Enameled rectangular wire film thickness measuring system
Through the combination of visual neural network and laser ranging, the problems of insufficient accuracy and insufficient internal void detection of traditional measurement methods in complex shapes and irregular surfaces are solved, and the automatic and accurate measurement of the thickness of enameled flat wire film is achieved. Intelligent detection.
Patent Information
- Application Number
- CN202411505642.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-28
- Publication Date
- 2024-11-26
AI Technical Summary
The traditional enameled flat wire film thickness measurement method is low in degree, making it difficult to accurately measure enameled flat wire with complex geometric shapes and irregular surfaces, and cannot effectively detect internal film thickness gaps.
The visual neural network is used to combine laser ranging and current voltage detection to identify surface defects through the deep residual network and generate test instructions, and combine laser ranging data and electrical characteristics to verify the film thickness to achieve automated and accurate measurement.
It improves the accuracy and reliability of film thickness measurement, can detect internal film thickness abnormalities in time, avoid invalid measurements caused by surface defects, and enhances the intelligence and accuracy of measurement.
Smart Images

Figure CN119022811A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of enameled flat wire film thickness measurement, in particular to an enameled flat wire film thickness measurement system. Background Art
[0002] Enameled flat wire is widely used in modern motors, transformers and other electromagnetic equipment. Its main function is to form an insulating layer on the surface of the wire to prevent short circuits between coils. With the efficiency and miniaturization of industrial equipment, the performance requirements for enameled flat wire are increasing. The uniformity and quality of the enameled film thickness are directly related to the stability and service life of the product. Therefore, how to accurately measure the film thickness of enameled flat wire has become a key technical issue in the industry.
[0003] In traditional preset scale measurement devices, the degree of measurement automation is low. When the enameled flat wire has bends or surface defects that affect the measurement results, the measurement will still be performed, resulting in a decrease in the credibility of the measurement data. The measurement accuracy depends on the resolution of the equipment and the fineness of the scale. These systems may have measurement deviations when faced with enameled flat wires with complex geometric shapes or irregular surfaces such as depressions or protrusions. Most current measurement methods rely only on physical dimensional parameters, while ignoring the relationship between electrical properties and film thickness. As a result, when there is a vacancy problem inside the enameled flat wire film, it is difficult to detect the abnormal vacancy in the internal film thickness through the surface thickness of the enameled flat wire. Summary of the invention
[0004] In order to solve the above problems, the present invention provides an enameled flat wire film thickness measurement system, which performs automated testing through visual neural networks and fixed signals, performs laser ranging data processing on the enameled flat wire, and realizes automated and accurate measurement of the enameled flat wire film thickness in combination with electrical characteristic verification of the enameled flat wire film.
[0005] To achieve the above object, the technical solution adopted by the present invention is: An enameled flat wire film thickness measurement system comprises: an enameled flat wire fixing module, a laser ranging module, a current and voltage detection module, a visual monitoring module and a film thickness measurement module, wherein the enameled flat wire fixing module, the laser ranging module, the current and voltage detection module and the visual monitoring module are respectively connected to the film thickness measurement module; The enameled flat wire fixing module is used to fix the two ends of the preset measuring length of the enameled flat wire and send the fixing information to the film thickness measuring module; The laser distance measuring module comprises two laser distance measuring devices, which are respectively arranged at spatial positions at preset distances from two sides of the enameled flat wire. The laser distance measuring module is used to obtain distance measurement data of the two sides of the fixed enameled flat wire relative to the two laser distance measuring devices, and the distance measurement data is a distance matrix; The current and voltage detection module is used to connect the two sides of the enameled flat wire to obtain the current and voltage values of the circuit; The visual monitoring module is used to determine whether the enameled flat wire has bending and surface defects based on the deep residual network, and generate a test instruction according to whether there are bending and surface defects; The film thickness measurement module is used to trigger the laser ranging module to obtain ranging data according to the fixed information and test instructions, calculate the film thickness of the enameled flat wire according to the ranging data of the two sides of the fixed enameled flat wire relative to the two laser ranging devices, the inner wire thickness and the actual distance between the two laser ranging devices, and verify the film thickness of the enameled flat wire according to the current and voltage values of the path.
[0006] Furthermore, the training steps of the deep residual network include: Collect enameled flat wire image data at different angles and lighting conditions; Performing data preprocessing on the image data, including image normalization and denoising; The enameled flat wire image data after data preprocessing is annotated, and the data annotation results include normal, curved, and surface defects; Build a deep residual network model containing several residual blocks, use the labeled image data to train the model through back propagation and adjust the number and depth of residual blocks; The deep residual network model is tested and validated using the test dataset.
[0007] Furthermore, the formula of the deep residual network model is as follows: ; Among them, y is the final mapping; Image data to complete data annotation; is the learnable parameter in the residual mapping; is the residual mapping function; is the weight matrix; is a non-linear activation function.
[0008] Further, generating a test instruction according to whether there are bends and surface defects includes: If the enameled flat wire has bends or surface defects, no test instructions will be generated; If the enameled flat wire has no bends or surface defects, a test instruction is generated.
[0009] Furthermore, the method of calculating the film thickness of the enameled flat wire according to the distance measurement data of the two sides of the fixed enameled flat wire relative to the two laser distance measuring devices, the inner wire thickness and the actual distance between the two laser distance measuring devices comprises the following steps: Filter out noise and outliers from the distance matrix; Based on the gradient edge detection algorithm, the boundary points of the enameled flat wire in the distance matrix are identified; Calculate the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device according to the boundary points; Based on the preset distance between two laser distance measuring devices, the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device and the inner wire thickness are screened and calculated to obtain the film thickness of the enameled flat wire.
[0010] Furthermore, filtering out noise and outliers from the distance matrix comprises the following steps: Apply Gaussian filtering algorithm to the obtained distance matrix to eliminate random noise; The data points that deviate from the trend of surrounding data are marked as outliers and removed through the threshold detection algorithm; The linear interpolation algorithm is used to fill in the missing positions in the distance matrix after removing outliers.
[0011] Furthermore, the calculation formula for the thickness of the enameled flat wire film is as follows: ; in, is the film thickness of the enameled flat wire; and They are respectively the vertical distances of the two sides of the enameled flat wire relative to the laser distance measuring device; The preset distance between two laser distance measuring devices; It is the thickness of the inner wire of the enameled flat wire.
[0012] Furthermore, the verification of the film thickness of the enameled flat wire according to the current and voltage values of the path includes the following steps: According to the acquired current and voltage data, the resistance value of the enameled flat wire at the current position is calculated; Calculate the film thickness value range of the resistance value of the enameled flat wire at the current position based on the resistance values of the enameled flat wires of several thicknesses as reference values; If the film thickness of the enameled flat wire is within the film thickness value range of the resistance value of the enameled flat wire at the current position, the film thickness verification of the enameled flat wire passes; If the film thickness of the enameled flat wire does not fall within the film thickness value range of the resistance value of the enameled flat wire at the current position, the enameled flat wire film thickness verification fails.
[0013] Furthermore, it also includes a thickness screening module, which is used to determine whether the current enameled flat wire film thickness is qualified according to a preset qualified thickness range.
[0014] The beneficial effect of the present invention is that the present invention fixes the two ends of the preset measurement length of the enameled flat wire, and sets two laser distance measuring devices on both sides of the enameled flat wire to obtain distance matrix data. Compared with the traditional preset scale measurement device, the laser distance measurement can accurately obtain the surface geometric features of the enameled flat wire and reduce the measurement deviation caused by surface irregularities such as depressions or protrusions. At the same time, after the distance matrix is filtered and optimized by the boundary detection algorithm, the smoothness and accuracy of the data are ensured, and the measurement accuracy of the film thickness is greatly improved. Then, by combining the current and voltage detection module, the electrical properties of the enameled flat wire are monitored in real time. By analyzing the relationship between resistance and film thickness, the accuracy of physical measurement can be further verified, especially when there is a vacancy problem inside the enameled film, the electrical properties can reflect this abnormality in time, thereby making up for the shortcomings of traditional methods in internal defect detection. Through the visual monitoring module, this scheme uses a deep residual network to monitor the surface of the enameled flat wire in real time to detect the bending and surface defects of the enameled flat wire. This module can automatically identify surface anomalies of enameled flat wires and generate test instructions based on the detection results, realizing an intelligent measurement process and effectively avoiding invalid measurements caused by bending or surface defects. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 It is a structural schematic diagram of a film thickness measurement system for enameled flat wire in the present invention.
[0016] Figure 2 It is a flowchart of the steps of training the deep residual network in the present invention. DETAILED DESCRIPTION
[0017] See also Figure 1-Figure 2 As shown, the present invention relates to an enameled flat wire film thickness measurement system, comprising: an enameled flat wire fixing module, a laser ranging module, a current and voltage detection module, a visual monitoring module and a film thickness measurement module, wherein the enameled flat wire fixing module, the laser ranging module, the current and voltage detection module and the visual monitoring module are respectively connected to the film thickness measurement module; The enameled flat wire fixing module is used to fix the two ends of the preset measuring length of the enameled flat wire and send the fixing information to the film thickness measuring module; The laser distance measuring module comprises two laser distance measuring devices, which are respectively arranged at spatial positions at preset distances from two sides of the enameled flat wire. The laser distance measuring module is used to obtain distance measurement data of the two sides of the fixed enameled flat wire relative to the two laser distance measuring devices, and the distance measurement data is a distance matrix; The current and voltage detection module is used to connect the two sides of the enameled flat wire to obtain the current and voltage values of the circuit; The visual monitoring module is used to determine whether the enameled flat wire has bending and surface defects based on the deep residual network, and generate a test instruction according to whether there are bending and surface defects; The film thickness measurement module is used to trigger the laser ranging module to obtain ranging data according to the fixed information and test instructions, calculate the film thickness of the enameled flat wire according to the ranging data of the two sides of the fixed enameled flat wire relative to the two laser ranging devices, the inner wire thickness and the actual distance between the two laser ranging devices, and verify the film thickness of the enameled flat wire according to the current and voltage values of the path.
[0018] It should be noted that the enameled flat wire fixing module can adopt a fixing fixture combined with a pulley structure. In order to adapt to the movement requirements of the enameled flat wire during the measurement process, the pulley structure is introduced as an auxiliary fixing component, allowing the long enameled flat wire to slide and adjust the part to be measured without affecting its lateral positioning in the measurement area. The pulley system reduces friction through precision bearings and low-friction materials (such as Teflon coating) to ensure the stability of the flat wire when running on the pulley. The size and spacing of the pulley are customized according to the enameled flat wire of different specifications to ensure that the pulley system can adapt to flat wires of different sizes and provide consistent fixing effects. Combined with the design of the fixing fixture and the pulley structure, the enameled flat wire maintains a stable tension during the measurement process and ensures its position accuracy, thereby providing a reliable foundation for subsequent laser ranging and visual inspection. This structural design not only improves the adaptability of the measurement system, but also effectively reduces the impact of external interference on the measurement results, and improves the overall measurement accuracy and efficiency of the system. The fixture can be designed to be adjustable to accommodate enameled flat wires of different specifications. The fixing force is monitored in real time by a pressure sensor, and when the preset pressure threshold is reached, the fixing information is sent to the film thickness measurement module. The laser distance measurement module is one of the core measurement components in the system. Each laser distance measurement device is positioned on both sides of the enameled flat wire through a precise optical system, and the distance is adjustable to accommodate enameled flat wires of different thicknesses and shapes. The laser distance measurement device usually adopts high-precision laser interferometry technology, which can provide sub-micron distance measurement accuracy. Through the synchronous acquisition of distance matrix data by the laser distance measurement devices on both sides, the system can obtain the geometric shape information of the surface of the enameled flat wire in real time. The measurement data is filtered and optimized by the boundary detection algorithm to eliminate noise and abnormal data, ensuring the reliability and stability of the measurement results. During the detection process, the current and voltage detection module first connects the upper and lower sides of the enameled flat wire to the circuit through the fixture to ensure that the current can pass through the flat wire smoothly. The current sensor monitors the current changes passing through the flat wire in real time and transmits the current value to the film thickness measurement module for analysis. At the same time, the voltage detection unit collects the voltage data at both ends of the flat wire, and combines the data from the current sensor to calculate the resistance or impedance value of the enameled flat wire. The change in resistance value can directly reflect the change in the thickness of the enameled film, especially when there are voids or unevenness inside the enameled film, the electrical characteristics will be abnormal, and the current and voltage detection module can detect and report these abnormalities in time. The visual monitoring module is designed based on a deep residual network, and is mainly used to identify defects and deformations on the surface of the enameled flat wire. The surface of the enameled flat wire is imaged by a high-definition industrial camera, and the system uses a trained deep residual network model to process the image. The model can accurately detect the defect information of the enameled flat wire, such as bending, surface scratches, and dents, and generate corresponding test instructions. The introduction of this module realizes the intelligent monitoring of the surface state of the enameled flat wire and avoids invalid measurements that may occur in traditional detection methods.The film thickness measurement module triggers the laser distance measurement module to perform measurement only when it obtains the test instructions and fixed information generated by the visual monitoring module at the same time. At the same time, the module also integrates the distance matrix data obtained by the laser distance measurement and the impedance data of the current and voltage detection module for verification. Specifically, the module uses a preset calculation model, combined with the inner wire thickness of the enameled flat wire and the actual distance measurement data, to accurately calculate the film thickness, and compare and analyze the physical measurement results with the electrical verification results to ensure the accuracy of the final measurement.
[0019] Furthermore, the training steps of the deep residual network include: Collect enameled flat wire image data at different angles and lighting conditions; Performing data preprocessing on the image data, including image normalization and denoising; The enameled flat wire image data after data preprocessing is annotated, and the data annotation results include normal, curved, and surface defects; Build a deep residual network model containing several residual blocks, use the labeled image data to train the model through back propagation and adjust the number and depth of residual blocks; The deep residual network model is tested and validated using the test dataset.
[0020] It should be noted that, first of all, the data collection process not only covers the images of enameled flat wire under different lighting and angles, but also generates additional synthetic image data through the Generative Adversarial Network (GAN). These data further expand the diversity of the training set and ensure that the deep residual network can handle various complex image features. In the process of data collection, factors such as uneven lighting, reflection and angle changes that may occur in the enameled flat wire under actual production conditions are taken into consideration to improve the generalization ability of the network. In the data preprocessing stage, image normalization is the first operation. Normalization not only standardizes the pixel values to the range of [0,1] or [-1,1], but also performs color space correction for the imaging differences of different camera devices. This step eliminates the image deviation caused by the imaging device and ambient light through contrast stretching, histogram equalization and other technologies. In addition, denoising usually uses the non-local means (NLM) denoising algorithm or wavelet denoising technology to retain image details to the greatest extent while eliminating high-frequency noise interference. These preprocessing steps ensure that the data input to the deep residual network is of high quality and has clear features. Data annotation relies on dedicated annotation tools and rule sets. The annotated images not only need to classify the overall morphology of the enameled flat wire, but also need to perform fine-grained annotation of specific defect types, such as scratches, dents, and local expansion. During the annotation process, semi-automatic annotation tools are used. These tools automatically identify potential defect areas through preliminary edge detection and region growing algorithms to reduce manual intervention. The accuracy of annotation is further improved through cross validation and expert review to ensure that the annotation results of each data sample can truly reflect the actual state of the enameled flat wire. The construction of the deep residual network model is based on the ResNet architecture, and a residual block structure with 50 layers or more is selected. In each residual block, the skip connection technology is used to alleviate the gradient vanishing problem in the deep network. Inside the residual block, the kernel size of the convolution layer is adjusted through experiments. Generally, a combination of 3x3 convolution kernels and 1x1 convolution kernels is used to reduce the computational complexity while ensuring the depth of the model. In order to improve the expressiveness of the model, some layers also introduce an attention mechanism, which enables the model to automatically focus on key areas in the image, such as the edges of the enameled flat wire and possible defects. During the model training process, the back-propagation algorithm is combined with the adaptive momentum estimation (Adam) optimizer to speed up the convergence and prevent falling into the local optimal solution. During training, a dynamic learning rate adjustment strategy is used to dynamically adjust the learning rate according to the convergence of the model to avoid overfitting caused by too fast convergence. Throughout the training process, the generalization ability of the model is further improved by using batch normalization and L2 regularization techniques. In the model testing and verification stage, an independent test data set is used to evaluate the trained deep residual network.This stage not only focuses on classification accuracy, but also uses multiple indicators such as precision, recall, and F1 score to comprehensively evaluate the performance of the model in various surface defect detection tasks. In particular, confusion matrix analysis is used to identify the strengths and weaknesses of the model in various defect classification tasks, guiding the subsequent model tuning process.
[0021] Furthermore, the formula of the deep residual network model is as follows: ; Among them, y is the final mapping; Image data to complete data annotation; is the learnable parameter in the residual mapping; is the residual mapping function; is the weight matrix; is a non-linear activation function.
[0022] Specifically, this formula describes the core computation process in a deep residual network. Input It is the image data of the enameled flat wire that has been preprocessed and annotated. After normalization and denoising, this data is used as the initial input feature of the network, representing the original information of the surface of the enameled flat wire. It is the core component in the deep residual network, which represents the network's response to the input data The feature extraction process. Here is a set of learnable parameters that are gradually updated through the back-propagation algorithm. The residual map captures the detailed features in the image through a combination of convolutional layers and activation functions, and is particularly suitable for processing image data with high-dimensional spatial features, such as the surface texture or defects of enameled flat wire. Represents the input data The linear transformation operation is performed. This transformation is done through the weight matrix Acts on the input data to capture linear features. Linear transformation enables the network to extract important linear relationships from the input data, thereby better understanding the global information in the image. Nonlinear activation function You can choose ReLU (Rectified Linear Unit) or Leaky ReLU. The role of the activation function is to introduce nonlinear features and improve the network's expressiveness, so that the network can capture more complex patterns and features. This step is particularly critical because simple linear transformations cannot effectively handle the complex structure in the image, and nonlinear activation functions can model this complexity. The final output y is obtained by transforming the input With residual mapping The residual connection design enables the network to directly transmit part of the input data and make improvements on this basis. This structure can effectively solve the gradient vanishing problem in deep neural networks and ensure that the network can still be effectively trained when the depth increases.
[0023] Further, generating a test instruction according to whether there are bends and surface defects includes: If the enameled flat wire has bends or surface defects, no test instructions will be generated; If the enameled flat wire has no bends or surface defects, a test instruction is generated.
[0024] Specifically, if bending or surface defects are detected: the system will not generate further test instructions. This is because bending or surface defects may cause measurement errors or inaccurate film thickness calculations, so the system will directly block the next measurement step to avoid unreliable results due to unqualified enameled flat wire data. If no bending or surface defects are detected: when the visual monitoring module confirms that the surface of the enameled flat wire is intact and there is no bending, the system will generate a test instruction to trigger the subsequent laser ranging module to obtain ranging data and calculate film thickness. At this point, the entire system enters the normal measurement process to ensure the validity of the measurement data.
[0025] Furthermore, the method of calculating the film thickness of the enameled flat wire according to the distance measurement data of the two sides of the fixed enameled flat wire relative to the two laser distance measuring devices, the inner wire thickness and the actual distance between the two laser distance measuring devices comprises the following steps: Filter out noise and outliers from the distance matrix; Based on the gradient edge detection algorithm, the boundary points of the enameled flat wire in the distance matrix are identified; Calculate the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device according to the boundary points; Based on the preset distance between two laser distance measuring devices, the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device and the inner wire thickness are screened and calculated to obtain the film thickness of the enameled flat wire.
[0026] In some embodiments, the system obtains the distance matrix data of both sides of the enameled flat wire through two laser ranging devices. Since the laser ranging process may be affected by factors such as environmental noise and light interference, the original ranging data often contains noise and outliers. To ensure the accuracy of the measurement, the system first pre-processes the distance matrix and uses a filtering algorithm (such as Gaussian filtering or median filtering) to eliminate the noise. At the same time, a statistical method is used to detect outliers, and the ranging data that deviates from the overall trend is marked as abnormal and eliminated. The distance matrix after noise filtering and outlier elimination has higher smoothness and reliability, ensuring a solid data foundation for subsequent processing steps. Next, the system uses a gradient-based edge detection algorithm (such as Sobel operator or Canny edge detection) to analyze the processed distance matrix and identify the boundary points of the enameled flat wire. These boundary points correspond to the mutation positions in the laser ranging data, usually marking the boundary between the surface of the enameled flat wire and the air. Through edge detection, the system can accurately extract the outer contour of the enameled flat wire, providing an important reference for subsequent distance calculations. After identifying the boundary points, the system calculates the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device.
[0027] Furthermore, filtering out noise and outliers from the distance matrix comprises the following steps: Apply Gaussian filtering algorithm to the obtained distance matrix to eliminate random noise; The data points that deviate from the trend of surrounding data are marked as outliers and removed through the threshold detection algorithm; The linear interpolation algorithm is used to fill in the missing positions in the distance matrix after removing outliers.
[0028] In some embodiments, the system applies a Gaussian filtering algorithm to the acquired distance matrix to eliminate random noise. Gaussian filtering is a commonly used smoothing technique that smoothes out high-frequency noise components in an image or data by convolving the data with a Gaussian kernel, thereby retaining a relatively smooth signal portion. Specifically, the Gaussian filter calculates a new data value based on the weighted average of the points around each data point in the distance matrix, where points with closer distances have a greater impact on the current point. The standard deviation (σ) parameter of the Gaussian filter can be adjusted according to actual measurement conditions to control the degree of smoothing. A larger σ will result in a stronger smoothing effect, but may blur details; while a smaller σ retains more details, but the noise removal effect is weakened. Therefore, in this embodiment, the σ value is set through experiments and system optimization to ensure effective elimination of noise and retention of valid signals. After noise elimination is completed, the system uses a threshold detection algorithm to mark and remove outliers in the distance matrix. Threshold detection is based on statistical principles and determines whether a data point is an outlier by comparing the change trend of a data point with its surrounding data points. Specifically, the system calculates the difference between each data point and its neighboring data points. If the difference exceeds a preset threshold, the data point is marked as an outlier. The setting of the threshold needs to be combined with the actual distribution of the data. Usually, the optimal threshold that can effectively distinguish normal data from abnormal data is determined by analyzing multiple sample data. Outliers may be caused by sensor errors or environmental interference. Removing these outliers can significantly improve the overall data quality of the distance matrix. After removing outliers, data gaps may appear in the distance matrix. To prevent these gaps from affecting subsequent boundary detection and thickness calculation, the system fills these gaps through a linear interpolation algorithm. The linear interpolation algorithm generates fill-in data based on the data points before and after the vacant position, assuming that the data changes along a linear trend. Specifically, for each vacant position, the system finds its nearest valid data point, and calculates the interpolation value of the vacant position based on the value and position of these data points. The advantage of linear interpolation is that it has low computational complexity and can better preserve the overall trend of the data, and is suitable for data structures with strong continuity such as distance matrices.
[0029] Furthermore, the calculation formula for the thickness of the enameled flat wire film is as follows: ; in, is the film thickness of the enameled flat wire; and They are respectively the vertical distances of the two sides of the enameled flat wire relative to the laser distance measuring device; The preset distance between two laser distance measuring devices; It is the thickness of the inner wire of the enameled flat wire.
[0030] Specifically, first, the laser distance measuring device collects distance measurement data from both sides of the enameled flat wire. and , and then the fixed distance between the two and the laser device Combined, the total thickness of the enameled flat wire is calculated. In order to eliminate the interference of the inner wire thickness on the film thickness calculation, the inner wire thickness of the enameled flat wire is further subtracted from the formula , thus obtaining the actual enamel film thickness .
[0031] Furthermore, the verification of the film thickness of the enameled flat wire according to the current and voltage values of the path includes the following steps: According to the acquired current and voltage data, the resistance value of the enameled flat wire at the current position is calculated; Calculate the film thickness value range of the resistance value of the enameled flat wire at the current position based on the resistance values of the enameled flat wires of several thicknesses as reference values; If the film thickness of the enameled flat wire is within the film thickness value range of the resistance value of the enameled flat wire at the current position, the film thickness verification of the enameled flat wire passes; If the film thickness of the enameled flat wire does not fall within the film thickness value range of the resistance value of the enameled flat wire at the current position, the enameled flat wire film thickness verification fails.
[0032] In some embodiments, first, the system obtains the current and voltage data of the enameled flat wire path in real time through the current and voltage detection module. In order to ensure the accuracy of the data, the system uses a high-precision current sensor and a low-noise voltage collector. These sensors can capture the transient current and voltage changes in the enameled flat wire path within microseconds. Based on these real-time collected data, the system calculates the resistance value of the enameled flat wire at the current position, and the calculation of the resistance value follows Ohm's law. Next, the system compares the calculated resistance value with the preset reference resistance interval. These reference values are based on the standard resistance value range measured in multiple experiments of enameled flat wires of different thicknesses. Through the accumulation of experimental data, the system can establish a correlation model between resistance value and film thickness for each enameled flat wire of different thicknesses. This model takes into account the conductivity of the enameled material, the size of the inner wire, and the influence of the external environment on the resistance. Based on these data, the system can calculate the corresponding film thickness value interval for each specific resistance value. When the film thickness value interval corresponding to the resistance value obtained by the system is consistent with the actual physically measured film thickness, the system determines that the film thickness verification of the enameled flat wire has passed. At this point, the physical measurement and electrical verification complement each other to ensure the accuracy and consistency of the film thickness. On the contrary, if the physically measured film thickness exceeds the film thickness range corresponding to the resistance value, the system determines that the verification has failed and may trigger an alarm or further troubleshooting procedures. This situation may indicate the presence of voids in the enamel film, uneven insulation, or anomalies in the measurement process.
[0033] Furthermore, it also includes a thickness screening module, which is used to determine whether the current enameled flat wire film thickness is qualified according to a preset qualified thickness range.
[0034] Specifically, after the confirmed film thickness data enters the thickness screening module, the system first determines whether the currently measured film thickness value falls within the preset qualified range. If the film thickness is within the range, the system will determine that the current enameled flat wire is qualified and allow it to enter the next production or inspection link. At the same time, the system will record the qualified data and generate a quality report for subsequent analysis. If the film thickness exceeds the preset range, whether the thickness is too thick or too thin, the system will determine that the enameled flat wire is unqualified. At this time, the thickness screening module will trigger the alarm mechanism to notify the operator to conduct further inspections or adjustments. In addition, the system will record the unqualified measurement data in detail, including the specific value of the film thickness deviation, the corresponding product number, production time and other information. These data will be used for product traceability and quality control analysis to help determine the root cause of the problem and ensure that deviations in the production process are corrected in a timely manner.
[0035] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.
Claims
1. An enameled flat wire film thickness measurement system, characterized in that: include: An enameled flat wire fixing module, a laser ranging module, a current and voltage detection module, a visual monitoring module and a film thickness measurement module, wherein the enameled flat wire fixing module, the laser ranging module, the current and voltage detection module and the visual monitoring module are respectively connected to the film thickness measurement module; The enameled flat wire fixing module is used to fix the two ends of the preset measuring length of the enameled flat wire and send the fixing information to the film thickness measuring module; The laser distance measuring module comprises two laser distance measuring devices, which are respectively arranged at spatial positions at preset distances from two sides of the enameled flat wire. The laser distance measuring module is used to obtain distance measurement data of the two sides of the fixed enameled flat wire relative to the two laser distance measuring devices, and the distance measurement data is a distance matrix; The current and voltage detection module is used to connect the two sides of the enameled flat wire to obtain the current and voltage values of the circuit; The visual monitoring module is used to determine whether the enameled flat wire has bending and surface defects based on the deep residual network, and generate a test instruction according to whether there are bending and surface defects; The film thickness measurement module is used to trigger the laser ranging module to obtain ranging data according to the fixed information and test instructions, calculate the film thickness of the enameled flat wire according to the ranging data of the two sides of the fixed enameled flat wire relative to the two laser ranging devices, the inner wire thickness and the actual distance between the two laser ranging devices, and verify the film thickness of the enameled flat wire according to the current and voltage values of the path.
2. The enameled flat wire film thickness measurement system according to claim 1, characterized in that: The training steps of the deep residual network include: Collect enameled flat wire image data at different angles and lighting conditions; Performing data preprocessing on the image data, including image normalization and denoising; The enameled flat wire image data after data preprocessing is annotated, and the data annotation results include normal, curved, and surface defects; Build a deep residual network model containing several residual blocks, use the labeled image data to train the model through back propagation and adjust the number and depth of residual blocks; The deep residual network model is tested and validated using the test dataset.
3. The enameled flat wire film thickness measurement system according to claim 2, characterized in that: The formula of the deep residual network model is as follows: ; Among them, y is the final mapping; Image data to complete data annotation; is the learnable parameter in the residual mapping; is the residual mapping function; is the weight matrix; is a non-linear activation function.
4. The enameled flat wire film thickness measurement system according to claim 1, characterized in that: The generating of the test instruction according to whether there is bending and surface defects comprises: If the enameled flat wire has bends or surface defects, no test instructions will be generated; If the enameled flat wire has no bends or surface defects, a test instruction is generated.
5. The enameled flat wire film thickness measurement system according to claim 1, characterized in that: The method of calculating the film thickness of the enameled flat wire according to the distance measurement data of the two sides of the fixed enameled flat wire relative to the two laser distance measuring devices, the inner wire thickness and the actual distance between the two laser distance measuring devices comprises the following steps: Filter out noise and outliers from the distance matrix; Based on the gradient edge detection algorithm, the boundary points of the enameled flat wire in the distance matrix are identified; Calculate the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device according to the boundary points; Based on the preset distance between two laser distance measuring devices, the vertical distance data of the two sides of the enameled flat wire relative to the laser distance measuring device and the inner wire thickness are screened and calculated to obtain the film thickness of the enameled flat wire.
6. The enameled flat wire film thickness measurement system according to claim 5, characterized in that: The filtering of noise and outliers from the distance matrix comprises the following steps: Apply Gaussian filtering algorithm to the obtained distance matrix to eliminate random noise; The data points that deviate from the trend of surrounding data are marked as outliers and removed through the threshold detection algorithm; The linear interpolation algorithm is used to fill in the missing positions in the distance matrix after removing outliers.
7. The enameled flat wire film thickness measurement system according to claim 5, characterized in that: The calculation formula for the thickness of the enameled flat wire film is as follows: ; in, is the film thickness of the enameled flat wire; and They are respectively the vertical distances of the two sides of the enameled flat wire relative to the laser distance measuring device; The preset distance between two laser distance measuring devices; It is the thickness of the inner wire of the enameled flat wire.
8. The enameled flat wire film thickness measurement system according to claim 5, characterized in that: The verification of the film thickness of the enameled flat wire according to the current and voltage values of the path comprises the following steps: According to the acquired current and voltage data, the resistance value of the enameled flat wire at the current position is calculated; Calculate the film thickness value range of the resistance value of the enameled flat wire at the current position based on the resistance values of the enameled flat wires of several thicknesses as reference values; If the film thickness of the enameled flat wire is within the film thickness value range of the resistance value of the enameled flat wire at the current position, the film thickness verification of the enameled flat wire passes; If the film thickness of the enameled flat wire does not fall within the film thickness value range of the resistance value of the enameled flat wire at the current position, the enameled flat wire film thickness verification fails.
9. The enameled flat wire film thickness measurement system according to claim 1, characterized in that: It also includes a thickness screening module, which is used to determine whether the current enameled flat wire film thickness is qualified according to a preset qualified thickness range.
Citation Information
Patent Citations
Device for measuring film thickness of silicon slice
CN102183198A
Thickness detector for metalized film of capacitor
CN108827167A
Method for manufacturing resistance-adjustable insulating layer on the surface of wire
CN110718341A
Flat enameled electromagnetic wire surface defect detection method based on deep learning
CN111398291A
Automatic online film thickness measurement system for enameled flat wire
CN113624120A
Cited By
Aluminum enameled flat wire online defect detection and self-adaptive deviation correction method and system
CN120908258A
An online defect detection and self-adaptive deviation correction method and system for aluminum enameled flat wire
CN120908258B
Visual inspection system and method for carbon dioxide laser depainting of flat wire stator
CN121073873A
On-line thickness measuring method and system for alloy hot-rolled plate
CN121252708A
