An inductance element quality evaluation method based on electrical parameter analysis

Abstract

The application relates to the field of measuring electric variables and discloses an inductive element quality evaluation method based on electric parameter analysis, which comprises the following steps: applying a step voltage pulse and synchronously collecting an actual input voltage and a loop response current, using the rising edge slope of the actual input voltage to normalize and correct the first-order change rate of the loop response current to generate a corrected change rate, calculating the corrected change rate by using time domain second-order difference to construct a second-order change rate, searching for local mutation extreme values to determine static inflection point time sequence characteristic values and slope variation coefficients and evaluate a turn-to-turn insulation deterioration state. The application realizes parameter evaluation without relying on external standard samples through self-referencing decoupling, cancels the front edge fluctuation of an excitation source and eliminates system dissipation caused by contact resistance jitter, and improves the detection sensitivity of early implicit defects.

Description

Technical Field

[0001] This invention relates to a method for evaluating the quality of inductor components based on electrical parameter analysis, belonging to the field of electrical variable measurement technology. Background Technology

[0002] In electrical parameter measurement and coil quality screening, component quality is typically judged by measuring static inductance and static quality factor. The basic approach is to use the lumped parameters measured under weak current excitation as an equivalent characterization of the component's circuit electrical response under high-frequency power pulse conditions. Failed components are then screened by comparing inductance deviations. However, when components operate under high-power alternating high-voltage pulse conditions, latent damage such as early-stage inter-turn insulation degradation, internal core defects, and abnormal parasitic capacitance often fails to manifest under weak current testing excitation due to insufficient excitation energy. Static parameters measured by conventional testing equipment usually do not show obvious abnormalities, allowing components with manufacturing defects to pass quality screening. Such components then continue to fail in subsequent operation. After being subjected to pulse voltage stress, inter-turn breakdown may occur, or the core loss may increase significantly, leading to heat generation and short-circuit failure. For example, Chinese invention patent application CN120722101A discloses a solenoid valve opening and closing state measurement system and method. It evaluates the opening and closing state of the solenoid valve by preprocessing the inductor sequence with Gaussian convolution filtering and extracting abrupt response factors based on the first and second derivatives of the dynamic inductor. However, the method is essentially an average smoothing process for the inductor trajectory. Although it can effectively suppress high-frequency noise, this preprocessing logic is prone to a significant decrease in evaluation sensitivity due to the loss of high-frequency information when capturing extremely weak local signal features in the early stage of inter-turn insulation degradation of the inductor element.

[0003] To identify nonlinear characteristics in transient waveforms, existing technologies typically use offline calibration curves of standard samples as a comparison benchmark. However, during automated continuous measurement, the contact resistance of the fixture contacts fluctuates randomly, high-frequency parasitic electromagnetic coupling exists in the test circuit, and the rising edge of the pulse excitation source is difficult to keep consistent. This makes it impossible for the offline calibration curve to accurately correspond to the real-time signal of the component under test. Using external standard samples for comparison will introduce interference components that are difficult to eliminate into the measurement data, and it is easy to misjudge the random fluctuation of contact resistance as an internal defect of the component. If it is improved by simply adding sensors or increasing the voltage amplitude, it will not only increase the complexity of the test circuit, but also cause high-frequency parasitic oscillations and aggravate waveform distortion.

[0004] Therefore, the technical problem to be solved by this invention is how to directly use the transient loop electrical variables of the inductor under test itself to correct the rising edge fluctuation of the excitation source without relying on the offline calibration curve of the external standard sample, and to combine the first-order transient rate of change sequence and the second-order transient rate of change sequence for analysis, so as to separate the nonlinear abnormal waveform from the high-frequency noise of the loop and accurately detect potential inter-turn insulation degradation and fine damage to the magnetic core. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A method for evaluating the quality of inductor components based on electrical parameter analysis, comprising the following steps: Step S1: Apply a step voltage pulse with a determined amplitude to the inductor under test, and control the rise time of the step voltage pulse to not exceed 20ns. Step S2: Within the transient time window after the step voltage pulse is applied, the loop response current of the inductor under test and the actual input voltage across the inductor under test are simultaneously acquired to obtain the discrete response current sequence and the actual input voltage sequence, wherein the sampling point numbers of the discrete response current sequence and the actual input voltage sequence are both integers. Step S3: Calculate the real-time rising edge slope of the actual input voltage sequence based on the ratio of the actual input voltage difference between adjacent sampling points to the sampling time interval, and calculate the first-order transient rate of change sequence of the discrete response current sequence based on the ratio of the loop response current difference between adjacent sampling points to the sampling time interval. Use the real-time rising edge slope to normalize and correct the amplitude of the first-order transient rate of change sequence to obtain the corrected first-order transient rate of change sequence. Step S4: The corrected first-order transient rate of change sequence is calculated by time-domain quadratic difference to construct the second-order transient rate of change sequence. Local abrupt extreme values ​​in the second-order transient rate of change sequence are retrieved to determine the static inflection point time series characteristic value and slope variation coefficient. The inter-turn insulation degradation state of the inductor under test is evaluated based on the static inflection point time series characteristic value and slope variation coefficient.

[0006] Preferably, the operation of constructing a second-order transient rate of change sequence by calculating the corrected first-order transient rate of change sequence through time-domain quadratic difference in step S4 includes the following sub-steps: Step S41, combining the first-order transient rate of change sequence and the second-order transient rate of change sequence to calculate the time-domain transient curvature value sequence corresponding to the sampling point number; Step S42, retrieving local abrupt extreme values ​​in the time-domain transient curvature value sequence to locate the static inflection point time-series characteristic value, and calculating the ratio of the standard deviation to the mean of the time-domain transient curvature value sequence in the neighborhood of the static inflection point time-series characteristic value as the slope variation coefficient, which is used to identify the nonlinear admittance distortion defect generated inside the inductor element under test.

[0007] Preferably, while obtaining the corrected first-order transient rate of change sequence in step S3, the method also includes the following extended sub-steps: Step S31, continuously acquiring the actual input voltage sequence and loop response current sequence of the inductor under test within the transient time window; Step S32, constructing a two-dimensional trajectory of voltage and current dynamic impedance by performing a two-dimensional projective mapping with the actual input voltage sequence as the vertical axis and the loop response current sequence as the horizontal axis.

[0008] Preferably, after constructing the two-dimensional trajectory of voltage and current dynamic impedance in step S32, the method further includes the following extended sub-steps: Step S33, projecting the two-dimensional trajectory of voltage and current dynamic impedance into a closed phase space, and calculating the area of ​​the two-dimensional geometric region enclosed by the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space, and evaluating the loss characteristics of the inductor under test under high-frequency alternating electric stress by using the area of ​​the two-dimensional geometric region.

[0009] Preferably, in step S33, after calculating the area of ​​the two-dimensional geometric region, the method further includes the following extended sub-steps: Step S34, calculate the position of the topological geometric center point of the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space, compare the static deviation value between the position of the topological geometric center point and the origin to obtain the geometric center drift, and nonlinearly quantify the abnormal magnetic hysteresis charge loss caused by batch differences of magnetic core material through the geometric center drift.

[0010] Preferably, the operation of applying a step voltage pulse with a determined amplitude to the inductor under test in step S1 includes the following sub-steps: Step S11, sequentially applying a series of step voltage pulses with equally increasing amplitudes to the inductor under test, and simultaneously acquiring the corresponding loop response current under each pulse input.

[0011] Preferably, under the state of applying a stepped voltage pulse sequence in step S11, the method further includes the following sub-step: Step S12, based on the difference in transient response of the electric charge under different amplitudes, extract the critical abrupt change point when the first-order transient rate of change sequence corresponding to each amplitude voltage pulse generates a sudden nonlinear jump, so as to determine the magnetic saturation threshold drift state of the inductor under test.

[0012] Preferably, the method further includes the following gain adjustment step: Step S5, using the time-domain local features of the corrected first-order transient rate of change sequence to perform feedforward feedback gain adjustment on the discrete response current sequence, so that the physical domain of the focus of electrical variable evaluation is transformed from static macroscopic inductance physical quantity to transient microscopic impedance topology network features.

[0013] Preferably, the following quantitative result output steps are included: Step S6, combining the slope variation coefficient to construct the electrical characteristic parameter matrix, inputting the electrical characteristic parameter matrix into a preset electrical parameter calibration mapping network for interval matching retrieval, and comparing and outputting the quality grading quantitative result of the inductor under test.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In the quality assessment of inductor components in electrical parameter analysis, the first-order transient rate of change sequence of the loop response current is normalized and corrected by using the real-time rising edge slope of the actual input voltage sequence. This can reduce the interference caused by the fluctuation of the contact resistance of the fixture contacts and the high-frequency parasitic oscillation of the excitation source in the initial stage of data processing. The corrected first-order transient rate of change sequence is less likely to deviate with the random changes of the rising edge of the excitation pulse, which can reduce false anomalies caused by the inconsistency of the signal source and improve the consistency and reliability of transient electrical parameter measurement results.

[0015] 2. The second-order transient rate of change sequence and time-domain transient curvature value are calculated based on the corrected first-order transient rate of change sequence. This allows for direct judgment using the transient waveform of the inductor under test itself, reducing reliance on offline calibration curves of external standard samples. When the inductor under test has inter-turn insulation degradation or core defects, nonlinear admittance distortion will disrupt the smooth change of the current waveform and form local abrupt changes in the time-domain transient curvature value sequence. The abrupt changes can be evaluated by using static inflection point time sequence characteristic values ​​and slope variation coefficients, which can improve the stability of early defect identification.

[0016] 3. The synchronously acquired voltage and current data are constructed into a two-dimensional trajectory of voltage and current dynamic impedance. The two-dimensional geometric area and geometric center drift of the trajectory are used for evaluation. The local changes of the time-domain waveform can be transformed into the geometric changes of the entire trajectory. This allows for a more comprehensive and accurate evaluation of the loss characteristics of the inductor under test under high-frequency alternating electrical stress. Furthermore, the geometric center drift can reflect the abnormal hysteresis charge loss caused by batch differences in the core material, reducing the interference of loss fluctuations on the judgment results over a wide temperature range. Attached Figure Description

[0017] Figure 1 This is a flowchart for assessing the inter-turn insulation degradation status of the inductor element according to the present invention; Figure 2 This is a working diagram for the automatic sorting and grading of inductor components according to the present invention.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A method for quality assessment of inductor components based on electrical parameter analysis includes the following steps: Step S1: Apply a step voltage pulse with a determined amplitude to the inductor under test, and control the rise time of the step voltage pulse to not exceed 20ns. Step S2: Within the transient time window after the step voltage pulse is applied, the loop response current of the inductor under test and the actual input voltage across the inductor under test are simultaneously acquired to obtain the discrete response current sequence and the actual input voltage sequence, wherein the sampling point numbers of the discrete response current sequence and the actual input voltage sequence are both integers. Step S3: Calculate the real-time rising edge slope of the actual input voltage sequence based on the ratio of the actual input voltage difference between adjacent sampling points to the sampling time interval, and calculate the first-order transient rate of change sequence of the discrete response current sequence based on the ratio of the loop response current difference between adjacent sampling points to the sampling time interval. Use the real-time rising edge slope to normalize and correct the amplitude of the first-order transient rate of change sequence to obtain the corrected first-order transient rate of change sequence. Step S4: The corrected first-order transient rate of change sequence is calculated by time-domain quadratic difference to construct the second-order transient rate of change sequence. Local abrupt extreme values ​​in the second-order transient rate of change sequence are retrieved to determine the static inflection point time series characteristic value and slope variation coefficient. The inter-turn insulation degradation state of the inductor under test is evaluated based on the static inflection point time series characteristic value and slope variation coefficient.

[0021] Preferably, the operation of constructing a second-order transient rate of change sequence by calculating the corrected first-order transient rate of change sequence through time-domain quadratic difference in step S4 includes the following sub-steps: Step S41, combining the first-order transient rate of change sequence and the second-order transient rate of change sequence to calculate the time-domain transient curvature value sequence corresponding to the sampling point number; Step S42, retrieving local abrupt extreme values ​​in the time-domain transient curvature value sequence to locate the static inflection point time-series characteristic value, and calculating the ratio of the standard deviation to the mean of the time-domain transient curvature value sequence in the neighborhood of the static inflection point time-series characteristic value as the slope variation coefficient, which is used to identify the nonlinear admittance distortion defect generated inside the inductor element under test.

[0022] Preferably, while obtaining the corrected first-order transient rate of change sequence in step S3, the method also includes the following extended sub-steps: Step S31, continuously acquiring the actual input voltage sequence and loop response current sequence of the inductor under test within the transient time window; Step S32, constructing a two-dimensional trajectory of voltage and current dynamic impedance by performing a two-dimensional projective mapping with the actual input voltage sequence as the vertical axis and the loop response current sequence as the horizontal axis.

[0023] Preferably, after constructing the two-dimensional trajectory of voltage and current dynamic impedance in step S32, the method further includes the following extended sub-steps: Step S33, projecting the two-dimensional trajectory of voltage and current dynamic impedance into a closed phase space, and calculating the area of ​​the two-dimensional geometric region enclosed by the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space, and evaluating the loss characteristics of the inductor under test under high-frequency alternating electric stress by using the area of ​​the two-dimensional geometric region.

[0024] Preferably, in step S33, after calculating the area of ​​the two-dimensional geometric region, the method further includes the following extended sub-steps: Step S34, calculate the position of the topological geometric center point of the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space, compare the static deviation value between the position of the topological geometric center point and the origin to obtain the geometric center drift, and nonlinearly quantify the abnormal magnetic hysteresis charge loss caused by batch differences of magnetic core material through the geometric center drift.

[0025] Preferably, the operation of applying a step voltage pulse with a determined amplitude to the inductor under test in step S1 includes the following sub-steps: Step S11, sequentially applying a series of step voltage pulses with equally increasing amplitudes to the inductor under test, and simultaneously acquiring the corresponding loop response current under each pulse input.

[0026] Preferably, under the state of applying a stepped voltage pulse sequence in step S11, the method further includes the following sub-step: Step S12, based on the difference in transient response of the electric charge under different amplitudes, extract the critical abrupt change point when the first-order transient rate of change sequence corresponding to each amplitude voltage pulse generates a sudden nonlinear jump, so as to determine the magnetic saturation threshold drift state of the inductor under test.

[0027] Preferably, the method further includes the following gain adjustment step: Step S5, using the time-domain local features of the corrected first-order transient rate of change sequence to perform feedforward feedback gain adjustment on the discrete response current sequence, so that the physical domain of the focus of electrical variable evaluation is transformed from static macroscopic inductance physical quantity to transient microscopic impedance topology network features.

[0028] Preferably, the following quantitative result output steps are included: Step S6, combining the slope variation coefficient to construct the electrical characteristic parameter matrix, inputting the electrical characteristic parameter matrix into a preset electrical parameter calibration mapping network for interval matching retrieval, and comparing and outputting the quality grading quantitative result of the inductor under test.

[0029] Example 1: The inductor under test is connected to a pulse excitation source via a test fixture for online quality screening under high-frequency, high-power alternating electric stress conditions. During continuous measurement, random fluctuations in the contact resistance of the fixture contacts and high-frequency parasitic electromagnetic coupling in the test circuit will cause random fluctuations in the leading edge of the voltage pulse across the inductor under test. The excitation energy used for measuring static inductance and static quality factor is insufficient to reveal early localized deterioration of inter-turn insulation and damage to the internal structure of the magnetic core. Inductors carrying the above defects may be missed and inter-turn breakdown may occur in subsequent operation. Therefore, the excitation application unit applies a step voltage pulse with a determined amplitude to the inductor under test. Through impedance matching control of the hardware circuit, the rise time of the step voltage pulse is limited to within 20ns. The nonlinear current response at the defect site is excited by transient high-frequency alternating electric stress.

[0030] In this embodiment, the output terminal of the excitation application unit is equipped with a pulse driver composed of a high-speed avalanche transistor and a low parasitic inductance metal-oxide-semiconductor field-effect transistor, and a two-stage impedance matching network is configured. The first stage of the impedance matching network consists of two non-inductive thin-film resistors with a resistance of 50Ω connected in parallel to match the transmission line characteristic impedance of the test cable. The second stage consists of a 10pF trimmer capacitor and a 10Ω carbon film resistor connected in series to form a series damping circuit, which is set at the input terminal of the test fixture to suppress high-frequency overshoot and parasitic oscillations at the pulse leading edge. Through the parameter matching of the pulse driver and the two-stage impedance matching network, the rising edge of the pulse waveform across the inductor under test is stabilized in the range of 15ns to 18ns, which meets the requirement that the rising edge time does not exceed 20ns.

[0031] The high-speed sampling unit starts synchronous sampling from the moment the step voltage pulse is applied and ends sampling when the loop response enters the stable phase, thus forming a transient time window. In this embodiment, the transient time window is set to 0 to 2 μs, and the original electrical signal in the circuit of the inductor under test is synchronously captured at a fixed sampling rate of 500 MSPS to obtain the discrete response current sequence. and actual input voltage sequence Discrete response current sequence This is the discrete form of the loop response current sequence. The integer sampling point number. and These represent the sampling point numbers respectively. The discrete response current value and the actual input voltage value at that time.

[0032] The differential processing unit processes the actual input voltage sequence. Perform time-domain discrete first-order difference to calculate the real-time rising edge slope. : ,in, This represents the real-time rising edge slope of the actual input voltage sequence. The sampling point number is The actual input voltage value at that time. The sampling point number is The actual input voltage value at that time. This represents the sampling time interval.

[0033] The differential processing unit simultaneously processes the discrete response current sequence Perform time-domain discretization with first-order difference to calculate the first-order transient rate of change sequence. : ,in, It is a first-order transient rate of change sequence of discrete response current sequence. The sampling point number is Discrete response current value at time, The sampling point number is Discrete response current value at time, This represents the sampling time interval.

[0034] To reduce contact resistance jitter and amplitude fluctuations caused by high-frequency parasitic oscillations of the excitation source, the differential processing unit utilizes the real-time rising edge slope. As a dynamic gain factor, for first-order transient rate of change sequences By applying amplitude normalization correction point by point, the corrected first-order transient rate of change sequence is obtained. : ,in, This is the corrected first-order transient rate of change sequence; It is a first-order transient rate of change sequence of discrete response current sequence; To determine the real-time rising edge slope of the actual input voltage sequence, the digital signal processor uniformly uses A / s and V / s as the measurement benchmarks for the corresponding rates of change, directly taking the ratio of their values ​​under a fixed measurement benchmark as the normalization result. In subsequent difference and curvature calculations, the dimensionless values ​​are used to process the real-time changes of the actual input voltage sequence to correct the change amplitude of the loop response current, thereby reducing the deviation of the corrected first-order transient rate of change sequence caused by the random deformation of the excitation source waveform.

[0035] While obtaining the corrected first-order transient rate of change sequence, the high-speed sampling unit continues to synchronously acquire the actual input voltage sequence and the loop response current sequence. A two-dimensional projective mapping is performed with the actual input voltage sequence as the vertical axis and the loop response current sequence as the horizontal axis to construct a two-dimensional trajectory of voltage and current dynamic impedance. The step response waveform tends to stabilize at the end of the transient time window. The processor linearly interpolates and connects the initial voltage and current data points at the start of the transient time window with the termination voltage and current data points at the end, closing the open-loop transient trajectory and projecting it into the closed phase space. Because a single unidirectional step voltage pulse is applied, the core undergoes a non-periodic unidirectional excitation transient process. The original voltage and current trajectory formed on the plane is essentially not closed. This scheme uses artificial linear interpolation to forcibly connect its start and end points, which is physically equivalent to... By subtracting the steady-state conduction impedance under a large DC signal, the kinetic energy dissipation caused by high-frequency parasitic electromagnetic coupling and microscopic domain wall movement during the very early transient process of 2 microseconds is isolated. Under ultra-high edge excitation within 20 nanoseconds, the magnetic domain walls inside the core undergo violent and irreversible elastic variations and transitions, which are equivalent to time-varying hysteresis admittance in the circuit. The damping of this microscopic irreversible domain wall movement and transient eddy current dissipation are directly manifested as the hysteresis offset of the trajectory relative to the undamaged state in the complex impedance projection plane. Therefore, the area of ​​the two-dimensional geometric region enclosed after linear interpolation closure is a characteristic measure of the very early microscopic irreversible magnetic excitation and transient parasitic loss. The two have physical conservation consistency in the energy dissipation trend, thus providing a closed-loop explanation of the objective physical basis for the high sensitivity of the transient forced closed phase space area to reflect microscopic core damage.

[0036] The processor uses Green's formula for discretization and performs loop integration along the direction of the discrete sampling points of the closed trajectory. It successively accumulates the product of the sum of the x-coordinates and the difference of the y-coordinates of adjacent sampling points, and then divides the accumulated result by 2 to obtain the area of ​​the two-dimensional geometric region enclosed by the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space. Under the condition that the pulse amplitude, transient time window and sampling rate are kept consistent, the area of ​​the two-dimensional geometric region increases with the increase of the hysteresis range of the voltage and current trajectory. Therefore, the area is used as the relative quantitative value of the loss characteristics under high-frequency alternating electric stress. The larger the area, the more obvious the loss characteristics of the inductor under test.

[0037] The processor also performs area integration on the x and y coordinates of each point within the envelope of the closed trajectory, and divides the integration result by the area of ​​the two-dimensional geometric region to obtain the position of the topological geometric center. The absolute distance between the position of the topological geometric center and the origin is used as the static deviation value, and the geometric center drift is obtained from the static deviation value. The larger the geometric center drift, the more obvious the deviation of the dynamic impedance two-dimensional trajectory relative to the origin. Based on this, the abnormal hysteresis charge loss caused by batch differences in magnetic core material is nonlinearly quantified.

[0038] Feature decoupling unit based on modified first-order transient rate of change sequence Run the multidimensional curvature decoupling program to Perform time-domain discrete difference again to construct a second-order transient rate of change sequence. : ,in, It is a second-order transient rate of change sequence. The sampling point number is The corrected first-order transient rate of change value at time; The sampling point number is The corrected first-order transient rate of change value at time; This represents the sampling time interval.

[0039] Feature decoupling unit point-by-point comparison of second-order transient rate of change sequence The values ​​of the current sampling point and its adjacent sampling points are used to determine the local abrupt extreme values. Sampling points that simultaneously satisfy the conditions of local maximum or local minimum and whose change amplitude exceeds the noise fluctuation range of the steady segment of the transient time window are identified as local abrupt extreme values. The sampling point index corresponding to the local abrupt extreme value is used to determine the time series characteristic value of the static inflection point and is used as the center position of the neighborhood when calculating the slope variation coefficient in the subsequent calculation.

[0040] When calculating the time-domain transient curvature value sequence, the corrected first-order transient rate of change sequence is used. As a normalized numerical representation of a first-order transient rate of change sequence, and a second-order transient rate of change sequence Combined, the transient curvature values ​​in the time domain corresponding to each sampling point are calculated according to the discrete geometric curvature equation. : ,in, The transient curvature value in the time domain. It is a second-order transient rate of change sequence. This is the corrected first-order transient rate of change value.

[0041] The feature decoupling unit searches for a sequence of time-domain transient curvature values ​​near the static inflection point time-series eigenvalues ​​determined by the second-order transient rate of change sequence. For local abrupt extrema, when multiple adjacent sampling points satisfy the local abrupt change condition of the second-order transient rate of change sequence, the static inflection point temporal feature value is located by the sampling point index corresponding to the local abrupt extrema in the time-domain transient curvature value sequence. The sampling point and its adjacent sampling points constitute the neighborhood of the static inflection point temporal feature value. The ratio of the standard deviation to the mean of the time-domain transient curvature value sequence within the neighborhood is calculated, and the ratio is used as the slope variation coefficient. The processor also retains the time-domain transient curvature values ​​within the entire transient time window. The ratio of the maximum value to the average value is used to check the stability of the local mutation amplitude during continuous measurement; the slope variation coefficient is still determined by the ratio of the standard deviation to the mean in the neighborhood of the static inflection point time series characteristic value.

[0042] When there is localized inter-turn insulation degradation or internal core structural defects in the inductor under test, high-frequency pulsed electrical stress drives nonlinear admittance distortion or local charge accumulation at the defect sites, causing local abrupt changes in the continuous evolution of the circuit response current waveform. Local abrupt extreme values ​​are formed in the second-order transient rate of change sequence and the time-domain transient curvature value sequence. The state quantization unit inputs the static inflection point time sequence characteristic value and slope variation coefficient into the multi-level classification judgment matrix, compares it with the preset boundary threshold range of 0.85 to 1.15, identifies whether the electrical parameter characteristics deviate from the preset operating condition boundary, and outputs the inter-turn insulation degradation state and structural quality classification evaluation results of the inductor under test.

[0043] During the feedforward feedback gain adjustment process, the gain adjustment module extracts the local peaks and abrupt changes of the corrected first-order transient rate of change sequence to form a feedforward control signal. Based on this, it dynamically adjusts the amplifier gain coefficient of the data acquisition channel. When a local surge occurs in the corrected first-order transient rate of change sequence, the gain adjustment module proportionally reduces the feedforward gain of subsequent sampling points in the current sampling period to reduce signal saturation distortion. The feedback branch generates a feedback error signal based on the deviation of the corrected first-order transient rate of change sequence from its local time-domain mean. It also uses the synchronously calculated time-domain transient curvature value as the feedback state input current control loop, correcting the amplitude reference of the discrete response current sequence through a closed negative feedback control function. To overcome the digital-to-analog conversion delay and digital computation lag at the extremely high sampling rate of 500 MSPS, the feedforward control channel employs an analog-level control network composed of an ultra-high-speed peak detector circuit and a high-speed voltage comparator. This network detects the instantaneous differential value at the leading edge of the input voltage pulse. When the voltage exceeds 5 volts per microsecond, a high-speed analog switch is triggered within 5 nanoseconds. A 50-ohm carbon film damping resistor is connected in parallel to the feedback resistor of the emitter follower amplifier, thereby directly attenuating the amplification gain of subsequent sampling points to 0.5 times the original value at the analog end, completing the feedforward anti-saturation hardware response. The feedback channel performs time-division iterative control based on the pulse period. The digital signal processor uses the mean value of the transient curvature in the time domain calculated by the previous pulse period as the digital deviation input, which is converted into an analog bias voltage by the high-speed digital-to-analog converter and applied to the non-inverting input of the main operational amplifier to calibrate the zero point of the loop before the sampling begins in the next pulse period. In addition, in order to prevent the hardware gain loop from self-oscillation caused by the rapid switching of the high-speed analog switch, a resistor-capacitor damping filter network is connected to the control logic of the analog switch to lock the transient delay and settling time of gain switching to 5 nanoseconds, and the time interval between the two switching within a single transient window is limited to no less than 100 nanoseconds.

[0044] The discrete response current sequence, after feedforward feedback gain adjustment, is directly used to calculate the corrected first-order transient rate of change sequence, second-order transient rate of change sequence, time-domain transient curvature value sequence, and two-dimensional trajectory of voltage and current dynamic impedance. This transforms the electrical variable evaluation object from static inductance physical quantity to transient impedance network characteristics. Amplitude normalization correction and spatial topology cascade calculation reduce the random jitter of contact resistance and edge fluctuations caused by high-frequency parasitic oscillations of the excitation source. It also converts the distortion of the loop response current waveform caused by minor core damage and inter-turn parasitic capacitance into local abrupt extreme values ​​in the second-order transient rate of change sequence and time-domain transient curvature value sequence. When the static inductance drift of the inductor under test is 0.5%, the extracted slope variation coefficient produces a step transition of more than 15%. The processing uses the actual input voltage sequence and loop response current sequence of the inductor under test as a reference, which can capture early latent insulation defects without introducing external offline calibration samples and heterogeneous reference curves. It can screen the dynamic electrical parameters and internal structural state of the inductor under test under high-frequency alternating electrical stress.

[0045] Example 2: This example establishes an electrical variable measurement platform consisting of a signal generator, a high-frequency data acquisition board, and a differential calibration module. The measurement results are verified under the condition of random contact resistance fluctuation of the test fixture. The signal generator outputs a step voltage pulse with a defined amplitude, and the rise time is limited to within 20ns through output impedance matching control. The resolution of the high-frequency data acquisition board is set to 14 bits, and the fixed sampling rate is 500MSPS. It is used to simultaneously acquire the discrete response current sequence and the actual input voltage sequence of the inductor under test.

[0046] The transient time window is determined based on the transient excitation process of the inductor under test under a step voltage pulse. For high-frequency power components with rated inductance of 10μH to 100μH, the equivalent series DC resistance is 0.1Ω to 1.0Ω. Based on the rated inductance and equivalent series DC resistance, the longest characteristic time constant of the transient process of the complete circuit is about 100μs. Defect detection uses the very early local response segment after the pulse rising edge ends. The nonlinear heterogeneous current response in the test waveform is mainly concentrated in the first 5 decay cycles of the local response segment, with a coverage time of about 0.5μs. Based on the coverage time, a safety margin of 4 times is taken, and the transient time window is set to 2μs to control the amount of data in discrete differential processing while retaining local mutation information.

[0047] To simulate electromagnetic interference in the production line, Gaussian white noise with a signal-to-noise ratio of 20dB and power frequency interference harmonics with a frequency of 50Hz were superimposed on the original circuit signal. The test samples were divided into the present invention sample group, the missing correction control group, the missing curvature control group, and the out-of-range control group. The present invention sample group retained the real-time rising edge slope normalization correction, time-domain quadratic difference, and time-domain transient curvature value calculation. The missing correction control group removed the amplitude normalization correction of the real-time rising edge slope to the first-order transient rate of change sequence. The missing curvature control group removed the time-domain quadratic difference geometric curvature cascade. The out-of-range control group extended the rise time of the step voltage pulse. The rise time of the step voltage pulses of samples 01 to 07 was 20ns, and the rise times of samples 08 and 09 were 25ns and 30ns, respectively.

[0048] The high-frequency data acquisition board uses a fixed sampling time interval of 2ns. Acquiring discrete response current sequences and actual input voltage sequence Within the transient time window, the superimposed Gaussian white noise causes the loop response current waveform to form a discrete envelope with high-frequency glitches. The differential processing unit calculates the real-time rising edge slope based on the actual input voltage difference between adjacent sampling points and the sampling time interval. The first-order transient rate of change sequence is calculated based on the difference in discrete response current between adjacent sampling points and the sampling time interval. ; and then utilize right Amplitude normalization correction is applied point by point to obtain the corrected first-order transient rate of change sequence. .

[0049] Feature decoupling unit pair Construct a second-order transient rate of change sequence by performing a second-order time-domain difference. and search The local extreme values ​​of mutations are used as candidate locations, and then the first-order transient rate of change sequences are combined. With second-order transient rate of change sequence Calculate the time-domain transient curvature value sequence Then search near the candidate locations. The local extreme values ​​of mutations are used to determine the time series characteristic values ​​of static inflection points. A neighborhood is selected centered on the time series characteristic values ​​of static inflection points, and the values ​​within the neighborhood are calculated. The ratio of the standard deviation to the mean yields the coefficient of variation of the slope.

[0050] Test sample 01 belongs to an embodiment of the present invention. Gaussian high-frequency noise is injected. At a time of 0.42μs, the nonlinear admittance distortion defect set inside the sample causes a local abrupt change in the loop response current. and When a local abrupt extreme value appears at the corresponding position, the slope variation coefficient measured by the above processing is 0.94. After matching with the preset boundary threshold range of 0.85 to 1.15, the graded output result is level 1.

[0051] Sample group B of this invention uses sample 02 and injects power frequency harmonic interference. After real-time rising edge slope normalization correction, time domain quadratic difference, and static inflection point time series characteristic value neighborhood calculation, the slope variation coefficient is measured to be 1.03, and the graded output result is level 1. Sample group C of this invention uses sample 03 and injects composite electromagnetic noise. The slope variation coefficient obtained by the same processing flow is measured to be 1.08, and the graded output result is also level 1. The slope variation coefficients obtained by the three sample groups of this invention under different interference conditions all fall within the preset boundary threshold range.

[0052] The missing data correction control group A and the missing data correction control group B used samples 04 and 05, respectively. Both were injected with composite electromagnetic noise. After removing the real-time rising edge slope normalization correction, the high-frequency parasitic oscillations caused by the actual input voltage rising edge fluctuation and the random jitter of the contact resistance directly entered the first-order transient rate of change sequence. The slope coefficient of variation measured by sample 04 was 0.62, and the slope coefficient of variation measured by sample 05 was 1.45. The graded output results of both were level 3.

[0053] The missing curvature control group A and the missing curvature control group B used samples 06 and 07, respectively, and injected composite electromagnetic noise. Since the time-domain quadratic difference geometric curvature cascade was removed, the local mutations in the discrete response current sequence did not continue to be converted into curvature changes in the neighborhood of the static inflection point time series characteristic value. The measured value of the slope variation coefficient obtained by sample 06 was 0.12, and the measured value of the slope variation coefficient obtained by sample 07 was 0.15. The graded output results of both were level 2.

[0054] The out-of-range control group A used sample 08, and under combined electromagnetic noise conditions, the rise time of the step voltage pulse was extended to 25 ns. The measured value of the slope coefficient of variation was 0.41, and the graded output result was level 2. The out-of-range control group B used sample 09, and the rise time was extended to 30 ns. The measured value of the slope coefficient of variation was 0.32, and the graded output result was also level 2. After the rise time was extended, the excitation degree of transient alternating electric stress on the defect site was reduced, and the local abrupt extreme values ​​in the second-order transient rate of change sequence and the time-domain transient curvature value sequence were weakened accordingly. The above data were used to determine the working boundary where the rise time of the step voltage pulse does not exceed 20 ns.

[0055] During continuous automated measurement, the state quantization unit outputs the structural quality grading evaluation results based on the static inflection point timing characteristic value and slope variation coefficient, and converts the grading results into action instructions that the production line sorting mechanism can execute. The sorting mechanism identifies the inductor components under test of different grades based on this, without having to directly analyze the local abrupt change patterns in the discrete response current waveform.

[0056] Example 3: During the sorting process on the inductor component production line, the digital signal processor (DSP) first acquires the electrical response data of a standard inductor component sample known to be free of internal defects. This data is used to correct the channel parameter temperature drift caused by continuous operation of the testing system. The DSP allocates a sliding hold memory area with a storage depth of 512 and receives the discrete response current sequence of the standard inductor component sample via a high-speed serial bus. and actual input voltage sequence Calculate the corrected first-order transient rate of change sequence sequentially. Second-order transient rate of change sequence and time-domain transient curvature value sequence ,Will The arithmetic mean within the transient time window of 0 to 2 μs is stored in memory as the baseline of the standard admittance distribution.

[0057] In the actual online sorting process, the standard admittance distribution baseline is used as a fixed global scalar threshold. After the digital signal processor calculates the time-domain transient curvature value sequence of the inductor under test, it does not perform point-by-point time-domain alignment with the standard sample sequence. Instead, it obtains the value from the current inductor under test within a 2μs transient time window. Five consecutive sampling points are selected sequentially to form a local observation window. The processor calculates the data within each local observation window. The arithmetic mean of the local admittance distribution is calculated and compared with the standard admittance distribution baseline. When the local arithmetic mean exceeds 1.5 times the standard admittance distribution baseline, it is determined that there is dynamic waveform distortion caused by nonlinear admittance distortion defect in the corresponding time period.

[0058] During each measurement, the digital signal processor reads the 14-bit binary code stream output by the external high-frequency data acquisition board through the serial peripheral interface, and cyclically writes the sampled data of each pulse cycle into the first-level data buffer register. Since the transient time window after the step voltage pulse is applied is 2μs, the high-frequency data acquisition board acquires 1000 sampling points in a single measurement. The high-capacity, high-speed, first-in-first-out buffer inside the board temporarily stores the sampled data at an instantaneous throughput of 7Gbit / s. After the transient time window closes, the temporarily stored binary data is asynchronously transmitted in batches to the first-level data buffer register of the digital signal processor through the serial peripheral interface at a communication clock frequency of 50MHz, so that the high-speed acquisition process and the subsequent data transmission process are separated in time.

[0059] The feature decoupling unit sequentially reads the discrete response current value and the actual input voltage value corresponding to each sampling point from the first-level data buffer register, and calculates the real-time rise slope. and first-order transient rate of change sequence reuse right After performing amplitude normalization correction, we obtain Digital signal processor will Sequentially write to the second-level differential storage matrix, and obtain the result through time-domain second-order difference. and search The local extreme values ​​of abrupt changes are used as the corresponding sampling points as candidate positions for the time series feature values ​​of static inflection points.

[0060] Near the candidate location, the feature decoupling unit combines the normalized numerical expression of the first-order transient rate of change sequence with the second-order transient rate of change sequence, and calculates point by point. , then search The local extreme values ​​of mutations are used to determine the static inflection point time series feature value based on the sampling point index corresponding to the local extreme values. Then, a neighborhood is selected centered on the static inflection point time series feature value, and the time series feature value within the neighborhood is calculated. The ratio of the standard deviation to the mean yields the coefficient of variation of the slope.

[0061] When the time axis advances to 0.42μs, the local inter-turn insulation degradation defect inside the inductor under test produces nonlinear admittance distortion under the action of high-frequency alternating electric stress, causing high-frequency abrupt distortion in the circuit response current waveform. Local extreme values ​​of abrupt change are formed at the corresponding sampling points. Local maxima are formed near the location, and the processor uses these to determine the timing characteristic value of the static inflection point and calculate the slope variation coefficient in its neighborhood.

[0062] The digital signal processor combines the static inflection point timing feature value and the slope variation coefficient into an electrical characteristic parameter matrix, and then inputs the matrix into a preset electrical parameter calibration mapping network. The network adopts a three-layer feedforward architecture with 2 input nodes, 5 hidden layer nodes and 3 output classification nodes. The 2 input nodes receive the static inflection point timing feature value and the slope variation coefficient, respectively, and the 3 output classification nodes correspond to different quality classification quantization results.

[0063] The electrical parameter calibration mapping network is trained using 1000 sets of known samples with different degrees of degradation. Multiple sets of weight coefficient matrices are stored in the internal memory. After the input electrical characteristic parameter matrix and the weight coefficient matrix are linearly combined, they are mapped to three output classification nodes through a nonlinear activation function. The network performs interval matching retrieval and outputs the quality grading quantification result corresponding to the inductor under test.

[0064] The preset boundary threshold range of 0.85 to 1.15 and the anomaly judgment threshold of 1.5 times the standard admittance distribution baseline were determined through statistical experiments. During the system calibration phase, continuous pulse tests were performed on 10,000 fully qualified inductor component samples, and the slope variation coefficient of each sample was calculated. The obtained slope variation coefficients followed a Gaussian normal distribution with a mean of 1.0 and a standard deviation of 0.05. The confidence interval for qualified products was determined according to 3 times the standard deviation, and the preset boundary threshold range was set to 0.85 to 1.15.

[0065] When the inductor under test has severe inter-turn insulation degradation, the nonlinear admittance distortion causes the time-series characteristic value of the static inflection point to be in the neighborhood of the inductance. Significant fluctuations occurred, with the slope coefficient of variation deviating from 0.85 to 1.15. Discrete tests on fault samples also showed that when five consecutive sampling points... When all values ​​reach more than 1.5 times the standard admittance distribution baseline, the local charge accumulation at the corresponding defect site has formed a continuous waveform abrupt change. The state quantization unit uses the slope variation coefficient, the static inflection point time series characteristic value, and the comparison results of five consecutive sampling points as input conditions for interval matching retrieval.

[0066] When the slope variation coefficient is within the range of 0.85 to 1.15, the state quantization unit determines that the inter-turn insulation of the inductor under test is normal. The digital signal processor controls the sorting control port to output a high level, driving the external relay of the production line to close and sending the inductor under test into the qualified product collection area. When the slope variation coefficient deviates from 0.85 to 1.15 and exceeds this range for 5 consecutive sampling points... When all values ​​are greater than 1.5 times the standard admittance distribution baseline, the state quantization unit determines that the inductor under test has a defect in the inter-turn insulation degradation. The digital signal processor adjusts the sorting control port to a low level, drives the external pneumatic discharge valve to act, intercepts the defective inductor and sends it to the scrap bin.

[0067] Example 4: This example combines Figures 1 to 2 This paper describes a method for evaluating the quality of inductor components based on electrical parameter analysis, such as... Figure 1As shown, in step S1, a step voltage pulse with a determined amplitude is applied to the inductor under test, and the rise time of the step voltage pulse is controlled to not exceed 20 ns. In step S2, within the transient time window after the step voltage pulse is applied, the loop response current of the inductor under test and the actual input voltage across the inductor under test are simultaneously acquired to obtain the discrete response current sequence and the actual input voltage sequence. In step S3, the real-time rise slope and the first-order transient rate of change sequence are calculated based on the difference between the actual input voltage and the difference between the loop response current, respectively. The amplitude is corrected by normalizing the real-time rise slope to obtain the corrected first-order transient rate of change sequence. In step S4, the corrected first-order transient rate of change sequence is calculated by time-domain quadratic difference to construct the second-order transient rate of change sequence. Local abrupt extreme values ​​are retrieved to determine the static inflection point timing characteristic value and the slope variation coefficient. Based on this, the inter-turn insulation degradation state of the inductor under test is evaluated.

[0068] like Figure 2 As shown, the environmental dissipation deviation value is obtained, and the measurement fixture is controlled to remain open-circuited in the open-circuit state; after the inductor under test is connected, the test circuit is switched to the closed-circuit state; after applying a step voltage pulse, a transient excitation process is entered, and transient high-frequency alternating electric stress is excited, and feedforward feedback gain adjustment is introduced during this process; under the action of a low-amplitude pulse, the system enters an unsaturated state and establishes a linear response baseline; after the voltage pulse amplitude increases, the system enters a magnetic saturation state and the incremental permeability of the magnetic core decreases; after extracting the critical abrupt change point, the system enters a magnetic saturation and threshold drift state that drifts towards low or high amplitude; After the transient excitation process, when the judgment result is within the range of 0.85 to 1.15, the judgment system is in a state where the inter-turn insulation is normal and the slope variation coefficient is not deviated, and the external relay is closed, sending the inductor under test into the qualified product collection area during the external operation of the production line; under the drive of high-frequency pulse electric stress, the system undergoes nonlinear admittance distortion and forms a continuous sudden nonlinear jump segment. When the judgment result deviates from 0.85 to 1.15, the judgment system is in a state of inter-turn insulation deterioration and generates a nonlinear distortion current response, and the pneumatic discharge valve is activated, sending the inductor under test into the scrap bin during the external operation of the production line.

[0069] Example 5: Before deploying the inductor component quality assessment method at a test station with different degrees of parasitic electromagnetic coupling, the digital signal processor first runs the station parasitic impedance alignment program, the measurement fixture is kept open, the excitation application unit applies a step voltage pulse to the input terminal of the fixture, and the high-speed sampling unit synchronously acquires the open-circuit input voltage sequence and the open-circuit response current sequence within a transient time window of 0 to 2 μs.

[0070] The digital signal processor performs time-domain discrete first-order difference on the open-circuit input voltage sequence and the open-circuit response current sequence, respectively. Then, it processes the two sets of open-circuit sampling data according to the normalization correction, time-domain second-order difference, and time-domain transient curvature value calculation process used in online measurement to obtain the time-domain transient curvature value sequence under the open-circuit state. The processor forms the open-circuit environmental dissipation basis quantity based on the average offset of the sequence within the transient time window, which is used to characterize the inherent deviation of the workstation caused by contact mechanical gaps and high-frequency electromagnetic radiation.

[0071] The open-circuit state and the closed-circuit state after the inductor under test is connected have different impedance boundaries. The basic dissipation of the open-circuit environment mainly corresponds to the equivalent parasitic capacitance formed by mechanical gaps and the equivalent parasitic inductance formed by spatial electromagnetic radiation. After the inductor under test is connected in the closed circuit, the total admittance network of the test circuit changes accordingly. Therefore, the impedance network transformation module reads the pre-measured high-frequency scattering parameter matrix of the work position, determines the normalization conversion factor based on the reflection and transmission parameters corresponding to the current test frequency band, and multiplies the basic dissipation of the open-circuit environment by the normalization conversion factor to obtain the environmental dissipation deviation value corresponding to the closed-circuit high-power alternating electric stress condition. The environmental dissipation deviation value is written into the calibration register of the digital signal processor.

[0072] After entering the online screening phase, the digital signal processor extracts the time-domain transient curvature value corresponding to each sampling point. The environmental dissipation deviation value is deducted. The calibrated time-domain transient curvature value is obtained. : ,in, The calibrated time-domain transient curvature value. This represents the transient curvature value in the time domain. To determine the environmental dissipation deviation value based on the open-circuit sampling data and the high-frequency scattering parameter matrix of the workstation, the normalized environmental dissipation deviation value corresponds to the dynamic impedance network baseline under closed-circuit test conditions. Based on this, the processor calibrates the parasitic electromagnetic coupling deviation of the circuit at different workstations.

[0073] After the measuring fixture completes 10,000 sorting cycles, the contact springs undergo mechanical wear under alternating power thermal stress. The circuit contact resistance slowly drifts in one direction with the number of runs. The digital signal processor runs an adaptive baseline sliding reconstruction program without interrupting the production line test. It extracts the slope variation coefficient measurement values ​​of 100 inductor components that are judged to be Grade 1 qualified products from the continuous test results and writes them into the shift register in the order of the test time.

[0074] After the shift register is filled with 100 slope variation coefficient measurements, the digital signal processor calculates the arithmetic mean of all the measurements and uses this mean as the dynamic compensation factor. Subsequently, whenever a new Level 1 qualified product measurement value is obtained, the processor writes it to the shift register, while shifting out the earliest written measurement value, and then recalculates the arithmetic mean of the updated 100 measurement values, so that the dynamic compensation factor changes slowly with the contact state of the fixture.

[0075] The state quantization unit utilizes a dynamic compensation factor Adjust the boundary threshold range, its upper limit Calculate using the following formula: ,in, This is the upper limit of the adjusted boundary threshold range; Both are dimensionless values ​​and are dynamic compensation factors; the lower limit of the boundary threshold interval is adjusted synchronously. The upper and lower limits are updated proportionally using the same dynamic compensation factor, so that the boundary threshold interval moves synchronously with the overall drift of the slope variation coefficient distribution, while maintaining the original interval width ratio and hierarchical judgment relationship.

[0076] The adaptive baseline sliding reconstruction program periodically refreshes the boundary threshold interval in the non-volatile memory according to the update results of the shift register. During the online screening process, the state quantization unit uses the updated boundary threshold interval to perform interval matching retrieval of the slope variation coefficient to compensate for the system bias caused by long-term wear of the fixture contacts.

[0077] Example 6: The current excitation application unit sequentially applies a stepped voltage pulse sequence with arithmetic progression to the inductor under test. Each pulse stage uses the same rise time and transient time window. An interval is reserved between two adjacent pulse stages to allow the loop response to recover to the initial state, so as to avoid the residual magnetization state of the previous stage pulse affecting the measurement of the next stage. Each time a pulse is applied, the high-speed sampling unit synchronously collects the actual input voltage sequence and the corresponding loop response current sequence across the inductor under test.

[0078] The digital signal processor determines the actual amplitude of each voltage pulse according to the actual input voltage sequence, and calculates the first-order transient rate of change sequence corresponding to each voltage pulse based on the difference in loop response current between adjacent sampling points and the sampling time interval. Each sequence is aligned with the starting sampling point of the rising edge of the pulse and the same transient time window is extracted for comparison, so that the difference in transient response of the electrical quantity under different amplitudes comes from the response change of the inductor under test itself.

[0079] Under the action of low amplitude pulses, the inductor under test is in an unsaturated state. The local peak values ​​of the first-order transient rate of change sequence of each stage change smoothly change with the increase of the actual input voltage amplitude. The digital signal processor calculates the local peak increments corresponding to two adjacent voltage pulse stages in turn, and establishes a linear response baseline by using the distribution range of the local peak increments of each low amplitude stage. The distribution range is determined by the continuous low amplitude stages in the stepped voltage pulse sequence that have not yet shown obvious nonlinear changes.

[0080] As the voltage pulse amplitude increases, the incremental permeability of the magnetic core decreases, and the rate of change of the loop response current changes accordingly. When the local peak increment of a certain first-order transient rate of change sequence exceeds the fluctuation range of the linear response baseline for the first time, and forms a continuous sudden nonlinear jump segment in the corresponding sequence, the digital signal processor determines the voltage pulse of the stage as the critical pulse stage and determines the sampling point corresponding to the starting position of the sudden nonlinear jump segment as the critical mutation point.

[0081] The actual input voltage amplitude corresponding to the critical abrupt change point is used as the current magnetic saturation threshold of the inductor under test. During the equipment calibration stage, the same stepped voltage pulse sequence and data processing procedure are used to measure qualified inductors of the same specification. The magnetic saturation reference threshold range is determined based on the critical abrupt change point of each qualified inductor. The range is stored in the non-volatile memory of the digital signal processor for online measurement.

[0082] The state quantization unit compares the current magnetic saturation threshold of the inductor under test with the magnetic saturation reference threshold range. If the current magnetic saturation threshold is within the magnetic saturation reference threshold range, it is determined that the magnetic saturation threshold has not drifted abnormally. If the current magnetic saturation threshold is lower than the lower limit of the range, it is determined that the magnetic saturation threshold has drifted towards a lower amplitude, indicating that the inductor under test has entered the magnetic saturation state earlier. If the current magnetic saturation threshold is higher than the upper limit of the range, it is determined that the magnetic saturation threshold has drifted towards a higher amplitude. In this way, the magnetic saturation threshold drift state of the inductor under test is obtained and used as a supplementary judgment basis for the quality grading quantization result.

[0083] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for evaluating the quality of inductor components based on electrical parameter analysis, characterized in that, Includes the following steps: Step S1: Apply a step voltage pulse with a determined amplitude to the inductor under test, and control the rise time of the step voltage pulse to not exceed 20ns. Step S2: Within the transient time window after the step voltage pulse is applied, the loop response current of the inductor under test and the actual input voltage across the inductor under test are simultaneously acquired to obtain the discrete response current sequence and the actual input voltage sequence, wherein the sampling point numbers of the discrete response current sequence and the actual input voltage sequence are both integers. Step S3: Calculate the real-time rising edge slope of the actual input voltage sequence based on the ratio of the actual input voltage difference between adjacent sampling points to the sampling time interval, and calculate the first-order transient rate of change sequence of the discrete response current sequence based on the ratio of the loop response current difference between adjacent sampling points to the sampling time interval. Use the real-time rising edge slope to normalize and correct the amplitude of the first-order transient rate of change sequence to obtain the corrected first-order transient rate of change sequence. Step S4: The corrected first-order transient rate of change sequence is calculated by time-domain quadratic difference to construct the second-order transient rate of change sequence. Local abrupt extreme values ​​in the second-order transient rate of change sequence are retrieved to determine the static inflection point time series characteristic value and slope variation coefficient. The inter-turn insulation degradation state of the inductor under test is evaluated based on the static inflection point time series characteristic value and slope variation coefficient.

2. The method for quality assessment of inductor components based on electrical parameter analysis according to claim 1, characterized in that, The operation of constructing the second-order transient rate of change sequence by calculating the corrected first-order transient rate of change sequence through time-domain quadratic difference in step S4 includes the following sub-steps: Step S41, combining the first-order transient rate of change sequence and the second-order transient rate of change sequence to calculate the time-domain transient curvature value sequence corresponding to the sampling point number; Step S42: Retrieve local abrupt extreme values ​​in the time-domain transient curvature value sequence to locate the static inflection point time-series characteristic value, and calculate the ratio of the standard deviation to the mean of the time-domain transient curvature value sequence in the neighborhood of the static inflection point time-series characteristic value as the slope variation coefficient, which is used to identify the nonlinear admittance distortion defect generated inside the inductor element under test.

3. The method for quality assessment of inductor components based on electrical parameter analysis according to claim 1, characterized in that, While obtaining the corrected first-order transient rate of change sequence in step S3, the method also includes the following extended sub-steps: Step S31, continuously acquire the actual input voltage sequence and loop response current sequence of the inductor under test within the transient time window; Step S32: Construct a two-dimensional trajectory of voltage and current dynamic impedance by using the actual input voltage sequence as the vertical axis and the loop response current sequence as the horizontal axis for two-dimensional projective mapping.

4. The method for evaluating the quality of inductor components based on electrical parameter analysis according to claim 3, characterized in that, After constructing the two-dimensional trajectory of voltage and current dynamic impedance in step S32, the method further includes the following extended sub-steps: Step S33, project the two-dimensional trajectory of voltage and current dynamic impedance onto the closed phase space, and calculate the area of ​​the two-dimensional geometric region enclosed by the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space, and evaluate the loss characteristics of the inductor under test under high-frequency alternating electric stress by using the area of ​​the two-dimensional geometric region.

5. The method for evaluating the quality of inductor components based on electrical parameter analysis according to claim 4, characterized in that, In step S33, after calculating the area of ​​the two-dimensional geometric region, the method further includes the following extended sub-steps: Step S34, calculate the position of the topological geometric center point of the two-dimensional trajectory of voltage and current dynamic impedance in the closed phase space, compare the static deviation value between the position of the topological geometric center point and the origin to obtain the geometric center drift, and nonlinearly quantify the abnormal magnetic hysteresis charge loss caused by batch differences of magnetic core material through the geometric center drift.

6. The method for quality assessment of inductor components based on electrical parameter analysis according to claim 1, characterized in that, The operation of applying a step voltage pulse with a determined amplitude to the inductor under test in step S1 includes the following sub-steps: Step S11, sequentially applying a series of step voltage pulses with equally increasing amplitudes to the inductor under test, and simultaneously acquiring the corresponding loop response current under each pulse input.

7. The method for evaluating the quality of inductor components based on electrical parameter analysis according to claim 6, characterized in that, In step S11, when a stepped voltage pulse sequence is applied, the method further includes the following sub-step: Step S12, based on the difference in transient response of the electrical charge under different amplitudes, extract the critical abrupt change point when the first-order transient rate of change sequence corresponding to each amplitude voltage pulse generates a sudden nonlinear jump, so as to determine the magnetic saturation threshold drift state of the inductor under test.

8. The method for quality assessment of inductor components based on electrical parameter analysis according to claim 1, characterized in that, The method also includes the following gain adjustment steps: Step S5, using the time-domain local features of the corrected first-order transient rate of change sequence to perform feedforward feedback gain adjustment on the discrete response current sequence, so that the physical domain of the focus of electrical variable evaluation is transformed from static macroscopic inductance physical quantity to transient microscopic impedance topology network features.

9. The method for evaluating the quality of inductor components based on electrical parameter analysis according to claim 2, characterized in that, The following quantitative result output steps are included: Step S6, combine the slope variation coefficient to construct the electrical characteristic parameter matrix, input the electrical characteristic parameter matrix into the preset electrical parameter calibration mapping network for interval matching retrieval, and compare and output the quality grading quantitative result of the inductor under test.

Images