Automotive wire harness terminal ultrasonic welding method

Abstract

The application discloses an automobile wire harness terminal ultrasonic welding method and belongs to the field of automobile manufacturing. The welding method specifically comprises the following steps: I, before welding, collecting information of the terminal and the wire, establishing a digital simulation model of each terminal, simulating the welding process, and predicting welding defects; II, in the initial stage of the terminal and the wire being pressed, performing dynamic scanning on the contact surface, constructing a micro contact resistance distribution map, and identifying a high contact resistance area; the application realizes systematic and closed-loop control of the welding process, avoids limitations caused by single-link optimization, ensures stable welding quality under process fluctuations and material differences, avoids energy waste, realizes higher energy efficiency utilization, reduces dependence on manual experience, improves the strength and conductivity of the welding point, and reduces the scrap rate caused by defects.

Description

Technical Field

[0001] This invention relates to the field of automobile manufacturing, and more particularly to an ultrasonic welding method for automobile wiring harness terminals. Background Technology

[0002] As a core component of a vehicle's electrical system, the quality of terminal welding in automotive wiring harnesses directly affects the vehicle's electrical performance and safety. While traditional crimping and soldering processes can meet connection requirements to some extent, they have shortcomings in conductivity, fatigue resistance, and production consistency. Ultrasonic welding, due to its ability to achieve strong metal-to-metal connections at low temperatures, avoid flux residue contamination, and ensure high conductivity, has gradually become the mainstream process for automotive wiring harness terminal manufacturing. However, with the continuous improvement of automotive electronics and intelligence, the number of wiring harness terminals has increased dramatically, placing higher demands on the stability and controllability of welding quality. The ultrasonic welding process is characterized by high transientity, nonlinearity, and multi-factor coupling, making it difficult to detect and repair weld defects in a timely manner. Traditional process control modes relying on experience and fixed parameters can no longer meet the high reliability requirements of the modern automotive industry; therefore, the invention of an ultrasonic welding method for automotive wiring harness terminals has become particularly important.

[0003] Existing ultrasonic welding methods for automotive wiring harness terminals cannot achieve systematic and closed-loop control of the welding process. Furthermore, they cannot maintain stable welding quality under process fluctuations and material differences, which easily leads to energy waste and reduced energy efficiency. In addition, they rely heavily on manual experience, which reduces the strength and conductivity of the weld joint and increases the scrap rate due to defects. Therefore, we propose an ultrasonic welding method for automotive wiring harness terminals. Summary of the Invention

[0004] The purpose of this invention is to address the deficiencies in the prior art by proposing an ultrasonic welding method for automotive wiring harness terminals.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] An ultrasonic welding method for automotive wiring harness terminals, the specific steps of which are as follows:

[0007] 1. Before welding begins, collect information on the terminals and wires, establish digital simulation models for each terminal, and simulate the welding process to predict welding defects.

[0008] II. In the initial stage of terminal and wire clamping, the contact surface is dynamically scanned and a microscopic contact resistance distribution map is constructed to identify areas with high contact resistance.

[0009] III. Real-time acquisition of multi-source sensor data during the welding process, followed by fusion analysis of the acquired multi-source sensor data, and dynamic generation of ultrasonic energy output curves;

[0010] IV. Based on the contact resistance distribution and real-time sensing data, ultrasonic energy is directed to promote local plastic deformation, and the energy control strategy is optimized in real time.

[0011] V. Perform spectrum analysis on the acoustic signals generated during the welding process, and determine in real time whether there are potential defects in the weld joint. At the same time, based on the determination result, automatically trigger the self-repair program.

[0012] VI. Monitor the inherent modes of the welding head in real time, and dynamically adjust the driving parameters of the welding head based on the monitoring results. When the welding head shows signs of fatigue or wear, remind maintenance personnel to replace the welding head.

[0013] As a further aspect of the present invention, the specific steps for establishing the digital simulation model of each terminal in step I are as follows:

[0014] S1.1: Various basic material performance parameters of terminals and wires are collected using hardness testers, tensile testing machines and microscopes. Then, point cloud data of terminals and wires are obtained using three-dimensional laser scanning or industrial CT. Noise in the original point cloud data is removed based on the voxel downsampling criterion, and outliers in the denoised point cloud data are also removed.

[0015] S1.2: Based on the preprocessed point cloud data, obtain the three-dimensional geometric dimensions and shapes of each terminal and wire, and establish a digital simulation model of each terminal and wire based on the three-dimensional geometric dimensions and shapes of each terminal and wire.

[0016] S1.3: Using a surface roughness meter or resistivity meter, collect the surface roughness and oxide layer resistance characteristics of the contact surface between the terminal and the wire. Then, import the collected surface roughness and oxide layer resistance characteristics into the corresponding digital simulation model of the terminal and the wire to establish a complete terminal-wire simulation model.

[0017] S1.4: Based on the main excitation frequency and the highest effective frequency component of the ultrasonic welding head in the simulation, calculate and set the simulation time. Then, according to the amplitude, frequency and phase angle of ultrasonic welding in the digital simulation, calculate the displacement of the welding head end face under the preset simulation time to obtain the corresponding ultrasonic vibration boundary. After that, according to the material properties of each terminal and wire, set the simulation boundary conditions of the corresponding terminal-wire simulation model.

[0018] As a further aspect of the present invention, the specific basic material performance parameters of the terminals and wires mentioned in S1.1 include density, elastic modulus, yield strength, thermal conductivity, etc.

[0019] As a further aspect of the present invention, the specific steps for predicting welding defects during the simulated welding process described in step I are as follows:

[0020] S2.1: Based on existing experience, apply normal compression and tangential relative slip conditions to the terminal-wire contact interface in each terminal-wire simulation model, and calculate the interfacial frictional heat flux generated by friction and adhesion-slip transformation based on the applied conditions.

[0021] S2.2: Calculate the structural displacement, velocity and acceleration response under preset ultrasonic excitation, and then solve the transient heat conduction by superimposing the generated interfacial frictional heat flux and the volume heat source converted by material plastic dissipation to obtain the thermal-mechanical coupling state of the corresponding terminal-wire simulation model.

[0022] S2.3: Based on the latest thermo-mechanical coupling state of the current simulation time step, update the material flow resistance and softening. Integrate the contact compaction and tangential work to calculate the interface energy density. If the interface energy density is lower than the preset process threshold, it is determined that the simulated welding has a weak welding trend under the current parameters.

[0023] S2.4: Monitor the peak temperature of the interface and its neighborhood. If the peak temperature is higher than the preset upper limit, it indicates that there is metal splashing or overheating in the current welding process. Based on the evolution relationship of equivalent porosity fraction with plastic accumulation and rapid temperature rise, calculate the formation trend of unwelded - inclusion - pore connectivity in the current welding process.

[0024] S2.5: The interface is combined with three types of indicators—insufficiency, overheating, and porosity—and normalized and weighted to generate a real-time defect risk score for the current welding simulation. At the same time, the welding process parameters are adjusted in real time based on the real-time risk score.

[0025] As a further aspect of the present invention, the specific steps for identifying the high contact resistance region in step II are as follows:

[0026] S3.1: The terminal-wire contact surface is scanned by a grid using current micro-excitation. Based on the scanning results, a set of two-dimensional coordinate scanning grids is established on the contact surface. At the same time, the excitation time window and weight of each grid are set. Then, the measurement signal of each grid is synchronously integrated by phase-locked loop to obtain the coherent potential value corresponding to each grid.

[0027] S3.2: Using a pre-set transfer matrix, the measured coherent potential values ​​are corrected by linear transformation, and a preset current perturbation is applied to each grid. The corresponding potential changes are read, and the obtained change values ​​are used as local incremental resistances. Then, the obtained local incremental resistances are convolved and corrected.

[0028] S3.3: The corrected local incremental resistances are inverted using the Tikhonov inversion method, and the inversion results are normalized using robust statistics to obtain the relative resistance anomaly of each grid. Then, the resistance distribution of each grid is calculated, and the corresponding high resistance judgment threshold is set.

[0029] S3.4: If the relative resistance anomaly is higher than the high resistance judgment threshold, the corresponding grid is divided into the high contact resistance initial screening set. Then, through area opening operation and connected component screening, each group of grids is processed into multiple groups of connected grids. Based on the preset retention area threshold, connected grids higher than the retention area threshold are retained to obtain the high contact resistance region in the micro contact resistance distribution map.

[0030] As a further aspect of the present invention, the specific steps for dynamically generating the ultrasonic energy output curve in step III are as follows:

[0031] S4.1: Multiple sensors are arranged on the welding equipment, and the various sensors are synchronously collected at a fixed sampling frequency to obtain multimodal sensing data. Then, the sensing data of different modes are preprocessed through various operations such as noise reduction, smoothing and scale normalization. After that, the continuously collected multimodal sensing data is divided into multiple sets of sensing datasets using a sliding window.

[0032] S4.2: Each set of sensor datasets is input into a one-dimensional convolutional network, and the data in each set of sensor datasets is convolved by the convolution kernels of each convolutional layer in the one-dimensional convolutional network to extract the local feature values ​​of each sensor data.

[0033] S4.3: The extracted local feature values ​​are transmitted to the LSTM network, and the dependency relationship between the local feature values ​​is identified through the gating mechanism of the LSTM network. At the same time, based on the identified dependency relationship, the dynamic evolution data of the welding process is acquired in real time. Then, the current dynamic evolution data is analyzed in real time through the Softmax classifier, and the welding physical state of the corresponding sensor dataset in each sliding window is determined.

[0034] S4.4: The identified welding physical state is compared with the expected process curve in real time, and based on the comparison results, a real-time control signal is generated, and the ultrasonic energy output curve corresponding to each control signal in each time period is output.

[0035] As a further aspect of the present invention, the specific steps for promoting localized plastic deformation in step IV are as follows:

[0036] S5.1: Collect the current contact resistance distribution map and the real-time data collected by various types of sensors. Then, fuse the contact resistance distribution map and the sensor data field to construct the corresponding joint state field, so as to establish a hot zone-impedance composite distribution model for the current welding interface.

[0037] S5.2: Divide the generated weld points into multiple meshes and calculate the joint state field intensity of each mesh element. If the joint state field intensity is higher than the preset threshold, mark the corresponding mesh element as the priority compensation area. Then, based on the marking results, calculate the ultrasonic energy weight of the ultrasonic head for each mesh element.

[0038] S5.3: After the ultrasonic energy weight calculation is completed, the ultrasonic energy is concentrated and loaded into the corresponding high resistance region according to the spatial distribution results of each ultrasonic energy weight by the modulation controller. Then, the ultrasonic energy is input into the material interface in the form of high frequency shear stress and friction work, and the output ultrasonic energy causes micro-cracks to occur and break the oxide film on the surface of the weld joint.

[0039] S5.4: After the oxide film is broken, the ultrasonic energy continues to apply force to the corresponding weld surface, causing the contact area of ​​the weld surface to undergo plastic deformation under the action of ultrasonic energy, while filling the micropores generated during the welding process.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] The ultrasonic welding method for automotive wiring harness terminals uses a hardness tester, tensile testing machine, and microscope to test material properties. Then, it uses three-dimensional laser scanning or industrial CT to obtain three-dimensional point cloud data of the terminals and wires. After noise reduction and outlier removal, an accurate digital simulation model is constructed, and surface roughness and oxide layer resistance characteristics are imported. Subsequently, under ultrasonic excitation, the interface frictional heat, plastic dissipation, and thermo-mechanical coupling state are simulated to predict defects such as weak welds, overheating, and porosity, and a real-time risk score is generated. Then, current micro-excitation and inversion algorithms are used to obtain the contact resistance distribution map and identify high-resistance areas. At the same time, multimodal sensors are deployed on the welding equipment to collect and preprocess data, perform fusion analysis, identify the welding state in real time, and output control signals. The contact resistance distribution and sensor data field are fused to construct a thermal-impedance joint model to determine the energy compensation requirements of high-risk areas. Then, the ultrasonic energy is spatially weighted and concentrated on the high-resistance area. The oxide film is broken through shear stress and frictional work, and the interface plastic deformation and porosity closure are promoted. This achieves systematic and closed-loop control of the welding process, avoids the limitations of single-stage optimization, ensures stable welding quality under process fluctuations and material differences, avoids energy waste, achieves higher energy efficiency, reduces reliance on human experience, improves weld strength and conductivity, and reduces scrap rate due to defects. Attached Figure Description

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0043] Figure 1 This is a flowchart of the ultrasonic welding method for automotive wiring harness terminals proposed in this invention. Detailed Implementation

[0044] Example

[0045] Reference Figure 1 An ultrasonic welding method for automotive wiring harness terminals, the specific steps of which are as follows:

[0046] Before welding begins, information about the terminals and wires is collected, and a digital simulation model of each terminal is established. At the same time, the welding process is simulated to predict welding defects.

[0047] Specifically, various methods, including hardness testing, tensile testing, and microscopy, are used to collect basic material performance parameters of terminals and wires. Then, 3D laser scanning or industrial CT is used to acquire point cloud data of the terminals and wires. Noise in the original point cloud data is removed based on voxel downsampling criteria, and outliers are eliminated. Based on the preprocessed point cloud data, the 3D geometric dimensions and shapes of each terminal and wire are obtained. Based on these 3D geometric dimensions and shapes, digital simulation models of each terminal and wire are established. Finally, surface roughness testers or resistivity meters are used to collect data on the terminals and wires. The surface roughness and oxide layer resistance characteristics of the contact surface are collected and then imported into the digital simulation model of the corresponding terminal and wire. A complete terminal-wire simulation model has been established. Based on the main excitation frequency and the highest effective frequency component of the ultrasonic welding head in the simulation, the simulation time is calculated and set. Then, according to the amplitude, frequency and phase angle of ultrasonic welding in the digital simulation, the displacement of the welding head end face under the preset simulation time is calculated to obtain the corresponding ultrasonic vibration boundary. Finally, according to the material properties of each terminal and wire, the simulation boundary conditions of the corresponding terminal-wire simulation model are set.

[0048] Specifically, based on existing experience, normal compression and tangential relative slip conditions are applied to the terminal-wire contact interface in each terminal-wire simulation model. Based on these conditions, the interfacial frictional heat flux generated by friction and adhesion-slip transformation is calculated. Under preset ultrasonic excitation, the structural displacement, velocity, and acceleration responses are calculated. Then, based on the generated interfacial frictional heat flux and the volumetric heat source converted from material plastic dissipation, transient heat conduction is solved by superposition to obtain the thermo-mechanical coupling state of the corresponding terminal-wire simulation model. According to the latest thermo-mechanical coupling state at the current simulation time step, the material flow resistance and softening are updated. Finally, the contact compression and tangential work are integrated to calculate... The interface energy density is calculated. If the interface energy density is lower than the preset process threshold, it is determined that the simulated welding has a weak welding trend under the current parameters. The peak temperature of the interface and its neighborhood is monitored. If the peak temperature is higher than the preset upper temperature limit, it indicates that there is metal spatter or overheating in the current welding process. Based on the evolution relationship of equivalent porosity fraction with plastic accumulation and rapid temperature rise, the formation trend of incomplete welding, inclusions and pore connectivity in the current welding process is calculated. The three types of indicators, namely insufficient interface bonding, overheating and porosity, are normalized and weighted, and a real-time defect risk score for the current welding simulation is generated. At the same time, the welding process parameters are adjusted in real time based on the real-time risk score.

[0049] It should be further explained that the specific material performance parameters of terminals and wires include density, elastic modulus, yield strength, thermal conductivity, etc.

[0050] In the initial stage of terminal and wire clamping, the contact surface is dynamically scanned and a microscopic contact resistance distribution map is constructed to identify areas with high contact resistance.

[0051] Specifically, a gridded scan of the terminal-wire contact surface is performed using current micro-excitation. Based on the scan results, a set of two-dimensional coordinate scanning grids is established on the contact surface. The excitation time window and weight of each grid are set. The measurement signal of each grid is then synchronously integrated using phase-locked loop (PLL) to obtain the coherent potential value corresponding to each grid. Using a pre-set transfer matrix, the measured coherent potential values ​​are corrected through linear transformation. A preset current perturbation is applied to each grid, and the corresponding potential change is read. The obtained change value is used as a local incremental resistance. The obtained local incremental resistances are then convolved and corrected using Tik. The Honov-type inversion method inverts the corrected local incremental resistances and then normalizes the inversion results using robust statistics to obtain the relative resistance anomaly of each grid. Next, it calculates the resistance distribution of each grid and sets a corresponding high-resistance threshold. If the relative resistance anomaly is higher than the high-resistance threshold, the corresponding grid is assigned to a high-contact-resistance initial screening set. Then, through area opening and connected component filtering, each group of grids is processed into multiple connected grids. Based on a preset area retention threshold, connected grids with areas higher than the threshold are retained to obtain high-contact-resistance regions in the microscopic contact resistance distribution map. Example

[0052] Reference Figure 1 An ultrasonic welding method for automotive wiring harness terminals, the specific steps of which are as follows:

[0053] Multi-source sensor data is acquired in real time during the welding process, and then the acquired multi-source sensor data is fused and analyzed to dynamically generate an ultrasonic energy output curve.

[0054] Specifically, multiple sensors are deployed on the welding equipment, and these sensors synchronously acquire data at a fixed sampling frequency to obtain multimodal sensing data. The sensing data from different modalities are then preprocessed through denoising, smoothing, and scale normalization. A sliding window is then used to divide the continuously acquired multimodal sensing data into multiple sets of sensing datasets. Each set of sensing datasets is input into a one-dimensional convolutional network, and the data in each set is convolved by the convolutional kernels of each convolutional layer to extract local feature values ​​from each sensor data. These extracted local feature values ​​are then transmitted to an LSTM network, and the gating mechanism of the LSTM network identifies the dependencies between these local feature values. Based on these dependencies, dynamic evolution data during the welding process is acquired in real time. A Softmax classifier is then used to analyze the current dynamic evolution data in real time and determine the welding physical state of the corresponding sensing dataset in each sliding window. The determined welding physical state is compared in real time with the expected process curve, and based on the comparison results, real-time control signals are generated, outputting the ultrasonic energy output curves corresponding to each control signal at each time period.

[0055] Based on the contact resistance distribution and real-time sensing data, ultrasonic energy is directed to promote local plastic deformation, and the energy control strategy is optimized in real time.

[0056] Specifically, the current contact resistance distribution map and real-time data collected by various types of sensors are collected. Then, the contact resistance distribution map and sensor data field are fused to construct a corresponding joint state field to establish a thermal-impedance composite distribution model for the current welding interface. The generated weld points are divided into multiple grids, and the joint state field intensity of each grid cell is calculated. If the joint state field intensity is higher than a preset threshold, the corresponding grid cell is marked as a priority compensation area. Then, based on the marking results, the ultrasonic energy weight of the ultrasonic head for each grid cell is calculated. After the ultrasonic energy weight is calculated, the ultrasonic energy is concentrated and loaded into the corresponding high resistance area according to the spatial distribution results of each ultrasonic energy weight through the modulation controller. Then, the ultrasonic energy is input into the material interface in the form of high-frequency shear stress and frictional work, and the output ultrasonic energy causes microcracks to occur and rupture in the oxide film on the surface of the weld point. After the oxide film ruptures, the ultrasonic energy continues to apply force to the corresponding weld point surface, causing the contact area on the surface of the weld point to undergo plastic deformation under the action of ultrasonic energy, while filling the micropores generated during the welding process.

[0057] The acoustic signals generated during the welding process are subjected to spectral analysis, and the potential defects of the weld joints are determined in real time. Based on the determination results, the self-repair program is automatically triggered.

[0058] The system monitors the inherent modes of the welding head in real time and dynamically adjusts the driving parameters of the welding head based on the monitoring results. When the welding head shows signs of fatigue or wear, the system reminds maintenance personnel to replace the welding head.

Claims

1. An ultrasonic welding method for automotive wiring harness terminals, characterized in that, The specific steps of this welding method are as follows: I. Before welding begins, collect information on the terminals and wires, and establish a digital simulation model for each terminal. Simultaneously, simulate the welding process and predict welding defects. The specific steps for establishing the digital simulation model for each terminal are as follows: S1.1: Various basic material performance parameters of terminals and wires are collected using hardness testers, tensile testing machines and microscopes. Then, point cloud data of terminals and wires are obtained using three-dimensional laser scanning or industrial CT. Noise in the original point cloud data is removed based on the voxel downsampling criterion, and outliers in the denoised point cloud data are also removed. S1.2: Based on the preprocessed point cloud data, obtain the three-dimensional geometric dimensions and shapes of each terminal and wire, and establish a digital simulation model of each terminal and wire based on the three-dimensional geometric dimensions and shapes of each terminal and wire. S1.3: Using a surface roughness meter or resistivity meter, collect the surface roughness and oxide layer resistance characteristics of the contact surface between the terminal and the wire. Then, import the collected surface roughness and oxide layer resistance characteristics into the corresponding digital simulation model of the terminal and the wire to establish a complete terminal-wire simulation model. S1.4: Based on the main excitation frequency and the highest effective frequency component of the ultrasonic welding head in the simulation, calculate and set the simulation time. Then, according to the amplitude, frequency and phase angle of ultrasonic welding in the digital simulation, calculate the displacement of the welding head end face under the preset simulation time to obtain the corresponding ultrasonic vibration boundary. After that, according to the material properties of each terminal and wire, set the simulation boundary conditions of the corresponding terminal-wire simulation model. II. In the initial stage of terminal and wire clamping, the contact surface is dynamically scanned and a microscopic contact resistance distribution map is constructed to identify areas with high contact resistance. III. Real-time acquisition of multi-source sensor data during the welding process, followed by fusion analysis of the acquired multi-source sensor data, and dynamic generation of ultrasonic energy output curves; IV. Based on the contact resistance distribution and real-time sensing data, ultrasonic energy is directed to promote local plastic deformation, and the energy control strategy is optimized in real time. V. Perform spectrum analysis on the acoustic signals generated during the welding process, and determine in real time whether there are potential defects in the weld joint. At the same time, based on the determination result, automatically trigger the self-repair program. VI. Monitor the inherent modes of the welding head in real time, and dynamically adjust the driving parameters of the welding head based on the monitoring results. When the welding head shows signs of fatigue or wear, remind maintenance personnel to replace the welding head.

2. The ultrasonic welding method for automotive wiring harness terminals according to claim 1, characterized in that, The specific steps for predicting welding defects during the simulated welding process described in Step I are as follows: S2.1: Based on existing experience, apply normal compression and tangential relative slip conditions to the terminal-wire contact interface in each terminal-wire simulation model, and calculate the interfacial frictional heat flux generated by friction and adhesion-slip transformation based on the applied conditions. S2.2: Calculate the structural displacement, velocity and acceleration response under preset ultrasonic excitation, and then solve the transient heat conduction by superimposing the generated interfacial frictional heat flux and the volume heat source converted by material plastic dissipation to obtain the thermal-mechanical coupling state of the corresponding terminal-wire simulation model. S2.3: Based on the latest thermo-mechanical coupling state of the current simulation time step, update the material flow resistance and softening. Integrate the contact compaction and tangential work to calculate the interface energy density. If the interface energy density is lower than the preset process threshold, it is determined that the simulated welding has a weak welding trend under the current parameters. S2.4: Monitor the peak temperature of the interface and its neighborhood. If the peak temperature is higher than the preset upper limit, it indicates that there is metal splashing or overheating in the current welding process. Based on the evolution relationship of equivalent porosity fraction with plastic accumulation and rapid temperature rise, calculate the formation trend of unwelded - inclusion - pore connectivity in the current welding process. S2.5: The interface is combined with three types of indicators—insufficiency, overheating, and porosity—and normalized and weighted to generate a real-time defect risk score for the current welding simulation. At the same time, the welding process parameters are adjusted in real time based on the real-time risk score.

3. The ultrasonic welding method for automotive wiring harness terminals according to claim 2, characterized in that, The specific steps for identifying high contact resistance areas in step II are as follows: S3.1: The terminal-wire contact surface is scanned by a grid using current micro-excitation. Based on the scanning results, a set of two-dimensional coordinate scanning grids is established on the contact surface. At the same time, the excitation time window and weight of each grid are set. Then, the measurement signal of each grid is synchronously integrated by phase-locked loop to obtain the coherent potential value corresponding to each grid. S3.2: Using a pre-set transfer matrix, the measured coherent potential values ​​are corrected by linear transformation, and a preset current perturbation is applied to each grid. The corresponding potential changes are read, and the obtained change values ​​are used as local incremental resistances. Then, the obtained local incremental resistances are convolved and corrected. S3.3: The corrected local incremental resistances are inverted using the Tikhonov inversion method, and the inversion results are normalized using robust statistics to obtain the relative resistance anomaly of each grid. Then, the resistance distribution of each grid is calculated, and the corresponding high resistance judgment threshold is set. S3.4: If the relative resistance anomaly is higher than the high resistance judgment threshold, the corresponding grid is divided into the high contact resistance initial screening set. Then, through area opening operation and connected component screening, each group of grids is processed into multiple groups of connected grids. Based on the preset retention area threshold, connected grids higher than the retention area threshold are retained to obtain the high contact resistance region in the micro contact resistance distribution map.

4. The ultrasonic welding method for automotive wiring harness terminals according to claim 3, characterized in that, The specific steps for dynamically generating the ultrasonic energy output curve in step III are as follows: S4.1: Multiple sensors are arranged on the welding equipment, and the various sensors are synchronously collected at a fixed sampling frequency to obtain multimodal sensing data. Then, the sensing data of different modes are preprocessed through various operations such as noise reduction, smoothing and scale normalization. After that, the continuously collected multimodal sensing data is divided into multiple sets of sensing datasets using a sliding window. S4.2: Each set of sensor datasets is input into a one-dimensional convolutional network, and the data in each set of sensor datasets is convolved by the convolution kernels of each convolutional layer in the one-dimensional convolutional network to extract the local feature values ​​of each sensor data. S4.3: The extracted local feature values ​​are transmitted to the LSTM network, and the dependency relationship between the local feature values ​​is identified through the gating mechanism of the LSTM network. At the same time, based on the identified dependency relationship, the dynamic evolution data of the welding process is acquired in real time. Then, the current dynamic evolution data is analyzed in real time through the Softmax classifier, and the welding physical state of the corresponding sensor dataset in each sliding window is determined. S4.4: The identified welding physical state is compared with the expected process curve in real time, and based on the comparison results, a real-time control signal is generated, and the ultrasonic energy output curve corresponding to each control signal in each time period is output.

5. The ultrasonic welding method for automotive wiring harness terminals according to claim 4, characterized in that, The specific steps for promoting localized plastic deformation described in step IV are as follows: S5.1: Collect the current contact resistance distribution map and the real-time data collected by various types of sensors. Then, fuse the contact resistance distribution map and the sensor data field to construct the corresponding joint state field, so as to establish a hot zone-impedance composite distribution model for the current welding interface. S5.2: Divide the generated weld points into multiple meshes and calculate the joint state field intensity of each mesh element. If the joint state field intensity is higher than the preset threshold, mark the corresponding mesh element as the priority compensation area. Then, based on the marking results, calculate the ultrasonic energy weight of the ultrasonic head for each mesh element. S5.3: After the ultrasonic energy weight calculation is completed, the ultrasonic energy is concentrated and loaded into the corresponding high resistance region according to the spatial distribution results of each ultrasonic energy weight by the modulation controller. Then, the ultrasonic energy is input into the material interface in the form of high frequency shear stress and friction work, and the output ultrasonic energy causes micro-cracks to occur and break the oxide film on the surface of the weld joint. S5.4: After the oxide film is broken, the ultrasonic energy continues to apply force to the corresponding weld surface, causing the contact area of ​​the weld surface to undergo plastic deformation under the action of ultrasonic energy, while filling the micropores generated during the welding process.

Images