A method and system for detecting and calibrating DC fluctuations in electroplating

Abstract

This invention discloses a method and system for detecting and calibrating DC fluctuations in electroplating, belonging to the technical field of electroplating process parameter detection and calibration. The method acquires real-time electroplating DC signals such as current density, ion concentration, temperature gradient, and pressure fluctuations within the anolyte circulation channel, and constructs a fluctuation feature atlas through time-series segmentation. Dynamic mode decomposition is performed on the fluctuation feature atlas to extract steady-state fluctuation components and transient disturbance features. Using the steady-state fluctuation components combined with preset voltage thresholds across the ion exchange membrane, a fluctuation calibration curve including a DC bias adjustment value and an anolyte flow rate compensation coefficient is generated and mapped to the anolyte circulation control unit and the rectifier adjustment terminal. Based on the transient disturbance features, the start time and duration of the disturbance are identified, and the fluctuation calibration curve is corrected segmentally to obtain an updated fluctuation calibration curve. The system executes the above steps to achieve dynamic detection and calibration of DC fluctuations in electroplating.

Description

Technical Field

[0001] This invention relates to the field of electroplating process parameter detection and calibration technology, specifically to an electroplating DC fluctuation detection and calibration method and system. Background Technology

[0002] In electroplating processes, dynamic fluctuations of various physical quantities exist within the anolyte circulation channel of the insoluble anode module, directly affecting the electric field balance across the ion exchange membrane and the uniformity of the coating. Existing DC fluctuation detection and calibration techniques mostly monitor single electrical parameters, such as only acquiring current or voltage signals for compensation. They fail to incorporate the current density at the anolyte inlet, the ion concentration at the outlet, the temperature gradient inside the anode cavity, and the pressure fluctuations at the anode module interface into a unified characteristic description framework. This results in a one-sided characterization of DC fluctuations and an inability to reflect the coupled changes of multiple physical quantities caused by anolyte circulation. When extracting fluctuation components, common methods rely on time-domain statistical indicators or fixed-band filtering. These methods struggle to distinguish between inherent steady-state fluctuations synchronized with the anolyte circulation cycle and transient abnormal disturbances caused by external interference. Steady-state compensation and transient detection become intertwined, and calibration results are easily contaminated by disturbed data. When generating calibration strategies, preset constant bias values ​​or manual experience-based adjustments are typically used. The calibration curve lacks the ability to dynamically adjust based on actual fluctuation characteristics. When transient disturbances occur, the adjustment action exhibits significant lag, and the compensation for anolyte flow rate and DC bias cannot precisely match the disturbance range, leading to overcompensation or undercompensation and disrupting the stable ion exchange state within the anolyte flow channel. To improve the real-time performance and accuracy of detection calibration, it is necessary to accurately decouple steady-state fluctuation components from transient disturbance characteristics from synchronous observation data of multiple physical quantities. Based on this, a calibration curve capable of autonomously adapting to changes in operating conditions should be constructed, achieving continuous dynamic adjustment in both DC bias and anolyte flow rate dimensions. Simultaneously, local corrections to the calibration curve should be performed for identified disturbance ranges to avoid interference from transient fluctuations on the steady-state compensation logic. Summary of the Invention

[0003] This invention provides a method and system for detecting and calibrating DC fluctuations in electroplating. The purpose is to separate steady-state fluctuation components and transient disturbance characteristics from the real-time signals of multiple physical quantities in the anolyte circulation channel, generate a fluctuation calibration curve that includes the DC bias adjustment value and the anolyte flow rate compensation coefficient, and perform segmented correction on the calibration curve based on the identification of the disturbance start time and the disturbance duration interval, so as to replace the single-parameter fixed compensation method in the prior art and adapt to the dynamic changes of multi-physical quantity coupled fluctuations and transient disturbances in the electroplating process.

[0004] The objective of this invention can be achieved through the following technical solutions: This invention provides a method for detecting and calibrating DC fluctuations in electroplating, comprising: acquiring real-time electroplating DC signals in the anolyte circulation channel of an insoluble anode module within an electroplating tank; performing time-series segmentation on the real-time electroplating DC signals to construct a fluctuation feature atlas, wherein the real-time electroplating DC signals include the current density at the anolyte inlet, the ion concentration at the anolyte outlet, the temperature gradient inside the anode cavity, and the pressure fluctuations at the anode module interface; performing dynamic mode decomposition on the fluctuation feature atlas to extract steady-state fluctuation components for DC fluctuation compensation and transient disturbance feature quantities for triggering detection and calibration; based on... The steady-state fluctuation component, combined with the preset voltage thresholds on both sides of the ion exchange membrane of the anode module, generates a fluctuation calibration curve for the anode liquid circulation channel. The fluctuation calibration curve includes a DC bias adjustment value and an anode liquid flow rate compensation coefficient. The fluctuation calibration curve is mapped to the anode liquid circulation control unit and the rectifier adjustment terminal. Based on the transient disturbance characteristics, the disturbance start time and disturbance duration interval of the real-time electroplating DC signal within the sampling period are identified. The fluctuation calibration curve is then segmented and corrected according to the disturbance start time and the disturbance duration interval to generate an updated fluctuation calibration curve.

[0005] By incorporating multi-dimensional signals such as current density, ion concentration, temperature gradient, and pressure fluctuations into a unified fluctuation feature atlas and performing dynamic mode decomposition, the steady-state fluctuation component reflecting the inherent drift of the system and the transient disturbance characteristic quantity characterizing sudden anomalies can be accurately separated. Using the fluctuation calibration curve generated by the steady-state fluctuation component and a preset voltage threshold, and simultaneously coordinating the adjustment of the rectifier output current and anolyte circulation velocity, continuous and smooth DC bias compensation can be achieved. Based on the transient disturbance characteristic quantity, the calibration curve is piecewise corrected, enabling rapid response and temporary compensation within the disturbance range, effectively suppressing the impact of disturbances on coating quality. After the disturbance ends, the system automatically reverts to the original calibration strategy, balancing system stability and dynamic response capability.

[0006] As a technical solution of the present invention, when performing time-series segmentation on the real-time electroplating DC signal to construct a fluctuation feature atlas, the real-time electroplating DC signal is divided into multiple continuous single-cycle signal segments according to the anolyte circulation cycle of the insoluble anode module. Within each single-cycle signal segment, the current density sequence at the anolyte inlet, the ion concentration sequence at the anolyte outlet, the temperature gradient sequence inside the anode cavity, and the pressure fluctuation sequence at the anode module interface are extracted respectively. The current density sequence, the ion concentration sequence, the temperature gradient sequence, and the pressure fluctuation sequence within the same single-cycle signal segment are aligned according to the sampling time and combined into a single-cycle feature matrix. All single-cycle feature matrices are stacked in chronological order to form the fluctuation feature atlas. This method preserves the periodicity of the electroplating process, making the modal components obtained from subsequent decomposition more physically interpretable.

[0007] Furthermore, when performing dynamic mode decomposition on the fluctuation feature map set to extract steady-state fluctuation components and transient disturbance features, adjacent single-cycle feature matrices in the fluctuation feature map set are paired sequentially, and a linear mapping matrix between each pair of adjacent single-cycle feature matrices is calculated. The linear mapping matrix is ​​then subjected to eigenvalue decomposition to obtain multiple dynamic modes and their corresponding eigenvalues. Dynamic modes whose eigenvalue magnitudes are close to the value one are identified as steady-state fluctuation components, and dynamic modes whose eigenvalue magnitudes significantly deviate from the value one are identified as transient disturbance features. Preferably, when identifying dynamic modes whose eigenvalue magnitudes are close to the value one as steady-state fluctuation components, the absolute difference between the magnitude of the eigenvalue corresponding to each dynamic mode and the value one is calculated. Dynamic modes whose absolute difference is less than a preset steady-state threshold are classified into a steady-state mode set. All dynamic modes in the steady-state mode set are superimposed and reconstructed according to the magnitude of their corresponding eigenvalues ​​to obtain the steady-state fluctuation components. This reliably distinguishes between long-term gradual trends and short-term impulsive disturbances in the system.

[0008] As a preferred method for generating the fluctuation calibration curve, the steady-state current density waveform at the anolyte inlet and the steady-state ion concentration waveform at the anolyte outlet are reconstructed based on the steady-state fluctuation components. A first deviation sequence between the steady-state current density waveform and preset voltage thresholds on both sides of the ion exchange membrane is calculated, and a second deviation sequence between the steady-state ion concentration waveform and the preset voltage thresholds is calculated. The first deviation sequence and the second deviation sequence are then weighted and fused over time to obtain the DC bias adjustment value. The steady-state temperature gradient change rate inside the anode cavity is extracted based on the steady-state fluctuation components, and the steady-state temperature gradient change rate is mapped to the anolyte flow rate compensation coefficient. The DC bias adjustment value and the anolyte flow rate compensation coefficient are combined in chronological order to form the fluctuation calibration curve. Specifically, when weighting and fusing the first deviation sequence and the second deviation sequence, the weighting coefficient of the ion concentration sequence is greater than that of the current density sequence to highlight the dominant influence of ion concentration on membrane voltage deviation. Further preferably, during the fusion process, a first confidence weight is obtained for the current density at the anolyte inlet within a historical period, and a second confidence weight is obtained for the ion concentration at the anolyte outlet within a historical period. Each deviation value in the first deviation sequence is multiplied by the first confidence weight to obtain a weighted first deviation sequence; each deviation value in the second deviation sequence is multiplied by the second confidence weight to obtain a weighted second deviation sequence; at each sampling moment, the corresponding value in the weighted first deviation sequence is added to the corresponding value in the weighted second deviation sequence to obtain the DC bias adjustment value at that sampling moment. This method utilizes historical data to dynamically evaluate the confidence level of each signal channel, making the DC bias adjustment value more robust.

[0009] In terms of disturbance identification, based on the transient disturbance feature quantity, when identifying the disturbance start time and disturbance duration interval, the disturbance feature waveform is extracted from the transient disturbance feature quantity, and the local matching degree between the disturbance feature waveform and each sampling time in the fluctuation feature set is calculated; the first sampling time where the local matching degree exceeds a preset matching threshold is marked as the disturbance start time; starting from the disturbance start time, the search continues along the time axis to the end sampling time where the local matching degree is continuously lower than the preset matching threshold, and the time interval between the disturbance start time and the end sampling time is taken as the disturbance duration interval. This matching method can accurately capture the start and end boundaries of the disturbance, avoiding missed detections or false triggers caused by improper threshold settings.

[0010] When performing segmented correction on the fluctuation calibration curve according to the disturbance interval, a first calibration point corresponding to the start time of the disturbance is located on the fluctuation calibration curve, and a second calibration point corresponding to the end time of the disturbance duration interval is located; the curve segment in the fluctuation calibration curve located between the first calibration point and the second calibration point is marked as the segment to be corrected; the peak value of the transient disturbance characteristic quantity in the disturbance duration interval is obtained, and a temporary DC bias adjustment value and a temporary anolyte flow rate compensation coefficient are generated in the segment to be corrected based on the peak value; the DC bias adjustment value and the anolyte flow rate compensation coefficient in the segment to be corrected are replaced with the temporary DC bias adjustment value and the temporary anolyte flow rate compensation coefficient, respectively, to obtain the updated fluctuation calibration curve. As a specific scheme for generating temporary compensation, the first difference between the peak value and the DC bias adjustment value at the beginning of the segment to be corrected is calculated. This first difference is divided by the time length of the segment to be corrected to obtain the unit-time bias adjustment step size. According to the unit-time bias adjustment step size, the DC bias adjustment value is increased incrementally at each sampling time within the segment to be corrected to generate the temporary DC bias adjustment value. The second difference between the peak value and the anolyte flow rate compensation coefficient at the beginning of the segment to be corrected is calculated. This second difference is divided by the time length of the segment to be corrected to obtain the unit-time flow rate compensation step size. According to the unit-time flow rate compensation step size, the anolyte flow rate compensation coefficient is increased incrementally at each sampling time within the segment to be corrected to generate the temporary anolyte flow rate compensation coefficient. This forms a smooth transition compensation curve within the disturbance range, avoiding secondary impacts on the electroplating process caused by step adjustments.

[0011] When mapping the fluctuation calibration curve to the execution unit, the DC bias adjustment value in the fluctuation calibration curve is encoded as a rectifier current regulation command, and the current regulation command is sent to the rectifier regulation terminal; the anolyte flow rate compensation coefficient in the fluctuation calibration curve is encoded as a circulation pump speed regulation command, and the speed regulation command is sent to the anolyte circulation control unit; while the rectifier regulation terminal adjusts the output current according to the current regulation command, the anolyte circulation control unit synchronously adjusts the anolyte circulation flow rate according to the speed regulation command. Preferably, the rectifier regulation terminal and the anolyte circulation control unit are clock synchronized via real-time industrial Ethernet to ensure that the current adjustment and flow rate adjustment are completed within the same control cycle, thereby achieving precise coordinated control of electrical parameters and flow field parameters.

[0012] This invention also provides an electroplating DC fluctuation detection and calibration system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-described electroplating DC fluctuation detection and calibration method. This system can be integrated into the electroplating production line control system to perform online real-time monitoring and adaptive calibration of insoluble anode modules, improving the stability of electroplating DC power supply and the uniformity of the plating layer.

[0013] The beneficial effects of this invention are: By acquiring real-time signals of current density at the anolyte inlet, ion concentration at the outlet, temperature gradient inside the anode cavity, and pressure fluctuations at the anode module interface, and performing time-series segmentation, a fluctuation feature atlas aligned with multiple physical quantities is constructed, overcoming the limitations of single electrical parameter detection. Dynamic mode decomposition is performed on the fluctuation feature atlas, decomposing it into steady-state fluctuation components corresponding to the inherent cycle mode of the anolyte circulation and transient disturbance features reflecting abnormal dynamics. The steady-state fluctuation components retain the fundamental and main harmonic structures required for comparison with the voltage thresholds on both sides of the ion exchange membrane, while the transient disturbance features independently characterize short-term impacts deviating from the cycle steady state. Based on the steady-state current density waveform and steady-state ion concentration waveform reconstructed from the steady-state fluctuation components, the deviation sequence between the current density waveform and the preset voltage threshold can be accurately calculated. After weighted fusion, a DC bias adjustment value varying with the anolyte circulation cycle is generated. Simultaneously, the steady-state temperature gradient change rate is extracted from the steady-state fluctuation components and mapped to the anolyte flow rate compensation coefficient, forming a fluctuation calibration curve that continuously coordinates the DC bias and flow rate on the time axis. This curve is directly mapped to the rectifier regulation terminal and the anode liquid circulation control unit, enabling current adjustment and flow rate adjustment to be executed synchronously within the same control cycle, thus avoiding the regulation rigidity problem that occurs when the traditional fixed bias changes the cyclic operating conditions.

[0014] Transient disturbance characteristics are determined by calculating the local matching degree between the disturbance characteristic waveform and each sampling time to locate the disturbance start time and the disturbance duration interval. Then, the curve segment corresponding to the disturbance interval is delineated on the generated fluctuation calibration curve as the segment to be corrected. The peak value of the transient disturbance characteristics within the disturbance duration interval is extracted. Based on the difference between the peak value and the DC bias adjustment value and the anolyte flow rate compensation coefficient at the beginning of the segment to be corrected, the adjustment step size per unit time is calculated. Temporary DC bias adjustment values ​​and temporary anolyte flow rate compensation coefficients are generated incrementally at each sampling time within the segment to be corrected, replacing the values ​​in the original curve segment. This segmented correction method uses temporary parameters to take over the compensation logic during the disturbance period and directly restores the original steady-state compensation curve after the disturbance ends. The current density spikes and ion concentration shifts caused by transient disturbances are confined to the disturbance interval and processed independently, without affecting the extraction of steady-state fluctuation components and the generation of subsequent calibration curves, achieving compensation isolation and autonomous recovery of the disturbance interval. The combination of dynamic mode decomposition and segmented correction enables the detection and calibration process to have the accuracy of multi-physical quantity coordinated adjustment in the steady state stage and the ability of autonomous segmented correction in the transient disturbance stage. The overall compensation logic remains dynamically consistent with the actual fluctuation state in the anolyte flow channel. Attached Figure Description

[0015] The invention will now be further described with reference to the accompanying drawings.

[0016] Figure 1 This is a flowchart of the electroplating DC fluctuation detection and calibration method; Figure 2 This is a flowchart of the process for constructing a fluctuation feature atlas from real-time electroplating DC signals; Figure 3 This is a flowchart of the extraction process for steady-state fluctuation components and transient disturbance features based on dynamic mode decomposition; Figure 4 This is a flowchart of the segmented correction method for the fluctuation calibration curve. Detailed Implementation

[0017] 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.

[0018] See Figure 1This invention provides a method for detecting and calibrating DC fluctuations in electroplating, comprising: acquiring real-time electroplating DC signals in the anolyte circulation channel of an insoluble anode module within an electroplating tank; performing time-series segmentation on the real-time electroplating DC signals to construct a fluctuation feature atlas; wherein the real-time electroplating DC signals include the current density at the anolyte inlet, the ion concentration at the anolyte outlet, the temperature gradient inside the anode cavity, and the pressure fluctuations at the anode module interface; performing dynamic mode decomposition on the fluctuation feature atlas to extract steady-state fluctuation components for DC fluctuation compensation and transient disturbance feature quantities for triggering detection and calibration; and based on... Based on the steady-state fluctuation component and combined with the preset voltage thresholds on both sides of the ion exchange membrane of the anode module, a fluctuation calibration curve for the anode liquid circulation channel is generated. The fluctuation calibration curve includes a DC bias adjustment value and an anode liquid flow rate compensation coefficient. The fluctuation calibration curve is mapped to the anode liquid circulation control unit and the rectifier adjustment terminal. Based on the transient disturbance characteristic quantity, the disturbance start time and disturbance duration interval of the real-time electroplating DC signal within the sampling period are identified. The fluctuation calibration curve is segmented and corrected according to the disturbance start time and the disturbance duration interval to generate an updated fluctuation calibration curve.

[0019] In specific implementation, please refer to Figure 2 The real-time electroplating DC signal from the anolyte circulation channel of the insoluble anode module in the electroplating tank is acquired. This real-time electroplating DC signal includes the current density at the anolyte inlet, the ion concentration at the anolyte outlet, the temperature gradient inside the anode cavity, and the pressure fluctuation at the anode module interface. The process of time-series segmenting the real-time electroplating DC signal to construct a fluctuation feature atlas is as follows: The real-time electroplating DC signal is divided into multiple continuous single-cycle signal segments according to the anolyte circulation cycle of the insoluble anode module. The anolyte circulation cycle is determined based on the rotational speed of the anolyte circulation pump, the volume of the anolyte flow channel, and the preset anolyte replacement flow rate. Alternatively, the cycle corresponding to the frequency component with the largest amplitude can be extracted from the spectral analysis results of the current density signal at the anolyte inlet. During segmentation, the starting time of the real-time electroplating DC signal is taken as the starting point of the first single-cycle signal segment. A single-cycle signal segment is divided every other anolyte circulation cycle. Adjacent single-cycle signal segments are continuous in time and do not overlap, forming multiple single-cycle signal segments arranged chronologically.

[0020] Within each single-cycle signal segment, the current density sequence at the anolyte inlet, the ion concentration sequence at the anolyte outlet, the temperature gradient sequence inside the anode cavity, and the pressure fluctuation sequence at the anode module interface are extracted. The extraction of the current density sequence at the anolyte inlet is as follows: Current density data belonging to the current single-cycle signal segment is extracted from the real-time electroplating DC signal. This current density data is acquired by a current density sensor installed at the anolyte inlet at a fixed sampling frequency. After extraction, the data is arranged in chronological order of sampling time to form the current density sequence at the anolyte inlet. Similarly, the extraction of the ion concentration sequence at the anolyte outlet is as follows: Ion concentration data belonging to the current single-cycle signal segment is extracted from the real-time electroplating DC signal. This ion concentration data is acquired by an ion-selective electrode installed at the anolyte outlet at the same fixed sampling frequency. After extraction, the data is arranged in chronological order of sampling time to form the ion concentration sequence at the anolyte outlet. The process of extracting the temperature gradient sequence inside the anode cavity is as follows: Temperature gradient data belonging to the current single-cycle signal segment is extracted from the real-time electroplating DC signal. This temperature gradient data is calculated from the temperature difference measured by multiple pairs of thermocouples arranged along the depth direction of the anode cavity. The acquisition frequency is consistent with the aforementioned fixed sampling frequency. After extraction, the data is arranged according to the chronological order of sampling time to form the temperature gradient sequence inside the anode cavity. The process of extracting the pressure fluctuation sequence at the anode module interface is as follows: Pressure data belonging to the current single-cycle signal segment is extracted from the real-time electroplating DC signal. This pressure data is acquired by a pressure transmitter installed at the anode module interface. The acquisition frequency is consistent with the aforementioned fixed sampling frequency. The average value of the pressure data within the single-cycle signal segment is subtracted from the extracted pressure data to obtain the pressure fluctuation sequence at the anode module interface.

[0021] The current density sequence at the anolyte inlet, the ion concentration sequence at the anolyte outlet, the temperature gradient sequence inside the anode cavity, and the pressure fluctuation sequence at the anode module interface within the same single-cycle signal segment are aligned according to their sampling times and combined into a single-cycle feature matrix. The alignment method is as follows: using the sampling time of the current density sequence at the anolyte inlet as the baseline sequence, linear interpolation resampling is performed on the ion concentration sequence at the anolyte outlet, the temperature gradient sequence inside the anode cavity, and the pressure fluctuation sequence at the anode module interface using timestamps, ensuring that all four sequences have corresponding values ​​at the same sampling time point. The aligned current density value at the anolyte inlet, the ion concentration value at the anolyte outlet, the temperature gradient value inside the anode cavity, and the pressure fluctuation value at the anode module interface at the same sampling time are treated as a sampling vector. The sampling vectors from all sampling times within the single-cycle signal segment are arranged in chronological order to form the single-cycle feature matrix. The single-cycle feature matrix is ​​expressed as follows:

[0022] in: Indicates the first The single-period feature matrix corresponding to a single-period signal segment This is the index of the sequence number of a single-cycle signal segment. The value range is from 1 to the total number of signal segments in a single period; Indicates the first In the first single-cycle signal segment Current density at the anolyte inlet at each sampling time; Indicates the first In the first single-cycle signal segment Ion concentration at the anolyte outlet at each sampling time; Indicates the first In the first single-cycle signal segment Temperature gradient inside the anode cavity at each sampling time; Indicates the first In the first single-cycle signal segment Pressure fluctuations at the anode module interface at each sampling time; This represents the total number of sampling times contained in each single-cycle signal segment. The value is obtained by rounding down the product of the anolyte circulation period and the fixed sampling frequency.

[0023] All single-period feature matrices are stacked in chronological order to form a fluctuation feature atlas. The stacking method is as follows: Arranged sequentially along the plane normal direction of the matrix, a three-dimensional data structure is formed. The three dimensions correspond to the single-cycle signal segment number, sampling time number, and feature channel number, respectively. Feature channel 1 stores the current density at the anolyte inlet, feature channel 2 stores the ion concentration at the anolyte outlet, feature channel 3 stores the temperature gradient inside the anode cavity, and feature channel 4 stores the pressure fluctuation at the anode module interface. The three-dimensional data structure is the fluctuation feature atlas.

[0024] In specific implementation, please refer to Figure 3 Dynamic mode decomposition is performed on the fluctuation feature atlas to extract steady-state fluctuation components for DC fluctuation compensation and transient disturbance features for triggering detection calibration. The fluctuation feature atlas consists of multiple single-cycle feature matrices arranged in chronological order. The first feature matrix in the fluctuation feature atlas is then used to extract the steady-state fluctuation components for DC fluctuation compensation and the transient disturbance features for triggering detection calibration. The single-period characteristic matrix is ​​denoted as , No. The single-period characteristic matrix is ​​denoted as ,in The value ranges from 1 to , This represents the total number of single-period characteristic matrices. The single-period characteristic matrices... and Pair them in sequence to form For adjacent single-period characteristic matrices, calculate the linear mapping matrix between each pair of adjacent single-period characteristic matrices. The linear mapping matrix is ​​calculated by: seeking the matrix... , making Approximately equal to and The product of . Using the least squares method, the linear mapping matrix. Determined by the following expression:

[0025] in: Indicates the first For the linear mapping matrix between adjacent single-period characteristic matrices, The dimension is ; Indicates the first There are single-period feature matrices with dimensions of . , The total number of sampling moments within a single-cycle signal segment; Indicates the first A single-period feature matrix, with dimension and same; Representation matrix The generalized inverse matrix with dimension . Calculate the generalized inverse matrix. When using singular value decomposition to... The matrix is ​​decomposed into the product of a left singular vector matrix, a singular value diagonal matrix, and a right singular vector matrix. Then, the reciprocal of each singular value in the singular value diagonal matrix is ​​taken, and singular values ​​less than a truncation threshold are set to zero. The truncation threshold is set to 0.01 times the maximum singular value. This value is chosen because, under the condition that the signal-to-noise ratio of the current density signal at the anolyte inlet is not less than 20 dB, singular value components less than 0.01 times the maximum singular value are mainly contributed by measurement noise. Multiplying the matrix after taking the reciprocals and setting them to zero with the left and right singular vector matrices yields the generalized inverse matrix. The above process is performed on each pair of adjacent single-period characteristic matrices to obtain... linear mapping matrices .

[0026] In some embodiments, eigenvalue decomposition is performed on the linear mapping matrix to obtain multiple dynamic modes and their corresponding eigenvalues. For each linear mapping matrix... Perform eigenvalue decomposition to solve for the condition that... eigenvalues and the corresponding feature vector Among them, subscript The component index is obtained from the eigenvalue decomposition. The value of ranges from 1 to the linear mapping matrix. The rank of the eigenvectors. As a dynamic modality, the dimension is , To and The corresponding eigenvalues. All All dynamic modes and eigenvalues ​​obtained from the decomposition of the linear mapping matrix are collected and merged into a dynamic mode set. and eigenvalue set ,in It is a global index.

[0027] Optionally, for each eigenvalue in the dynamic mode set, calculate the magnitude of the eigenvalue. And calculate the absolute difference between the modulus and the value one. The preset steady-state threshold is denoted as... , The value is set to 0.05, based on the following: when the eigenvalue magnitude is within the interval... At this time, the time evolution characteristics of the corresponding dynamic mode exhibit periodic or near-steady-state fluctuations, consistent with the steady-state fluctuation characteristics of the anolyte circulation channel. The time evolution corresponding to characteristic values ​​exceeding this range shows a rapid decay or divergence trend, corresponding to transient processes caused by bubble disturbances or external interference in signals such as the current density at the anolyte inlet or the ion concentration at the anolyte outlet. This will satisfy... The dynamic modes are identified as steady-state ripple components for DC ripple compensation, and these dynamic modes are categorized into the steady-state mode set. Will satisfy The dynamic mode determination is used as a transient perturbation characteristic quantity to trigger detection calibration.

[0028] In practical implementation, the steady-state mode set All dynamic modes are superimposed and reconstructed according to the magnitude of their corresponding eigenvalues ​​to obtain the steady-state fluctuation components. During reconstruction, each dynamic mode belonging to the steady-state mode set is obtained. initial coefficients , This is an index within the set of steady-state modes. Initial coefficients. By analyzing the first single-period characteristic matrix The solution is obtained by least squares fitting, that is, by obtaining Each column and To minimize the sum of squared residuals, solve for each... Then, for the first... A single-cycle signal segment, the steady-state fluctuation component is constructed by superimposing all dynamic modes in the steady-state mode set according to time evolution. The superposition method is as follows: dynamic modes... Multiply by its initial coefficient Then multiply by the eigenvalue of The power of 1, and finally all Summing, we get the first... The steady-state fluctuation component reconstruction matrix corresponds to a single-cycle feature matrix. Each column of this reconstruction matrix corresponds to the steady-state fluctuation component sequence of the current density at the anolyte inlet, the ion concentration at the anolyte outlet, the temperature gradient inside the anode cavity, and the pressure fluctuation at the anode module interface, which constitutes the steady-state fluctuation component for DC fluctuation compensation.

[0029] In practical implementation, a fluctuation calibration curve for the anolyte circulation channel is generated based on the steady-state fluctuation components and the preset voltage thresholds on both sides of the ion exchange membrane of the anode module. The method for reconstructing the steady-state current density waveform at the anode inlet and the steady-state ion concentration waveform at the anode outlet based on the steady-state fluctuation components is as follows: extract the sequence of the current density characteristic dimension at the anode inlet from the steady-state fluctuation components, arrange the extracted sequence according to the time sequence of single-cycle signal segments, and splice them into a continuous steady-state current density waveform, with each point on the waveform corresponding to the steady-state current density value at a sampling time; extract the sequence of the ion concentration characteristic dimension at the anode outlet from the steady-state fluctuation components, and splice them into a continuous steady-state ion concentration waveform in the same way.

[0030] When calculating the first deviation sequence between the steady-state current density waveform and the preset voltage thresholds across the ion exchange membrane, the preset voltage thresholds are first obtained. These thresholds are calculated from the electrochemical model in the electroplating process specification, which uses the ion concentration at the anolyte outlet and the current density at the anolyte inlet under steady-state operating conditions for the insoluble anode module. The conversion relationship is determined based on the Nernst equation and the ion migration characteristics across the ion exchange membrane. The steady-state current density value at each sampling moment is subtracted from the standard current density value corresponding to the preset voltage threshold to obtain the deviation value for the corresponding sampling moment in the first deviation sequence. The standard current density value is the ratio of the preset voltage threshold to the equivalent resistance of the ion exchange membrane. When calculating the second deviation sequence between the steady-state ion concentration waveform and the preset voltage threshold, the steady-state ion concentration value at each sampling moment is subtracted from the standard ion concentration value corresponding to the preset voltage threshold to obtain the deviation value for the corresponding sampling moment in the second deviation sequence. The standard ion concentration value is the reference ion concentration value obtained by back-calculating the preset voltage threshold and the ion-selective electrode calibration curve.

[0031] The first and second deviation sequences are weighted and fused over time to obtain the DC bias adjustment value. During the weighted fusion process, the weighting coefficient of the ion concentration sequence is greater than that of the current density sequence.

[0032] In some embodiments, a first confidence weight of the current density at the anolyte inlet end within a historical period is obtained, and a second confidence weight of the ion concentration at the anolyte outlet end within a historical period is obtained. The first and second confidence weights are determined by the signal-to-noise ratio data of the current density measurement channel at the anolyte inlet end and the ion concentration measurement channel at the anolyte outlet end within the historical period. Within the historical period, the noise variance of the current density signal at the anolyte inlet end under steady-state conditions is statistically analyzed and denoted as... The noise variance of the ion concentration signal at the anolyte outlet under steady-state conditions is statistically analyzed and denoted as . The first confidence weight is determined by the proportion of the inverse of the measurement channel noise variance to the total weight, and the second confidence weight is also determined by the proportion of the inverse of the measurement channel noise variance to the total weight. The calculation method for the first confidence weight is as follows: The second confidence weight is calculated as follows: Because the ion concentration measurement channel at the anolyte outlet has a lower noise level in the low-frequency range, which is directly related to the electrochemical equilibrium across the ion exchange membrane, while the current density measurement channel at the anolyte inlet suffers from a larger noise variance due to high-frequency ripple interference from the rectifier, this results in... ,therefore That is, the weighting coefficient of the ion concentration sequence is greater than the weighting coefficient of the current density sequence.

[0033] Each deviation value in the first deviation sequence is multiplied by a first confidence weight to obtain a weighted first deviation sequence; each deviation value in the second deviation sequence is multiplied by a second confidence weight to obtain a weighted second deviation sequence. At each sampling time, the corresponding value in the weighted first deviation sequence is added to the corresponding value in the weighted second deviation sequence to obtain the DC bias adjustment value at that sampling time. The DC bias adjustment value is generated by the following expression:

[0034] in: Indicates the first Each sampling time DC bias adjustment value; This represents the first confidence weight, which is a coefficient calculated from the historical periodic noise variance. The specific value determined based on the noise variance ratio is between 0.2 and 0.4. This represents the second confidence weight, with a value of 1 minus... The value is between 0.6 and 0.8; This indicates the first deviation sequence at sampling time. The deviation value, in amperes per square decimeter; This indicates the second bias sequence at sampling time. The deviation value is expressed in moles per liter. and The specific value is based on the current density at the anolyte inlet collected over a historical period, with a channel noise standard deviation of 0.05 A / dm. 2 The calculation was performed when the standard deviation of the ion concentration channel noise at the anolyte outlet was 0.02 mol / L. The value is 0.3. The value is 0.7.

[0035] The steady-state temperature gradient rate of change inside the anode cavity is extracted based on the steady-state fluctuation components. The extraction method involves extracting a sequence corresponding to the temperature gradient feature dimension inside the anode cavity from the steady-state fluctuation components, and then calculating the first-order backward difference of the extracted sequence along the time direction to obtain the steady-state temperature gradient rate of change at each sampling moment. When mapping the steady-state temperature gradient rate of change to the anode fluid flow rate compensation coefficient, a piecewise linear mapping relationship is used. The mapping relationship is constructed based on the anode fluid heat exchange model. The mapping method for the anode fluid flow rate compensation coefficient is as follows: when the steady-state temperature gradient rate of change is within a preset normal range, the anode fluid flow rate compensation coefficient is set to a value of one, indicating no additional compensation; when the steady-state temperature gradient rate of change exceeds the upper limit of the preset normal range, the anode fluid flow rate compensation coefficient increases linearly with the increase of the steady-state temperature gradient rate of change, and the linear slope is the reciprocal of the product of the anode fluid specific heat capacity and the rated flow rate of the circulating pump; when the steady-state temperature gradient rate of change is below the lower limit of the preset normal range, the anode fluid flow rate compensation coefficient decreases linearly with the decrease of the steady-state temperature gradient rate of change, and the linear slope is symmetrical to the slope when exceeding the upper limit. The preset normal variation range is set based on the internal thermal balance test data of the anode cavity, with an upper limit of +0.5℃ / s and a lower limit of -0.5℃ / s.

[0036] In some embodiments, the DC bias adjustment value and the anolyte flow rate compensation coefficient are combined in chronological order to form a fluctuation calibration curve. The fluctuation calibration curve is a two-dimensional ordered set of points, with each sampling time corresponding to a calibration point. The first dimension of the calibration point is the DC bias adjustment value, and the second dimension is the anolyte flow rate compensation coefficient. The calibration points for all sampling times are arranged in chronological order to form a complete fluctuation calibration curve.

[0037] In specific implementation, please refer to Figure 4Based on transient disturbance characteristics, the starting time and duration of disturbances in the real-time electroplating DC signal within the sampling period are identified. Disturbance feature waveforms are extracted from the transient disturbance characteristics by reconstructing each dynamic mode identified as a transient disturbance characteristic in the time domain according to its corresponding feature value. This yields the time-domain waveform of each transient disturbance characteristic within the entire single-cycle signal segment. Then, the time-domain waveforms of all transient disturbance characteristics are superimposed at the same sampling time to form the disturbance feature waveform. The disturbance feature waveform covers the entire sampling period of the real-time electroplating DC signal, with each sampling time corresponding to a disturbance feature amplitude.

[0038] Calculate the local matching degree between the perturbation feature waveform and the wave feature map at each sampling time. The local matching degree is measured using the sliding window inner product similarity. The perturbation feature waveform is then compared with the wave feature map at each sampling time. Cut off a width of A window segment of sampling points is denoted as the perturbation feature window vector; the sampling time is extracted from the fluctuation feature map set. The corresponding feature vectors, in the fluctuation feature set, consist of the current density value at the anolyte inlet, the ion concentration value at the anolyte outlet, the temperature gradient value inside the anode cavity, and the pressure fluctuation value at the anode module interface. Similarly, at each sampling time... The width of the front and back sections is A window segment of sampling points is used to construct a fluctuation feature window matrix, where each row of the fluctuation feature window matrix corresponds to a feature vector at a sampling time. The disturbance feature window vector is then multiplied by the feature vector at each corresponding sampling time in the fluctuation feature window matrix, and the average value of the result within the window is taken as the sampling time. Local matching degree. Sliding window half width. It is set to one-tenth of the number of sampling points within the anolyte circulation cycle. This value is set based on the process experience that the shortest duration of a disturbance event within one anolyte circulation cycle is one-tenth of the cycle.

[0039] The first sampling time when the local matching degree exceeds the preset matching threshold is marked as the disturbance start time. The preset matching threshold is denoted as... , The threshold is set to 0.35. The rationale is as follows: when the local matching degree reaches 0.35, it indicates a significant structural similarity between the disturbance characteristic waveform and the wave characteristic atlas of the signal segment near that sampling time. This corresponds to an abnormal change in the current density at the anolyte inlet or the ion concentration at the anolyte outlet, deviating significantly from the steady-state wave characteristics. The value of 0.35 is determined by adding three times the standard deviation to the statistical upper limit of the local matching degree under normal operating conditions. The statistical upper limit of the local matching degree under normal operating conditions is 0.15, and the standard deviation is 0.067. Starting from the initial sampling time of the real-time electroplating DC signal, the local matching degree of each sampling time is checked sequentially along the positive time axis. The first sampling time to satisfy a local matching degree greater than the preset matching threshold is selected. The sampling time is marked as the disturbance start time, denoted as . .

[0040] Starting from the initial time of the disturbance, search forward along the time axis for the last sampling time when the local matching degree continuously falls below a preset matching threshold. Starting from this point, the local matching degree is checked at each sampling time along the positive direction of the time axis. When a continuous... The local matching degree at each sampling time point is less than the preset matching threshold. At that time, it will be continuous The first sampling time in a set of sampling times is marked as the end sampling time, denoted as . Number of consecutive sampling times The sampling time is set to one-fifth of the number of sampling points within the anolyte circulation cycle. This setting is based on the fact that after a disturbance event subsides, the current density at the anolyte inlet and the ion concentration at the anolyte outlet need a sufficiently long confirmation interval to recover to their steady-state fluctuation characteristics. This is to avoid misinterpreting short-term fluctuations during the disturbance process as the end of the disturbance. A confirmation time of one-fifth of the cycle can cover a complete response transition process of the anolyte circulation system. The disturbance start time is set to... With the end of sampling time The time interval between these points is considered the disturbance continuation interval, and all sampling times included within the disturbance continuation interval are periods affected by transient disturbances.

[0041] In some embodiments, the fluctuation calibration curve is segmented and corrected based on the disturbance start time and the disturbance duration interval to generate an updated fluctuation calibration curve. A first calibration point corresponding to the disturbance start time is located on the fluctuation calibration curve, and a second calibration point corresponding to the end time of the disturbance duration interval is located. The location method is as follows: the fluctuation calibration curve is represented by an ordered set with a one-to-one correspondence between sampling time parameters and calibration points. The sampling time value is searched within the ordered set to find the calibration point corresponding to the disturbance start time. The calibration point that is equal to or closest to the original calibration point is designated as the first calibration point; the sampling time is then used to find the end sampling time. The calibration point that is equal to or closest to the calibration point is designated as the second calibration point.

[0042] The curve segment in the fluctuation calibration curve located between the first calibration point and the second calibration point is marked as the segment to be corrected. All calibration points within the segment to be corrected constitute an ordered subsequence, with the first calibration point as the starting calibration point and the second calibration point as the ending calibration point, including both the first and second calibration points.

[0043] The peak value of the transient disturbance characteristic quantity within the disturbance duration interval is obtained, and the peak value is denoted as . The acquisition method is as follows: within the entire sampling time range covered by the disturbance duration interval, the amplitude of the transient disturbance characteristic quantity is compared point by point, and the maximum absolute value of the amplitude is taken as the peak value of the transient disturbance characteristic quantity. Based on the peak value of the transient disturbance characteristic quantity, a temporary DC bias adjustment value and a temporary anolyte flow rate compensation coefficient are generated for the segment to be corrected.

[0044] Optionally, the first difference between the peak value of the transient disturbance characteristic and the DC bias adjustment value at the beginning of the segment to be corrected is calculated. This first difference is then divided by the time length of the segment to be corrected to obtain the unit-time bias adjustment step size. The DC bias adjustment value at the beginning of the segment to be corrected is the DC bias adjustment value corresponding to the first calibration point, denoted as... The first difference is calculated as follows: The method for calculating the unit time offset adjustment step size is as follows: According to the unit time offset adjustment step size, the DC offset adjustment value is increased incrementally at each sampling time within the segment to be corrected to generate a temporary DC offset adjustment value. The generation method is as follows: for the first sampling time within the segment to be corrected... Temporary DC bias adjustment value at each sampling time ,in The value of is an integer ranging from 0 to the total number of sampling times within the segment to be corrected minus one.

[0045] The second difference between the peak value of the transient disturbance characteristic and the anolyte velocity compensation coefficient at the beginning of the segment to be corrected is calculated. This second difference is then divided by the time length of the segment to be corrected to obtain the velocity compensation step size per unit time. The anolyte velocity compensation coefficient at the beginning of the segment to be corrected is the same as the anolyte velocity compensation coefficient corresponding to the first calibration point, denoted as... The second difference is calculated as follows: ,in The velocity mapping factor represents the increment of the anolyte velocity compensation coefficient required per unit amplitude of the peak value of the transient disturbance characteristic quantity. The value of is determined by the following formula:

[0046] in: This represents the flow rate mapping factor, with units of (flow rate compensation coefficient) / (current density units). This represents the maximum change in the anolyte flow rate compensation coefficient that the anolyte circulation control unit can respond to under transient disturbance conditions. It is set according to the speed range and response characteristics of the anolyte circulation pump and is taken as 0.3. This represents the maximum magnitude of change in current density at the anolyte inlet recorded in historical operating data due to transient disturbances, and is valued at 0.8 A / dm². 2 Therefore, the flow rate mapping factor The specific value is 0.375. The calculation method for the unit time flow rate compensation step size is as follows: According to the unit time flow rate compensation step size, the anolyte flow rate compensation coefficient is increased incrementally at each sampling time within the segment to be corrected, generating a temporary anolyte flow rate compensation coefficient. The generation method is as follows: for the first sampling time within the segment to be corrected... Temporary anolyte flow rate compensation coefficient at each sampling time .

[0047] The DC bias adjustment value and anolyte flow rate compensation coefficient in the segment to be corrected are replaced with temporary DC bias adjustment values ​​and temporary anolyte flow rate compensation coefficients, respectively, to obtain an updated fluctuation calibration curve. The replacement method is as follows: the calibration point values ​​before the first calibration point and after the second calibration point in the fluctuation calibration curve are kept unchanged; the DC bias adjustment value at each sampling time between the first and second calibration points is replaced with the temporary DC bias adjustment value at the corresponding sampling time; and the anolyte flow rate compensation coefficient at each sampling time is replaced with the temporary anolyte flow rate compensation coefficient at the corresponding sampling time. After the replacement is completed, an updated fluctuation calibration curve is formed.

[0048] In practice, the fluctuation calibration curve is mapped to the anolyte circulation control unit and the rectifier adjustment terminal. The DC bias adjustment value in the fluctuation calibration curve is encoded as the rectifier's current adjustment command. The encoding process is as follows: extract the DC bias adjustment value corresponding to each sampling moment in the fluctuation calibration curve. The DC bias adjustment value is expressed as the current density deviation in amperes per square decimeter. Multiply the DC bias adjustment value by the effective plating area of ​​the insoluble anode module to convert it into the absolute adjustment amount of the rectifier output current. The conversion formula is:

[0049] in: Indicates at the sampling time The amount of current regulation that needs to be applied to the rectifier output, in amperes; Indicates at the sampling time DC bias adjustment value, in amperes per square decimeter; This indicates the effective electroplating area of ​​the insoluble anode module, expressed in square decimeters. The value is the sum of the surface areas of the anode plates actually immersed in the anolyte in the insoluble anode module, calculated from the dimensions of the insoluble anode module design drawings. The rectifier's current set output current value is added to the current adjustment amount. The target output current value is obtained, and then encoded into a current regulation command according to the data frame format specified by the rectifier communication protocol. The rectifier communication protocol adopts the Modbus RTU protocol. The data frame of the current regulation command includes the rectifier slave address, function code, register start address, number of registers, number of data bytes, and CRC check value. The encoded current regulation command is sent to the rectifier regulation terminal through the RS-485 physical link.

[0050] The anolyte flow rate compensation coefficient in the fluctuation calibration curve is encoded as the speed adjustment command for the circulating pump. The encoding process is as follows: Extract the anolyte flow rate compensation coefficient corresponding to each sampling moment in the fluctuation calibration curve. The anolyte flow rate compensation coefficient is a dimensionless multiplicative coefficient, representing the speed adjustment ratio that needs to be applied based on the current reference speed of the anolyte circulating pump. Obtain the current reference speed of the anolyte circulating pump. The current reference speed is obtained by converting the real-time operating frequency read by the anolyte circulation control unit from the circulating pump frequency converter driver. The conversion relationship is: the circulating pump speed equals the output frequency of the circulating pump frequency converter driver multiplied by 60 and then divided by the number of pole pairs of the circulating pump motor. Multiply the current reference speed by the anolyte flow rate compensation coefficient to obtain the target speed value. Then, convert the target speed value into a frequency setting value that the circulating pump frequency converter driver can recognize. The conversion method is: the frequency setting value of the circulating pump frequency converter driver equals the target speed value multiplied by the number of pole pairs of the circulating pump motor and then divided by 60. The frequency setpoint of the circulating pump inverter driver is encoded into a speed adjustment command according to the data frame format specified in the circulating pump inverter driver communication protocol. The circulating pump inverter driver communication protocol adopts the CANopen protocol. The data frame of the speed adjustment command includes COB-ID, command word, index, sub-index, and data field. The encoded speed adjustment command is then sent to the anolyte circulation control unit via the CAN bus.

[0051] While the rectifier regulating terminal adjusts the output current according to the current regulation command, the anolyte circulation control unit synchronously adjusts the anolyte circulation flow rate according to the speed regulation command. Upon receiving the current regulation command, the rectifier regulating terminal parses the target output current value in the data frame and writes it into the setpoint register of the rectifier's current closed-loop controller. The rectifier's current closed-loop controller adjusts the thyristor firing angle or IGBT duty cycle according to the setpoint, making the rectifier output current track the target output current value. Upon receiving the speed regulation command, the anolyte circulation control unit parses the frequency setpoint in the data frame and writes it into the frequency command channel of the circulation pump frequency converter. The circulation pump frequency converter adjusts the output frequency according to the frequency setpoint, making the circulation pump speed reach the target speed value, thereby adjusting the anolyte flow rate in the anolyte circulation channel.

[0052] In some embodiments, the rectifier regulating terminal and the anolyte circulation control unit synchronize their clocks via real-time industrial Ethernet to ensure that current adjustment and flow rate adjustment are completed within the same control cycle. The real-time industrial Ethernet uses the EtherCAT protocol, with both the rectifier regulating terminal and the anolyte circulation control unit acting as EtherCAT slave devices connected to the same EtherCAT master station. The EtherCAT master station periodically sends synchronization frames containing the system clock reference to all slave devices. These synchronization frames carry a distributed clock reference timestamp at a specific sub-message location within the EtherCAT data frame. Upon receiving the synchronization frame, the EtherCAT slave controllers within both the rectifier regulating terminal and the anolyte circulation control unit latch their local clock count, calculate the deviation and transmission delay between their local clock and the reference clock based on the distributed clock synchronization mechanism, and compensate and calibrate their local clocks to ensure that the synchronization error between the local clock of the rectifier regulating terminal and the local clock of the anolyte circulation control unit does not exceed 1 microsecond.

[0053] Optionally, the control cycle is determined by the cyclic scan cycle of the EtherCAT master station, which is set to 500 microseconds. At the beginning of each control cycle, the EtherCAT master station simultaneously sends process data frames to the rectifier regulation terminal and the anolyte circulation control unit. These process data frames contain the current regulation command and speed regulation command corresponding to the current control cycle. Within the same control cycle after receiving the process data frame, the rectifier regulation terminal and the anolyte circulation control unit respectively adjust the rectifier output current and the anolyte circulation flow rate, ensuring that the current adjustment and flow rate adjustment actions are synchronized in time.

[0054] It can be understood that the electroplating DC fluctuation detection and calibration system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the electroplating DC fluctuation detection and calibration method. The memory stores instruction codes for a fluctuation feature map construction module, a dynamic mode decomposition module, a fluctuation calibration curve generation module, a transient disturbance identification and segmentation correction module, and a mapping execution module. The processor calls the instruction code of the fluctuation feature map construction module to perform time-series segmentation of the real-time electroplating DC signal to construct a fluctuation feature map. The processor calls the instruction code of the dynamic mode decomposition module to perform dynamic mode decomposition on the fluctuation feature map and extract steady-state fluctuation components and transient disturbance features. The processor calls the instruction code of the fluctuation calibration curve generation module to generate a fluctuation calibration curve based on the steady-state fluctuation components and a preset voltage threshold. The processor calls the instruction code of the transient disturbance identification and segmentation correction module to identify the disturbance start time and the disturbance duration interval and perform segmentation correction on the fluctuation calibration curve. The processor calls the instruction code of the mapping execution module to perform the operation of mapping the fluctuation calibration curve to the anolyte circulation control unit and the rectifier regulation terminal.

[0055] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for detecting and calibrating DC fluctuations in electroplating, characterized in that, include: The real-time electroplating DC signal in the anolyte circulation channel of the insoluble anode module in the electroplating tank is acquired, and the real-time electroplating DC signal is time-series segmented to construct a fluctuation feature atlas. The real-time electroplating DC signal includes the current density at the anolyte inlet, the ion concentration at the anolyte outlet, the temperature gradient inside the anode cavity, and the pressure fluctuation at the anode module interface. Dynamic mode decomposition is performed on the wave feature map to extract the steady-state wave component for DC wave compensation and the transient disturbance feature for triggering detection calibration. Based on the steady-state fluctuation component and combined with the preset voltage thresholds on both sides of the ion exchange membrane of the anode module, a fluctuation calibration curve for the anode liquid circulation channel is generated. The fluctuation calibration curve includes a DC bias adjustment value and an anode liquid flow rate compensation coefficient. The fluctuation calibration curve is then mapped to the anode liquid circulation control unit and the rectifier adjustment terminal. Based on the transient disturbance characteristics, the disturbance start time and disturbance duration interval of the real-time electroplating DC signal within the sampling period are identified. The fluctuation calibration curve is then segmented and corrected according to the disturbance start time and the disturbance duration interval to generate an updated fluctuation calibration curve.

2. The electroplating DC fluctuation detection and calibration method according to claim 1, characterized in that, The specific steps of performing time-series segmentation on the real-time electroplating DC signal to construct a fluctuation feature atlas include: The real-time electroplating DC signal is divided into multiple continuous single-cycle signal segments according to the anolyte circulation cycle of the insoluble anode module. Within each single-cycle signal segment, the current density sequence at the anolyte inlet, the ion concentration sequence at the anolyte outlet, the temperature gradient sequence inside the anode cavity, and the pressure fluctuation sequence at the anode module interface are extracted respectively. The current density sequence, ion concentration sequence, temperature gradient sequence, and pressure fluctuation sequence within the same single-cycle signal segment are aligned according to the sampling time and combined into a single-cycle feature matrix; All single-cycle feature matrices are stacked in chronological order to form the fluctuation feature atlas.

3. The electroplating DC fluctuation detection and calibration method according to claim 2, characterized in that, The steps of performing dynamic mode decomposition on the wave feature map to extract the steady-state wave component for DC wave compensation and the transient disturbance feature quantity for triggering detection calibration specifically include: Pair adjacent single-cycle feature matrices in the wave feature set in sequence, and calculate the linear mapping matrix between each pair of adjacent single-cycle feature matrices. Eigenvalue decomposition is performed on the linear mapping matrix to obtain multiple dynamic modes and their corresponding eigenvalues; Dynamic modes with eigenvalues ​​whose magnitudes are close to one are identified as the steady-state fluctuation component, while dynamic modes with eigenvalues ​​whose magnitudes deviate significantly from one are identified as the transient disturbance characteristic.

4. The electroplating DC fluctuation detection and calibration method according to claim 3, characterized in that, The steps for generating a fluctuation calibration curve for the anolyte circulation channel based on the steady-state fluctuation component and in conjunction with the preset voltage thresholds across the ion exchange membrane of the anode module specifically include: The steady-state current density waveform at the anolyte inlet and the steady-state ion concentration waveform at the anolyte outlet are reconstructed based on the steady-state fluctuation components. Calculate the first deviation sequence between the steady-state current density waveform and the preset voltage thresholds on both sides of the ion exchange membrane, and calculate the second deviation sequence between the steady-state ion concentration waveform and the preset voltage thresholds; The first deviation sequence and the second deviation sequence are weighted and fused by time to obtain the DC bias adjustment value; The steady-state temperature gradient change rate inside the anode cavity is extracted based on the steady-state fluctuation component, and the steady-state temperature gradient change rate is mapped to the anode liquid flow rate compensation coefficient. The DC bias adjustment value and the anolyte flow rate compensation coefficient are combined in chronological order to form the fluctuation calibration curve.

5. The electroplating DC fluctuation detection and calibration method according to claim 4, characterized in that, When the first deviation sequence and the second deviation sequence are fused together by time weighting, the weighting coefficient of the ion concentration sequence is greater than the weighting coefficient of the current density sequence.

6. The electroplating DC fluctuation detection and calibration method according to claim 3, characterized in that, Based on the transient disturbance characteristics, the steps for identifying the disturbance start time and disturbance duration interval of the real-time electroplating DC signal within the sampling period specifically include: Extract the disturbance feature waveform from the transient disturbance feature quantity, and calculate the local matching degree between the disturbance feature waveform and each sampling time in the fluctuation feature set; The first sampling moment in which the local matching degree exceeds the preset matching threshold is marked as the disturbance start moment; Starting from the disturbance initiation time, search backward along the time axis for the end sampling time when the local matching degree is continuously lower than the preset matching threshold, and take the time interval between the disturbance initiation time and the end sampling time as the disturbance continuation interval.

7. The electroplating DC fluctuation detection and calibration method according to claim 6, characterized in that, The steps of segmenting and correcting the fluctuation calibration curve based on the disturbance start time and the disturbance duration interval to generate an updated fluctuation calibration curve specifically include: Locate a first calibration point on the fluctuation calibration curve corresponding to the start time of the disturbance, and locate a second calibration point corresponding to the end time of the disturbance duration interval; The curve segment in the fluctuation calibration curve located between the first calibration point and the second calibration point is marked as the segment to be corrected; Obtain the peak value of the transient disturbance characteristic quantity within the disturbance duration interval, and generate a temporary DC bias adjustment value and a temporary anolyte flow rate compensation coefficient within the segment to be corrected based on the peak value; The DC bias adjustment value and the anolyte flow rate compensation coefficient in the segment to be corrected are replaced with the temporary DC bias adjustment value and the temporary anolyte flow rate compensation coefficient, respectively, to obtain the updated fluctuation calibration curve.

8. The electroplating DC fluctuation detection and calibration method according to claim 1, characterized in that, The specific steps of mapping the fluctuation calibration curve to the anolyte circulation control unit and the rectifier regulation terminal include: The DC bias adjustment value in the fluctuation calibration curve is encoded into a current regulation command for the rectifier, and the current regulation command is sent to the rectifier regulation terminal. The anolyte flow rate compensation coefficient in the fluctuation calibration curve is encoded as a speed adjustment command for the circulation pump, and the speed adjustment command is sent to the anolyte circulation control unit. While the rectifier regulating terminal adjusts the output current according to the current regulating command, the anolyte circulation control unit synchronously adjusts the anolyte circulation flow rate according to the speed regulating command.

9. The electroplating DC fluctuation detection and calibration method according to claim 7, characterized in that, The steps of generating the temporary DC bias adjustment value and the temporary anolyte flow rate compensation coefficient within the segment to be corrected based on the peak value specifically include: Calculate the first difference between the peak value and the DC bias adjustment value at the beginning of the segment to be corrected, and divide the first difference by the time length of the segment to be corrected to obtain the unit time bias adjustment step size; According to the unit time offset adjustment step size, the DC offset adjustment value is increased at each sampling time in the segment to be corrected to generate the temporary DC offset adjustment value; Calculate the second difference between the peak value and the anolyte flow rate compensation coefficient at the beginning of the segment to be corrected, and divide the second difference by the time length of the segment to be corrected to obtain the flow rate compensation step size per unit time. According to the unit time flow rate compensation step size, the anolyte flow rate compensation coefficient is increased at each sampling time within the segment to be corrected to generate the temporary anolyte flow rate compensation coefficient.

10. An electroplating DC fluctuation detection and calibration system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the electroplating DC fluctuation detection and calibration method as described in any one of claims 1 to 9.

Images