during physical testing and non-destructive testing, which helps to effectively compare the experimental stress field with the simulated stress field.

[0035] The connector product itself is designed to be more easily evaluated by the specified testing methods. The added (micro)structural features serve the purpose of "facilitating internal damage detection" and do not change the original basic structure and connection function of the product.

[0036] In another embodiment, an electric meter box is also proposed, including a (modular) electric meter box connector as in any of the other embodiments.

[0037] In another embodiment, a method for detecting test damage to meter box connectors is also proposed, comprising the following steps:

[0038] S11: Use an ultrasonic phased array probe to perform a full matrix capture scan on the physically tested connector to obtain raw internal ultrasonic data of the connector;

[0039] S12: Based on the original internal ultrasound data, the three-dimensional acoustic image of the connector is reconstructed using the full focusing method, and the first three-dimensional stress field is calculated by inversion based on the acoustoelastic theory.

[0040] S13: Establish a parameterized three-dimensional finite element model of the connector, and run an explicit dynamic simulation on the parameterized three-dimensional finite element model to calculate the second three-dimensional stress field simulating the insertion, clamping and unclamping operations.

[0041] S14: Perform a voxel-to-voxel subtraction operation between the first three-dimensional stress field and the second three-dimensional stress field to generate a three-dimensional stress difference field;

[0042] S15: Based on the local stress difference peak value in the three-dimensional stress difference field, diagnose whether there is damage inside the connector and locate the damage location.

[0043] Optionally, step S12 further includes: reconstructing the three-dimensional acoustic image of the connector using the full-focusing method based on the original internal ultrasound data, and calculating the first three-dimensional stress field based on the acoustoelastic theory.

[0044] Optionally, the calculation of the first three-dimensional stress field in step S12 is an inversion calculation based on the acoustoelastic theory, specifically including:

[0045] S21: Based on the original internal ultrasonic data, the stress tensor at each point inside the connector is calculated using the formula relating sound velocity and stress in the theory of acoustoelasticity.

[0046] S22: Based on the calculated stress tensor, generate the first three-dimensional stress field.

[0047] Optionally, the step of performing a full matrix capture scan in step S11 specifically includes:

[0048] S31: Control each crystal of the ultrasonic phased array probe to emit ultrasonic pulses in sequence, and synchronously control all crystals to receive full-waveform echo signals from inside the connector;

[0049] S32: Arrange the full waveform echo signal according to the transmit-receive chip pairs and store it in a full matrix data format;

[0050] S33: Based on the full matrix data, the ultrasonic propagation time delay law is used to perform pixel focusing calculation on a specific three-dimensional spatial region of the connector to generate a high-resolution three-dimensional acoustic image.

[0051] Optionally, the step of establishing and running explicit dynamic simulation in step S13 (i.e., establishing a parameterized three-dimensional finite element model of the connector and running an explicit dynamic simulation on the parameterized three-dimensional finite element model) specifically includes:

[0052] S41: Based on the CAD geometric model of the connector, perform geometric cleanup while retaining the C-shaped gradient curvature feature of the self-locking spring;

[0053] S42: Assign an elastoplastic material constitutive model and Johnson-Cook dynamic yield criterion parameters to the cleaned geometry;

[0054] S43: Define the surface contact pairs and coefficient of friction in the contact area between the self-locking spring and the wire, and at the contact interface between the adapter plug and the guide groove;

[0055] S44: Apply boundary conditions simulating insertion and extraction speeds to the model, solve the transient dynamic equations using an explicit central difference algorithm, and output the second three-dimensional stress field.

[0056] In the explicit dynamics simulation, high-fidelity modeling is crucial for ensuring the accuracy of the stress field. In this embodiment, step S41 cleans up the CAD geometric model of the connector while precisely preserving the C-shaped gradient curvature characteristic of the self-locking spring, a geometric feature that decisively influences the bending deformation and stress distribution of the spring. Step S42 assigns the cleaned model to the constitutive model of the elastoplastic material and the Johnson-Cook dynamic yield criterion parameters, such as setting the hardening coefficient and strain rate sensitivity coefficient for the phosphor bronze material, accurately describing its nonlinear response under dynamic loads. Step S43 defines the surface contact pairs and friction coefficients in the contact area between the self-locking spring and the wire, and at the contact interface between the adapter plug and the guide groove, simulating the sliding and clamping behaviors during actual insertion and removal. Step S44 solves the transient dynamic equations using the explicit central difference algorithm, outputting the second three-dimensional stress field. The synergy of these steps ensures the consistency between the simulation results and the physical experiments.

[0057] The geometric processing in step S41 avoids the impact of redundant features on computational efficiency while retaining key mechanical features, such as the curvature variation region at the root of the reed, which is often a stress concentration point. The Johnson-Cook model in step S42 quantifies the dynamic yielding behavior of the material through the hardening and strain rate terms, for example, simulating the strain hardening effect of the reed under rapid loading. The contact definition in step S43 uses the penalty function method to handle interfacial interactions and calibrates the friction coefficient based on experimental data, for example, setting the metal-insulator interface friction coefficient to the range of 0.15 to 0.18 to closely approximate actual working conditions. The explicit algorithm in step S44 balances computational accuracy and stability through adaptive time step control. This modeling process solves common problems in complex assembly simulations, such as excessive geometric simplification, material model distortion, or difficulty in contact convergence.

[0058] The preprocessing steps S41 to S43 provide physically reasonable input for the dynamic solution, while the algorithm selection in step S44 ensures reliable capture of the transient process. For example, when simulating the insertion and removal operation, the explicit central difference method can effectively handle the large deformation and the contact nonlinearity, avoiding the iterative divergence that may occur with the implicit algorithm.

[0059] The method significantly enhances the root cause analysis capability of damage diagnosis. Through high-precision simulation, the mechanical environment in which damage originates can be reproduced, thereby distinguishing whether stress concentration is caused by structural design defects or material fatigue. For example, the stress field output in step S44 can reveal the plastic accumulation trend of the spring root under cyclic loading, providing a basis for design improvement. Overall, this simulation scheme forms a closed loop with ultrasonic testing, driving damage diagnosis through experimental-simulation differences. Its technical effects are reflected in three aspects: first, the refined modeling reduces simulation errors; second, the dynamic analysis captures transient mechanical behavior; and third, the standardized parameter settings enhance the comparability of the results. This integrated method provides solid support for the reliability design and life prediction of the connector.

[0060] Optionally, the step of diagnosing the injury in step S15 specifically includes:

[0061] S51: Perform spatial Gaussian filtering on the three-dimensional stress difference field to suppress noise and retain local stress difference peaks;

[0062] S52: Set a dynamic damage determination threshold based on the material yield strength of the self-locking spring;

[0063] S53: Identify the continuous regions in the filtered three-dimensional stress difference field that exceed the dynamic damage determination threshold as potential damage regions;

[0064] S54: Classify the damage pattern by combining the geometry of the potential damage area with the stress distribution gradient direction.

[0065] Optionally, the method also includes a step of verifying the explicit dynamic simulation:

[0066] S61: Extract the contact force-time history curve of the self-locking spring clamping section in the explicit dynamic simulation;

[0067] S62: Measure the clamping force-time history data of the actual connector under the same working conditions on a physical test platform;

[0068] S63: Calculate the correlation coefficient and root mean square error of the simulated and measured clamping force-time history curves;

[0069] S64: When the correlation coefficient is greater than 0.95 and the root mean square error is less than a set threshold, the validity of the parameterized three-dimensional finite element model is determined.

[0070] Optionally, the step of classifying damage patterns in step S54 specifically includes:

[0071] S71: Extract the volume, surface area, aspect ratio feature vector, and principal stress direction distribution of the potential damage area;

[0072] S72: Input the feature vector into a pre-trained convolutional neural network damage classifier for pattern recognition;

[0073] S73: Determine the type of microcrack, plastic deformation, or fatigue damage based on the output probability distribution of the convolutional neural network damage classifier;

[0074] S74: Map the classification results to the spatial coordinates of the potential damage area and generate a three-dimensional damage distribution map.

[0075] In the damage diagnosis process, accurate identification and classification of the internal damage patterns are crucial to improving the reliability of the detection. Step S71 provides a multi-dimensional quantitative description of the damage pattern analysis by extracting geometric feature vectors such as the volume, surface area, and aspect ratio of the potential damage area, as well as the principal stress direction distribution. These features can comprehensively characterize the morphology and mechanical state of the damage, overcoming the misjudgment problem caused by traditional methods relying on only a single parameter. Subsequently, in step S72, these feature vectors are input into a pre-trained convolutional neural network damage classifier for pattern recognition. This classifier automatically learns the complex mapping relationship between the damage features and the patterns through a deep learning algorithm, avoiding the subjectivity of manually setting the threshold. In step S73, the microcrack, plastic deformation, or fatigue damage type is determined based on the output probability distribution of the classifier, thereby achieving objective classification of the damage pattern. Finally, in step S74, the classification result is mapped to the spatial coordinates of the damage area to generate a three-dimensional damage distribution map, making the damage information spatially visible.

[0076] The synergistic effect of this series of steps solves the technical problems of low accuracy in damage pattern recognition and reliance on human expert experience in traditional nondestructive testing. The feature extraction step S71 ensures the comprehensiveness of the input data, the neural network classification steps S72 and S73 achieve automated and high-precision pattern discrimination, and the 3D atlas generation step S74 enhances the interpretability and practicality of the results. The entire process, through a data-driven approach, organically combines geometric features with mechanical parameters, significantly improving the accuracy and efficiency of damage classification.

[0077] In step S71, the extraction of the feature vector includes not only basic geometric parameters but also the principal stress direction distribution, which helps distinguish different types of damage mechanisms. For example, microcracks tend to propagate along the direction of the maximum principal stress, while the plastic deformation region exhibits specific stress gradient characteristics. Step S72 utilizes the processing power of the convolutional neural network to automatically filter out the most discriminative indicators from high-dimensional features, avoiding information loss in feature engineering. Step S73 provides a confidence assessment of the classification results through the probability distribution output, enhancing the reliability of the decision. The spatial mapping function in step S74 integrates damage localization with pattern classification, providing direct support for maintenance decisions.

[0078] Through the close coordination of the above steps, the method can effectively identify minute damage inside the connector caused by the fatigue test, solving the problem that traditional visual or electrical inspections cannot detect stress damage. The integration of feature extraction and intelligent classification achieves a deep fusion of data preprocessing and machine learning algorithms.

[0079] Optionally, the method further includes a step of predicting damage evolution based on the three-dimensional damage distribution map:

[0080] S81: Input the size parameters, spatial orientation, and stress intensity factor of the microcrack into the Paris law crack propagation model;

[0081] S82: The evolution of damage variables with load cycles is calculated using a continuous damage mechanics model for the plastic deformation region;

[0082] S83: Integrate the outputs of the crack propagation model and the damage mechanics model to predict the remaining service life of the connector under specified working conditions;

[0083] S84: Couple the remaining service life with the three-dimensional damage distribution map to generate a dynamic damage evolution cloud map.

[0084] After completing the damage mode classification, preferably, it is necessary to further predict the damage evolution trend to assess the remaining service life of the connector. Step S81 inputs the size parameters, spatial orientation, and stress intensity factor of the microcrack into the Paris law crack propagation model, which quantifies the crack growth behavior under cyclic loading based on fracture mechanics principles. Step S82 uses the continuous damage mechanics model to calculate the evolution of the damage variables with the load cycles in the plastic deformation region, thereby describing the gradual degradation process of the material. Step S83 integrates the outputs of the crack propagation model and the damage mechanics model, and predicts the remaining service life of the connector under specified operating conditions through multiphysics coupling analysis. Step S84 couples the remaining service life with the three-dimensional damage distribution map for display, generating a dynamic damage evolution cloud map to achieve visual tracking of the damage state.

[0085] The interconnected steps described above resolve the technical problem of insufficient dynamic prediction capability in damage assessment. The Paris law model provides a quantitative description of crack propagation in step S81, while the continuous damage mechanics model covers the evolution of non-crack damage in step S82. The ensemble method in step S83 overcomes the limited applicability of a single model, making the prediction results more comprehensive. The visualization output in step S84 enhances the engineering applicability of the results.

[0086] In step S81, the calculation based on the stress intensity factor ensures the accuracy of the crack propagation driving force analysis, for example, by adaptively adjusting the prediction parameters for cracks with different orientations. Step S82 quantifies the cumulative effect of the plastic deformation region through the evolution of the damage variable, reflecting the performance degradation of the material under repeated loading. Step S83 fuses the outputs of different models using probabilistic or deterministic methods, for example, by handling parameter uncertainties through Monte Carlo simulation, thereby providing the confidence interval for the remaining life. The coupled display function in step S84 clearly presents the damage evolution process in the spatiotemporal dimension, assisting the user in formulating the predictive maintenance strategy.

[0087] This proposed solution, through multi-model collaboration and dynamic visualization, achieves a leap from static detection to dynamic prediction, solving the problem that traditional methods cannot assess the long-term evolution of damage.

[0088] Optionally, after step S33, a time-frequency joint analysis step is further included:

[0089] S91: Perform wavelet packet transform on the full matrix data to extract the multi-band acoustic energy attenuation coefficient distribution matrix;

[0090] S92: Perform pixel-level fusion of the multi-band acoustic energy attenuation coefficient distribution matrix with the three-dimensional acoustic image;

[0091] S93: Reconstructing the abrupt change in acoustic impedance of anisotropic materials based on the fused multidimensional data field;

[0092] S94: Correct the wave propagation path calculation in the acoustoelastic theory inversion by using the abrupt change interface of acoustic impedance.

[0093] Optionally, after step S44, a multi-scale stress field fusion step is further included:

[0094] S101: Extract the micro-stress concentration factor distribution of the self-locking spring region in the second three-dimensional stress field;

[0095] S102: A grain-scale representative volume element model of the self-locking spring material is established using the crystal plasticity finite element method;

[0096] S103: Input the micro-stress concentration factor as a boundary condition into the representative volume element model to calculate local stress redistribution;

[0097] S104: Map the stress redistribution results at the grain scale back to the second three-dimensional stress field to generate multi-scale fused stress field data.

[0098] Optionally, the process of running an explicit dynamic simulation in step S44 includes:

[0099] S111: Define the nonlinear stress-strain relationship of the elastoplastic constitutive model based on the thin-wall bending characteristics of the self-locking spring;

[0100] S112: Establish a dynamic contact algorithm based on a penalty function for the sliding contact characteristics between the adapter plug and the guide groove;

[0101] S113: Define the contact pair and load transfer path based on the interaction between the abutment and clamping section in the unwinding mechanism;

[0102] S114: The second three-dimensional stress field distribution is obtained by solving the dynamic equilibrium equations that include geometric nonlinearity and material nonlinearity.

[0103] Optionally, step S44 employs an auxiliary numerical method to ensure computational stability, specifically including:

[0104] S121: Optimize the mesh quality of the thin-walled region of the self-locking spring, and use an advanced algorithm to control the warpage and aspect ratio of the control unit within the allowable range;

[0105] S122: Introducing hourglass energy control modes to suppress the zero-energy deformation mode that may occur in the reduced integral unit;

[0106] S123: Define a contact algorithm based on a penalty function for the contact interface between the adapter plug and the guide groove, and balance the penetration amount and calculation efficiency by automatically adjusting the contact stiffness coefficient;

[0107] S124: An energy balance monitor is used to track the ratio of kinetic energy, internal energy, and dissipated energy in real time. When an abnormal increase in energy is detected, the calculation is automatically terminated and a prompt to adjust the parameters is given.

[0108] Optionally, step S44 employs a core algorithm to solve the dynamic equations, specifically including:

[0109] S131: Discretize the dynamic equilibrium equations containing inertial and damping terms using an explicit central difference scheme, and establish the nodal motion state update formula;

[0110] S132: Based on the strain rate sensitive characteristics of the self-locking spring, the stress tensor increment at the current time step is calculated using the Johnson-Cook constitutive model;

[0111] S133: The contact force vector is updated in real time according to the contact state between the adapter plug and the guide groove, and the tangential force component is corrected by introducing an anisotropic friction model.

[0112] S134: Monitor the numerical stability of the explicit dynamic simulation using the energy conservation criterion, and automatically trigger the energy consumption control algorithm when the hourglass energy ratio exceeds the threshold.

[0113] In the explicit dynamic simulation, ensuring the stability and accuracy of the numerical calculation is crucial for obtaining a reliable stress field. Step S131 discretizes the dynamic equilibrium equations containing the inertia and damping terms using the explicit central difference scheme, establishing the nodal motion state update formula. This scheme avoids the iterative convergence problem in the implicit algorithm and is suitable for short-time transient analysis. Step S132, based on the strain rate sensitivity of the self-locking spring, uses the Johnson-Cook constitutive model to calculate the stress tensor increment at the current time step, thereby accurately describing the mechanical response of the material under high strain rates. Step S133 updates the contact force vector in real time according to the contact state between the adapter plug and the guide groove, and introduces the anisotropic friction model to correct the tangential force component, ensuring a realistic simulation of the contact interaction. Step S134 monitors the numerical stability of the simulation using the energy conservation criterion. When the hourglass energy ratio exceeds the threshold, the energy consumption control algorithm is automatically triggered to prevent numerical divergence.

[0114] The combination of these steps addresses the technical problems of numerical instability and insufficient accuracy that easily occur in complex dynamic simulations. The discretization method provides an efficient time integration scheme in step S131, the constitutive model incorporates the material strain rate effect in step S132, the contact treatment captures the details of the interface behavior in step S133, and the stability monitoring ensures the reliability of the calculation process in step S134.

[0115] In step S131, the explicit central difference scheme significantly improves computational efficiency by directly recursively solving the equations of motion, making it particularly suitable for transient processes involving large deformations and contact, such as connector insertion and removal. The Johnson-Cook model in step S132 quantifies the dynamic yielding behavior of the material through the hardening coefficient and strain rate-sensitive parameters, accurately reflecting the stress-strain relationship of the phosphor bronze spring under rapid loading. The anisotropic friction model in step S133 considers the contact surface orientation dependence, more realistically simulating the sliding behavior during insertion and removal. The energy monitoring mechanism in step S134 tracks the balance of kinetic energy, internal energy, and hourglass energy in real time, automatically adjusting the algorithm parameters upon detecting anomalies to ensure the physical validity of the results.

[0116] Through the progressive steps and feedback adjustment described above, the simulation method can achieve high-fidelity stress field prediction while maintaining computational efficiency, thus solving the problem of balancing accuracy and stability in traditional simulations.

[0117] Optionally, the stress calculation model established based on finite element theory in step S44 specifically includes:

[0118] S141: The dynamic equilibrium equations of a discrete system are established by the principle of virtual work, where the internal force term is expressed by the convolution of the full stress tensor with the strain tensor.

[0119] S142: For the nonlinear material behavior of the self-locking spring, the plastic strain increment is calculated using the Von Mises yield criterion and the associated flow rule;

[0120] S143: Based on the contact geometry characteristics of the adapter plug and the guide groove, the contact force is defined as a function of the contact penetration amount, and a nonlinear damping term is introduced to suppress numerical oscillations.

[0121] S144: The dynamic equations are solved using an explicit time integration algorithm, where the acceleration term is discretized using a central difference scheme and the velocity term is updated using a half-step scheme.

[0122] Optionally, explicit dynamic simulation can be coupled with a three-dimensional finite element stress model for analysis, specifically including:

[0123] S151: Input the dynamic contact force time history obtained from the explicit dynamic simulation as the boundary condition into the parameterized three-dimensional finite element model;

[0124] S152: Based on the strain rate sensitive characteristics of the self-locking spring, the strain rate effect of the static stress field is corrected by the Johnson-Cook constitutive model;

[0125] S153: The local stress field of the contact area between the adapter plug and the guide groove is embedded into the global finite element model for reanalysis using sub-model technology;

[0126] S154: The dynamic stress amplitude is coupled and evaluated with the hydrostatic pressure component of the three-dimensional finite element stress field through the stress intensity factor calculation module.

[0127] Optionally, the real-time monitoring data of the digital twin is fused and analyzed with the three-dimensional finite element stress model, specifically including:

[0128] S161: The temperature-stress coupling data stream under actual working conditions is continuously collected through the built-in micro-state monitor of the digital twin;

[0129] S162: Input the temperature-stress coupled data stream as a boundary condition into the parameterized three-dimensional finite element model for online stress field correction;

[0130] S163: Based on the extended Kalman filter algorithm, the real-time acquired vibration spectrum data is dynamically matched with the finite element modal analysis results;

[0131] S164: Reconstruct the transient stress distribution cloud map of the self-locking spring during actual insertion and removal operations using data assimilation technology.

[0132] Optionally, the digital twin performs a health status assessment through the following steps:

[0133] S171: The contact resistance-temperature coupling data stream is collected in real time by the micro status monitor and input into the health index calculation model;

[0134] S172: Based on the inversion formula of acoustic elasticity theory, the multi-band acoustic energy attenuation coefficient is converted into real-time microscopic stress tensor components.

[0135] S173: An extended Kalman filter is used to assimilate and fuse the real-time micro-stress tensor with the finite element stress field prediction;

[0136] S174: Dynamically update the remaining service life probability distribution function of the connector using a health decay model.

[0137] Optionally, the digital twin enhances the reliability of state assessment through auxiliary algorithms, specifically including:

[0138] S181: Establish a dynamic adjustment mechanism for each weight coefficient in the health index calculation model, and automatically optimize parameter sensitivity according to the ambient temperature gradient;

[0139] S182: The sliding time window algorithm is used to extract the trend of the real-time micro-stress tensor to eliminate the influence of transient interference on the health assessment.

[0140] S183: The observation noise matrix of the extended Kalman filter is calibrated in real time using covariance matching technology to improve the convergence stability of the data assimilation process;

[0141] S184: Establish a confidence interval for the predicted remaining useful life based on historical degradation data, and trigger self-learning update of model parameters when the prediction error exceeds the threshold.

[0142] Optionally, step S44 employs a core algorithm to solve the stress field, specifically including:

[0143] S191: Discretize the dynamic equilibrium equations using an explicit central difference scheme, where the acceleration term is calculated using the following formula:

[0144] = -1 ( - );

[0145] This formula discretizes the dynamic equilibrium equations using an explicit central difference scheme and is used to calculate nodal accelerations. The meanings of the parameters are as follows:

[0146] : Indicates at time step The node acceleration vector, used to update the node's motion state (velocity and displacement), is a key variable in explicit time integration.

[0147] : Represents the mass matrix, usually a diagonal matrix or a lumped mass matrix, whose elements correspond to the mass of each node and are used to convert force into acceleration.

[0148] : Indicates at time step The external force vector includes applied loads, boundary condition forces, and other external forces.

[0149] : Indicates at time step The internal force vector is mainly generated by internal interactions such as stress and contact force, and is obtained by calculating the element stress integral.

[0150] S192: Solve for the stress tensor increment based on the Johnson-Cook constitutive model, where the flow stress is determined by the following formula:

[0151] ;

[0152] This formula, based on the Johnson-Cook constitutive model, solves for the flow stress in the stress tensor increment and is used to describe the plastic behavior of materials under dynamic loading. The meanings of the parameters are as follows:

[0153] : Represents flow stress (dynamic yield stress), which is the yield strength of a material under specific plastic strain and strain rate, and is used to determine whether a material has entered the plastic state.

[0154] : Represents the initial yield stress, which is the yield stress of the material under reference conditions (such as zero plastic strain and reference strain rate), and the unit is MPa.

[0155] : Represents the hardening coefficient, which quantifies the strain hardening effect of a material caused by plastic deformation, and is measured in MPa.

[0156] : Represents equivalent plastic strain, describes the degree of cumulative plastic deformation of a material, and is a dimensionless quantity.

[0157] : Represents the hardening index, which controls the nonlinearity of strain hardening behavior and is a dimensionless parameter.

[0158] : Represents the strain rate sensitivity coefficient, which describes the dependence of the material's yield stress on the strain rate. It is a dimensionless parameter.

[0159] : Represents the equivalent plastic strain rate, describing the rate of plastic deformation, with units of seconds (s). -1 .

[0160] : Represents the reference strain rate, typically taken as 1.0 s. -1 , used to normalize the strain rate term.

[0161] This formula uses the hardening term ( ) and strain rate term ( Together they determine the flow stress.

[0162] These parameters are typically calibrated through material testing; for example, typical values ​​for phosphor bronze self-locking springs in connectors might be: , , , , The precise use of the formula ensures the accuracy of the stress field calculations and avoids errors caused by simplified models.

[0163] S193: Calculate the normal contact force Fn=kn based on the contact penetration amount δ. δ, and an anisotropic friction model is used to correct the tangential force component;

[0164] S194: The ratio of hourglass energy to internal energy is tracked in real time by an energy conservation monitor, and the dissipation algorithm is automatically triggered when the threshold is exceeded.

[0165] In the stress field calculation process, the accuracy of the core algorithm directly determines the reliability of the simulation results. Step S191 discretizes the dynamic equilibrium equations using the explicit central difference scheme, where the acceleration term is calculated based on the relationship between the nodal forces and the mass matrix, providing the basis for updating the kinematic quantities. Step S192 solves for the stress tensor increment based on the Johnson-Cook constitutive model, where the flow stress is jointly determined by the hardening term and the strain rate term, accurately characterizing the dynamic plastic behavior of the material. Step S193 calculates the normal contact force based on the contact penetration amount and corrects the tangential force component using the anisotropic friction model, thereby refining the simulation of the interface interaction. Step S194 uses the energy conservation monitor to track the ratio of hourglass energy to internal energy in real time, automatically triggering the dissipation algorithm when it exceeds the threshold to suppress the accumulation of numerical errors.

[0166] This sequence of steps addresses the accuracy loss and numerical divergence issues that are prone to occur in the stress field calculation. The discretization step S191 ensures the numerical feasibility of solving the equations, the constitutive model step S192 introduces the material nonlinear effects, the contact processing step S193 covers the complexity of the boundary conditions, and the energy monitoring step S194 provides a convergence guarantee for the calculation process.

[0167] In step S191, the explicit calculation of acceleration avoids the burden of matrix inversion for large systems, making it suitable for solving large-scale problems. The flow stress formula in step S192, using experimentally calibrated parameters (e.g., the hardening index and the strain rate coefficient), ensures the accuracy of the model prediction and reflects the stress response of the self-locking spring in actual operation. The contact force algorithm in step S193, based on dynamic adjustment of the penetration amount and the stiffness coefficient, effectively handles the abrupt changes in the contact state. The monitoring mechanism in step S194, by setting a reasonable threshold (e.g., the hourglass energy percentage not exceeding 10%), maintains numerical stability while preserving computational efficiency.

[0168] The coordinated implementation of the above steps ensures that the stress field calculation takes into account both the physical properties of material nonlinearity and contact nonlinearity, while also avoiding the risk of computational failure through stability control measures. The tight coupling of each component of the algorithm and its adaptive adjustment capability enhance the robustness and practicality of the entire method.

[0169] Optionally, stress evolution analysis is performed on the self-locking spring structure and dynamic operation process of the connector, specifically including:

[0170] S201: Establish a C-shaped curved surface finite element model of the self-locking spring, and use reduced integral shell elements to simulate large deformation behavior, wherein the element stress is calculated using the following formula:

[0171] ;

[0172] This formula is used to calculate the element stress in the C-shaped surface finite element model of a self-locking spring, based on elastoplastic theory, where the stress is determined only by the elastic strain. The meanings of the various parameters are as follows:

[0173] : Represents the element stress tensor, which describes the stress state at a point inside the material, including normal stress and shear stress components, and the unit is MPa.

[0174] : Represents the elasticity matrix or elastic tensor, which contains the elastic properties of the material (such as Young's modulus and Poisson's ratio), and is used to convert elastic strain into stress, with units of MPa.

[0175] : Represents the total strain tensor, which describes the total degree of deformation of a material, including elastic strain and plastic strain, and is a dimensionless quantity.

[0176] : Represents the plastic strain tensor, which describes the irreversible plastic deformation that occurs in a material, and is a dimensionless quantity.

[0177] This formula obtains the elastic strain by subtracting the plastic strain from the total strain, and then multiplies it by the elastic matrix to calculate the stress, thus accurately simulating the stress response of the material under elastoplastic behavior.

[0178] S202: Define the velocity-dependent friction model of the contact interface between the adapter plug and the guide groove. The friction coefficient is calculated as follows, varying with sliding speed:

[0179] ;

[0180] This formula defines a velocity-dependent friction model for the contact interface between the adapter plug and the guide groove, simulating the variation of the friction coefficient with sliding speed. The meanings of each parameter are as follows:

[0181] : Represents the instantaneous coefficient of friction, which changes dynamically with the sliding speed. It is used to calculate the frictional force between the contact interfaces and is a dimensionless quantity.

[0182] : Represents the steady-state friction coefficient, the friction coefficient during high-speed sliding, and is a dimensionless quantity.

[0183] : Represents the static friction coefficient, the friction coefficient at zero speed or extremely low speed, and is a dimensionless quantity.

[0184] : Represents the attenuation coefficient, which controls the exponential attenuation rate of the friction coefficient from its static value to its steady-state value, and is measured in s / m.

[0185] : Indicates the sliding speed, the relative speed between the contact surfaces, in m / s.

[0186] This model uses an exponential function to describe the behavior of the friction coefficient as it transitions from a static value to a steady-state value, thus more realistically reflecting the friction characteristics during dynamic contact.

[0187] S203: Solving the dynamic equations using the explicit central difference method To capture the stress wave propagation effect during the insertion and removal process;

[0188] This formula is a dynamic equilibrium equation solved using the explicit central difference method, used to capture the stress wave propagation effect during the insertion and extraction process. The meanings of the various parameters are as follows:

[0189] : Represents the mass matrix, which is assembled from the mass of individual units and is used for the inertia term. The unit is kg.

[0190] : Represents the nodal acceleration vector, describing the acceleration state of the node, with units of m / s².

[0191] : Represents the damping matrix, which describes the energy dissipation characteristics of the system (such as viscous damping), and the unit is N·s / m.

[0192] : Represents the node velocity vector, describing the velocity state of the node, in m / s.

[0193] : Represents the internal force vector, obtained by the element stress integral, including elastic force and damping force, with the unit being N.

[0194] : Represents the external force vector, including applied loads, boundary condition forces, and other external forces, with units of N.

[0195] This equation is based on Newton's second law and solves for the motion response of the nodes by balancing inertial forces, damping forces, and internal and external forces.

[0196] S204: Predict the fatigue damage evolution of the self-locking spring under cyclic loading based on the cumulative plastic strain criterion. The cumulative plastic strain criterion is...

[0197]

[0198] This formula is a fatigue damage prediction model based on the cumulative plastic strain criterion, used to evaluate the damage evolution of self-locking springs under cyclic loading. The meanings of each parameter are as follows:

[0199] : Represents the equivalent plastic strain rate, describing the rate of plastic deformation, with units of seconds (s). -1 .

[0200] t: represents time, the integral variable, and the unit is seconds (s).

[0201] : Represents the critical cumulative plastic strain, the maximum plastic strain that a material can withstand before fatigue failure, and is a dimensionless quantity.

[0202] Integral term: Represents the cumulative plastic strain from time 0 to t, that is, the integral of plastic strain with time, and is a dimensionless quantity.

[0203] This criterion determines whether a material has experienced fatigue failure by monitoring whether the accumulated plastic strain reaches a critical value, and is applicable to predicting low-cycle fatigue damage.

[0204] The technical effects of this invention are as follows:

[0205] Modular meter box connectors are prone to internal damage, such as microcracks or plastic deformation, due to stress concentration during long-term insertion and removal operations. Traditional testing methods struggle to quantitatively assess these defects. This connector, through structural innovation and optimized material acoustic properties, achieves feasibility and reliability for non-destructive testing. In some embodiments, the connector includes a frame and a detachable adapter plug, wherein the adapter plug integrates a current bar, a wiring hole, and a clamping mechanism. The current bar is positioned near the meter interface to ensure the stability of the electrical connection; the wiring hole allows for the insertion of external wires, simplifying the installation process; the clamping mechanism, made of a conductive material, automatically clamps and conducts current after wire insertion. A key improvement is that at least a portion of the adapter plug's housing is made of a material with specific acoustic properties, whose acoustic impedance matches that of the internal structural material. This design optimizes the penetration and reflection of ultrasonic waves, reduces energy loss at the interface, and provides a clear signal basis for subsequent ultrasonic testing. This acoustic matching solves the problem of ultrasonic signal attenuation caused by material acoustic impedance mismatch in traditional connectors, thereby improving the sensitivity of damage identification.

[0206] Preferably, the selection of the acoustic material is based on the acoustic impedance value of the internal metal components (such as the current strip and the clamping mechanism). For example, the acoustic impedance of the copper alloy is approximately 40 MRayl, and the acoustic impedance of the shell material is made close to this value through composite modification, with the deviation controlled within a small range (e.g., ±1 MRayl). This matching reduces the multiple interface reflections of the ultrasonic waves from the coupling agent to the shell and then to the internal metal, allowing the sound waves to efficiently penetrate the internal structure and capture the stress distribution information of the key components. Simultaneously, the acoustic optimization of the shell not only serves the testing phase but also enhances the connector's condition monitoring capabilities during operation. For example, stress changes can be inverted in real time through periodic ultrasonic scanning, enabling predictive maintenance. The entire solution, through the deep integration of electromechanical design and non-destructive testing requirements, solves the technical problem of early detection of internal connector damage. Its innovation lies in using acoustic parameters as core design indicators, breaking through the limitations of traditional connectors that only focus on electrical performance.

[0207] The connector's frame guide groove design ensures precise positioning of the adapter plug, avoiding mechanical stress caused by installation deviations. The modular structure of the adapter plug facilitates replacement and maintenance, reducing lifecycle costs. The conductivity of the clamping mechanism ensures low contact resistance, while the acoustic housing provides "detectability," upgrading the connector from a passive component to a smart monitoring node. The technical effects are reflected in three aspects: first, the acoustic matching improves the signal-to-noise ratio of ultrasonic testing, enhancing the identifiability of minor damage; second, the simplified structure reduces testing complexity, allowing internal assessment without disassembly; and third, the synergistic design of materials and functions provides a new paradigm for power equipment health management. This overall optimization significantly improves the connector's reliability and service life.

[0208] The modular meter box connector is prone to internal stress concentration or micro-damage due to insertion and removal operations during long-term use. Traditional detection methods, such as visual inspection or electrical testing, are insufficient to quantitatively assess these invisible defects. This method, by integrating ultrasonic non-destructive testing and numerical simulation technology, achieves accurate diagnosis of internal damage. In step S11, a full-matrix capture scan is performed using an ultrasonic phased array probe to acquire the raw full waveform data inside the connector, ensuring the integrity of the data source and a high signal-to-noise ratio. In step S12, based on this data, the three-dimensional acoustic image is reconstructed using the full-focusing algorithm, and the first three-dimensional stress field is calculated by inversion according to the acoustoelastic theory, converting the sound wave propagation characteristics into mechanical parameters and establishing the experimental reference field. In step S13, the parametric finite element model is independently established, and the explicit dynamic simulation of the insertion and removal process is run, outputting the second three-dimensional stress field as the theoretical reference. In step S14, the two stress fields are subtracted at the voxel level to generate the three-dimensional stress difference field, highlighting the deviation between the actual and ideal state. Step S15 locates the damage region by analyzing the local peaks in the difference field. This process, through cross-validation of experimental measurements and simulation predictions, solves the technical problem that traditional methods cannot quantitatively identify stress damage. Its innovation lies in coupling acoustic inversion with dynamic simulation, realizing a closed-loop analysis from data acquisition to damage diagnosis.

[0209] The full-matrix capture scan in step S11 ensures the spatial integrity of the ultrasound data, laying the foundation for high-precision imaging. The full-focusing algorithm in step S12 improves image resolution through coherent superposition, while the acoustoelastic inversion utilizes the constitutive relationship between material sound velocity and stress to quantitatively convert the acoustic signal into a stress tensor. The explicit dynamic simulation in step S13 provides the theoretical stress distribution under non-damaged conditions by reproducing the physical testing process. The voxel-level contrast in step S14 eliminates systematic errors, highlighting minute stress anomalies. The peak analysis in step S15, combined with the material yield threshold, enables automatic damage identification. These progressively advancing steps constitute a complete technical chain from physical measurement to numerical analysis and then back to engineering diagnosis, improving the reliability and quantification level of the detection through multi-source data fusion.

[0210] Preferably, the scanning strategy in step S11 optimizes the probe layout and frequency selection, for example, using an array with a center frequency of 10MHz to balance the penetration depth and resolution. The inversion algorithm in step S12 introduces the anisotropic acoustoelastic coefficient, accurately characterizing the stress state at the multi-material interface. The finite element model in step S13 meticulously considers the connector's geometric topology, such as the spatial distribution of the guide groove and the spring, ensuring consistency between the simulation and the actual object. Before the subtraction operation in step S14, spatial registration is required, for example, aligning the two stress fields based on feature points to avoid miscomparison. Step S15 uses the region growing algorithm to extract continuous abnormal regions from the difference field and combines this with the stress gradient direction to determine the damage type. The entire method reduces human interference through the standardized process, making the detection results repeatable, and is particularly suitable for the quality assessment of batch connectors.

[0211] This method solves the long-standing problem of the difficulty in quantifying and assessing the internal damage of the meter box connector after fatigue testing. Traditional methods, such as industrial CT, can only identify geometric defects, while this method directly reflects the abnormal mechanical state through stress field differences, achieving an upgrade from "geometric detection" to "mechanical state diagnosis." It reduces the false negative rate by complementing ultrasound and simulation, and improves spatial positioning accuracy through voxel-level analysis. Attached Figure Description

[0212] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0213] Figure 1 This is a schematic flowchart of a method for detecting test damage to a connector according to an embodiment of the present invention.

[0214] Figure 2 This is a schematic flowchart of a method for detecting test damage to a connector according to another embodiment of the present invention.

[0215] Figure 3 This is a structural block diagram of a connector damage detection device according to an exemplary embodiment.

[0216] Figure 4 This is a schematic diagram of the multi-angle three-dimensional structure of a connector according to an embodiment of the present invention;

[0217] Figure 5 This is an exploded perspective view of the power supply box and its connector according to another embodiment of the present invention. Detailed Implementation

[0218] The following are several embodiments of this application to specifically implement the corresponding technical solutions of the present invention.

[0219] Example 1:

[0220] The following detailed explanation of the implementation process of the modular meter box connector internal damage detection method provided by this invention is illustrated through a specific application scenario. This embodiment uses a certain model of modular meter box connector as the testing object. After completing the standard insertion and removal life test, it is necessary to evaluate whether its key components, such as the self-locking spring, have suffered internal damage due to repeated operation.

[0221] like Figure 1The flowchart shown discloses the logical chain of this invention: "Experimental Field - Theoretical Field - Difference Field - Damage Diagnosis". Step S11: Ultrasonic Phased Array (FMC) Scan to acquire internal raw data. This is the starting point of the method and the source of experimental data. A full-matrix capture scan is performed on the physically tested connector using an ultrasonic phased array probe to acquire complete internal ultrasonic raw data containing amplitude and phase information. Step S12: Focusing Method Reconstruction to Calculate the First Three-Dimensional Stress Field (Experimental Stress Field). This step is the inversion calculation of the experimental stress field. Based on the full-matrix data acquired in Step S11, a high-precision algorithm such as the full-focusing method is used to reconstruct the three-dimensional acoustic image of the connector, and based on the acoustoelastic theory, information such as sound velocity changes is inverted and calculated into the true first three-dimensional stress field inside the connector. Step S13: Establish a parametric finite element model and run an explicit dynamic simulation to calculate the second three-dimensional stress field (theoretical stress field). This step is the simulation calculation of the theoretical stress field. It can be performed in parallel with or independently of Step S12. A parametric three-dimensional finite element model of the connector is established, and the insertion and removal processes are reproduced through explicit dynamic simulation to calculate the second three-dimensional stress field under an ideal, damage-free state. Step S14: Voxel-level subtraction operation to generate a three-dimensional stress difference field. This step is a comparison and fusion of experiment and theory. After accurately spatially registering the experimental stress field obtained in step S12 with the theoretical stress field obtained in step S13, a voxel-to-voxel subtraction operation is performed to generate a three-dimensional stress difference field. This difference field highlights the abnormal stress distribution caused by actual damage. Step S15: Analyze local stress difference peaks to diagnose and locate internal damage. This is the ultimate goal and output of the method. By analyzing the local stress difference peaks in the three-dimensional stress difference field obtained in step S14, a threshold is set to quantitatively diagnose whether there are microcracks, plastic deformations, or other damage inside the connector, and to accurately locate their three-dimensional spatial positions.

[0222] Specifically, firstly, the connector sample that has completed physical testing is pre-treated by cleaning its surface to ensure good acoustic coupling. A 64-chip ultrasonic phased array probe with a center frequency of 10MHz is used and brought into contact with the connector housing surface using a coupling agent. The ultrasonic system is controlled to perform the full-matrix capture scan, i.e., each chip is sequentially excited to emit ultrasonic pulses, while all chips simultaneously receive the full waveform echo signal. The raw data acquired in this process contains the amplitude and phase information and is stored as complete full-matrix data according to the chip pair numbering. This step ensures the acquisition of complete wavefield information inside the connector, laying the foundation for subsequent high-precision imaging. Next, the acquired full-matrix data is processed. The full-focusing algorithm is used to perform three-dimensional imaging of the connector interior. The full-focusing algorithm significantly improves the signal-to-noise ratio and resolution of the image by calculating the acoustic wave propagation delay of all the chip pairs corresponding to each pixel and superimposing them. Based on the obtained high-resolution three-dimensional acoustic image, the first three-dimensional stress field is calculated by inversion according to the acoustoelastic theory. Specifically, by precisely measuring the changes in the propagation velocities of the ultrasonic longitudinal and transverse waves within the material, and combining this with the acoustoelastic coefficient of the material, the stress tensor components at various internal points are calculated. This process quantitatively converts the acoustic information into mechanical parameters, establishing an experimental reference field for the stress within the connector.

[0223] To obtain the theoretical benchmark, a parametric finite element model of the connector needs to be established. Based on the CAD geometric model of the connector, appropriate geometric cleanup is performed, while precisely preserving key features such as the C-shaped gradient curvature of the self-locking spring. The model is meshed using C3D8R elements, with local refinement in the spring contact area, and the minimum element size controlled at 0.1 mm. The Johnson-Cook plastic model is selected as the material model, and the parameters are set based on the phosphor bronze material test data: A=200MPa, B=250MPa, n=0.35, C=0.015. Surface-to-surface contact is defined between the self-locking spring and the wire, and between the adapter plug and the guide groove, with the coefficient of friction set to 0.15.

[0224] After completing the modeling, the explicit dynamic simulation is run to reproduce the mechanical behavior during the test. A displacement load of 50 mm / s is applied to the adapter plug to simulate the actual insertion and removal operation. The dynamic equilibrium equations are solved using the explicit central difference algorithm, with the time step automatically adjusted according to the smallest element size of the model. The ratio of kinetic energy to internal energy is monitored during the calculation to ensure computational stability. The simulation output includes stress, strain, and contact force data throughout the dynamic process, from which a stable second three-dimensional stress field is extracted as the theoretical reference benchmark. The first three-dimensional stress field obtained from the experiment is compared with the second three-dimensional stress field obtained from the simulation at the voxel level. The two stress field data are first spatially registered to ensure consistency of the comparison benchmark. A three-dimensional stress difference field is generated using point-by-point subtraction, which highlights the deviation between the stress distribution inside the connector after the actual test and the prediction of the ideal model. To eliminate noise, the difference field is processed by Gaussian filtering with a filter kernel size of 3×3×3 pixels.

[0225] Damage diagnosis is performed based on the processed stress difference field. First, a damage criterion threshold is set, which is set to 20% of the theoretical stress value based on the yield strength of the self-locking spring material (approximately 400 MPa). Continuous regions where the stress difference exceeds the threshold are identified as potential damage zones. The geometric features (volume, shape index) and mechanical features (principal stress direction, stress gradient) of these regions are extracted and input into a pre-trained convolutional neural network classifier for damage pattern recognition. The classifier output shows obvious microcrack features in the spring root region, while the contact area shows plastic deformation features.

[0226] To verify the reliability of the diagnostic results, supplementary analysis was conducted. The spring clamping force-time curve obtained from the explicit dynamic simulation was extracted and compared with the measured data recorded by the force sensor in the physical experiment. The correlation coefficient was calculated to be 0.97, and the root mean square error was 3.2N, proving that the finite element model has high reliability. Simultaneously, local mesh refinement analysis was performed on the damaged area, and the stress field of the potential damaged area was embedded into the global model for reanalysis using sub-model technology, further confirming the accuracy of the damage assessment.

[0227] The final result is a comprehensive inspection report containing damage information. The report includes a 3D damage distribution map, with different colors indicating different types of damage and their severity. Combining the Paris law crack propagation model and the continuous damage mechanics model, the remaining service life of the connector under rated operating conditions is predicted. All results are integrated into a digital twin platform, enabling visualization of the inspection data and long-term trend analysis.

[0228] The method described in this embodiment effectively detects internal stress damage in connectors after physical testing, achieving a leap from qualitative testing to quantitative assessment. Through the deep integration of ultrasonic testing and numerical simulation, a complete technical solution is provided for connector reliability assessment and predictive maintenance. This method is particularly suitable for quality inspection and lifespan assessment of precision electrical connectors.

[0229]

Example 2

[0230] This embodiment details the specific implementation process of explicit dynamic simulation of modular meter box connectors. The process aims to reproduce the mechanical behavior of the connector during insertion, removal, clamping, and unplugging operations through high-fidelity numerical simulation, providing accurate theoretical stress field data for subsequent damage diagnosis.

[0231] In this embodiment, in step S41, the original CAD assembly model of the connector is imported, and a geometric repair tool is used to automatically identify and stitch together surface gaps with minute gaps or overlaps, ensuring that all contact interfaces form a closed solid. Next, a feature recognition algorithm is used to locate the self-locking spring component, and its C-shaped structure's centerline and cross-sectional profile are extracted, accurately preserving the root transition fillet radius R=0.3mm and the gradient curvature features at the end of the clamping section. Furthermore, non-critical structures such as bolt hole edge chamfers and label grooves are simplified and filled, reducing the number of model facets by approximately 40% to improve mesh generation efficiency.

[0232] Then, in step S41, the cleaned geometry is inspected for defects, and the G2 continuity of the C-shaped reed surface is verified using a curvature continuity analysis tool to ensure there are no abrupt changes or distortions. Preferably, the equivalent surface undulations with micron-level roughness texture are preserved in the contact area between the reed and the connecting claw of the current bar. Simultaneously, the simplified geometric model is converted into a neutral file in STEP format, with key dimensional tolerances (such as reed thickness tolerance ±0.05mm) noted to provide standardized input for subsequent finite element modeling.

[0233] In this embodiment, in step S41, all geometric operation steps (including stitching tolerance of 0.01mm, simplification threshold of 5°, etc.) are finally recorded through a parameterized script to achieve traceability of model modifications. The final geometric model is then compared with the original CAD to ensure that the maximum deviation is less than 0.1mm, thereby optimizing the use of computational resources while preserving mechanical characteristics.

[0234] In this embodiment, in step S42, an elastoplastic constitutive model is created for the phosphor bronze material of the self-locking spring, defining the elastic modulus E=110 GPa, Poisson's ratio ν=0.34, and density ρ=8.8 g / cm³ as basic parameters. Next, the Johnson-Cook model parameters are assigned: initial yield stress A=200 MPa, strain hardening coefficient B=250 MPa, hardening exponent n=0.35, strain rate sensitivity coefficient C=0.015, and thermal softening exponent m=1.2. Further, a reference strain rate is set. =1.0 s -1 The melting point temperature T_melt = 920°C is used to fully describe the dynamic response under high strain rates.

[0235] Then, in step S42, an isotropic elastic model (E=3.2 GPa, ν=0.38) is assigned to the engineering plastic of the adapter plug housing, and a bilinear kinematic hardening model (yield strength 180 MPa, tangent modulus 2 GPa) is defined for the copper alloy of the current bar. Preferably, a thin film material property with a thickness of 5 μm is assigned separately for the contact interface plating (such as a silver plating) based on the material library. Simultaneously, the Johnson-Cook parameters are coupled with the temperature field through a user subroutine to achieve thermo-mechanical co-simulation capability.

[0236] In this embodiment, in step S42, the unit consistency of the material parameters is finally verified, and a mesh sensitivity test is performed on all material models to ensure that the stress calculation result deviation does not exceed 5% when the element size change is ±20%. The material property set is then exported as an XML format configuration file for reuse in subsequent simulation cases.

[0237] In this embodiment, in step S43, a surface-to-surface contact pair is created between the inner surface of the clamping section of the self-locking spring and the outer surface of the wire, and the normal contact stiffness coefficient is defined as 1×10 using a penalty function algorithm. 6 The coefficient of friction is set to N / mm, and the tangential friction coefficient is set to 0.18. Next, a sliding contact pair is established between the guide ribs of the adapter housing and the guide groove of the frame, and the automatic contact detection function is enabled, setting the search radius to 1.5 times the preferred unit size. Furthermore, an anisotropic friction model is assigned to the metal-insulator contact interface, making the axial friction coefficient (0.15) lower than the radial coefficient (0.22).

[0238] Then, in step S43, a critical separation stress of 5 MPa is set for possible contact separation, and a viscous damping coefficient of 0.01 is enabled during the rebound phase to suppress oscillation. Preferably, the density of contact detection points in the stress concentration area at the root of the reed is increased to three times that of the conventional area. Simultaneously, a failure criterion is defined for all contact pairs, automatically disabling the contact unit when the equivalent plastic strain exceeds 0.4.

[0239] In this embodiment, in step S43, the stability of the contact state is finally verified through pre-loading analysis to ensure uniform contact pressure distribution under initial assembly conditions. The contact definition parameters (including stiffness, friction coefficient, and damping) are then packaged into a contact attribute library, supporting batch application in multi-condition analysis.

[0240] In this embodiment, in step S44, a uniform displacement load is applied to the insertion end of the adapter plug, and the velocity curve is set to a trapezoidal wave of 50 mm / s (including a 0.1s acceleration segment and a 0.1s deceleration segment) to simulate actual insertion and removal operations. Next, the six degrees of freedom of the frame guide groove are constrained, and a spring-damped boundary (stiffness 1×10⁻⁶) is applied to the meter interface side. 5 The damping coefficient was 200 N·s / m (N / m) to simulate the actual installation environment. Furthermore, the explicit analysis time was set to 0.5 s, and automatic time step control was used, with an initial step size of 1 × 10⁻⁶. -7 s.

[0241] Then, in step S44, the explicit central difference algorithm is used to solve the dynamic equations, and the nodal accelerations are updated at each time step as follows:

[0242] a n =M -1 (F ext -F int );

[0243] The node acceleration update formula is based on Newton's second law and is used to calculate the node acceleration at each time step.

[0244] a n : Represents the nodal acceleration vector at time step n, describing the instantaneous acceleration of the mesh nodes. It is a fundamental physical quantity for updating velocity and displacement, and its unit is m / s². 2 .

[0245] M: Represents the lumped mass matrix, which is a diagonal matrix whose diagonal elements are the mass allocated to each node, in kg.

[0246] F ext : Represents the external force vector, including all external loads applied to the model, such as the displacement load and the reaction force of the spring-damped boundary in this example, and the unit is N (Newton).

[0247] F int : Represents the internal force vector, which is obtained by integrating (combining) the element stress field at the nodes. It reflects the stress state inside the material and is measured in N (Newtons).

[0248] The velocity and displacement are updated using the following formula:

[0249]

[0250] The above speed update formula uses an explicit format of "half-step" to update the node speed.

[0251] : Indicates at time step The nodal displacement vector, in units of .

[0252] : Indicates at time step The nodal displacement vector, in units of .

[0253] : Indicates at time step The nodal velocity vector (calculated from the previous formula), in units of .

[0254] : indicates half step size arrive The time increment. Under the case of constant step size... The unit is (Second).

[0255] The displacement update formula described above uses the updated half-step velocity to calculate the new nodal displacement.

[0256] : Indicates at time step The nodal displacement vector, in units of .

[0257] : Indicates at time step The nodal displacement vector, in units of .

[0258] : Indicates at time step The nodal velocity vector (calculated from the previous formula), in units of .

[0259] : indicates half step size arrive The time increment, under the condition of constant step size. The unit is (Second).

[0260] This series of formulas constitutes the core iterative process of the explicit central difference method, and its solution flow is: acceleration → velocity → displacement. This method avoids solving large linear equation systems and is very suitable for analyzing short-lived transient dynamic processes involving large deformations and complex contacts, such as connector insertion and removal. All parameters need to be updated at each time step of the simulation to dynamically capture the propagation effect of stress waves.

[0261] Preferably, the ratio of kinetic energy to internal energy is monitored, and an energy dissipation algorithm is automatically triggered when it exceeds 10%. Simultaneously, full-field stress and strain data are output every 0.001 seconds.

[0262] In this embodiment, in step S44, the stress field time series during the stable insertion / removal phase (0.2s-0.4s) is finally extracted, and the time-averaged stress field is calculated as the second three-dimensional stress field. Furthermore, MPI parallel computing is used to increase the solution speed to 3.5 times that of single-machine mode, ensuring that the transient analysis of a model with tens of millions of elements is completed within 8 hours.

[0263]

Example 3

[0264] This embodiment details the implementation process of an artificial intelligence-based damage pattern recognition method. This method achieves automatic identification and classification of damage types through intelligent analysis of a three-dimensional stress difference field, significantly improving the accuracy and efficiency of damage diagnosis.

[0265] In this embodiment, in step S71, based on the three-dimensional stress difference field, a region growing algorithm is used to identify all connected voxel sets whose stress values ​​exceed the dynamic damage determination threshold, serving as the initial spatial domain of the potential damage region. Next, morphological closing operations are applied to each connected voxel set to fill internal pores and smooth boundaries, obtaining a geometrically continuous binary three-dimensional mask of the potential damage region. Further, the volume V and surface area S of the region enclosed by each binary three-dimensional mask are calculated, and the sphericity index Ψ = (π^(1 / 3)(6V)^(2 / 3)) / S is derived based on the ratio of V to S to quantify the geometric compactness of the potential damage region.

[0266] Then, in step S71, for each potential damage region, the dimensions (length L, width W, height H) of its minimum circumscribed cuboid are extracted, and the aspect ratio feature AR = max(L, W, H) / min(L, W, H) is calculated. Simultaneously, principal component analysis is performed based on the voxel coordinate covariance matrix within the region to obtain the first principal direction vector as an estimate of the principal stress direction distribution. Preferably, the stress tensor eigenvalues ​​of each voxel point within the potential damage region are calculated, and the average orientation angle (θ, φ) of the unit vector of the maximum principal stress direction is statistically analyzed throughout the entire region.

[0267] In this embodiment, in step S71, the volume V, surface area S, sphericity Ψ, aspect ratio AR, and average orientation angles (θ, φ) are combined into a six-dimensional geometric-mechanical feature vector. Simultaneously, the root mean square value of the stress gradient within the region is additionally calculated as a mechanical fluctuation descriptor and incorporated into the feature vector. Finally, the feature vector is Z-score standardized so that the mean of each feature component is 0 and the standard deviation is 1, to eliminate the influence of dimensions and prepare for subsequent classifier input.

[0268] In step S72, the standardized six-dimensional geometric-mechanical feature vector is reshaped into a 2x3 two-dimensional matrix and expanded to an 8x8 input size with zero padding to accommodate the input requirements of the convolutional neural network. Next, the expanded matrix is ​​input to a pre-trained ResNet-18 architecture convolutional neural network damage classifier, which contains four residual blocks, each with two 3x3 convolutional layers and shortcut connections.

[0269] Further, in step S72, the first convolutional layer of the convolutional neural network damage classifier uses 32 filters with a stride of 1 and employs the ReLU activation function; subsequent residual blocks sequentially multiply the number of filters to 64, 128, and 256. Then, a fully connected layer with 128 neurons is followed by a global average pooling layer, and finally, a softmax output layer generates the probability distributions of three damage modes (microcracks, plastic deformation, and fatigue damage). Preferably, the network weights are initialized using transfer learning with weights pre-trained on ImageNet and fine-tuned using 5000 samples from the connector historical damage dataset.

[0270] In this embodiment, the Adam optimizer is used during training, with an initial learning rate of 0.001 and a cross-entropy loss function applied. Every 10 training epochs, if the validation set loss does not decrease, the learning rate is halved. Simultaneously, to prevent overfitting, L2 regularization (weight decay coefficient λ = 0.01) and Dropout (dropout rate 0.5) are applied during training. Finally, the feature vector to be identified is forward-propagated through the network to obtain its probability value belonging to each damage category.

[0271] In step S73, a probability threshold P is set. th =0.85, if the probability value of a certain damage category output by the network exceeds P th If the probability of any of the categories is lower than P, then the potential damage area is directly determined to belong to that damage type. Next, if the probability of all categories is lower than P... thThen, a multi-model ensemble strategy is initiated: the same feature vector is input in parallel to a support vector machine classifier and a random forest classifier (both of which have been trained on the same historical dataset).

[0272] Further, in step S73, the support vector machine classifier uses a radial basis function kernel with a penalty parameter C=1.0; the random forest classifier contains 100 decision trees with a maximum depth of 10. Then, the classification results of the convolutional neural network, support vector machine, and random forest are collected, and a weighted voting method is used for the final decision (weights are 0.5, 0.25, and 0.25, respectively). Simultaneously, the case classified as low-confidence is recorded, and its feature vector and the judgment result are added to the historical dataset for subsequent model retraining.

[0273] In this embodiment, in step S73, for regions determined to be microcracks, an additional probability confidence score CS = (P max - P second ) / P max , where P max For the highest probability, P second The probability is the second highest. If CS < 0.3, the area is marked as a disputed case requiring manual review. Finally, all judgment results (damage type, confidence level, disputed label) are associated and stored with the corresponding potential damage area identifier.

[0274] In step S74 of this embodiment, the center point coordinates (x, y, z) of each potential damage region are associated with the damage type label and confidence level output in step S73 to construct a triplet dataset containing spatial location, damage category, and confidence level value. Then, the Marching Cubes algorithm is used to interpolate the triplet data in the three-dimensional space of the connector to generate a continuous isosurface mesh model.

[0275] Further, in step S74, color codes are assigned according to the damage type: red for microcracks (RGB: 255,0,0), yellow for plastic deformation (RGB: 255,255,0), and blue for fatigue damage (RGB: 0,0,255). Confidence is represented by color transparency (Alpha channel); the higher the confidence level, the less transparent the color. Then, the colored isosurface mesh is overlaid with the CAD 3D model of the connector, and rotation, zoom, and cross-sectional viewing functions are provided in the visualization interface.

[0276] Finally, in this embodiment, a volumetric bar chart and confidence level labels for each damaged region are additionally annotated on the three-dimensional damage distribution map. Simultaneously, a lightweight, web-compatible format (such as GLTF) file of the map is generated, embedding timestamps and detection batch information. Finally, the map file is automatically uploaded to a cloud platform digital twin system and associated with the real-time operational data of the connector, enabling historical traceability of damage status and visualization of predictive maintenance.

[0277]

Example 4

[0278] This embodiment details the implementation process of the damage evolution prediction and life assessment method. Based on damage pattern recognition, this method further predicts the development trend of damage and the remaining service life of the equipment, providing a scientific basis for predictive maintenance.

[0279] In step S81, the current size parameters of each identified microcrack, including crack length, depth, and opening displacement, are extracted from the three-dimensional damage distribution map, and their spatial orientation angle is determined based on the principal axis analysis of the crack voxel cluster. Next, using the local stress field data of the connector under the rated insertion and extraction load spectrum, the range of the Type I stress intensity factor at the tip of each microcrack is calculated, considering the influence of geometric correction factors and far-field stress amplitude. Furthermore, for non-planar cracks, the range of equivalent stress intensity factors in the hybrid mode is calculated using the superposition principle as the driving force for crack propagation.

[0280] Then, the dimensional parameters, spatial orientation angle, and equivalent stress intensity factor range are used as input variables and substituted into the Paris law crack propagation model, where the material constants are calibrated through fatigue tests of the phosphor bronze self-locking spring. Preferably, considering the load sequence effect, the Wheeler delay model is used to correct the crack propagation rate, introducing a shape factor and a delay parameter. Simultaneously, for cracks whose orientation angle changes exceed a threshold, a redirection algorithm is automatically initiated to update the crack propagation path.

[0281] In this embodiment, in step S81, based on the Paris law model, the crack size evolution trajectory is iteratively calculated at discrete time steps using the current crack size as the initial condition. Finally, the propagation path of each microcrack is associated with the remaining number of cycles and stored as a time series dataset for subsequent lifetime ensemble analysis.

[0282] In step S82, plastic deformation regions are identified from the three-dimensional damage distribution map, and their average equivalent plastic strain, strain amplitude, and current damage state variables are extracted. Next, based on the local stress-strain history output by the explicit dynamic simulation, the increment of plastic strain energy density under each load cycle is calculated. Further, the Lemaitre continuous damage mechanics model is used to define the damage evolution law as related to the increment of plastic strain energy density and the material fracture energy. Then, for the phosphor bronze material, the model parameters, including fracture energy density and damage index, are calibrated through uniaxial fatigue tests, and a correction term is introduced considering the influence of average stress. Preferably, for multiaxial stress states, the increment of equivalent plastic strain energy density is calculated using the Von Mises equivalent stress conversion. Simultaneously, the damage variables are updated at each load step using the implicit Euler method numerical integration damage evolution equation.

[0283] In this embodiment, when the damage variable approaches a critical value, the nonlinear acceleration module is automatically activated, and the Coffin-Manson relationship is used to correct the damage accumulation rate. Finally, the damage evolution curve of each plastic deformation region is bound to spatial coordinates to generate damage time field data.

[0284] Next, the microcrack propagation trajectory output in step S81 is spatiotemporally aligned with the plastic damage evolution curve output in step S82, and a joint failure criterion is established based on the number of common load cycles. Then, the failure probability distribution of each damage mode is calculated using the first-pass probability theory, where the failure rate function is derived from the damage growth rate.

[0285] Furthermore, in step S83, considering the interaction between microcracks and plastic damage, a coupling factor is introduced to correct the joint failure probability. Then, multiple random load sequence samples are generated through Monte Carlo simulation, the distribution function of the remaining service life is statistically analyzed, and the expected value and confidence interval are output.

[0286] In this embodiment, in step S83, the prediction result is fused with real-time monitored insertion / removal counts and ambient temperature data, and a Bayesian update algorithm is used to dynamically adjust the remaining service life distribution parameters. Finally, a comprehensive service life report is generated, including expected service life, confidence interval, and failure probability curve.

[0287] Subsequently, in step S84, the expected remaining service life value output in step S83 is mapped to each voxel point of the three-dimensional damage distribution map, and a color gradient is assigned according to the remaining service life value. Then, based on the crack propagation path and damage field data, a continuous time series three-dimensional stress-damage coupled field is generated using a linear interpolation algorithm.

[0288] Then, in step S84, volume rendering technology is used to render the damage state at each time step as a semi-transparent cloud map, which is then overlaid on the connector CAD model. Preferably, an interactive visualization interface is developed using an integrated WebGL engine, supporting animation playback control via a time slider and detailed parameter queries for damaged areas. Simultaneously, an alarm module is embedded, automatically highlighting and flashing and pushing warning information when the remaining service life falls below a preset threshold.

[0289] Finally, the dynamic cloud map data is compressed into a time-series file, including metadata such as timestamps, load conditions, and model version. Finally, the cloud map is synchronized in real-time to the meter box digital twin platform via the OPC UA protocol, enabling predictive maintenance decision support.

[0290]

Example 5

[0291] This embodiment focuses on describing the specific implementation process of the explicit dynamics core algorithm. This algorithm is the technical foundation for connector dynamic stress analysis and directly affects the accuracy and reliability of simulation results.

[0292] In this embodiment, in step S131, the input parameters required for calculating the nodal acceleration are initialized based on the nodal mass matrix and initial load conditions of the parameterized three-dimensional finite element model of the connector. Through step S131, the lumped mass values ​​of the mass matrix and the nodal force vector at the current time step are read to prepare data for explicit time integration. Then, the initial time step is automatically determined according to the Courant-Friedrichs-Lewy stability condition to ensure numerical stability. In this embodiment, in step S131, the nodal acceleration vector at the current time step is calculated by multiplying the nodal force vector by the inverse of the mass matrix. Further, a half-step velocity update format is adopted, updating the current half-step velocity based on the half-step velocity of the previous time step and the current acceleration. Simultaneously, the nodal displacement vector is updated using the current half-step velocity and the current time step.

[0293] Optionally, in step S131, the change in the time step is monitored in real time, and the step size is dynamically adjusted according to the minimum element size of the model and the material wave velocity. Step S131 also stores the updated nodal motion state data, including displacement, velocity, and acceleration vectors, for calculation in the next time step. Furthermore, in step S131, the energy balance relationship is verified to ensure that the sum of kinetic energy and internal energy is within the allowable error range. Finally, the discretized solution of the motion equations is output to provide input for stress calculation.

[0294] In this embodiment, in step S132, the equivalent plastic strain and plastic strain rate data of the current integration point are read from the database of the explicit dynamic simulation. Through step S132, the pre-calibrated Johnson-Cook model material parameters, including the initial yield stress A, hardening coefficient B, and strain rate sensitivity coefficient C, are invoked. Then, the strain hardening term is calculated based on the equivalent plastic strain, and the strain rate strengthening term is calculated in conjunction with the plastic strain rate. Further, in step S132, the thermal softening effect is integrated, and the dimensionless temperature term is calculated by querying the current temperature field data. Optionally, the dynamic yield stress value is obtained by multiplying the strain hardening term, strain rate strengthening term, and thermal softening term.

[0295] Next, in step S132, the elastic prediction-plastic correction algorithm is used to calculate the trial stress tensor and check whether it exceeds the dynamic yield stress. Through step S132, if it exceeds the yield stress, the stress is corrected to the yield surface via radial return mapping; otherwise, the stress is directly updated. Simultaneously, the updated stress tensor increment and plastic strain increment are stored for the next time step. Finally, the stress tensor increment is output to the global stress field array.

[0296] Further, in step S133, based on the contact geometry data of the connector's adapter plug and guide groove, a global contact search algorithm is executed to identify potential contact pairs. In step S133, the normal penetration amount of each contact pair is calculated, and the magnitude of the normal contact force is determined based on the penalty function method. Then, according to the anisotropic friction coefficient model of the contact material, the friction coefficient value varying with the direction angle is queried. Optionally, in step S133, the tangential friction force vector is calculated using the normal contact force and the friction coefficient, taking into account the influence of sliding speed.

[0297] In this embodiment, in step S133, the normal contact force and tangential friction force vectors are integrated to generate a total contact force vector and update the nodal force array. Step S133 employs a smooth transition algorithm to handle abrupt changes in the contact state, avoiding numerical oscillations. Simultaneously, changes in contact energy are monitored to ensure energy conservation in contact interactions. Finally, the updated contact force vector is output for solving the dynamic equations.

[0298] Optionally, in step S134, the kinetic energy, internal energy, contact energy, and hourglass energy component data from the explicit dynamic simulation are collected in real time. In step S134, the total energy balance error is calculated and compared with a preset tolerance threshold. Then, if the hourglass energy percentage exceeds the threshold of 10%, the viscous damped hourglass control algorithm is automatically activated. Optionally, in step S134, the artificial damping coefficient is adjusted to suppress the development of the zero-energy mode.

[0299] Finally, in step S134, the historical trend of energy error is monitored, and if it continuously exceeds the limit, an adaptive time step reduction mechanism is triggered. In step S134, an energy monitoring log is recorded, including time history data for each component. Simultaneously, when the total energy deviation exceeds the safe range, the calculation is automatically terminated and a warning message is output. Finally, a numerical stability report is generated for subsequent analysis.

[0300]

Example 6

[0301] This embodiment details the specific implementation of the core formulas for stress field calculation. These formulas form the theoretical basis for connector stress analysis and directly determine the accuracy and reliability of the simulation results.

[0302] In step S191, based on the nodal mass matrix and initial load conditions of the parameterized three-dimensional finite element model of the connector, the input parameters required for nodal acceleration calculation are initialized, including reading the lumped mass values ​​of the mass matrix and the nodal force vector at the current time step, to prepare data for explicit time integration. Then, the initial time step is automatically determined according to the Courant-Friedrichs-Lewy stability condition to ensure numerical stability.

[0303] In this embodiment, the node acceleration vector at the current time step is calculated by multiplying the nodal force vector by the inverse of the mass matrix. The mass matrix employs a lumped mass scheme to simplify the inversion operation. Furthermore, a half-step velocity update scheme is used, updating the current half-step velocity based on the half-step velocity of the previous time step and the current acceleration to ensure time integration accuracy. Simultaneously, the node displacement vector is updated using the current half-step velocity and the current time step to capture the dynamic deformation of the connector.

[0304] Optionally, in step S191, the change in the time step is monitored in real time, and the step size is dynamically adjusted according to the minimum element size of the model and the material wave velocity to adapt to the large deformation effect during the connector insertion and removal process. Through step S191, the updated node motion state data, including displacement, velocity, and acceleration vectors, are stored for calculation in the next time step, ensuring data consistency. Furthermore, in step S191, the energy balance relationship is verified, and the sum of kinetic energy and internal energy is calculated to ensure it is within the allowable error range and to prevent numerical divergence. Finally, the discretized solution of the motion equations is output to provide input for subsequent stress calculations.

[0305] Next, in step S192, the equivalent plastic strain and plastic strain rate data at the current integration point are read from the database of the explicit dynamic simulation to provide input for the Johnson-Cook constitutive model. Through step S192, pre-calibrated Johnson-Cook model material parameters, including the initial yield stress A, hardening coefficient B, and strain rate sensitivity coefficient C, are invoked. These parameters are determined based on experimental data of the phosphor bronze material of the self-locking spring.

[0306] Then, a strain hardening term is calculated based on the equivalent plastic strain, and a strain rate strengthening term is calculated in conjunction with the plastic strain rate, wherein the strain hardening term is described in power-law form. Further, in step S192, the thermal softening effect is integrated, and a dimensionless temperature term is calculated by querying the current temperature field data to reflect the softening behavior of the material at high temperatures.

[0307] Optionally, the dynamic yield stress value is obtained by multiplying the strain hardening term, strain rate strengthening term, and thermal softening term, and is used to determine whether the material has yielded. In step S192, an elastic prediction-plastic correction algorithm is used to calculate the trial stress tensor and check whether it exceeds the dynamic yield stress. If it does, the stress is corrected to the yield surface through radial return mapping; otherwise, the stress is directly updated.

[0308] Simultaneously, the updated stress tensor increment and plastic strain increment are stored for the next time step to ensure the continuity of the stress history. Finally, the stress tensor increment is output to the global stress field array to provide data for subsequent stress analysis.

[0309] Further, in step S193, based on the contact geometry data of the adapter plug and guide groove of the connector, a global contact search algorithm is executed to identify potential contact pairs, taking into account the contact interface between the self-locking spring and the wire. Through step S193, the normal penetration of each contact pair is calculated, and the magnitude of the normal contact force is determined based on the penalty function method, wherein the contact stiffness is automatically optimized according to the material properties.

[0310] Then, based on the anisotropic friction coefficient model of the contact material, the friction coefficient value varying with the direction angle is queried to simulate the actual sliding contact behavior of the adapter plug. Optionally, in step S193, the tangential friction force vector is calculated using the normal contact force and the friction coefficient, and the influence of sliding speed is considered to introduce a speed-dependent friction model.

[0311] In this embodiment, the normal contact force and tangential friction force vectors are integrated to generate a total contact force vector, which is then updated to the nodal force array for solving the dynamic equations. In step S193, a smooth transition algorithm is used to handle abrupt changes in the contact state, avoiding numerical oscillations and ensuring the stability of the connector insertion and removal process. Simultaneously, changes in contact energy are monitored to ensure energy conservation in contact interactions and prevent energy anomalies. Finally, the updated contact force vector is output, providing input for the next calculation.

[0312] Alternatively, in step S194, the kinetic energy, internal energy, contact energy, and hourglass energy components of the explicit dynamic simulation are collected in real time to establish a complete energy balance equation. Step S194 is then used to calculate the total energy balance error and compare it with a preset tolerance threshold to detect numerical stability issues.

[0313] Then, in step S194, if the hourglass energy percentage is detected to exceed a threshold of 10%, the viscous damping hourglass control algorithm is automatically activated to suppress the development of the zero-energy mode. Preferably, the artificial damping coefficient is then adjusted, and the damping parameters are dynamically optimized based on the historical trend of the hourglass energy.

[0314] In this embodiment, the historical trend of energy error is monitored. If the error continuously exceeds the limit, an adaptive time step reduction mechanism is triggered to improve calculation accuracy. In step S194, an energy monitoring log is recorded, including time history data for each component, for subsequent analysis. Simultaneously, when the total energy deviation exceeds the safe range, the calculation is automatically terminated and a warning message is output to prevent invalid results. Finally, a numerical stability report is generated to provide a basis for simulation verification.

[0315]

Example 7

[0316] This embodiment details the implementation process of the explicit dynamics core algorithm in the stress analysis of modular meter box connectors. This algorithm provides crucial technical support for the numerical simulation of the connector's dynamic operation, ensuring the accuracy and reliability of the stress field calculations.

[0317] In this embodiment, step S131 initializes the input parameters required for nodal acceleration calculation based on the nodal mass matrix and initial load conditions of the parameterized three-dimensional finite element model of the connector. Step S131 also reads the lumped mass values ​​of the mass matrix and the nodal force vector at the current time step to prepare data for explicit time integration. Then, the initial time step is automatically determined according to the Courant-Friedrichs-Lewy stability condition to ensure numerical stability. In this embodiment, in step S131, the nodal acceleration vector at the current time step is calculated by multiplying the nodal force vector by the inverse of the mass matrix. Further, a half-step velocity update format is used to update the current half-step velocity based on the half-step velocity of the previous time step and the current acceleration. Simultaneously, the nodal displacement vector is updated using the current half-step velocity and the current time step.

[0318] Optionally, in step S131, the change in the time step is monitored in real time, and the step size is dynamically adjusted according to the minimum element size of the model and the material wave velocity. Step S131 also stores the updated nodal motion state data, including displacement, velocity, and acceleration vectors, for calculation in the next time step. Furthermore, in step S131, the energy balance relationship is verified to ensure that the sum of kinetic energy and internal energy is within the allowable error range. Finally, the discretized solution of the motion equations is output to provide input for stress calculation.

[0319] In this embodiment, in step S132, the equivalent plastic strain and plastic strain rate data of the current integration point are read from the database of the explicit dynamic simulation. Through step S132, the pre-calibrated Johnson-Cook model material parameters, including the initial yield stress A, hardening coefficient B, and strain rate sensitivity coefficient C, are invoked. Then, the strain hardening term is calculated based on the equivalent plastic strain, and the strain rate strengthening term is calculated in conjunction with the plastic strain rate. Further, in step S132, the thermal softening effect is integrated, and the dimensionless temperature term is calculated by querying the current temperature field data. Optionally, the dynamic yield stress value is obtained by multiplying the strain hardening term, strain rate strengthening term, and thermal softening term.

[0320] Next, in step S132, the elastic prediction-plastic correction algorithm is used to calculate the trial stress tensor and check whether it exceeds the dynamic yield stress. Through step S132, if it exceeds the yield stress, the stress is corrected to the yield surface via radial return mapping; otherwise, the stress is directly updated. Simultaneously, the updated stress tensor increment and plastic strain increment are stored for the next time step. Finally, the stress tensor increment is output to the global stress field array.

[0321] Further, in step S133, based on the contact geometry data of the connector's adapter plug and guide groove, a global contact search algorithm is executed to identify potential contact pairs. In step S133, the normal penetration amount of each contact pair is calculated, and the magnitude of the normal contact force is determined based on the penalty function method. Then, according to the anisotropic friction coefficient model of the contact material, the friction coefficient value varying with the direction angle is queried. Optionally, in step S133, the tangential friction force vector is calculated using the normal contact force and the friction coefficient, taking into account the influence of sliding speed.

[0322] In this embodiment, in step S133, the normal contact force and tangential friction force vectors are integrated to generate a total contact force vector and update the nodal force array. Step S133 employs a smooth transition algorithm to handle abrupt changes in the contact state, avoiding numerical oscillations. Simultaneously, changes in contact energy are monitored to ensure energy conservation in contact interactions. Finally, the updated contact force vector is output for solving the dynamic equations.

[0323] Optionally, in step S134, the kinetic energy, internal energy, contact energy, and hourglass energy component data from the explicit dynamic simulation are collected in real time. In step S134, the total energy balance error is calculated and compared with a preset tolerance threshold. Then, if the hourglass energy percentage exceeds the threshold of 10%, the viscous damped hourglass control algorithm is automatically activated. Optionally, in step S134, the artificial damping coefficient is adjusted to suppress the development of the zero-energy mode.

[0324] Next, in step S134, the historical trend of energy error is monitored. If the error continuously exceeds the limit, an adaptive time step reduction mechanism is triggered. In step S134, an energy monitoring log is recorded, including time history data for each component. Simultaneously, when the total energy deviation exceeds the safe range, the calculation is automatically terminated and a warning message is output. Finally, a numerical stability report is generated for subsequent analysis.

[0325] [Example 8]: Implementation Method of Core Formula for Stress Field Calculation

[0326] This embodiment focuses on the specific implementation of the core formulas for stress field calculation. These formulas are the theoretical basis for evaluating the mechanical performance of connectors and directly determine the accuracy and reliability of the simulation results.

[0327] In this embodiment, in step S191, based on the nodal mass matrix and initial load conditions of the parameterized three-dimensional finite element model of the connector, the input parameters required for nodal acceleration calculation are initialized, including reading the lumped mass values ​​of the mass matrix and the nodal force vector at the current time step, to prepare data for explicit time integration. Then, the initial time step is automatically determined according to the Courant-Friedrichs-Lewy stability condition to ensure numerical stability.

[0328] In this embodiment, in step S191, the node acceleration vector for the current time step is calculated by multiplying the nodal force vector by the inverse of the mass matrix, wherein the mass matrix adopts a lumped mass format to simplify the inversion operation. Further, a half-step velocity update format is used to update the current half-step velocity based on the half-step velocity of the previous time step and the current acceleration, ensuring time integration accuracy. Simultaneously, the node displacement vector is updated using the current half-step velocity and the current time step to capture the dynamic deformation of the connector.

[0329] Optionally, in step S191, the change in the time step is monitored in real time, and the step size is dynamically adjusted according to the minimum element size of the model and the material wave velocity to adapt to the large deformation effect during the connector insertion and removal process. Through step S191, the updated node motion state data, including displacement, velocity, and acceleration vectors, are stored for calculation in the next time step, ensuring data consistency. Furthermore, in step S191, the energy balance relationship is verified, and the sum of kinetic energy and internal energy is calculated to ensure it is within the allowable error range and to prevent numerical divergence. Finally, the discretized solution of the motion equations is output to provide input for subsequent stress calculations.

[0330] Next, in step S192, the equivalent plastic strain and plastic strain rate data at the current integration point are read from the database of the explicit dynamic simulation to provide input for the Johnson-Cook constitutive model. Through step S192, pre-calibrated Johnson-Cook model material parameters, including the initial yield stress A, hardening coefficient B, and strain rate sensitivity coefficient C, are invoked. These parameters are determined based on experimental data of the phosphor bronze material of the self-locking spring.

[0331] Then, in step S192, a strain hardening term is calculated based on the equivalent plastic strain, and a strain rate strengthening term is calculated in conjunction with the plastic strain rate, wherein the strain hardening term is described in power-law form. Further, in step S192, the thermal softening effect is integrated, and a dimensionless temperature term is calculated by querying the current temperature field data to reflect the softening behavior of the material at high temperatures.

[0332] Optionally, in step S192, the dynamic yield stress value is obtained by multiplying the strain hardening term, strain rate strengthening term, and thermal softening term, and is used to determine whether the material has yielded. Through step S192, an elastic prediction-plastic correction algorithm is used to calculate the trial stress tensor and check whether it exceeds the dynamic yield stress. If it does, the stress is corrected to the yield surface through radial return mapping; otherwise, the stress is directly updated.

[0333] Simultaneously, in step S192, the updated stress tensor increment and plastic strain increment are stored for the next time step to ensure the continuity of the stress history. Finally, the stress tensor increment is output to the global stress field array to provide data for subsequent stress analysis.

[0334] Further, in step S193, based on the contact geometry data of the adapter plug and guide groove of the connector, a global contact search algorithm is executed to identify potential contact pairs, taking into account the contact interface between the self-locking spring and the wire. Through step S193, the normal penetration of each contact pair is calculated, and the magnitude of the normal contact force is determined based on the penalty function method, wherein the contact stiffness is automatically optimized according to the material properties.

[0335] Then, in step S193, based on the anisotropic friction coefficient model of the contact material, the friction coefficient value varying with the direction angle is queried to simulate the actual sliding contact behavior of the adapter plug. Optionally, in step S193, the tangential friction force vector is calculated using the normal contact force and the friction coefficient, and the influence of sliding speed is considered to introduce a speed-dependent friction model.

[0336] In this embodiment, in step S193, the normal contact force and tangential friction force vectors are integrated to generate a total contact force vector, which is then updated to the nodal force array for solving the dynamic equations. Step S193 employs a smooth transition algorithm to handle abrupt changes in the contact state, avoiding numerical oscillations and ensuring the stability of the connector insertion and removal process. Simultaneously, changes in contact energy are monitored to ensure energy conservation in contact interactions and prevent energy anomalies. Finally, the updated contact force vector is output, providing input for the next calculation.

[0337] Alternatively, in step S194, the kinetic energy, internal energy, contact energy, and hourglass energy components of the explicit dynamic simulation are collected in real time to establish a complete energy balance equation. Step S194 is then used to calculate the total energy balance error and compare it with a preset tolerance threshold to detect numerical stability issues.

[0338] Then, in step S194, if the hourglass energy percentage is detected to exceed a threshold of 10%, the viscous damping hourglass control algorithm is automatically activated to suppress the development of the zero-energy mode. Optionally, in step S194, the artificial damping coefficient is adjusted, and the damping parameters are dynamically optimized based on the historical trend of the hourglass energy.

[0339] In this embodiment, step S194 monitors the historical trend of energy error. If the error continuously exceeds the limit, an adaptive time step reduction mechanism is triggered to improve calculation accuracy. Step S194 also records an energy monitoring log, including time history data for each component, for subsequent analysis. Simultaneously, when the total energy deviation exceeds the safe range, the calculation is automatically terminated and a warning message is output to prevent invalid results. Finally, a numerical stability report is generated to provide a basis for simulation verification.

[0340]

Example 9

[0341] Please refer to Figure 2 :

[0342] This embodiment describes a complete application scenario for internal damage detection of a modular meter box connector. The connector has completed a 100,000-cycle mating test, and it is necessary to evaluate whether key components such as its self-locking spring have developed internal microcracks or plastic deformation. The detection process employs a method combining ultrasonic non-destructive testing and numerical simulation, specifically including the following key steps:

[0343] First, preparatory work is performed. After removing the connector to be tested from the test bench, its outer shell surface is cleaned with anhydrous ethanol to ensure it is free of oil and dust. Special attention is paid to cleaning the standardized test window area pre-drilled on the frame. A layer of ultrasonic coupling agent of appropriate thickness is evenly applied to the surface of the test window. The 64-chip ultrasonic phased array probe is precisely fixed above the test window using a six-axis robotic arm, ensuring that the probe is perpendicular to the window surface and the contact pressure is stable within a specific range (e.g., 5 ± 0.5 N). The center frequency of the ultrasonic testing instrument is set to 10 MHz, the sampling depth to 50 mm, and the scanning angle range to -30° to +30°. This preparation process lays the foundation for subsequent high-quality data acquisition.

[0344] In this embodiment, in step S11, after the modular meter box connector, which has completed 100,000 mating cycles life test, is removed from the test bench, its outer shell surface is cleaned with anhydrous ethanol, with particular attention paid to cleaning the standardized detection window area reserved on the frame to remove oil and dust. Next, an ultrasonic coupling agent with a thickness of 0.5 mm to 1.0 mm is uniformly coated on the surface of the detection window to ensure efficient sound wave transmission. Further, a 64-chip ultrasonic phased array probe is precisely fixed above the detection window using a six-axis robotic arm. The robotic arm is adjusted so that the probe's central axis is perpendicular to the window surface, and the contact pressure is controlled to remain stable within the range of 5 N ± 0.5 N.

[0345] Then, the center frequency of the ultrasonic testing instrument is set to 10MHz, the sampling depth to 50mm, and the scanning angle range to -30° to +30° to cover the area where critical components inside the connector, such as the self-locking spring, are located. In this embodiment, the ultrasonic system is controlled to sequentially excite each chip to emit ultrasonic pulses with a pulse width of 100ns, while all chips synchronously receive the full waveform echo signal. The signal of each transmit-receive pair is accumulated and averaged 256 times to suppress random noise. Optionally, the acquired full waveform data is arranged by chip number and time sequence and stored in a three-dimensional array format, with the array dimensions corresponding to the transmitting chip index, receiving chip index, and time sampling point.

[0346] Furthermore, the ambient temperature was maintained at 23℃±2℃ throughout the scanning process, and a temperature control system was used to reduce the impact of temperature fluctuations on sound velocity measurements. Simultaneously, the coupling agent layer thickness and probe contact pressure were monitored in real time; if the deviation exceeded the set tolerance, the robotic arm's fine-tuning program was automatically triggered. Subsequently, the stored 3D array data underwent preliminary verification, checking data integrity and marking abnormal channels to ensure that subsequent processing was based on a valid dataset.

[0347] In step S12 of this embodiment, waveform signals for each transmitter-receiver pair are extracted from the raw internal ultrasonic data in the three-dimensional array format and bandpass filtered. The passband frequency range is 8MHz to 12MHz to remove low-frequency noise and high-frequency interference. Next, for the three-dimensional spatial region to be imaged inside the connector, the region is divided into a 0.1mm × 0.1mm × 0.1mm voxel grid, and the acoustic wave propagation delay for all transmitter-receiver pairs is calculated for each voxel. Further, based on the acoustic wave propagation velocity model in materials, which is preset according to the sound velocity values ​​of the connector shell and internal metal materials, the theoretical time delay from each transmitter-receiver pair to the current voxel is calculated.

[0348] Then, for each voxel, the waveform signals of all corresponding transmit-receive pairs are aligned according to the calculated time delay and coherently superimposed. The amplitude of the superimposed signal is used as the acoustic intensity value of that voxel, thereby reconstructing a high-resolution three-dimensional acoustic image. In this embodiment, based on the reconstructed three-dimensional acoustic image, the longitudinal wave velocity and transverse wave velocity values ​​of each voxel are extracted. Using the linear relationship formula σ = K×(V - V0) / V0 in the acoustoelastic theory, where σ is stress, K is the acoustoelastic coefficient, V is the measured sound velocity, and V0 is the reference sound velocity under stress-free conditions, the stress tensor components of each voxel are calculated by inversion. Optionally, the acoustoelastic coefficient is obtained through calibration experiments. For the phosphor bronze material of the self-locking spring, the longitudinal wave acoustoelastic coefficient K11 is taken as 2.5×10⁻⁶. -5 MPa -1 .

[0349] In step S12, the calculated stress tensors at each point are combined to generate a first three-dimensional stress field. This first three-dimensional stress field is then smoothed using a 3×3×3 mean filter kernel to eliminate local fluctuations. Simultaneously, the first three-dimensional stress field is spatially aligned with the connector's CAD model, and feature point registration ensures that the stress field coordinates match the actual object.

[0350] In step S13 of this embodiment, the CAD assembly model of the modular meter box connector is imported, and the model is geometrically cleaned to remove non-critical features such as small chamfers and threaded holes, while precisely preserving the C-shaped gradient curvature structure of the self-locking spring. Next, the cleaned geometry is meshed using C3D8R elements, with local mesh refinement in the contact area between the self-locking spring and the wire, and the minimum element size set to 0.1 mm. Furthermore, the self-locking spring is given a Johnson-Cook plastic constitutive model made of phosphor bronze material, with the following parameters: initial yield stress A = 200 MPa, hardening coefficient B = 250 MPa, hardening exponent n = 0.35, and strain rate sensitivity coefficient C = 0.015.

[0351] Then, in step S13, surface-to-surface contact pairs are defined between the self-locking spring and the wire, and between the adapter and the guide groove. The penalty function method is selected as the contact algorithm, and the normal contact stiffness is set to 1×10. 6 The friction coefficient is set to 0.15 N / mm. In this embodiment, a displacement load along the guide groove direction is applied to the adapter plug, and the load curve is a trapezoidal wave with a uniform velocity of 50 mm / s, simulating actual plugging and unplugging operations. Optionally, six degrees of freedom of the frame are constrained, and spring damping boundary conditions are applied on the meter interface side, with a stiffness set to 1×10 N / mm. 5 The damping coefficient is set to 200 N·s / m to simulate the actual installation environment.

[0352] Furthermore, in step S13, the explicit central difference algorithm is used to solve the dynamic equations, with the time step automatically adjusted and the initial step size set to 1×10⁻⁶. -7 The simulation duration was 0.5 s. Simultaneously, the kinetic energy to internal energy ratio was monitored, and an energy dissipation algorithm was automatically triggered when it exceeded 10%. Subsequently, stress field data for the stable insertion / removal phase (0.2 s to 0.4 s) was extracted from the simulation results, and the time-averaged stress field was calculated as the second three-dimensional stress field.

[0353] In step S14 of this embodiment, the first three-dimensional stress field and the second three-dimensional stress field are resampled to the same voxel grid resolution with a grid size of 0.1mm × 0.1mm × 0.1mm to ensure spatial alignment. Next, an image registration algorithm based on mutual information is used to perform fine registration of the two stress fields to eliminate displacement errors caused by coordinate system differences. Further, a voxel-level subtraction operation is performed on the registered two stress fields, that is, the stress values ​​at corresponding coordinate positions are subtracted to generate the original three-dimensional stress difference field.

[0354] Then, the original three-dimensional stress difference field is subjected to Gaussian filtering with a kernel size of 3×3×3 pixels and a standard deviation of 0.5 to suppress random noise and retain local stress difference peaks. In this embodiment, the global mean and standard deviation of the filtered difference field are calculated, and the difference values ​​are standardized to Z-score form for easier subsequent thresholding. Optionally, the range of the difference field data is checked to remove outliers caused by boundary effects.

[0355] Furthermore, the processed three-dimensional stress difference field is stored as a three-dimensional matrix and associated with the connector's three-dimensional CAD model for easy visualization. Simultaneously, statistical characteristics of the difference field, such as the maximum difference value, average difference value, and difference distribution histogram, are output to provide a reference for damage diagnosis.

[0356] Subsequently, in step S51, the three-dimensional stress difference field data is read, and its data range and voxel size are determined. Next, a three-dimensional Gaussian filter is applied, with the filter kernel size set to 3×3×3 voxels and the standard deviation σ of the Gaussian function set to 0.5 voxel units, to smooth noise while preserving local peaks with feature scales greater than 1 mm. Further, gradient calculation is performed on the filtered difference field, and the stress gradient amplitude at each voxel point is obtained using the Sobel operator.

[0357] Then, based on the yield strength (400 MPa) of the self-locking spring material, the dynamic damage determination threshold is set to 20% of the theoretical stress value, i.e., 80 MPa. In this embodiment, voxels with stress differences exceeding the threshold in the filtered difference field are marked as candidate damage points. Optionally, a region growing algorithm is used, with each candidate damage point as a seed point, to aggregate adjacent voxels with differences exceeding half the threshold based on the 6-connectivity criterion, forming a continuous potential damage region.

[0358] Furthermore, the geometric features of each potential damage region are calculated, including volume, surface area, and aspect ratio based on the minimum circumscribed cuboid. Simultaneously, the average stress difference, maximum stress difference, and stress gradient direction distribution within each region are extracted. These feature parameters for each region are then stored as feature vectors for subsequent classification.

[0359] In this embodiment, in step S72, the geometric and mechanical feature vectors of the potential damage region are standardized so that the mean of each feature component is 0 and the variance is 1. Then, the standardized feature vectors are reshaped into an 8×8 two-dimensional matrix, with zero values ​​padded to any insufficient parts. Further, the two-dimensional matrix is ​​input into a pre-trained convolutional neural network classifier. This network structure includes two convolutional layers (with 32 and 64 filters respectively, and a 3×3 kernel size) and two fully connected layers (with 128 and 3 neurons respectively).

[0360] Subsequently, the convolutional neural network employs the ReLU activation function, and the output layer uses the Softmax function to generate probability distributions for three damage modes (microcracks, plastic deformation, and fatigue damage). In this embodiment, the network weights have been trained on a historical damage dataset, using the cross-entropy loss function and the Adam optimizer during training. Optionally, if the highest probability of the network output is lower than a preset threshold of 0.85, an auxiliary classification strategy is activated, inputting the same feature vector in parallel into a support vector machine classifier.

[0361] Furthermore, in step S72, the support vector machine classifier uses a linear kernel function with a penalty parameter C=1.0 for secondary classification. Simultaneously, combining the outputs of the convolutional neural network and the support vector machine, a weighted voting method (with weights of 0.7 and 0.3 respectively) is used to determine the final damage pattern label. Then, the classification result is associated with the spatial coordinates of the potential damage region.

[0362] In step S83, for the damaged area classified as a microcrack, the current crack length is extracted. ,depth and orientation angle Next, based on the second three-dimensional stress field data, the range of Type I stress intensity factor at the crack tip is calculated. Considering the geometric correction factor Y(a), further, input ΔK and crack parameters into the Paris law model: ,in and The material constant is determined by phosphor bronze fatigue testing (C = 1.2 × 10⁻⁶). -11 (m=3.0), predicting cracks with increasing insertion and extraction cycles. The extended trajectory.

[0363] Then, in step S83, for the damaged areas classified as plastic deformation, their equivalent plastic strain is extracted. In this embodiment, the Lemaitre continuous damage mechanics model is adopted, and the damage evolution law is as follows: ,in For fracture strain, Calculate damage variables for material parameters. With the number of loops The growth of [damage]. Optionally, a damage threshold can be set. ,when Material failure is determined at that time.

[0364] Furthermore, in step S83, an evolution model of microcracks and plastic damage is integrated, and the damage development process is simulated with the number of insertion / removal cycles as the time axis. Simultaneously, 1000 sets of random load sequences are generated through Monte Carlo simulation, the distribution function of the remaining service life is statistically analyzed, and the expected life and 90% confidence interval are output.

[0365] Finally, the prediction results are coupled with the number of insertions and removals monitored in real time to update the remaining lifetime estimate.

[0366] [Example 10]: Optimized design and application of the modular meter box connector

[0367] like Figure 4 , Figure 5 As shown, this embodiment details a specific implementation of the modular meter box connector. Through structural innovation and material optimization, this connector achieves reliable electrical connections and convenient non-destructive testing, making it particularly suitable for connecting energy metering devices to external lines in smart grids. The connector includes a frame 1 and a detachable adapter 2, wherein the adapter 2 integrates current conduction, wire clamping, and acoustic detection auxiliary functions. The components are described in detail below with reference to the accompanying drawings.

[0368] The frame 1 serves as the support structure for the connector, and its interior is machined with multiple guide grooves. These guide grooves are formed using precision injection molding to ensure dimensional consistency and guiding accuracy. The frame 1 is typically made of flame-retardant engineering plastics, such as reinforced nylon or polycarbonate, to meet the safety standards of the electrical equipment. The guide grooves are designed to allow the adapter plug 2 to be inserted and removed along a specific path, preventing misalignment or tilting. Standardized inspection windows are also provided on the sides of the frame 1 to facilitate contact with the ultrasonic probe for non-destructive scanning. The positions of these windows are aligned with the key components of the adapter plug 2 to optimize inspection efficiency.

[0369] The adapter plug 2 is the core component of the connector and is detachably mounted within the guide groove of the frame 1. The housing of the adapter plug 2 is made of the high-performance engineering plastic, but its acoustic properties have been specifically optimized. Specifically, the acoustic impedance of the housing material is gradient-distributed from the outer surface inwards, smoothly transitioning from an acoustic impedance value close to that of the ultrasonic coupling agent (e.g., approximately 2.5 MRayl) to an acoustic impedance value close to that of the internal metal components (e.g., approximately 40 MRayl), with an impedance difference greater than 30 MRayl. This gradient design is achieved through co-injection molding of the multilayer composite material, with the acoustic impedance of each layer varying exponentially, thereby significantly reducing the reflection loss of the ultrasonic waves at the interface and improving the signal penetration. Furthermore, on a specific outer surface of the adapter plug 2 housing, a plurality of pits or protrusions are formed through precision machining. The diameter D of these microstructures satisfies D = V_sound / (2f) with respect to the ultrasonic detection center frequency f, where V_sound is the velocity of sound in the material (e.g., for the ultrasonic wave with a center frequency of 10 MHz, the pit diameter is approximately 0.3 mm). These arrays together constitute the embedded acoustic lens, which can focus the ultrasonic beam and improve the detection resolution. Furthermore, acoustic reinforcement material strips arranged in a grid pattern are embedded in the outer shell material. The absolute value of the difference between the acoustic impedance value Z2 of the strips and the average acoustic impedance Z1 of the internal metal is not less than 3 MRayl. For example, zirconia ceramic strips (Z2≈22 MRayl) are embedded in a polyetheretherketone matrix (Z1≈3 MRayl) to generate controllable acoustic contrast, facilitating image recognition.

[0370] The current strip 201 is housed within the internal cavity of the adapter plug 2. This component is made of a highly conductive copper alloy and silver-plated to reduce contact resistance. The current strip 201 is located at the end of the adapter plug 2 closest to the meter interface, with its end forming a terminal block for establishing an electrical connection with the socket of the meter interface. The main body of the current strip 201 includes a top connecting strip and a vertical dividing strip, these structures being formed by stamping and bending to ensure mechanical strength and current carrying capacity. At the end of the current strip 201 furthest from the meter interface, a wiring hole is provided for inserting an external wire. The diameter of the wiring hole is designed according to the wire specification; for example, for a wire with a cross-sectional area of ​​2.5 mm², the diameter is approximately 3 mm, and it features a guide bevel for insertion.

[0371] The clamping mechanism is a key functional component of the adapter plug 2, which uses the self-locking spring 202 to automatically clamp the wires. The self-locking spring 202 is made of phosphor bronze, a material that not only has good conductivity and elasticity, but also has a calibrated acoustoelastic coefficient (e.g., the longitudinal wave acoustoelastic coefficient K11 is approximately 2.5 × 10⁻⁶). -5 MPa -1 This allows for the calculation of the stress state experienced by the ultrasonic wave by measuring the change in its propagation speed within the reed. The self-locking reed 202 has a C-shaped structure, including a top section, a vertical connecting section, and a clamping section. One end of the vertical connecting section is fixed to the top connecting strip of the current bar 201 via a connecting claw, while the other end transitions into the clamping section in an arc shape. When an external wire is inserted through the wiring hole and contacts the top section, the clamping section elastically deforms and pops outward, pressing the wire against the vertical dividing strip of the current bar 201, forming a reliable electrical connection. This self-locking mechanism requires no additional tools, is easy to operate, and provides stable contact pressure.

[0372] The wire removal mechanism works in conjunction with the self-locking spring 202 to release the wire clamping. The wire removal mechanism includes the wire removal button, the spring, and the abutment. The exposed surface of the wire removal button is made of an acoustic impedance material different from the surrounding housing (e.g., a stainless steel insert with an acoustic impedance of approximately 45 MRayl), or has a special geometric texture (e.g., a cross groove) to make it clearly identifiable in the ultrasound image, thus serving as a reference benchmark for verifying the operational status. When the wire removal button is pressed, the spring transmits force to the abutment, which contacts the clamping section of the self-locking spring 202, causing it to spring back and release the wire, allowing it to exit. The entire wire removal process is smooth and controllable, avoiding mechanical damage to the wire.

[0373] The adapter plug 2 also has a signal hole on its housing for leading out signal lines from built-in sensors (such as temperature or strain sensors). These sensors can monitor the connector's operating status in real time, providing data support for the digital twin model. The signal hole is sealed with a rubber plug to ensure that the protection level reaches IP54 or higher.

[0374] The connector operates as follows: During installation, the adapter plug 2 is inserted along the guide groove of the frame 1 until the terminal of the current bar 201 engages with the meter interface. Then, an external wire is inserted through the wiring hole. The wire pushes the top section of the self-locking spring 202, deforming the clamping section and clamping the wire. For disassembly, pressing the wire release button releases the wire, allowing the adapter plug 2 to be pulled out. Throughout its lifespan, the connector can undergo periodic ultrasonic non-destructive testing. During testing, the ultrasonic phased array probe is placed on the detection window of the frame 1, and internal data is acquired through full-matrix capture scanning. Due to the acoustically optimized design of the housing, the ultrasonic waves can penetrate efficiently, and the stress distribution is inverted using the acoustoelastic properties of the self-locking spring 202, thereby assessing damage.

[0375] The connector described in this embodiment solves the problem of quantitative detection of internal damage in traditional connectors through a multi-dimensional combination of structural innovation and materials science. The gradient acoustic impedance shell and acoustic lens structure improve the quality of the ultrasonic signal, making the identification of minute defects possible; the acoustoelastic calibration of the self-locking spring 202 enables quantitative inversion of the stress state; and the acoustic marking of the unwind button assists in determining the detection benchmark. These features work synergistically to not only ensure the reliability of the electrical connection but also provide a technical basis for predictive maintenance, extending the service life of the equipment. Furthermore, the modular design allows for individual replacement of the adapter plug 2, reducing maintenance costs.

[0376] In the specific manufacturing process, the frame 1 is injection molded, and the mold precision requirements are high to ensure the parallelism of the guide groove. The outer shell of the adapter plug 2 achieves the acoustic impedance gradient through two-color injection molding, wherein the high impedance layer is filled with fillers such as tungsten powder, and the low impedance layer uses a microporous structure to regulate the sound velocity. The self-locking spring 202 is formed by stamping and heat treatment of phosphor bronze strip, and its acoustic elastic coefficient is calibrated by ultrasonic resonance method. During assembly, the current bar 201 and the self-locking spring 202 are first pre-assembled into sub-modules, then embedded into the outer shell, and finally the wire removal mechanism is installed. The finished product undergoes insertion and removal life testing (e.g., 10,000 cycles) and electrical performance verification to ensure compliance with industry standards.

[0377] The connector described in this embodiment has been applied in actual meter boxes and has demonstrated excellent performance. For example, in a smart grid demonstration project, the connector successfully achieved a zero-failure record during long-term operation, and the ultrasonic testing data was highly consistent with the simulation results, proving the rationality of its design. In the future, by integrating IoT sensors, the connector can be further upgraded into a smart node, enabling real-time status monitoring and big data analysis.

[0378] By incorporating the aforementioned acoustic impedance gradient design, the acoustic performance of the connector housing is further enhanced. The acoustic impedance value exhibits a gradient distribution from the outer surface of the housing inwards, smoothly transitioning from the acoustic impedance value Z_out (approximately 2.5 MRayl) close to that of the coupling agent to the acoustic impedance value Z_in (approximately 40 MRayl) of the internal metal components, with an impedance difference greater than 30 MRayl. This gradient is achieved through co-injection molding of the multilayer composite material, with the acoustic impedance of each layer varying exponentially. For example, from the outer layer to the inner layer, polyetheretherketone (3 MRayl), epoxy resin (5 MRayl), and the alumina composite material (15 MRayl) are used sequentially. This gradient structure effectively suppresses the reflection and scattering of ultrasonic waves at the interface, allowing the sound wave energy to penetrate the key internal areas more concentratedly, such as the location of the self-locking spring. The gradient design and the acoustoelastic calibration of the self-locking spring work synergistically: the gradient layer optimizes the sound wave propagation path, while the acoustoelastic properties make the spring a built-in stress sensor. The clamping force or residual stress can be quantitatively assessed through ultrasonic velocity measurement without the need for additional sensors. This integrated solution overcomes the difficulty of invasive placement required for stress measurement in traditional methods, achieving non-contact mechanical state monitoring.

[0379] In the aforementioned operation process, the deformation of the self-locking spring during wire insertion generates transient stress fluctuations, and the gradient acoustic impedance housing ensures that the ultrasonic signal can capture these details. For example, in the insertion / removal life test, the ultrasonic phased array scanning can invert the stress history through changes in the spring's sound velocity, and, combined with the reference point of the wire removal button, accurately identify the location of fatigue damage. In terms of technical effectiveness, the combined design enhances the connector's intelligence: the self-locking mechanism ensures the mechanical reliability of the electrical connection, the acoustic impedance gradient enhances the quality of the detection signal, and the acoustoelastic inversion provides the mechanical state data. The interaction of these three elements forms a closed loop of "operation-detection-evaluation," enabling the connector to not only possess the connection function but also become a state-sensing unit. This innovative integration solves the problem of balancing mechanical reliability and state monitoring in power equipment, providing technical support for predictive maintenance.

[0380] The acoustic lens and the reinforcing strip enhance the detection accuracy by further optimizing the spatial resolution and signal contrast of the ultrasonic detection. The connector housing adopts a composite design of the microstructure array and the acoustic reinforcement material strip. On a specific outer surface of the adapter housing, a plurality of pits or protrusions are precisely machined. These microstructures have specific geometries and dimensions, with their diameter D satisfying D = V_sound / (2f) with the center frequency f of the ultrasonic detection, where V_sound is the velocity of sound in the material. For example, for the ultrasonic wave with a center frequency of 10MHz, if the velocity of sound in the material is 3000 m / s, the pit diameter is designed to be 0.15mm. The array together constitutes the embedded acoustic lens, which can focus the ultrasonic beam, reduce sound energy diffusion, and thus improve the resolution of the detection area. The lens principle is based on wavefront modulation. The depth and spacing of the pits or protrusions are optimized through simulation to concentrate the sound wave energy at the focal point, making it particularly suitable for fine imaging of small components such as self-locking springs. Simultaneously, the acoustic enhancement material strips are embedded in the outer shell in a grid-like or array-like arrangement, positioned corresponding to the key internal components (such as the interface between the current strip and the clamping mechanism). The absolute value of the difference between the acoustic impedance value Z2 of the strip and the average acoustic impedance Z1 of the internal material is not less than 3 MRayl. For example, when Z1 is approximately 30 MRayl, the strip is made of silicon carbide (Z2≈35 MRayl) or polymer foam (Z2≈0.5 MRayl) to generate controllable acoustic contrast. This design allows the key areas to present high contrast in the ultrasound image, facilitating automatic identification and positioning.

[0381] The synergistic effect of the acoustic lens and the enhancement strip solves the technical problem of signal ambiguity within complex structures. The lens's focusing function reduces the ultrasonic beam width, concentrating the detection energy on the damage-prone areas, while the enhancement strip highlights the structural boundaries through impedance contrast. The combination of these two achieves a balance between "global scanning and local focusing." For example, in full-matrix capture scanning, the lens enhances the ability to identify microcracks, while the strip helps distinguish between different material interfaces, reducing false positives. Furthermore, the design is compatible with the acoustic impedance gradient layer: the gradient layer is responsible for the efficient penetration of broadband acoustic waves, the lens for focus optimization, and the strip for feature enhancement, forming a multi-layered acoustic management system. This integrated solution overcomes the resolution limitations of traditional ultrasonic testing in complex components, and is particularly suitable for multi-material, small-scale scenarios such as connectors.

[0382] In this implementation, the microstructure array is formed through micro-injection molding or laser processing to ensure dimensional accuracy and consistency. The arrangement of the reinforcing strips is determined based on finite element acoustic simulation to maximize the signal-to-noise ratio of the critical region. The technical effects are reflected in three aspects: first, the lens focusing improves the spatial resolution by approximately 30%, enabling the identification of sub-millimeter level damage; second, the strip contrast simplifies the image segmentation algorithm and improves the efficiency of automated diagnosis; and third, the overall design reduces the requirements for probe position during detection, enhancing operational robustness. The acoustic optimization not only serves damage detection but can also be used for installation verification, such as confirming whether the wire is clamped in place using ultrasonic images. The ingenious essence lies in the deep integration of acoustic engineering and structural design, transforming the connector housing from a passive enclosure into an active detection interface, solving the problem of poor adaptability of non-destructive testing in power equipment.

[0383] The following is a glossary of the key terms in the relevant embodiments, including textual descriptions and necessary formula models.

[0384] 1. Ultrasonic Phased Array Full Matrix Capture Scan

[0385] This refers to a measurement method that uses each crystal of an ultrasonic phased array probe as a sequential transmitting unit, with all crystals simultaneously acting as receiving units, to obtain a complete ultrasonic response data matrix. This method obtains full waveform data of any point inside an object by controlling the transmission-reception timing, providing a data foundation for subsequent advanced imaging algorithms.

[0386] 2. Acoustoelastic Theory Inversion

[0387] The stress measurement method based on the acoustoelastic effect establishes a physical relationship between the sound wave propagation characteristics (such as wave velocity and attenuation) and the stress state of the material, and then quantitatively reconstructs the stress field from the ultrasonic measurement data.

[0388] 3. Parametric 3D Finite Element Model

[0389] This is a computer model built based on the geometric features of the connector, which can be quickly modified by adjusting parameters (such as dimensions, material properties, boundary conditions, etc.). The model is discretized using the finite element method and includes complete definitions of nodes, elements, material constitutives, and boundary conditions. Its geometry can be represented as: G(p) = {N(p), E(p), C(p)}, where p is the parameter vector, N is the node set, E is the element set, and C is the constraint set.

[0390] 4. Explicit Dynamics Simulation

[0391] The numerical simulation process of solving the dynamic equations using the explicit time integration method is particularly suitable for analyzing short-time, transient problems involving large deformations and complex contacts.

[0392] 5. Voxel-by-Voxel Subtraction Operation

[0393] A mathematical operation that involves subtracting the values ​​of two three-dimensional data fields at the same spatial coordinate positions point by point.

[0394] 6. Local Stress Difference Peak

[0395] In a three-dimensional stress difference field Δσ(x,y,z), points whose values ​​are significantly higher than the local maxima of the surrounding adjacent regions are considered. Mathematically, this requires the following simultaneous conditions:

[0396] Δσ(x,y,z)>Δσ(x±i,y±j,z±k) and Δσ(x,y,z)>θ

[0397] Where i, j, k take the value of 1 or 2, and θ is a threshold value set according to the yield strength of the material.

[0398] 7. Johnson-Cook Dynamic Yield Criterion

[0399] A constitutive model describing the yielding behavior of metallic materials at high strain rates, whose expression comprehensively considers strain hardening, strain rate strengthening, and thermal softening effects:

[0400]

[0401] Where A is the initial yield stress, B is the hardening coefficient, n is the hardening exponent, C is the strain rate sensitivity coefficient, and m is the thermal softening exponent. For equivalent plastic strain, For plastic strain rate, For reference strain rate, T = (T-T_room) / (T_melt-T_room) is dimensionless temperature.

[0402] 8. Explicit Central Difference Algorithm

[0403] The explicit time integration method has the following recursive formula:

[0404] a_n = M -1 (F_ext,n - F_int,n)

[0405] v_{n+1 / 2} = v_{n-1 / 2} + a_n Δt_n

[0406] u_{n+1} = u_n + v_{n+1 / 2} Δt_{n+1 / 2}

[0407] Where M is the mass matrix, F_ext is the external force vector, and F_int is the internal force vector.

[0408] 9. Convolutional Neural Network Damage Classifier

[0409] A machine learning model employing a deep convolutional neural network architecture is used to automatically identify and classify damage patterns.

[0410] 10. Paris' Law Crack Growth Model

[0411] An empirical model describing the fatigue crack propagation rate is in the following form:

[0412] da / dN = C(ΔK) m

[0413] Where a is the crack length, N is the number of load cycles, ΔK is the stress intensity factor range, and C and m are material constants.

[0414] 11. Continuum Damage Mechanics Model

[0415] A theoretical framework for describing material damage evolution using continuous internal variables. The damage evolution equation is usually expressed as:

[0416] dD / dt = f(σ, (, D, T, ...)

[0417] Where D is the damage variable (0 ≤ D ≤ 1), and σ is the stress tensor. Let T be the strain tensor and T be the temperature.

[0418] 12. Remaining Useful Life Probability Distribution

[0419] The remaining lifetime estimate described using probabilistic methods is expressed as a function of the failure probability over time:

[0420] P_f(t) = 1 - exp[-∫0 t λ(τ)dτ]

[0421] Where λ(τ) is the failure rate function related to the damage state.

[0422] In addition, the following rules are adopted in the mathematical formula expressions of some embodiments of this application:

[0423] 1. Subscript Representation Rules

[0424] Subscripts are represented using an underscore symbol (_) followed by the subscript identifier:

[0425] Single-character subscript: σ_d indicates that d is the subscript of σ.

[0426] Multi-character subscripts: W_ij indicates that ij is the subscript of W (Note: here ij should be regarded as a whole identifier).

[0427] It is recommended to use explicit parentheses for multi-character subscripts: W_{ij} more clearly indicates that ij is a compound subscript.

[0428] 2. Superscript Representation Rules

[0429] Superscript is indicated by using the caret character ^ followed by the superscript content:

[0430] Square: x^2 represents x raised to the power of two.

[0431] Exponent: e^x represents the natural exponent raised to the power of x.

[0432] Compound superscript: x^{n+1} represents x raised to the power of (n+1).

[0433] Figure 3This is a block diagram illustrating a connector damage detection device 300 according to an exemplary embodiment. For example, the detection device 300 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0434] Reference Figure 3 The detection device 300 may include one or more of the following components: processing component 302, memory 304, power supply component 306, multimedia component 308, audio component 310, input / output interface 312, sensor component 314, and communication component 316.

[0435] Figure 3 The detection device 300 shown, with its processing component 302 and memory 304, is particularly suitable for executing the coordinated control method described in this invention. For example, the processing component 302 can perform the steps of data fusion, passenger flow prediction, and collaborative decision-making, and the memory 304 can store the fused data, prediction models, optimization algorithms, and computer programs. The communication component 316 is used for data interaction with the multimodal sensor network and elevator group control system within the building.

[0436] Processing component 302 typically controls the overall operation of detection device 300, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 302 may include one or more processors 320 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 302 may include one or more modules to facilitate interaction between processing component 302 and other components. For example, processing component 302 may include a multimedia module to facilitate interaction between multimedia component 308 and processing component 302.

[0437] Memory 304 is configured to store various types of data to support the operation of detection device 300. Examples of this data include instructions for any application or method operating on detection device 300, contact data, phone book data, messages, pictures, videos, etc. Memory 304 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0438] The power supply assembly 306 provides power to the various components of the detection device 300. The power supply assembly 306 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the detection device 300.

[0439] The multimedia component 308 includes a screen that provides an output interface between the detection device 300 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, the multimedia component 308 includes a front-facing camera and / or a rear-facing camera. When the detection device 300 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0440] Audio component 310 is configured to output and / or input audio signals. For example, audio component 310 includes a microphone (MIC) configured to receive external audio signals when the detection device 300 is in an operating mode, such as a call mode, recording mode, or voice recognition mode. The received audio signals may be further stored in memory 304 or transmitted via communication component 316. In some embodiments, audio component 310 also includes a speaker for outputting audio signals.

[0441] Input / output interface 312 provides an interface between processing component 302 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, start buttons, and lock buttons.

[0442] Sensor assembly 314 includes one or more sensors for providing status assessments of various aspects of detection device 300. For example, sensor assembly 314 can detect the on / off state of detection device 300, the relative positioning of components such as the display and keypad of detection device 300, changes in position of detection device 300 or a component of detection device 300, the presence or absence of user contact with detection device 300, orientation or acceleration / deceleration of detection device 300, and temperature changes of detection device 300. Sensor assembly 314 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 314 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 314 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0443] Communication component 316 is configured to facilitate wired or wireless communication between detection device 300 and other devices. Detection device 300 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 316 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 316 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0444] In an exemplary embodiment, the detection device 300 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0445] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 304 including instructions, which can be executed by the processor 320 of the detection device 300 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

Claims

1. A method for detecting test damage to a connector, characterized in that, Includes the following steps: S11: Use an ultrasonic phased array probe to perform a full matrix capture scan on the physically tested connector to obtain raw internal ultrasonic data of the connector; S12: Based on the original internal ultrasound data, the three-dimensional acoustic image of the connector is reconstructed using the full focusing method, and the acoustic elasticity of the connector and the corresponding first three-dimensional stress field are calculated based on the three-dimensional acoustic image. S13: Establish a parameterized three-dimensional finite element model of the connector, and run an explicit dynamic simulation on the parameterized three-dimensional finite element model to calculate the second three-dimensional stress field simulating the insertion, clamping and unclamping operations. S14: Perform a voxel-to-voxel subtraction operation between the first three-dimensional stress field and the second three-dimensional stress field to generate a three-dimensional stress difference field; S15: Based on the local stress difference peak value in the three-dimensional stress difference field, diagnose whether there is damage inside the connector and locate the damage location. Specifically, step S11, which involves full-matrix capture scanning, includes: S31: Control each crystal of the ultrasonic phased array probe to emit ultrasonic pulses in sequence, and synchronously control all crystals to receive full-waveform echo signals from inside the connector; S32: Arrange the full waveform echo signal according to the transmit-receive chip pairs and store it in a full matrix data format; S33: Based on the full matrix data, the ultrasonic propagation time delay law is used to perform pixel focusing calculation on a specific three-dimensional spatial region of the connector to generate a high-resolution three-dimensional acoustic image; Following step S33, a time-frequency joint analysis step is also included: S91: Perform wavelet packet transform on the full matrix data to extract the multi-band acoustic energy attenuation coefficient distribution matrix; S92: Perform pixel-level fusion of the multi-band acoustic energy attenuation coefficient distribution matrix with the three-dimensional acoustic image; S93: Reconstructing the abrupt change in acoustic impedance of anisotropic materials based on the fused multidimensional data field; S94: Correct the wave propagation path calculation in the acoustic elasticity theory inversion by the abrupt change interface of acoustic impedance; The step S13, which involves calculating the second three-dimensional stress field to simulate the insertion, clamping, and retraction processes, specifically includes: S41: Based on the CAD geometric model of the connector, perform geometric cleanup and retain the C-shaped gradient curvature feature of the self-locking spring; S42: Assign an elastoplastic material constitutive model and Johnson-Cook dynamic yield criterion parameters to the cleaned geometry; S43: Define the surface contact pairs and coefficient of friction in the contact area between the self-locking spring and the wire, and at the contact interface between the adapter plug and the guide groove; S44: Apply boundary conditions simulating insertion and extraction speeds to the model, solve the transient dynamic equations using an explicit central difference algorithm, and output the second three-dimensional stress field.

2. The method for detecting damage to a connector according to claim 1, characterized in that, The steps in step S15 for diagnosing the injury specifically include: S51: Perform spatial Gaussian filtering on the three-dimensional stress difference field to suppress noise and retain local stress difference peaks; S52: Setting a dynamic damage assessment threshold based on the material yield strength of the self-locking spring; S53: Identify the continuous regions in the filtered three-dimensional stress difference field that exceed the dynamic damage determination threshold as potential damage regions; S54: Classify the damage pattern by combining the geometry of the potential damage area with the stress distribution gradient direction.

3. The method for detecting damage to a connector according to claim 2, characterized in that, The step of classifying damage patterns in step S54 specifically includes: S71: Extract the volume, surface area, aspect ratio feature vector, and principal stress direction distribution of the potential damage area; S72: Input the feature vector into a pre-trained convolutional neural network damage classifier for pattern recognition; S73: Determine the classification result of microcracks, plastic deformation, or fatigue damage type based on the output probability distribution of the convolutional neural network damage classifier; S74: Map the classification results to the spatial coordinates of the potential damage area and generate a three-dimensional damage distribution map.

4. A connector suitable for the detection method as described in any one of claims 1 to 3, characterized in that, include: The frame has multiple guide grooves formed inside; The adapter plugs are detachably disposed within the plurality of guide slots; The adapter plug includes: A current bar is located at the end of the adapter plug near the meter interface, and is used to establish an electrical connection with the meter interface; A wiring hole is provided at the end of the adapter plug away from the meter interface for inserting external wires; A clamping mechanism, located in the internal cavity of the adapter plug, is made of conductive material and connected to the current bar. It is used to clamp the external wire after it is inserted through the wiring hole, so as to fix it and make it conductive with the current bar. The adapter plug has at least a portion of its housing made of a material with adaptive acoustic properties, the acoustic impedance of which matches the acoustic impedance of the internal structural material to optimize the transmission and reflection of ultrasonic waves.

5. The connector according to claim 4, characterized in that, The clamping mechanism includes a self-locking spring; The self-locking spring has a C-shaped structure and includes: Top section, The vertical connecting section has one end connected to the current bar via a connecting claw and fixed to the top of the internal cavity of the adapter plug. Clamping section; The self-locking spring is made of phosphor bronze, and its acoustoelastic coefficient has been calibrated so that the stress state it experiences can be calculated by measuring the change in the propagation speed of ultrasonic waves within it. When an external wire is inserted and contacts the top segment, the clamping segment is configured to elastically deform to pop out and clamp the external wire into the internal cavity; The connector also includes a wire unwinding mechanism that contacts the clamping section; The wire ejection mechanism is configured to apply an external force to the clamping section, causing it to spring back and release the clamped external wire, thereby allowing the external wire to exit through the terminal hole; The exposed surface of the unwinding button in the unwinding mechanism is made of an acoustic impedance material different from the surrounding housing or has a target geometric texture, so that it can be clearly identified in the ultrasound image, thus serving as a reference benchmark for verifying the operational status.

6. The connector according to claim 4, characterized in that, The material with adaptive acoustic properties has an acoustic impedance value that gradually changes from the outer surface of the shell inward, smoothly transitioning from the acoustic impedance value Z_out of the coupling agent to the acoustic impedance value Z_in of the internal metal component, and |Z_out - Z_in|>30 MRayl.

7. The connector according to claim 6, characterized in that, On the target outer surface of the adapter plug housing, an array of pits or protrusions is machined. These pits or protrusions have the target geometry and size, and their diameter D satisfies the following with the center frequency f of ultrasonic testing: D = V_sound / (2f), where V_sound is the sound velocity in the material. These microstructure arrays together constitute an embedded acoustic lens.

8. The connector according to claim 4, characterized in that, The acoustic impedance of the material with adaptive acoustic properties is controlled within the range of Z1 ± 0.5 MRayl, where Z1 is the average acoustic impedance of the material used for the current bar and the clamping mechanism in the internal cavity of the adapter plug. Furthermore, the material with adaptive acoustic properties is composited in the housing of the adapter plug in a set three-dimensional distribution manner. The distribution manner is as follows: acoustic reinforcement material strips arranged in a grid or array are embedded in the housing wall corresponding to the key components in the internal cavity; the acoustic impedance value Z2 of the acoustic reinforcement material strips satisfies |Z2- Z1| ≥ 3MRayl.

9. An electric meter box, characterized in that, include: The connector as described in any one of claims 4 to 8.

Images