Electromagnetic heating equipment control method based on load sensing

Abstract

The invention discloses an electromagnetic heating equipment control method based on load sensing, and particularly relates to the technical field of electromagnetic heating equipment control, which comprises the following steps: acquiring current and voltage when a special deck thick coating reaches a preset stripping temperature, and dividing point by point in a sliding time window to obtain an impedance sequence; identifying abnormal energy fluctuation of frequency points in the impedance sequence based on a fast Fourier transform tool to form a frequency fluctuation curve; by fusing coating thickness change, electromagnetic load response characteristics and temperature rise rate indexes, a dynamic regulation and control and multi-parameter analysis mechanism is constructed, continuous tracking of the degeneration process of the interface layer and judgment of the stripping completion state are achieved, and therefore it is ensured that the stripping conclusion is consistent with the real state of the interface; the problem that an existing judgment link depends on a macroscopic form and lacks an interface verification mechanism is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electromagnetic heating equipment control, and more particularly to an electromagnetic heating equipment control method based on load sensing. BACKGROUND

[0002] In actual application, the existing electromagnetic heating stripping technology usually relies on the shedding form of the surface coating, the change curve of the load impedance and the distribution of the surface temperature to judge the stripping progress. These criteria can reflect whether the thick coating separates as a whole to a certain extent, but the core limitation is that it cannot directly confirm whether the interface adhesive layer is synchronized with the coating to complete degradation and complete removal.

[0003] In actual operation, when the coating is lifted or slipped as a whole under the action of high temperature, the operator and the system often regard it as a signal of stripping completion, but the adhesive residue of the interface layer often adheres to the surface of the steel shell in irregular patches, still retaining part of the adhesion and adsorption.

[0004] These residual adhesive layers are extremely easy to capture salt mist, water vapor, oil stains or particulate matter in subsequent environmental exposure, thereby forming a new pollution bonding surface between the steel substrate and the external environment, seriously affecting the adhesion and uniformity of the recoating process, reducing the overall corrosion resistance, and even possibly failing to peel off in a short service period, causing rework and quality uncontrollable risks.

[0005] As can be seen, the existing technology's determination link excessively relies on macroscopic visible changes and indirect electromagnetic signals, lacks accurate verification mechanism for the complete failure and removal of the interface adhesive layer, resulting in an uncertain gap between the stripping completion conclusion and the real state of the interface. SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide an electromagnetic heating equipment control method based on load sensing, which fuses coating thickness change, electromagnetic load response characteristics and temperature rise rate indicators to build a dynamic regulation and multi-parameter analysis mechanism, realizes continuous tracking of the interface layer degradation process and judgment of the stripping completion state, thereby ensuring that the stripping conclusion is consistent with the real state of the interface, and solving the problem that the existing determination link relies on macroscopic form and lacks interface verification mechanism.

[0007] To achieve the above purpose, the present application provides the following technical scheme: an electromagnetic heating equipment control method based on load sensing, comprising:

[0008] S1: obtaining the current and voltage of the special deck thick coating when reaching the preset stripping temperature, dividing the current and voltage point by point in the sliding time window to obtain an impedance sequence, identifying the energy abnormal fluctuation of the frequency point in the impedance sequence based on a fast Fourier transform tool to form a frequency fluctuation curve;

[0009] S2: Compare the peaks and valleys in the frequency fluctuation curve with the preset threshold to identify adhesive residue points. Based on the frequency fluctuation range corresponding to each adhesive residue point, extract the center frequency position of the frequency fluctuation range and the corresponding threshold deviation and combine them in a fixed field order to form a feature input vector.

[0010] S3: Input the feature input vector into the secure Bayesian optimization architecture, using the center frequency as the initial sampling point. In each round, the position of the next sampling point is calculated based on the threshold deviation calculated from the previous round's sampling points.

[0011] S4: Calculate the threshold deviation of the estimated sampling points and add the existing results. At the same time, correct the probability distribution of the entire frequency range, narrow the allowable sampling range until the probability distribution is concentrated at a certain frequency point, and take the corresponding frequency point as the target frequency of the corresponding adhesive residue point.

[0012] S5: Using the target frequency as the main frequency, superimposing integer multiple harmonics and applying pulse modulation, a composite electromagnetic signal is generated and applied to the adhesive residue point, causing the molecular chain to vibrate continuously under resonance and be destroyed at the pulse peak.

[0013] S6: While loading the corresponding composite electromagnetic signal at the adhesive residue point, collect the current and voltage of the corresponding area, generate a new frequency fluctuation curve, and compare it with the preset threshold to identify whether the adhesive residue point has completely disintegrated and fallen off.

[0014] In a preferred embodiment, in S1, the instantaneous values ​​of current and voltage are acquired point by point within a preset sliding time window and a division operation is performed to obtain an impedance sequence arranged in the order of sampling time.

[0015] Input the impedance sequence into the Fast Fourier Transform (FFT) tool, multiply the value of each sampling point in the impedance sequence with the value of each frequency basis function point by point, and sum the product results of all sampling points at the same frequency to obtain the complex number result corresponding to the frequency;

[0016] Extract the real and imaginary parts of the complex number result, and then perform squaring operations on the real and imaginary parts respectively and add them together to obtain the energy amplitude corresponding to the frequency.

[0017] The frequency interval is calculated based on the ratio of the sampling rate to the number of sampling points of the impedance sequence, and the frequency intervals are accumulated sequentially to determine the frequency position of each complex result. A one-to-one correspondence is established between the frequency position and the corresponding energy amplitude, and the energy spectrum is output.

[0018] Connecting all frequency positions in the energy spectrum with the energy amplitude in order of frequency position forms a curve, which is the frequency fluctuation curve.

[0019] In a preferred embodiment, in S2, the energy amplitude of each peak and valley point in the frequency fluctuation curve is extracted sequentially and compared with a preset threshold point by point. Points with energy amplitude exceeding the preset threshold are marked as adhesive residue points when the special deck thick coating is peeled off.

[0020] For each adhesive residue point, the energy amplitude of adjacent frequency positions is retrieved sequentially from the adhesive residue point to the left and to the right until the energy amplitude is lower than the preset threshold, thus obtaining the frequency fluctuation range of the corresponding adhesive residue point.

[0021] The adjacent inner frequency positions of the stopping point are respectively used as the lower limit frequency position and the upper limit frequency position of the frequency fluctuation range where the adhesive residue point is located, and the average value of the two is used as the center frequency position of the frequency fluctuation range.

[0022] Calculate the difference between the energy amplitude corresponding to each adhesive residue point and the preset threshold, and use it as the degree of threshold deviation corresponding to the adhesive residue point;

[0023] The center frequency position and the corresponding threshold deviation are arranged in a fixed field order to form a feature input vector.

[0024] In a preferred embodiment, in S3, the feature input vector corresponding to the adhesive residue point is input into the safe Bayesian optimization architecture, which includes a prediction distribution model that predicts the relationship between the sampling point frequency position and the degree of deviation from the threshold, and a safe boundary condition for limiting the range of sampling point frequency position.

[0025] The center frequency position in the feature input vector is used as the initial sampling point frequency position. The corresponding threshold deviation is calculated for the initial sampling point frequency position, and the corresponding threshold deviation is used as the objective function value input into the prediction distribution model.

[0026] In the predictive distribution model, the center frequency position is used as the benchmark, and the corresponding frequency fluctuation range is used as the search range. Several candidate sampling point frequency positions are generated within the frequency fluctuation range at preset intervals.

[0027] Calculate the expected objective function value for each candidate sampling point frequency position, compare the candidate sampling point frequency positions according to the magnitude of the expected objective function value, and select the candidate sampling point frequency position with the upper limit of the expected objective function value as the next sampling point frequency position.

[0028] In a preferred embodiment, in S4, the threshold deviation of the frequency position of the next sampling point is calculated and re-input into the prediction distribution model as a new objective function, triggering the recalculation of the frequency position of the next sampling point, eliminating candidate sampling point frequency positions whose expected objective function value is lower than a preset threshold, narrowing the calculation range of the frequency position of the next sampling point, and completing the correction of the prediction distribution model.

[0029] The correction process is repeated in the predicted distribution model until the calculated frequency position of the next sampling point no longer changes. The frequency corresponding to the sampling point frequency position is then determined as the target frequency of the adhesive residue point.

[0030] In a preferred embodiment, in S5, the target frequency corresponding to the adhesive residue point when the special deck thick coating is peeled off is used as the main frequency electromagnetic signal. Based on the main frequency electromagnetic signal, the target frequency is multiplied by an integer factor in sequence to obtain multiple integer multiple harmonic frequency values.

[0031] For each harmonic frequency value, a periodic waveform signal is generated by calling a standard sine or cosine function, and the start time of the harmonic signal is aligned with the main frequency signal on the time axis.

[0032] Based on the energy amplitude corresponding to the target frequency in the frequency fluctuation curve, and combined with the preset amplitude conversion factor, the energy amplitude is converted into the main frequency signal amplitude that can be output by the electromagnetic heating device.

[0033] The amplitude of the main frequency signal is used as the reference amplitude, and the reference amplitude is scaled according to a preset ratio to obtain the amplitude of each harmonic signal. The periodic waveforms of the main frequency signal and each harmonic signal are multiplied by the corresponding amplitude, and then added point by point in the time domain to generate a composite frequency component containing the main frequency and each harmonic.

[0034] In a preferred embodiment, S5 further includes applying pulse modulation to the composite frequency component so that the composite frequency component forms a composite electromagnetic signal with an instantaneous peak value in the time domain.

[0035] The composite electromagnetic signal is applied to the adhesive residue point, causing the molecular chains of the residual adhesive to generate a gradually increasing vibration amplitude under the resonance of the target frequency and harmonic frequency.

[0036] The instantaneous peak power based on pulse modulation acts on the vibration of the residual adhesive molecular chain, causing the vibrational energy of the residual adhesive molecular chain to exceed the stability threshold of the molecular chain chemical bond.

[0037] In a preferred embodiment, in S6, while applying a composite electromagnetic signal to the adhesive residue point, the current and voltage of the corresponding area are collected to generate a new frequency fluctuation curve.

[0038] The energy amplitude corresponding to each frequency position in the new frequency fluctuation curve is compared with the preset threshold point by point. If all energy amplitudes are lower than the preset threshold, the adhesive residue point is completely disintegrated and detached. If there is an energy amplitude greater than or equal to the preset threshold, the adhesive residue point is still present and has not detached.

[0039] The technical effects and advantages of this invention are as follows:

[0040] 1. This solution integrates multiple physical parameters such as load voltage change, coating thickness change and load temperature rise trend to construct a continuous judgment path for the degradation state of the interface adhesive layer. It breaks through the traditional indirect judgment that relies on macroscopic peeling and surface temperature, and realizes the confirmation of whether the adhesive residue has been completely removed, directly solving the problem of the peeling completion signal and the actual state of the interface.

[0041] 2. Dynamically construct an electromagnetic heating load response feature library under different working conditions, perform attribution analysis and trend extrapolation on the physical signal features at different times during the stripping process, and improve the intelligent identification capability of stripping integrity and interface cleanliness.

[0042] 3. During implementation, the coating thickness, power intensity and heating time are systematically linked and controlled to ensure that the high-power pulse output only acts on the stage when the interface layer enters the critical transition zone, avoiding excessive energy application that may cause interface micro-damage or local carbonization, and ensuring the safety and adaptability of the heating process to the substrate and interface structure.

[0043] 4. For stages where the load temperature rise response fluctuates, a mapping relationship is established between the relative temperature rise rate and the preset adhesive layer thickness gradient to accurately identify the thermal plateau segment in the peeling process. This allows for power injection adjustment within the critical softening window of the adhesive layer, enabling directional peeling of the adhesive residue area. Attached Figure Description

[0044] Figure 1 This is a flowchart outlining the method steps of the present invention;

[0045] Figure 2 This is a flowchart illustrating the initial extraction of impedance sequence and frequency fluctuation in this invention.

[0046] Figure 3 This is a flowchart of the residual point identification and feature vector construction process of the present invention;

[0047] Figure 4 This is a flowchart of the Bayesian sampling and target frequency convergence process of the present invention;

[0048] Figure 5 This is a flowchart illustrating the composite signal construction and loading process of the present invention.

[0049] Figure 6 This is a flowchart of the peeling determination process of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Refer to the instruction manual appendix Figures 1-6 An embodiment of the present invention provides a control method for an electromagnetic heating device based on load sensing, comprising:

[0052] S1: Obtain the current and voltage when the special deck thick coating reaches the preset peeling temperature, divide the impedance sequence point by point within the sliding time window, identify the abnormal energy fluctuations at the frequency points in the impedance sequence based on the fast Fourier transform tool, and form a frequency fluctuation curve with frequency coordinates and amplitude correspondence. The frequency fluctuation curve is used to sense the peeling process of the special deck thick coating and find potential adhesive residue points in the electromagnetic heating equipment control in real time.

[0053] S2: Compare the peaks and valleys in the frequency fluctuation curve with the preset threshold to identify adhesive residue points. Based on the frequency fluctuation range corresponding to each adhesive residue point, extract the center frequency position of the frequency fluctuation range and the corresponding threshold deviation and combine them in a fixed field order to form a feature input vector. The feature input vector is used to guide the control logic of the electromagnetic heating equipment to ensure that the response characteristics of adhesive residue points can be accurately locked during the peeling process of thick coating.

[0054] S3: Input the feature input vector into the secure Bayesian optimization architecture, with the center frequency as the initial sampling point. In each round, based on the threshold deviation calculated from the previous round's sampling point, calculate and recommend the position of the next sampling point, so that the electromagnetic heating equipment can dynamically adjust the adhesive residue points after the special deck thick coating is peeled off.

[0055] S4: Calculate the threshold deviation of the estimated sampling point and add the existing results. At the same time, correct the probability distribution of the entire frequency range, shrink the allowable sampling range until the probability distribution is stably concentrated at a certain frequency point, and take the corresponding frequency point as the target frequency of the corresponding adhesive residue point. The target frequency is used as the control output signal of the electromagnetic heating equipment so as to accurately heat the adhesive residue point after the thick coating is peeled off.

[0056] S5: Using the target frequency as the main frequency, superimposing integer harmonics and applying pulse modulation generates a composite electromagnetic signal covering the resonance range of the adhesive residue point. The corresponding composite electromagnetic signal is applied to the adhesive residue point, causing the molecular chains within the residue point to continuously amplify vibrations under resonance. Under the action of the pulse peak, the molecular chains break through their stability threshold, allowing the electromagnetic heating equipment to completely remove the adhesive residue point during the peeling of thick coatings on special decks. This causes the polymer structure of the residual adhesive to gradually disintegrate. Here, the molecular chain refers to the polymer structural unit of the residual adhesive, whose chemical bonds have inherent vibration frequencies; the pulse peak refers to the instantaneous peak power generated by pulse modulation, which can release energy in a short time.

[0057] S6: While loading the corresponding composite electromagnetic signal at the adhesive residue point, the current and voltage of the corresponding area are collected to generate a new frequency fluctuation curve, which is compared with a preset threshold to identify whether the adhesive residue point has been completely disintegrated and detached. This is used to confirm whether the adhesive residue point has been completely removed during the peeling process of thick coating on special decks by electromagnetic heating equipment control.

[0058] In S1, the instantaneous values ​​of current and voltage are acquired point by point within the preset sliding time window and a division operation is performed to obtain an impedance sequence arranged in the order of sampling time.

[0059] Input the impedance sequence into the Fast Fourier Transform (FFT) tool, multiply the value of each sampling point in the impedance sequence with the value of each frequency basis function point by point, and sum the product results of all sampling points at the same frequency to obtain the complex result corresponding to that frequency. The frequency basis functions refer to a set of predefined standard sine and cosine waves, which are used as reference waveforms to multiply the impedance sequence point by point to decompose the strength of different frequency components.

[0060] Extract the real and imaginary parts of the complex number result, and perform squaring and summing the real and imaginary parts respectively to obtain the energy amplitude corresponding to the frequency. The real and imaginary parts refer to the two components in the complex number result calculated by the Fast Fourier Transform. The real part reflects the similarity between the impedance sequence and the cosine wave, and the imaginary part reflects the similarity between the impedance sequence and the sine wave.

[0061] The frequency interval is calculated according to the ratio of the sampling rate to the number of sampling points of the impedance sequence, and the frequency intervals are accumulated sequentially to determine the frequency position of each complex number result. A one-to-one correspondence is established between the frequency position and the corresponding energy amplitude, and the energy spectrum of the frequency position and energy amplitude is output. The sampling rate refers to the number of times the instantaneous values ​​of current and voltage are acquired per unit time within a preset sliding time window.

[0062] Connecting all frequency positions in the energy spectrum with the energy amplitude in order of frequency position forms a curve, which is the frequency fluctuation curve.

[0063] In S2, the energy amplitude of each peak and valley point in the frequency fluctuation curve is extracted sequentially and compared with the preset threshold point by point. Points with energy amplitude exceeding the preset threshold are marked as adhesive residue points when the special deck thick coating is peeled off.

[0064] For each adhesive residue point, the energy amplitude of adjacent frequency positions is sequentially retrieved from the adhesive residue point to the left and to the right until the energy amplitude is lower than the preset threshold, thus obtaining the frequency fluctuation range of the corresponding adhesive residue point.

[0065] The adjacent inner frequency positions of the stopping point are respectively used as the lower limit frequency position and the upper limit frequency position of the frequency fluctuation range where the adhesive residue point is located, and the average value of the two is used as the center frequency position of the frequency fluctuation range.

[0066] Calculate the difference between the energy amplitude corresponding to each adhesive residue point and the preset threshold, and use it as the degree of threshold deviation corresponding to that adhesive residue point;

[0067] The center frequency position and the corresponding threshold deviation are arranged in a fixed field order to form a feature input vector.

[0068] It should be noted that in the formula structure involved in this scheme, dimensionless terms can be used as proportional or structural adjustment factors. When combined with quantities with units, they only play a role in numerical scaling and do not introduce new physical dimensions. Therefore, they will not change or confuse the overall unit system. This combination of "dimensionless terms and terms with units" can be understood as a composite structural expression commonly used in mathematical physics modeling. It conforms to the principle of dimensional consistency and has a clear physical interpretation basis.

[0069] Secondly, in the formula structure of this scheme, if multiple variables with different physical units are involved, including but not limited to time, mass or energy variables, their joint appearance is to express the collaborative modeling relationship of multiple physical mechanisms. Each variable can form a unified structure through function mapping, ratio combination or normalization adjustment, with clear units and clear meaning. The overall expression conforms to the principle of dimensional consistency and the conventional formula of engineering modeling.

[0070] In this solution, constants, weights, adjustment factors, threshold parameters, proportional coefficients, etc., are all adjustable control parameters for different application environments. Their values ​​depend on the target equipment configuration, data input characteristics, and performance optimization goals. During the implementation phase, they are set to converge within a reasonable range through model verification, performance constraints, or engineering calibration. Although these parameters do not have a unique preset value, they have clear adjustment logic and calculation paths. They belong to the deterministic setting process in engineering implementation. The purpose of this setting is to ensure that the solution is both universally adaptable and reproducible and operable, without affecting its technical clarity and feasibility.

[0071] In S3, the feature input vector corresponding to the adhesive residue point is input into the safe Bayesian optimization architecture. The safe Bayesian optimization architecture includes a prediction distribution model that predicts the relationship between the sampling point frequency position and the degree of deviation from the threshold, and a safety boundary condition for limiting the sampling point frequency position range. The safety boundary condition refers to the value limit set based on the frequency fluctuation range corresponding to the adhesive residue point and combined with the output frequency range allowed by the device, in order to avoid the sampling point frequency position from exceeding the physical safety range.

[0072] The center frequency position in the feature input vector is used as the initial sampling point frequency position. The corresponding threshold deviation of the initial sampling point frequency position is calculated, and the corresponding threshold deviation is used as the objective function value input into the prediction distribution model.

[0073] In the predictive distribution model, the center frequency position is used as the benchmark, and the corresponding frequency fluctuation range is used as the search range. Several candidate sampling point frequency positions are generated within the frequency fluctuation range at preset intervals.

[0074] The expected objective function value is calculated for each candidate sampling point frequency position. The candidate sampling point frequency positions are compared according to the magnitude of the expected objective function value. The candidate sampling point frequency position with the upper limit of the expected objective function value is selected as the next sampling point frequency position. The expected objective function value refers to the predicted value obtained by extrapolating the objective function value of the initial sampling point frequency position and the relative difference between the candidate sampling point frequency position and the center frequency position based on the prediction distribution model. It is used to characterize the possible deviation intensity of the candidate sampling point frequency position.

[0075] Define the expected objective function value Ψ(f) i ):

[0076]

[0077] Where f0 represents the initial sampling point frequency position; f i Indicates the frequency position of the candidate sampling point; φ(f0) represents the response index; Δ(f i ,f0) represents the normalized frequency difference; Γ(f i ) represents the historical perturbation integral; Λ(·) represents the weighted mapping function; This represents the partial derivative of the sampling path parameter θ. Used to estimate the guiding trend of the current candidate sampling point on the change of the sampling path; dθ is the integration variable, which represents the integration process with respect to the variable θ;

[0078] Furthermore, in Ψ(f i In the formula, the frequency position f of the candidate sampling point is used. i The response index φ(f0) and the normalized frequency difference Δ(f0) of the initial sampling point frequency position f0 are...i ,f0) and historical perturbation integral Γ(f i The partial derivatives of the common input sampling path parameter θ Obtain the frequency position f of the candidate sampling point i The expected objective function value Ψ(f) i );

[0079]

[0080] Where E(f0) represents the frequency energy amplitude at the initial sampling point frequency position f0; ω(f0) represents the local frequency fluctuation density in the frequency fluctuation range where the initial sampling point frequency position f0 is located;

[0081] Furthermore, in the formula φ(f0), the response index φ(f0) of the initial sampling point frequency position f0 is obtained by performing square operations on the frequency energy amplitude E(f0) corresponding to the initial sampling point frequency position f0 and the local frequency fluctuation density ω(f0) respectively, adding them together and taking the square root.

[0082]

[0083] Where κ(f) i ) represents the frequency position of the candidate sampling point f i The boundary sensitivity of the frequency fluctuation range; log(·) represents the logarithmic function; This represents the relative difference in frequency positions of candidate sampling points;

[0084] Furthermore, in Δ(f i In the formula, f0), the frequency position of the candidate sampling point is f i The relative difference between the initial sampling point frequency position f0 and the boundary sensitivity κ(f) i After summing the results and taking the natural logarithm, we obtain the normalized frequency difference Δ(f). i ,f0);

[0085]

[0086] Where δ τ (f i ) represents the frequency position of the candidate sampling point f i The deviation residual between the predicted energy amplitude and the actual observed energy amplitude in the τth iteration; η(τ) represents the time decay function of the τth iteration, and η(τ) is used to reduce the impact of the deviation residual of earlier iterations on the current prediction result; dτ represents the set of all historical sampling rounds executed before the current round; dτ represents the infinitesimal unit of integration performed in iteration rounds for variable τ;

[0087] Furthermore, in Γ(f iIn the formula, the frequency position f of the candidate sampling point is used. i In each historical cycle τ∈ Deviation residual δ τ (f i The square of f is multiplied by its corresponding decay time function η(τ), and then integrated and summed to obtain the historical disturbance integral Γ(f). i );

[0088]

[0089] Where a = φ(f0) and b = Δ(f i ,f0), c=Γ(f i ); exp(·) represents the exponential function;

[0090] Furthermore, in equation Λ(a,b,c), the response index a=φ(f0) and the normalized frequency difference b=Δ(f i The product of f(0) and the historical perturbation integral c = Γ(f i The nonlinear suppression terms formed by the exponential function exp(·) are added together to construct the relative ratio between the numerator and denominator in the fractional structure, thus obtaining the weighted mapping function Λ(·).

[0091] In S4, the threshold deviation of the frequency position of the next sampling point is calculated and re-inputted into the prediction distribution model as a new objective function, triggering the recalculation of the frequency position of the next sampling point. Candidate sampling point frequency positions with expected objective function values ​​lower than the preset threshold are eliminated, narrowing the calculation range of the frequency position of the next sampling point and completing the correction of the prediction distribution model.

[0092] The correction process is repeated in the predicted distribution model until the frequency position of the next sampling point no longer changes. The frequency corresponding to the frequency position of this sampling point is then determined as the target frequency of the adhesive residue point.

[0093] In S5, the target frequency corresponding to the adhesive residue point when the special deck thick coating is peeled off is used as the main frequency electromagnetic signal. Based on the main frequency electromagnetic signal, the target frequency is multiplied by an integer factor in turn to obtain multiple integer multiple harmonic frequency values.

[0094] For each harmonic frequency value, a periodic waveform signal is generated by calling a standard sine or cosine function, and the start time of the harmonic signal is aligned with the main frequency signal on the time axis.

[0095] Based on the energy amplitude corresponding to the target frequency in the frequency fluctuation curve, and combined with the preset amplitude conversion factor, the energy amplitude is converted into the main frequency signal amplitude that the electromagnetic heating device can output. The amplitude conversion factor is a coefficient that maps the energy amplitude obtained from the frequency fluctuation curve analysis to the output amplitude range of the electromagnetic heating device according to a preset ratio.

[0096] The amplitude of the main frequency signal is used as the reference amplitude, and the reference amplitude is scaled according to a preset ratio to obtain the amplitude of each harmonic signal. The periodic waveforms of the main frequency signal and each harmonic signal are multiplied by their corresponding amplitudes and added point by point in the time domain to generate a composite frequency component containing the main frequency and each harmonic.

[0097] S5 also includes applying pulse modulation to the composite frequency components so that the composite frequency components form a composite electromagnetic signal with an instantaneous peak value in the time domain;

[0098] The composite electromagnetic signal is applied to the adhesive residue point, causing the molecular chains of the residual adhesive to generate a gradually increasing vibration amplitude under the resonance of the target frequency and its harmonic frequencies.

[0099] The instantaneous peak power based on pulse modulation acts on the vibration of the molecular chain of the residual adhesive, causing the vibrational energy of the molecular chain to exceed the stability threshold of the chemical bond of the molecular chain, thereby causing the polymer structure of the residual adhesive to disintegrate.

[0100] In S6, while applying a composite electromagnetic signal to the adhesive residue point, the current and voltage of the corresponding area are collected to generate a new frequency fluctuation curve.

[0101] The energy amplitude corresponding to each frequency position in the new frequency fluctuation curve is compared with the preset threshold point by point. If all energy amplitudes are lower than the preset threshold, the adhesive residue is completely disintegrated and detached. This is used to control and confirm that the peeling process of the special deck thick coating has been completed. If there is an energy amplitude greater than or equal to the preset threshold, the adhesive residue is still present and has not detached. The process is then returned to the target frequency calculation and composite electromagnetic signal generation process for reloading.

[0102] It should be noted that, including but not limited to: this solution was developed based on an analysis of the problem that traditional electromagnetic heating peeling processes cannot accurately identify the residual state of the adhesive layer. Existing technologies mainly rely on the coating peeling morphology, load impedance changes, or surface temperature distribution as criteria for peeling completion. However, the above information only reflects surface changes and cannot provide a direct perception of the true state of the adhesive layer at the interface. Especially when there are invisible patches of residue at the interface between thick coatings and steel shells, the peeling determination is highly uncertain.

[0103] To address the aforementioned issues, a load-aware frequency control path is constructed, integrating frequency fluctuation sensing, feature modeling, frequency prediction, energy output, and feedback verification to achieve accurate identification and complete removal of adhesive residue.

[0104] The implementation path consists of six steps:

[0105] The first step is to collect current and voltage in real time through a sliding time window, construct an impedance sequence, and use a frequency domain conversion tool to convert the impedance sequence into a frequency fluctuation curve, which is used to capture abnormal energy fluctuations in real time during the stripping process.

[0106] The second step is to extract the energy amplitude of all peaks and valleys in the frequency fluctuation curve and compare it with a preset threshold to identify the frequency positions where the energy is higher than the threshold and determine them as potential adhesive residue points. For each residue point, the frequency positions are retrieved sequentially from its left and right sides to determine the boundary of the fluctuation range. The deviation between the center frequency position and the amplitude is calculated to form a combination of frequency position and response characteristics, which constitutes the feature input vector.

[0107] The third step is to input the feature input vector into the prediction model with a safety boundary, generate several candidate frequency points based on the center frequency position, perform extrapolation according to the initial deviation, calculate the expected objective function value of the candidate points, and select the frequency point with the greatest deviation potential as the sampling position for the next round.

[0108] The fourth step is to calculate the degree of deviation for the new sampling points, correct the prediction model, eliminate inefficient candidate frequency points in the frequency range, and continue iterative calculation until the target frequency is stably converged.

[0109] The fifth step involves using the target frequency as the main frequency to generate a composite electromagnetic signal, superimposing integer multiple harmonics, and applying pulse modulation. Under the combined action of the target frequency and its harmonics, the molecular chains of the residual polymer structure are excited to vibrate, gradually increasing their amplitude. Under the action of the pulse peak, the stable structure of the chemical bonds is broken, thus completing the resonant deconstruction of the residual polymer.

[0110] The sixth step involves real-time acquisition of regional current and voltage during signal loading to generate a new frequency fluctuation curve. It is then determined whether all frequency responses have dropped below the threshold. If no response is confirmed, the loading process is terminated. If residual responses still exist, the process returns to the frequency calculation and composite signal generation steps to execute a new loading cycle until the residue is completely cleared.

[0111] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A control method for electromagnetic heating equipment based on load sensing, characterized in that, include: S1: Obtain the current and voltage when the special deck thick coating reaches the preset peeling temperature, divide them point by point within the sliding time window to obtain the impedance sequence, and identify the abnormal energy fluctuations at the frequency points in the impedance sequence based on the fast Fourier transform tool to form a frequency fluctuation curve. S2: Compare the peaks and valleys in the frequency fluctuation curve with the preset threshold to identify adhesive residue points. Based on the frequency fluctuation range corresponding to each adhesive residue point, extract the center frequency position of the frequency fluctuation range and the corresponding threshold deviation and combine them in a fixed field order to form a feature input vector. S3: Input the feature input vector into the secure Bayesian optimization architecture, using the center frequency as the initial sampling point. In each round, the position of the next sampling point is calculated based on the threshold deviation calculated from the previous round's sampling points. S4: Calculate the threshold deviation of the estimated sampling points and add the existing results. At the same time, correct the probability distribution of the entire frequency range, narrow the allowable sampling range until the probability distribution is concentrated at a certain frequency point, and take the corresponding frequency point as the target frequency of the corresponding adhesive residue point. S5: Using the target frequency as the main frequency, superimposing integer multiple harmonics and applying pulse modulation, a composite electromagnetic signal is generated and applied to the adhesive residue point, causing the molecular chain to vibrate continuously under resonance and be destroyed at the pulse peak. S6: While loading the corresponding composite electromagnetic signal at the adhesive residue point, collect the current and voltage of the corresponding area, generate a new frequency fluctuation curve, and compare it with the preset threshold to identify whether the adhesive residue point has completely disintegrated and fallen off.

2. The control method for an electromagnetic heating device based on load sensing according to claim 1, characterized in that: In S1, the instantaneous values ​​of current and voltage are acquired point by point within the preset sliding time window and a division operation is performed to obtain an impedance sequence arranged in the order of sampling time. Input the impedance sequence into the Fast Fourier Transform (FFT) tool, multiply the value of each sampling point in the impedance sequence with the value of each frequency basis function point by point, and sum the product results of all sampling points at the same frequency to obtain the complex number result corresponding to the frequency; Extract the real and imaginary parts of the complex number result, and then perform squaring operations on the real and imaginary parts respectively and add them together to obtain the energy amplitude corresponding to the frequency. The frequency interval is calculated based on the ratio of the sampling rate to the number of sampling points of the impedance sequence, and the frequency intervals are accumulated sequentially to determine the frequency position of each complex result. A one-to-one correspondence is established between the frequency position and the corresponding energy amplitude, and the energy spectrum is output. Connecting all frequency positions in the energy spectrum with the energy amplitude in order of frequency position forms a curve, which is the frequency fluctuation curve.

3. The control method for an electromagnetic heating device based on load sensing according to claim 2, characterized in that: In S2, the energy amplitude of each peak and valley point in the frequency fluctuation curve is extracted sequentially and compared with the preset threshold point by point. Points with energy amplitude exceeding the preset threshold are marked as adhesive residue points when the special deck thick coating is peeled off. For each adhesive residue point, the energy amplitude of adjacent frequency positions is retrieved sequentially from the adhesive residue point to the left and to the right until the energy amplitude is lower than the preset threshold, thus obtaining the frequency fluctuation range of the corresponding adhesive residue point. The adjacent inner frequency positions of the stopping point are respectively used as the lower limit frequency position and the upper limit frequency position of the frequency fluctuation range where the adhesive residue point is located, and the average value of the two is used as the center frequency position of the frequency fluctuation range. Calculate the difference between the energy amplitude corresponding to each adhesive residue point and the preset threshold, and use it as the degree of threshold deviation corresponding to the adhesive residue point; Arrange the center frequency position and the corresponding threshold deviation in a fixed field order. Form a feature input vector.

4. The control method for an electromagnetic heating device based on load sensing according to claim 3, characterized in that: In S3, the feature input vector corresponding to the adhesive residue point is input into the safe Bayesian optimization architecture. The safe Bayesian optimization architecture includes a prediction distribution model that predicts the relationship between the sampling point frequency position and the degree of deviation from the threshold, and a safe boundary condition for limiting the range of sampling point frequency position. The center frequency position in the feature input vector is used as the initial sampling point frequency position. The corresponding threshold deviation is calculated for the initial sampling point frequency position, and the corresponding threshold deviation is used as the objective function value input into the prediction distribution model. In the predictive distribution model, the center frequency position is used as the benchmark, and the corresponding frequency fluctuation range is used as the search range. Several candidate sampling point frequency positions are generated within the frequency fluctuation range at preset intervals. Calculate the expected objective function value for each candidate sampling point frequency position, compare the candidate sampling point frequency positions according to the magnitude of the expected objective function value, and select the candidate sampling point frequency position with the upper limit of the expected objective function value as the next sampling point frequency position.

5. The control method for an electromagnetic heating device based on load sensing according to claim 4, characterized in that: In S4, the threshold deviation of the frequency position of the next sampling point is calculated and re-inputted into the prediction distribution model as a new objective function, triggering the recalculation of the frequency position of the next sampling point. Candidate sampling point frequency positions with expected objective function values ​​lower than the preset threshold are eliminated, narrowing the calculation range of the frequency position of the next sampling point and completing the correction of the prediction distribution model. The correction process is repeated in the predicted distribution model until the calculated frequency position of the next sampling point no longer changes. The frequency corresponding to the sampling point frequency position is then determined as the target frequency of the adhesive residue point.

6. The control method for an electromagnetic heating device based on load sensing according to claim 5, characterized in that: In S5, the target frequency corresponding to the adhesive residue point when the special deck thick coating is peeled off is used as the main frequency electromagnetic signal. Based on the main frequency electromagnetic signal, the target frequency is multiplied by an integer factor in turn to obtain multiple integer multiple harmonic frequency values. For each harmonic frequency value, a periodic waveform signal is generated by calling a standard sine or cosine function, and the start time of the harmonic signal is aligned with the main frequency signal on the time axis. Based on the energy amplitude corresponding to the target frequency in the frequency fluctuation curve, and combined with the preset amplitude conversion factor, the energy amplitude is converted into the main frequency signal amplitude that can be output by the electromagnetic heating device. The amplitude of the main frequency signal is used as the reference amplitude, and the reference amplitude is scaled according to a preset ratio to obtain the amplitude of each harmonic signal. The periodic waveforms of the main frequency signal and each harmonic signal are multiplied by the corresponding amplitude, and then added point by point in the time domain to generate a composite frequency component containing the main frequency and each harmonic.

7. The control method for an electromagnetic heating device based on load sensing according to claim 6, characterized in that: S5 also includes applying pulse modulation to the composite frequency components so that the composite frequency components form a composite electromagnetic signal with an instantaneous peak value in the time domain; The composite electromagnetic signal is applied to the adhesive residue point, causing the molecular chains of the residual adhesive to generate a gradually increasing vibration amplitude under the resonance of the target frequency and harmonic frequency. The instantaneous peak power based on pulse modulation acts on the vibration of the residual adhesive molecular chain, causing the vibrational energy of the residual adhesive molecular chain to exceed the stability threshold of the molecular chain chemical bond.

8. The control method for an electromagnetic heating device based on load sensing according to claim 7, characterized in that: In S6, while applying a composite electromagnetic signal to the adhesive residue point, the current and voltage of the corresponding area are collected to generate a new frequency fluctuation curve. The energy amplitude corresponding to each frequency position in the new frequency fluctuation curve is compared with the preset threshold point by point. If all energy amplitudes are lower than the preset threshold, the adhesive residue point is completely disintegrated and detached. If there is an energy amplitude greater than or equal to the preset threshold, the adhesive residue point is still present and has not detached.

Images