New energy profile anodic oxidation film thickness closed loop control method and anodic oxidation line

Abstract

The application is suitable for the technical field of anodic oxidation, and particularly relates to a new energy profile anodic oxidation film thickness closed-loop control method and an anodic oxidation line. The method comprises the following steps: only using the voltage / current signals output by the rectifier power supply to extract the average rising slope in the initial observation window, so that the online quantitative characterization of the film thickness growth rate is realized without adding any sensors or hardware; the batch difference of the measurable process parameters is eliminated through temperature and surface area double-factor compensation; the residual oxidation time is dynamically adjusted through the current efficiency correction coefficient; the second end point prediction channel independent and complementary to the initial slope method is formed through the slot voltage waveform shape analysis and the matching with the historical batch data, the fusion judgment of the two channels takes into account the rapid response and long-term experience correction; and finally, the online closed-loop control of the anodic oxidation film thickness in batches is realized without changing the existing production line hardware configuration, and the film thickness consistency and the production line intelligent level are significantly improved.

Description

Technical Field

[0001] This application belongs to the field of anodizing technology, and particularly relates to a closed-loop control method for the thickness of anodized film on new energy profiles and an anodizing line. Background Technology

[0002] The thickness of the aluminum alloy anodized film on new energy profiles (such as power battery trays, battery pack housings, and electric drive housings) directly affects the corrosion resistance, insulation performance, and service life of the components. Taking power battery trays as an example, the required oxide film thickness is typically 10–15 μm, with a uniformity tolerance of ±1.5 μm. Currently, the industry commonly uses the fixed ampere-hour method (i.e., setting the current density × fixed oxidation time), but this method faces a long-neglected problem: as the Al³⁺ concentration in the bath accumulates from less than 1 g / L in a new bath to over 15 g / L, the current efficiency decreases from approximately 85% to approximately 60%, and the film thickness deviation between different batches under the same set of process parameters can reach ±30%. The factory's countermeasure is "over-oxidation," setting the time according to the worst-case scenario to ensure that the standard is met even when the efficiency is at its lowest, resulting in excessive film thickness and 15–25% energy waste under normal operating conditions. The voltage / current output waveform of the rectifier power supply itself continuously encodes information about the entire film growth process, but this data has never been used for real-time endpoint determination.

[0003] In existing technologies, while offline eddy current thickness measurement offers high accuracy under dry conditions, it cannot be performed online in electrolytes, essentially representing batch-to-batch post-feedback rather than intra-batch closed-loop control. Indirect monitoring of electrolyte conductivity is affected by multiple factors such as temperature and alloy composition; a single conductivity index cannot be quantitatively converted into a film thickness correction value, and the conductivity electrode experiences drift issues after long-term immersion in strong acid. Online estimation methods based on electrochemical impedance spectroscopy require additional frequency response analysis hardware, resulting in high equipment costs and complex installation. Detection methods based on machine vision or optical sensors require additional hardware such as industrial cameras and waterproof optical windows, and are significantly affected by electrolyte bubbles and the complexity of the profile cross-section.

[0004] In summary, existing technologies are limited by the open-loop mode of post-processing inspection and batch adjustment, or by the cost barriers of adding sensors / analysis hardware. This results in the long-term reliance on operational experience and over-oxidation strategies for controlling the thickness of anodized films. This leads to poor consistency in film thickness between batches, significant energy waste, and the inability to achieve automated closed-loop endpoint determination within a single batch. This severely restricts the improvement of yield and intelligent upgrading of anodized production lines for new energy profiles. Summary of the Invention

[0005] This application provides a closed-loop control method for the thickness of anodized film on new energy profiles and an anodizing line, which can solve the problems of existing technologies that rely on offline post-detection or open-loop fixed ampere-hour method, which cannot achieve closed-loop control within batches, resulting in poor film thickness consistency and energy waste.

[0006] In a first aspect, embodiments of this application provide a closed-loop control method for the thickness of the anodic oxide film on new energy profiles, including: Within a preset initial observation window after constant current anodizing is energized, the cell voltage and cell current values ​​output by the rectifier power supply are continuously collected at a preset sampling period. The average rising slope of the cell voltage within the initial observation window is calculated based on the cell voltage values. Obtain the current bath temperature and the current total surface area of ​​the profile, and compensate the average rise slope based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average rise slope; The ratio of the compensated average rising slope to the pre-calibrated baseline rising slope is determined as the current efficiency correction coefficient for the current batch. Based on the current efficiency correction coefficient, the remaining oxidation time required for the current batch to reach the target film thickness is dynamically calculated. Within each preset control cycle, the continuously collected cell voltage values ​​are analyzed for waveform shape. Based on the matching results of the waveform shape characteristics of the current batch and the historical batch data, a recommended endpoint time is obtained to help determine the final value of the remaining oxidation time. The recommended endpoint time and the time when the remaining oxidation time is reduced to zero are merged and determined, and the endpoint signal is triggered at the merged endpoint time.

[0007] The technical solutions described in this application embodiment have at least the following technical effects: The closed-loop control method for anodized film thickness of new energy profiles provided in this application extracts the average rising slope within the initial observation window using only the voltage / current signal output by the rectifier power supply itself, achieving online quantitative characterization of film thickness growth rate without adding any sensors or hardware. It eliminates batch-to-batch differences in measurable process parameters through dual-factor compensation of temperature and surface area, ensuring that the compensated slope only reflects the net impact of non-measurable factors such as bath aging on current efficiency. It dynamically adjusts the remaining oxidation time using a current efficiency correction coefficient, automatically following changes in the bath state to solve the film thickness drift problem of the fixed ampere-hour method at different bath life stages. It forms a second endpoint prediction channel that is independent and complementary to the initial slope method by analyzing the shape of the bath voltage waveform and matching it with historical batch data; the fusion of these two channels balances rapid response and long-term experience correction. Finally, without altering the existing production line hardware configuration, it achieves online closed-loop control of anodized film thickness within a batch, significantly improving film thickness consistency and production line intelligence.

[0008] Secondly, embodiments of this application provide a closed-loop control system for the thickness of anodized film on new energy profiles, comprising: The acquisition unit is used to continuously acquire the cell voltage and cell current values ​​output by the rectifier power supply within a preset initial observation window after the constant current anodizing is energized, at a preset sampling period, and calculate the average rising slope of the cell voltage within the initial observation window based on the cell voltage value. The compensation unit is used to obtain the current bath temperature and the current total surface area of ​​the profile, and to compensate the average rise slope based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average rise slope. The calculation unit is used to determine the ratio of the compensated average rising slope to the pre-calibrated reference rising slope as the current efficiency correction coefficient for the current batch, and dynamically calculate the remaining oxidation time required for the current batch to reach the target film thickness based on the current efficiency correction coefficient. The recommendation unit is used to perform waveform shape analysis on the continuously collected cell voltage value within each preset control cycle, and obtain a recommended endpoint time to assist in determining the final value of the remaining oxidation time based on the matching results of the waveform shape characteristics of the current batch and the historical batch data. The result unit is used to merge and determine the recommended endpoint time with the time when the remaining oxidation time is zero, and to trigger the endpoint signal at the merged endpoint time.

[0009] Thirdly, embodiments of this application provide an anodizing line, including a control system and a production system. The control system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method as described in any of the foregoing aspects.

[0010] Fourthly, embodiments of this application provide a computer program product that, when run on an anodizing line, causes the anodizing line to perform the method described in any of the above aspects.

[0011] It is understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant descriptions in the above aspects, and will not be repeated here. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic flowchart of a closed-loop control method for the thickness of anodized film on new energy profiles provided in an embodiment of this application; Figure 2This is a schematic diagram of the operation of a closed-loop control method for the thickness of anodized film on new energy profiles provided in an embodiment of this application; Figure 3 This is a parameter curve diagram of a closed-loop control method for the thickness of anodized film on new energy profiles provided in an embodiment of this application; Figure 4 This is a schematic diagram of the closed-loop control system for the thickness of the anodized film on new energy profiles provided in one embodiment of this application; Figure 5 This is a schematic diagram of the structure of an anodized line provided in an embodiment of this application. Detailed Implementation

[0014] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0015] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0016] The thickness of the aluminum alloy anodized film on new energy profiles (such as power battery trays, battery pack housings, and electric drive housings) directly affects the corrosion resistance, insulation performance, and service life of the components. Taking power battery trays as an example, the required oxide film thickness is typically 10–15 μm, with a uniformity tolerance of ±1.5 μm. Currently, the industry commonly uses the fixed ampere-hour method (i.e., setting the current density × fixed oxidation time), but this method faces a long-neglected problem: as the Al³⁺ concentration in the bath accumulates from less than 1 g / L in a new bath to over 15 g / L, the current efficiency decreases from approximately 85% to approximately 60%, and the film thickness deviation between different batches under the same set of process parameters can reach ±30%. The factory's countermeasure is "over-oxidation," setting the time according to the worst-case scenario to ensure that the standard is met even when the efficiency is at its lowest, resulting in excessive film thickness and 15–25% energy waste under normal operating conditions. The voltage / current output waveform of the rectifier power supply itself continuously encodes information about the entire film growth process, but this data has never been used for real-time endpoint determination.

[0017] In existing technologies, while offline eddy current thickness measurement offers high accuracy under dry conditions, it cannot be performed online in electrolytes, essentially representing batch-to-batch post-feedback rather than intra-batch closed-loop control. Indirect monitoring of electrolyte conductivity is affected by multiple factors such as temperature and alloy composition; a single conductivity index cannot be quantitatively converted into a film thickness correction value, and the conductivity electrode experiences drift issues after long-term immersion in strong acid. Online estimation methods based on electrochemical impedance spectroscopy require additional frequency response analysis hardware, resulting in high equipment costs and complex installation. Detection methods based on machine vision or optical sensors require additional hardware such as industrial cameras and waterproof optical windows, and are significantly affected by electrolyte bubbles and the complexity of the profile cross-section.

[0018] In summary, existing technologies are limited by the open-loop mode of post-processing inspection and batch adjustment, or by the cost barriers of adding sensors / analysis hardware. This results in the long-term reliance on operational experience and over-oxidation strategies for controlling the thickness of anodized films. This leads to poor consistency in film thickness between batches, significant energy waste, and the inability to achieve automated closed-loop endpoint determination within a single batch. This severely restricts the improvement of yield and intelligent upgrading of anodized production lines for new energy profiles.

[0019] To address the aforementioned issues, this application provides a closed-loop control method for anodized film thickness in new energy profiles. This method extracts the average rising slope within the initial observation window using only the voltage / current signal output from the rectifier power supply itself, achieving online quantitative characterization of film thickness growth rate without adding any sensors or hardware. It eliminates batch-to-batch differences in measurable process parameters through dual-factor compensation of temperature and surface area, ensuring that the compensated slope only reflects the net impact of non-measurable factors such as bath aging on current efficiency. The remaining oxidation time is dynamically adjusted using a current efficiency correction coefficient, allowing the oxidation endpoint to automatically follow changes in the bath state, thus solving the film thickness drift problem of the fixed ampere-hour method at different bath life stages. A second endpoint prediction channel, independent and complementary to the initial slope method, is formed by analyzing the bath voltage waveform shape and matching it with historical batch data. The fusion of these two channels balances rapid response and long-term empirical correction. Finally, without altering the existing production line hardware configuration, online closed-loop control of anodized film thickness within a batch is achieved, significantly improving film thickness consistency and production line intelligence.

[0020] The closed-loop control method for the thickness of the anodic oxide film of new energy profiles provided in this application embodiment can be applied to the anodic oxide line. In this case, the anodic oxide line is the main body for executing the closed-loop control method for the thickness of the anodic oxide film of new energy profiles provided in this application embodiment. This application embodiment does not impose any restrictions on the specific type of anodic oxide line.

[0021] For example, an anodizing line can be various types of horizontal or vertical aluminum alloy profile anodizing production lines. Specifically, an anodizing line can include a production system and a control system. The production system is used to perform the anodizing function on the aluminum alloy profiles, and the control system is used to control the operation of the production system. The production system can include an anodizing tank, a sulfuric acid electrolyte circulation system, a tank solution heat exchange and temperature control system, a rectifier power supply, conductive beams and fixtures, and cathode plate assemblies. The anodizing tank is a rectangular acid-resistant tank containing sulfuric acid electrolyte, and the tank body is made of polypropylene or stainless steel lined with acid-resistant rubber. The sulfuric acid electrolyte circulation system is connected to the inlet and outlet of the anodizing tank through an acid-resistant circulation pump and pipelines, used to create directional flow of the electrolyte in the tank to maintain uniform temperature and concentration. The tank solution heat exchange and temperature control system includes a heat exchanger and a temperature controller. The heat exchanger is connected in series in the circulation pipeline of the sulfuric acid electrolyte circulation system. An online temperature sensor for the tank solution is installed at the outlet of the tank solution circulation pipeline. The temperature controller adjusts the cooling water flow rate of the heat exchanger according to the real-time reading of the temperature sensor to maintain the tank solution temperature within a preset range. The rectifier power supply is installed in a power cabinet next to the anodizing tank. The positive output terminal of the rectifier power supply is electrically connected to the conductive beam above the anodizing tank via a DC bus, and the negative output terminal is electrically connected to the cathode plate assemblies on both sides of the inner wall of the anodizing tank via a DC bus. The conductive beam is erected above the anodizing tank and has multiple mounting interfaces. Each mounting interface suspends a titanium alloy or aluminum alloy fixture containing the profile to be oxidized. After the profile is electrically connected to the conductive beam via the fixture, it acts as the anode and is immersed in the electrolyte. The cathode plate assemblies consist of multiple parallel lead or stainless steel plates, fixedly installed on both sides of the inner wall of the anodizing tank, maintaining a preset electrode spacing with the profile anode.

[0022] The control system may include a PLC or industrial computer, communication equipment, and storage devices. The PLC or industrial computer integrates a processor and memory, which are electrically connected via an internal bus. The memory stores the computer program. The PLC or industrial computer is installed in a control cabinet next to the anodizing line. It connects to the rectifier power supply via an RS-485 communication interface or analog input module to read the real-time tank voltage and current values ​​output by the rectifier power supply; it connects to the online temperature sensor of the tank solution via the analog input module to read the real-time tank solution temperature; and it connects to the start / stop control terminal of the rectifier power supply via a digital output module to send a stop output command to the rectifier power supply at the end of the cycle. The communication equipment may be an industrial Ethernet switch or a wireless communication module, connecting the PLC or industrial computer to a host computer or MES system for data exchange. The storage device may be an external solid-state drive connected to the PLC or industrial computer or a hard drive built into the host computer, communicating with the processor to store the computer program and complete process records for each batch.

[0023] To better understand the closed-loop control method for the thickness of the anodic oxide film on new energy profiles provided in this application, the specific implementation process of the closed-loop control method for the thickness of the anodic oxide film on new energy profiles provided in this application will be described exemplarily below.

[0024] Figure 1 A schematic flowchart of the closed-loop control method for the thickness of the anodized film on new energy profiles provided in this application embodiment is shown. Figure 2 This illustration shows an operational schematic of a closed-loop control method for the thickness of anodized film on new energy profiles provided in an embodiment of this application. The method includes: S100 continuously collects the cell voltage and cell current values ​​output by the rectifier power supply within a preset initial observation window after the constant current anodizing is energized, and calculates the average rise slope of the cell voltage within the initial observation window based on the cell voltage value.

[0025] It can be understood that constant current anodizing refers to the process where a rectifier power supply outputs current in constant current mode, causing an aluminum alloy profile to undergo an anodizing reaction in a sulfuric acid electrolyte to form an aluminum oxide film. In constant current mode, as the oxide film thickness increases, the resistance of the oxide film continuously increases. To maintain a constant current, the rectifier power supply automatically increases the output voltage. Therefore, the cell voltage monotonically increases with time, and the trajectory of the cell voltage increase directly reflects the growth process of the oxide film. The preset initial observation window refers to a preset fixed time interval calculated from the moment of constant current energization. Please refer to [link to relevant documentation]. Figure 3This time interval is specifically used to collect and analyze the changes in tank voltage during the initial oxidation phase. The duration of the preset initial observation window was determined as follows: Five standard current density anodizing experiments were conducted in each of the three conditions: new tank solution (Al³⁺ concentration less than 1 g / L), intermediate tank solution (Al³⁺ concentration 8 to 12 g / L), and aging tank solution (Al³⁺ concentration 15 to 20 g / L). The entire tank voltage curve was recorded. For each condition, the transition time from the rapid rise phase to the linear slow rise phase of the voltage curve was manually marked. The minimum value of the transition time from the 15 experiments was rounded down to 80% and used as the preset initial observation window duration. Following the above calibration process, the window duration was determined to be 3 minutes. The determination of the preset sampling period needs to consider both the statistical accuracy of the slope calculation and the data processing load of the host computer or PLC. After comparing the standard deviation of the slope calculation and the PLC CPU utilization rate within a 3-minute window under three sampling periods of 0.5 seconds, 1 second, and 2 seconds, a 1-second sampling period was selected as the optimal sampling period. Under this period, the standard deviation of the slope is less than 0.05V / min, and the PLC CPU utilization rate is less than 15%. The slot voltage and slot current values ​​output by the rectifier power supply are transmitted to the host computer or PLC via the Modbus RTU communication protocol or a 4–20mA analog interface. After receiving the data for each sampling period, the host computer or PLC adds a timestamp and stores it in a circular buffer. The average rising slope is calculated by reading the slot voltage value at the end of the initial observation window and the slot voltage value at the beginning of the window in the buffer. The voltage increment is obtained by subtracting the slot voltage value at the beginning of the window from the slot voltage value at the end of the window. The average rising slope is obtained by dividing the voltage increment by 3 minutes, with the unit being V / min. The average rising slope is calculated using the formula k_raw = (V_end − V_start) / 3. This average rising slope is an online measurable surrogate indicator of the film thickness growth rate. A larger average rising slope indicates faster film thickness growth and higher current efficiency under the current operating conditions, and the subsequent oxidation time should be shortened accordingly. Conversely, a smaller average rising slope indicates slower film thickness growth and lower current efficiency, and the subsequent oxidation time should be extended accordingly.

[0026] For example, the rectifier power supply adopts a three-phase thyristor rectifier or a high-frequency switching power supply, with a rated output current of not less than 1000A, and has a constant current control mode with a current control accuracy better than ±1%. The host computer can adopt an industrial-grade embedded motherboard, equipped with a Linux or Windows Embedded operating system, and configured with data acquisition software and storage modules. The PLC adopts a Siemens S7-1200 series or Mitsubishi FX5U series, configured with an analog input module for reading 4–20mA signals of slot voltage and slot current, or configured with an RS-485 communication module for data reading using the Modbus RTU protocol. The memory length of the annular buffer should at least cover all sampling points within the preset initial observation window, i.e., at least 180 sampling data pairs (3-minute window, 1-second sampling period).

[0027] It should be noted that the existing fixed ampere-hour method only uses current density and oxidation time as control parameters, completely ignoring the film thickness growth information contained in the tank voltage waveform. While some solutions use online detection equipment such as eddy current thickness gauges or laser thickness gauges for film thickness feedback, these devices require additional hardware investment and installation modifications. The eddy current method requires a dry surface and cannot be used for online measurement in the electrolyte; the laser method is severely affected by bubbles and turbidity in the electrolyte, resulting in insufficient measurement stability. This application utilizes the voltage / current signals that the rectifier power supply itself must output. These two signals are standard on any anodizing production line and do not require any additional sensors. The initial voltage rise slope is extracted from these signals, transforming the previously unused electrical signals into feedback for film thickness control. This achieves closed-loop film thickness control with zero additional hardware, avoiding the technical bottlenecks and cost barriers of online thickness sensors.

[0028] In one possible implementation, in step S100, the slot voltage and slot current values ​​output by the rectified power supply are continuously acquired at a preset sampling period, and the average rise slope of the slot voltage within the initial observation window is calculated based on the slot voltage values, including: S110: Collect the tank voltage and tank current values ​​at each sampling time according to the preset sampling period, and mark the sampling points that deviate from the target current value by more than the preset current fluctuation threshold as invalid sampling points and discard them.

[0029] It can be understood that the target current value refers to the constant current setting value calculated based on the preset current density and the total surface area of ​​the current batch of profiles. The calculation formula is I_target=J0×S, where J0 is the preset current density (taken as 1.5A / dm²), and S is the total surface area of ​​the current batch of profiles. The preset current fluctuation threshold is determined as follows: when the rectifier power supply is running stably in constant current mode, the instantaneous value of the tank current is continuously collected for 5 minutes (sampling period of 1 second, a total of 300 sampling points). The standard deviation of the 300 sampling points is calculated, and 3 times the standard deviation is taken as the allowable range of current fluctuation. After performing the above calibration test on multiple thyristor rectifier power supplies and high-frequency switching power supplies of different models, the range of 3 times the standard deviation is between ±3% and ±5% of the target current value. To ensure the universality and robustness of the control method, 5% is uniformly taken as the preset current fluctuation threshold. At each sampling moment, the host computer or PLC reads the instantaneous value of the tank current and calculates the percentage deviation of this instantaneous value from the target current value. When the deviation percentage is greater than 5%, the corresponding tank voltage and tank current values ​​at that sampling moment are simultaneously marked as invalid and removed from the valid sampling point sequence. The reason for removing invalid sampling points is that when the output current of the rectifier power supply deviates instantaneously due to grid voltage fluctuations, load switching, or rectifier component commutation, the tank voltage value at that moment no longer accurately reflects the true electrochemical process of film thickness growth. When the current is too high, the tank voltage reading is artificially inflated; when the current is too low, the tank voltage reading is depressed. If these abnormal data points are not removed, the calculation result of the average rising slope will be skewed, thus affecting the accuracy of the subsequent current efficiency correction coefficient.

[0030] S120: Calculate the proportion of invalid sampling points that have been removed to the total number of sampling points in the initial observation window. When the proportion exceeds the preset proportion threshold, extend the initial observation window until the number of valid sampling points reaches the preset valid number threshold.

[0031] It is understandable that the preset proportion threshold refers to the upper limit of the ratio of invalid sampling points to the theoretical total number of sampling points within the initial observation window. Exceeding this upper limit means that there is a severe shortage of valid data within the initial observation window, and the confidence level of the voltage rise slope calculated based on the remaining valid data decreases significantly. The preset proportion threshold is determined by continuously recording the proportion of invalid sampling points in 20 batches under normal operating conditions (no abnormalities in the rectifier power supply and stable grid voltage), calculating the mean and standard deviation of the invalid sampling proportions for the 20 batches, and taking the mean plus twice the standard deviation as the proportion threshold. After recording 20 batches of data on each of three different anodizing production lines, the mean was between 6% and 10%, the standard deviation was between 3% and 5%, and the upper limit of the mean plus twice the standard deviation was between 12% and 20%. To accommodate the differences in rectifier power supply quality across different production lines and to allow for a safety margin, a uniform 20% is taken as the preset proportion threshold. The preset effective number threshold refers to the minimum number of valid sampling points that must be met within the initial observation window. The calculation formula is N_valid_min = N_total × (1 − 20%), where N_total is the theoretical total number of sampling points (3 minutes × 60 seconds / minute ÷ 1 second / time = 180). Therefore, N_valid_min = 180 × 0.80 = 144. After the initial observation window timer expires, the host computer or PLC automatically counts the number of invalid sampling points and calculates the invalidity ratio. If the invalidity ratio does not exceed 20% and the number of valid sampling points reaches or exceeds 144, the subsequent slope calculation step is initiated. If the invalidity ratio exceeds 20%, the PLC continues to extend the sampling period by 1 second, recalculating the cumulative number of valid sampling points for each newly added valid sampling point. When the cumulative number of valid sampling points first reaches 144, the extended sampling stops. If the extended total duration exceeds 6 minutes (i.e., twice the original 3-minute window), and the cumulative number of valid sampling points still does not reach 144, the PLC triggers an alarm signal for abnormal rectifier power supply, and simultaneously prompts the operator on the HMI (Human Machine Interface) to check the operating status of the rectifier power supply and the communication line.

[0032] S200: Obtain the current bath temperature and the current total surface area of ​​the profile. Compensate the average rise slope based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average rise slope.

[0033] It can be understood that the bath temperature refers to the real-time temperature of the sulfuric acid electrolyte in the anodizing bath. This temperature directly affects the dynamic balance between the growth rate and dissolution rate of the oxide film. The higher the temperature, the faster the chemical dissolution rate of the oxide film by the electrolyte, the lower the net film formation rate, and the thinner the film thickness generated under the same current density and oxidation time. The bath temperature value is read from the PT100 platinum resistance temperature sensor already configured on the production line. This sensor is installed at the outlet of the bath circulation pipeline or in the middle of the bath, with a measurement accuracy of ±0.5°C. The sensor transmits the temperature value to the host computer or PLC via a 4–20mA analog signal or RS-485 digital communication. The total surface area of ​​the profiles refers to the sum of the effective conductive areas of all profiles hung in the anodizing bath in this batch. This value is provided by the production work order, which clearly records the unfolded surface area of ​​a single profile (calculated from the profile cross-section CAD model) and the number of profiles hung in this batch. The total surface area is obtained by multiplying the two, and the unit is square meters. Changes in total surface area affect the rate of rise of the cell voltage. In constant current mode, a larger total surface area results in a lower current density distributed per unit area, leading to a slower rate of voltage rise. Therefore, the original average rise slopes of batches with different surface areas are not directly comparable and must be eliminated through surface area normalization. The compensated average rise slope is the result of normalizing the original average rise slope using both temperature and surface area factors. The batch-to-batch differences in the compensated slope only reflect the combined effect of factors such as the degree of aging of the bath solution and differences in alloy composition, which cannot be directly measured, on current efficiency. The formula for calculating the compensated average rise slope is k_comp = k_raw × f_T × f_S, where k_raw is the original average rise slope, f_T is the temperature compensation coefficient, f_S is the surface area normalization coefficient, and k_comp is the compensated average rise slope, all in V / min.

[0034] For example, the typical operating range of the bath temperature is 15°C to 25°C. The total surface area of ​​the profile is obtained by the host computer or PLC reading the individual surface area and the quantity of material hanging in the production order through the OPCUA interface or the MES database query interface, and multiplying the two to obtain the total surface area.

[0035] It should be noted that the existing ampere-hour integration method completely ignores the impact of temperature and surface area changes on the voltage slope. Temperature fluctuations of ±3°C or changes in material load of ±20% can lead to a slope deviation of 15–25%. Operators typically do not adjust process parameters separately for these factors because there is a lack of quantitative correction models. This application incorporates temperature and surface area—two measurable parameters directly provided by existing sensors and work order systems on the production line—into the slope compensation logic. Without adding any hardware or manual input, it establishes a quantitative coupling relationship between existing but never-connected process data and the rectifier power supply electrical signal, eliminating the influence of the two main known sources of interference on film thickness control.

[0036] In one possible implementation, S200, the current bath temperature and the current total surface area of ​​the profile are obtained, and the average rise slope is compensated based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average rise slope, including: S210, obtain the current bath temperature and the current total surface area of ​​the profile, and calculate the difference between the reciprocal of the absolute temperature value corresponding to the current bath temperature and the reciprocal of the absolute temperature value corresponding to the pre-calibrated reference temperature.

[0037] It is understandable that the absolute temperature value refers to the temperature value in Kelvin, which is converted from the Celsius temperature value by adding 273.15. The reason for using the absolute temperature value is that the chemical reaction rate constant and the reciprocal of the absolute temperature follow an Arrhenius exponential relationship, rather than a linear relationship with the Celsius temperature. When the temperature rises from 15°C to 25°C, the Celsius temperature scale shows an increase of 10°C, but the reciprocal of the absolute temperature decreases from about 0.00347 Kelvin to about 0.00336 Kelvin. This tiny difference in reciprocal, multiplied by the ratio of the activation energy gas constant, is exponentially amplified to produce a temperature compensation coefficient range of about 1.4 to 0.7 times, accurately reproducing the amplitude of the effect of temperature on the film formation rate. The pre-calibrated reference temperature is the standard bath temperature used when calibrating the reference slope. Based on historical operating temperature data from three production lines across different seasons throughout the year, the bath temperature is mainly stable between 19°C and 21°C during the air-conditioned season. The temperature control system exhibits the smallest fluctuation range (within ±0.3°C) at 20°C; therefore, 20°C was selected as the reference temperature. The absolute temperature corresponding to the reference temperature is 293.15 K. Let the current bath temperature be T, the current absolute temperature be T plus 273.15 K, and the reference absolute temperature be 293.15 K. The reciprocal of this difference is 1 divided by the current absolute temperature minus 1 divided by 293.15. The sign of this difference indicates the direction of deviation of the current temperature from the reference temperature. When the current temperature is higher than the reference temperature, the difference is negative, and the temperature compensation coefficient should be less than 1; when the current temperature is lower than the reference temperature, the difference is positive, and the temperature compensation coefficient should be greater than 1.

[0038] S220 determines the temperature compensation coefficient based on the current bath temperature and the pre-calibrated reference temperature.

[0039] The temperature compensation coefficient is understood to be a dimensionless multiplier used to convert the average rise slope under the current bath temperature conditions to an equivalent value under the reference temperature conditions. The rate constants of aluminum ion oxidation and alumina deposition during anodizing both follow the Arrhenius equation, meaning the rate constant is proportional to an exponential function whose exponent is the negative apparent activation energy divided by the product of the ideal gas constant and the absolute temperature. Under constant current conditions, the film formation rate is proportional to the product of current density and current efficiency. Current efficiency is significantly affected by temperature; the higher the temperature, the more intense the oxygen evolution side reaction, and the lower the proportion of effective current used for film formation. This temperature dependence also follows an Arrhenius-type exponential law. Therefore, the basic form of the temperature compensation coefficient is an exponential function: f_T = exp[(Ea / R) × (1 / T − 1 / T_ref)], where Ea / R is the ratio of activation energy to the gas constant. When the current temperature equals the reference temperature, the exponent term is zero and the temperature compensation coefficient is 1; when the current temperature is higher than the reference temperature, the exponent term is negative and the temperature compensation coefficient is less than 1; when the current temperature is lower than the reference temperature, the exponent term is positive and the temperature compensation coefficient is greater than 1.

[0040] Optionally, S220 determines a temperature compensation coefficient based on the current bath temperature and a pre-calibrated reference temperature, including: S221, calculate the difference between the reciprocal of the absolute temperature value corresponding to the current tank temperature and the reciprocal of the absolute temperature value corresponding to the reference temperature.

[0041] It can be understood that this difference, in the Arrhenius coordinate system (with the reciprocal of absolute temperature as the horizontal axis), physically represents the horizontal distance between the current temperature point and the reference temperature point. For a specific numerical example: if the current bath temperature is 25°C, the absolute temperature is 298.15 K, and its reciprocal (1 divided by 298.15) is approximately 0.003355 Kelvin; the reference temperature of 20°C corresponds to an absolute temperature of 293.15 K, and its reciprocal (1 divided by 293.15) is approximately 0.003411 Kelvin. The difference between the two is 0.003355 minus 0.003411, which equals -5.6 multiplied by 10 to the power of -5 per Kelvin. If the current bath temperature is 15°C, the absolute temperature is 288.15 K. Its reciprocal, 1 divided by 288.15, is approximately 0.003470 Kelvin. The difference between this and the reciprocal of the reference value is 0.003470 minus 0.003411, which equals +5.9 multiplied by 10 to the power of -5 Kelvin. In the host computer or PLC, this difference is stored and calculated as a double-precision floating-point number.

[0042] S222, the difference is multiplied by the ratio of the pre-calibrated activation energy gas constant to obtain the exponent, and the exponential function value is calculated with the natural constant e as the base to obtain the temperature compensation coefficient.

[0043] The activation energy-to-gas constant ratio is understood to be the ratio of the apparent activation energy to the ideal gas constant. This ratio reflects the sensitivity of the anodic oxidation film formation process to temperature changes; the larger the ratio, the more drastic the fluctuation in the film formation rate caused by small temperature changes. The calibration method for the activation energy-to-gas constant ratio is as follows: In a freshly prepared bath solution with a sulfuric acid concentration of 180 g / L and an Al³⁺ concentration below 1 g / L, three constant-current anodic oxidation experiments were conducted at three temperature points: 15°C, 20°C, and 25°C. The experimental conditions for each experiment were: current density 1.5 A / dm², profile surface area 1.0 m², and oxidation time equal to the standard total oxidation time. After each experiment, the final film thickness was measured using an eddy current thickness gauge. Using the reciprocals of the absolute temperatures corresponding to the three temperature points as the x-axis and the natural logarithm of the ratio of the final film thickness value of each experiment to the mean film thickness of the three experiments at the reference temperature (20°C) as the y-axis, a least-squares linear fit was performed on the nine data points. The slope of the fitted line is the calibration value of the activation energy gas constant ratio. After performing the above standard calibration experiment (27 independent experiments) on 6063 alloy in 180 g / L sulfuric acid electrolyte, the calibration value of the activation energy gas constant ratio is approximately 5410 K, with a standard deviation of ±120 K. The complete formula for calculating the temperature compensation coefficient is: f_T=exp[5410×(1 / T−1 / 293.15)]. In a PLC, the exponential function with the natural constant e as the base can be calculated using a preset exponential function lookup table. 181 pre-calculated values ​​are stored in the PLC memory in array form, covering the range of 5410 multiplied by the reciprocal difference from -0.9 to +0.9 (step size 0.01). During operation, the actual calculated product is used as the index to look up the table. Linear interpolation is used between two adjacent step sizes in the table, and the interpolation error is less than 0.5%.

[0044] S230, determine the surface area normalization coefficient based on the ratio of the current total surface area of ​​the profile to the pre-calibrated total surface area of ​​the reference profile.

[0045] It is understandable that the surface area normalization coefficient is used to eliminate the influence of differences in the amount of material applied to profiles from different batches on the average rise rate. In constant current anodizing, the total current output by the rectified power supply is distributed across the entire effective anode area of ​​the profile. When the total current is fixed, the larger the total surface area of ​​the profile, the lower the actual current density distributed per unit area, resulting in a negative correlation between the tank voltage rise rate and the surface area. The total surface area of ​​the reference profile is the total surface area of ​​the standard profile used in the reference slope calibration experiment, and 1.0 m² is selected as the total surface area of ​​the reference profile. The formula for calculating the surface area normalization coefficient is f_S = S_ref / S_current, where S_ref is the total surface area of ​​the reference profile (1.0 m²), and S_current is the total surface area of ​​the current batch of profiles. A value greater than 1 indicates that the total surface area of ​​the current batch is less than the reference value, and the original slope that is too large needs to be adjusted downward; a value less than 1 indicates that the total surface area of ​​the current batch is greater than the reference value, and the original slope that is too small needs to be adjusted upward. The host computer or PLC reads the surface area of ​​a single piece and the quantity of material to be hung from the production work order, multiplies the two numbers to get S_current, and then performs a division operation to get f_S.

[0046] S240 is calculated based on the average rise slope, temperature compensation coefficient, and surface area normalization coefficient to obtain the compensated average rise slope.

[0047] It can be understood that the physical meaning of the compensated average rise slope is to normalize the original average rise slope obtained by the current batch under specific bath temperature and specific surface area of ​​the substrate to the equivalent value under the reference temperature (20°C) and reference surface area (1.0m²). This double normalization eliminates the known influence of measurable process parameters and retains the net influence of factors that cannot be directly measured on the slope. The formula for calculating the compensated average rise slope is k_comp=k_raw×f_T×f_S. For a specific numerical example: Under the conditions of a bath temperature of 23°C and a total surface area of ​​1.3m², the original average rise slope of a batch is 1.80V / min. The temperature compensation coefficient (at 23°C) is approximately 1.36, and the surface area normalization coefficient is 1.0 divided by 1.3, which is approximately 0.77. The compensated average rise slope is 1.80 multiplied by 1.36 multiplied by 0.77, which is approximately 1.88V / min. The compensated slope provides a unified basis for comparison in subsequent calculations of the current efficiency correction factor.

[0048] S300 determines the ratio of the compensated average rising slope to the pre-calibrated baseline rising slope as the current efficiency correction coefficient for the current batch, and dynamically calculates the remaining oxidation time required for the current batch to reach the target film thickness based on the current efficiency correction coefficient.

[0049] It is understood that the baseline rise slope is the average rise slope obtained under strict standard conditions: freshly prepared electrolyte (sulfuric acid concentration 180 g / L, Al³⁺ concentration less than 1 g / L), standard 6063 alloy profile (surface area 1.0 m²), baseline electrolyte temperature 20°C, and current density 1.5 A / dm². The calibration method is as follows: five independent constant current anodizing experiments were conducted under the above standard conditions. Each experiment followed the complete procedure of the initial observation window, and the average rise slope of each experiment was calculated. The arithmetic mean of the five experiments was taken as the baseline rise slope. After multiple calibrations, the baseline rise slope of 6063 alloy under these conditions is approximately 2.13 V / min. It is recommended that the baseline rise slope be calibrated once after each fresh electrolyte preparation, or periodically recalibrated every six months. The formula for calculating the current efficiency correction factor is η_corr = k_comp / k_ref. This factor reflects the current efficiency level of the current batch relative to the baseline operating conditions. When the equivalent conductivity decreases due to aging of the bath solution in the current batch, k_comp is lower than k_ref, and η_corr is less than 1, requiring an extension of the oxidation time. When k_comp is higher than k_ref, η_corr is greater than 1, requiring a shortening of the oxidation time. Dynamic calculation means that the remaining oxidation time is not a fixed value calculated once in the initial stage, but is continuously refreshed at preset control intervals (30 seconds) throughout the entire oxidation process.

[0050] For example, assuming a batch is placed at a bath temperature of 23°C and a total surface area of ​​1.3 m², the initial observation window calculates an original average rise rate of approximately 1.80 V / min. After temperature and surface area compensation, the compensated average rise rate is approximately 1.88 V / min. Dividing this compensated rise rate by the baseline rise rate of 2.13 V / min yields a current efficiency correction factor of approximately 0.88, which is less than 1. This indicates that the current efficiency of the current batch is lower than the baseline value, requiring a corresponding extension of the oxidation time. The host computer or PLC refreshes the calculated values ​​of the current efficiency correction factor and the remaining oxidation time every 30 seconds.

[0051] It should be noted that existing ampere-hour integration methods assume that the current efficiency is a constant value (usually taken as 85-90% of the theoretical value) and does not change with the state of the bath solution. However, in actual production, as the Al³⁺ concentration in the bath solution continuously increases from less than 1 g / L to more than 15 g / L, and the H₂SO₄ concentration decreases from 180 g / L to about 150 g / L, the current efficiency will gradually decrease from about 85% in the new bath solution to about 60% in the aged bath solution, with a deviation of more than 25 percentage points. This application quantifies this drift in real time and incorporates it into the control logic through a current efficiency correction coefficient, so that the oxidation time is automatically adjusted according to the changes in the state of the bath solution. This fundamentally solves the industry pain points of poor film thickness consistency at different life stages of the fixed ampere-hour method and the need to rely on excessive oxidation to maintain the lower limit.

[0052] In one possible implementation, step S300 involves dynamically calculating the remaining oxidation time required for the current batch to reach the target film thickness based on a current efficiency correction factor, including: S310, multiply the current efficiency correction factor, the pre-calibrated reference current efficiency, the pre-calibrated Faraday film formation factor, the nominal current density, and the cumulative oxidation time from the time of energization to the current time to obtain the estimated value of the generated film thickness.

[0053] It can be understood that the generated film thickness estimate is an online estimate of the average thickness of the alumina film layer formed on the profile surface from the start of energization to the current moment, in μm. The current efficiency correction factor is the ratio of the compensated average rise slope to the reference rise slope. The pre-calibrated reference current efficiency refers to the anodizing current efficiency calibrated experimentally under reference operating conditions. The calibration method is as follows: under reference operating conditions, five independent anodizing experiments are conducted with a fixed oxidation time. Each time, the film thickness is measured at 20 evenly distributed measuring points on the profile surface using an eddy current thickness gauge, and the average value is taken as the actual film thickness. At the same time, the theoretical film thickness is calculated according to Faraday's law. The reference current efficiency is equal to the average actual film thickness divided by the theoretical film thickness. The arithmetic mean of the five experiments is approximately 0.85. The pre-calibrated Faraday film-forming coefficient is the theoretical film-forming coefficient for aluminum oxide formation by anodizing. It is obtained by dividing the molar mass of aluminum oxide by the product of the electron transfer number, the Faraday constant, and the density of aluminum oxide, and then multiplying by a unit conversion factor. It is approximately 1.04 x 10⁻⁵ cubic centimeters per ampere-second. The nominal current density is taken as 1.5 A / dm². The cumulative oxidation time is calculated from the moment of energization, in seconds. The formula for calculating the estimated film thickness is d_elapsed = η_corr × η_ref × K × J0 × t_elapsed.

[0054] For example, under baseline conditions (η_corr equals 1.00), the film thickness per unit time is approximately 0.30 μm / min. If the current efficiency correction factor for the current batch drops to 0.70 (corresponding to a condition with severe bath aging), the film thickness per unit time decreases to approximately 0.21 μm / min. The host computer or PLC refreshes the calculation of the generated film thickness estimate every 30 seconds.

[0055] S320, subtract the generated film thickness estimate from the preset target film thickness to obtain the remaining required film thickness.

[0056] It is understandable that the target film thickness is a pre-set design thickness value for the oxide film based on product technical drawings or customer technical specifications. For aluminum alloy profiles (6063 alloy) of new energy power battery trays, the target film thickness is usually set at 12μm, with an allowable tolerance range of ±1.5μm (i.e., the acceptable range is 10.5μm to 13.5μm). The target film thickness is set based on two aspects: according to the GB / T8013 series specifications, the minimum requirement for the sulfuric acid anodic oxide film thickness of power battery trays is 10μm to meet the requirements of insulation strength above 60V and corrosion resistance performance of more than 96 hours of neutral salt spray test; at the same time, engineering practice shows that the brittleness of the oxide film layer increases significantly when the film thickness exceeds 15μm. Therefore, in engineering, 12μm is taken as the target value between the minimum requirement of 10μm and the excessive thickness risk threshold of 15μm. The target film thickness is preset in the process formula database of the host computer or PLC. The formula for calculating the remaining required film thickness is d_remaining=d_target−d_elapsed, in μm. If d_remaining is less than or equal to 0, the endpoint signal should be triggered immediately. If d_remaining is greater than 0, the remaining oxidation time should continue to be calculated.

[0057] S330, divide the remaining required film thickness by the product of the current efficiency correction factor, the reference current efficiency, the Faraday film formation factor, and the nominal current density to obtain the remaining oxidation time at the current moment.

[0058] The remaining oxidation time, in minutes, is the estimated time required to maintain constant current oxidation under the current operating conditions to achieve the target film thickness, starting from the current moment. The calculation of the remaining oxidation time is based on the fundamental electrochemical relationship that the film thickness growth rate is proportional to the effective charge flux. The film thickness per unit time at the current moment is equal to the product of the current efficiency correction factor, the reference current efficiency, the Faraday film formation factor, and the nominal current density. The formula for calculating the remaining oxidation time is t_remaining = d_remaining / (η_corr × η_ref × K × J0). The host computer or PLC refreshes the calculated value of the remaining oxidation time once every preset control cycle (30 seconds). When the remaining oxidation time is less than or equal to zero after a refresh, the host computer or PLC transmits the time when the remaining oxidation time reaches zero to the fusion determination module. The remaining oxidation time is displayed on the human-machine interface in countdown form.

[0059] S400 performs waveform shape analysis on the continuously collected cell voltage values ​​within each preset control cycle. Based on the matching results of the waveform shape characteristics of the current batch and the historical batch data, it obtains a recommended endpoint time to help determine the final value of the remaining oxidation time.

[0060] Waveform shape analysis can be understood as the process of identifying stages and extracting morphological features from the complete time series curve of the tank voltage from the moment of energization to the present moment. In the constant current anodizing process, the tank voltage curve exhibits a typical two-stage morphological characteristic. The first stage is the barrier layer formation stage, where the voltage rises rapidly and linearly, lasting from tens of seconds to two minutes. The second stage is the stable growth stage of the porous layer, where the voltage transitions to a slower linear rise, covering most of the oxidation time. The location and shape of the transition point between the two stages contain comprehensive information about the tank state, alloy characteristics, and film thickness progression. Waveform shape features refer to the normalized multi-feature vector extracted from the voltage curve. Historical batch data refers to the database of completed batches accumulated during long-term production line operation. Each record contains the normalized multi-feature vector of that batch and the actual total oxidation time verified by offline eddy current thickness measurement. The basic matching logic is: extract the current waveform feature vector within each control cycle of the current batch, retrieve the K most similar historical batches from the historical database, and calculate the recommended endpoint time based on the total oxidation time of these historical batches.

[0061] For example, the preset control cycle is set to 30 seconds. At the beginning of each control cycle, the host computer or PLC reads all currently accumulated slot voltage sampling data and performs stage segmentation and feature extraction operations. The recommended endpoint time is an absolute time value calculated from the moment of power-on, in minutes. The recommended channel based on complete waveform shape analysis and historical experience and the fast response channel based on the initial slope operate independently and complement each other.

[0062] It should be noted that while the method of calculating the remaining oxidation time based on the initial slope can quickly provide an endpoint estimate, the initial slope only utilizes voltage information from the first few minutes and cannot reflect anomalies in the later stages of oxidation. By continuously analyzing the complete voltage waveform shape and combining it with empirical data from historical batches, an endpoint prediction channel is formed that runs parallel to and complements the initial slope method. The core advantage of the waveform matching channel is that, with the continuous accumulation and incremental updates of the historical batch database, the recommendation accuracy will continuously improve with usage time, achieving an online learning effect that becomes more accurate with use. This effect is achieved without any machine learning models or neural networks, relying solely on similarity retrieval based on labeled historical production data.

[0063] In one possible implementation, S400 performs waveform shape analysis on the continuously acquired cell voltage values ​​within each preset control cycle. Based on the matching results between the waveform shape characteristics of the current batch and historical batch data, a recommended endpoint time is obtained to assist in determining the final value of the remaining oxidation time, including: S410 performs stage segmentation on the time curve corresponding to the continuously acquired slot voltage value within each preset control cycle, and extracts a normalized multi-feature vector that reflects the waveform shape of the current batch.

[0064] Stage segmentation can be understood as automatically determining the boundary between the barrier layer formation stage and the stable growth stage of the porous layer on the slot voltage-time curve using algorithms. Stage segmentation is the foundation of waveform shape analysis; only by accurately locating the transition point can the characteristic parameters of the first and second stages be extracted separately. Normalized multi-feature vector refers to a vector composed of a set of dimensionless values ​​obtained by dividing the extracted six waveform characteristic parameters (transition time, transition point voltage, average slope of the first stage, average slope of the second stage, characteristic resistance, and slope ratio) by their respective preset reference values. The purpose of normalization is to eliminate the dimensional differences between the six characteristic parameters, enabling characteristic parameters of different orders of magnitude to be fairly compared in the same Euclidean distance space. The normalized reference values ​​are obtained by taking the arithmetic mean after extracting the corresponding characteristic parameters through more than 5 calibration experiments under the reference operating conditions. The reference values ​​for the calibration transition time are approximately 90 seconds, the reference value for the transition point voltage is approximately 20V, the reference value for the average slope of the first stage is approximately 0.036V / s, the reference value for the average slope of the second stage is approximately 0.0071V / s, the reference value for the characteristic resistance is approximately 0.095Ω, and the reference value for the slope ratio is approximately 0.2.

[0065] For example, at the beginning of each preset control cycle (30 seconds), the host computer or PLC extracts all slot voltage sampling data from the power-on moment to the current moment and executes the sliding window stage segmentation algorithm. After the transition moment is successfully detected for the first time, the stage segmentation is not repeated in subsequent control cycles. Only the average slope of the second stage is re-extracted, because its value will be gradually updated due to factors such as temperature drift during the oxidation process.

[0066] Optionally, S410, within each preset control cycle, performs stage segmentation on the time curve corresponding to the continuously acquired slot voltage value, and extracts a normalized multi-feature vector reflecting the waveform shape of the current batch, including: S411, along the time curve of the slot voltage value, calculate the first average slope within a window length before the current sampling time and the second average slope within a window length after the current sampling time using a sliding window of preset duration. Determine the transition time and transition point voltage between the barrier layer formation stage and the porous layer stable growth stage based on the degree of difference between the first average slope and the second average slope.

[0067] It is understandable that the preset duration of the sliding window is set to 30 seconds. The selection of the window duration must simultaneously meet two conditions: the window must be short enough to ensure the timing accuracy of the stage segmentation, and the window must be long enough to contain sufficient sampling points (30 one-second sampling points within a 30-second window) to ensure statistical stability. After comparing and testing four window durations of 10 seconds, 20 seconds, 30 seconds, and 60 seconds, 30 seconds was determined to be the optimal window duration. The detection delay did not exceed 30 seconds, and no false detections occurred under both extreme conditions of new and aged tank solutions. At each sampling moment, the first average slope (the tank voltage increment within the 30-second window before that moment divided by 30 seconds) and the second average slope (the tank voltage increment within the 30-second window after that moment divided by 30 seconds) were calculated. The degree of difference was quantitatively measured by the slope change rate r: r equals the maximum value among the absolute value of the difference between the two slopes divided by the absolute value of the two slopes and the preset decimal value (0.001). When r first exceeds the preset slope change threshold and r exceeds the threshold for at least three consecutive sampling points, the current moment is determined to be a transition moment. The preset slope change threshold is calibrated as follows: the transition moment is visually determined by the process engineer in 5 benchmark operating condition experiments, and the corresponding r value is then calculated using a sliding window algorithm. The average of the 5 calculated values ​​is multiplied by 80% as the threshold, which is then calibrated and set to 0.3.

[0068] S412, based on the transition time and transition point voltage, extract the first stage average slope of the tank voltage before the transition time, the second stage average slope after the transition time, the characteristic resistance of the transition point voltage and the target current value, and the slope ratio of the second stage average slope to the first stage average slope.

[0069] It can be understood that the average slope k1 in the first stage is the result of dividing the total increase in tank voltage from the energizing moment to the transition moment by the transition moment, calculated as k1=(V_trans−V0) / t_trans. k1 is simultaneously affected by the conductivity of the bath, the surface area of ​​the profile, the alloy composition, and the kinetics of barrier layer formation. The average slope k2 in the second stage is the result of dividing the total increase in tank voltage from the transition moment to the current sampling moment by the corresponding duration, calculated as k2=(V_current−V_trans) / (t_current−t_trans). k2 mainly reflects the kinetics of stable growth of the porous layer. The characteristic resistance R_trans is the ratio of the transition point voltage to the target current value, calculated as R_trans=V_trans / I_target. The slope ratio ratio is the result of k2 divided by k1, reflecting the degree to which the oxide film changes from the barrier layer growth mode to the porous layer growth mode. The smaller the ratio, the more significant the difference between the two stages (usually corresponding to a fresher bath), and the larger the ratio, the smoother the transition (usually corresponding to a more aged bath).

[0070] S413, normalize the transition time, transition point voltage, first stage average slope, second stage average slope, characteristic resistance, and slope ratio by their respective preset reference values ​​to obtain normalized multi-feature vectors.

[0071] It can be understood that the normalized multi-feature vector is a 6-dimensional vector, where each dimension is equal to the corresponding original feature parameter divided by the baseline value of that dimension. The normalization formula is to divide each original feature parameter by its corresponding baseline value. The fluctuation of the normalized value of each dimension around 1.00 indicates the direction and degree of deviation of the current batch from the baseline condition in that dimension; a value greater than 1 indicates a deviation that is too high, and a value less than 1 indicates a deviation that is too low. After the host computer or PLC completes 6 division operations, it stores the 6-dimensional normalized values ​​into an array for subsequent Euclidean distance calculations.

[0072] S420 matches the normalized multi-feature vector with the corresponding feature parameters in the historical batch database, and determines the recommended endpoint time based on the matching results.

[0073] It is understandable that the historical batch database is a production record database gradually accumulated during the long-term operation of the production line, stored in an SQLite or MySQL database on the host computer. Each record represents a completed batch that has been verified offline. The archiving time is after the oxidation of the batch is completed and the film thickness is measured at 20 measuring points on the profile surface by an eddy current thickness gauge. If the average measured film thickness is within the tolerance range of the target film thickness, the record is marked as a qualified batch and stored in the valid reference record set. The archiving fields of each valid reference record include batch number, profile type, alloy grade, total surface area, average bath temperature, initial bath voltage value, normalized multi-feature vector (a 6-dimensional array stored as a JSON string serialization), actual total oxidation time, average measured film thickness, and standard deviation of measured film thickness. The database construction rules are as follows: the first construction is triggered when the number of qualified batch archive records of the same profile type and the same alloy grade first reaches the preset database construction quantity threshold (set to 50 records); subsequent additions of each new qualified record are incrementally appended.

[0074] Optionally, S420 matches the normalized multi-feature vector with the corresponding feature parameters in the historical batch database, and determines the recommended endpoint time based on the matching results, including: S421, calculate the similarity between the normalized multi-feature vector and the feature vector of each historical batch in the pre-built historical batch database, and take the K historical batches with the highest similarity as the nearest neighbor batch set; where K is a preset integer not less than five.

[0075] It is understandable that Euclidean distance is used as a metric for similarity calculation; the smaller the Euclidean distance, the closer the waveform shapes of the two batches are. The Euclidean distance is calculated as the square root of the sum of the squares of the differences in the six dimensions. After receiving the normalized multi-feature vector of the current batch, the host computer filters out all qualified records of the same profile type and alloy grade from the historical batch database and calculates the Euclidean distance for each record. The value of K is determined through the following cross-validation method: leave-one-out cross-validation is performed in a database with more than 100 historical records of the same model and alloy, and the root mean square error is taken for the validation results of different K values ​​(3 to 15). After performing the above cross-validation on the historical databases of different profile types on the three production lines, the average root mean square error was the smallest when K was 10 (about 1.2 minutes), slightly higher when K was 5 (about 1.5 minutes), and started to rise when K was 15 (about 1.8 minutes). Considering both accuracy and statistical stability, K was set to 10. The nearest neighbor batch set is a subset consisting of the first 10 historical batch records with the smallest Euclidean distance after sorting them in ascending order of Euclidean distance.

[0076] S422: Based on the total actual oxidation time of the K nearest neighbor batches, a weighted average is calculated using the similarity of each historical batch as the weight to obtain the recommended endpoint time.

[0077] It's understandable that the similarity weight can be the reciprocal of the Euclidean distance, with smaller distances resulting in larger weights for historical batches. The reason for using the reciprocal of the Euclidean distance is that when two batches are almost completely similar, the Euclidean distance approaches 0, and its reciprocal approaches infinity, thus this historical batch almost exclusively accounts for the weighted average. When two batches are significantly different, the Euclidean distance is large, and its reciprocal is small, correspondingly reducing its contribution to the weighted average. The weighted average is calculated as follows: the recommended endpoint time equals the sum of the products of the actual total oxidation time of each nearest neighbor batch and their respective weights, divided by the sum of the weights, where the weight is equal to the reciprocal of the Euclidean distance of the corresponding batch. The recommended endpoint time is the absolute time value calculated from the moment of power-on, in minutes. The time complexity of the host computer performing the above weighted average calculation is proportional to K; when K is 10, the calculation time is less than 1 millisecond. As the oxidation process progresses, the stability of the feature vector and the recommendation accuracy improve simultaneously. Typically, when half of the total oxidation time has passed, the recommended endpoint time has converged to within ±2 minutes of the final value.

[0078] S500 merges the recommended endpoint time with the time when the remaining oxidation time is zero, and triggers the endpoint signal at the merged endpoint time.

[0079] It is understandable that the recommended endpoint time is derived from the recommendation results based on complete waveform shape analysis and historical batch matching, while the time when the remaining oxidation time reaches zero is derived from the dynamic calculation results based on the initial slope method and current efficiency correction coefficient. The two endpoint times originate from independent control logic channels. The initial slope channel responds quickly based on the initial slope of the first few minutes but is insensitive to abnormal changes in the later stages of oxidation. The waveform matching channel is based on complete waveform matching and historical experience but is sensitive to the uneven distribution of historical data. The probability of both independent channels simultaneously malfunctioning with completely identical error directions and magnitudes is extremely low. Therefore, fusion judgment can effectively reduce the risk of systematic out-of-tolerance errors caused by independent errors in any single channel. The formula for calculating the endpoint time after fusion is t_fusion = (t_zero + t_recommend) / 2. The trigger endpoint signal refers to the host computer or PLC sending a stop output command to the rectifier power supply and driving the audible and visual alarm to indicate the tank exit when it detects that the cumulative oxidation time has first reached or exceeded t_fusion.

[0080] For example, suppose that when a batch has been oxidizing for 15 minutes, the initial slope channel calculates that the remaining oxidation time is 25 minutes (the zeroing time is 40 minutes after power-on). The waveform matching channel recommends an endpoint time of 43 minutes after power-on. The final endpoint time after fusion is (40+43) / 2, which equals 41.5 minutes after power-on. When the timer accumulates to 41.5 minutes, the PLC outputs an endpoint signal.

[0081] It should be noted that existing technologies for film thickness control either employ a completely open-loop fixed ampere-hour method (offline detection only after oxidation is complete) or rely on additional hardware sensors for online thickness measurement. This application achieves online endpoint control of film thickness without adding any hardware by fusing two complementary soft measurement channels: an initial slope rapid response channel and a waveform matching historical experience channel. When the recommended endpoint deviation between the two channels exceeds a preset deviation threshold (e.g., 5 minutes), the host computer or PLC triggers an alarm mechanism requiring manual intervention to prevent systematic deviations in the automatic control system within data blind spots.

[0082] In one possible implementation, S500 merges the recommended endpoint time with the time when the remaining oxidation time returns to zero, and triggers an endpoint signal at the merged endpoint time, including: S510, the recommended endpoint time and the time when the remaining oxidation time is reduced to zero are arithmetically averaged to obtain the endpoint time after fusion.

[0083] Arithmetic averaging fusion can be understood as adding the recommended endpoint time of the waveform matching channel output and the time when the remaining oxidation time of the initial slope channel output is zeroed, and then dividing by 2. Arithmetic averaging fusion is the most robust fusion method that gives equal decision weight to two independent information sources, provided there is no known systematic difference in the estimation accuracy of the two independent channels. The formula for calculating the endpoint time after fusion is t_fusion = (t_zero + t_recommend) / 2. The host computer or PLC synchronously reads the latest outputs of the two channels and recalculates the fused endpoint time within each preset control cycle (30 seconds), ensuring that the endpoint determination is always based on the latest information.

[0084] S520 triggers an endpoint signal when the cumulative oxidation time reaches the endpoint after fusion.

[0085] It can be understood that triggering the endpoint signal refers to the automatic execution of the following sequence of operations when the host computer or PLC detects that the cumulative oxidation time is greater than or equal to the endpoint time after fusion for the first time. First, a stop output command is sent to the rectifier power supply. The rectifier power supply will turn off the output current at a preset slow-decreasing rate (usually 50% of the target current per second, dropping to zero within 2 seconds) to prevent the inductor backflow voltage spike caused by the sudden current change from damaging the oxide film layer. Second, the audible and visual alarm is activated to flash twice per second to notify the operator that the current batch oxidation is complete. At the same time, a pop-up window on the HMI (Human Machine Interface) displays a text message indicating that oxidation is complete, as well as a summary of the key process parameters for this batch. Third, the complete process data and final results for this batch are packaged and stored, including the entire tank voltage and current acquisition sequence, average rise slope, average rise slope after compensation, current efficiency correction coefficient, 6 waveform characteristic parameters, recommended endpoint time, endpoint time after fusion, and actual trigger time. This data is then stored in the host computer's batch record database and marked as pending verification. Finally, wait for the preset confirmation time for exiting the tank (set to 10 seconds): If the operator presses the confirmation button within 10 seconds, the PLC resets the alarm state and records the exit timestamp; if no confirmation is received within 10 seconds, the alarm continues to prevent the profile from being chemically dissolved and damaged by prolonged immersion in the electrolyte without current protection. After exiting the tank, the profile undergoes two stages of counter-current water washing and compressed air drying. Quality inspectors then use an eddy current thickness gauge to measure the film thickness at 20 evenly distributed measuring points on the profile surface. The measurement data is entered into the host computer quality management system and automatically associated with the corresponding batch, completing the data loop. This closed-loop data feedback mechanism provides continuous empirical support for the continuous self-optimization of the historical batch database. With each new qualified record, the coverage and matching accuracy of the database are incrementally improved, thus achieving an online evolution effect where the waveform matching channel becomes more accurate with use.

[0086] Corresponding to the closed-loop control method for the thickness of the anodized film on new energy profiles in the above embodiments, this application also provides a closed-loop control system for the thickness of the anodized film on new energy profiles. Each unit of this system can realize each step of the closed-loop control method for the thickness of the anodized film on new energy profiles. Figure 4 The diagram shows a structural block diagram of the closed-loop control system for the thickness of the anodized film on new energy profiles provided in this application embodiment. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0087] Reference Figure 4 The closed-loop control system for the thickness of the anodized film on this new energy profile includes: The acquisition unit is used to continuously acquire the cell voltage and cell current values ​​output by the rectifier power supply within a preset initial observation window after the constant current anodizing is energized, at a preset sampling period, and calculate the average rising slope of the cell voltage within the initial observation window based on the cell voltage value. The compensation unit is used to obtain the current bath temperature and the current total surface area of ​​the profile, and to compensate the average rise slope based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average rise slope. The calculation unit is used to determine the ratio of the compensated average rising slope to the pre-calibrated reference rising slope as the current efficiency correction coefficient for the current batch, and dynamically calculate the remaining oxidation time required for the current batch to reach the target film thickness based on the current efficiency correction coefficient. The recommendation unit is used to perform waveform shape analysis on the continuously collected cell voltage value within each preset control cycle, and obtain a recommended endpoint time to assist in determining the final value of the remaining oxidation time based on the matching results of the waveform shape characteristics of the current batch and the historical batch data. The result unit is used to merge and determine the recommended endpoint time with the time when the remaining oxidation time is zero, and to trigger the endpoint signal at the merged endpoint time.

[0088] It should be noted that the information interaction and execution process between the above systems / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0089] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit module can exist physically separately, or two or more unit modules can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0090] This application also provides an anodizing line. Figure 5 This is a schematic diagram of the structure of an anodized line provided in one embodiment of this application. Figure 5 As shown, the anodizing line 6 in this embodiment includes: at least one processor 60 ( Figure 5 Only one is shown in the image), at least one memory 61 ( Figure 5 (Only one is shown in the image) and a computer program 62 stored in the at least one memory 61 and executable on the at least one processor 60. When the processor 60 executes the computer program 62, it causes the anodizing line 6 to perform the steps in any of the above embodiments of the closed-loop control method for anodized film thickness of new energy profiles, or causes the anodizing line 6 to perform the functions of each unit in the above embodiments of the system.

[0091] Exemplarily, the computer program 62 may be divided into one or more units, which are stored in the memory 61 and executed by the processor 60 to complete this application. The one or more units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 62 in the anodizing line 6.

[0092] Anodizing line 6 can be any type of horizontal or vertical aluminum alloy profile anodizing production line. This anodizing line may include, but is not limited to, processor 60 and memory 61. Those skilled in the art will understand that... Figure 5 This is merely an example of anodized line 6 and does not constitute a limitation on anodized line 6. It may include more or fewer components than shown, or combine certain components, or different components, such as input / output devices, network access devices, buses, etc.

[0093] The processor 60 can be a Central Processing Unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0094] In some embodiments, the memory 61 can be an internal storage unit of the anodizing line 6, such as a hard disk or memory of the anodizing line 6. In other embodiments, the memory 61 can be an external storage device of the anodizing line 6, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the anodizing line 6. Furthermore, the memory 61 can include both internal storage units and external storage devices of the anodizing line 6. The memory 61 is used to store operating systems, applications, bootloaders, data, and other programs, such as the program code of computer programs. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0095] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0096] This application provides a computer program product that, when run on an anodizing line, causes the anodizing line to perform the steps described in any of the above method embodiments.

[0097] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to the anodizing line, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0098] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0099] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0100] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A closed-loop control method for the thickness of anodized film on new energy profiles, characterized in that, The method includes: Within a preset initial observation window after constant current anodizing is energized, the cell voltage and cell current values ​​output by the rectifier power supply are continuously collected at a preset sampling period. The average rising slope of the cell voltage within the initial observation window is calculated based on the cell voltage values. Obtain the current bath temperature and the current total surface area of ​​the profile, and compensate the average rise slope based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average rise slope; The ratio of the compensated average rising slope to the pre-calibrated baseline rising slope is determined as the current efficiency correction coefficient for the current batch. Based on the current efficiency correction coefficient, the remaining oxidation time required for the current batch to reach the target film thickness is dynamically calculated. Within each preset control cycle, the continuously collected cell voltage values ​​are analyzed for waveform shape. Based on the matching results of the waveform shape characteristics of the current batch and the historical batch data, a recommended endpoint time is obtained to help determine the final value of the remaining oxidation time. The recommended endpoint time and the time when the remaining oxidation time is reduced to zero are merged and determined, and the endpoint signal is triggered at the merged endpoint time.

2. The method as described in claim 1, characterized in that, The rectified power supply outputs slot voltage and slot current values ​​are continuously acquired at a preset sampling period, including: The tank voltage and tank current values ​​are collected at each sampling time according to the preset sampling period. Sampling points that deviate from the target current value by more than the preset current fluctuation threshold are marked as invalid sampling points and discarded. The proportion of invalid sampling points removed to the total number of sampling points in the initial observation window is counted. When the proportion exceeds a preset proportion threshold, the initial observation window is extended until the number of valid sampling points reaches a preset valid number threshold.

3. The method as described in claim 1, characterized in that, The process of obtaining the current bath temperature and the current total surface area of ​​the profile, and compensating the average upward slope based on the bath temperature and the total surface area of ​​the profile to obtain the compensated average upward slope includes: Obtain the current bath temperature and the current total surface area of ​​the profile, and calculate the difference between the reciprocal of the absolute temperature value corresponding to the current bath temperature and the reciprocal of the absolute temperature value corresponding to the pre-calibrated reference temperature. The temperature compensation coefficient is determined based on the current bath temperature and the pre-calibrated reference temperature. The surface area normalization coefficient is determined based on the ratio of the current total surface area of ​​the profile to the pre-calibrated total surface area of ​​the reference profile. The compensated average slope is calculated based on the average upward slope, the temperature compensation coefficient, and the surface area normalization coefficient.

4. The method as described in claim 3, characterized in that, The process of determining the temperature compensation coefficient based on the current bath temperature and the pre-calibrated reference temperature includes: Calculate the difference between the reciprocal of the absolute temperature value corresponding to the current tank temperature and the reciprocal of the absolute temperature value corresponding to the reference temperature; The temperature compensation coefficient is obtained by multiplying the difference by the pre-calibrated ratio of the activation energy gas constant and using it as the exponent, and calculating the exponential function value with the natural constant e as the base.

5. The method as described in claim 1, characterized in that, The remaining oxidation time required for the current batch to reach the target film thickness is dynamically calculated based on the current efficiency correction coefficient, including: Multiply the current efficiency correction factor, the pre-calibrated reference current efficiency, the pre-calibrated Faraday film formation factor, the nominal current density, and the cumulative oxidation time from the time of energization to the current time to obtain the estimated value of the generated film thickness. Subtract the generated film thickness estimate from the preset target film thickness to obtain the remaining required film thickness; The remaining required film thickness is divided by the product of the current efficiency correction factor, the reference current efficiency, the Faraday film formation factor, and the nominal current density to obtain the remaining oxidation time at the current moment.

6. The method as described in claim 1, characterized in that, Within each preset control cycle, waveform shape analysis is performed on the continuously collected tank voltage values. Based on the matching results between the waveform shape characteristics of the current batch and historical batch data, a recommended endpoint time is obtained to assist in determining the final value of the remaining oxidation time, including: Within each preset control cycle, the time curve corresponding to the continuously acquired slot voltage value is segmented into stages, and a normalized multi-feature vector reflecting the waveform shape of the current batch is extracted. The normalized multi-feature vector is matched with the corresponding feature parameters in the historical batch database, and the recommended endpoint time is determined based on the matching results.

7. The method as described in claim 6, characterized in that, Within each preset control cycle, the time curve corresponding to the continuously acquired slot voltage value is segmented into stages, and a normalized multi-feature vector reflecting the waveform shape of the current batch is extracted, including: The time curve of the slot voltage value is used to calculate the first average slope within a window length before the current sampling time and the second average slope within a window length after the current sampling time using a sliding window of a preset duration. The transition time and transition point voltage between the barrier layer formation stage and the porous layer stable growth stage are determined based on the degree of difference between the first average slope and the second average slope. Based on the transition time and the transition point voltage, extract the first-stage average slope of the tank voltage before the transition time, the second-stage average slope after the transition time, the characteristic resistance of the transition point voltage and the target current value, and the slope ratio of the second-stage average slope to the first-stage average slope. The transition time, the transition point voltage, the average slope of the first stage, the average slope of the second stage, the characteristic resistance, and the slope ratio are each normalized by dividing by their respective preset reference values ​​to obtain normalized multi-feature vectors.

8. The method as described in claim 6, characterized in that, The step of matching the normalized multi-feature vector with the corresponding feature parameters in the historical batch database, and determining the recommended endpoint time based on the matching result, includes: The normalized multi-feature vector is compared with the feature vectors of each historical batch in the pre-constructed historical batch database. The K historical batches with the highest similarity are taken as the nearest neighbor batch set; where K is a preset integer not less than five. Based on the actual total oxidation time of the K nearest neighbor batches, a weighted average is performed using the similarity of each historical batch as the weight to obtain the recommended endpoint time.

9. The method as described in claim 1, characterized in that, The step of fusing the recommended endpoint time with the time when the remaining oxidation time is reduced to zero, and triggering the endpoint signal at the fused endpoint time, includes: The recommended endpoint time and the time when the remaining oxidation time is reduced to zero are arithmetically averaged to obtain the fusion endpoint time. When the cumulative oxidation time reaches the endpoint time after fusion, the endpoint signal is triggered.

10. An anodizing line, comprising a control system and a production system, the control system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 9.

Images