Search

CN-121981948-A - Quality detection method and system for wire harness terminal and wire harness after welding

CN121981948ACN 121981948 ACN121981948 ACN 121981948ACN-121981948-A

Abstract

The invention provides a quality detection method and a system for a wire harness terminal and a wire harness after welding, which are characterized in that color images and near infrared images of welding spots and surrounding areas are acquired through multi-source image acquisition, thermal physical characteristics and geometric characteristics are extracted, a thermal model is established based on a Fourier thermal conduction equation, welding thermal input parameters are inverted through a particle swarm optimization algorithm, process anomalies are judged, multi-dimensional characteristics are fused and then input into a trained variational self-encoder, potential defects are identified by utilizing a Markov distance.

Inventors

  • WU YINLONG

Assignees

  • 江西省龙邦电器有限公司

Dates

Publication Date
20260505
Application Date
20251219

Claims (10)

  1. 1. A method for detecting quality of a wire harness terminal after welding with a wire harness, the method comprising: synchronously acquiring a color image and a near infrared image of a welding area of a wire harness terminal through a double camera, and realizing pixel level alignment of the two images by using a SIFT feature matching algorithm; extracting color and oxidation characteristics, heat affected zone characteristics and infrared intensity characteristics from the aligned images to obtain thermophysical characteristics; establishing a simplified thermal model based on a Fourier thermal conduction equation, inverting welding heat input parameters through finite element discrete solving and a particle swarm optimization algorithm, calculating the relative deviation between inversion parameters and actual process parameters, and marking abnormal thermal process; Extracting geometric features of welding spots, carrying out standardized treatment on the thermophysical features, the geometric features, the relative deviation and the thermal process abnormal marks, and then splicing to form fusion feature vectors; and training the variation self-encoder by adopting the qualified samples, calculating the mahalanobis distance of the fusion feature vector of the new samples in the hidden space of the variation self-encoder, and identifying potential defects and high-confidence defects according to marked thermal process anomalies.
  2. 2. The method for detecting the quality of a welded wire harness terminal and the wire harness according to claim 1, wherein in the step of extracting color and oxidation characteristics, heat affected zone characteristics and infrared intensity characteristics from the aligned images to obtain thermophysical characteristics, converting the color image into an HSV color space, extracting a hue mean value, a saturation gradient and a brightness variance of a welding spot area, quantifying oxidation color distribution, determining oxidation coverage according to an oxidation area pixel area and a welding spot total pixel area, and obtaining the color and oxidation characteristics; Dividing the boundary of the heat affected zone by adopting a Canny edge detection algorithm, and calculating the width of the heat affected zone and the profile irregularity of the heat affected zone to obtain the characteristics of the heat affected zone; And extracting the infrared intensity difference between the center of the welding spot and the edge of the heat affected zone to obtain the infrared intensity characteristic.
  3. 3. The method for detecting the quality of a welded wire harness terminal and the wire harness according to claim 2, wherein the simplified thermal model is established based on a fourier heat conduction equation, the welding heat input parameters are inverted by a finite element discrete solving and particle swarm optimization algorithm, the relative deviation between the inversion parameters and the actual process parameters is calculated, and in the step of marking the abnormal thermal process, the simplified thermal model is expressed as: ; Where k is the thermal conductivity of the material, ρ is the density of the material, c is the specific heat capacity of the material, Is the temperature of the (x, y) point at the time t, Is the welding heat input density.
  4. 4. The method for detecting the quality of a welded wire harness terminal and wire harness according to claim 3, wherein the step of establishing a simplified thermal model based on a fourier heat conduction equation, inverting welding heat input parameters through a finite element discrete solving and particle swarm optimization algorithm, calculating relative deviation of inversion parameters and actual process parameters, and marking abnormal thermal process comprises the steps of: Based on a Fourier heat conduction equation, the simplified thermal model is established by considering the material heat conductivity, specific heat capacity and density of the wire harness terminal and the soldering tin; Dispersing a welding area by adopting a triangular grid, and converting the simplified thermal model into a linear equation set; constructing an objective function by taking the thermophysical characteristics as constraint targets; minimizing the objective function by adopting a particle swarm optimization algorithm, and inverting optimal heat input parameters, wherein the heat input parameters comprise peak heat input density and heat input duration; Calculating the relative deviation between the inverted heat input parameter and the actual process parameter, and judging whether the relative deviation is larger than a threshold value or not; If yes, the thermal process is marked as abnormal.
  5. 5. The method for detecting the quality of a welded wire harness terminal and wire harness according to claim 4, wherein the geometric features include an area, roundness of a solder joint extracted from a color image, and a wire harness insertion depth.
  6. 6. The method for detecting the quality of the welded wire harness terminal and the wire harness according to claim 5, wherein the structure of the variation self-encoder comprises an encoder and a decoder, the encoder and the decoder are both composed of a 3-layer full-connection network, and the activation functions are ReLU and Sigmoid respectively; The training loss function is expressed as: ; ; ; Wherein L is a training loss function, In order to reconstruct the loss of the device, For KL divergence, beta is the trade-off coefficient, N is the number of samples, For the fused feature vector of the i-th sample, A reconstructed fusion feature vector for the ith sample, the reconstructed fusion feature vector being a fusion feature vector that is generated in reverse direction by a decoder of the variable self-encoder based on the hidden vector output by the encoder, Is the square of the euclidean norms, The sign of the calculation for the KL divergence, To vary the distribution of hidden vectors output from the encoder, Is an a priori distribution of hidden vectors.
  7. 7. The method for detecting the quality of a welded wire harness terminal and wire harness according to claim 6, wherein the training of the variable self-encoder by using the qualified samples, calculating the mahalanobis distance of the fusion feature vector of the new sample in the variable self-encoder hidden space, and identifying the potential defect and the high confidence defect according to the marked thermal process abnormality, wherein the mahalanobis distance of the fusion feature vector of the new sample in the variable self-encoder hidden space is expressed as: ; ; Wherein, the The hidden vector obtained after the variation from the encoder is input for the fusion feature vector of the new sample, As the mean value of the hidden vectors, For covariance matrix, M is the number of hidden vectors, z o is the o-th hidden vector, and T is the transpose symbol.
  8. 8. A system for detecting the quality of a welded wire harness terminal and wire harness, for realizing the method for detecting the quality of a welded wire harness terminal and wire harness according to any one of claims 1 to 7, comprising: The alignment module is used for synchronously collecting a color image and a near infrared image of a welding area of the wire harness terminal through a double camera and realizing pixel level alignment of the two images by using a SIFT feature matching algorithm; the extraction module is used for extracting color and oxidation characteristics, heat affected zone characteristics and infrared intensity characteristics from the aligned images to obtain thermophysical characteristics; the first calculation module is used for establishing a simplified thermal model based on a Fourier thermal conduction equation, inverting welding heat input parameters through finite element discrete solution and a particle swarm optimization algorithm, calculating the relative deviation between inversion parameters and actual process parameters, and marking abnormal thermal process; The splicing module is used for extracting geometric features of welding spots, carrying out standardized processing on the thermophysical features, the geometric features, the relative deviation and the thermal process abnormal marks, and then splicing to form fusion feature vectors; And the second calculation module is used for training the variation self-encoder by adopting the qualified samples, calculating the mahalanobis distance of the fusion feature vector of the new sample in the hidden space of the variation self-encoder, and identifying potential defects and high-confidence defects according to marked thermal process anomalies.
  9. 9. A computer-readable storage medium, comprising: the readable storage medium stores one or more programs which, when executed by a processor, implement the method for quality detection after welding of a wire harness terminal to a wire harness according to any one of claims 1 to 7.
  10. 10. An electronic device comprising a memory and a processor, wherein: The memory is used for storing a computer program; the processor is configured to implement the method for detecting quality of a welded wire harness terminal and wire harness according to any one of claims 1 to 7 when executing the computer program stored in the memory.

Description

Quality detection method and system for wire harness terminal and wire harness after welding Technical Field The invention belongs to the technical field of quality detection, and particularly relates to a quality detection method and system for a wire harness terminal and a wire harness after welding. Background Wire harness terminal to wire harness welding is a process of permanently joining together metal terminals (typically joints made of copper or copper alloy) and wires (wire harnesses) by means of molten solder (e.g., tin alloy). The process utilizes a high temperature heat source (such as an electric soldering iron, laser or resistance welding) to melt the solder, so that the solder wets and fills the terminal and the metal surface of the lead, a stable and low-resistance electrical connection point is formed after cooling, and good mechanical strength and long-term reliability are ensured, for example, the welding of a wire harness terminal and a wire harness exists in a mobile phone charging wire. The welding quality of the wire harness terminal and the wire harness directly affects the reliability and safety of electrical equipment, and the traditional welding quality detection method mainly relies on manual visual detection or single 2D image recognition technology and has the following technical defects: 1. the hidden defect identification capability is insufficient, namely, the manual detection and the traditional 2D image identification can only judge the dominant defects such as the outline size of a welding spot and the like, and the hidden defects such as the false welding, the internal looseness and the like caused by insufficient heat input are difficult to identify; 2. The existing detection method based on supervised learning needs to collect a large number of marked defect samples for model training, and the defect samples are scarce in actual production, so that the model generalization capability is poor; 3. the traceability of the welding process is lacking, namely the quality is judged only by the static image after welding, the heat input parameters in the welding process cannot be correlated, and the technological root of the defect generation is difficult to position; 4. The error judgment rate is high, namely the defect judgment is carried out by singly depending on geometric features or data statistics rules, and the error judgment is easily caused by environmental interference due to the lack of physical mechanism support. Disclosure of Invention Based on the above, the embodiment of the invention provides a method and a system for detecting the quality of a wire harness terminal after welding the wire harness terminal, which aim to solve the technical problems of difficult hidden defect identification, dependence on defect sample labeling, lack of heat process traceability and high misjudgment rate in the prior art. A first aspect of an embodiment of the present invention provides a method for detecting quality after welding a wire harness terminal and a wire harness, the method including: synchronously acquiring a color image and a near infrared image of a welding area of a wire harness terminal through a double camera, and realizing pixel level alignment of the two images by using a SIFT feature matching algorithm; extracting color and oxidation characteristics, heat affected zone characteristics and infrared intensity characteristics from the aligned images to obtain thermophysical characteristics; establishing a simplified thermal model based on a Fourier thermal conduction equation, inverting welding heat input parameters through finite element discrete solving and a particle swarm optimization algorithm, calculating the relative deviation between inversion parameters and actual process parameters, and marking abnormal thermal process; Extracting geometric features of welding spots, carrying out standardized treatment on the thermophysical features, the geometric features, the relative deviation and the thermal process abnormal marks, and then splicing to form fusion feature vectors; and training the variation self-encoder by adopting the qualified samples, calculating the mahalanobis distance of the fusion feature vector of the new samples in the hidden space of the variation self-encoder, and identifying potential defects and high-confidence defects according to marked thermal process anomalies. Further, in the step of extracting color and oxidation characteristics, heat affected zone characteristics and infrared intensity characteristics from the aligned images to obtain thermophysical characteristics, converting a color image into an HSV color space, extracting a hue mean value, a saturation gradient and a brightness variance of a welding spot area, quantifying oxidation color distribution, and determining oxidation coverage according to an oxidation area pixel area and a welding spot total pixel area to obtain color and oxidation characteristics; Dividing the bou