Plug-in inductor winding process control method based on production parameter dynamic compensation

Abstract

The invention discloses a plug-in inductor winding process control method based on production parameter dynamic compensation, and provides a closed-loop compensation control strategy taking a physical field evolution model as a core aiming at the problems of unstable sensing quantity and non-uniform structure in the traditional process. The method comprises the following steps: firstly, constructing an ideal physical field reference model according to target inductance parameters and a structure, and describing an evolution law of characteristics such as a magnetic field, an electric field or impedance in an ideal state; in the actual winding process, a physical field signal is collected through a sensor, production parameters such as wire tension and main shaft rotating speed are obtained, and a physical field distribution snapshot of the current winding state is generated. And the system compares the extracted actual features with the reference model, identifies a dominant disturbance source causing deviation in combination with real-time production parameters, and outputs a disturbance type label carrying causal information. And according to a label and a production parameter change trend, a compensation instruction is dynamically generated and executed, fine adjustment of tension, a winding displacement track or density is realized, and consistency and precision of an inductor finished product are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of inductor process control, and in particular to a plug-in inductor winding process control method based on dynamic compensation of production parameters. Background Art

[0002] In existing technology, the winding process for plug-in inductors typically uses fixed parameter settings or tension-based closed-loop control, controlling the finished inductor's inductance by adjusting wire tension, the number of winding turns, and the wire arrangement speed. Some high-end equipment incorporates tension monitoring and feedback mechanisms to correct for local tension anomalies, thereby improving inductance repeatability. These methods often rely on empirical formulas and static settings, primarily using the number of turns and tension as input variables. This makes it difficult to respond promptly to dynamic disturbances in the inductance evolution during the actual winding process.

[0003] The main problem with existing technologies is that their control logic typically lacks direct sensing of changes in the inductor's internal electromagnetic field structure, making it impossible to assess in real time changes in the spatial electromagnetic distribution and structural consistency deviations during the winding process. Furthermore, most current systems fail to incorporate multiple disturbances during the production process, such as temperature and humidity, wire diameter fluctuations, and equipment vibration, as modeling inputs. This leads to issues such as inductance drift, uneven distribution, and localized layer overlap when material batches vary or environmental conditions fluctuate, further impacting product consistency and yield.

[0004] Based on the above background, it is urgent to propose a plug-in inductor winding control method that can combine production parameters and has dynamic compensation capabilities to improve control accuracy and process adaptability. Summary of the Invention

[0005] The present application provides a plug-in inductor winding process control method based on dynamic compensation of production parameters to improve control accuracy.

[0006] The present application provides a method for controlling a plug-in inductor winding process based on dynamic compensation of production parameters, comprising:

[0007] Based on the target inductor parameters and the plug-in inductor structure, a benchmark model describing the ideal physical field evolution characteristics is constructed. The benchmark model is used to characterize the physical field changes in the inductor core area under ideal winding conditions with the number of turns or over time.

[0008] A non-contact sensor array placed in the winding area collects physical field signals, including magnetic, electric, or impedance signals. Combined with sampled data of current production parameters, including wire tension, spindle speed, wire pitch, and ambient temperature and humidity, this generates a snapshot of the physical field distribution at the current level and extracts evolutionary characteristic parameters that match the benchmark model.

[0009] The extracted evolutionary characteristic parameters are compared with the ideal characteristic parameters under the same winding schedule in the benchmark model. Combined with the collected production parameters, a mapping relationship between physical field deviations and production disturbances is established. The dominant disturbance source causing the deviation is identified through a preset deviation attribution mechanism, and a disturbance type label is obtained.

[0010] Based on the disturbance type label and the corresponding production parameter change trend, compensation instructions for fine-tuning the winding behavior are generated. The compensation instructions include: adjusting the tension curve of the wire feeding mechanism to offset the influence of tension disturbance, correcting the wire arrangement rhythm to alleviate displacement deviation, optimizing the wire arrangement path to restore distribution uniformity, or adjusting the wire arrangement density to compensate for wire diameter changes; executing the compensation instructions to compensate for the winding behavior.

[0011] The beneficial effects of the technical solution provided by this application include:

[0012] (1) By constructing a benchmark model of the ideal physical field evolution characteristics, it can provide an accurate reference for the electromagnetic field behavior of the inductor winding process, improve the ability to judge the structural consistency during the winding process, and provide a theoretical basis for subsequent deviation control. (2) Using a non-contact sensor array to collect multi-dimensional physical field signals in real time, and combining them with production parameters such as tension, speed, temperature and humidity, it can fully reflect the actual winding state and significantly enhance the process's perception and response speed to disturbance changes. (3) By establishing a causal mapping relationship between physical field deviations and production disturbances, and guiding compensation control with the help of disturbance type labels, the compensation measures are targeted and explainable, overcoming the problem of easy inaccuracy of traditional empirical parameter adjustment methods. (4) The generated compensation instructions can achieve multi-dimensional fine-tuning of tension, wiring rhythm, path and density, thereby achieving a dual improvement in inductance precision control and finished product consistency without changing the mechanical structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of a plug-in inductor winding process control method based on dynamic compensation of production parameters provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0014] The following description sets forth many specific details to facilitate a thorough understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of the present application. Therefore, the present application is not limited to the specific implementations disclosed below.

[0015] The first embodiment of the present application provides a method for controlling the winding process of an insert inductor based on dynamic compensation of production parameters. Figure 1 , which is a schematic diagram of the first embodiment of this application. Figure 1The first embodiment of the present application provides a detailed description of a plug-in inductor winding process control method based on dynamic compensation of production parameters.

[0016] Step S101: During the winding task initialization phase, a benchmark model describing the ideal physical field evolution characteristics is constructed based on the target inductor parameters and the plug-in inductor structure. The benchmark model is used to characterize the physical field variation pattern of the inductor core region under ideal winding conditions with the number of turns or over time, serving as a reference for subsequent winding deviation judgments.

[0017] During the initialization phase of the winding task, a baseline model of idealized physical field evolution characteristics must be constructed to enable real-time monitoring and compensation of dynamic deviations during the plug-in inductor winding process. This model aims to characterize the changing trends of key physical quantities such as magnetic field, electric field, and impedance in the core region of the plug-in inductor under ideal winding conditions, as a function of the number of turns or the passage of time. This provides an accurate and quantifiable reference for subsequent feature extraction, deviation assessment, and compensation control.

[0018] The input information required to build this benchmark model includes the target inductor's inductance value, allowable error, operating frequency range and other performance parameters, as well as the inductor's structural information, such as the skeleton form, core size, wire diameter, insulation layer thickness and wiring method. These parameters can be obtained from the inductor product design drawings. In addition, the electromagnetic properties data of the relevant materials are required, such as the initial magnetic permeability of the core, the saturation magnetic flux density, the resistivity of the wire and the dielectric constant of the dielectric. These data can be derived from the specifications provided by the material supplier or experimental measurements. In order to simulate the operating state in the actual winding environment, initial boundary conditions should also be introduced, such as the current excitation method, winding speed setting and the assumption of stable ambient temperature, to ensure that the modeling process is consistent with the actual working conditions.

[0019] The model can be implemented in two ways. The first is numerical simulation, which involves constructing a three-dimensional geometric model of the inductor to be wound on an electromagnetic field simulation platform. Combining the aforementioned input parameters and physical constraints, the winding step strategy is set so that the simulation system simulates the inductor winding process turn by turn. At each stepping node, the system calculates the physical field distribution inside and around the inductor, extracts representative characteristic quantities, such as the magnetic field intensity, field gradient, unit volume magnetic flux density, or impedance phase angle change at the inductor center, and records their continuous evolution trend over time or the number of turns. The second is sample inversion modeling, which involves using several groups of plug-in inductor samples with stable performance as the basis, using non-contact testing equipment such as micro-Hall sensors, electric field probes, or impedance analyzers, to collect physical field response data at different levels of sample winding. The correspondence between the inductor quantity and the physical field is established through timestamps or turn number marks, and then an ideal evolution curve is constructed through algorithms such as data fitting and principal component analysis.

[0020] Regardless of whether simulation or inversion is used, the output of the model should be structured and reusable. The output results mainly include the expected eigenvalue sequence of the physical field under multiple winding progress nodes, as well as the alignment criteria and difference quantification method between the actual measured data and the benchmark eigenvalues. For the convenience of subsequent use, the model can be saved in the form of a database, function expression or normalized table, and a unified data interface is provided for calling and comparison. In order to improve the adaptability and robustness of the model, it is recommended to perform noise filtering, standardization and spatial interpolation optimization on the original data during the modeling process to ensure that the model still has good versatility under different material batches and different environmental conditions.

[0021] By constructing this benchmark model, the system can establish a complete set of expected physical field trajectories in the early stages of winding, and use it as the basis for deviation judgment and compensation control of the actual winding state in subsequent processes, significantly enhancing the precision control capability and environmental adaptability of the plug-in inductor winding process, and providing key support for the dynamic compensation mechanism of the present invention.

[0022] Furthermore, the benchmark model is established through electromagnetic field finite element simulation. The simulation process uses the magnetic flux density distribution, magnetic field intensity distribution, and local electric field response within the capacitive coupling surface after each winding of the plug-in inductor structure is completed as reference indicators. The simulation input includes the target inductor parameters, the plug-in inductor structure, and the wire arrangement. The material permeability, saturation magnetic flux density, and dielectric constant of the insulating material are introduced as modeling parameters.

[0023] During the simulation process, multiple characteristic points are extracted from the impedance frequency response curve, including the maximum phase difference point, the minimum impedance point, the impedance mutation inflection point, and the critical frequency band boundary. This set of frequency response characteristics is used as auxiliary reference parameters to describe the winding evolution process under ideal conditions, and is used together with the magnetic flux density distribution and the electric field distribution to construct the physical field evolution path of the benchmark model.

[0024] The output of the benchmark model is in the form of a structured vector. The structured vector includes an index of the number of winding turns, an index of the evolutionary characteristic parameter corresponding to each turn, and a tolerance range field corresponding to each characteristic parameter. The structured vector is aligned and compared with the evolutionary characteristic parameters extracted in real time during the actual winding process of the plug-in inductor.

[0025] The structured vector serves as the only reference for comparing the extracted evolution characteristic parameters with the ideal characteristic parameters under the same winding progress in the benchmark model in the subsequent steps, and is consistent with the deviation input data structure required to identify the dominant disturbance source causing the deviation through the preset deviation attribution mechanism.

[0026] To ensure the benchmark model possesses sufficient physical realism and control accuracy during the plug-in inductor winding process, it was constructed using electromagnetic field finite element simulation. This modeling process simulates the physical field evolution of the wire after winding, turn by turn, based on the specific structural characteristics of the plug-in inductor. The magnetic flux density distribution, magnetic field intensity distribution, and the local electric field response within the capacitive coupling surface serve as the primary simulation output metrics to characterize the spatial accumulation path of the inductance and the consistency of the structural electrical performance.

[0027] Specifically, during the model initialization phase, the key performance parameters of the target inductor, such as the target inductance value, operating frequency, expected tolerance, etc., should be input first, and combined with the detailed geometric model of the plug-in inductor skeleton structure, including the core shape, center column size, winding window size, installation form, etc. In terms of wire arrangement, information such as wire density, cross-layer method, wire routing sequence, and interlayer insulation thickness should be defined. Subsequently, at the material level, the initial magnetic permeability and saturation flux density of the magnetic conductor, the electrical conductivity and surface roughness of the wire, and the physical parameters such as the dielectric constant and loss factor of the insulating material should be introduced. All input parameters must have sufficient resolution to meet the requirements for restoring the details of electromagnetic behavior during turn-by-turn modeling.

[0028] During the modeling process, each time a simulation of a wire winding is completed, the system automatically records the magnetic flux density distribution diagram and magnetic field vector diagram under the corresponding structure, and generates a capacitive coupling distribution diagram in combination with the potential gradient field of the dielectric region between the wires. At the same time, in order to capture the detailed characteristics of the inductor's frequency domain response, it is necessary to apply AC frequency sweep boundary conditions to the model at each stage, simulate and extract the corresponding impedance frequency response curve, and extract several typical feature points from it, including the maximum phase difference point (corresponding to the behavioral change near the resonance peak), the minimum impedance point (corresponding to the conduction characteristics in the low frequency band), the impedance mutation inflection point (reflecting the response discontinuity caused by structural or material mutation), and the critical frequency band boundary (used to describe the inductance-capacitance switching threshold of the winding structure). These frequency feature points not only reflect the full-band performance of the inductor, but also reveal its structural consistency and boundary behavior. They are indispensable auxiliary references in the ideal evolution path.

[0029] The data generated by the above simulations is uniformly encoded and output as a structured vector. This structured vector uses the number of winding turns as the primary index, representing the physical progression of each wire's position along the evolutionary path. Each turn contains several evolutionary characteristic parameter indices, identifying core metrics such as magnetic flux intensity, electric field coupling, and impedance variation. Each characteristic parameter is accompanied by a preset tolerance range field, indicating the physical deviation allowed during the actual winding process, ensuring engineering applicability and judgment tolerance in the subsequent comparison process.

[0030] As the only ideal reference object in the entire control system, the structured vector will be precisely aligned in space and time with the evolutionary characteristic parameters extracted turn by turn during the real-time process of plug-in inductor winding, and the deviation will be output through difference calculation, threshold judgment, etc. At the same time, the field structure and index rules of the structured vector are consistent with the deviation input data structure used in the subsequent deviation attribution module, ensuring that the comparison results can be directly used as input for attribution judgment without the need for additional format conversion, thereby achieving seamless connection of the data processing link. This approach not only improves the overall response speed of the system, but also enhances the physical interpretability and semantic correspondence between the model results and real-time data, making the entire compensation control mechanism rigorous, inter-module data compatible, and practical engineering implementability.

[0031] Furthermore, during the simulation process, multiple characteristic points are extracted from the impedance frequency response curve, including:

[0032] During the electromagnetic field finite element simulation process, the impedance frequency response curve of each coil of wire is first simulated over the entire frequency range, and a set of characteristic points including the maximum phase difference point, the minimum impedance point, the impedance mutation inflection point, and the critical frequency band boundary are identified in the simulation results;

[0033] In the set of identified feature points, the frequency response slope change, impedance phase angle fluctuation range and amplitude gradient curvature of each feature point are further calculated, and based on this, a frequency response perturbation sensitivity matrix is constructed to characterize the response amplification factor of each feature point to structural micro-perturbations;

[0034] The frequency response disturbance sensitivity matrix is coupled with the local magnetic flux density distribution diagram of the plug-in inductor structure to analyze and filter out abnormal frequency mutation points appearing in the weak coupling section of the structure, ensuring that only effective response characteristic points with structural physical relevance are retained;

[0035] Perform frequency band clustering processing on the retained valid feature point set, group and merge them based on bandwidth coverage and frequency proximity, and calculate the center frequency, main response mode and representative response characteristic value for each cluster segment to form a frequency band representation subset;

[0036] The frequency band representation subset is used as an auxiliary reference parameter to construct the benchmark model. After being aligned with the magnetic flux density distribution and electric field distribution in the turn dimension, it is encoded in a unified data format as a frequency domain evolution path field in a structured vector. This allows for turn-by-turn comparison of the evolutionary characteristic parameters extracted in real time during the actual winding process of the plug-in inductor.

[0037] In the electromagnetic field finite element simulation process described in the present invention, the goal is to generate an impedance frequency response curve for the structural state after each winding of the wire. This curve generally represents the amplitude and phase information of the complex impedance at multiple frequency points, and formally contains the relationship between the real and imaginary parts and the frequency variation. The simulation uses a three-dimensional finite element method based on the magnetic vector potential, and obtains the frequency response results by applying a swept current source to the model. The frequency sweep range should cover a bandwidth of at least two orders of magnitude, for example, 10kHz to 10MHz, to ensure that the main operating frequency band and boundary behavior of the plug-in inductor are included.

[0038] After obtaining the frequency response curve for each winding structure, the system begins to identify characteristic points. First, the derivative of the phase curve is detected to find the point of maximum phase difference, which is usually located near resonance and is most sensitive to the winding structure. Secondly, the minimum impedance point is found, indicating that certain paths in the structure exhibit near-electrical short-circuit behavior. Furthermore, the region of sharp gradient change at the intersection of the real and imaginary parts is analyzed to extract the impedance mutation inflection point, which is used to reflect the change in the electrical coupling response at the structural boundary. Finally, the flat response turning area is searched at the end of the curve or in the high-frequency portion, which is defined as the critical frequency band boundary. The above set of points constitutes the first-level identification result of the frequency response characteristic points and is stored as the original characteristic point set containing the frequency value, complex impedance value, and number of turns index.

[0039] In order to improve the model's response sensitivity to disturbances, the system will enter the second stage, which is to calculate the local response parameters of the above-mentioned characteristic points. In this process, a certain frequency interval (such as ±5% bandwidth) is selected near each characteristic point to construct a local window, and the first-order derivative of the complex impedance curve in the window (frequency response slope), the maximum variation range of the complex phase angle (phase angle fluctuation amplitude), and the second-order derivative of the amplitude with respect to frequency (curvature) are calculated to evaluate the point's ability to amplify micro-perturbations. All calculation results constitute a disturbance response sensitivity matrix, which is indexed by characteristic points and has column vectors including four dimensions: frequency, phase angle fluctuation, response slope, and curvature.

[0040] Next, the system performs a spatial correspondence analysis between the frequency response perturbation sensitivity matrix and the magnetic flux density distribution map for the same structure. This requires the introduction of a three-dimensional structural grid from the electromagnetic simulation model and its positional binding to the structural segment represented by the frequency response point. This allows the system to determine whether the frequency response feature point is located in an area of concentrated magnetic flux density or a region of significant magnetic field gradient variation. If the structural location corresponding to a feature point has a flat magnetic flux distribution with no apparent response concentration, it is considered to have no direct physical coupling with the structure and the system will identify it as a pseudo-response point or boundary error point and remove it.

[0041] After completing the screening of valid feature points, the system clusters the remaining feature points according to their frequency values. The clustering algorithm can use K-means or bandwidth-aware clustering mechanisms, combining bandwidth coverage (whether the frequency spacing between points is dense enough to form a continuous band) and frequency proximity (the mean and variance of the distance between points) for grouping. Within each cluster segment, the system will further calculate the center frequency (geometric mean), the main response mode (mainly the point with the maximum response slope), and the representative eigenvalue (the maximum value of the product of response intensity and stability), and finally output the representation subset of the frequency band. This representation subset not only compresses the original data dimension, but also retains the main frequency area that is most sensitive to structural disturbances.

[0042] Finally, the frequency band representation subset is encoded in a structured format as part of the benchmark model. Specifically, it is in the form of a "frequency domain evolution path field" in a structured vector. Its data structure should include the turn index, the center frequency of the frequency band, the response indicator triplet (slope, phase angle amplitude, curvature), and the tolerance range (defining the allowable comparison error window). The frequency domain evolution path field will be aligned with the magnetic flux density distribution field and the electric field distribution field in the turn dimension, and unified timing and structural indexes will be used to participate in the comparison of the evolution characteristic parameters extracted during the actual winding process of the plug-in inductor. The comparison logic is not limited to amplitude comparison, but should also include dimensions such as frequency shift deviation, phase angle offset, and response change rate change to fully support deviation diagnosis and disturbance attribution.

[0043] Step S102: During the actual winding process of the plug-in inductor, a non-contact sensor array arranged in the winding area is used to collect physical field signals such as magnetic field, electric field or impedance. Combined with the real-time sampling data of the current production parameters, including wire tension, spindle speed, wiring pitch, ambient temperature and humidity, a snapshot of the physical field distribution at the current level is generated, and evolutionary characteristic parameters matching the benchmark model are extracted to reflect the local gradient, coupling morphology or change rate of the wound area.

[0044] During the actual winding process of plug-in inductors, in order to achieve dynamic perception of the winding state and basic support for subsequent deviation judgment, it is necessary to perform real-time physical field information collection and evolution feature parameter extraction operations. To this end, the deployment of a non-contact sensor array in the winding area is a key implementation method for this step. These sensors are usually installed in multiple spatial locations near the winding station, which can cover the main electromagnetic active areas of the inductor winding area and have multi-channel, high-time resolution sensing capabilities. Commonly used sensor types include miniature Hall effect magnetic field sensors, high-frequency electric field probe arrays, broadband impedance sampling probes, etc. The specific selection can be determined according to the structural size, operating frequency range and process cycle of the target inductor. The above-mentioned sensor array needs to have non-contact measurement capabilities to avoid mechanical interference with the coil formation process, and is connected to the data acquisition main control system through a high-speed sampling interface.

[0045] During the winding process, the system uses the number of winding turns or time as a synchronization index to collect data on the transient physical field distribution generated during the formation of each layer or each coil. The acquired data is then filtered, denoised, and normalized. To ensure timing consistency, this physical field acquisition process must be precisely synchronized with the winding equipment's spindle speed, wire feed speed, and wire arrangement rhythm. The system typically uses encoder signals and tension closed-loop feedback signals to time-bind the physical field data acquisition instructions at the hardware and software levels, ensuring a one-to-one correspondence between the measured electromagnetic signals and the specific winding stage of the physical process. After data acquisition is complete, the system generates a spatial field distribution snapshot reflecting the physical characteristics of the layer, taking into account the structural state of the current layer. This "physical field distribution snapshot" refers to the reconstruction of the physical quantities measured at each sensor point at a specific moment into a two-dimensional or three-dimensional electromagnetic field map of the inductor winding region, such as a magnetic induction intensity distribution map, an electric field coupling intensity map, or an equivalent impedance phase map, through spatial interpolation and structural mapping. This snapshot not only displays the spatial distribution of the physical characteristics of the wound inductor region but also can be used to further extract core evolution parameters that match the benchmark model.

[0046] The evolutionary characteristic parameters corresponding to the benchmark model refer to quantitative indicators extracted from the actual physical field distribution snapshot that can accurately reflect the difference between the winding state and the ideal inductance evolution trend. These parameters are based on the physical quantity path pre-defined in the benchmark model and are consistent in terms of time sequence, spatial structure and physical dimensions. Specifically, they include but are not limited to: the evolution curve of the axial magnetic induction intensity at the center of the inductor, the local magnetic field gradient distribution, the attenuation rate of the magnetic flux density at the edge of the winding, the inter-wire electric field coupling energy density, the displacement trajectory of the maximum electric field response point, the rate of change of the capacitive coupling path, the trend change of the impedance phase angle with the increase of the number of turns, etc. These characteristic parameters can be extracted from the snapshot map through algorithms such as spatial weighted averaging, principal component analysis, template fitting, Fourier transform, etc., and uniformly output as structured vectors or tensors for high-precision comparison with the benchmark model in the next step.

[0047] To improve the stability and environmental adaptability of the feature extraction results, real-time production parameters need to be introduced as synchronous input variables during the extraction process, including wire tension, spindle speed, wire pitch, ambient temperature and humidity, etc. The above production parameters directly affect the micro-forming quality and electromagnetic response of the inductor. For example, tension fluctuations may cause abnormal magnetic field gradients, humidity changes may affect the dielectric properties of the insulating medium, and uneven pitch may lead to unbalanced capacitive coupling. Therefore, during the feature parameter extraction process, the system jointly models the production parameters and field data, and sets dynamic weighting coefficients or sensitivity factors in the algorithm, so that the final extracted evolutionary feature parameters have both physical interpretability and robustness to disturbed environments.

[0048] Ultimately, within each winding cycle, the system outputs a set of evolutionary characteristic parameters with clear structure, explicit physical meaning, and a one-to-one correspondence with the baseline model. These parameters describe the key physical response trends within the current wound region of the inductor. These parameters serve as the foundation for deviation calculation and disturbance attribution, as well as the direct basis for generating subsequent compensation control strategies. They constitute the core perception layer of the plug-in inductor intelligent winding control system.

[0049] Furthermore, the sampling data of current production parameters, including wire tension, spindle speed, wire pitch, and ambient temperature and humidity, is combined to generate a snapshot of the physical field distribution at the current level and extract evolutionary characteristic parameters that match the benchmark model, including:

[0050] Synchronously processing the raw physical field signals collected by the non-contact sensor array and the sampled data, normalizing the outputs of different sensing channels based on the reference physical response values corresponding to the number of winding turns, and obtaining standardized multi-channel equivalent physical response curves for subsequent analysis;

[0051] Based on the standardized equivalent physical response curve, combined with the arrangement coordinates of each turn of the wire and the trend of the magnetic flux density change, the local physical field boundary in the current inductor winding is identified, and based on the boundary information, the entire winding area is divided into multiple local areas;

[0052] In each local area, the magnetic field response path, capacitance coupling trajectory, and impedance phase change curve are extracted based on the spatial distribution of the equivalent physical response curve. The magnetic field gradient trajectory, capacitance energy trajectory, and complex impedance trajectory are generated through fitting processing, and a set of trajectory vectors representing the spatial physical evolution trend is output.

[0053] Based on this set of orbital vectors, and in combination with the aforementioned standard deviation of wire tension, the first-order derivative of ambient humidity changes, the fluctuation rate of the wiring pitch, and the rate of change of the spindle speed, the dynamic threshold parameters used to adjust the comparison tolerance of each type of evolving orbit are calculated. This reconstructs the allowable offset range of the magnetic field orbit, the response anomaly tolerance of the capacitance orbit, and the phase error threshold of the impedance orbit.

[0054] The orbital vector set adjusted by dynamic threshold is used as the final evolution characteristic parameter.

[0055] During the actual winding process of the plug-in inductor, a non-contact sensor array continuously collects raw physical field signals covering the winding space, including raw physical quantities such as magnetic induction intensity, electric field response, capacitive coupling, and current path impedance. Simultaneously, a control system samples key production parameters such as current wire tension, spindle speed, wire pitch, and ambient temperature and humidity at a high frequency. These production parameters are real-time and perturbative, and must be processed synchronously with the physical field data to ensure physical consistency and temporal alignment accuracy in the subsequent inference process.

[0056] During synchronous processing, the system uses the current lap number as the time index and establishes a binding relationship between the sensor channel number and the lap sequence for each sampling point. The system then combines the various physical field response values collected during that lap with the tension curve, wire tracing rhythm changes, humidity change slope, and other factors within the corresponding time period, applying a normalization function to unify the multi-channel responses. This normalization process not only unifies the dimensions but also corrects for the dynamic sensitivity offset of each channel, ensuring that the final output is a standardized multi-channel response curve that can be used for subsequent structural reconstruction and physical track fitting.

[0057] After generating the normalized response, the system enters the winding local structure identification phase. This phase involves reading the spatial arrangement of the wire loop in the skeleton coordinate system and coupling this with the magnetic flux density trend at that location to identify the boundary position of the current winding in the physical response map. By analyzing the spatial differences and gradient trends between multiple sets of boundary points, the system can delineate local response regions, each with essentially consistent electromagnetic response behavior. These local regions serve as the spatial basis for the physical field distribution snapshot, providing boundary constraints for subsequent orbital vector extraction.

[0058] Within each local region, the system calculates the spatial distribution of the magnetic induction intensity vector field, electric field potential gradient field, and equivalent impedance phase change function within the region, fitting the main gradient track reflecting the magnetic field transmission path, the energy track of the capacitive coupling aggregation path, and the phase change track of the impedance track. Each track is fitted by a set of continuous coordinate points, and its physical meaning corresponds to the contribution track of the coil structure to the inductance, parasitic capacitance, and equivalent impedance performance. These track vectors not only express the trend of physical state changes with position, but also have parameters such as spatial directionality, amplitude change rate, and response distribution width, which are used to quantify the order and offset degree of the physical structure in space.

[0059] After the track extraction is completed, the system enters the dynamic threshold calculation stage. In this stage, the system introduces the high-order statistical characteristics of the aforementioned production parameters as adjustment factors, including the standard deviation of wire tension (reflecting mechanical stability), the first-order derivative of ambient humidity changes (reflecting medium environment fluctuations), the wire pitch fluctuation rate (reflecting the wire control accuracy) and the spindle speed change rate (reflecting the winding rate stability). Based on these parameters, the system constructs a set of physical offset sensitivity functions and applies them to the extracted track vectors, setting different comparison tolerance upper limits for the magnetic field track, capacitance track and impedance track. The calculation of these tolerance thresholds follows the setting logic of "the greater the disturbance, the tighter the tolerance" or "the more rapid the offset trend, the dynamic enhancement of the threshold", thereby realizing dynamic adaptation of the deviation comparison rules to the actual disturbance scenario.

[0060] Ultimately, the system outputs the aforementioned set of orbital vectors, adjusted for dynamic tolerances, as the evolutionary characteristic parameters for the current number of turns or current level. These evolutionary characteristic parameters not only fully record the structural trends and response behaviors of the current winding region within the electromagnetic space, but also possess numerical stability after perturbation adaptation. They are the only high-fidelity inputs required for subsequent comparisons with the ideal characteristic parameters under the same winding schedule in the benchmark model. Through this processing path, the system can establish a clear and operational mapping relationship between the physical field space and production disturbance changes, ensuring that the entire deviation diagnosis and compensation path maintains a high degree of precision control and evolutionary consistency even under complex operating conditions.

[0061] Furthermore, the calculation method of the dynamic threshold parameter for adjusting the tolerance of each type of evolving orbit comparison includes a weighted fusion model based on the disturbance attribution factor and the orbit response sensitivity, and the dynamic threshold ΔT(i,j) is calculated as follows:

[0062]

[0063] Where ΔT(i,j) represents the comparison tolerance threshold of the i-th track in the j-th region, and the unit is the normalized response; S(i,j) represents the perturbation sensitivity coefficient of the track vector in the circle number i and the spatial region j, which comes from the perturbation response gradient in the early track feature modeling process; Indicates the tension fluctuation variance of the tension sensor in the current winding cycle;

[0064] It represents the first-order derivative of the magnetic field as the number of turns changes, reflecting the trend of local magnetic flux mutation;

[0065] Var(P j ) represents the historical variance of the line pitch in the j-th spatial region, reflecting the fluctuation amplitude of the structural density; It represents the spatial gradient of the electric field in the jth spatial region, reflecting the steepness of the boundary of the dielectric coupling response in this region; α is the basic scaling factor, which is used to adjust the overall tolerance level and has a value range of [0.5, 2.0]; β1, β2, and β3 are the perturbation type attribution weighting factors, which dynamically select the weighted priority according to the perturbation type indicated in the perturbation label to satisfy the normalization constraint of β1+β2+β3=1.

[0066] The calculated result ΔT(i, j) will serve as the basis for determining the offset of each trajectory node in the orbit vector and will be used to set the comparison thresholds for the main magnetic field path, the capacitive energy path, and the complex impedance path. This will ensure detection sensitivity while improving the ability to suppress false alarms of low-amplitude disturbances.

[0067] In the plug-in inductor winding process control method proposed in the present invention, in order to achieve high-precision comparison of different types of physically evolved tracks and effectively deal with the local error amplification effect caused by tension fluctuations, environmental disturbances or structural unevenness, the system introduces a dynamic threshold parameter calculation method based on the weighted fusion of disturbance attribution factors and track response sensitivity, which is used to determine the response offset tolerance range allowed for each type of track during the comparison process.

[0068] The dynamic threshold uses the number of circles and the spatial area as dual indexes to form a two-dimensional tolerance matrix. Its core calculation formula is as follows:

[0069]

[0070] Wherein, ΔT(i,j) represents the comparison tolerance threshold of the i-th orbit in the j-th spatial region. Its unit is the normalized response amplitude and it is a positive real value. Its physical meaning is the range within which the offset of the orbit vector at this position can be regarded as an invalid disturbance or a normal fluctuation, which is used to determine whether to trigger the compensation mechanism.

[0071] S(i,j) is the track perturbation sensitivity coefficient. It is the quantified result of the local perturbation response gradient extracted after fitting the historical perturbation response during track modeling. It is typically based on the rate of change of the magnetic field gradient track, capacitance track, and complex impedance track to the perturbation input. It has a unit of 1 and is a dimensionless ratio. S(i,j) reflects the susceptibility of the response at that turn and location to the perturbation input. The recommended value range is [0.8, 1.5]. When the winding area is at the edge of the structure or across layers, this value should be increased to improve monitoring sensitivity. The track perturbation sensitivity coefficient S(i,j) is determined based on the magnitude of the sensor response change at that location in historical samples when a perturbation occurs. The more severe the response, the higher the coefficient. The system automatically sets a higher value (e.g., 1.3) for structural edges, across layers, or areas prone to tension fluctuations; it sets a lower value (e.g., 0.8) for central areas where the response is stable. The system establishes a table of position-specific coefficients through pre-calibration, which can be directly accessed by lookup based on the number of turns and location during the actual winding process.

[0072] The fluctuation variance of the conductor tension in the current winding cycle, in (N·m) 2 Continuous real-time sampling from the tension sensor is used to calculate the variance using a sliding window. The recommended time window is ±2 turns of the current number of turns. A larger value indicates significant tension fluctuations during the winding process, which can easily lead to changes in winding density. Therefore, the local comparison threshold should be relaxed.

[0073] It is the square of the first derivative of the magnetic induction intensity along the number of turns, which is used to measure the sudden change of the magnetic field along the winding time direction. This quantity is calculated by analyzing the difference of the B value sequence of the non-contact Hall sensor along the time axis, and the unit is (T / s) 2 If a flux jump occurs in a local area, this value will increase significantly.

[0074] Var(P j ) is the historical variance of the wiring pitch in the jth spatial region, in mm 2 , an indicator of cable routing consistency. This value is obtained by reading the cable routing pitch in the corresponding area over multiple winding cycles and calculating the statistical variance. A larger variance indicates significant fluctuations in the routing density in that area, poor structural stability, and a high likelihood of accumulating capacitive coupling errors.

[0075] is the square norm of the electric field gradient in the region, in (V / m) 2 , characterizing the spatial intensity of electric field variations in the dielectric layer. This term is acquired using a high-frequency electric field probe array, reconstructed into a continuous field distribution through interpolation, and then calculated using the local gradient norm. If this value is large, there may be a region of enhanced interlayer coupling, requiring increased tolerance to avoid false triggering.

[0076] α is the basic scaling factor, a global scaling factor set by the user, used to adjust the looseness of the overall tolerance. The recommended value is 1.0. During the debugging phase, it can be adjusted between 0.5 and 2.0 to adapt to different devices and product specifications.

[0077] β1, β2, and β3 are the normalized weighting coefficients for the three types of physical deviation factors, determining their contribution to the dynamic tolerance. The sum of these three factors must be 1 and should be dynamically set based on the currently identified disturbance type label. For example, if the label indicates that the disturbance is primarily caused by tension fluctuations, β1 is recommended to be set above 0.5. If the disturbance is primarily caused by cross-layer structural offset, β2 is recommended to be increased to 0.4-0.6. If electric field coupling anomalies are the primary cause, increasing β3 to 0.5 is more effective.

[0078] The resulting ΔT(i,j) is then applied to the comparison of each orbital vector type to determine whether the current orbital node offset exceeds a reasonable range. If the offset values of multiple consecutive nodes exceed the threshold, the system determines that an actual disturbance has occurred and triggers the corresponding compensation action. Compared to a fixed threshold mechanism, this method has greater environmental adaptability and interference robustness, significantly reducing false alarm rates and improving actual winding consistency.

[0079] Furthermore, in each local area, according to the spatial distribution of the equivalent physical response curve, the magnetic field response path, the capacitance coupling trajectory and the impedance phase change curve are extracted, and the magnetic field gradient trajectory, the capacitance energy trajectory and the complex impedance trajectory are generated through fitting processing, and a set of trajectory vectors representing the spatial physical evolution trend is output, including:

[0080] In each local area, based on the multi-channel equivalent physical response curves collected and normalized by the contactless sensor array, directional projection is performed in the structural coordinate system. The magnetic field response, electric field coupling, and complex impedance phase change are projected onto the axial, radial, and planar transverse vector channels of the inductor winding, respectively, thereby constructing a response matrix set with direction labels.

[0081] Using the aforementioned directional-labeled response matrix set, the main response path is identified along the three-dimensional spatial grid nodes in each local area. The main path identification algorithm adopts the principle of local maximum response growth rate and selects the continuous point chain with the largest absolute value of the first-order derivative of the response change as the candidate segment of the main response trajectory. In addition, the abnormal segments with inconsistent physical logic are eliminated by combining the magnetic flux continuity, electric field enclosure, and impedance-admittance consistency conditions. Finally, three types of spatial response trajectories are formed: the main magnetic field gradient trajectory, the capacitance energy aggregation trajectory, and the complex impedance offset path.

[0082] Based on the three types of spatial response trajectories obtained, trajectory fitting models are established respectively based on the number of conductor turns and the spatial hierarchical structure. The fitting models use B-spline curves or polynomial fitting to smooth the trajectory curves and discretize each type of trajectory path into a set of trajectory nodes with a turn number index, path direction vector, response amplitude and change rate.

[0083] For abnormal change points in the track node set, combined with the aforementioned production parameter sampling data, including the standard deviation of conductor tension, the fluctuation rate of the wire pitch, and the first-order derivative of humidity change, the weight of the disturbance influence factor is set on the fitting track, and dynamic compensation fitting is performed on the local track segment. This ensures that each track not only reflects the physical response itself, but also has structural sensitivity that is strongly correlated with external disturbance sources.

[0084] The three types of orbital node sets are fused and encoded into an orbital vector set. The orbital vector set uses the number of circles as the main index, the spatial path segment as the sub-index, and the orbit type as the classification label. It also includes a disturbance sensitivity field, a position confidence field, and a spatial scale normalization factor field. This serves as the physical structure reference basis for subsequent deviation judgment, dynamic threshold calculation, and disturbance attribution comparison.

[0085] During the actual plug-in inductor winding process, a sensor array continuously and contactlessly collects spatially distributed magnetic and electric field signals, as well as complex impedance response signals, across the winding area. These signals are synchronized with the number of turns and the current wiring rhythm for sampling. The output data from all sensor channels is uniformly normalized to form a multi-channel equivalent physical response curve. Each channel corresponds to a spatial location, and this response curve exhibits significant directional differences at different winding stages, necessitating spatial decomposition within the structural coordinate system.

[0086] Specifically, the system will perform three-dimensional projection processing on the collected response curves in each local area (for example, within a range of ± several millimeters around the center of the current winding). The magnetic field response is projected to the axial direction of the inductor winding, that is, it is unfolded along the centerline of the winding; the capacitive coupling characteristics are unfolded radially, reflecting the distribution density and dielectric response of the insulating medium between the wires; and the complex impedance phase change is more reflected in the winding cross-sectional area, which is unfolded in the planar direction. After three-dimensional projection, a response matrix set with direction labels is formed. Each column of the matrix corresponds to a response type, and each row corresponds to a spatial sampling point, while retaining the turn index and timestamp.

[0087] Based on the above response matrix set, the system performs the main response path identification operation. Each response dimension will be sorted on the spatial grid nodes in the corresponding direction according to the absolute value of the first-order derivative of the response value between adjacent points. The continuous point chain with the largest response growth rate is identified as the candidate segment of the main response path. In order to eliminate the erroneous path caused by sensor noise or isolated disturbances, three physical consistency constraints need to be introduced: the magnetic flux path should be continuous and satisfy the closed loop; the electric field trajectory should conform to the consistency of the direction change of the potential gradient in space; and the complex impedance path should follow the admittance continuity principle in phase change. Only the response paths that meet the above physical conditions are retained, and finally three types of spatial trajectories are output: the main magnetic field gradient trajectory, the capacitance energy aggregation trajectory, and the complex impedance offset path.

[0088] A track fitting model is further established for these three types of trajectories. Taking into account the discreteness of actual sensor data and the presence of certain errors, the fitting process uses B-spline or polynomial fitting to make the path physically continuous and smooth in the structural space. The fitting trajectory is discretized into a set of track nodes according to the circle number dimension. Each node records the circle number index of the current position, the direction vector of the track at that point (indicating the direction of track propagation), the instantaneous response amplitude (such as magnetic field strength, capacitance density, complex impedance modulus) and the rate of change between adjacent nodes to form a four-tuple data structure. The number of track nodes is set in proportion to the spatial resolution of the sensor array to ensure fitting accuracy.

[0089] Because disturbances during the production process can cause nonlinear deformations in certain trajectory segments, the system further performs dynamic disturbance compensation on the trajectory fitting process to improve the expressiveness of the trajectory in abnormal sections. This process uses previously collected production parameters, including the standard deviation of wire tension, the fluctuation rate of the wire pitch, and the first-order derivative of the ambient humidity change, to perform weighted corrections to the disturbance factors at points where the rate of change in the trajectory suddenly changes. For example, if a sudden change in the magnetic field direction occurs in a certain trajectory segment and a localized drop in tension is detected at the same time, the trajectory fitting weight at that node will be increased, and the amplitude of the change in the fitting curve at this point will be moderately enhanced, thereby making the trajectory expression sensitive to disturbances.

[0090] All processed orbital nodes are fused and encoded into a set of orbital vectors. This set is indexed by the number of circles, with each circle containing several path segments as sub-indexes. Each path segment is categorized by orbital type: magnetic field, capacitance, and impedance. Each orbital segment contains the following fields: orbital type label, path direction vector, response amplitude, rate of change, perturbation sensitivity score, spatial position confidence of the node, and a scale normalization factor within the unit space. This orbital vector set will be used in subsequent steps to compare with the baseline model to determine whether the evolution deviation exceeds the tolerance range. It will also serve as a physical structure reference for tasks such as dynamic threshold calculation and perturbation attribution comparison.

[0091] Through the above-mentioned continuous, nested, and layer-dependent processing path, the system not only realizes the spatial behavior modeling of multi-source physical signals during the plug-in inductor winding process, but also constructs a physical orbit vector representation system with highly expressive capabilities.

[0092] Step S103: Compare the extracted evolutionary characteristic parameters with the ideal characteristic parameters under the same winding schedule in the benchmark model, and establish a mapping relationship between the physical field deviation and the production disturbance in combination with the collected production parameters. Identify the dominant disturbance source causing the deviation through the preset deviation attribution mechanism, and obtain a disturbance type label. The disturbance type label carries causal information related to the change in production parameters, which is used to guide subsequent compensation control.

[0093] After completing real-time acquisition of the plug-in inductor's physical field and extracting its evolving characteristic parameters, the system then compares the extracted actual characteristic values against the previously constructed ideal benchmark model to identify the sources of deviations and disturbance mechanisms during the winding process. The core of this step is to establish a stable and generalizable deviation judgment and attribution mechanism, enabling the system to automatically identify the primary cause of the current winding deviation without human intervention and generate information tags to guide subsequent compensation decisions.

[0094] To ensure physical comparability, the system must ensure that the evolutionary characteristic parameters are aligned with the ideal values in the baseline model based on the number of turns, time, or hierarchical index. Specifically, the characteristic quantities such as the magnetic field distribution, electric field characteristics, or impedance response corresponding to each turn or winding section must be compared with the corresponding indicators at the same position or progress in the baseline model. The comparison process can be performed through various methods such as setting deviation tolerance intervals, Euclidean distance measurement, cosine similarity analysis, and principal component offset trajectories. The system can select the appropriate comparison strategy based on different modeling methods and quantify the comparison results.

[0095] After completing the feature comparison, the system will obtain a set of data results that describe the degree of deviation in the direction of each physical quantity. These results themselves are only a reflection of static differences and cannot directly explain the physical reasons behind the deviation. Therefore, it is necessary to introduce a disturbance attribution mechanism based on experience and simulation. This mechanism identifies the dominant physical disturbance source behind the deviation by matching the deviation characteristics with a set of preset disturbance response patterns. The disturbance attribution mechanism can be implemented based on an expert system, a rule matching algorithm or a training model (such as a discriminant model based on a decision tree or fuzzy logic). Its knowledge base comes from the "deviation form-disturbance type" mapping library established by historical process data, experimental research or simulation analysis.

[0096] In actual applications, when the system identifies phenomena such as an abnormal increase in the magnetic field gradient, asymmetric electric field coupling flux, or impedance spectrum drift in a certain winding cycle, it will match these feature combinations with typical disturbance patterns in the mapping library. For example, if the magnetic field is detected to exhibit multi-order distortion in the axial direction, the system can determine that its source is a short-term fluctuation in the winding tension; if the electric field energy concentration is detected to deviate from the central area and is accompanied by a decrease in the local wire arrangement density, it may be attributed to an instability in the wiring rhythm or a cross-layer displacement error caused by mechanical vibration. Each identification result will correspond to a disturbance type label, which not only indicates the physical source of the disturbance, such as "tension fluctuation", "pitch deviation", "surge in ambient humidity", "uneven wire diameter", etc., but also carries causal chain information between the disturbance and a certain type of production parameter change.

[0097] To ensure operability, the disturbance type labels must be organized in a structured manner, including the disturbance type name, potentially controlled process parameters, the affected evolutionary feature categories, recommended compensation directions, and logical associations with production parameters. For example, a label might indicate: "The current deviation pattern corresponds to a tension disturbance. It is recommended to increase the initial tension setting of the wire feed mechanism for the next several turns. The deviation amount is positively correlated with the tension change trend over the past five turns." Such labels facilitate automatic interpretation by the control system and provide direct decision support for subsequent fine-tuning strategies.

[0098] Through this comparison and attribution mechanism, the system no longer simply determines whether the inductance deviates from the target state, but can further explore the causes of the deviation and form a structured representation.

[0099] Furthermore, the extracted evolutionary characteristic parameters are compared with the ideal characteristic parameters under the same winding schedule in the benchmark model. Combined with the collected production parameters, a mapping relationship between physical field deviations and production disturbances is established. The dominant disturbance source causing the deviation is identified through a preset deviation attribution mechanism, and a disturbance type label is obtained, including:

[0100] After comparing the extracted evolutionary characteristic parameters with the ideal characteristic parameters under the same winding schedule in the benchmark model, a set of multidimensional deviation index vectors is output. The deviation index vectors are composed of magnetic field gradient difference, capacitive coupling energy difference, and impedance phase angle offset as the main dimensions, and are accompanied by confidence weights and spatial position index information for each feature;

[0101] The deviation index vector is jointly analyzed with a set of synchronous production parameters currently collected, including wire tension, spindle speed, wire arrangement pitch, and ambient temperature and humidity. This set of joint data is used as input to call a pre-established deviation disturbance mapping database for matching reasoning. The database is obtained by fusion training of experimental data, finite element simulation data, and historical process data. The mapping database is constructed according to a hierarchical index of the disturbance feature space and presets a causal structure between disturbance type and deviation pattern.

[0102] During the mapping reasoning process, a combined attribution path calculation based on fuzzy rule reasoning and gradient decision tree algorithm is performed. For each set of deviation input data, the corresponding possible disturbance type, the causal trigger chain of the corresponding production parameters, the physical field evolution disturbance direction and the response time window prediction results are output;

[0103] After obtaining the attribution reasoning result, a structured disturbance type label is generated. The disturbance type label includes at least a disturbance type field, a source parameter field, a presumed causal chain field, a recommended intervention direction field, and a disturbance judgment confidence score field. Each field remains independent in structure and has a calling interface identifier.

[0104] The disturbance type label is encoded into a standardized JSON data structure and embedded in the process control instruction frame of the current winding control cycle, serving as input content for subsequently generating compensation instructions for fine-tuning the winding behavior based on the disturbance type label and the corresponding production parameter change trend.

[0105] During the winding process, the system extracts evolutionary characteristic parameters, typically expressed as multidimensional vectors. These include, but are not limited to, the spatial gradient curvature distribution of the magnetic field, the variation in capacitive coupling energy density with layer evolution, and the phase offset of the complex impedance at a specified test frequency. These parameters should be indexed by the number of turns, spatial coordinates, and sampling timestamps to ensure a one-to-one correspondence with the ideal characteristic parameters in the benchmark model.

[0106] During the comparison, the system calculates the difference between the actual evolving characteristic parameters for each winding turn or time segment and the ideal characteristic parameters for the same winding schedule in the baseline model, generating a set of deviation indicators. For example, if the measured magnetic field main direction gradient corresponding to a winding turn exhibits a discontinuous abrupt change, the system identifies it as a magnetic field gradient deviation exceeding the preset offset tolerance. Alternatively, if the extracted capacitive coupling flux in a layer of wire structure shows an abnormal increase in concentration intensity in a certain adjacent region, accompanied by an increase in wire spacing, the system uses this as a key quantitative indicator of capacitive coupling offset. Similarly, under AC testing, if the complex impedance phase angle corresponding to a particular winding turns exhibits a positive or negative inflection point compared to the ideal curve, the system identifies an abnormal impedance behavior. All these deviation indicators are organized into a unified data structure, forming a multidimensional deviation indicator vector. Each dimension contains not only the difference in the physical quantity itself, but also a corresponding feature confidence score (calculated from the sampling signal-to-noise ratio or fitting residuals) and an index of its spatial location, supporting the subsequent localization and analysis of spatial attribution paths.

[0107] After obtaining the deviation index vector, the system synchronizes it with the currently collected production parameters. The production parameters required here are not limited to single momentary values but should include short-term trends, such as the instantaneous fluctuation amplitude of wire tension, short-term acceleration of spindle speed, variance of wire pitch, and the first-order derivative of ambient temperature and humidity. These parameters are normalized and input into the attribution analysis logic module along with the deviation vector.

[0108] The core of this attribution analysis module lies in the use of a pre-established deviation disturbance mapping database. This database contains data from at least three sources: first, a large amount of historical process data collected through actual winding equipment, which records disturbance events that have occurred during the production process and their corresponding deviation patterns; second, a data set obtained by simulating the plug-in inductor structure under different disturbance conditions using a finite element electromagnetic simulation platform, specifically used to cover the deviation response morphology under extreme working conditions; and finally, disturbance-response matching samples obtained through laboratory control variable experiments, which are used to verify model stability. The entire database is indexed and classified according to multi-dimensional feature spaces such as disturbance type, disturbance occurrence area, disturbance intensity level, and deviation dimension index to ensure query efficiency.

[0109] When performing mapping reasoning, the system calls the above-mentioned database and makes a joint judgment based on the input deviation index vector and synchronous production parameters through the fuzzy rule reasoning system and the gradient decision tree model. The fuzzy reasoning system mainly targets problems with continuity and fuzzy boundaries in physical relationships, such as whether a certain type of slight tension fluctuation is sufficient to trigger outlier behavior of coupled energy, while the gradient decision tree algorithm is used to handle the classification of disturbance types, identify disturbance-dominant parameters, and provide branch decision paths. Joint reasoning can output disturbance attribution results, including the identified disturbance type (such as tension disturbance, pitch fluctuation, structural vibration, or temperature and humidity shock), possible triggering causal chains (such as tension disturbance → density drop in the winding layer → magnetic field gradient offset → inductance drop), the main direction of physical field interference (such as axial degradation or radial deflection), and the estimated impact duration interval of the disturbance.

[0110] After obtaining the aforementioned attribution inference results, the system constructs a set of structured disturbance type labels. Each label must include a disturbance type field (e.g., "local capacitive coupling peak drift caused by uneven cross-layer wiring"), a source parameter field (e.g., "instantaneous fluctuation in wiring pitch exceeds the mean ±8%"), an inferred causal chain field (e.g., "pitch fluctuation → increased inter-wire spacing → enhanced capacitive coupling → local impedance anomaly"), a recommended intervention direction field (e.g., "tighten the current layer wiring pitch and compress the lateral margin of the wiring window"), and a disturbance judgment confidence score field (typically using a probability output in the range of 0–1, or a consistency score for multi-model fusion). The data type of each field must be predefined, for example, using a string enumeration for the type field and a floating-point number for the confidence score.

[0111] After the structure of the above-mentioned disturbance type tags is completed, unified data encapsulation is required. The recommended organizational structure is JSON format to ensure that the fields have a key-value correspondence and can be called and distributed by the control platform's standard protocol. In the actual system, this JSON-formatted disturbance tag will be encapsulated and embedded in the process control instruction frame generated by the current winding cycle. It will be aligned with the control platform's digital communication protocol, ensuring that the lower-level computer driver layer can directly read the tag information and execute subsequent compensation instructions for fine-tuning the winding behavior without increasing the control communication burden.

[0112] Through the above complete process, the system can accurately identify the disturbance type and trigger mechanism before the physical field deviation just occurs or exceeds the process tolerance, providing the control system with advance feedback and targeted intervention basis, significantly improving the process adaptability, finished product consistency and process closed-loop adjustment capabilities of inductor products.

[0113] Furthermore, during the mapping reasoning process, a combined attribution path calculation based on fuzzy rule reasoning and gradient decision tree algorithm is performed, including:

[0114] The deviation index vector is subjected to feature selection and importance ranking. By introducing a set of dynamic feature correlation evaluation mechanisms, a feature weight vector of the current sample is generated based on the contribution of the disturbance response fluctuation of each dimensional feature (including magnetic field gradient difference, capacitive coupling energy difference, and impedance phase angle offset) in the historical disturbance sample. This is used to guide the priority of the main cause fields in the subsequent reasoning path.

[0115] Based on the feature weight vector, a multi-conditional fuzzy logic rule set constructed by integrating expert experience and training data is loaded into the fuzzy rule inference system. Each rule outputs an intermediate causal trigger level label based on the joint membership relationship between one or more physical deviation dimensions and their corresponding production parameters, which is used to preliminarily narrow down the candidate set of disturbance types.

[0116] The intermediate disturbance labels and their confidence scores output by the fuzzy inference results are input as prior conditions into a gradient decision tree model. The gradient decision tree model uses the disturbance causal graph as a training structure, and uses the disturbance label confidence, the first-order derivative change rate of the production parameter time series curve, and the multi-scale characteristics of the disturbance occurrence time window as input nodes to determine the most likely disturbance type and its space-time causal path step by step.

[0117] During the model inference process, the structure mapping table in the perturbation feature space index database is called in real time to verify whether there is a spatiotemporal overlap between the interruption point of the current inference path and the physical field evolution trajectory. If there is a match, the confidence of the inference path is enhanced; otherwise, the confidence is downgraded and a suboptimal branch is triggered to retry;

[0118] The output results of the combined reasoning, including the disturbance type, the corresponding production parameter name and fluctuation direction, the predicted physical field interference propagation direction, and the predicted disturbance impact duration window, are output in a unified data structure. At the same time, the feature dimensions, model branch sequence, and maximum information gain point number called by the current reasoning path are recorded for subsequent disturbance identification review based on path consistency.

[0119] During the actual plug-in inductor winding process, the system continuously extracts physical field evolution characteristic parameters obtained by comparison with the baseline model. Combined with real-time production parameters such as wire tension, wire pitch, spindle speed, and ambient temperature and humidity, it generates a multidimensional deviation index vector for deviation identification. This deviation index vector serves as input for disturbance attribution analysis and first undergoes feature selection and importance ranking.

[0120] In order to effectively avoid the problem of information dilution caused by indiscriminate participation in calculations in a multi-dimensional parameter space, this method introduces a dynamic feature correlation evaluation mechanism. This mechanism is based on a pre-built disturbance label sample library, and statistically analyzes the contribution of each dimension's deviation features in historical disturbance events to the fluctuation of the disturbance type judgment results. For example, if the magnetic field gradient difference shows a significant response in all tension disturbance samples, while the phase angle offset changes less in such disturbances, the magnetic field gradient will receive a higher weight. This correlation evaluation is usually quantified by information gain or mutual information value, and the mean weight within the category is extracted based on the disturbance sample clustering results to output a set of feature weight vectors. This vector not only reflects the deviation dimension that the current sample should focus on, but also determines the priority matching order of subsequent fuzzy rules and decision paths.

[0121] After completing the feature weight sorting, the system inputs the vector into the fuzzy rule inference system for the first stage of attribution calculation. The fuzzy rule base consists of two parts: one is a set of rules constructed by experts based on long-term experience, such as "when the magnetic field gradient is high, the tension fluctuation is medium, and the ambient humidity change is high, there may be a loose winding disturbance caused by a sudden drop in tension"; the other is a membership rule automatically summarized from historical annotated data through supervised machine learning. Each rule maps one or more physical deviation dimensions and their corresponding production parameters into fuzzy membership (such as "low", "medium", "high"), and then outputs an intermediate label of the disturbance type (such as "suspected tension disturbance"), accompanied by a confidence score. The score is determined based on the strength of the membership overlap interval of the current sample matching rule and the historical hit rate statistics.

[0122] The intermediate disturbance labels and confidence scores output by fuzzy inference serve as prior inputs to a gradient decision tree model for the second stage of refined path judgment. This gradient tree model is trained based on a disturbance causal graph structure, where nodes represent the combined features of physical deviation dimensions and production parameters, and edges represent logical causal relationships. During the prediction phase, the model uses the disturbance label confidence, the rate of change of the first-order derivative of each production parameter's time series curve (e.g., the instantaneous slope of tension decline), and multi-scale windows of the disturbance's occurrence time window (e.g., mean shift within 0.5s, 1.0s, and 2.0s) as input variables. Information gain is evaluated step by step, and the most discriminative decision path is automatically selected. If multiple candidate branches with similar outputs exist during this process, the system prioritizes the optimal primary path based on the path's historical hit rate, retaining the remaining paths for subsequent retries.

[0123] To enhance the structural consistency judgment of attribution reasoning, this method calls the built-in disturbance feature space index database in real time during the execution of model reasoning. This database associates and maps each disturbance pattern with its corresponding physical field evolution trajectory through space-time labels. When the time node of the current reasoning path coincides with the physical field trajectory, for example, if a magnetic field gradient anomaly point in the trajectory matches the disturbance outbreak time point predicted by the model, the system will increase the overall confidence of the reasoning path; if there is no obvious match, the model will automatically reduce the weight of the path and trigger the retry mechanism of the next alternative branch to prevent the wrong path from solidifying into the result.

[0124] Ultimately, the output of the combined reasoning will include: disturbance type (such as tension disturbance, pitch drift disturbance, etc.), the name of the corresponding main production parameter (such as wire tension) and its fluctuation direction (such as continuous decline), the predicted physical field disturbance propagation direction (such as radial expansion of the winding), and the duration window of the predicted disturbance impact (such as 6 to 10 turns or 0.8 to 1.2 seconds). At the same time, the system records the deviation indicator dimension number called by the current reasoning path, the branch sequence number of the gradient tree model passed through, and the feature point number where the maximum information gain appears in the path, and saves them to the disturbance identification cache for subsequent review or path consistency comparison in the case of continuous disturbance superposition.

[0125] Step S104: Generate compensation instructions for fine-tuning the winding behavior based on the disturbance type label and the corresponding production parameter change trend. The compensation instructions include: adjusting the tension curve of the wire feeding mechanism to offset the influence of the tension disturbance, correcting the wire arrangement rhythm to alleviate the displacement deviation, optimizing the wire arrangement path to repair the distribution uniformity, or adjusting the wire arrangement density to compensate for the wire diameter change; executing the compensation instructions to compensate for the winding behavior.

[0126] After generating the disturbance type label, the system enters the compensation control phase. Its core task is to generate and execute a set of targeted winding compensation instructions in real time based on the disturbance type identified in the previous phase and the corresponding production parameter change trends. These compensation instructions must not only reflect the response relationship to the disturbance but also the logical compatibility between the specific execution action and the disturbance characteristics. This ensures that the winding behavior can be finely corrected without stopping operation, thereby guiding the physical field evolution state to gradually return to the target trajectory defined by the ideal model.

[0127] The generation of compensation instructions uses the disturbance type label as input. This label typically contains information such as the disturbance category (such as tension fluctuation, pitch anomaly, sudden change in ambient humidity, wire diameter deviation, etc.), the occurrence section (corresponding to the range of turns or structural location), the impact direction (such as the magnetic field distribution form, electric field coupling path or impedance phase trajectory), and the recommended intervention method (active compensation or delayed adjustment). The system first parses the label content and combines it with the real-time updated production parameter curve (such as the tension history sequence, spindle speed trend, wire feed rate, etc.) to assess whether the disturbance impact is continuing, expanding, or stabilizing, thereby determining the dynamic strength and execution range of the compensation instruction.

[0128] For example, if the system detects a brief drop in tension during a certain section of the winding process, and this tension disturbance has been identified as the primary source of magnetic field gradient anomalies, the control system will immediately generate a compensation instruction to adjust the tension curve of the wire feed mechanism. This instruction is not simply executed by "increasing the tension setpoint," but rather by fine-tuning the instantaneous acceleration of the wire feed mechanism in the form of a curve, or locally modifying the closed-loop PID parameters of the constant tension motor to ensure that the tension recovery action matches the local density of the winding, thereby compensating for the wire density gap on a turn-by-turn basis without causing reverse shock.

[0129] If the disturbance type label indicates an unstable wiring rhythm, such as an offset in the stacking of wire layers due to mechanical gap fluctuations, the system will fine-tune the wiring stepping rhythm by correcting it. This instruction acts on the drive pulse logic of the wiring stepper motor, adjusting its lateral displacement per revolution or optimizing the wire laying carry path through interpolation algorithms, achieving a smoother and more consistent wiring rhythm per unit time, thereby suppressing discontinuous deformation of the winding structure.

[0130] In some special disturbance types, such as abnormal electric field coupling caused by a sudden increase in ambient humidity or a change in wire batches, the system may need to implement compensation strategies such as "optimizing the wiring path" or "adjusting the wiring density." These strategies usually do not involve direct acceleration of moving parts, but rather modify the local structural micromorphology by controlling the wiring starting position, carry interval, and the transposition method between layers. For example, the system can slightly offset the wiring coordinates to achieve a more uniform spacing distribution across the coil layer interface, or actively insert compensation pitches and reduce edge empty windings when a slight deviation in wire diameter is detected, thereby balancing the impact of volume errors on inductance parameters.

[0131] All compensation instructions are transmitted to each actuator drive control unit in the form of structured control data and executed in real time without causing winding interruptions. The executed winding behavior is then re-collected by the system and fed back to the physical field distribution snapshot extraction module, forming a closed-loop process of perception, judgment, execution, and feedback. The system continuously evaluates whether the compensated physical field is converging toward the baseline model, converging the compensation behavior after reaching the tolerance threshold, or re-triggering the disturbance identification and strategy update process if compensation failure trends emerge.

[0132] Through this closed-loop dynamic compensation mechanism, triggered by disturbance tags, modulated based on production parameter trends, and fine-tuned using control curves, the present invention enables precise control of multiple uncertainties during the winding process, including inductance deviation, structural non-uniformity, and electrical performance drift. This approach, independent of static rules and without mechanical interruptions, significantly enhances the adaptability and product consistency of the plug-in inductor manufacturing process. This compensation approach can be implemented through software upgrades of existing winding equipment and is suitable for the precise manufacture of plug-in I-shaped inductors in automated mass production lines.

[0133] Furthermore, generating compensation instructions for fine-tuning the winding behavior according to the disturbance type label and the corresponding production parameter change trend includes:

[0134] After identifying the disturbance type tag, a specific fine-tuning control method is selected based on the disturbance type field and source parameter field in the tag, including tension change trajectory reconstruction for tension disturbance, lateral step trajectory correction for line tracing rhythm disturbance, and path interpolation adjustment for line tracing path uniformity deviation;

[0135] For tension disturbances, obtain the starting number of disturbance turns and the disturbance direction. Based on the recommended intervention direction field in the disturbance type tag, set the target tension adjustment interval. Construct a time interval of three to five turns starting from the disturbance starting turn. Within this interval, adjust the rate of change of tension output by the wire feeding mechanism so that the tension increases or decreases at a fixed slope in the number of turns dimension, so that it falls within the target tension range at the end of the compensation cycle. The tension adjustment process does not experience step changes, and the first-order derivative remains continuous.

[0136] In response to the disturbance in the winding rhythm, the lateral winding position index of the disturbance location is combined with the lateral stepping amount of the winding driver to be adjusted circle by circle within the range of the remaining number of turns before the winding layer is changed. The adjustment amount is determined by the product of the disturbance confidence score and the mean square error of the current winding pitch. This makes the lateral spacing of the wire arrangement in the layer gradually return to the historical average pitch, avoiding the phenomenon of dense and uneven coils in the edge area of the layer.

[0137] To address capacitance distribution anomalies caused by non-uniform wiring paths, the system reads the deviation spatial position index to locate the specific disturbance area. Two to three transition point coordinates are inserted near this area, and the subsequent wiring trajectory is adjusted using linear interpolation or equidistant interpolation to slightly offset the centerline of the conductor in space, thereby spatially expanding the capacitance concentration area and smoothing out changes in coupling strength.

[0138] During the compensation process, the evolution characteristic parameters corresponding to the disturbance type label are continuously monitored, and based on the changing trends of the magnetic field gradient, capacitive coupling energy, and impedance phase angle offset in the comparison results, it is determined whether the disturbance has naturally subsided. When it is detected that the variation amplitude of the physical deviation remains within the tolerance range of the reference model for two consecutive winding cycles, and the disturbance type label has not been updated for two consecutive cycles, the compensation is considered to be completed and the compensation process is terminated.

[0139] The parameters of all adjustment paths during the compensation period will be written into the dynamic record stack and used to determine the correlation between the historical compensation action and the current disturbance during the next disturbance identification process, thereby achieving historical closed-loop reinforcement between the deviation path and the compensation trajectory.

[0140] In this method, after the system generates a disturbance type tag through the previous steps, it enters the compensation instruction generation phase. The system first parses the multiple fields in the disturbance type tag, specifically the "disturbance type field" and the "source parameter field." The disturbance type field identifies the type of physical or structural anomaly to which the current deviation belongs, such as tension disturbance, wire routing rhythm disturbance, or wire routing path uniformity disturbance. The source parameter field indicates which type of production parameter fluctuation triggered the disturbance, such as a short-term drop in tension, an abnormal increase in wire routing pitch, or a sudden change in humidity causing a change in insulation performance.

[0141] Based on the combined information from these two fields, the system selects a corresponding control strategy and generates fine-tuning instructions that can be directly sent to the driver layer. Specifically, in the case of tension disturbances, the system first reads the winding number at the time of the disturbance and identifies whether the disturbance indicates "tension below expected" or "tension above expected." The system then refers to the adjustment suggestions provided in the "Suggested Intervention Direction" field in the tag (such as incremental compensation or incremental relief) to set a clear tension target range, for example, within a target window of ±5%. The system then sets a compensation cycle starting from the current winding number and extending back three to five turns. During this cycle, the control signal to the wire feed motor is adjusted so that the wire tension gradually approaches the target tension window at a rate that is approximately linear as the winding number increases, rather than abruptly changing. The control logic requires that the rate of change of tension remain continuous, meaning that the first-order derivative of the tension compensation curve does not undergo abrupt changes to avoid inertial overshoot and secondary disturbances.

[0142] When a disturbance in the wiring rhythm is detected, the system will first locate the lateral wiring position index where the disturbance occurs, that is, the lateral coordinate interval of the wire on the wiring platform. On this basis, the system will evaluate whether the number of remaining turns in the current inductor structure is sufficient for rhythm correction, and select a reasonable number of remaining turns as the compensation action area. Within this compensation action area, the system adjusts the lateral step amount of the wiring motor turn by turn. The adjustment amplitude is not fixed, but is obtained by multiplying the statistical volatility of the current wiring pitch (usually calculated using the mean square error) by the confidence score in the disturbance label. This strategy can achieve a larger correction response in areas with high disturbance intensity, while maintaining minimal intervention in areas with slight disturbance intensity. The goal is to gradually return the lateral wire spacing in the current layer to the historical average wiring pitch to restore structural uniformity and avoid the formation of cross-layer pressure or winding deflection caused by rhythm misalignment.

[0143] For abnormal capacitance distribution caused by local non-uniformity in the wiring path, the system will locate the disturbance specifically to an actual position in the wire path based on the deviation spatial position index provided in the disturbance label. Then, two to three transition points are inserted before and after the local area, and linear interpolation or equidistant interpolation is used to reconstruct the wire path. This means that without changing the winding level or number of turns, the spatial direction of the wire is slightly offset, so that the wiring shows a slight expansion or compression in the disturbance area, thereby alleviating capacitance concentration and weakening the sudden change in electric field coupling strength in the electromagnetic space.

[0144] During the execution of any of the above compensation strategies, the system does not automatically terminate the compensation action immediately. Instead, it initiates a real-time monitoring mechanism. This monitoring mechanism continuously compares the evolutionary characteristic parameters corresponding to the compensation process with the ideal characteristic values of the circle or region in the baseline model. Characteristic parameters include the rate of change of the magnetic field gradient, the rate of change of the capacitive coupling energy density, and the trajectory of the impedance phase angle change. Once the system detects that the deviations of all corresponding dimensions are within the tolerance range of the baseline model for two consecutive winding cycles, and the disturbance type label has not been updated for two consecutive cycles, it is considered that the disturbance has naturally subsided and the compensation has achieved its purpose. The current compensation behavior is then terminated to avoid ineffective or excessive intervention.

[0145] At the same time, the system writes all adjustment path parameters used in this compensation process, including the tension adjustment path, step correction coefficient, interpolation coordinate offset, etc., into a dynamic record stack with time index and disturbance type as the primary key. This record stack will be called up during the next round of disturbance identification to analyze whether the newly identified disturbance is similar to a previously compensated path. If the similarity meets the set threshold, the system can directly reference the historical compensation path, further reducing response time and adjustment cost, thereby forming an enhanced closed-loop matching mechanism between the disturbance path and the compensation trajectory.

[0146] Through the above steps, the system can not only accurately classify and grade specific disturbances, but also achieve continuous, nonlinear, and target-controllable compensation behavior throughout the entire process through multi-circle range, dynamic curvature, feedback convergence, and historical reinforcement mechanisms.

[0147] Although the present application is disclosed as above with the preferred embodiments, it is not intended to limit the present application. Any person skilled in the art may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A plug-in inductor winding process control method based on dynamic compensation of production parameters, characterized in that: include: Based on the target inductor parameters and the plug-in inductor structure, a benchmark model describing the ideal physical field evolution characteristics is constructed. The benchmark model is used to characterize the physical field changes in the inductor core area under ideal winding conditions with the number of turns or over time. A non-contact sensor array placed in the winding area collects physical field signals, including magnetic, electric, or impedance signals. Combined with sampled data of current production parameters, including wire tension, spindle speed, wire pitch, and ambient temperature and humidity, this generates a snapshot of the physical field distribution at the current level and extracts evolutionary characteristic parameters that match the benchmark model. The extracted evolutionary characteristic parameters are compared with the ideal characteristic parameters under the same winding schedule in the benchmark model. Combined with the collected production parameters, a mapping relationship between physical field deviations and production disturbances is established. The dominant disturbance source causing the deviation is identified through a preset deviation attribution mechanism, and a disturbance type label is obtained. Based on the disturbance type label and the corresponding production parameter change trend, compensation instructions for fine-tuning the winding behavior are generated. The compensation instructions include: adjusting the tension curve of the wire feeding mechanism to offset the influence of tension disturbance, correcting the wire arrangement rhythm to alleviate displacement deviation, optimizing the wire arrangement path to restore distribution uniformity, or adjusting the wire arrangement density to compensate for wire diameter changes; executing the compensation instructions to compensate for the winding behavior.

2. The plug-in inductor winding process control method based on dynamic compensation of production parameters according to claim 1 is characterized in that: The benchmark model was established using electromagnetic field finite element simulation. The simulation process used the magnetic flux density distribution, magnetic field intensity distribution, and local electric field response within the capacitive coupling surface after each winding of the plug-in inductor structure as reference indicators. The simulation input included the target inductor parameters, the plug-in inductor structure, and the wire arrangement. The material permeability, saturation magnetic flux density, and dielectric constant of the insulating material were also introduced as modeling parameters. During the simulation process, multiple characteristic points are extracted from the impedance frequency response curve, including the maximum phase difference point, the minimum impedance point, the impedance mutation inflection point, and the critical frequency band boundary. This set of frequency response characteristics is used as auxiliary reference parameters to describe the winding evolution process under ideal conditions, and is used together with the magnetic flux density distribution and the electric field distribution to construct the physical field evolution path of the benchmark model. The output of the benchmark model is in the form of a structured vector. The structured vector includes an index of the number of winding turns, an index of the evolutionary characteristic parameter corresponding to each turn, and a tolerance range field corresponding to each characteristic parameter. The structured vector is aligned and compared with the evolutionary characteristic parameters extracted in real time during the actual winding process of the plug-in inductor. The structured vector serves as the only reference for comparing the extracted evolution characteristic parameters with the ideal characteristic parameters under the same winding progress in the benchmark model in the subsequent steps, and is consistent with the deviation input data structure required to identify the dominant disturbance source causing the deviation through the preset deviation attribution mechanism.

3. The plug-in inductor winding process control method based on dynamic compensation of production parameters according to claim 1, characterized in that: The sampling data of the current production parameters, including wire tension, spindle speed, wire pitch, and ambient temperature and humidity, is combined to generate a snapshot of the physical field distribution at the current level and extract evolutionary characteristic parameters that match the benchmark model, including: Synchronously processing the raw physical field signals collected by the non-contact sensor array and the sampled data, normalizing the outputs of different sensing channels based on the reference physical response values corresponding to the number of winding turns, and obtaining standardized multi-channel equivalent physical response curves for subsequent analysis; Based on the standardized equivalent physical response curve, combined with the arrangement coordinates of each turn of the wire and the trend of the magnetic flux density change, the local physical field boundary in the current inductor winding is identified, and based on the boundary information, the entire winding area is divided into multiple local areas; In each local area, the magnetic field response path, capacitance coupling trajectory, and impedance phase change curve are extracted based on the spatial distribution of the equivalent physical response curve. The magnetic field gradient trajectory, capacitance energy trajectory, and complex impedance trajectory are generated through fitting processing, and a set of trajectory vectors representing the spatial physical evolution trend is output. Based on this set of orbital vectors, and in combination with the aforementioned standard deviation of wire tension, the first-order derivative of ambient humidity changes, the fluctuation rate of the wiring pitch, and the rate of change of the spindle speed, the dynamic threshold parameters used to adjust the comparison tolerance of each type of evolving orbit are calculated. This reconstructs the allowable offset range of the magnetic field orbit, the response anomaly tolerance of the capacitance orbit, and the phase error threshold of the impedance orbit. The orbital vector set adjusted by dynamic threshold is used as the final evolution characteristic parameter.

4. The plug-in inductor winding process control method based on dynamic compensation of production parameters according to claim 1, characterized in that: The extracted evolutionary characteristic parameters are compared with the ideal characteristic parameters under the same winding schedule in the benchmark model. Combined with the collected production parameters, a mapping relationship between physical field deviation and production disturbance is established. The dominant disturbance source causing the deviation is identified through a preset deviation attribution mechanism to obtain a disturbance type label, including: After comparing the extracted evolutionary characteristic parameters with the ideal characteristic parameters under the same winding schedule in the benchmark model, a set of multidimensional deviation index vectors is output. The deviation index vectors are composed of magnetic field gradient difference, capacitive coupling energy difference, and impedance phase angle offset as the main dimensions, and are accompanied by confidence weights and spatial position index information for each feature; The deviation index vector is jointly analyzed with a set of synchronous production parameters currently collected, including wire tension, spindle speed, wire arrangement pitch, and ambient temperature and humidity. This set of joint data is used as input to call a pre-established deviation disturbance mapping database for matching reasoning. The database is obtained by fusion training of experimental data, finite element simulation data, and historical process data. The mapping database is constructed according to a hierarchical index of the disturbance feature space and presets a causal structure between disturbance type and deviation pattern. During the mapping reasoning process, a combined attribution path calculation based on fuzzy rule reasoning and gradient decision tree algorithm is performed. For each set of deviation input data, the corresponding possible disturbance type, the causal trigger chain of the corresponding production parameters, the physical field evolution disturbance direction and the response time window prediction results are output; After obtaining the attribution reasoning result, a structured disturbance type label is generated. The disturbance type label includes at least a disturbance type field, a source parameter field, a presumed causal chain field, a recommended intervention direction field, and a disturbance judgment confidence score field. Each field remains independent in structure and has a calling interface identifier. The disturbance type label is encoded into a standardized JSON data structure and embedded in the process control instruction frame of the current winding control cycle, serving as input content for subsequently generating compensation instructions for fine-tuning the winding behavior based on the disturbance type label and the corresponding production parameter change trend.

5. The plug-in inductor winding process control method based on dynamic compensation of production parameters according to claim 1, characterized in that: Generating compensation instructions for fine-tuning winding behavior according to the disturbance type label and the corresponding production parameter change trend includes: After identifying the disturbance type tag, a specific fine-tuning control method is selected based on the disturbance type field and source parameter field in the tag, including tension change trajectory reconstruction for tension disturbance, lateral step trajectory correction for line tracing rhythm disturbance, and path interpolation adjustment for line tracing path uniformity deviation; For tension disturbances, obtain the starting number of disturbance turns and the disturbance direction. Based on the recommended intervention direction field in the disturbance type tag, set the target tension adjustment interval. Construct a time interval of three to five turns starting from the disturbance starting turn. Within this interval, adjust the rate of change of tension output by the wire feeding mechanism so that the tension increases or decreases at a fixed slope in the number of turns dimension, so that it falls within the target tension range at the end of the compensation cycle. The tension adjustment process does not experience step changes, and the first-order derivative remains continuous. In response to the disturbance in the winding rhythm, the lateral winding position index of the disturbance location is combined with the lateral stepping amount of the winding driver to be adjusted circle by circle within the range of the remaining number of turns before the winding layer is changed. The adjustment amount is determined by the product of the disturbance confidence score and the mean square error of the current winding pitch. This makes the lateral spacing of the wire arrangement in the layer gradually return to the historical average pitch, avoiding the phenomenon of dense and uneven coils in the edge area of the layer. For abnormal capacitance distribution caused by non-uniform wiring paths, the deviation spatial position index is read to locate the specific disturbance area, and two to three transition point coordinates are inserted near the area. The subsequent wiring trajectory is adjusted using linear interpolation or equidistant interpolation to cause a slight spatial offset in the center line of the wire, thereby spatially expanding the capacitance concentration area and smoothing the change in coupling strength.

6. The method for controlling the winding process of an insert inductor based on dynamic compensation of production parameters according to claim 5, characterized in that: The method of generating a compensation instruction for fine-tuning the winding behavior according to the disturbance type label and the corresponding production parameter change trend further includes: During the compensation process, the evolution characteristic parameters corresponding to the disturbance type label are continuously monitored, and based on the changing trends of the magnetic field gradient, capacitive coupling energy, and impedance phase angle offset in the comparison results, it is determined whether the disturbance has naturally subsided. When it is detected that the variation amplitude of the physical deviation remains within the tolerance range of the reference model for two consecutive winding cycles, and the disturbance type label has not been updated for two consecutive cycles, the compensation is considered to be completed and the compensation process is terminated. The parameters of all adjustment paths during the compensation period will be written into the dynamic record stack and used to determine the correlation between the historical compensation action and the current disturbance during the next disturbance identification process, thereby achieving historical closed-loop reinforcement between the deviation path and the compensation trajectory.

7. The plug-in inductor winding process control method based on dynamic compensation of production parameters according to claim 2, characterized in that: During the simulation process, multiple characteristic points are extracted from the impedance frequency response curve, including: During the electromagnetic field finite element simulation process, the impedance frequency response curve of each coil of wire is first simulated over the entire frequency range, and a set of characteristic points including the maximum phase difference point, the minimum impedance point, the impedance mutation inflection point, and the critical frequency band boundary are identified in the simulation results; In the set of identified feature points, the frequency response slope change, impedance phase angle fluctuation range and amplitude gradient curvature of each feature point are further calculated, and based on this, a frequency response perturbation sensitivity matrix is constructed to characterize the response amplification factor of each feature point to structural micro-perturbations; The frequency response disturbance sensitivity matrix is coupled with the local magnetic flux density distribution diagram of the plug-in inductor structure to analyze and filter out abnormal frequency mutation points appearing in the weak coupling section of the structure, ensuring that only effective response characteristic points with structural physical relevance are retained; Perform frequency band clustering processing on the retained valid feature point set, group and merge them based on bandwidth coverage and frequency proximity, and calculate the center frequency, main response mode and representative response characteristic value for each cluster segment to form a frequency band representation subset; The frequency band representation subset is used as an auxiliary reference parameter to construct the benchmark model. After being aligned with the magnetic flux density distribution and electric field distribution in the turn dimension, it is encoded in a unified data format as a frequency domain evolution path field in a structured vector. This allows for turn-by-turn comparison of the evolutionary characteristic parameters extracted in real time during the actual winding process of the plug-in inductor.

8. The plug-in inductor winding process control method based on dynamic compensation of production parameters according to claim 3 is characterized in that: In each local area, according to the spatial distribution of the equivalent physical response curve, the magnetic field response path, the capacitance coupling trajectory and the impedance phase change curve are extracted, and the magnetic field gradient trajectory, the capacitance energy trajectory and the complex impedance trajectory are generated through fitting processing, and a set of trajectory vectors representing the spatial physical evolution trend is output, including: In each local area, based on the multi-channel equivalent physical response curves collected and normalized by the contactless sensor array, directional projection is performed in the structural coordinate system. The magnetic field response, electric field coupling, and complex impedance phase change are projected onto the axial, radial, and planar transverse vector channels of the inductor winding, respectively, thereby constructing a response matrix set with direction labels. Using the aforementioned directional-labeled response matrix set, the main response path is identified along the three-dimensional spatial grid nodes in each local area. The main path identification algorithm adopts the principle of local maximum response growth rate and selects the continuous point chain with the largest absolute value of the first-order derivative of the response change as the candidate segment of the main response trajectory. In addition, the abnormal segments with inconsistent physical logic are eliminated by combining the magnetic flux continuity, electric field enclosure, and impedance-admittance consistency conditions. Finally, three types of spatial response trajectories are formed: the main magnetic field gradient trajectory, the capacitance energy aggregation trajectory, and the complex impedance offset path. Based on the three types of spatial response trajectories obtained, trajectory fitting models are established respectively based on the number of conductor turns and the spatial hierarchical structure. The fitting models use B-spline curves or polynomial fitting to smooth the trajectory curves and discretize each type of trajectory path into a set of trajectory nodes with a turn number index, path direction vector, response amplitude and change rate. For abnormal change points in the track node set, combined with the aforementioned production parameter sampling data, including the standard deviation of conductor tension, the fluctuation rate of the wire pitch, and the first-order derivative of humidity change, the weight of the disturbance influence factor is set on the fitting track, and dynamic compensation fitting is performed on the local track segment. This ensures that each track not only reflects the physical response itself, but also has structural sensitivity that is strongly correlated with external disturbance sources. The three types of orbital node sets are fused and encoded into an orbital vector set. The orbital vector set uses the number of circles as the main index, the spatial path segment as the sub-index, and the orbit type as the classification label. It also includes a disturbance sensitivity field, a position confidence field, and a spatial scale normalization factor field. This serves as the physical structure reference basis for subsequent deviation judgment, dynamic threshold calculation, and disturbance attribution comparison.

9. The method for controlling the winding process of an insert inductor based on dynamic compensation of production parameters according to claim 4, characterized in that: During the mapping reasoning process, a combined attribution path calculation based on fuzzy rule reasoning and gradient decision tree algorithm is performed, including: Perform feature selection and importance ranking on the deviation index vector. By introducing a set of dynamic feature correlation evaluation mechanisms, a feature weight vector for the current sample is generated based on the contribution of the disturbance response fluctuation of each dimensional feature in the historical disturbance sample. This is used to guide the priority of the main cause fields in the subsequent reasoning path. Based on the feature weight vector, a multi-conditional fuzzy logic rule set constructed by integrating expert experience and training data is loaded into the fuzzy rule inference system. Each rule outputs an intermediate causal trigger level label based on the joint membership relationship between one or more physical deviation dimensions and their corresponding production parameters, which is used to preliminarily narrow down the candidate set of disturbance types. The intermediate disturbance labels and their confidence scores output by the fuzzy inference results are input as prior conditions into a gradient decision tree model. The gradient decision tree model uses the disturbance causal graph as a training structure, and uses the disturbance label confidence, the first-order derivative change rate of the production parameter time series curve, and the multi-scale characteristics of the disturbance occurrence time window as input nodes to determine the most likely disturbance type and its space-time causal path step by step. During the model inference process, the structure mapping table in the perturbation feature space index database is called in real time to verify whether there is a spatiotemporal overlap between the interruption point of the current inference path and the physical field evolution trajectory. If there is a match, the confidence of the inference path is enhanced; otherwise, the confidence is downgraded and a suboptimal branch is triggered to retry; The output results of the combined reasoning, including the disturbance type, the corresponding production parameter name and fluctuation direction, the predicted physical field interference propagation direction, and the predicted disturbance impact duration window, are output in a unified data structure. At the same time, the feature dimensions, model branch sequence, and maximum information gain point number called by the current reasoning path are recorded for subsequent disturbance identification review based on path consistency.

Images