Intelligent production control method and system for electromagnetic coils

Abstract

The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent production control method and system for electromagnetic coils. The method comprises the following steps: performing coil digitization processing according to the electromagnetic coil design drawing to generate a three-dimensional coil digital model; performing voxel winding trajectory mapping according to the three-dimensional coil digital model, and performing adaptive winding trajectory adjustment to obtain optimized coil winding trajectory data; performing real-time coil winding image acquisition based on the optimized coil winding trajectory data to generate real-time coil winding image data; performing turn layer wire diameter change trend prediction according to the real-time coil winding image data to generate wire diameter change trend data; performing tension feedback control strategy processing according to the wire diameter change trend data, and performing multi-layer wire diameter consistency analysis to obtain electromagnetic coil quality score data. The present invention achieves precise control of electromagnetic coil tension by enhancing multi-layer wire diameter consistency during electromagnetic coil winding.

Description

Technical Field

[0001] The present invention relates to the field of electronic manufacturing technology, and in particular to an intelligent production control method and system for electromagnetic coils. Background Art

[0002] Electromagnetic coils, as a basic electromagnetic component, are widely used in various fields such as motors, sensors, medical devices, and communications equipment. In recent years, with the development trend of miniaturization, intelligence, and high performance of electronic equipment, electromagnetic coils, as core components of these devices, have become increasingly demanding in terms of performance, such as higher efficiency, smaller size, and more precise magnetic field distribution. Intelligent production of electromagnetic coils can not only improve production efficiency, but also ensure the consistency and stability of product quality, reduce production costs, and enhance market competitiveness. However, traditional electromagnetic coil production methods often rely on manual operation or simple automated equipment, making it difficult to achieve precise tension control. This is especially difficult when winding wires of different wire diameters and materials, which can easily lead to loose coils, short circuits between layers, or breakage. At the same time, for multi-layer electromagnetic coils, slight differences in wire diameter between different layers can cause deviations in the coil's inductance and Q value, affecting the coil's performance and reliability. Summary of the Invention

[0003] Based on this, the present invention provides an electromagnetic coil intelligent production control method and system to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for intelligent production control of electromagnetic coils includes the following steps:

[0005] Step S1: performing coil geometric feature recognition on the electromagnetic coil design drawing to obtain coil topology data; performing key process parameter supplementation processing based on the coil topology data to obtain key coil process parameters; performing coil digitization processing on the coil topology data using the key coil process parameters to generate a three-dimensional coil digital model;

[0006] Step S2: performing coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; performing voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; performing adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data;

[0007] Step S3: Based on the optimized coil winding trajectory data, intelligent manufacturing is performed using an intelligent electromagnetic coil production device, and real-time coil winding image acquisition is performed to generate real-time coil winding image data; based on the real-time coil winding image data, a turn layer wire diameter change trend is predicted to generate wire diameter change trend data;

[0008] Step S4: performing tension feedback control strategy processing based on the wire diameter change trend data to obtain tension feedback control data; performing wire diameter consistency compensation control based on the tension feedback control data, and collecting wire diameter compensation control effect data to obtain wire diameter compensation effect data; performing multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index;

[0009] Step S5: Score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain electromagnetic coil quality score data.

[0010] The present invention identifies coil geometry features from electromagnetic coil design drawings, accurately capturing the coil's geometry, dimensions, and wiring topology. Key process parameter supplementation based on the coil topology data allows for the addition of necessary process parameters, such as wire diameter and winding tension, based on the coil's actual structural characteristics. Using key coil process parameters to digitize the coil topology data, the coil geometry and process information are integrated to construct a highly accurate three-dimensional digital coil model. By refining the three-dimensional model down to the voxel level, precise planning of the coil winding path is achieved, improving winding accuracy and feasibility and providing direct guidance for intelligent manufacturing. Dynamically optimizing the path based on equipment capabilities, material properties, and process requirements reduces physical friction and material waste during the winding process, improving production efficiency and coil consistency. Executing the optimized winding path through an intelligent electromagnetic coil production device achieves automation and intelligentization of the production process, significantly improving production efficiency and product quality. Real-time coil winding image acquisition not only monitors the production process but also promptly identifies potential issues. By analyzing the captured images, changes in wire diameter can be predicted in advance. Based on the wire diameter change trend data, the tension during winding is automatically adjusted to ensure the consistency of the wire diameter during the coil winding process, which is crucial for the stability of the electromagnetic coil performance. The wire diameter consistency compensation control and effect acquisition mechanism further optimizes the accuracy of wire diameter control, reduces the performance fluctuations caused by uneven wire diameter, evaluates the consistency of the wire diameter between each layer of the coil as a whole, and provides a quantitative indicator for the final quality of the product. Therefore, the intelligent production control method of an electromagnetic coil of the present invention constructs four-dimensional voxel winding topology network data, performs geometric topology and timing information analysis of the coil winding, and obtains optimized winding trajectory data; controls point calibration, classification and processing instruction encoding are performed according to the optimized winding trajectory data, thereby realizing refined programming control of the winding process; obtains wire diameter change trend data in real time through high-speed visual monitoring, accurately predicts the wire diameter change trend, performs comprehensive analysis of pre-winding trajectory deviation and real-time tension with the optimized winding trajectory data, adjusts the control parameters of the winding system in real time, and performs closed-loop compensation control on the next turn layer of the coil, effectively solving the tension control problem and improving the consistency of the wire diameter of the multi-layer winding of the electromagnetic coil.

[0011] Preferably, step S1 includes the following steps:

[0012] Step S11: performing image denoising preprocessing on the electromagnetic coil design drawing to generate a standard electromagnetic coil design drawing; extracting original drawing annotations based on the standard electromagnetic coil design drawing to obtain original drawing annotation data;

[0013] Step S12: performing multi-level drawing analysis based on the standard electromagnetic coil design drawing and performing vector graphics analysis to generate coil vector structure data;

[0014] Step S13: performing coil geometric feature recognition on the coil vector structure data using preset geometric constraint relationship rules to obtain coil topology structure data;

[0015] Step S14: performing graphic element annotation processing on the coil topology structure data using the original drawing annotation data, and establishing a mapping relationship between graphic elements and parameters to generate preliminary topology graphic element annotation data;

[0016] Step S15: performing key process parameter supplementation processing based on the preliminary topological primitive annotation data and the coil topological structure data to obtain key coil process parameters, wherein the key coil structure parameters include coil material attribute data, coil winding attribute data, and coil electrical attribute data;

[0017] Step S16: digitize the key coil structure parameters using the coil topology data to generate a three-dimensional coil digital model.

[0018] The present invention eliminates noise and unnecessary interference information in electromagnetic coil design drawings through image denoising preprocessing, making the design drawings clearer and more accurate. The generated standard electromagnetic coil design drawings improve the standardization of drawing information, facilitate unified processing and reduce recognition errors, and accelerate the conversion process from design to production. Multi-level drawing analysis and vector graphics parsing can not only meticulously analyze the various complex elements in the design drawings, but also improve the resolution and scaling invariance of the graphics through vectorization processing, making the representation of the coil structure more accurate and efficient. By processing the coil vector structure data according to preset geometric constraint relationship rules, the key geometric features of the coil, such as coil diameter, height, number of layers, etc., can be accurately identified, improving the accuracy and efficiency of recognition. The graphic element annotation processing and the establishment of mapping relationships associate abstract graphic elements with specific process parameters, not only ensuring the faithful transmission of design intent, but also facilitating the rapid adjustment of parameters according to different production requirements, thereby improving the adaptability of the production system. The supplementary processing of key process parameters comprehensively considers the material properties, winding properties, and electrical properties of the coil. By utilizing coil topology data and key process parameters, a high-precision three-dimensional coil digital model is generated through digital processing, providing direct processing guidance for subsequent intelligent manufacturing, ensuring the entire process from design to manufacturing is digitalized and precise, and greatly improving production efficiency and product consistency.

[0019] Preferably, step S15 includes the following steps:

[0020] Step S151: Analyze the electrical performance of the target coil based on the preliminary topological element annotation data to generate coil electrical property data;

[0021] Step S152: extracting coil winding geometric parameters based on the coil topology data to generate coil winding geometric parameters, wherein the coil winding geometric parameters include coil wire diameter, coil outer diameter, coil inner diameter, and coil winding height or length;

[0022] Step S153: performing wire material property matching on the coil electrical property data and the coil winding geometric parameters through a preset material database to generate coil material property data;

[0023] Step S154: Calculate the number of coil layers according to the coil winding geometric parameters to generate coil layer data;

[0024] Step S155: using the coil material attribute data to estimate the coil winding geometric parameters, and generating coil turn estimation data; wherein the coil turn estimation is calculated using a coil turn estimation formula, which is as follows:

[0025]

[0026] Where N is the estimated number of coil turns, V is the volume of the coil winding area, and A w It is expressed as the cross-sectional area of a single wire diameter, π is pi, D is the average winding diameter of the coil, d is the wire diameter, and α is the winding process coefficient;

[0027] Step S156: identifying the coil winding mode of the coil topology data using the coil layer number data and the coil turn number estimation data to generate coil winding mode category data;

[0028] Step S157: integrating the coil winding type data, the coil layer number data, and the coil turn number estimation data into winding data to generate coil winding attribute data.

[0029] This invention uses preliminary topological element annotation data to perform electrical performance analysis, directly linking design intent with actual electrical performance. This ensures that key electrical characteristics of electromagnetic coils, such as inductance and resistance, can be predicted and optimized early in the design process. Extracting coil winding geometry parameters not only clarifies the coil's physical form, such as wire diameter, inner and outer diameters, and height. By matching against a material database, optimal material selection is achieved based on the coil's electrical properties and winding geometry, ensuring that the selected material meets both electrical performance requirements and adapts to the specific geometry. Calculating the number of coil layers based on the winding geometry ensures a compact and rational coil structure. Accurately estimating the number of coil turns not only optimizes electromagnetic performance but also takes into account the impact of winding process factors on the actual number of turns, ensuring that coil performance during actual manufacturing is consistent with design expectations and improving the first-time success rate of the product. The most suitable coil winding method is identified, and a customized winding solution is selected based on the coil's structural characteristics and turn requirements. Key coil winding parameters, including winding method, number of layers, and number of turns, are integrated into unified winding attribute data.

[0030] Preferably, step S2 includes the following steps:

[0031] Step S21: performing finite element meshing processing on the three-dimensional coil digital model to generate a finite element coil digital model;

[0032] Step S22: performing coil voxel winding network analysis based on the finite element coil digital model to generate voxel winding topology network data;

[0033] Step S23: performing a windability analysis on the voxel winding topology network data and performing voxel winding trajectory mapping to obtain preliminary coil winding trajectory data;

[0034] Step S24: Decomposing the winding path according to the preliminary coil winding trajectory data to generate a coil winding sub-path set;

[0035] Step S25: performing path conflict detection on the coil winding sub-path set to obtain winding conflict sub-path data;

[0036] Step S26: performing magnetic field constraint analysis on the preliminary coil winding trajectory data using a finite element coil digital model based on a preset magnetic field uniformity distribution strategy, and performing finite element magnetic field distribution simulation calculation to generate coil magnetic field constraint simulation data;

[0037] Step S27: identifying the magnetic field inhomogeneity region according to the coil magnetic field constraint simulation data, and adaptively adjusting the preliminary coil winding trajectory data through the winding conflict sub-path data to obtain optimized coil winding trajectory data.

[0038] The present invention discretizes the continuous three-dimensional coil digital model into a finite number of units, so that the coil model can perform accurate electromagnetic field distribution calculation and structural optimization in a finite element solution environment. The geometric structure of the coil is converted into a voxel-level topological network representation, so that the winding path can be accurately mapped to the voxel structure of the coil. Through the windability analysis, the winding feasibility of the coil structure is evaluated, the feasibility of the planned winding trajectory in actual production is ensured, and physically infeasible paths are avoided. The windable topological network is mapped into preliminary winding trajectory data, and the complex overall winding path is subdivided into multiple sub-path sets, which facilitates independent analysis and optimization of each sub-path, thereby enhancing the flexibility and pertinence of path planning. By identifying and marking the sub-paths that cause physical conflicts in the coil winding process, a clear goal is provided for eliminating potential production obstacles, which helps to reduce the failure rate and material waste in the production process and ensure smooth production. The magnetic field constraint analysis based on the magnetic field uniformity distribution strategy ensures that the winding trajectory not only meets physical feasibility but also achieves ideal electromagnetic performance. By identifying areas of magnetic field inhomogeneity and combining winding conflict sub-path data to perform adaptive winding trajectory adjustments, performance bottlenecks and physical obstacles encountered in production are specifically addressed. The optimized coil winding trajectory data not only ensures the conflict-free winding process, but also maximizes the magnetic field uniformity of the coil.

[0039] Preferably, step S22 includes the following steps:

[0040] Step S221: constructing a four-dimensional voxel space for the finite element coil digital model to generate a coil four-dimensional voxel space, wherein three spatial dimensions in the coil four-dimensional voxel space correspond to the length, width, and height of the collar, and the fourth dimension corresponds to the number of layers of the coil;

[0041] Step S222: performing tensor element initialization processing according to the coil four-dimensional voxel space to generate an initialized four-dimensional tensor value;

[0042] Step S223: Mark the voxel occupancy of the coil's four-dimensional voxel space by initializing the four-dimensional tensor value to generate coil occupancy voxel data. For each voxel, determine whether its center point is within the geometric range of the layer. If so, iteratively update the tensor value of the voxel space to 1, indicating that it is occupied. Otherwise, the tensor value of the voxel space remains unchanged.

[0043] Step S224: performing connected domain component analysis on the coil four-dimensional voxel space using a breadth-first search algorithm based on the coil occupied voxel data to generate coil voxel connected domain data;

[0044] Step S225: performing adjacent layer overlap analysis based on the coil voxel connected domain data to obtain a directed acyclic graph of the voxel connected layer;

[0045] Step S226: performing discrete time series calculation on the voxel connected layer directed acyclic graph using a preset winding speed, and assigning a timestamp step size to generate voxel winding topology network data.

[0046] The present invention discretizes the continuous three-dimensional coil digital model into a four-dimensional voxel representation, improving the accuracy and resolution of the model. The fourth dimension introduces layer number information, and the initialization processing of the tensor elements establishes the basic numerical representation of the coil voxel space. Through voxel occupancy marking, it is possible to accurately identify and record which voxels are occupied by the coil during each layer of winding, and to accurately mark the voxel units occupied by the coil in the four-dimensional voxel space, thereby generating occupied voxel data describing the geometric structure of the coil. Through an efficient breadth-first search algorithm, it is possible to extract the connected domain component information of the coil from the occupied voxel data, that is, which voxels are connected. The generated connected domain data reflects the topological structural characteristics of the coil. By analyzing the voxel connectivity relationship between different layers, a directed acyclic graph describing the topological connection between layers is constructed. This graph can accurately express the layer-by-layer connection relationship between different layers during the coil winding process. The geometric topological structure of the coil is combined with the winding timing information. According to the preset winding speed, a discrete timestamp step value is assigned to each node in the directed acyclic graph. This concretizes the time dimension of the coil winding process and fully describes the four-dimensional topological network data of the coil winding process.

[0047] Preferably, step S3 includes the following steps:

[0048] Step S31: calibrating the control path points of the optimized coil winding trajectory data using the three-dimensional coil digital model to generate winding path control point data;

[0049] Step S32: performing path point type identification on the winding path control point data to generate winding path point type data, wherein the winding path point type data includes a starting point, an end point, a turning point, an intersection point, and a smoothing point;

[0050] Step S33: encoding the optimized coil winding trajectory data into processing control instructions using the winding path point type data, and distributing the servo control instructions to obtain coil servo control instruction data;

[0051] Step S34: performing intelligent manufacturing execution using an intelligent electromagnetic coil production device based on the coil servo control instruction data, and using a high-speed industrial camera to perform real-time coil winding image acquisition to generate real-time coil winding image data;

[0052] Step S35: predicting the wire diameter change trend of the turn layer according to the real-time coil winding image data, and generating wire diameter change trend data.

[0053] The present invention combines optimized coil winding trajectory data with a three-dimensional coil digital model to calibrate control path points, ensuring precise control of each key position during the actual production process and improving the execution accuracy of the winding path. The winding path control point data is identified and distinguished into different types, such as starting points, ending points, turning points, intersection points, and smoothing points. This allows control instructions to adopt the most appropriate processing strategy for different types of points, improving processing flexibility and efficiency. Based on the path point type and optimized trajectory, corresponding processing control instructions, such as motion instructions, speed instructions, and direction instructions, are encoded and generated. These instructions are then distributed to the servo control system to generate directly executable coil servo control instruction data. Based on the servo control instructions, the coil winding process is precisely controlled by an intelligent production device, while high-speed cameras are used to capture real-time image data of the winding process. These real-time image data can reflect the actual state of the coil winding. Based on the real-time captured winding images, image processing and machine learning techniques are used to predict the changing trend of the turn layer wire diameter during the coil winding process. This not only reflects the coil quality status but also provides an important basis for tension feedback control and wire diameter compensation.

[0054] Preferably, step S35 includes the following steps:

[0055] Step S351: using an edge detection operator to perform real-time coil winding layer edge detection on the real-time coil winding image data, and performing dynamic edge pixel focusing to generate winding layer edge mask data;

[0056] Step S352: performing pixel-level motion estimation of edges of adjacent frames based on the winding layer edge mask data, and performing continuous frame displacement vector superposition to obtain real-time winding layer edge trajectory data;

[0057] Step S353: performing sub-pixel edge positioning processing on the real-time winding coil edge trajectory data to generate a high-precision coil edge coordinate set;

[0058] Step S354: performing time-series multi-frame image fitting based on the high-precision circle edge coordinate set to obtain dynamic line diameter curve fitting data;

[0059] Step S355: performing spectrum analysis on the dynamic wire diameter curve fitting data to generate wire diameter fluctuation characteristic vector data;

[0060] Step S356: Based on the preset different wire winding fluctuation data and the corresponding future wire diameter change trend label data, a long short-term memory network model is used to perform transfer learning processing to obtain a wire diameter prediction model;

[0061] Step S357: transmitting the wire diameter fluctuation characteristic vector data to the wire diameter prediction model to predict the wire diameter change trend of the next turn layer, and generating wire diameter change trend data.

[0062] The present invention processes real-time image data using an edge detection operator, enabling rapid identification and focusing of the edges of the winding coil layer. It can accurately detect the edge contour of the current winding coil layer from the real-time winding image, focus on the edge pixels, and generate mask data describing the coil layer edge. By estimating the motion of edge pixels in adjacent video frames and superimposing the displacement vectors of multiple consecutive frames, the complete edge trajectory of the current winding coil layer can be accurately reconstructed. A sub-pixel-level fine positioning algorithm can obtain a high-precision coordinate set of the coil layer edge, significantly improving the resolution and accuracy of wire diameter measurement. Curve fitting of multiple frames of high-precision edge coordinates in a time series can reconstruct the dynamic wire diameter change curve of the current coil winding. The dynamic wire diameter curve data can reflect the real-time change trend of the wire diameter. By performing spectral analysis on the dynamic wire diameter curve, the key frequency characteristics of the wire diameter fluctuation can be extracted and quantified into feature vectors. The wire diameter fluctuation feature vector data can accurately describe the periodicity and regularity of the current wire diameter change. Leveraging existing wire diameter fluctuation data and labeled data, a long short-term memory neural network model capable of predicting future wire diameter trends was trained through transfer learning. This model can learn from time-series wire diameter fluctuation data and predict future trends. Inputting the wire diameter fluctuation feature vector into the pre-trained wire diameter prediction model enables accurate prediction of future wire diameter trends for one or more turns.

[0063] Preferably, step S4 includes the following steps:

[0064] Step S41: performing wire diameter trend quantification processing according to the wire diameter change trend data to generate quantified wire diameter change rate data;

[0065] Step S42: performing eigenmode analysis on the quantized wire diameter change rate data to obtain wire diameter transformation characteristic data;

[0066] Step S43: Calculating the trend Gaussian cumulative distribution of the wire diameter transformation feature data using a preset wire diameter change threshold, and determining the wire diameter change level. When the wire diameter change level is lower than the wire diameter change threshold, the winding layer is marked as a stable winding layer; when the wire diameter change level is higher than or equal to the wire diameter change threshold, the winding layer is marked as an abnormal winding layer.

[0067] Step S44: performing tension feedback control strategy processing on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data based on the abnormal winding layer to obtain tension feedback control data;

[0068] Step S45: Using the intelligent electromagnetic coil production device to perform real-time wire diameter consistency compensation control on the next turn layer of the coil based on the tension feedback control data, and collecting the wire diameter compensation control effect to obtain wire diameter compensation effect data;

[0069] Step S46: performing multi-layer wire diameter consistency calculation based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index.

[0070] The present invention converts complex wire diameter change information into intuitive quantitative wire diameter change rate data through quantitative processing of wire diameter change trend data. Intrinsic mode analysis extracts the core features of wire diameter transformation from the quantitative wire diameter change rate data. These features reflect the essential pattern of wire diameter change and help to deeply understand the inherent law of wire diameter change. By setting the wire diameter change threshold and performing Gaussian cumulative distribution calculation of the trend, an objective classification of wire diameter change levels is achieved, and stable winding layers and abnormal winding layers are distinguished, so that abnormal situations that affect product quality can be identified in a timely manner. Adjustments are made to the causes of wire diameter changes, such as controlling the wire diameter by changing the winding tension, which effectively suppresses the occurrence of wire diameter inconsistency and ensures the consistency and quality of the product. The intelligent electromagnetic coil production device is used to implement a tension feedback control strategy and implement real-time wire diameter consistency compensation control on the next turn layer of the coil, which significantly improves the self-adjustment ability of the production process and the efficiency of responding to abnormal situations. The multi-layer wire diameter consistency calculation based on the wire diameter compensation effect data provides a quantitative evaluation standard for the comprehensive quality of the entire electromagnetic coil. It can not only be used to instantly evaluate the control effect of the production process, but also promote the dual improvement of production efficiency and product quality.

[0071] Preferably, step S44 includes the following steps:

[0072] Step S441: performing preset trajectory deviation calculation on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data to generate a winding trajectory deviation value;

[0073] Step S442: performing deviation time series analysis on the winding trajectory deviation value through a preset winding time window to obtain dynamic deviation time series characteristic data;

[0074] Step S443: Calculating fuzzy membership based on the dynamic deviation time series feature data to obtain a deviation fuzzy feature matrix;

[0075] Step S444: using the tension sensor to obtain real-time tension monitoring data; performing fuzzy inference calculation on the real-time tension monitoring data using a preset tension fuzzy control rule library to generate fuzzy tension control data;

[0076] Step S445: Defuzzifying the deviation fuzzy feature matrix using the tension fuzzy control rule library based on the fuzzy tension control data to obtain precise tension adjustment amount sequence data;

[0077] Step S446: Adaptively adjust control parameters according to the precise tension adjustment amount sequence data to generate tension feedback control data.

[0078] The present invention reveals the difference between the actual winding process and the ideal trajectory by optimizing the combination of coil winding trajectory data and wire diameter transformation characteristic data, thereby enhancing the pertinence of the control strategy. The introduction of the winding time window for time series analysis of the deviation value not only takes into account the immediate state of the deviation, but also the trend of the deviation over time, which helps to deeply understand the development pattern of the deviation. Complex nonlinear relationships can be simplified, and the application of fuzzy set theory improves the ability to process uncertain information and enhances the flexibility and robustness of control decisions. Through fuzzy reasoning calculation, real-time tension data is combined with a preset tension fuzzy control rule library to achieve intelligent processing of tension control, which can respond to dynamic changes in the production process more quickly and accurately. The deviation fuzzy characteristic matrix is defuzzified and converted to obtain precise tension adjustment sequence data, and the decision results of fuzzy control are converted into actual executable control instructions, which improves the accuracy and execution efficiency of the control instructions and ensures accurate tension adjustment. The control parameters are adaptively adjusted according to the precise tension adjustment sequence data, directly guiding the production equipment to perform real-time tension control. The tension can be dynamically adjusted according to the real-time status of the coil winding, effectively controlling the wire diameter change, and ensuring the stability of the coil winding process and the consistency of product quality.

[0079] The present invention further provides an electromagnetic coil intelligent production control system, which executes the electromagnetic coil intelligent production control method described above. The electromagnetic coil intelligent production control system includes:

[0080] The coil digitization module is used to identify the coil geometric features of the electromagnetic coil design drawing to obtain the coil topology data; perform key process parameter supplementation processing based on the coil topology data to obtain key coil process parameters; and use the key coil process parameters to digitize the coil topology data to generate a three-dimensional coil digital model;

[0081] The winding trajectory optimization module is used to perform coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; perform voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; and perform adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data;

[0082] An intelligent manufacturing execution module is used to perform intelligent manufacturing execution using an intelligent electromagnetic coil production device based on optimized coil winding trajectory data, and to collect real-time coil winding images to generate real-time coil winding image data; based on the real-time coil winding image data, it predicts the trend of wire diameter changes in the turn layer and generates wire diameter change trend data;

[0083] The tension compensation control module is used to process the tension feedback control strategy based on the wire diameter change trend data to obtain tension feedback control data; perform wire diameter consistency compensation control based on the tension feedback control data, and collect wire diameter compensation control effect data to obtain wire diameter compensation effect data; perform multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index;

[0084] The coil quality evaluation module is used to score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain the electromagnetic coil quality score data. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 This is a schematic diagram of the steps of an electromagnetic coil intelligent production control method and system of the present invention;

[0086] Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0087] Figure 3 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0088] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0089] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0090] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0091] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0092] To achieve this, please refer to Figures 1 to 3 The present invention provides an electromagnetic coil intelligent production control method, comprising the following steps:

[0093] Step S1: performing coil geometric feature recognition on the electromagnetic coil design drawing to obtain coil topology data; performing key process parameter supplementation processing based on the coil topology data to obtain key coil process parameters; performing coil digitization processing on the coil topology data using the key coil process parameters to generate a three-dimensional coil digital model;

[0094] Step S2: performing coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; performing voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; performing adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data;

[0095] Step S3: Based on the optimized coil winding trajectory data, intelligent manufacturing is performed using an intelligent electromagnetic coil production device, and real-time coil winding image acquisition is performed to generate real-time coil winding image data; based on the real-time coil winding image data, a turn layer wire diameter change trend is predicted to generate wire diameter change trend data;

[0096] Step S4: performing tension feedback control strategy processing based on the wire diameter change trend data to obtain tension feedback control data; performing wire diameter consistency compensation control based on the tension feedback control data, and collecting wire diameter compensation control effect data to obtain wire diameter compensation effect data; performing multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index;

[0097] Step S5: Score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain electromagnetic coil quality score data.

[0098] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of an electromagnetic coil intelligent production control method of the present invention. In this embodiment, the electromagnetic coil intelligent production control method includes the following steps:

[0099] Step S1: performing coil geometric feature recognition on the electromagnetic coil design drawing to obtain coil topology data; performing key process parameter supplementation processing based on the coil topology data to obtain key coil process parameters; performing coil digitization processing on the coil topology data using the key coil process parameters to generate a three-dimensional coil digital model;

[0100] In an embodiment of the present invention, computer-aided design (CAD) software is used to identify the geometric features of the electromagnetic coil design drawing. Edge detection and shape analysis algorithms are used to extract the coil's contour, cross-sectional shape, number of turns, and other geometric features to generate the coil's topological structure data. Then, based on design specifications and process requirements, key process parameters such as winding tension, wire diameter, and insulation grade are added. Next, parametric modeling technology is used to integrate the topological structure data and process parameters to construct a three-dimensional digital model of the coil. For example, for a cylindrical coil, a three-dimensional solid model can be generated through rotational scanning, and information such as material properties and winding direction can be added.

[0101] Step S2: performing coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; performing voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; performing adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data;

[0102] In an embodiment of the present invention, a three-dimensional coil digital model is voxelized, discretizing the continuous geometric body into a uniform cubic grid. A graph theory algorithm is then used to perform topological analysis on the voxelized model, generating a network structure describing the coil winding path. Next, a shortest path algorithm is used to perform trajectory planning on the voxel network to obtain a preliminary coil winding trajectory. For example, for a multi-layer coil, the winding path for each layer can be planned first, and then the transition trajectory between layers can be determined. Subsequently, the preliminary trajectory is locally optimized and smoothed based on actual process constraints, such as minimum bending radius and obstacle avoidance. The entire trajectory is then globally optimized to obtain the optimal winding trajectory that meets all constraints.

[0103] Step S3: Based on the optimized coil winding trajectory data, intelligent manufacturing is performed using an intelligent electromagnetic coil production device, and real-time coil winding image acquisition is performed to generate real-time coil winding image data; based on the real-time coil winding image data, a turn layer wire diameter change trend is predicted to generate wire diameter change trend data;

[0104] In an embodiment of the present invention, the optimized coil winding trajectory data is input into an intelligent electromagnetic coil production device. The device is equipped with a high-precision servo motor and a tension control system, which can accurately execute complex three-dimensional winding trajectories. During the winding process, a high-speed camera is used to capture the coil forming image in real time to form an image sequence. Then, a computer vision algorithm is used to process the image sequence to extract the real-time change data of the wire diameter of each turn. Then, a time series analysis method, such as an autoregressive moving average (ARMA) model, is used to model and predict the wire diameter change data to obtain the wire diameter change trend for the next few turns. For example, it can be predicted that the coil wire diameter will increase slightly during the subsequent winding process.

[0105] Step S4: performing tension feedback control strategy processing based on the wire diameter change trend data to obtain tension feedback control data; performing wire diameter consistency compensation control based on the tension feedback control data, and collecting wire diameter compensation control effect data to obtain wire diameter compensation effect data; performing multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index;

[0106] In an embodiment of the present invention, an adaptive tension feedback control strategy is designed based on wire diameter trend data. A fuzzy PID control algorithm is employed to dynamically adjust the tension setpoint based on the predicted wire diameter trend. For example, when the wire diameter is predicted to increase, the tension is appropriately increased to suppress expansion. The tension adjustment instructions output by the controller are converted into specific execution instructions by a digital signal processor (DSP), driving the tension control device to adjust the winding tension in real time. For example, if a particular wire turn is detected to have an excessively large diameter, the system immediately increases the tension to restore the diameter of the electromagnetic coil winding turn to normal.

[0107] Step S5: Score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain electromagnetic coil quality score data.

[0108] In an embodiment of the present invention, a quality scoring strategy for electromagnetic coils is established based on the multi-layer wire diameter consistency index combined with other key quality parameters, such as insulation performance and resistance. The analytic hierarchy process (AHP) is used to determine the weights of each indicator and construct a comprehensive scoring function. For example, the wire diameter consistency index can be weighted 50%, insulation performance can be weighted 30%, and resistance can be weighted 20%. A weighted sum is taken based on the weights to obtain the final quality score of the coil.

[0109] Preferably, step S1 includes the following steps:

[0110] Step S11: performing image denoising preprocessing on the electromagnetic coil design drawing to generate a standard electromagnetic coil design drawing; extracting original drawing annotations based on the standard electromagnetic coil design drawing to obtain original drawing annotation data;

[0111] Step S12: performing multi-level drawing analysis based on the standard electromagnetic coil design drawing and performing vector graphics analysis to generate coil vector structure data;

[0112] Step S13: performing coil geometric feature recognition on the coil vector structure data using preset geometric constraint relationship rules to obtain coil topology structure data;

[0113] Step S14: performing graphic element annotation processing on the coil topology structure data using the original drawing annotation data, and establishing a mapping relationship between graphic elements and parameters to generate preliminary topology graphic element annotation data;

[0114] Step S15: performing key process parameter supplementation processing based on the preliminary topological primitive annotation data and the coil topological structure data to obtain key coil process parameters, wherein the key coil structure parameters include coil material attribute data, coil winding attribute data, and coil electrical attribute data;

[0115] Step S16: digitize the key coil structure parameters using the coil topology data to generate a three-dimensional coil digital model.

[0116] In an embodiment of the present invention, an electromagnetic coil design image file is read using image processing software such as OpenCV. The image is then filtered using a median filter algorithm to remove noise. The filtered image is then binarized using an adaptive threshold segmentation algorithm to generate a standard black-and-white binary image. Morphological operations such as opening and closing are used to refine and smooth the binary image to obtain a standardized electromagnetic coil design drawing. Based on the standard electromagnetic coil design drawing, optical character recognition (OCR) technology is used to extract text annotation information from the drawing to generate original drawing annotation data. An image segmentation algorithm such as the Watershed algorithm is used to segment the image to separate the coil graphic from other elements. The segmented coil graphic is then vectorized and converted into vector data using a vector tracking algorithm such as the Potrace algorithm. The various components of the coil, such as the coil windings and leads, are identified and their topological relationships are established. The coil vector structure data is analyzed based on pre-defined geometric constraints to extract geometric features of the coil windings, such as the coil diameter and the number of winding layers. The coil winding topology is then identified based on the geometric characteristics of the coil winding, combined with pre-defined constraints such as coil layer spacing and coil end curvature radius. Next, the geometric characteristics of the coil leads, including lead length and bend angle, are analyzed. Based on the positional relationship between the leads and the coil winding, the topological connection between them is determined. The textual information in the original drawing annotation data is preprocessed, such as removing irrelevant characters and standardizing the annotation format. Then, based on pre-established annotation rules, the processed annotation information is matched and mapped to the corresponding primitives in the coil topology data. For example, the annotation "Coil diameter: 10mm" is mapped to the corresponding coil winding primitive. A mapping relationship between primitives and parameters is established, generating preliminary topological primitive annotation data. For example, based on the coil winding's geometric dimensions, a coil process specification library is queried to supplement parameters such as the winding method (manual or machine) and winding tension. Furthermore, based on the coil lead length and bend angle, parameters such as the lead forming method (stamping or bending) are supplemented. For some parameters that cannot be obtained directly from the drawings, such as the electrical properties of the coil, it is necessary to calculate based on existing parameters or consult the database to obtain them. The coil is digitally modeled in three dimensions. First, based on the coil topology data, the basic framework of the coil is constructed, including the three-dimensional geometry of the coil winding, the three-dimensional path of the lead wire, etc. Then, the coil material property data is applied to the model, and corresponding material properties are assigned to different parts of the coil, such as conductor material, insulation material, etc. Next, based on the coil winding property data, the coil winding process is simulated to generate an accurate winding model.

[0117] Preferably, step S15 includes the following steps:

[0118] Step S151: Analyze the electrical performance of the target coil based on the preliminary topological element annotation data to generate coil electrical property data;

[0119] Step S152: extracting coil winding geometric parameters based on the coil topology data to generate coil winding geometric parameters, wherein the coil winding geometric parameters include coil wire diameter, coil outer diameter, coil inner diameter, and coil winding height or length;

[0120] Step S153: performing wire material property matching on the coil electrical property data and the coil winding geometric parameters through a preset material database to generate coil material property data;

[0121] Step S154: Calculate the number of coil layers according to the coil winding geometric parameters to generate coil layer data;

[0122] Step S155: using the coil material attribute data to estimate the coil winding geometric parameters, and generating coil turn estimation data; wherein the coil turn estimation is calculated using a coil turn estimation formula, which is as follows:

[0123]

[0124] Where N is the estimated number of coil turns, V is the volume of the coil winding area, and A w It is expressed as the cross-sectional area of a single wire diameter, π is pi, D is the average winding diameter of the coil, d is the wire diameter, and α is the winding process coefficient;

[0125] Step S156: identifying the coil winding mode of the coil topology data using the coil layer number data and the coil turn number estimation data to generate coil winding mode category data;

[0126] Step S157: integrating the coil winding type data, the coil layer number data, and the coil turn number estimation data into winding data to generate coil winding attribute data.

[0127] In an embodiment of the present invention, the preliminary topological element annotation data includes information such as coil voltage, current, and resistance extracted from the design drawings, representing the coil's electrical performance indicators such as inductance and AC resistance at different operating frequencies. The outer contour of the coil winding is determined to calculate the coil's outer diameter. The boundary of the center hole is then identified from within the winding to calculate the coil's inner diameter. Next, the winding height or length is measured based on the winding's three-dimensional shape. Wire diameter parameters are obtained from the original drawing annotation data or a material database. A pre-established material database is queried to match the appropriate wire material properties. The material database contains electrical, physical, and mechanical performance parameters for various wire materials (such as copper and aluminum) at different specifications. First, based on the coil's electrical requirements (such as operating frequency and current load), materials that meet the electrical performance requirements are selected from the database. Based on the coil's outer and inner diameters, the average diameter of the coil winding is calculated. The average diameter is then divided by the wire diameter (wire diameter) to obtain the theoretical number of coil turns. The coil winding height or length is then divided by the wire diameter to obtain the theoretical number of layers. The wire diameter d is extracted from the coil material attribute data. Then, according to the coil winding geometric parameters, the value of the coil average winding diameter D is calculated. Then, the volume V of the coil winding area and the cross-sectional area A of a single wire are calculated. w Substitute this into the formula. For the winding process coefficient α, an empirical value can be assigned based on the actual process level. During the estimation process, some special cases need to be considered. For example, when the coil is rectangular or other non-circular, the formula needs to be modified accordingly. Based on the estimated data for the number of coil layers and turns, combined with a preset coil winding method rule library, the coil winding method can be identified. For example, if the number of layers is 1, the winding method is single-layer winding; if the number of layers is greater than 1 and the number of turns per layer is equal, the winding method is multi-layer dense winding; if the number of layers is greater than 1 and the number of turns per layer is unequal, the winding method is multi-layer random winding. The identified winding method category data is stored in a database, such as "single-layer winding," "multi-layer dense winding," or "multi-layer random winding." The coil layer data, coil turn estimation data, and coil winding method category data are integrated to form complete coil winding attribute data.

[0128] Preferably, step S2 includes the following steps:

[0129] Step S21: performing finite element meshing processing on the three-dimensional coil digital model to generate a finite element coil digital model;

[0130] Step S22: performing coil voxel winding network analysis based on the finite element coil digital model to generate voxel winding topology network data;

[0131] Step S23: performing a windability analysis on the voxel winding topology network data and performing voxel winding trajectory mapping to obtain preliminary coil winding trajectory data;

[0132] Step S24: Decomposing the winding path according to the preliminary coil winding trajectory data to generate a coil winding sub-path set;

[0133] Step S25: performing path conflict detection on the coil winding sub-path set to obtain winding conflict sub-path data;

[0134] Step S26: performing magnetic field constraint analysis on the preliminary coil winding trajectory data using a finite element coil digital model based on a preset magnetic field uniformity distribution strategy, and performing finite element magnetic field distribution simulation calculation to generate coil magnetic field constraint simulation data;

[0135] Step S27: identifying the magnetic field inhomogeneity region according to the coil magnetic field constraint simulation data, and adaptively adjusting the preliminary coil winding trajectory data through the winding conflict sub-path data to obtain optimized coil winding trajectory data.

[0136] In an embodiment of the present invention, the coil model is meshed by setting an appropriate mesh size and type based on the complexity of the coil structure and the required simulation accuracy. For example, a regular hexahedral mesh can be used for coil regions with regular shapes, while a more adaptable tetrahedral mesh can be used for coil regions with complex shapes. During the meshing process, it is important to properly refine the mesh in key areas (such as the coil windings and lead bends) to ensure computational accuracy. The boundaries of the coil winding area are determined and divided into small voxels (3D pixels). Each voxel represents a tiny spatial unit of the coil winding and can store attribute information related to it, such as whether it is windable and the winding layer it belongs to. Then, based on the coil topology, the connectivity between each voxel is analyzed to establish a topological network between the voxels. For example, a graph data structure from graph theory can be used to represent the voxel winding topological network, where voxels are nodes and the connections between adjacent voxels are edges. Based on preset windability rules, such as the minimum curvature radius and the minimum inter-layer spacing, non-windable voxels in the topological network are marked. Then, a preliminary winding path is constructed based on the windable voxels, determining the starting and ending points of the winding and its path within the voxel network. During path construction, the coil's topological constraints must be considered, such as avoiding non-windable areas and meeting interlayer spacing requirements. Examples include voxels with excessively narrow spaces or small bending radii. Based on pre-defined winding strategies, such as shortest path or uniform distribution, a winding path that meets the requirements is searched for within the windable voxel network. This winding path is then mapped to the actual 3D coordinate system to generate preliminary coil winding trajectory data. Based on coil winding process requirements, such as winding machine reversal and wire breakage reconnection, the preliminary coil winding trajectory data is decomposed into multiple, continuously executable sub-paths. For example, the winding path can be segmented based on changes in winding direction or number of winding layers, with each sub-path representing an independent winding action. All coil winding sub-paths are then compared pairwise to detect conflicts, such as path intersections and collisions. For example, the minimum distance between two sub-paths can be calculated to determine whether it is less than a preset safety distance threshold. The conflicting sub-path pairs are recorded to form winding conflict sub-path data. For example, the maximum deviation and gradient of the magnetic field intensity in the target area can be set. The preliminary coil winding trajectory data is imported into the finite element analysis software, and the magnetic field distribution simulation calculation is performed in combination with the finite element coil digital model. For example, the finite element method can be used to solve Maxwell's equations to obtain parameters such as the magnetic field intensity and magnetic induction intensity in the space around the coil. According to the preset magnetic field uniformity distribution strategy, the coil magnetic field constraint simulation data is analyzed to identify areas of magnetic field inhomogeneity, such as areas where the magnetic field intensity deviation is too large or the gradient is too high. Combined with the winding conflict sub-path data, the preliminary coil winding trajectory data is adjusted to improve the magnetic field uniformity.For example, the order, direction, shape, and other parameters of the conflicting winding sub-paths can be adjusted to avoid magnetic field-sensitive areas, or the winding density can be increased to enhance the local magnetic field strength. Ultimately, the optimized coil winding trajectory data is obtained. For example, the winding order of one sub-path can be advanced so that it is wound before another sub-path to avoid the coil winding density in this area being too low; the winding direction of one sub-path can be reversed to use the opposite magnetic field generated by the current to enhance the magnetic field strength in this area; the shape of one sub-path can be fine-tuned, such as increasing the number of winding turns or changing the curvature of the winding path to increase the winding density in this area.

[0137] Preferably, step S22 includes the following steps:

[0138] Step S221: constructing a four-dimensional voxel space for the finite element coil digital model to generate a coil four-dimensional voxel space, wherein three spatial dimensions in the coil four-dimensional voxel space correspond to the length, width, and height of the collar, and the fourth dimension corresponds to the number of layers of the coil;

[0139] Step S222: performing tensor element initialization processing according to the coil four-dimensional voxel space to generate an initialized four-dimensional tensor value;

[0140] Step S223: Mark the voxel occupancy of the coil's four-dimensional voxel space by initializing the four-dimensional tensor value to generate coil occupancy voxel data. For each voxel, determine whether its center point is within the geometric range of the layer. If so, iteratively update the tensor value of the voxel space to 1, indicating that it is occupied. Otherwise, the tensor value of the voxel space remains unchanged.

[0141] Step S224: performing connected domain component analysis on the coil four-dimensional voxel space using a breadth-first search algorithm based on the coil occupied voxel data to generate coil voxel connected domain data;

[0142] Step S225: performing adjacent layer overlap analysis based on the coil voxel connected domain data to obtain a directed acyclic graph of the voxel connected layer;

[0143] Step S226: performing discrete time series calculation on the voxel connected layer directed acyclic graph using a preset winding speed, and assigning a timestamp step size to generate voxel winding topology network data.

[0144] In an embodiment of the present invention, a four-dimensional voxel space is constructed. The first three dimensions correspond to the length, width, and height of the coil, dividing the three-dimensional geometric space of the coil into a large number of small cubes, namely three-dimensional voxels. The fourth dimension corresponds to the number of winding layers of the coil, that is, each layer of winding is further subdivided into multiple two-dimensional layers along the height direction. Through this four-dimensional voxel discretization, the topological structure of the coil winding can be accurately described. A four-dimensional tensor of the same size as the four-dimensional voxel space of the coil is created to store the occupancy status information of each voxel. In the initial state, the values of all tensor elements are set to 0, indicating that all voxels are unoccupied. Each voxel in the four-dimensional voxel space of the coil is traversed. For each voxel, its four-dimensional coordinates (x, y, z, l) are used to determine whether its center point is within the geometric range of the coil of that layer (l). The judgment method can be selected according to the shape of the coil. For example, for a circular coil, the distance from the voxel center point to the coil center is calculated to determine whether it is less than or equal to the coil radius. For a rectangular coil, the x and y coordinates of the voxel center point are determined whether they are within the boundary range of the coil. If the voxel center point falls within the geometric range of the coil layer, the tensor element corresponding to that voxel is updated to 1, indicating that the voxel is occupied by the coil; otherwise, the tensor element remains at 0. Using the coil occupied voxel data as input, a breadth-first search (BFS) algorithm is used to perform a connected domain analysis in the four-dimensional voxel space. Starting from any occupied voxel, its six adjacent voxels (up, down, left, right, front, and back) are searched. If an adjacent voxel is also occupied, it is added to the current connected domain, and the search continues until all adjacent voxels have been searched. All searched voxels are marked as belonging to the same connected domain, and the domain number, number of voxels contained, and other information are recorded. The above steps are repeated until all occupied voxels are marked in the corresponding connected domain. All connected domains of coil voxels are traversed, and the overlap between adjacent layers is analyzed. If two connected domains belonging to different layers have overlapping voxels, a directed edge is established between the two connected domains, from the connected domain with the lower layer to the connected domain with the higher layer. The result is a voxel-connected layer directed acyclic graph (DAG), where nodes represent connected domains at different layers, and directed edges represent the winding order between layers. The graph is discretely timed using a preset winding speed, and each node (connected domain) is assigned a reasonable timestamp step value. First, the starting connected domain for coil winding is determined, and its timestamp is set to an initial value (e.g., 0). Then, based on the preset winding speed, the time increment required to wind from the starting connected domain to the adjacent connected domain is calculated. This increment is assigned to the directed edges between the starting and adjacent connected domains. By traversing all edges and gradually accumulating the time increments, the theoretical timestamp value for each connected domain can be obtained. The calculated timestamp step value is assigned to the corresponding connected domain, and together with the spatial position information of the connected domain, it forms the complete voxel winding topology network data.

[0145] Preferably, step S3 includes the following steps:

[0146] Step S31: calibrating the control path points of the optimized coil winding trajectory data using the three-dimensional coil digital model to generate winding path control point data;

[0147] Step S32: performing path point type identification on the winding path control point data to generate winding path point type data, wherein the winding path point type data includes a starting point, an end point, a turning point, an intersection point, and a smoothing point;

[0148] Step S33: encoding the optimized coil winding trajectory data into processing control instructions using the winding path point type data, and distributing the servo control instructions to obtain coil servo control instruction data;

[0149] Step S34: performing intelligent manufacturing execution using an intelligent electromagnetic coil production device based on the coil servo control instruction data, and using a high-speed industrial camera to perform real-time coil winding image acquisition to generate real-time coil winding image data;

[0150] Step S35: predicting the wire diameter change trend of the turn layer according to the real-time coil winding image data, and generating wire diameter change trend data.

[0151] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S3 are shown in the flowchart. In this example, step S3 includes:

[0152] Step S31: calibrating the control path points of the optimized coil winding trajectory data using the three-dimensional coil digital model to generate winding path control point data;

[0153] In an embodiment of the present invention, within a three-dimensional coil digital model, optimized coil winding trajectory data is sampled based on preset control accuracy and the winding machine's motion characteristics, extracting a series of key control path points. For example, different sampling intervals can be set based on trajectory characteristics such as changes in curvature, number of winding layers, and speed. In areas with greater curvature, varying number of layers, or varying speed, a tighter sampling interval is used to ensure winding accuracy. Each control path point contains information such as three-dimensional coordinates, winding speed, and winding direction, forming winding path control point data.

[0154] Step S32: performing path point type identification on the winding path control point data to generate winding path point type data, wherein the winding path point type data includes a starting point, an end point, a turning point, an intersection point, and a smoothing point;

[0155] In an embodiment of the present invention, the winding path control point data is analyzed based on the geometric features and motion features of the control path points in the winding trajectory to identify different types of path points. For example, the starting point: the starting point of the winding trajectory, corresponding to the first control point where the winding machine starts winding; the ending point: the end point of the winding trajectory, corresponding to the last control point where the winding machine completes winding; the turning point: the point where the direction of the winding trajectory changes, such as from horizontal winding to vertical winding; the intersection point: the point where different layers of winding trajectories intersect, which requires avoidance or crossing processing; the smooth point: the point on the winding trajectory where the curvature changes less, and continuous motion control can be used. For each control path point, based on its position on the trajectory, the distribution of adjacent points, etc., pre-set recognition rules are applied to perform type judgment and marking, such as when adjacent points are too close or the trajectory has self-intersections, to ensure the accuracy of recognition.

[0156] Step S33: encoding the optimized coil winding trajectory data into processing control instructions using the winding path point type data, and distributing the servo control instructions to obtain coil servo control instruction data;

[0157] In an embodiment of the present invention, corresponding motion control instructions are determined according to different types of control path points, such as linear interpolation, circular interpolation, precise positioning, etc. Then, these control instructions are encoded in the order of trajectory points to generate a series of machining program codes. Starting point: Use the G00 rapid positioning instruction to move the winding head to the starting point position; End point: Use the M30 program end instruction to stop the winding machine; Turning point: Use the G01 linear interpolation instruction to control the winding head to change direction; Intersection point: Depending on the specific situation of the intersection, use the G01 instruction to avoid, or use the G04 pause instruction to cross over; Smoothing point: Use the G01 instruction to control the winding head for continuous winding. The generated machining control instruction codes are sorted in chronological order, and corresponding servo control parameters are added, such as acceleration and deceleration control, position feedback control, etc., to form complete coil servo control instruction data.

[0158] Step S34: performing intelligent manufacturing execution using an intelligent electromagnetic coil production device based on the coil servo control instruction data, and using a high-speed industrial camera to perform real-time coil winding image acquisition to generate real-time coil winding image data;

[0159] In an embodiment of the present invention, an intelligent electromagnetic coil production device receives coil servo control instruction data and drives the winding machine to perform the winding operation according to the instruction. At the same time, a high-speed industrial camera captures the winding process in real time at a preset frame rate and transmits the captured image data to an image processing system to form real-time coil winding image data. For example, four 20-megapixel linear array CCD industrial cameras are used, installed at four angles of the coil winding machine, to ensure that clear images of the coil winding area can be captured. The camera acquisition rate needs to be high enough to capture the dynamic process of coil winding.

[0160] Step S35: predicting the wire diameter change trend of the turn layer according to the real-time coil winding image data, and generating wire diameter change trend data.

[0161] In an embodiment of the present invention, the image is pre-processed, such as denoising and enhancement, to improve the image quality. Then, edge detection, image segmentation and other algorithms are applied to accurately extract the coil winding area from the image. Next, pattern recognition technology is used to identify and measure the wire diameter size in the extracted winding area. By tracking and analyzing the changing trend of the wire diameter size in continuous image frames, the change of the wire diameter in the future period can be predicted. For example, algorithms such as image edge detection and circular fitting can be used to identify the outline of the coil and calculate the diameter of the coil. Based on multiple frames of image data, the changing trend of the coil diameter over time or the number of winding layers is analyzed, such as linear growth, exponential growth, etc., and the diameter change trend of the next few turns of the coil is predicted to generate wire diameter change trend data.

[0162] Preferably, step S35 includes the following steps:

[0163] Step S351: using an edge detection operator to perform real-time coil winding layer edge detection on the real-time coil winding image data, and performing dynamic edge pixel focusing to generate winding layer edge mask data;

[0164] Step S352: performing pixel-level motion estimation of edges of adjacent frames based on the winding layer edge mask data, and performing continuous frame displacement vector superposition to obtain real-time winding layer edge trajectory data;

[0165] Step S353: performing sub-pixel edge positioning processing on the real-time winding coil edge trajectory data to generate a high-precision coil edge coordinate set;

[0166] Step S354: performing time-series multi-frame image fitting based on the high-precision circle edge coordinate set to obtain dynamic line diameter curve fitting data;

[0167] Step S355: performing spectrum analysis on the dynamic wire diameter curve fitting data to generate wire diameter fluctuation characteristic vector data;

[0168] Step S356: Based on the preset different wire winding fluctuation data and the corresponding future wire diameter change trend label data, a long short-term memory network model is used to perform transfer learning processing to obtain a wire diameter prediction model;

[0169] Step S357: transmitting the wire diameter fluctuation characteristic vector data to the wire diameter prediction model to predict the wire diameter change trend of the next turn layer, and generating wire diameter change trend data.

[0170] In an embodiment of the present invention, the original image is subjected to Gaussian filtering for denoising. The image gradient amplitude and direction are then calculated, and non-maximum suppression is performed on the gradient amplitude. Finally, a double-threshold detection and edge connection method are used to obtain the final set of edge pixels. During the detection process, operator parameters, such as the Gaussian kernel size and threshold, need to be adjusted based on the characteristics of the coil winding image to achieve optimal results. For example, the Lucas-Kanade optical flow method can be used, combined with a pyramid multi-scale strategy and Gaussian filtering, to accurately estimate small movements of the coil edge. Displacement vectors from consecutive frames are superimposed to obtain the motion trajectory of each edge pixel over a time series, forming real-time winding layer edge trajectory data. For example, a sub-pixel edge location method based on grayscale moments can be used to accurately locate the edge position at the sub-pixel level by calculating the grayscale moment information within the edge pixel neighborhood. Ultimately, the sub-pixel coordinates of the coil edge at each moment are obtained, forming a high-precision coil edge coordinate set. For example, methods such as least squares fitting or spline interpolation can be used to fit the edge coordinate points to obtain a functional expression for the coil edge. This function can calculate geometric parameters such as the coil diameter and circumference at any given moment, generating dynamic wire diameter curve fitting data. For example, the wire diameter data sequence can be decomposed into a superposition of sinusoidal waves of different frequencies, and the amplitude and phase of each frequency component can be analyzed. The extracted frequency domain characteristic parameters are combined into a vector form to generate wire diameter fluctuation feature vector data. Pre-set fluctuation data for different wire winding processes is collected, including actual wire diameter fluctuations under various process parameters. This fluctuation data is then paired with corresponding label data for future wire diameter change trends to serve as the model training dataset. For example, wire diameter change trends can be categorized as stable, thickening, or thinning, with each data sample annotated with the corresponding category label. Next, a long short-term memory (LSTM) network is selected as the model architecture. Using transfer learning, the LSTM model is pre-trained on a large amount of wire diameter fluctuation training data to generate an initial wire diameter prediction model. Based on the input feature vector, the model outputs a prediction of the wire diameter change trend for the next turn layer, for example, predicting whether the wire diameter will thicken, thin, or remain stable, and provides a confidence level for each prediction. The prediction results and confidence information are integrated to generate wire diameter change trend data.

[0171] Preferably, step S4 includes the following steps:

[0172] Step S41: performing wire diameter trend quantification processing according to the wire diameter change trend data to generate quantified wire diameter change rate data;

[0173] Step S42: performing eigenmode analysis on the quantized wire diameter change rate data to obtain wire diameter transformation characteristic data;

[0174] Step S43: Calculating the trend Gaussian cumulative distribution of the wire diameter transformation feature data using a preset wire diameter change threshold, and determining the wire diameter change level. When the wire diameter change level is lower than the wire diameter change threshold, the winding layer is marked as a stable winding layer; when the wire diameter change level is higher than or equal to the wire diameter change threshold, the winding layer is marked as an abnormal winding layer.

[0175] Step S44: performing tension feedback control strategy processing on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data based on the abnormal winding layer to obtain tension feedback control data;

[0176] Step S45: Using the intelligent electromagnetic coil production device to perform real-time wire diameter consistency compensation control on the next turn layer of the coil based on the tension feedback control data, and collecting the wire diameter compensation control effect to obtain wire diameter compensation effect data;

[0177] Step S46: performing multi-layer wire diameter consistency calculation based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index.

[0178] As an example of the present invention, refer to Figure 3 As shown, Figure 1 Detailed implementation steps of step S4 are shown in the flowchart. In this example, step S4 includes:

[0179] Step S41: performing wire diameter trend quantification processing according to the wire diameter change trend data to generate quantified wire diameter change rate data;

[0180] Step S42: performing eigenmode analysis on the quantized wire diameter change rate data to obtain wire diameter transformation characteristic data;

[0181] Step S43: Calculating the trend Gaussian cumulative distribution of the wire diameter transformation feature data using a preset wire diameter change threshold, and determining the wire diameter change level. When the wire diameter change level is lower than the wire diameter change threshold, the winding layer is marked as a stable winding layer; when the wire diameter change level is higher than or equal to the wire diameter change threshold, the winding layer is marked as an abnormal winding layer.

[0182] Step S44: performing tension feedback control strategy processing on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data based on the abnormal winding layer to obtain tension feedback control data;

[0183] Step S45: Using the intelligent electromagnetic coil production device to perform real-time wire diameter consistency compensation control on the next turn layer of the coil based on the tension feedback control data, and collecting the wire diameter compensation control effect to obtain wire diameter compensation effect data;

[0184] Step S46: performing multi-layer wire diameter consistency calculation based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index.

[0185] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S4 are shown in the flowchart. In this example, step S4 includes:

[0186] Step S41: performing wire diameter trend quantification processing according to the wire diameter change trend data to generate quantified wire diameter change rate data;

[0187] In an embodiment of the present invention, first, a quantization level of the wire diameter change rate is set, such as 5% as a quantization level. Then, the wire diameter change trend data is segmented, and the wire diameter change rate in each time period is calculated. The calculated change rate value is mapped to the preset quantization level, thereby obtaining the quantized wire diameter change rate of the time period. For example, the predicted diameter value of the coil for the next few turns is compared with the current coil diameter to calculate the wire diameter change rate. The wire diameter change rate can be quantized, for example, it can be divided into several levels, such as: rapid thickening: the wire diameter change rate is greater than the preset threshold A; slow thickening: the wire diameter change rate is between the preset threshold B and A; basically stable: the wire diameter change rate is between the preset threshold C and B; slow thinning: the wire diameter change rate is between the preset threshold D and C; rapid thinning: the wire diameter change rate is less than the preset threshold D.

[0188] Step S42: performing eigenmode analysis on the quantized wire diameter change rate data to obtain wire diameter transformation characteristic data;

[0189] In an embodiment of the present invention, for example, the quantized wire diameter change rate data of a multi-turn coil can be constructed into a data matrix, and the covariance matrix of the quantized wire diameter change rate data can be constructed. Then, the covariance matrix is subjected to eigenvalue decomposition to obtain a series of eigenvalues and corresponding eigenvectors. These eigenvalues reflect the importance of each pattern in the data, while the eigenvectors describe the specific form of each pattern. For example, the quantized wire diameter change rate data of a multi-turn coil can be constructed into a data matrix, and then the matrix is subjected to singular value decomposition to obtain eigenvalues and eigenvectors. The first few eigenvectors with the largest contribution rate are selected as the wire diameter transformation feature data, which are used to characterize the main trends and patterns of wire diameter changes.

[0190] Step S43: Calculating the trend Gaussian cumulative distribution of the wire diameter transformation feature data using a preset wire diameter change threshold, and determining the wire diameter change level. When the wire diameter change level is lower than the wire diameter change threshold, the winding layer is marked as a stable winding layer; when the wire diameter change level is higher than or equal to the wire diameter change threshold, the winding layer is marked as an abnormal winding layer.

[0191] In an embodiment of the present invention, the wire diameter variation level of the current winding layer is determined. First, the Gaussian cumulative distribution function value of the wire diameter variation of the layer is calculated based on the wire diameter transformation characteristic data. This function value reflects the degree of deviation of the wire diameter variation amplitude from a normal distribution. Then, this function value is compared with a preset wire diameter variation threshold. If the function value is lower than the threshold, the current winding layer is marked as a stable winding layer, indicating that its wire diameter variation is within a controllable range. Conversely, if the function value is higher than or equal to the threshold, the current winding layer is marked as an abnormal winding layer, requiring compensation and adjustment of winding parameters. For example, the wire diameter variation level can be divided into three levels: mild, moderate, and severe, based on factors such as the amplitude, speed, and duration of the wire diameter variation. If the calculated wire diameter variation level is lower than the preset threshold, the wire diameter variation of the winding layer is considered to be within a controllable range and the layer is marked as a stable winding layer. Otherwise, the wire diameter variation of the winding layer is considered to affect coil performance and the layer is marked as an abnormal winding layer.

[0192] Step S44: performing tension feedback control strategy processing on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data based on the abnormal winding layer to obtain tension feedback control data;

[0193] In an embodiment of the present invention, the selected control strategy is applied to the optimized coil winding trajectory data, the trajectory of the abnormal winding layer is compensated and adjusted, and tension feedback control data is generated. For example, if the cause of the abnormality is excessive wire feeding tension, a buffer section can be added to the trajectory of the abnormal layer to reduce the local feed speed, thereby reducing the tension; if the cause of the abnormality is temperature change, the winding sequence around the abnormal layer can be adjusted to avoid temperature accumulation. The coil winding trajectory data is optimized for adjustment, and tension feedback control data is generated. For example: if the wire diameter change trend is getting thicker, the winding tension can be reduced, or the winding speed can be increased. If the wire diameter change trend is getting thinner, the winding tension can be increased, or the winding speed can be reduced. The tension feedback control strategy can be adjusted according to the specific wire diameter change and control target, such as PID control, fuzzy control, etc. The tension feedback control data contains information such as the adjusted winding tension and winding speed.

[0194] Step S45: Using the intelligent electromagnetic coil production device to perform real-time wire diameter consistency compensation control on the next turn layer of the coil based on the tension feedback control data, and collecting the wire diameter compensation control effect to obtain wire diameter compensation effect data;

[0195] In this embodiment of the present invention, upon receiving tension feedback control data, the intelligent electromagnetic coil production device adjusts the winding machine's tension and speed control systems in real time to wind the next coil turn. Simultaneously, a high-speed industrial camera continues to capture image data of the winding process and transmits it to an image processing system. Based on this real-time image data, the image processing system calculates the change in wire diameter after wire diameter compensation, such as the wire diameter change rate and wire diameter deviation, generating data on the wire diameter compensation effect.

[0196] Step S46: performing multi-layer wire diameter consistency calculation based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index.

[0197] In an embodiment of the present invention, the wire diameter difference between different layers of the same coil is calculated to obtain the inter-layer wire diameter deviation value. Taking into account the presence of certain noise in the wire diameter measurement, the deviation value can be statistically filtered, such as median filtering, to eliminate the influence of wild values. The statistical quantities such as the mean, variance, and range of the wire diameter deviation values of all layers are further calculated to construct an evaluation index that comprehensively describes the wire diameter consistency. The wire diameter consistency index is converted into a multi-layer wire diameter consistency index between 0 and 1. The higher the index, the better the wire diameter consistency. For example, the multi-layer wire diameter consistency index can be calculated using the following formula: consistency index = 1-(wire diameter standard deviation / average wire diameter).

[0198] Preferably, step S44 includes the following steps:

[0199] Step S441: performing preset trajectory deviation calculation on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data to generate a winding trajectory deviation value;

[0200] Step S442: performing deviation time series analysis on the winding trajectory deviation value through a preset winding time window to obtain dynamic deviation time series characteristic data;

[0201] Step S443: Calculating fuzzy membership based on the dynamic deviation time series feature data to obtain a deviation fuzzy feature matrix;

[0202] Step S444: using the tension sensor to obtain real-time tension monitoring data; performing fuzzy inference calculation on the real-time tension monitoring data using a preset tension fuzzy control rule library to generate fuzzy tension control data;

[0203] Step S445: Defuzzifying the deviation fuzzy feature matrix using the tension fuzzy control rule library based on the fuzzy tension control data to obtain precise tension adjustment amount sequence data;

[0204] Step S446: Adaptively adjust control parameters according to the precise tension adjustment amount sequence data to generate tension feedback control data.

[0205] In an embodiment of the present invention, the deviation value between the actual winding trajectory and the expected trajectory is calculated based on a pre-set wire diameter-trajectory mapping relationship. For example, the wire diameter change can be mapped to the adjustment amount of the winding radius, and a thicker wire diameter corresponds to an increase in the winding radius, and a thinner wire diameter corresponds to a decrease in the winding radius. According to the size of the wire diameter change, the corresponding winding radius adjustment amount is calculated, which is the winding trajectory deviation value. A winding time window is set, for example, including the trajectory deviation values of the past 5 winding cycles. The trajectory deviation value of the current winding cycle is combined with the historical deviation value in the time window to form dynamic deviation time series feature data. For example, a sliding window method can be used to form a 5-dimensional vector with the trajectory deviation value of each time point and the deviation values of the previous 4 time points as the dynamic deviation time series feature data of the time point. According to the pre-set fuzzy rules, the dynamic deviation time series feature data is fuzzified, and the membership of each feature value to different fuzzy sets is calculated. For example, the trajectory deviation values can be divided into five fuzzy sets: "negative large," "negative small," "zero," "positive small," and "positive large." A triangular membership function or a Gaussian membership function is used to calculate the membership of each eigenvalue to each fuzzy set. The memberships of all eigenvalues are combined into a matrix to obtain the deviation fuzzy feature matrix. A tension sensor installed on the winding machine monitors tension changes in real time during the winding process and transmits the data to the control system. A tension fuzzy control rule base is pre-established. Based on the real-time tension monitoring data and preset tension fuzzy sets, such as "too small," "moderate," and "too large," the membership of the tension value to each fuzzy set is calculated. Fuzzy inference mechanisms, such as Mamdani or Sugeno reasoning, are then used to infer the tension fuzzy control rule base and obtain fuzzy tension control data, such as "reduce tension," "maintain tension unchanged," and "increase tension." For example, based on the memberships of the fuzzy sets in the deviation fuzzy feature matrix, the rules in the fuzzy tension control rule base can be weighted averaged to obtain the final tension adjustment value. For more precise control, the tension adjustment can be set as a sequence, for example, containing the tension adjustment values for the next five winding cycles, to dynamically compensate for changes in wire diameter. For example, the tension adjustment sequence data can be input into a PID controller, which automatically adjusts the controller's parameters, such as the proportional coefficient, integral time, and differential time, based on the error signal to achieve rapid and accurate compensation for changes in wire diameter. Ultimately, tension feedback control data containing information such as the adjusted winding tension and winding speed is generated to guide the winding machine for the next stage of winding operations.

[0206] The present invention further provides an electromagnetic coil intelligent production control system, which executes the electromagnetic coil intelligent production control method described above. The electromagnetic coil intelligent production control system includes:

[0207] The coil digitization module is used to identify the coil geometric features of the electromagnetic coil design drawing to obtain the coil topology data; perform key process parameter supplementation processing based on the coil topology data to obtain key coil process parameters; and use the key coil process parameters to digitize the coil topology data to generate a three-dimensional coil digital model;

[0208] The winding trajectory optimization module is used to perform coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; perform voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; and perform adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data;

[0209] An intelligent manufacturing execution module is used to perform intelligent manufacturing execution using an intelligent electromagnetic coil production device based on optimized coil winding trajectory data, and to collect real-time coil winding images to generate real-time coil winding image data; based on the real-time coil winding image data, it predicts the trend of wire diameter changes in the turn layer and generates wire diameter change trend data;

[0210] The tension compensation control module is used to process the tension feedback control strategy based on the wire diameter change trend data to obtain tension feedback control data; perform wire diameter consistency compensation control based on the tension feedback control data, and collect wire diameter compensation control effect data to obtain wire diameter compensation effect data; perform multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index;

[0211] The coil quality evaluation module is used to score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain the electromagnetic coil quality score data.

[0212] The beneficial effect of this application lies in the crucial importance of accurately identifying and controlling the geometric features, dimensions, and wiring topology of electromagnetic coils through digital processing and three-dimensional modeling. By supplementing key process parameters such as wire diameter and winding tension based on the actual structural characteristics of the coil, the coil topology data is digitally processed, which optimizes winding path planning, improves winding accuracy and feasibility, and effectively resolves performance fluctuations caused by uneven wire diameter. By real-time monitoring and automatic adjustment of winding tension, wire diameter consistency is ensured, significantly improving the stability and performance consistency of electromagnetic coil production.

[0213] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0214] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent production control method for electromagnetic coils, characterized in that: The following steps are involved: Step S1: performing coil geometric feature recognition on the electromagnetic coil design drawing to obtain coil topology data; Performing supplementary processing on key process parameters based on coil topology data to obtain key coil process parameters; utilizing the key coil process parameters to perform coil digitization processing on the coil topology data to generate a three-dimensional coil digital model; wherein the key coil structure parameters include coil material property data, coil winding property data, and coil electrical property data; Step S2: performing coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; performing voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; performing adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data; Step S3: Based on the optimized coil winding trajectory data, intelligent manufacturing is performed using an intelligent electromagnetic coil production device, and real-time coil winding image acquisition is performed to generate real-time coil winding image data; based on the real-time coil winding image data, a turn layer wire diameter change trend is predicted to generate wire diameter change trend data; Step S4: performing tension feedback control strategy processing according to the wire diameter change trend data to obtain tension feedback control data; Perform wire diameter consistency compensation control based on tension feedback control data, and collect wire diameter compensation control effect data to obtain wire diameter compensation effect data; perform multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index; wherein, step S4 is specifically as follows: Step S41: performing wire diameter trend quantification processing according to the wire diameter change trend data to generate quantified wire diameter change rate data; Step S42: performing eigenmode analysis on the quantized wire diameter change rate data to obtain wire diameter transformation characteristic data; Step S43: Calculating the trend Gaussian cumulative distribution of the wire diameter transformation feature data using a preset wire diameter change threshold, and determining the wire diameter change level. When the wire diameter change level is lower than the wire diameter change threshold, the corresponding winding layer is marked as a stable winding layer; when the wire diameter change level is higher than or equal to the wire diameter change threshold, the corresponding winding layer is marked as an abnormal winding layer. Step S44: Based on the abnormal winding layer, the wire diameter transformation characteristic data is processed by optimizing the coil winding trajectory data to obtain tension feedback control data; wherein, step S44 is specifically as follows: Step S441: performing preset trajectory deviation calculation on the wire diameter transformation characteristic data by optimizing the coil winding trajectory data to generate a winding trajectory deviation value; Step S442: performing deviation time series analysis on the winding trajectory deviation value through a preset winding time window to obtain dynamic deviation time series characteristic data; Step S443: Calculating fuzzy membership based on the dynamic deviation time series feature data to obtain a deviation fuzzy feature matrix; Step S444: using the tension sensor to obtain real-time tension monitoring data; performing fuzzy inference calculation on the real-time tension monitoring data using a preset tension fuzzy control rule library to generate fuzzy tension control data; Step S445: Defuzzifying the deviation fuzzy feature matrix using the tension fuzzy control rule library based on the fuzzy tension control data to obtain precise tension adjustment amount sequence data; Step S446: Adaptively adjust control parameters according to the precise tension adjustment amount sequence data to generate tension feedback control data; Step S45: Using the intelligent electromagnetic coil production device to perform real-time wire diameter consistency compensation control on the next turn layer of the coil based on the tension feedback control data, and collecting the wire diameter compensation control effect to obtain wire diameter compensation effect data; Step S46: performing multi-layer wire diameter consistency calculation based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index; Step S5: Score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain electromagnetic coil quality score data.

2. The electromagnetic coil intelligent production control method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: performing image denoising preprocessing on the electromagnetic coil design drawing to generate a standard electromagnetic coil design drawing; extracting original drawing annotations based on the standard electromagnetic coil design drawing to obtain original drawing annotation data; Step S12: performing multi-level drawing analysis based on the standard electromagnetic coil design drawing and performing vector graphics analysis to generate coil vector structure data; Step S13: performing coil geometric feature recognition on the coil vector structure data using preset geometric constraint relationship rules to obtain coil topology structure data; Step S14: performing graphic element annotation processing on the coil topology structure data using the original drawing annotation data, and establishing a mapping relationship between graphic elements and parameters to generate preliminary topology graphic element annotation data; Step S15: performing key process parameter supplementation processing based on the preliminary topological element annotation data and the coil topological structure data to obtain key coil process parameters; Step S16: digitize the key coil structure parameters using the coil topology data to generate a three-dimensional coil digital model.

3. The electromagnetic coil intelligent production control method according to claim 2, characterized in that: Step S15 includes the following steps: Step S151: Analyze the electrical performance of the target coil based on the preliminary topological element annotation data to generate coil electrical property data; Step S152: extracting coil winding geometric parameters based on the coil topology data to generate coil winding geometric parameters, wherein the coil winding geometric parameters include coil wire diameter, coil outer diameter, coil inner diameter, and coil winding height or length; Step S153: performing wire material property matching on the coil electrical property data and the coil winding geometric parameters through a preset material database to generate coil material property data; Step S154: Calculate the number of coil layers according to the coil winding geometric parameters to generate coil layer data; Step S155: using the coil material attribute data to estimate the coil winding geometric parameters, and generating coil turn estimation data; wherein the coil turn estimation is calculated using a coil turn estimation formula, which is as follows: ； in, Expressed as an estimate of the number of coil turns, Expressed as the coil winding area volume, Expressed as the cross-sectional area of a single wire diameter, Expressed as pi, Expressed as the average winding diameter of the coil, Expressed as the wire diameter, Expressed as the winding process coefficient; Step S156: identifying the coil winding mode of the coil topology data using the coil layer number data and the coil turn number estimation data to generate coil winding mode category data; Step S157: integrating the coil winding type data, the coil layer number data, and the coil turn number estimation data into winding data to generate coil winding attribute data.

4. The electromagnetic coil intelligent production control method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing finite element meshing processing on the three-dimensional coil digital model to generate a finite element coil digital model; Step S22: performing coil voxel winding network analysis based on the finite element coil digital model to generate voxel winding topology network data; Step S23: performing a windability analysis on the voxel winding topology network data and performing voxel winding trajectory mapping to obtain preliminary coil winding trajectory data; Step S24: Decomposing the winding path according to the preliminary coil winding trajectory data to generate a coil winding sub-path set; Step S25: performing path conflict detection on the coil winding sub-path set to obtain winding conflict sub-path data; Step S26: performing magnetic field constraint analysis s on the preliminary coil winding trajectory data using a finite element coil digital model based on a preset magnetic field uniformity distribution strategy, and performing finite element magnetic field distribution simulation calculation to generate coil magnetic field constraint simulation data; Step S27: identifying the magnetic field inhomogeneity region according to the coil magnetic field constraint simulation data, and adaptively adjusting the preliminary coil winding trajectory data through the winding conflict sub-path data to obtain optimized coil winding trajectory data.

5. The electromagnetic coil intelligent production control method according to claim 4, characterized in that: Step S22 includes the following steps: Step S221: constructing a four-dimensional voxel space for the finite element coil digital model to generate a coil four-dimensional voxel space, wherein three spatial dimensions in the coil four-dimensional voxel space correspond to the length, width, and height of the coil, and the fourth dimension corresponds to the number of layers of the coil; Step S222: performing tensor element initialization processing according to the coil four-dimensional voxel space to generate an initialized four-dimensional tensor value; Step S223: Marking the voxel occupancy of the coil four-dimensional voxel space by initializing the four-dimensional tensor value to generate coil occupied voxel data. For each voxel, determine whether its center point is within the geometric range of the corresponding layer. If so, iteratively update the tensor value of the voxel space to 1, indicating that it is occupied. Otherwise, the tensor value of the voxel space remains unchanged. Step S224: performing connected domain component analysis on the coil four-dimensional voxel space using a breadth-first search algorithm based on the coil occupied voxel data to generate coil voxel connected domain data; Step S225: performing adjacent layer overlap analysis based on the coil voxel connected domain data to obtain a directed acyclic graph of the voxel connected layer; Step S226: performing discrete time series calculation on the voxel connected layer directed acyclic graph using a preset winding speed, and assigning a timestamp step size to generate voxel winding topology network data.

6. The electromagnetic coil intelligent production control method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: calibrating the control path points of the optimized coil winding trajectory data using the three-dimensional coil digital model to generate winding path control point data; Step S32: performing path point type identification on the winding path control point data to generate winding path point type data, wherein the winding path point type data includes a starting point, an end point, a turning point, an intersection point, and a smoothing point; Step S33: encoding the optimized coil winding trajectory data into processing control instructions using the winding path point type data, and distributing the servo control instructions to obtain coil servo control instruction data; Step S34: performing intelligent manufacturing execution using an intelligent electromagnetic coil production device based on the coil servo control instruction data, and using a high-speed industrial camera to perform real-time coil winding image acquisition to generate real-time coil winding image data; Step S35: predicting the wire diameter change trend of the turn layer according to the real-time coil winding image data, and generating wire diameter change trend data.

7. The electromagnetic coil intelligent production control method according to claim 6, characterized in that: Step S35 includes the following steps: Step S351: using an edge detection operator to perform real-time coil winding layer edge detection on the real-time coil winding image data, and performing dynamic edge pixel focusing to generate winding layer edge mask data; Step S352: performing pixel-level motion estimation of edges of adjacent frames based on the winding layer edge mask data, and performing continuous frame displacement vector superposition to obtain real-time winding layer edge trajectory data; Step S353: performing sub-pixel edge positioning processing on the real-time winding coil edge trajectory data to generate a high-precision coil edge coordinate set; Step S354: performing time-series multi-frame image fitting based on the high-precision circle edge coordinate set to obtain dynamic line diameter curve fitting data; Step S355: performing spectrum analysis on the dynamic wire diameter curve fitting data to generate wire diameter fluctuation characteristic vector data; Step S356: Based on the preset different wire winding fluctuation data and the corresponding future wire diameter change trend label data, a long short-term memory network model is used to perform transfer learning processing to obtain a wire diameter prediction model; Step S357: transmitting the wire diameter fluctuation characteristic vector data to the wire diameter prediction model to predict the wire diameter change trend of the next turn layer, and generating wire diameter change trend data.

8. An intelligent production control system for electromagnetic coils, characterized in that: For executing the electromagnetic coil intelligent production control method according to claim 1, the electromagnetic coil intelligent production control system comprises: The coil digitization module is used to identify the coil geometric features of the electromagnetic coil design drawing to obtain coil topology data; perform key process parameter supplementation processing based on the coil topology data to obtain key coil process parameters; and use the key coil process parameters to digitize the coil topology data to generate a three-dimensional coil digital model. The winding trajectory optimization module is used to perform coil voxel winding network analysis on the three-dimensional coil digital model to generate voxel winding topology network data; perform voxel winding trajectory mapping based on the voxel winding topology network data to obtain preliminary coil winding trajectory data; and perform adaptive winding trajectory adjustment on the preliminary coil winding trajectory data to obtain optimized coil winding trajectory data; An intelligent manufacturing execution module is used to perform intelligent manufacturing execution using an intelligent electromagnetic coil production device based on optimized coil winding trajectory data, and to collect real-time coil winding images to generate real-time coil winding image data; based on the real-time coil winding image data, it predicts the trend of wire diameter changes in the turn layer and generates wire diameter change trend data; The tension compensation control module is used to process the tension feedback control strategy based on the wire diameter change trend data to obtain tension feedback control data; perform wire diameter consistency compensation control based on the tension feedback control data, and collect wire diameter compensation control effect data to obtain wire diameter compensation effect data; perform multi-layer wire diameter consistency analysis based on the wire diameter compensation effect data to generate a multi-layer wire diameter consistency index; The coil quality evaluation module is used to score the electromagnetic coil quality according to the multi-layer wire diameter consistency index to obtain the electromagnetic coil quality score data.

Images