A copper material surface on-line coating thickness detection system

Abstract

The application relates to the technical field of industrial automation detection, and particularly discloses a copper material surface online coating thickness detection system. The system comprises a physical thickness measuring device, an electrochemical impedance spectroscopy micro-area detection device, a data fusion processing unit, a coating defect prediction model and an early warning output unit; by synchronously acquiring the coating geometric thickness and the multi-frequency electrochemical impedance response, a multi-dimensional feature data set is constructed, and a convolutional neural network is used for pattern recognition on the impedance spectrum to determine whether defects such as insufficient adhesion, microscopic pores or insufficient solidification exist; when the thickness is qualified but there is an early failure risk, a graded early warning is triggered, and a closed-loop regulation and control is realized by linking the production line. The application can find hidden quality hazards in advance, and improve the reliability and consistency of copper material coating products.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation testing technology, and specifically relates to an online coating thickness detection system for copper materials. Background Technology

[0002] With the continuous improvement of industrial manufacturing automation, copper surface coating technology is increasingly widely used in fields such as power conduction and precision electronic packaging. As a key process to ensure the oxidation resistance and physical isolation effect of metal substrates, the stability of online coating processes determines the durability and environmental adaptability of the final products. To meet the requirements of high-precision fields for the surface performance of metal materials, establishing a comprehensive quality monitoring and performance evaluation system for the entire coating process has become the core technology for the current surface treatment industry to pursue high-quality development.

[0003] Online detection of coating thickness and film integrity is a core step in ensuring that the coating process meets pre-design specifications. By integrating automated detection modules into a continuous production line, the coating status of the copper surface can be scanned in real time and multi-dimensional data can be acquired, achieving real-time closed-loop control of production quality. This technology focuses on using high-sensitivity sensors to capture coating coverage, with the basic goal of reducing material scrap rates due to process fluctuations and ensuring consistent quality of finished products in large-scale production through rapid feedback of key physical parameters.

[0004] Existing coating inspection methods largely focus on single physical thickness analysis such as eddy current induction or magnetic measurement, resulting in inspection dimensions remaining at the geometric level and failing to delve into the intrinsic functional state of the coating. Traditional methods lack the means to perceive the microscopic density of the coating, interfacial bonding strength, and the degree of curing of polymer chains. This makes it difficult to identify potential adhesion defects or micropore corrosion risks when process parameters such as curing temperature deviate even if the thickness meets the standard. Furthermore, the lack of current technology to simultaneously correlate the electrochemical characteristics of the coating with its geometric parameters prevents the system from providing intelligent early warnings for early failures under complex operating conditions, creating a technological gap between inspection results and actual protective performance.

[0005] There is an urgent need for an online coating thickness detection system for copper materials. Summary of the Invention

[0006] The purpose of this invention is to provide an online coating thickness detection system for copper materials, which can solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An online coating thickness detection system for copper surfaces includes a physical thickness measuring device, an electrochemical impedance spectroscopy micro-area detection device, a data fusion processing unit, a coating defect prediction model, and an early warning output unit, as follows: The physical thickness measuring device is configured to perform non-contact thickness measurement on the coating of a continuously moving copper material surface to obtain the geometric thickness parameters of the coating. The electrochemical impedance spectroscopy micro-area detection device is configured to apply a weak alternating electrical signal to the coating on the copper surface while the physical thickness measurement device completes the thickness acquisition, and simultaneously acquire the electrochemical impedance response data of the coating at multiple frequencies to form an impedance spectrum. The data fusion processing unit is configured to perform time alignment and spatial matching between the thickness data output by the physical thickness measurement device and the impedance spectrum output by the electrochemical impedance spectroscopy micro-area detection device, thereby constructing a multidimensional feature dataset containing geometric and electrochemical dimensions. The coating defect prediction model is configured to receive the multidimensional feature dataset and perform pattern recognition on the spatial distribution features of the impedance spectrum based on a pre-trained convolutional neural network structure to determine whether the coating has inherent defects such as insufficient adhesion, micropores, or incomplete curing. The early warning output unit is configured to generate a quality anomaly early warning signal and output it to the production line control system when the coating thickness is within a preset range but there is an early failure risk, based on the judgment result of the coating defect prediction model.

[0008] Preferably, the physical thickness measuring device adopts either the eddy current induction principle or the magnetic measurement principle, and its measurement accuracy meets the real-time monitoring requirements of industrial online inspection for the thickness fluctuation of the copper surface coating.

[0009] Furthermore, the electrochemical impedance spectroscopy micro-area detection device includes a miniature three-electrode probe array, which is non-destructively attached to the coating on the copper surface to complete the acquisition of electrochemical excitation and response in a local area without affecting the continuous operation of the production line.

[0010] Furthermore, the amplitude of the alternating electrical signal applied by the electrochemical impedance spectroscopy micro-area detection device is controlled within a range that does not cause electrochemical reactions in the coating, ensuring that the detection process does not damage the coating itself and does not affect subsequent processes.

[0011] Preferably, the data fusion processing unit has a built-in spatiotemporal synchronization module, which is used to perform precise spatiotemporal coordinate mapping of thickness data and impedance spectrum based on the copper material's travel speed and the sensor's installation position, ensuring that the two correspond to the same detection area on the copper material's surface.

[0012] Furthermore, the coating defect prediction model is a convolutional neural network model trained based on historical production data. Its input layer receives a normalized impedance spectrum matrix, the intermediate layer extracts local correlation features in the spectrum through multi-scale convolution kernels, and the output layer generates coating state classification results. The classification results include at least three categories: "normal and dense", "weak adhesion" and "presence of micropores".

[0013] Furthermore, the coating defect prediction model is trained with adversarial sample enhancement before deployment to improve its robustness under noise interference or signal drift conditions, ensuring that it can still maintain a high accuracy rate in defect identification in complex industrial environments.

[0014] Preferably, the early warning output unit is equipped with a graded alarm mechanism. When the coating thickness deviates from a preset threshold, a first-level alarm is triggered. When the thickness is within the acceptable range but the coating defect prediction model determines that there is an inherent defect, a second-level early warning is triggered. The second-level early warning signal includes a defect type identifier and risk level information.

[0015] Furthermore, the early warning output unit is communicatively connected to the automatic sorting device or process parameter adjustment module of the production line, enabling the system to automatically trigger closed-loop control actions such as isolating defective products or dynamically adjusting process parameters such as coating curing temperature when a potential failure risk is detected.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The online coating thickness detection system for copper materials provided by this invention breaks through the limitations of traditional thickness measurement technology that only focuses on geometric dimensions. By simultaneously integrating physical thickness measurement and electrochemical impedance spectroscopy micro-area detection, it achieves dual verification of coating thickness and internal functional status.

[0017] 2. The system can not only confirm whether the coating has reached the designed thickness, but also sense its micro-density, interfacial bonding strength and curing integrity through electrochemical signals. When the thickness index is normal but the process parameters (such as insufficient curing temperature) deviate, it can identify early failure risks such as poor adhesion or micro-pores in advance.

[0018] 3. By leveraging the deep pattern recognition capabilities of convolutional neural networks for impedance spectrum, the system can automatically distinguish different types of inherent defects, thereby enhancing the scientific rigor and foresight of coating quality assessment.

[0019] 4. The system supports real-time linkage with the production line, forming a complete closed loop from detection and diagnosis to control, reducing the probability of product failure due to latent defects and improving the reliability and consistency of high-end copper products in service environment. Attached Figure Description

[0020] Figure 1This is a schematic diagram of the overall technical solution architecture of the present invention; Figure 2 This is a schematic diagram of the core principle framework of the coating defect prediction model in this invention, which is based on extracting impedance spectrum features using a convolutional neural network. Figure 3 This is a flowchart illustrating the logical process framework of the data fusion processing unit in this invention for spatiotemporal alignment and multidimensional feature construction. Figure 4 This is a schematic diagram of the multi-level interaction relationship and data flow of the physical thickness measurement and electrochemical impedance acquisition process in this invention; Figure 5 This is a schematic diagram illustrating the technical principle of the early warning output unit triggering hierarchical alarms and closed-loop control in this invention. Detailed Implementation

[0021] Example 1: Please refer to the appendix Figure 1 To be continued Figure 5 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0022] An online coating thickness detection system for copper surfaces includes a physical thickness measuring device, an electrochemical impedance spectroscopy micro-area detection device, a data fusion processing unit, a coating defect prediction model, and an early warning output unit. The physical thickness measurement device is configured to perform non-contact thickness measurement on the surface coating of continuously moving copper materials, obtaining the geometric thickness parameters of the coating. The physical thickness measurement device is installed above the conveyor rollers after the coating and curing process in the copper production line. The physical thickness measurement device integrates a high-frequency excitation circuit, an electromagnetic induction probe, and a signal demodulation circuit. The high-frequency excitation circuit is configured to generate an alternating current with a constant frequency and controlled amplitude, and injects this current into the excitation coil of the electromagnetic induction probe, exciting a high-frequency alternating magnetic field in the space around the probe. When the copper material coated with an insulating coating passes under the electromagnetic induction probe, due to the good conductivity of the copper substrate, it generates an eddy current effect under the action of the alternating magnetic field. The magnitude of the eddy current effect is inversely proportional to the physical distance between the probe and the copper surface. The detection coil inside the electromagnetic induction probe captures the reverse magnetic field signal generated by the eddy current and converts it into an induced voltage signal. The signal demodulation circuit receives this induced voltage signal, and through synchronous rectification and low-pass filtering, extracts the voltage component reflecting the distance from the probe to the copper substrate surface. Since the coating itself is non-magnetic and insulating, it occupies part of the space between the probe and the substrate. Therefore, the physical thickness measurement device calculates the geometric thickness value of the coating by subtracting a preset system zero-point offset from the detected spatial distance using a preset calibration curve. The physical thickness measurement device is also equipped with an air curtain protection assembly to blow away floating dust or residual solvents on the copper surface, ensuring the purity of the optical or magnetic path during non-contact measurement.

[0023] The electrochemical impedance spectroscopy micro-area detection device is configured to apply a weak alternating electrical signal to the copper surface coating while the physical thickness measurement device completes thickness acquisition, and simultaneously acquires electrochemical impedance response data of the coating at multiple frequencies to form an impedance spectrum. The electrochemical impedance spectroscopy micro-area detection device includes a miniature three-electrode probe array, an electrolyte isothermal circulation unit, a broadband potentiostat, and a digital signal processing module. The miniature three-electrode probe array adopts a flexible support structure to ensure that it can adhere to the continuously operating copper coating surface with minimal constant pressure. Each detection unit in the array includes a platinum disk electrode as an auxiliary electrode, a saturated calomel electrode or a silver / silver chloride electrode as a reference electrode, and the copper substrate itself as a working electrode. To form an electrochemical circuit, the electrolyte isothermal circulation unit continuously provides a highly conductive and wettable neutral electrolyte to the tiny area where the probe contacts the coating. The electrolyte enters the contact interface through a microfluidic channel and is prevented from overflowing by a vacuum suction system. The broadband potentiostat is configured, under the control of the digital signal processing module, to generate a set of weak sinusoidal voltage excitation signals with a frequency range from 100 kHz to 0.01 Hz, the amplitude of which is limited to between 10 mV and 20 mV to ensure that the electrical signal is insufficient to cause electrochemical decomposition of the coating material or severe corrosion of the substrate. During the application of the excitation signal, the digital signal processing module simultaneously records the current response signal through the loop and calculates the amplitude ratio and phase difference angle between the voltage vector and the current vector. By aggregating the impedance magnitude and phase angle at different frequency points, impedance spectrum data in the form of a Nyquist plot or Bode plot, containing both real and imaginary impedance characteristics, is constructed.

[0024] The data fusion processing unit is configured to perform time alignment and spatial matching between the thickness data output by the physical thickness measurement device and the impedance spectrum output by the electrochemical impedance spectroscopy micro-area detection device, constructing a multidimensional feature dataset containing geometric and electrochemical dimensions. The data fusion processing unit integrates a high-speed cache, a clock synchronization logic controller, and a spatial coordinate transformation module. Since the physical thickness measurement device and the electrochemical impedance spectroscopy micro-area detection device have a preset distance in the physical layout of the production line, and the copper material's travel speed may fluctuate dynamically, the spatial coordinate transformation module acquires the encoder feedback pulses of the production line drive motor in real time. The clock synchronization logic controller adds a precise timestamp to each set of collected thickness and impedance spectrum data. The spatial coordinate transformation module calculates the time deviation value of different sensors reaching the same position by integrating the travel speed over time, and uses this deviation value to retrieve corresponding historical data from the high-speed cache for matching. Each set of matched data includes the geometric thickness at the same coordinate point on the copper material surface, the impedance magnitude at different frequencies, and the corresponding phase shift, forming a multidimensional vector. The data fusion processing unit also performs data cleaning operations to remove outliers caused by vibration of the copper surface, and uses a linear interpolation algorithm to compensate for missing data points caused by inconsistent sampling frequencies.

[0025] The coating defect prediction model is configured to receive the multidimensional feature dataset and perform pattern recognition on the spatial distribution features of the impedance spectrum based on a pre-trained convolutional neural network structure to determine whether the coating has inherent defects such as insufficient adhesion, micropores, or incomplete curing. The coating defect prediction model is deployed on a computing node equipped with a high-performance graphics processor. Its core algorithm structure includes an input layer, multiple alternating convolutional and pooling layers, a fully connected layer, and a classification output layer. The input layer is configured to normalize the received impedance spectrum matrix, mapping the real impedance, imaginary impedance, and frequency index to a numerical range between 0 and 1. The convolutional layers use multi-scale convolutional kernels to perform sliding weighted summation operations along the impedance spectrum dimension to extract local correlation features between different frequency bands, such as the coating pore resistance features reflected in the high-frequency region and the interface double-layer capacitance features reflected in the mid-to-low-frequency region. The pooling layer employs a maximum downsampling strategy to reduce the dimensionality of the feature map and enhance the model's robustness to small phase drifts in the impedance signal. The fully connected layer concatenates the extracted deep electrochemical features with the geometric thickness features output by the data fusion processing unit to form a global feature vector. The classification output layer uses a normalized exponential function to calculate the probability values ​​of the current coating belonging to multiple preset categories such as "normally dense," "weak adhesion," "microscopic porosity," or "insufficient curing," and uses the category with the highest probability value as the final recognition result. In the offline stage, the model is trained by collecting a large number of samples with known artificial defects, and the weight parameters of the convolutional kernel are continuously optimized using the backpropagation algorithm.

[0026] The early warning output unit is configured to generate a quality anomaly early warning signal and output it to the production line control system when the coating thickness is within a preset range but there is a risk of early failure, based on the judgment result of the coating defect prediction model. The early warning output unit is connected to the production line's programmable logic controller (PLC) via an industrial Ethernet interface and has two levels of alarm logic internally. The first-level alarm logic targets geometric thickness anomalies; when the thickness value output by the physical thickness measuring device exceeds preset upper and lower thresholds, a shutdown and maintenance signal is triggered. The second-level early warning logic targets inherent latent defects; when the thickness value is within the acceptable range, but the coating defect prediction model outputs a "weak adhesion" or "micropore" level exceeding a preset risk value, the early warning output unit generates an early warning message containing a defect type code, coordinate location information, and risk level score. This message is sent to the production line control system to guide the automatic sorting mechanism to mark or isolate the copper section during the winding process. The warning output unit also sends adjustment commands to the temperature control unit of the coating machine through the process closed-loop interface. If insufficient curing is detected, the working temperature of the drying zone is automatically increased to achieve real-time feedback control of detection and manufacturing.

[0027] The physical thickness measurement device also includes an environmental compensation module, which is configured to monitor the ambient air temperature and the real-time temperature of the copper surface using a built-in temperature sensor. Since the eddy current induction intensity is affected by the conductivity of the substrate, and conductivity exhibits a linear correlation with temperature, the environmental compensation module stores a conductivity temperature compensation coefficient specific to the copper material. Based on the monitored temperature value, the environmental compensation module corrects the original induced voltage collected by the physical thickness measurement device, ensuring the stability of thickness measurement under high-temperature or variable-temperature conditions.

[0028] The micro-electrode probe array of the electrochemical impedance spectroscopy micro-area detection device employs a multi-channel independent parallel acquisition architecture. Each channel is equipped with an independent transimpedance amplifier to convert the weak current response signal at the nanoampere level into a millivolt-level voltage signal suitable for analog-to-digital converter processing. The transimpedance amplifier uses a high input impedance operational amplifier design, and its input bias current is controlled at the picoampere level to reduce disturbances to the electrochemical equilibrium state of the micro-area. The device also includes an electrolyte level monitoring circuit, which determines whether a liquid junction filled with electrolyte has been formed between the probe and the coating surface by measuring the DC conduction state between the counter electrode and the reference electrode. If the liquid junction is broken, the system automatically triggers a liquid replenishment operation and suspends the current impedance scanning task.

[0029] The data fusion processing unit is also configured to execute fusion logic based on feature weight allocation. Internally, the unit includes a dynamic weight calculator that adjusts the input weights of geometric and electrochemical features in the defect prediction model in real time based on the thickness stability index output by the physical thickness measurement device and the signal-to-noise ratio index of the electrochemical impedance spectroscopy. When a large surface roughness of the coating is detected, causing drastic changes in physical thickness measurement, the calculator automatically increases the weight ratio of the impedance spectral features, utilizing the insensitivity of electrochemical signals to surface morphology to maintain the continuity of defect identification.

[0030] The coating defect prediction model incorporates an attention mechanism module within its convolutional layers. This module is configured to reweight the extracted feature maps in terms of both spatial and channel dimensions, enabling the model to automatically focus on the low-frequency impedance regions that contribute most to adhesion evaluation. The model also integrates a drift adaptive subnetwork, which uses real-time statistical analysis of historical impedance data distribution characteristics to fine-tune the model's forward propagation parameters online, eliminating systematic biases caused by slight changes in electrolyte concentration or probe aging.

[0031] The early warning output unit is also connected to a local visualization terminal. The visualization terminal is configured to render and display in real time the trend curve of coating thickness, the equivalent circuit fitting parameters of impedance spectrum, and a heat map of defect distribution. The heat map uses different colors to represent different defect probabilities, and provides process engineers with a spatiotemporal distribution view of coating quality by corresponding to the production line length coordinates.

[0032] Example 2: This example describes an online coating thickness detection system for copper surfaces using a distributed edge computing architecture. This architecture is particularly suitable for copper processing workshops with very long production lines, aiming to improve system response speed by distributing computational load and localizing processing.

[0033] An online coating thickness detection system for copper materials includes multiple edge physical thickness measuring devices distributed at various nodes of the production line, multiple edge electrochemical impedance acquisition units, a regional data aggregation hub, a distributed defect analysis array, and a collaborative early warning center. The edge physical thickness measurement devices are deployed in a distributed manner, installed at the discharge port of the coating machine, the middle of the drying zone, and the end of the cooling zone, forming a multi-point monitoring network. Each edge physical thickness measurement device integrates a microcontroller responsible for local signal acquisition and preprocessing. These devices utilize the principle of magnetic thickness measurement to perform high-frequency sampling on thick protective coatings. Their internal magnetic sensors detect magnetic flux fluctuations caused by changes in the magnetic reluctance of the copper substrate and convert them into digital thickness signals. Each device is connected to the regional data aggregation hub via a fieldbus, uploading a thickness data stream containing node numbers and timestamps.

[0034] The edge electrochemical impedance spectroscopy (EIS) acquisition units are geographically paired with the edge physical thickness measurement devices. Each acquisition unit is equipped with a closed electrolyte circulation trolley, which can move synchronously along a guide rail for short distances, completing a full frequency scan without stopping the production line. Each EIS acquisition unit integrates a Fast Fourier Transform (FFT) processor to locally convert the acquired time-series current and voltage signals into frequency domain impedance data in real time. This local conversion reduces the amount of data uploaded to the regional data aggregation hub, avoiding communication congestion caused by large-scale raw sampling points.

[0035] The regional data aggregation hub is configured as an industrial-grade gateway, responsible for coordinating data synchronization across all edge devices within its jurisdiction. Internally, this hub operates a virtual synchronization clock, which maintains synchronization with the production line's master clock via a network time protocol, with errors controlled to the millisecond level. The regional data aggregation hub receives thickness and impedance data from different nodes and, based on the physical span of each node, uses a pipeline scheduling algorithm to aggregate data from different process stages of the same copper section, generating a full lifecycle coating feature package.

[0036] The distributed defect analysis array consists of multiple parallel neural network acceleration units. Unlike the centralized model in Example 1, the distributed defect analysis array in this example employs a task-parallel processing strategy. Each acceleration unit is configured to process data of a specific type of feature or a specific region. For example, some units specialize in analyzing high-frequency impedance data to detect surface pinholes, while others combine physical thickness data to specifically analyze the risk of interface peeling. The units exchange intermediate feature vectors via an internal high-speed bus, and finally, a consensus decision layer aggregates the judgment results from each unit, outputting the final defect type and location.

[0037] The collaborative early warning center is configured to integrate diagnostic reports from multiple distributed analysis arrays. Internally, this center establishes a logical judgment matrix, configured to comprehensively compare local defects at a single node with the cumulative process trends across multiple nodes. If multiple nodes in a certain area detect thickness fluctuations accompanied by a decrease in impedance modulus, the collaborative early warning center determines this as a systemic process failure and immediately cuts off the production line's power supply via a high-priority control message. If the defect is merely an isolated, accidental defect, a quality tracking file is generated, and this information is pushed to the downstream automatic labeling machine to print the defect location code on the copper material edge.

[0038] The edge physical thickness measurement device also integrates an automatic calibration module. This module is configured to automatically move the thickness probe to the position of the built-in standard thickness gauge during production line downtime, perform measurements, and calculate the deviation between the measured value and the standard value. The microcontroller automatically updates its internal proportional coefficient and zero-point offset based on this deviation value, achieving self-maintenance of thickness measurement accuracy.

[0039] The electrolyte circulation trolley of the edge electrochemical impedance acquisition unit employs magnetic adsorption positioning technology. When the trolley receives an acquisition command, its bottom permanent magnet adheres to the supporting steel plate below the copper material, keeping the trolley relatively stationary. After acquisition, the electromagnet is energized to generate a reverse magnetic field to counteract the adsorption force, and the trolley, assisted by a spring, quickly returns to its starting position, ready for the next acquisition. This reciprocating following design solves the measurement noise problem caused by friction between the probe and the coating during continuous movement.

[0040] The regional data aggregation hub also features data integrity verification. It is configured to verify each data packet using a cyclic redundancy check algorithm. If data corruption or loss is detected during transmission, the hub automatically sends a retransmission request to the corresponding edge device. The hub internally stores a circular buffer capable of storing up to 4 hours of production data in the event of a temporary network connection interruption, ensuring the continuity of detection records.

[0041] The distributed defect analysis array employs a lightweight convolutional neural network obtained through model distillation. This network, while maintaining recognition accuracy, compresses the number of weight parameters to 1 / 5 of the original model, enabling it to run on resource-constrained edge acceleration units. The acceleration unit also supports dynamic reconstruction, allowing it to download and load optimized network parameter sets for specific copper grades from the cloud, based on the specifications of the copper currently being produced.

[0042] The collaborative early warning center also integrates an augmented reality interface. This interface is configured to project detected defect data in real time onto the large-screen monitoring system in the workshop or the mobile terminals of inspection personnel. By establishing a digital twin model of the production line, the collaborative early warning center can display the current thickness uniformity distribution of the copper coating and potential defect risk points in the form of 3D animation, improving the intuitiveness of manual intervention.

[0043] Example 3: This example describes an online coating thickness detection system for copper surfaces with environmental adaptive adjustment and multi-sensor redundancy verification functions. This system is particularly enhanced for reliability under extreme industrial environments (such as high temperature, high humidity, and strong electromagnetic interference).

[0044] An online coating thickness detection system for copper materials includes a sealed thickness sensor module, a corrosion-resistant impedance monitoring array, an environmental interference elimination unit, a multi-dimensional cross-validation calculation center, and an intelligent process scheduling gateway. The sealed thickness sensor module has a shell made of special stainless steel with a Teflon coating, and its interior is filled with insulating and thermally conductive adhesive. This structural configuration is designed to prevent acid mist and volatile organic solvents in the workshop from corroding the internal precision coils. The module employs a dual-coil differential structure; one coil measures the coating thickness, while the other is positioned in a reference environment. The signal processing circuitry within the module is configured to automatically compensate for measurement errors caused by fluctuations in the external magnetic field by calculating the difference between the output signals of the two coils.

[0045] The corrosion-resistant impedance monitoring array employs a multi-electrode integrated chip on a ceramic substrate. A platinum working electrode and a gold auxiliary electrode are fabricated on the chip surface using photolithography. To meet the extremely high cleanliness requirements of high-end copper surfaces, the array utilizes non-contact inductively coupled electrochemical technology. This technology is configured to induce weak eddy currents within the coating via high-frequency electromagnetic coupling, and the electrochemical characteristics of the coating are deduced from the impedance changes at the coil terminals, thus avoiding contamination of the copper surface by electrolyte liquids.

[0046] The environmental interference cancellation unit is configured to collect electromagnetic spectrum data and vibration spectrum data in the workshop in real time. This unit integrates an adaptive digital filter, configured to use the environmental interference frequency as a reference input to filter out electromagnetic interference components of the same frequency from the physical thickness measurement signal and impedance signal. This unit utilizes an accelerometer to monitor the mechanical vibration of the production line and uses an active compensation algorithm to eliminate noise caused by probe gap changes due to copper material vibration, thereby improving the signal-to-noise ratio of ultra-thin coating inspection.

[0047] The multidimensional cross-validation calculation center is configured to perform logical correlation analysis on physical thickness and electrochemical impedance data. Internally, this center runs a fusion algorithm based on evidence theory. When the physical thickness measurement shows a thinner thickness and the high-frequency phase angle of the electrochemical impedance increases, the center determines with high confidence that the coating is incomplete or insufficient in thickness. If the physical thickness is normal, but the low-frequency modulus of the impedance spectrum decreases and the capacitive arc radius shrinks, the center determines that there are microscopic conductive pathways within the coating, i.e., penetrating pore defects have occurred. This cross-validation mechanism reduces the false alarm rate of a single sensor.

[0048] The intelligent process scheduling gateway is configured to convert detected quality information into specific production control instructions. This gateway integrates a knowledge base based on an expert system. When the system identifies a coating defect, the gateway automatically queries the knowledge base based on the defect type to obtain possible causes of failure (such as insufficient spraying pressure, short curing time, or excessively high ambient humidity). The gateway sends adjustment messages to the coating process system via the production bus, for example, automatically increasing the voltage value of electrostatic spraying or extending the conveyor belt dwell time in the curing oven, thus achieving closed-loop management from quality monitoring to process optimization.

[0049] The sealed thickness sensor module is also equipped with a built-in self-heating component. In winter or humid environments, the self-heating component is configured to maintain the internal temperature of the sensor at a constant 40 degrees Celsius to prevent condensation on the internal optical or electronic components. A humidity monitoring point is also installed inside the sensor; when the humidity inside the sealed housing exceeds a safe threshold, a maintenance request is automatically sent to the central system.

[0050] The inductive coupling frequency of the corrosion-resistant impedance monitoring array is configured between 1 MHz and 10 MHz. Within this frequency range, electromagnetic waves can penetrate a polymer coating of a certain thickness and generate strong physical interaction with the underlying copper surface. The impedance analysis module within the array calculates the equivalent capacitance value of the coating interface by extracting the higher-order harmonic components of the induction coil. Since the equivalent capacitance value is closely related to the dielectric constant and thickness of the coating, this characteristic is used as a redundant backup of the physical thickness measurement data, further improving the reliability of the detection.

[0051] The environmental interference cancellation unit also employs fiber optic transmission technology. Analog signals from all sensors are converted into digital optical signals for transmission at the node locations. The fiber optic transmission path is completely immune to spatial electromagnetic interference generated by high-power frequency converters, motors, and other strong electrical equipment. A high-speed photoelectric conversion interface is configured at the multi-dimensional cross-validation calculation center to ensure the real-time performance and synchronization accuracy of large-scale concurrent data acquisition.

[0052] The multidimensional cross-validation calculation center also established a coating aging prediction model. This model uses the low-frequency limiting impedance value in electrochemical impedance spectroscopy as an input parameter, combined with historical accelerated corrosion test data, to calculate the expected service life of the current coating under standard service conditions. This function enables the system not only to perform real-time quality inspection but also to provide a scientific basis for subsequent product reliability assessment.

[0053] The intelligent process scheduling gateway also has an external communication channel, supporting integration with the enterprise's production execution system and quality management platform. The gateway is configured to automatically generate a quality analysis report for each batch of copper material, detailing key indicators such as average thickness, defect rate distribution, and process stability score. This data is signed using an encryption algorithm to ensure the immutability of the archived records, providing strong support for product quality traceability.

[0054] In summary, the online coating thickness detection system for copper surfaces provided by this invention solves the technical problem of traditional single-dimensional detection's inability to detect inherent hidden defects in coatings by integrating physical and electrochemical sensing methods and using deep learning-based pattern recognition technology. The system demonstrates its application flexibility and technological depth in different industrial scenarios through the integrated architecture in Example 1, the distributed edge computing architecture in Example 2, and the environmentally adaptive redundancy architecture in Example 3. The close collaboration and real-time interaction between modules construct a fully intelligent monitoring system from low-level signal acquisition to high-level process decision-making. This not only improves the delivery quality of copper coating products but also reduces resource waste and scrap rates during production through timely process feedback adjustments. Regarding high-precision analysis of electrochemical signals, the system achieves the capture and pattern mapping of nanoampere-level current responses through complex logic and algorithm configurations described in text, providing a solid technical guarantee for the long-life operation of high-end copper materials in fields such as power and electronics. This multi-dimensional detection capability of the system...

[0055] Those skilled in the art should understand that the above-described embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the claims of the present invention.

Claims

1. An online coating thickness detection system for copper materials, characterized in that, include: A physical thickness measurement device is configured to perform non-contact thickness measurement on the surface coating of a continuously moving copper material to obtain the geometric thickness parameters of the coating. An electrochemical impedance spectroscopy micro-area detection device is configured to apply an alternating electrical signal to the coating on the surface of the copper material while the physical thickness measurement device is collecting thickness data, and simultaneously collect the electrochemical impedance response data of the coating at multiple frequencies to construct an impedance spectrum. A data fusion processing unit is configured to perform time-alignment and spatial matching of the geometric thickness parameters and the impedance spectrum to construct a multidimensional feature dataset containing geometric and electrochemical dimensions. A coating defect prediction model is configured to receive the multidimensional feature dataset and perform pattern recognition on the distribution characteristics of the impedance spectrum based on a pre-trained convolutional neural network structure to determine whether the coating has inherent defects. The early warning output unit is configured to generate a quality anomaly early warning signal and output it to the production line control system when the coating thickness meets the preset range and the judgment result indicates an early failure risk, based on the judgment result of the coating defect prediction model.

2. The online coating thickness detection system for copper materials according to claim 1, characterized in that, The physical thickness measurement device includes a high-frequency excitation circuit, an electromagnetic induction probe, and a signal demodulation circuit. The high-frequency excitation circuit is connected to the excitation coil of the electromagnetic induction probe to generate an alternating current with a constant frequency and controlled amplitude, and to generate a high-frequency alternating magnetic field around the electromagnetic induction probe. The electromagnetic induction probe is equipped with a detection coil, which is configured to capture the reverse magnetic field signal generated by the eddy current effect of the copper substrate and convert the reverse magnetic field signal into an induced voltage signal. The signal demodulation circuit is connected to the detection coil and is used to perform synchronous rectification and low-pass filtering on the induced voltage signal to extract the voltage component that reflects the distance from the electromagnetic induction probe to the surface of the copper substrate. The physical thickness measuring device has a built-in calculation module. The calculation module calculates the geometric thickness parameters of the coating by subtracting the preset system zero offset from the measured spatial distance using a preset calibration curve.

3. The online coating thickness detection system for copper materials according to claim 2, characterized in that, The physical thickness measuring device also includes an environmental compensation module and an air curtain protection component; The air curtain protection assembly is disposed on the outer periphery of the detection end face of the electromagnetic induction probe, and blows away the floating dust or residual solvent on the surface of the copper material by spraying compressed air. The environmental compensation module includes a temperature sensor and a compensation calculator. The temperature sensor is configured to monitor the air temperature of the detection environment and the real-time temperature of the copper surface. The compensation arithmetic unit stores a conductivity temperature compensation coefficient for copper materials, which is used to correct the original induced voltage output by the signal demodulation circuit based on the monitored temperature value, thereby eliminating the interference of substrate conductivity fluctuations caused by temperature changes on measurement accuracy.

4. The online coating thickness detection system for copper materials according to claim 3, characterized in that, The electrochemical impedance spectroscopy micro-region detection device includes a miniature three-electrode probe array, an electrolyte isothermal cycling unit, a broadband potentiostat, and a digital signal processing module. The miniature three-electrode probe array includes multiple detection units, each of which contains a platinum disk electrode as an auxiliary electrode, a saturated calomel electrode or a silver / silver chloride electrode as a reference electrode, and uses the copper substrate as the working electrode. The electrolyte isothermal circulation unit continuously supplies neutral electrolyte to the tiny area in contact with the coating through a microfluidic channel, and works in conjunction with a vacuum suction system to prevent the neutral electrolyte from overflowing. The wideband potentiostat is controlled by the digital signal processing module to generate a set of sinusoidal voltage excitation signals with a frequency range from 100 kHz to 0.01 Hz, the amplitude of which is limited to between 10 mV and 20 mV.

5. The online coating thickness detection system for copper materials according to claim 4, characterized in that, The electrochemical impedance spectroscopy micro-area detection device also includes a magnetic adsorption positioning mechanism and an electrolyte level monitoring circuit. The magnetic adsorption positioning mechanism includes a permanent magnet and an electromagnet disposed at the bottom of the miniature three-electrode probe array. The magnetic adsorption positioning mechanism is configured to, upon receiving a collection command, adsorb onto a support steel plate below the copper material via the permanent magnet, so that the miniature three-electrode probe array moves synchronously with the copper material. The electrolyte level monitoring circuit is configured to measure the DC conduction state between the auxiliary electrode and the reference electrode, thereby determining whether a liquid junction filled with electrolyte has been formed between the miniature three-electrode probe array and the coating surface. The digital signal processing module calculates the amplitude ratio and phase difference angle between the excitation signal voltage vector and the current response vector, and uses this to construct an impedance spectrum that includes the characteristics of the real and imaginary impedances.

6. The online coating thickness detection system for copper materials according to claim 5, characterized in that, The data fusion processing unit integrates a high-speed cache, a clock synchronization logic controller, and a spatial coordinate transformation module. The spatial coordinate transformation module is connected to the encoder of the production line drive motor, and is used to acquire feedback pulses in real time and calculate the dynamic travel speed of the copper material. The clock synchronization logic controller is configured to timestamp each set of acquired geometric thickness parameters and impedance spectrum. The spatial coordinate transformation module calculates the time deviation between the physical thickness measurement device and the electrochemical impedance spectroscopy micro-area detection device reaching the same position by integrating the dynamic travel speed over time. The module then uses the time deviation value to retrieve corresponding historical data from the cache for matching, ensuring that the geometric dimension and the electrochemical dimension correspond to the same coordinate region on the copper surface.

7. The online coating thickness detection system for copper materials according to claim 6, characterized in that, The data fusion processing unit is also equipped with a dynamic weight calculator, which is used to execute fusion logic based on feature weight allocation; The dynamic weight calculator adjusts the input weights of geometric features and electrochemical features in the coating defect prediction model in real time based on the data stability index output by the physical thickness measuring device and the signal-to-noise ratio index of the impedance spectrum. When the surface roughness of the coating is detected to cause the data jump value of the physical thickness measuring device to exceed the preset deviation threshold, the dynamic weight calculator automatically increases the weight ratio of the impedance spectrum feature in the model input layer to maintain the continuity of defect identification. The data fusion processing unit also uses a linear interpolation algorithm to numerically compensate for missing data points caused by inconsistent sampling frequencies.

8. The online coating thickness detection system for copper materials according to claim 7, characterized in that, The coating defect prediction model is deployed in a computing node equipped with a graphics processor, and its algorithm architecture includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and a classification output layer. The input layer is configured to normalize the received impedance spectrum matrix, mapping the real impedance, imaginary impedance, and frequency index to a numerical range between 0 and 1. The convolutional layer uses multi-scale convolutional kernels to perform sliding weighted summation operations on the impedance spectrum dimension to extract local correlation features between different frequency bands. The high-frequency features reflect the pore resistance of the coating, while the mid-to-low frequency features reflect the capacitance of the interface double layer. The pooling layer employs a maximum downsampling strategy to reduce the dimension of the feature map; the fully connected layer concatenates the extracted deep electrochemical features with the geometric thickness parameters to form a global feature vector. The classification output layer uses a normalized exponential function to calculate the probability value of the current coating being normally dense, having weak adhesion, micropores, or being insufficiently cured.

9. The online coating thickness detection system for copper materials according to claim 8, characterized in that, The coating defect prediction model also integrates an attention mechanism module and a drift adaptive subnetwork; The attention mechanism module is configured to reweight the feature maps extracted by the convolutional layer in terms of spatial and channel dimensions, so that the model automatically focuses on the features of the low-frequency inductive region. The drift adaptive subnetwork fine-tunes the forward propagation parameters of the model online by statistically analyzing the distribution characteristics of historical impedance data in real time, in order to eliminate systematic deviations caused by changes in electrolyte concentration or probe aging. Before deployment, the coating defect prediction model is also trained with adversarial sample enhancement. By introducing impedance feature samples with preset noise interference into the training set, the defect identification accuracy of the model under signal drift conditions is improved.

10. The online coating thickness detection system for copper materials according to claim 9, characterized in that, The early warning output unit is equipped with two levels of alarm logic and is connected to a process closed-loop interface and a local visualization terminal. The first-level alarm logic targets geometric thickness anomalies. When the geometric thickness parameter exceeds the preset upper and lower thresholds, a shutdown signal is triggered. The level 2 alarm logic targets inherent latent defects. When the thickness value is within the acceptable range but the coating defect prediction model determines that there is an inherent defect, an early warning message containing the defect type code, coordinate location information, and risk level score is generated. The process closed-loop interface is connected to the temperature control unit of the production line and is used to automatically increase the working temperature of the drying zone when insufficient curing defects are detected. The local visualization terminal is configured to render and display in real time the trend curve of the coating thickness, the equivalent circuit fitting parameters of the impedance spectrum, and the heat map of the defect distribution. The heat map uses different colors to represent different defect probability values.

Images