Lower carving processing control method and system for mica groove

Through the combination of vision system and motion controller, the problems of insufficient positioning accuracy and robustness in mica groove engraving processing were solved, high-precision and high-quality mica groove processing was achieved, and the performance and production efficiency of motor products were improved.

CN120630860APending Publication Date: 2025-09-12NINGBO XIANGYAN TECH SERVICE DEV CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202510631760.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-05-16
Publication Date
2025-09-12

AI Technical Summary

Technical Problem

The existing mica groove engraving technology has shortcomings in positioning accuracy, robustness and adaptability to complex working conditions, and it is difficult to meet the high-precision and high-quality production needs.

Method used

A vision system is used to acquire rotor images. Through image preprocessing, template matching, convolutional neural networks, and feature fine-grained reconstruction technologies, the centerline position of the mica groove is identified and calculated. The motion controller is used to adjust the servo motor to achieve precise positioning and engraving.

Benefits of technology

It achieves more precise and robust mica groove positioning and engraving control, adapts to high-quality production needs under complex working conditions, and improves the quality and production efficiency of motor products.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120630860A_ABST
    Figure CN120630860A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of mica groove processing, and particularly discloses a lower carving processing control method and system for a mica groove, which are used for acquiring a rotor image acquired by a visual system and realizing accurate positioning and adjustment of the mica groove through a series of image processing steps. Firstly, an acquired rotor image is subjected to preprocessing such as filtering, contrast enhancement and geometric correction so as to reduce noise interference and highlight characteristics of a mica groove; next, candidate mica slot regions are identified using a template matching technique, and a confidence score for each candidate region is calculated to determine the most trusted target slot. Based on the position information of the target groove, the deviation of the target groove relative to the ideal machining position is calculated, deviation data is fed back to the motion controller, the servo motor is driven to conduct accurate adjustment, and after the center line of the target groove is aligned to the lower portion of the tool, the next carving program is executed. In this way, more accurate and robust mica groove positioning and undercutting control can be achieved, and the high-quality production requirement under the complex working condition is met.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of mica groove processing, and more specifically, to a method and system for controlling the engraving process of mica grooves. Background Art

[0002] The motor rotor is a core component of the electric motor. The mica slots in its commutator are crucial for ensuring proper commutation, improving operational stability, and extending the motor's service life. However, in actual manufacturing, due to process limitations or material variations, the mica slots in the motor rotor commutator may exhibit problems such as uneven indexing, varying widths, and irregular edges. Furthermore, after long-term operation, conductive powder, such as copper shavings, generated by friction between the brushes and the commutator can accumulate in the mica slots, degrading insulation and even causing inter-slot shorts, seriously impacting motor performance and safety. To address these issues and ensure commutator quality, the mica slots are typically milled (also known as slot cleaning) to remove debris, correct the slot shape, and ensure uniform slot width and depth, thereby improving the overall performance and reliability of the motor. Therefore, developing a precise and efficient mica slot milling control solution is crucial in automated production lines, as it directly impacts motor product quality and production efficiency.

[0003] At present, there are some positioning methods for mica groove engraving, such as traditional manual positioning of the dividing head, positioning of the automatic dividing mechanism based on a preset step, or auxiliary positioning using lasers, photoelectric sensors, etc. The manual positioning of the dividing head is inefficient and its accuracy is easily affected by human factors; the automatic positioning method based on the mechanical dividing mechanism, although it improves efficiency, its positioning accuracy often depends on the rotor's own (possibly deviated) physical structure or indexing accuracy, and it is difficult to completely eliminate the positioning errors caused by manufacturing or accumulated wear, especially for rotors with uneven indexing, it is difficult to achieve accurate on-demand processing; and the laser or photoelectric tracking positioning method, although with a high degree of automation, may be affected by complex working conditions such as burrs on the edge of the mica groove, contamination on the groove, uneven reflection, or fine scratches on the surface. It is difficult to ensure the accurate capture of the center line of each mica groove, which in turn affects the quality of engraving. These existing solutions still have room for improvement in positioning accuracy, robustness, and adaptability to complex working conditions, and it is difficult to fully meet the needs of high-precision, high-quality engraving.

[0004] Therefore, an optimized control scheme for the undercutting process of mica grooves is desired. Summary of the Invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a method and system for controlling the engraving of mica grooves, which can achieve more accurate and robust mica groove positioning and engraving control, and meet the high-quality production requirements under complex working conditions.

[0006] According to one aspect of the present application, a method for controlling the engraving of mica grooves is provided, comprising: acquiring a rotor image captured by a visual system; locating any identifiable mica groove from the rotor image, and using the mica groove as a current target groove; calculating the centerline position of the current target groove, and comparing the centerline position of the current target groove with a preset ideal processing position to obtain positioning deviation data; inputting the positioning deviation data into a motion controller, the motion controller being used to drive a servo motor to adjust based on the deviation data so as to align the centerline of the current target groove below the tool; and executing the engraving program for the current target groove after the centerline of the current target groove is aligned below the tool.

[0007] According to another aspect of the present application, a control system for engraving mica grooves is provided, comprising: a rotor image acquisition module for acquiring a rotor image captured by a visual system; a target groove positioning module for locating any identifiable mica groove from the rotor image, and using the mica groove as the current target groove; a centerline calculation and deviation detection module for calculating the centerline position of the current target groove, and comparing the centerline position of the current target groove with a preset ideal processing position to obtain positioning deviation data; a motion control module for inputting the positioning deviation data into a motion controller, and the motion controller for driving a servo motor to adjust based on the deviation data so as to align the centerline of the current target groove below the tool; and an engraving execution module for executing the engraving program for the current target groove after the centerline of the current target groove is aligned below the tool.

[0008] The present application provides a method and system for controlling the engraving of mica grooves, which obtains a rotor image collected by a visual system and realizes precise positioning and adjustment of the mica grooves through a series of image processing steps. First, the collected rotor image is pre-processed by filtering, contrast enhancement, and geometric correction to reduce noise interference and highlight the characteristics of the mica groove. Then, the template matching technology is used to identify the candidate mica groove areas, and the confidence score of each candidate area is calculated to determine the most credible target groove. Based on the position information of the target groove, its deviation from the ideal processing position is calculated, and this deviation data is fed back to the motion controller to drive the servo motor for precise adjustment so that the center line of the target groove is aligned with the bottom of the tool and then the engraving program is executed. In this way, more accurate and robust mica groove positioning and engraving control can be achieved to meet the high-quality production requirements under complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings of the embodiments of the present application. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0010] Figure 1 This is a schematic flow chart of a method for controlling the engraving process of a mica groove according to an embodiment of the present application.

[0011] Figure 2 This is a schematic flow chart of step S2 in the method for controlling the undercutting of mica grooves according to an embodiment of the present application.

[0012] Figure 3 This is a schematic flow chart of step S23 in the mica groove engraving control method according to an embodiment of the present application.

[0013] Figure 4 This is a schematic flowchart of step S232 in the mica groove engraving control method according to an embodiment of the present application.

[0014] Figure 5 This is a schematic diagram of the material transfer motor ejector and the rotating ejector at both ends pressing the rotor in an embodiment of the present application.

[0015] Figure 6 This is a schematic diagram of mica tank image acquisition according to an embodiment of the present application.

[0016] Figure 7 This is a schematic diagram of the engraved mica groove according to an embodiment of the present application.

[0017] Figure 8 This is a schematic block diagram of a control system for engraving mica grooves according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without making creative efforts also fall within the scope of protection of this application.

[0019] In order to solve the above technical problems, a method for controlling the engraving process of mica grooves is proposed in the technical solution of this application. Figure 1 Schematic flow chart of the method for controlling the engraving process of mica grooves according to an embodiment of the present application. Figure 1As shown, the engraving control method for mica grooves includes: S1, acquiring a rotor image captured by a visual system; S2, locating any identifiable mica groove from the rotor image, and using the mica groove as the current target groove; S3, calculating the centerline position of the current target groove, and comparing the centerline position of the current target groove with a preset ideal processing position to obtain positioning deviation data; S4, inputting the positioning deviation data into a motion controller, and the motion controller is used to drive a servo motor to adjust based on the deviation data so as to align the centerline of the current target groove below the tool; S5, after the centerline of the current target groove is aligned below the tool, executing the engraving program for the current target groove.

[0020] For example, in step S1, a rotor image captured by a visual system is obtained. It should be understood that during the production, assembly and subsequent operation of modern motor rotors, the surface may inevitably produce a certain degree of wear, dust accumulation, scratches or material residues, and these factors are likely to cause great interference to the accurate positioning and measurement of the slots in actual operation. In addition, the slight differences in the commutator structure and the microscopic changes in the shape, width and edge position of the slot area have a significant impact on the accuracy of subsequent engraving and the batch consistency of the product. Traditional manual division and reliance on scale jigs are not only time-consuming and labor-intensive, but also have a high misjudgment rate. By collecting images through a visual system, not only can fully automatic, high-resolution real-time observation be achieved, but also various details of the original slot morphology can be quantitatively reflected, providing a basis for subsequent algorithms to eliminate human errors and significantly improving the closed-loop control capabilities from detection and positioning to processing.

[0021] In one embodiment, a high-resolution industrial camera (such as an area or line scan CCD / CMOS camera) is installed vertically above the processing area, equipped with a custom-designed, highly uniform diffuse ring light source to avoid local reflections and light-dark distortion. Once the material is loaded, the device automatically focuses and simultaneously captures a bird's-eye view of the entire rotor, capturing all the mica groove surface conditions. Supporting software automatically extracts valid image content based on pre-defined ROI regions. All image data is cached in real time and fed into the subsequent groove segmentation and recognition process.

[0022] Exemplarily, in step S2, any identifiable mica groove is located from the rotor image, and the mica groove is used as the current target groove. Specifically, in order to achieve more accurate and robust mica groove positioning and engraving control, the present technical solution adopts a control method based on a visual system. During the visual processing process, it is not enough to find candidate mica groove areas through template matching alone, because the actual collected image may contain noise, interference, and non-target features similar to mica grooves, resulting in multiple candidate areas. In order to accurately identify the true mica groove that is most suitable for the current processing target from these candidate areas, it is necessary to perform a reliability assessment on each candidate area. Specifically, in the technical solution of the present application, by comparing the confidence scores of each candidate mica groove area, the most credible candidate area can be selected, thereby improving the accuracy and robustness of positioning. Furthermore, in order to improve the accuracy of confidence calculation, especially in the face of complex situations such as lighting changes, surface texture differences, and subtle deformation of the groove shape, using only conventional visual features (such as features extracted by standard convolutional coding) may not be sufficient to capture the key subtle information to distinguish between true and false mica grooves or evaluate the groove quality. Therefore, it is necessary to perform fine-grained reshaping of the visual features of each candidate mica groove region. This fine-grained reshaping technique aims to deeply optimize and enhance the initially extracted visual features, focusing more on the structural information and key details within the features, thereby generating more robust and discriminative visual enhancement features of the candidate mica groove regions. Inputting these enhanced visual features of the candidate mica groove regions into the classifier can help improve the accuracy of mica groove recognition and the reliability of confidence assessment, laying a solid foundation for subsequent calculation of positioning deviations and achieving high-quality engraving.

[0023] In one embodiment, Figure 2 As shown, locating any identifiable mica slot from the rotor image and taking the mica slot as the current target slot includes: S21, performing image preprocessing on the rotor image to obtain a preprocessed rotor image, wherein the preprocessing includes filtering, contrast enhancement and geometric correction; S22, sliding matching the mica slot template image in the preprocessed rotor image to obtain a set of candidate mica slot areas; S23, calculating the confidence score of each candidate mica slot area in the set of candidate mica slot areas as a mica slot to obtain a set of confidence scores; S24, taking the candidate mica slot area corresponding to the largest one in the set of confidence scores as the current target slot.

[0024] For example, in step S21, the rotor image is subjected to image preprocessing to obtain a preprocessed rotor image, wherein the preprocessing includes filtering, contrast enhancement, and geometric correction. That is, in an actual industrial environment, the motor rotor image directly acquired by the visual system is often not ideal and is subject to interference from various factors. Specifically, the surface of the rotor commutator may have defects from the manufacturing process (such as irregular edges), dirt accumulated after operation (such as copper chips), and fine scratches on the surface. At the same time, the original image may contain noise and interference. Therefore, the rotor image needs to be subjected to image preprocessing.

[0025] Furthermore, the lighting conditions during imaging may be uneven, or the relative position between the camera and rotor, or the lens itself, may introduce geometric distortion. These factors can severely impact the subsequent accurate identification and positioning of the mica slot. Therefore, the purpose of image preprocessing is to eliminate or mitigate the impact of these adverse factors, improve image quality, and lay the foundation for subsequent precise mica slot positioning. Specifically, filtering operations (such as mean filtering and median filtering) are designed to remove random noise (such as sensor noise or artifacts caused by fine dust and scratches) from the image and smooth the image. Contrast enhancement is used to highlight the difference between the mica slot (usually darker in color or made of a different material) and the surrounding commutator copper bars (usually brighter), making the outline and structure of the mica slot more clearly discernible, especially in poor lighting or uneven reflections. Geometric correction is used to correct image distortion caused by shooting angle, lens distortion, and other factors, restoring the true geometric proportions and positional relationships of objects in the image, and ensuring the accuracy of subsequent measurements.

[0026] For example, in step S22, the mica groove template image is slidably matched in the pre-processed rotor image to obtain a set of candidate mica groove regions. It should be understood that after the rotor image is pre-processed, such as filtering, contrast enhancement, and geometric correction, the image quality is improved, laying the foundation for subsequent mica groove identification. However, simply completing the pre-processing does not mean that the mica groove has been located. In other words, considering that the quality of the pre-processed image is improved, it is still a complete rotor image containing a large number of non-target areas. Directly performing complex feature analysis and positioning calculations on the entire image is inefficient. Therefore, a fast and effective method is needed to preliminarily screen out areas in the image that may contain mica grooves. Based on this, in the technical solution of the present application, the mica groove template image is slidably matched in the pre-processed rotor image to obtain a set of candidate mica groove regions. It should be understood that mica grooves generally have relatively fixed shape features. Using a pre-defined mica groove template image to perform a sliding comparison (i.e., template matching) on ​​the entire pre-processed image can quickly find areas with a high similarity to the template shape at the pixel level.

[0027] In a specific embodiment, a standard mica slot template (e.g., 20px*60px size) is entered in advance, and a sliding window is traversed across the entire image with a step size of 5px. For each window area, the similarity score between the target window and the template is calculated by the normalized correlation coefficient. Areas with scores exceeding a set threshold (e.g., 0.8) are considered candidate mica slot candidate areas, and all starting points and polar coordinate positions that meet the conditions are marked. The candidate areas screened in batches are arranged into sets for further processing by the subsequent confidence judgment and feature extraction modules.

[0028] Exemplarily, in step S23, the confidence score of each candidate mica groove region in the set of candidate mica groove regions is calculated to obtain a set of confidence scores. It should be understood that although the previous template matching process can quickly screen out areas in the image that are similar in morphology to the mica groove and form a set of candidate regions, this matching may not be completely accurate. The actual rotor image may contain noise, uneven lighting, reflections, surface scratches, stains (such as copper scrap accumulation), and structures that are similar in shape to the mica groove but are not the target, causing the template matching to produce misjudgments and include some non-target areas in the candidate set. If these candidate regions are used indiscriminately, it will directly affect the accuracy of the subsequent centerline calculation, thereby reducing the accuracy and quality of the engraving process. Therefore, in the technical solution of the present application, the confidence score of each candidate mica groove region in the set of candidate mica groove regions is further calculated to obtain a set of confidence scores. A reliability assessment is performed on each candidate mica groove region initially screened out by template matching. By calculating the likelihood that each candidate region is a genuine mica groove (i.e., its confidence score), we can quantitatively determine its effectiveness as a target. This helps the system identify the most reliable mica groove from among many candidates, thereby improving the accuracy and robustness of the entire visual positioning process and addressing various complex working conditions (such as uneven scale and irregular edges).

[0029] In one embodiment, Figure 3 As shown, the confidence score of each candidate mica groove area in the set of candidate mica groove areas being a mica groove is calculated to obtain a set of confidence scores, including: S231, performing convolution coding on each candidate mica groove area in the set of candidate mica groove areas to obtain a set of candidate mica groove area visual feature coding vectors; S232, performing fine-grained visual feature reshaping on each candidate mica groove area visual feature coding vector in the set of candidate mica groove area visual feature coding vectors to obtain a set of candidate mica groove area visual feature enhancement coding vectors; S233, inputting the set of candidate mica groove area visual feature enhancement coding vectors into a classifier-based mica groove identifier to obtain the set of confidence scores.

[0030] Exemplarily, in step S231, each candidate mica groove region in the set of candidate mica groove regions is convolutionally encoded to obtain a set of candidate mica groove region visual feature encoding vectors. Specifically, in the process of calculating the confidence score of each candidate mica groove region in the set of candidate mica groove regions being a mica groove, first, it is considered that after the set of candidate mica groove regions is obtained by template matching, these regions are merely original pixel blocks in the image that are similar in shape to the template. However, the actual image may contain noise, interference, illumination changes, surface texture differences, scratches, and features that are similar to the real mica groove morphology but are not the target. It is difficult to ensure the robustness and accuracy of confidence judgment directly based on these original pixel blocks or simple geometric features, and it is impossible to effectively meet the recognition requirements under complex working conditions. Therefore, a more powerful method is needed to extract deep and more discriminative information from these candidate regions. Based on this, each candidate mica groove region in the set of candidate mica groove regions is further convolutionally encoded to obtain a set of candidate mica groove region visual feature encoding vectors. Convolution operations, particularly convolutional layers derived from convolutional neural networks (CNNs), excel at automatically learning local patterns and spatial hierarchical features in images, from low-level edges and textures to high-level shape combinations. They can capture richer and more essential visual information than raw pixels or simple template matching. Each candidate mica groove region is converted into a compact and information-rich mathematical representation—the candidate mica groove region visual feature encoding vector. This vector is no longer a simple list of pixel values, but rather a collection of features extracted through convolutional layers that represent the core visual content of the region. The goal is to generate a feature representation that is invariant (robust) to interference factors such as lighting, noise, and slight deformations, while effectively distinguishing between mica grooves of different confidence levels, providing high-quality input for subsequent, more refined recognition (such as the fine-grained reconstruction and classifier recognition mentioned in the technical solution).

[0031] In a specific embodiment, ResNet-18 is used to perform convolution encoding on each candidate mica groove region in the set of candidate mica groove regions to obtain a set of candidate mica groove region visual feature encoding vectors. Specifically, the candidate mica groove region is used as the network input, and after various levels of convolution, batch normalization, ReLU activation and pooling, each candidate mica groove region finally outputs a candidate mica groove region visual feature encoding vector to obtain a set of candidate mica groove region visual feature encoding vectors.

[0032] Exemplarily, in step S232, each candidate mica groove region visual feature coding vector in the set of candidate mica groove region visual feature coding vectors is subjected to fine-grained visual feature reshaping to obtain a set of candidate mica groove region visual feature enhanced coding vectors. It should be understood that the candidate mica groove region visual feature coding vectors obtained solely through preliminary convolutional coding may not be sufficient to cope with the complex situations existing in actual industrial scenarios. In other words, due to problems such as irregular edges, copper chip contamination, scratches, uneven reflection, and subtle deformation of the groove shape on the rotor surface, these factors will cause the initially extracted features to contain noise and redundant information, or fail to fully capture the key subtle information for distinguishing true and false mica grooves and evaluating groove quality. The relative relationship and change pattern between the internal dimensions of the candidate mica groove region visual feature coding vectors contain key information, and conventional coding may not be able to fully and effectively utilize this information, and may even have dimensional redundancy and information mixing. Therefore, a more in-depth and sophisticated processing method is needed to optimize these preliminary features. Based on this, in the technical solution of the present application, each candidate mica groove area visual feature coding vector in the set of candidate mica groove area visual feature coding vectors is further subjected to fine-grained visual feature reshaping to obtain a set of candidate mica groove area visual feature enhanced coding vectors. Specifically, the process of fine-grained visual feature reshaping decouples and reshapes the structured information inside the feature vector, and generates a higher information content, stronger anti-interference ability and better discrimination through a series of refined operations such as feature phase reconstruction (using one-dimensional convolution to capture local patterns), information squeezing (distillation features, de-redundancy and denoising), counting of key elements of visual features, calculation of phase reconfiguration amplification factors (building fine control tools) and final feature phase importance reconstruction (adjusting structure, highlighting important information, and suppressing unimportant information). This aims to overcome the limitations of preliminary features, improve the efficiency, effectiveness, robustness and expressiveness of feature representation, so that it can better accurately reflect the true state of the mica groove under complex conditions.

[0033] In one embodiment, Figure 4As shown, each candidate mica groove area visual feature coding vector in the set of candidate mica groove area visual feature coding vectors is subjected to fine-grained visual feature reshaping to obtain a set of candidate mica groove area visual feature enhancement coding vectors, including: S2321, feature phase reconstruction and information squeezing processing are performed on the candidate mica groove area visual feature coding vectors to obtain a set of candidate mica groove area visual feature extraction local phase coding vectors; S2322, the candidate mica groove area visual feature of each candidate mica groove area visual feature extraction local phase coding vector in the set of candidate mica groove area visual feature extraction local phase coding vectors is calculated. Key element counting; S2323, based on the candidate mica groove area visual feature extraction of the local phase coding vector of each candidate mica groove area visual feature, the candidate mica groove area visual feature phase reconfiguration amplification factor of each candidate mica groove area visual feature local phase coding vector is calculated; S2324, based on the candidate mica groove area visual feature phase reconfiguration amplification factor of each candidate mica groove area visual feature local phase coding vector, the set of candidate mica groove area visual feature local phase coding vectors is reconstructed based on the feature phase importance to obtain the candidate mica groove area visual feature enhanced coding vector.

[0034] In one embodiment, in step S2321, the candidate mica groove region visual feature coding vector is subjected to feature phase reconstruction and information compression processing to obtain a set of candidate mica groove region visual feature extraction local phase coding vectors, including: performing feature phase reconstruction based on one-dimensional convolution coding on the candidate mica groove region visual feature coding vector to obtain a set of candidate mica groove region visual feature local phase coding vectors, expressed as:

[0035] Conv l×1 (X)={x1,x2,...,x i ,...,x n}

[0036] Where X is the visual feature encoding vector of the candidate mica groove area, Conv l×1 It is a one-dimensional convolutional coding process, l is the characteristic phase reconstruction step, x1,x2,x i ,x n They are respectively the 1st, 2nd, i-th and n-th candidate mica groove region visual feature local phase encoding vectors in the set of candidate mica groove region visual feature local phase encoding vectors.

[0037] Specifically, by reconstructing the feature phase of the candidate mica groove region visual feature encoding vector based on one-dimensional convolutional coding, the internal structure of the original convolutional feature vector is reprocessed and deeply modeled at the deep feature level. Convolutional coding not only expands the original global spatial features into a sliding window in one dimension, capturing the transitions, changes, and correlation patterns between each subsequence of the feature vector with a local receptive field, but also focuses on the implicit relationships and interactions between adjacent dimensions. The so-called phase reconstruction is theoretically based on the local coupling state of feature components at different spatial orders and scales. This coupling directly reflects the physical properties of the mica groove structure, such as continuity, boundary undulations, and defect fractures. The sliding characteristics of the one-dimensional convolution kernel enable the system to adaptively focus on structural information such as mutation points, pattern boundaries, and periodic features. Whether it is a micro-transformation in groove width, electrochemical corrosion texture, or mechanical scratch mutation, it can be encoded into a new local phase expression.

[0038] Information compression is performed on each candidate mica groove region visual feature local phase encoding vector in the set of candidate mica groove region visual feature local phase encoding vectors to obtain a set of candidate mica groove region visual feature extracted local phase encoding vectors, which is expressed as:

[0039]

[0040] Among them, ‖·‖ is the norm of the vector, v i The i-th candidate mica groove region visual feature extraction local phase encoding vector in the set of candidate mica groove region visual feature extraction local phase encoding vectors is extracted.

[0041] Specifically, information compression is performed on the obtained set of local phase encoding vectors of the visual features of the candidate mica slot regions, further achieving deep purification and selection of noise, redundancy, and weak similar information within the feature space. Information compression significantly reduces information in the local phase encoding that does not significantly contribute to target resolution or is even misleading. In industrial practice, rotor surfaces often experience feature aliasing due to copper chip accumulation, stray light, silver reflections, and surface oxidation color differences. Direct reliance on uncompressed features can easily lead to target recognition misjudgments and centerline fitting deviations. However, through information compression, highly representative components truly related to the physical mica slot structure, such as edge fluctuations, width extremes, and concavity patterns, are retained and enhanced. Simultaneously, the algorithm automatically suppresses the impact of subtle unstructured noise on the discriminant boundaries, significantly improving the effectiveness and anti-interference performance of subsequent recognition and segmentation. Ultimately, the resulting set of feature-extracted local phase encoding vectors after compression is a highly concentrated, discriminative, and easily analyzed refined information carrier.

[0042] Specifically, in step S2322, the count of key elements of the candidate mica groove region visual feature extraction local phase encoding vector of each candidate mica groove region visual feature extraction local phase encoding vector in the set of candidate mica groove region visual feature extraction local phase encoding vectors is calculated, which is expressed as:

[0043]

[0044] in, Extract the feature value of the jth position in the local phase encoding vector for the visual feature of the i-th candidate mica groove area, count i represents the effective component count, ε is the trainable preset threshold, en i v i The corresponding key element counts of visual features in the candidate mica groove area.

[0045] Specifically, the local phase coding vector extracted from the visual features of the candidate mica groove area after one-dimensional convolution coding and information compression can highly concentrate the main spatial texture, boundary changes and detail features of the mica groove target area, eliminate a large amount of redundant and noise components, and is a fine purification of the preliminary convolution features. At this time, the implementation of key element counting further explores the intrinsic significance of these highly refined local phase coding vectors. By counting the key elements of the visual features of the candidate mica groove area, the originally abstract high-dimensional feature representation can be converted into a clear, quantifiable indicator that is highly correlated with the physical performance of the groove structure. The key elements counted essentially represent the important information units that must be relied upon to distinguish between real mica grooves and non-target intervals within the candidate area, including but not limited to strong gradient changes at the groove edge, continuity or discontinuity of the groove bottom texture, local morphological complexity, and microscopic defect points.

[0046] In one embodiment, in step S2323, based on the count of key elements of the candidate mica groove region visual feature of the local phase coding vector of each candidate mica groove region visual feature extraction, calculating the candidate mica groove region visual feature phase reconfiguration amplification factor of each candidate mica groove region visual feature local phase coding vector includes: based on the count of key elements of the candidate mica groove region visual feature of the local phase coding vector of each candidate mica groove region visual feature extraction, determining the suppression factor corresponding to each candidate mica groove region visual feature extraction local phase coding vector, expressed as:

[0047]

[0048] λ i =1-θ i

[0049] Where n is the number of vectors in the set of local phase encoding vectors of the visual features of the candidate mica groove area, θi v i The corresponding polar angle, λ i v i The corresponding suppression factor, π represents pi, and arctan represents the inverse tangent function; based on the suppression factor corresponding to the local phase coding vector extracted from the visual features of each candidate mica groove area, the candidate mica groove area visual feature phase reconfiguration amplification factor of the local phase coding vector of the visual features of each candidate mica groove area is calculated.

[0050] Specifically, if the number of key elements is high and the distribution is reasonable, it means that the component that effectively characterizes the groove structure in the feature vector is dominant and can reflect the real mica groove boundary, width change and surface morphology. If the number is sparse or the distribution is abnormal, it indicates that the candidate area represented by the vector is more likely to reflect noise, artifacts, impurities or other types of structures. Therefore, the key element counting results are combined with the statistical analysis model, and the system dynamically assigns a corresponding inhibition factor to each feature vector. The level of the inhibition factor reflects the contribution and credibility of the visual features of the candidate area in the overall target discrimination. A high inhibition factor means that the information value of the area is low, which will lead to a weakening of the subsequent expression weight; a low inhibition factor means that the feature expression of the area is rich and the discrimination is strong, and its subsequent expression will be strengthened.

[0051] In one embodiment, calculating the candidate mica groove region visual feature phase reconfiguration amplification factor of each candidate mica groove region visual feature local phase coding vector based on the suppression factor corresponding to the extraction of each candidate mica groove region visual feature local phase coding vector includes: calculating the initial candidate mica groove region visual feature phase reconfiguration amplification factor of each candidate mica groove region visual feature local phase coding vector based on the suppression factor corresponding to the extraction of each candidate mica groove region visual feature local phase coding vector, expressed as:

[0052]

[0053] Among them, e i Reconfigure the amplification factor for the visual feature phase of the initial candidate mica groove region.

[0054] Specifically, the core significance of calculating the initial visual feature phase reconfiguration amplification factor by extracting the suppression factor corresponding to the local phase encoding vector for each feature lies in dynamically prioritizing the feature representations obtained from early recognition and analysis. The suppression factor generation process incorporates a quantitative assessment of the number and spatial distribution of key elements within the candidate region's feature vector, representing a comprehensive assessment of the validity and discriminative power of the information carried by the current feature dimension. Low suppression factors reflect feature components with rich structural information, dense key elements, and a significant fit to the actual boundaries or texture features of the mica groove. These dimensions are assigned high initial amplification weights, multiplying their contribution to the subsequent overall feature representation. High suppression factors represent feature components in the feature space that are information-poor, have weak structural representation, or are noise-dominated, with no clear groove shape. The corresponding initial amplification weights are reduced or suppressed. Through this adaptive weighting, the system proactively enhances key feature signals directly related to process target discrimination within the feature space while minimizing interference signals caused by factors such as industrial environmental changes, camera noise, and workpiece defects.

[0055] The visual feature phase reconfiguration amplification factor of the initial candidate mica groove region is corrected for characteristic phase dispersion omission to obtain the visual feature phase reconfiguration amplification factor of the candidate mica groove region, which is expressed as:

[0056]

[0057] Among them, K i is the analytic flatness factor of the visual space holomorphic structure of the candidate mica groove region, T i is the visual flat decomposition metric characterization factor of the candidate mica groove area, e i ′ Reconfigure the amplification factor for the visual feature phase of the candidate mica groove region.

[0058] Specifically, the distribution of the phase reconfiguration amplification factor of the visual features of the initial candidate mica groove area will inevitably be affected by problems such as discrete sample features, uneven spatial distribution, and omission of actual defects. Specifically, in the feature space, some local important structures or abnormal features may not obtain the required amplification due to averaging processing, resulting in the hidden dangers of phase dispersion and element omission. These dispersions and omissions are likely to manifest as interruptions in the expression of mica groove boundary features, weakening of the groove texture signal, or loss of key geometric information, thereby weakening the subsequent classifier's ability to resolve real grooves, and also affecting the accuracy of motion control and mechanical positioning. To address this problem, the present application introduces a feature phase dispersion omission correction step to adjust the phase reconfiguration amplification factor of the visual features of the initial candidate mica groove area. First, the set space is analyzed as a holomorphic structure metric to obtain the candidate mica groove region visual space holomorphic structure analysis flatness factor, then the effectiveness enhancement of local phase encoding is used as the flatness decomposition metric representation based on the holomorphic structure to construct the single-mode coupling representation as a gauge field to obtain the candidate mica groove region visual flatness decomposition metric characterization factor, and then, through the effectiveness enhancement representation of the initial candidate mica groove region visual feature phase reconfiguration amplification factor, it is shown that its individual mode is the highest weight state of flatness excitation in the spatial holomorphic structure, so that the initial candidate mica groove region visual feature phase reconfiguration amplification factor can be updated. Thus, while maintaining the phase direction symmetry of the set of the initial candidate mica groove region visual feature phase reconfiguration amplification factors based on the flatness space stability condition, the feature phase dispersion loss is avoided in the subsequent feature phase importance reconstruction process, thereby improving the expression effect of the candidate mica groove region visual feature enhancement coding vector.

[0059] Specifically, in step S2324, based on the candidate mica groove region visual feature phase reconfiguration amplification factor of each candidate mica groove region visual feature local phase encoding vector, the set of candidate mica groove region visual feature local phase encoding vectors is reconstructed based on feature phase importance to obtain the candidate mica groove region visual feature enhancement encoding vector, which is expressed as:

[0060]

[0061] Among them, exp is the value of the natural exponential function with a natural constant as the base, i Reshape the gain weight for the visual feature phase of the candidate mica groove region, v c Enhance the encoding vector with visual features for the candidate mica groove regions.

[0062] Specifically, the energy distribution of feature expression is dynamically adjusted by normalizing the candidate mica groove region visual feature phase reconfiguration amplification factor of each candidate mica groove region visual feature local phase encoding vector and then performing element-by-element product with the original candidate mica groove region visual feature local phase encoding vector. The values ​​of those components that are judged to have high discriminatory power in characterizing important parameters such as the mica groove boundary, width, and groove depth are greatly improved; while the corresponding background, weak texture, disturbance signal or non-structural redundant components have their expression weights significantly reduced or even close to suppression due to the smaller amplification factor. Such an operation ensures that after the weighted reconstruction of the entire feature set, all key information structures are highly prominent, and noise and redundancy are extremely weakened, thereby forming a candidate mica groove region visual feature enhanced coding vector that is information-intensive, highly discriminative and highly robust to engineering.

[0063] Exemplarily, in step S233, the set of the candidate mica groove region visual feature enhanced coding vectors is input into the mica groove identifier based on the classifier to obtain the set of the confidence scores. It should be understood that due to the internal structure of these enhanced candidate mica groove region visual feature enhanced coding vectors, key information is highlighted, redundancy and noise are suppressed, and thus have higher discrimination and stability. Inputting these high-quality enhanced feature vectors into the mica groove identifier based on the classifier can significantly improve the accuracy of mica groove recognition and the reliability of confidence assessment, laying a solid data foundation for the subsequent precise calculation of positioning deviation data (X-axis and rotation axis adjustment), the realization of precise tool setting and high-quality engraving operations.

[0064] In a specific embodiment, the classifier-based mica groove identifier uses a fully connected classifier, which takes the visual feature enhancement coding vectors of each candidate mica groove area as input, and after being processed by multiple layers of fully connected weight mapping and nonlinear activation units, realizes the output of the discrimination probability of the feature expression of different regions. Specifically, after the last level of fully connected layer, the model maps the visual feature enhancement coding vector of the candidate mica groove area to a probability distribution in the interval [0,1] through the Softmax normalization layer, and then obtains the confidence score. Each input candidate mica groove area visual feature enhancement coding vector corresponds to a confidence score to obtain a set of confidence scores. Finally, in step S24, the candidate mica groove area corresponding to the largest confidence score in the set is used as the current target groove.

[0065] Exemplarily, in step S3, the centerline position of the current target groove is calculated, and the centerline position of the current target groove is compared with the preset ideal processing position to obtain positioning deviation data. It should be understood that, due to the distribution of the mica grooves of the motor rotor during actual production, assembly and transportation, the grooves often have small but substantial offsets due to various factors such as material stress release, deformation of process parts, uneven clamping, and fixture errors. This offset may be negligible under the scale of a single groove, but with the iterative effect of the tool and the groove position during engraving, it will directly lead to inconsistency between the processing benchmark and the tool path, and ultimately produce defects such as skewed grooves, uneven depths, or widened copper bars, which not only affect the performance of the entire machine, but also cause major safety hazards such as short circuits in severe cases. In fact, traditional schemes such as manual alignment of dividing heads and mechanical step indexing often cannot perceive the spatial position changes of the actual workpiece in real time because they rely on physical reference points or process clamping scales. Positioning errors cannot be compensated in real time, resulting in product consistency and quality being easily limited. Therefore, the centerline position of the current target groove is calculated, and the centerline position of the current target groove is compared with the preset ideal processing position to obtain positioning deviation data.

[0066] Specifically, the Canny edge detection algorithm is first used to identify the boundary points of the target slot. Then, the least squares method is used to fit a straight line to represent the slot's centerline. The fitting process can be described as follows: The edge detection algorithm is used to obtain the coordinates of all boundary points on both sides of the mica slot. These coordinates are then input into the least squares model to calculate a best-fit straight line, which becomes the target slot's centerline. Mathematically, the least squares method aims to minimize the sum of the squared distances between the actual points and the fitted straight line.

[0067] Next, the preset ideal processing position is obtained. Those skilled in the art will appreciate that the preset ideal processing position is not a fixed value, but rather a reference position pre-set based on the motor's design specifications and processing standards. Setting the preset ideal processing position is a routine process. Specifically, the preset ideal processing position is typically generated based on a CAD (Computer-Aided Design) model, containing detailed dimensional specifications and technical parameters. The following are the specific steps: Extract key parameters such as the number of mica slots, distribution angle, and radial spacing from the motor's design documentation. For example, for a specific motor model, the ideal centerline position of its mica slots may be a certain distance from the rotor center in the rotor diameter direction, and distributed at predetermined angular intervals in the circumferential direction. These parameters are programmed into the control system to serve as a basis for subsequent adjustments. For example, in CNC programming, G-code instructions can be used to define the ideal processing path and position for each mica slot. Given the potential for subtle manufacturing variations between motor rotor batches, the preset ideal processing position may require fine-tuning based on actual conditions. For example, during mass production, statistical analysis can be used to assess the average deviation of the previous batch of products and adjust the processing parameters of subsequent batches accordingly to improve overall processing accuracy and consistency.

[0068] Specifically, after the centerline position of the current target groove is calculated, it is compared with the preset ideal processing position to obtain positioning deviation data, which includes the precise adjustment amount of the X-axis and the rotation axis.

[0069] For example, in step S4, the positioning deviation data is input into a motion controller, which is used to drive the servo motor based on the deviation data to adjust the centerline of the current target slot to align below the tool. It should be understood that the motion controller is responsible for receiving the positioning deviation data from the vision system and calculating the required adjustment instructions for the servo motor in real time based on a built-in algorithm. To achieve this, in one embodiment of the present application, the motion controller employs a PID (proportional-integral-differential) control strategy, a widely used feedback control method that effectively reduces errors and ensures both rapid and stable system response. PID control optimizes the servo motor's response characteristics by continuously adjusting three parameters: the proportional coefficient, the integral time, and the differential time. For example, during the adjustment process, if a large deviation between the actual position and the target position is detected, the proportional component generates a large adjustment signal. As the deviation gradually decreases, the integral component becomes more important, helping to eliminate steady-state errors. Finally, the differential component is primarily used to predict future error trends and prevent overshoot. After receiving the specific adjustment instruction, the servo motor will move in the predetermined direction and speed until the center line of the current target groove is accurately aligned under the tool.

[0070] Those skilled in the art will appreciate that the specific parameters of PID control can be set according to specific circumstances, and parameter setting is a routine process. Here, for ease of understanding, a specific embodiment is provided, which is merely an example and does not constitute a specific limitation. Specifically, the ideal machining position is 50 mm from the rotor center in the rotor diameter direction and 60 degrees in the circumferential direction. However, the actual measurement results show that the centerline position of the current target slot deviates from the ideal position, offset by +2 mm in the X-axis direction and -3 degrees in the rotation axis direction. The deviation data (+2 mm, -3 degrees) calculated above is transmitted to the motion controller. Setting the proportional gain to 1.2, the integral gain to 0.1, and the differential gain to 0.05, the calculated result for the X-axis adjustment is +2.4 mm, and the calculated result for the rotation axis adjustment is -3.6 degrees. Based on the calculated control signal, the servo motor drives the rotor to move +2.4 mm along the X-axis and rotate -3.6 degrees around its center.

[0071] For example, in step S5, after the centerline of the current target groove is aligned below the tool, the engraving process for the current target groove is executed. It should be understood that after confirming that the centerline of the current target groove has been accurately aligned below the tool, executing the engraving process can ensure that each mica groove is accurately machined to the ideal size and shape, thereby improving the operating stability of the motor and extending its service life.

[0072] In a specific embodiment, the motor rotor is sequentially sent to the slot of the stepper feeder body by an external mechanism, the piston rod of the supporting cylinder extends to lift the feeding bracket, and the feeding bracket slot lifts the motor rotor from the slot of the stepper feeder body, the feeding motor is started, the feeding bracket and the feeding cylinder move forward to the set position, the piston rod of the supporting cylinder retracts, the feeding bracket and the motor rotor descend, and the rear two rotors are placed in the slot of the stepper feeder body (advanced one slot), the main motor drives the parallelogram double crank mechanism to rotate to the material taking position, the ejection motor is started, the jacking linear slide moves forward with the transfer motor, the transfer motor ejector hits the center hole at one end of the rotor main shaft, and pushes the rotor out at a suitable speed until the center hole at the other end of the rotor main shaft is pushed by the fixed rotating ejector, see Figure 5 The main motor drives the parallelogram double crank mechanism to rotate until the double crank is in the vertical position. The main motor brake device works to keep the parallelogram double crank mechanism stationary. The material transfer motor rotates at the set speed. The linear array camera collects the image of the mica groove. Figure 6The detection program system detects the width of the mica groove and calculates the center line distance of each groove. (If the groove width exceeds the maximum allowable size, the rotor is unqualified and the program system sends an unqualified signal; if the distance between two adjacent center lines exceeds the maximum allowable value, the rotor is unqualified and the program system sends an unqualified signal.) Determine the starting groove for the CNC milling tool to carve and send the result information. The material transfer motor rotates to make the starting groove for carving rotate to the top, the main motor rotates, and the parallelogram double crank mechanism rotates to the bottom of the tool. The CNC milling machine works and carves the mica groove. Please refer to Figure 7 After each mica groove is carved, the tool is lifted, the material transfer motor rotates one grid, and the next mica groove is carved, until all the mica grooves are carved. The CNC milling machine is reset, the tightening motor reverses, the center is withdrawn, and the rotor falls into the plastic return chute, and the external material receiving mechanism takes the rotor away.

[0073] In summary, the present application provides a method for controlling the engraving of mica grooves, which obtains a rotor image captured by a visual system and realizes precise positioning and adjustment of the mica grooves through a series of image processing steps. First, the acquired rotor image is pre-processed by filtering, contrast enhancement, and geometric correction to reduce noise interference and highlight the characteristics of the mica groove. Then, the template matching technology is used to identify the candidate mica groove areas, and the confidence score of each candidate area is calculated to determine the most credible target groove. Based on the position information of the target groove, its deviation from the ideal processing position is calculated, and this deviation data is fed back to the motion controller to drive the servo motor for precise adjustment so that the center line of the target groove is aligned with the bottom of the tool and then the engraving program is executed. In this way, more accurate and robust mica groove positioning and engraving control can be achieved to meet the high-quality production requirements under complex working conditions.

[0074] This application also provides a control system for the engraving of mica grooves, such as Figure 8 As shown, the engraving control system 800 for mica grooves includes: a rotor image acquisition module 810, which is used to acquire a rotor image collected by a visual system; a target groove positioning module 820, which is used to locate any identifiable mica groove from the rotor image, and use the mica groove as the current target groove; a centerline calculation and deviation detection module 830, which is used to calculate the centerline position of the current target groove, and compare the centerline position of the current target groove with a preset ideal processing position to obtain positioning deviation data; a motion control module 840, which is used to input the positioning deviation data into a motion controller, and the motion controller is used to drive the servo motor to adjust based on the deviation data to align the centerline of the current target groove below the tool; an engraving execution module 850, which is used to execute the engraving program for the current target groove after the centerline of the current target groove is aligned below the tool.

[0075] In one embodiment, the positioning deviation data includes precise adjustment values ​​of the X-axis and the rotation axis.

[0076] Here, those skilled in the art will appreciate that the specific operations of the various modules and units in the above-mentioned mica groove engraving control system have been described in detail above. Figures 1 to 7 The following has been described in detail in the description of the undercut processing control method for mica grooves, and therefore, its repeated description will be omitted.

[0077] Finally, it should be noted that the above description has been provided for purposes of illustration and description. Furthermore, the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to be limiting. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art will appreciate that the technical solutions of the present invention may be modified or replaced with equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for controlling the engraving process of a mica groove, characterized in that: include: acquiring a rotor image captured by a vision system; Locating any identifiable mica slot from the rotor image and using the mica slot as a current target slot; Calculating the centerline position of the current target groove, and comparing the centerline position of the current target groove with a preset ideal processing position to obtain positioning deviation data; Inputting the positioning deviation data into a motion controller, wherein the motion controller is configured to drive a servo motor to adjust the positioning deviation data so as to align the center line of the current target slot below the tool; After the center line of the current target groove is aligned with the bottom of the tool, the engraving procedure for the current target groove is executed.

2. The method for controlling the engraving process of a mica groove according to claim 1, characterized in that: The positioning deviation data includes the precise adjustment amounts of the X-axis and the rotation axis.

3. The method for controlling the engraving process of a mica groove according to claim 1, characterized in that: Locating any identifiable mica slot from the rotor image and using the mica slot as a current target slot, including: performing image preprocessing on the rotor image to obtain a preprocessed rotor image, wherein the preprocessing includes filtering, contrast enhancement, and geometric correction; Sliding matching the mica groove template image in the preprocessed rotor image to obtain a set of candidate mica groove regions; Calculating a confidence score that each candidate mica groove region in the set of candidate mica groove regions is a mica groove to obtain a set of confidence scores; The candidate mica slot region corresponding to the one with the largest confidence score in the set is used as the current target slot.

4. The method for controlling the engraving process of a mica groove according to claim 3, characterized in that: Calculating the confidence score of each candidate mica groove region in the set of candidate mica groove regions being a mica groove to obtain a set of confidence scores, including: performing convolution coding on each candidate mica groove region in the set of candidate mica groove regions to obtain a set of candidate mica groove region visual feature coding vectors; Performing fine-grained visual feature reshaping on each candidate mica groove region visual feature coding vector in the set of candidate mica groove region visual feature coding vectors to obtain a set of candidate mica groove region visual feature enhanced coding vectors; The set of candidate mica groove region visual feature enhanced encoding vectors is input into a classifier-based mica groove identifier to obtain the set of confidence scores.

5. The method for controlling the engraving process of a mica groove according to claim 4, characterized in that: Performing fine-grained visual feature reshaping on each candidate mica groove region visual feature coding vector in the set of candidate mica groove region visual feature coding vectors to obtain a set of candidate mica groove region visual feature enhanced coding vectors, including: performing feature phase reconstruction and information squeezing processing on the candidate mica groove region visual feature coding vector to obtain a set of candidate mica groove region visual feature extraction local phase coding vectors; Calculating the number of key elements of the candidate mica groove region visual feature extraction local phase coding vector of each candidate mica groove region visual feature extraction local phase coding vector in the set of the candidate mica groove region visual feature extraction local phase coding vector; Extracting the key element count of the candidate mica groove region visual feature of the local phase coding vector based on the visual feature of each candidate mica groove region, and calculating the candidate mica groove region visual feature phase reconfiguration amplification factor of the local phase coding vector of each candidate mica groove region visual feature; Based on the candidate mica groove area visual feature phase reconfiguration amplification factor of each candidate mica groove area visual feature local phase coding vector, the set of candidate mica groove area visual feature local phase coding vectors is reconstructed in terms of feature phase importance to obtain the candidate mica groove area visual feature enhanced coding vector.

6. The method for controlling the engraving process of a mica groove according to claim 5, characterized in that: Performing feature phase reconstruction and information squeezing processing on the candidate mica groove region visual feature coding vector to obtain a set of candidate mica groove region visual feature extraction local phase coding vectors, including: Performing feature phase reconstruction based on one-dimensional convolution coding on the candidate mica groove region visual feature coding vector to obtain a set of candidate mica groove region visual feature local phase coding vectors; Information compression is performed on each candidate mica groove region visual feature local phase coding vector in the set of candidate mica groove region visual feature local phase coding vectors to obtain the set of candidate mica groove region visual feature extracted local phase coding vectors.

7. The method for controlling the engraving process of a mica groove according to claim 6, characterized in that: Extracting a count of key elements of the candidate mica groove region visual features of the local phase coding vector based on the visual features of each candidate mica groove region, and calculating a candidate mica groove region visual feature phase reconfiguration magnification factor of the local phase coding vector of each candidate mica groove region visual feature, comprising: Determining the suppression factor corresponding to the local phase coding vector extracted from the visual feature of each candidate mica groove region based on the count of the key elements of the visual feature of the candidate mica groove region extracted from the local phase coding vector of the visual feature of each candidate mica groove region; Based on the suppression factors corresponding to the local phase coding vectors extracted from the visual features of each candidate mica groove region, the candidate mica groove region visual feature phase reconfiguration amplification factors of the local phase coding vectors of the visual features of each candidate mica groove region are calculated.

8. The method for controlling the engraving process of a mica groove according to claim 7, characterized in that: The method comprises: extracting a suppression factor corresponding to a local phase coding vector based on each candidate mica groove region visual feature, and calculating a candidate mica groove region visual feature phase reconfiguration amplification factor of each candidate mica groove region visual feature local phase coding vector, comprising: Calculate the initial candidate mica groove region visual feature phase reconfiguration amplification factor of the local phase encoding vector of each candidate mica groove region visual feature based on the suppression factor corresponding to the local phase encoding vector of each candidate mica groove region visual feature; The initial candidate mica groove region visual feature phase reconfiguration magnification factor is corrected for feature phase dispersion omission to obtain the candidate mica groove region visual feature phase reconfiguration magnification factor.

9. A control system for engraving mica grooves, characterized in that: include: A rotor image acquisition module, used for acquiring a rotor image collected by a visual system; a target slot positioning module, configured to locate any identifiable mica slot from the rotor image and use the mica slot as a current target slot; A centerline calculation and deviation detection module is used to calculate the centerline position of the current target groove and compare the centerline position of the current target groove with a preset ideal processing position to obtain positioning deviation data; a motion control module, configured to input the positioning deviation data into a motion controller, wherein the motion controller is configured to drive a servo motor to adjust the positioning deviation data so as to align the center line of the current target groove below the tool; The engraving execution module is used to execute the engraving program for the current target groove after the center line of the current target groove is aligned below the tool.

10. The control system for the engraving process of mica groove according to claim 9, characterized in that: The positioning deviation data includes the precise adjustment amounts of the X-axis and the rotation axis.

Citation Information

Patent Citations

  • Automatic forming machine for insulating groove between rotor commutator segments based on visual identification technology

    CN112615233A