Electroplating solution additive automatic supply control method based on conductivity monitoring data

Abstract

The invention relates to the technical field of electrochemical machining, in particular to an electroplate liquid additive automatic supply control method based on conductivity monitoring data. The method comprises the steps that the temperature, the original conductivity and the carbonate concentration of electroplating liquid are obtained, an activation energy coefficient is determined based on the change trend of the original conductivity and the temperature, dynamic thermodynamic decoupling treatment is conducted on the original conductivity, and the net conductivity of the electroplating liquid is generated; according to the cumulative variation of the net conductivity in the historical time period and the total amount of the additive in the historical time period, the transmission inertia factor of the electroplating liquid is calculated in combination with the carbonate concentration, and the retardation risk index is determined based on the evolution trend of the transmission inertia factor, so that a PID control algorithm is reconstructed, and the supply amount of the additive is obtained; and performing automatic replenishment control based on the replenishment amount. According to the method, the accuracy of automatic supply control is improved by evaluating the degradation state of the bath solution components and self-adaptive integral saturation suppression.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrochemical processing, in particular to an electroplating solution additive automatic replenishment control method based on conductivity monitoring data. BACKGROUND

[0002] Gas insulated switchgear (GIS) is the core equipment of ultra-high voltage power transmission and transformation system, and the surface treatment quality of the aluminum alloy shell directly relates to the insulation performance and service life of the equipment. In the electroplating process of the aluminum alloy shell, the concentration of additives such as brightener and leveling agent in the electroplating solution is a key factor determining the microcrystalline density and flatness of the plating layer. Therefore, in actual production, an automatic dosing system is usually configured to monitor the solution state by using an online conductivity sensor and replenish the additives accordingly to maintain the dynamic balance of the electroplating solution components.

[0003] Currently, PID (proportional-integral-derivative) control algorithm is generally used in industrial sites to adjust the replenishment amount of additives according to the conductivity deviation. PID algorithm is a classic feedback control strategy, and its typical composition includes a signal acquisition unit, a deviation calculation unit and a control execution unit. The algorithm calculates the control output based on the linear combination of the proportion, integral and derivative of the error, and is widely used in scenes such as automobile manufacturing and chemical production, which can reduce manual intervention and improve production efficiency.

[0004] However, in the large-scale continuous electroplating scene of GIS aluminum alloy shell, the existing PID control method faces serious physical adaptability problems. As the electroplating period is prolonged, aluminum ion impurities and carbonate by-products will inevitably accumulate in the solution, resulting in gradual increase of the viscosity and ion migration resistance of the electroplating solution, and causing component degradation. This change in physical and chemical properties causes a mass transfer delay in the reading response of the conductivity sensor after the addition of additives. In this component degradation condition, since the conductivity change lags behind the dosing action, the standard PID algorithm will misjudge as invalid replenishment, and then through the continuous accumulation of the integral term, it will cause integral saturation and issue excessive dosing instructions. When the conductivity suddenly changes due to the final diffusion of the additives, the concentration of the solution has already exceeded the standard, resulting in pinhole or scorching defects on the surface of the aluminum alloy shell. The PID algorithm cannot identify the time-varying physical parameter of component degradation, and executes fixed gain control, which is the main cause of the instability of the plating solution.

[0005] Therefore, there is an urgent need for a control method that can accurately identify the component degradation state of the solution and adaptively suppress excessive replenishment in high-lag conditions, to solve the problems of integral saturation and unstable plating layer quality caused by response lag in the prior art. SUMMARY

[0006] To solve the integral saturation and excessive compensation problem caused by response lag of the PID algorithm in the tank liquid aging scene, the application proposes an electroplating solution additive automatic compensation control method based on conductivity monitoring data, which includes: The temperature and original conductivity and carbonate concentration of the electroplating solution are obtained, and the activation energy coefficient of the electroplating solution is determined based on the change trend of the original conductivity and temperature, to reflect the sensitivity of ion migration in the electroplating solution to temperature change; the original conductivity is dynamically thermodynamically decoupled to eliminate the influence of temperature fluctuation on conductivity, and the net conductivity of the electroplating solution is generated; According to the cumulative change amount of the net conductivity in the preset historical time period and the total amount of additives in the historical time period, combined with the carbonate concentration, the transmission inertia factor of the electroplating solution is calculated to reflect the lag degree of the electroplating solution to the additive compensation response; The PID control algorithm is used to control the additive compensation of the electroplating solution, and the retardation risk index of the electroplating solution is calculated based on the transmission inertia factor and the evolution trend of the transmission inertia factor over time; The integral term of the PID control algorithm is reconstructed in real time using the retardation risk index, and the compensation amount of the additive is obtained based on the real-time reconstructed PID control algorithm, and the electroplating solution is automatically compensated based on the compensation amount.

[0007] This technical solution first removes temperature interference through dynamic thermodynamic decoupling at the data source, ensuring the purity of the control signal, and quantifies the component degradation degree and response lag characteristics of the tank liquid from the physical layer by calculating the transmission inertia factor, and then uses the integral confidence to intelligently inhibit and attenuate the blind accumulation of the integral term in the PID algorithm when the mass transfer of the tank liquid is blocked. This mechanism avoids integral saturation and system overshoot caused by lag, ensuring the accuracy of additive compensation.

[0008] Preferably, the activation energy coefficient of the electroplating solution is determined based on the following method: At each dosing time of the electroplating process, the natural logarithm value of the original conductivity of all historical dosing times in the preset historical time period and the reciprocal value of the temperature of all historical dosing times at the dosing time are obtained; the natural logarithm value and the reciprocal value are linearly regressed and fitted by using the least squares method, and the slope of the fitted straight line is obtained. The activation energy coefficient at the dosing time is determined based on the slope through the Arrhenius law.

[0009] This technical solution performs thermodynamic decoupling by introducing a dynamic Arrhenius equation based on real-time multi-dimensional data regression of the electroplating solution, and calculates the real-time activation energy coefficient, so that the algorithm can adapt to the influence of component changes in the electroplating solution on the thermal sensitivity characteristics, thereby achieving high-precision temperature compensation of the electroplating solution throughout its life cycle.

[0010] Preferably, the net conductivity of the electroplating solution is determined based on the following manner:

[0011] wherein, is the time to be dosed, is the net conductivity of the electroplating solution at the time to be dosed, is the original conductivity of the electroplating solution at the time to be dosed, is a natural exponential function, is the activation energy coefficient of the electroplating solution at the time to be dosed, is the ideal gas constant, is the temperature of the electroplating solution at the time to be dosed, is a preset standard temperature.

[0012] This technical solution introduces a real-time calculated activation energy coefficient, constructs a dynamic thermodynamic decoupling mechanism consistent with the Arrhenius ion migration law. In the GIS aluminum alloy shell electroplating process, the bath temperature fluctuates due to the Joule heat effect, and the aging of the bath causes the thermal sensitivity characteristics (i.e. activation energy) to drift. According to the current micro-ion migration energy barrier, the physical disturbance of the temperature change to the original conductivity is offset, so that the calculated net conductivity can maintain a linear response to the solute concentration even under severe temperature fluctuations, thereby decoupling the thermodynamic noise and providing an accurate chemical component feedback signal for subsequent control.

[0013] Preferably, the transmission inert factor satisfies the following relationship:

[0014] wherein, is the transmission inert factor of the electroplating solution at the time, is the time, is a hyperbolic tangent function, is a preset dimensionless coefficient, used to convert the calculation item in the parentheses into a dimensionless value, is the length of the historical time period, is the integral variable, is the dosing amount at the time, is the change rate of the net conductivity at the time, is a positive number preset to prevent the denominator from being 0, is the relative growth rate of the carbonate concentration of the electroplating solution at the time, is a preset positive number to prevent the denominator from being 0, is the relative growth rate of the carbonate concentration of the electroplating solution at the time, is a preset dimensionless coefficient, used to convert the calculation item in the parentheses into a dimensionless value, is a weight coefficient of the influence of the carbonate concentration on the viscosity of the electroplating solution, which is determined in advance through experiments.

[0015] The technical solution quantifies the mass transfer resistance state of the electroplating solution from the physical dynamics level, constructs a dynamic ratio of historical dosing excitation and net conductivity response, and introduces a relative growth rate of carbonate concentration as a correction term. In actual working conditions, as the carbonate accumulates, the viscosity of the bath solution increases, hindering the Brownian motion and convective diffusion of the additives, resulting in a lag in the sensor response. By calculating the dynamic gain ratio and combining viscosity correction, this macroscopic response delay is mapped to a normalized inertia factor. When the transmission inertia factor approaches 1, the system is accurately identified as being in a high-viscosity, high-damping physical obstruction zone, thereby providing a basis for the control algorithm to determine whether the dosing is ineffective due to sufficient concentration or diffusion obstruction.

[0016] Preferably, the carbonate concentration at the dosing time is obtained by an online ion concentration analyzer or is determined by receiving periodic test data input through a human-computer interaction interface.

[0017] Preferably, the relative growth rate of the carbonate concentration of the electroplating solution at the dosing time is obtained by first obtaining the carbonate concentration determined at the initial tank preparation time, calculating the relative increment of the carbonate concentration at the dosing time and the carbonate concentration determined at the initial tank preparation time, and taking the ratio of the relative increment and the carbonate concentration determined at the initial tank preparation time as the relative growth rate of the carbonate concentration of the electroplating solution at the dosing time.

[0018] Preferably, the resistance risk index of the electroplating solution is determined based on the following method: obtaining the transmission inertia factor at each historical dosing time in the historical time period at the dosing time, and performing linear regression fitting to obtain a fitting slope to reflect the evolution gradient of the transmission inertia factor; calculating the resistance risk index:

[0019] wherein, is the resistance risk index of the electroplating solution at the dosing time, is the transmission inertia factor of the electroplating solution at the dosing time, is a sign function, is a nonlinear normalization function, is an absolute value symbol.

[0020] The technical scheme effectively distinguifies different states of stable high viscosity and transient retardation deterioration by introducing a product model of static reference and dynamic modulation, uses a sign function to accurately judge the evolution gradient direction, automatically degrades to a static reference in a stable state, ensures the continuity of control, uses the product effect to amplify the risk signal in the deterioration state, realizes acute capture of acute failure, uses negative gradient attenuation risk in the recovery period, realizes advanced prediction of system improvement, and such mechanism makes the PID control algorithm no longer blindly rely on current values, but introduces historical trend analysis, and ensures high consistency of risk assessment and actual physical mass transfer process.

[0021] Preferably, the integral term of the PID control algorithm is reconstructed in real time by using the retardation risk index, including: using a nonlinear attenuation function to map the retardation risk index into a value in the interval of 0 to 1 as an integral confidence, and using the integral confidence to perform a weighting operation on the integral term of the PID control algorithm to obtain a reconstructed PID control algorithm.

[0022] Preferably, the replenishment amount of the additive based on the real-time reconstructed PID control algorithm is determined based on the following relationship: ; wherein, is the replenishment amount of the additive of the electroplating solution at the time of adding the additive, are respectively a proportional gain constant, an integral gain constant and a differential gain constant of the PID control algorithm preset in advance, is the deviation amount of the net conductivity at the time and the preset standard conductivity, is the integral confidence at the time, is the derivative of the deviation amount with respect to time, is the integral operation with respect to time. The technical scheme reconstructs the integral core of the traditional PID control algorithm by using the calculated integral confidence, which is different from the blind accumulation of errors in the traditional algorithm. The dynamic weight is implanted into the integral term, which gives the system state perception ability. When the bath is in the causal lag area of component deterioration, the algorithm automatically freezes the integral action to prevent integral saturation caused by sensor reading delay. Such control strategy ensures that the final output of the replenishment control amount can match the actual ion diffusion rate of the electroplating solution, avoids excessive dosing caused by pinhole and scorching defects of the plated layer, and improves the robustness of the process.

[0023] Preferably, the electroplating solution is automatically controlled based on the replenishment amount, including: at the time of adding the additive, driving an additive replenishment device to release the additive into the electroplating solution according to the replenishment amount.

[0024] Preferably, the electroplating solution is automatically controlled based on the replenishment amount, including: at the time of adding the additive, driving an additive replenishment device to release the additive into the electroplating solution according to the replenishment amount.

[0025] ​The present application has the following effects: The present application overcomes the limitations of the traditional PID algorithm in time-varying and lagging systems, realizes high-precision temperature decoupling by dynamically calculating ion migration activation energy, and endows the control system with the ability to identify the deterioration state of the bath solution composition and its microscopic dynamics mechanism by introducing a transmission inert factor containing the influence of carbonate, realizes adaptive defense against integral saturation by reconstructing integral confidence based on the retardation risk index. This dual mechanism of dynamic physical identification and environmental adaptive inhibition solves the problems of dosing lag and excessive replenishment caused by the deterioration of the electroplating solution composition. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 The present application has the following effects: Figure 2 The present application has the following effects: DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0028] The present application provides an electroplating solution additive automatic replenishment control method based on conductivity monitoring data, as shown in Figure 1 The present application has the following effects: S1: Obtain the temperature and original conductivity of the electroplating solution, calculate the activation energy coefficient, and perform dynamic thermodynamic decoupling processing on the original conductivity to generate the net conductivity of the electroplating solution.

[0029] At the dosing time of the electroplating process link, considering that the conductivity of the electroplating solution is extremely sensitive to temperature changes, and this thermal sensitivity is not constant, as the use time of the electroplating solution increases, the accumulation of impurity ions and the decomposition of organic additives will change the microstructure of the solution, and then change its ion migration activation energy. This step aims to perform adaptive thermodynamic decoupling preprocessing, use the naturally occurring temperature fluctuation data in the production process to perform real-time regression analysis of the activation energy characteristics at the current time, use the dynamic Arrhenius equation to strip the temperature influence, and extract the high-fidelity net conductivity to provide a pure concentration feedback signal for subsequent control.

[0030] First, the activation energy coefficient is dynamically obtained: For the dosing time, it is denoted as At any given moment, the temperature, initial conductivity, and carbonate concentration of the electroplating solution are acquired. The temperature is obtained in real-time using a temperature sensor, such as a PT100 resistance temperature detector (RTD) or a type K thermocouple, within the electroplating bath's circulation pipeline or the bath itself. The carbonate concentration is obtained using an online ion concentration analyzer, and the initial conductivity is obtained in real-time using an immersion or flow-through online conductivity sensor, such as an electromagnetic induction conductivity probe.

[0031] A historical time period moving forward along the time axis is constructed as the historical time period for when the additive is to be added. To ensure the validity of the statistics and to adapt to the slow time-varying characteristics of the tank solution changes, this historical time period includes 10 historical addition times, meaning that the additive has been added 10 times within this historical time period. For each historical addition time, a data set is constructed. ,in, The original conductivity at that historical moment of drug administration. The temperature at that historical moment of administering the medication. The function is logarithmic, which will result in a series of data combinations. We then use the least squares method to perform linear regression on these data combinations to construct a linear equation: ,in, The slope The intercept term is obtained through regression calculation. and According to Arrhenius's law, the following relation is satisfied: ,in, The activation energy coefficient of the electroplating solution at the moment of chemical addition is given. The ideal gas constant Therefore, the activation energy coefficient at the time of drug administration The physical significance of this mechanism lies in its ability to capture the sensitivity of the electroplating solution's conductivity to temperature changes. When the electroplating solution is relatively new and contains few impurities, The temperature is relatively low, with moderate temperature compensation. However, when the bath solution ages or its viscosity increases, ion migration is hindered, typically resulting in greater sensitivity to temperature. Increase.

[0032] Next, an activation energy coefficient is introduced to perform dynamic thermodynamic decoupling on the original conductivity, so as to eliminate the influence of temperature fluctuations on conductivity and generate the net conductivity of the electroplating solution.

[0033] Specifically, the net conductivity of the electroplating solution at the moment of chemical addition satisfies the following relationship:

[0034] in, For the electroplating solution at the moment when the chemical is to be added ( Net conductivity at time (time) G0 is the original conductivity of the electroplating solution at the moment when the additive is to be added, G0 is the original conductivity of the electroplating solution at the moment when the additive is to be added, G0 is the original conductivity of the electroplating solution at the moment when the additive is to be added, G0 is the original conductivity of the electroplating solution at the moment when the additive is to be added, G0 is the original conductivity of the electroplating solution at the moment when the additive is to be added, G0 is the original conductivity of the electroplating solution at the moment when the additive is to be added,

[0035] In this relationship, represents the difference between the reciprocal of the real-time temperature and the reciprocal of the reference temperature, which reflects the degree of thermodynamic deviation of the current temperature relative to the standard temperature, represents the ratio of the activation energy coefficient to the gas constant, which reflects the sensitivity coefficient of the conductivity of the electroplating solution to temperature changes, represents a dimensionless temperature compensation index, which reflects the theoretical deviation amplitude of the temperature deviation to the conductivity under the current activation energy characteristics, which constitutes a temperature compensation coefficient, which reflects the correction multiple of the original conductivity, regardless of changes, the temperature compensation coefficient always follows the thermodynamic law: when , is negative, the exponential term is less than 0, the temperature compensation coefficient is less than 1, and the temperature compensation coefficient has a shrinking effect on the original conductivity , offsetting the increase in ion mobility at high temperatures and eliminating the false high component in the original conductivity; when , is positive, the exponential term is greater than 0, the temperature compensation coefficient is greater than 1, and the temperature compensation coefficient has an amplification effect on the original conductivity , compensating for the decrease in ion mobility at low temperatures and eliminating the false low component in the original conductivity.

[0036] In the formula, represents the original conductivity value directly measured by the sensor at the current temperature, which is multiplied by and , indicating that the original measurement value is multiplied by the dynamically calculated compensation coefficient, eliminating the interference caused by temperature fluctuations through physical layer correction, and restoring the true conductivity determined only by solute concentration. This dynamic closed loop ensures that the solute concentration is always accurately reflected and is not affected by the thermal sensitivity drift caused by the aging of the bath solution.

[0037] As the electroplating process proceeds, the temperature of the electroplating solution gradually rises due to the Joule heating effect. On a physical level, the temperature rise gives ions higher kinetic energy, causing the raw conductivity read by the sensor to naturally rise even if the additive concentration has not changed. If there is a lack of dynamic decoupling in this step, the control system will mistakenly believe that the additive concentration is already high enough, and will falsely stop replenishment due to the false high conductivity. The net conductivity calculated in real time reflects the thermal sensitivity of the electroplating solution at the time of dosing. In the case of an increase in viscosity and thermal sensitivity of the electroplating solution due to aging, the relationship will automatically apply a larger correction factor to exclude the false increase caused by temperature, and the final net conductivity will truly reflect the conductivity fluctuations caused by changes in chemical composition under temperature fluctuations, providing an accurate basis for subsequent precise control.

[0038] S2: Calculate the transport inertia factor of the electroplating solution according to the cumulative change of the net conductivity in a preset historical time period and the total amount of additive in the historical time period, combined with the carbonate concentration.

[0039] After obtaining the net conductivity, in order to solve the response lag problem caused by the inability of the PID algorithm to identify the deterioration of the bath composition, the current mass transfer resistance must be quantified. This step aims to establish a dynamic state identification mechanism that integrates multiple physical characteristics. By comprehensively analyzing the dynamic proportion of dosing amount and conductivity change, and introducing the carbonate concentration as a correction factor, the transport inertia factor is accurately calculated. The transport inertia factor not only reflects the macroscopic response lag degree, but also reveals the microscopic electrochemical blockage degree, providing a reliable physical basis for the integral reconstruction of the PID control algorithm.

[0040] Specifically, at the time of dosing, the total amount of already replenished additive at all historical dosing times in the historical time period is obtained, and the change rate of the net conductivity at each historical dosing time is calculated, i.e. the difference between the net conductivity at the historical dosing time and the net conductivity at the previous historical dosing time. The difference between the net conductivity at the historical dosing time and the net conductivity at the previous historical dosing time is calculated, and the difference is divided by the time interval between the two historical dosing times to obtain the change rate of the net conductivity, which reflects the transient response speed of the electroplating solution to the additive.

[0041] Then, the transport inertia factor of the electroplating solution at the time of dosing is calculated:

[0042] wherein, is the transport inertia factor of the electroplating solution at the time of dosing, with a value range of , is a hyperbolic tangent function used to map the calculation result to the interval of 0 to 1, is a preset dimensional balance coefficient, whose scalar value is 1, and its unit is the reciprocal of the unit of the calculation item in the parentheses, and its physical meaning is to offset the dimension of the fraction, so that The input of the function becomes a dimensionless value, is the length of the historical time period at the time of dosing, that is, how many historical dosing times are included, is the integral variable, is the dosing amount at the time, is the rate of change of the net conductivity at the time, reflecting the dynamic response sensitivity of the electroplating solution to the addition of the additive, a high rate of change means that after the additive is added, it quickly spreads throughout the bath, which usually means that the bath is in good condition, with low viscosity, good flowability and sufficient agitation, and a low rate of change means that after the additive is added, it spreads slowly, and the sensor cannot read the value change for a long time, which reflects that the bath may have component degradation, high viscosity, high accumulation of carbonate and high mass transfer resistance. is a preset positive number to prevent the denominator from being 0, usually set to to prevent the denominator from being 0 due to no change in conductivity, is the relative growth rate of the carbonate concentration of the electroplating solution at the time of dosing, is a weight coefficient of the influence of the carbonate concentration on the viscosity of the electroplating solution, which is determined by experiment in advance, and reflects the degree of influence of viscosity on diffusion, and the value is 0.5, which physically corrects the physical diffusion lag caused by the increase in viscosity.

[0043] In this relationship, the relative growth rate of the carbonate concentration of the electroplating solution at the time of dosing is obtained by: first obtaining the carbonate concentration of the electroplating solution determined at the initial tank preparation, calculating the relative increment of the carbonate concentration at the time of dosing and the carbonate concentration determined at the initial tank preparation, and taking the ratio of the relative increment to the carbonate concentration determined at the initial tank preparation as the relative growth rate of the carbonate concentration of the electroplating solution at the time of dosing. Specifically, wherein, is the carbonate concentration of the electroplating solution at the time of dosing, is the carbonate concentration at the initial tank preparation.

[0044] In this relationship, is the total amount of additive in the historical time period at the time of dosing, is the length of the historical time period at the time of dosing, is the historical time period at the time of dosing, The ratio of the total amount of additive and the cumulative change of net conductivity forms a dynamic gain ratio, the numerator is the recent input, reflecting the excitation of the system, and the denominator is the change of net conductivity, reflecting the response of the system. The larger the ratio, the more "medicine is injected, but the conductivity does not change", indicating that the system has high damping or high lag characteristics. Using integral form is more resistant to noise than using instantaneous value, which can smooth out short-term fluctuations.

[0045] In the formula, Part is a correction term based on the change of carbonate concentration. The accumulation of carbonate concentration is a typical feature of aluminum alloy electroplating aging, which will increase the viscosity and hinder ion diffusion. Multiplying the correction term and the dynamic gain ratio forms a composite lag index that combines macro input-output efficiency and micro viscosity resistance. The larger the dynamic gain ratio, the more medicine is injected but the less response, and the larger the relative growth rate of carbonate concentration, the larger the viscosity of the electroplating solution. At this time, it means that the electroplating solution is in a mass transfer blockage and response lag state. The smaller the two, the more responsive the electroplating solution is, and the faster the additive can produce results after being added. The smaller the two, the more responsive the electroplating solution is, and the faster the additive can produce results after being added. The tanh function is used to normalize the physical calculation value which may have a large range of values to the [0, 1) interval. Considering the smooth saturation characteristics of the function can prevent extreme abnormal values from causing factor overflow, while retaining the change trend.

[0046] When the electroplating solution is in good condition, the response is sensitive (the denominator is large), the carbonate concentration is low, and the calculated composite lag index value is small. After tanh mapping, it tends to 0, and the system determines that the electroplating solution has low inertia, meaning "input has output", and the electroplating solution is in a mass transfer state. At this time, the diffusion channel of the additive is unobstructed, the sensor can feedback the concentration change in real time, and the control system can operate according to the normal logic. When the bath deteriorates, the conductivity does not change after dosing, and the impurity accumulation further amplifies the value. tanh quickly saturates to 1, determining that the electroplating solution is in a high inertia state, meaning "more input but less output", and the system is in a mass transfer blockage state. The additive diffusion lags, and the control system needs to be vigilant about the risk of integral saturation.

[0047] The transmission inertia factor of the electroplating solution at the dosing time is no longer dependent on the pre-set static reference, but is completely based on the current dynamic input-output relationship and electrochemical state of the system, achieving real-time and accurate quantification of the lag characteristics.

[0048] This step is a dynamic diagnosis of the electroplating bath in physical essence. When the bath is aging, carbonate ions are enriched, and the concentration is at a high level, which will significantly increase the macroscopic viscosity of the solution, forming a paste-like fluid environment that will hinder the micro Brownian motion and convective diffusion of additive molecules like a damper. When the control system performs dosing operations, When the additive is wrapped by the high-viscosity bath, the cumulative amount of the net conductivity of the electroplating solution is small, and the dynamic gain ratio will expand sharply, and the further amplification of the correction term based on the change in the carbonate concentration will cause the transmission inertia factor to become large and approach 1, which means that the control system is warned that the current conductivity does not remain unchanged because of no addition of the additive, but because the system is in a physically blocked state with extremely high damping, and if the blind addition of the additive continues, serious supersaturation will occur. When the bath is in a good state, the concentration of carbonate ions is low, and the bath remains a low-viscosity fluid environment similar to water, which enables the additive molecules to perform unhindered Brownian motion and efficient convective diffusion. At this time, the additive is not wrapped by the high-viscosity bath, the sensor responds quickly, and under the same addition of the additive, the cumulative amount of the net conductivity of the electroplating solution is large, and the dynamic gain ratio will become small, and the further attenuation of the carbonate correction term will cause the transmission inertia factor to become small and approach 0, which means that the current conductivity remains unchanged because of no addition of the additive or insufficient addition of the additive, and the system mass transfer is normal, and at this time, the addition operation can be normally performed, and the integral term is allowed to accumulate to eliminate the steady-state error.

[0049] S3: Based on the transmission inertia factor, the retardation risk index of the electroplating solution is calculated in combination with the evolution trend of the transmission inertia factor over time.

[0050] After the transmission inertia factor is determined, in order to accurately adjust the integral action of the PID algorithm, this step considers that when the electroplating solution is in a stable high-viscosity state and a transient retardation deterioration state, the transmission inertia factor of the electroplating solution is high. Therefore, this step further analyzes the evolution gradient of the transmission inertia factor in the time dimension, and uses a static reference and a dynamic modulation model to calculate the retardation risk of the electroplating solution at the time of addition of the additive, so as to distinguish whether the electroplating solution is in a stable high-viscosity state or a transient retardation deterioration state.

[0051] Firstly, the transmission inertia factor of each historical addition time in the historical time period at the time of addition of the additive is obtained, and linear regression fitting is performed, which can specifically use the least squares method to obtain a fitting slope to reflect the evolution gradient of the transmission inertia factor. If , it means that the electroplating solution is in a continuous retardation deterioration state, if , it means that the electroplating solution is in a continuous improvement state of diffusion recovery, and if , it means that the electroplating solution is in a stable equilibrium state.

[0052] Then, the retardation risk index is calculated as follows:

[0053] wherein, is the retardation risk index of the electroplating solution at the moment of dosing, is the transport inertia factor of the electroplating solution at the moment of dosing, is the sign function, if , if , if , , is a non-linear normalization function, used to map to the range 0 to 1, specifically the hyperbolic tangent function , is the absolute value sign.

[0054] In this relation, reflects the magnitude of the evolution gradient of the transport inertia factor, reflects the direction of the evolution gradient of the transport inertia factor, reflects the reference state of the electroplating solution.

[0055] If the transport inertia factor is high, but the evolution gradient direction is almost non-existent, it means that the viscosity of the electroplating solution at the moment of dosing is high, but the trend has been stable in the historical time period, and there has been no instantaneous sudden retardation. It means that at the moment of dosing, the electroplating solution is more likely to be in a stable high viscosity state, and the viscosity of the electroplating solution is high due to long-term use, which is an inherent characteristic. At this time, becomes , the retardation risk index strictly degenerates into the transport inertia factor at the moment of dosing, which is dominated by the transport inertia factor. The larger the transport inertia factor, the larger the retardation risk index. For the PID control algorithm, the strategy to be adopted in this case is to maintain a moderate integral confidence. Since the electroplating solution is in a steady state, the deviation of the conductivity truly reflects the gap in concentration, but due to the high background viscosity, the mass transfer is slow. At this time, by setting the retardation risk index equal to the transport inertia factor, a medium-low integral confidence is maintained, allowing the integral term to slowly accumulate the amount of supply, which can ensure that the target concentration is finally reached, and can also avoid the instantaneous excess caused by viscosity.

[0056] If the transport inertia factor is high, and the evolution gradient direction is positive, This indicates that not only the viscosity of the plating solution at the moment of dosing is large, but also the trend is deteriorating over the historical period, and the response ability of the plating solution to the additive is showing a continuous downward trend, and the response of the plating solution to the additive is deteriorating sharply, and the input of the additive increases but does not produce effect, which is usually due to the large amount of colloid-like by-products generated in the bath or the failure of the filtration system, resulting in a sharp rise in the viscosity of the bath, and the ion migration channel is blocked, at this time, , the amplification effect of is realized, and at this time the retardation risk index is greater than the transmission inertial factor. For the PID control algorithm, in this case, the strategy to be adopted is to greatly reduce the integral confidence or even completely block the integral, at this time the conductivity deviation is not reliable, and if the integral is allowed, the algorithm will mistakenly think that the additive is too little and will accumulate the output crazily, by amplifying the retardation risk index, the integral confidence is forced to approach 0, and the integral action is frozen at the physical level until the drug diffusion and gradient fall, thereby avoiding integral saturation.

[0057] If the evolution gradient direction is negative, , it indicates that the transmission inertial factor is decreasing over the historical period, and the response ability of the plating solution is recovering, although the transmission inertial factor at the moment of dosing may still be high, but due to the improving trend over the historical period, if only the transmission inertial factor at the moment of dosing is considered, it is easy to lead to a slow response to replenishment. Therefore, the improving trend is used to offset the high base inertia to release the integral ability in advance, at this time , the reduction effect on is realized, making it lower than the transmission inertial factor at the moment of dosing. For the PID control algorithm, in this case, the strategy to be adopted is to improve the integral confidence, which enables the PID control algorithm to predict the recovery of the plating solution and allows the integral term to intervene quickly to compensate for the concentration deficit accumulated during the previous retardation period, thereby shortening the adjustment time of the system.

[0058] S4: Reconstruct the integral term of the PID control algorithm in real time using the retardation risk index, and obtain the replenishment amount of the additive based on the reconstructed PID control algorithm.

[0059] The traditional PID algorithm is prone to integral saturation under lagging conditions, essentially because the algorithm unconditionally trusts the conductivity deviation feedback by the sensor. This step uses the retardation risk index to calculate a dynamic integral confidence for real-time adjustment of the participation degree of the integral term.

[0060] First, a nonlinear decay function is used to map the retardation risk index to a value in the interval of 0 to 1 as an integral confidence. Specifically, the following relationship is satisfied:

[0061] wherein, is the integral confidence at time t, representing the effectiveness of the control layer, with a value range of (0, 1], is a natural exponential function, is the retardation risk index of the electroplating solution at time t, representing the degree of unreliability of the physical layer.

[0062] By using , any retardation risk index can be smoothly mapped to the dimensionless interval of (0, 1], compared to the linear function, , when is small, the function curve is gentle, , maintaining a high level close to 1, ensuring high sensitivity of the system under normal working conditions, and once the retardation risk index exceeds the critical value, the function curve drops sharply, , quickly falling to 0, this nonlinear mechanism ensures that the integral term is not blindly accumulated under unreliable signals in the moment when serious retardation is detected. Then, the integral confidence is used to weight the integral term of the PID control algorithm to obtain a reconstructed PID control algorithm, and the replenishment amount of the additive is obtained based on the real-time reconstructed PID control algorithm. The larger the retardation risk index, the more likely the electroplating solution is in a retardation deterioration state at the time of adding the drug, which is mapped to a smaller integral confidence by a negatively correlated exponential function, which means that the PID control algorithm no longer trusts the current deviation signal, considering it a false appearance caused by lag, and the error accumulation of the integral term is frozen, avoiding the integral saturation effect, leading to excessive addition of drugs; the smaller the retardation risk index, the more likely the electroplating solution is in a stable state or improving state at the time of adding the drug, which is mapped to a larger integral confidence by a negatively correlated exponential function, which means that the PID control algorithm trusts the current deviation signal, considering it a real state change of the electroplating solution, at which time the error accumulation of the integral term is restored, which can quickly eliminate steady-state error and ensure accurate addition of drugs to maintain the net conductivity.

[0063] Through this confidence-weighted integral mechanism, the system can automatically suspend integration in the blind area of sensor response lag, and restore integration when the response recovers or stabilizes, achieving adaptive control of complex electrochemical working conditions.

[0064] In order to output the final dosing machine control instruction, this physical constraint must be returned to the execution link of the control algorithm, and this step aims to use the integral confidence to real-time reconstruct the integral link of the traditional PID control algorithm, outputting a final control quantity that can quickly respond and avoid component degradation lag overshoot.

[0065]

[0066] ​​Specifically, the amount of additive to be supplied, obtained from the real-time reconstructed PID control algorithm, is determined based on the following relationship:

[0067] in, This refers to the amount of additives to be replenished in the electroplating solution at the time of chemical addition. These are the proportional gain constant, integral gain constant, and derivative gain constant of the PID control algorithm, which are usually set based on empirical values. , for The deviation between the net conductivity at a given time and the preset standard conductivity is specifically the preset standard conductivity minus the net conductivity at that time. A positive deviation indicates a lack of additives, requiring pump replenishment. The PID control algorithm outputs a positive replenishment amount, and vice versa. yes The integral confidence level at time step. This is the derivative of the deviation with respect to time. This indicates that integration is performed over time.

[0068] in, For the proportional term, it provides basic corrective force. The differential term is used to predict trends and suppress oscillations. For the integral term, by introducing Dynamic start-stop is achieved through integral action, whereas the integral term of the traditional PID control algorithm is a function of the deviation. direct integral This invention integrates confidence levels. When injected into the integral term, and the electroplating solution is in a state of deteriorating resistance, the resistance risk index is very high, leading to... Extremely small, even close to zero, the accumulation rate of the integral term is slowed down or even stopped at a physical level, even if the error is at this point. It persists, and the integral terms will not accumulate indefinitely; this is the control quantity. It will not surge blindly, thus avoiding integral saturation; when the electroplating solution is in diffusion recovery or steady state, the resistance risk index is very small, leading to... With the integral term's accumulation function restored to normal, it can quickly respond and eliminate deviations, thereby ensuring that the conductivity accurately returns to the set target value.

[0069] In this relation, , is the standard conductivity, which represents the plating solution at the best component concentration, i.e. the theoretical conductivity at the standard temperature with the best additive and main salt concentration as specified in the initial bath preparation or standard process formula, and is obtained by first preparing a fresh plating solution standard sample with accurate component content according to the standard formula of the plating process, placing the standard sample in a constant temperature environment, controlling the temperature to be stable at the standard temperature, and measuring the conductivity value at this state using a high-precision conductivity meter, and then setting the value as the standard conductivity. In actual control, is a static reference, and the target of the PID control algorithm is to make the net conductivity after temperature decoupling always approach this static reference by adding additives, is the net conductivity at the moment, represents the magnitude of the control requirement, and represents the credibility of the deviation at the time of adding the additive, and in the integral term , is an error term responsible for providing direction and size, telling the controller whether to add or reduce the additive and how much the gap is, is a weight term responsible for determining whether the error should be accumulated. If the plating solution is in a state of poor mass transfer, the transmission inertia factor is large, and the deviation

[0070] at this time is often a false sustained error due to not being diffused, and tends to 0, which means that the error is not qualified to enter the integral accumulation, and the deviation is not counted in the integral accumulation, which means that the integral action is automatically suspended during physical blockage, preventing integral oversaturation. If the plating solution is in a good mass transfer state, the transmission inertia factor is small, and the deviation at this time is often a real sustained error due to the diffusion of the additive, and tends to 1, which means that the error is qualified to enter the integral accumulation, and the integral term resumes normal accumulation and performs regular control. This mechanism mathematically eliminates the integral saturation phenomenon caused by the sustained error during sensor response lag, ensuring that the additive supply is only driven by the effective error. It should be noted that since the additive metering pump is usually a one-way actuator that can only perform the additive action, if the calculated supply amount is negative, it means that the current net conductivity is already higher than the standard conductivity, and the additive action is stopped until the supply amount returns to positive.

[0071] S5: automatically supply the plating solution based on the supply amount.

[0072] S5: automatically supply the plating solution based on the supply amount.

[0073] After obtaining the replenishment amount at the time of chemical addition, the additive replenishment device is driven to release the additive into the electroplating solution according to the replenishment amount.

[0074] Specifically, additive replenishment equipment is typically a metering pump. The replenishment amount is mapped to a control signal for the metering pump, and the replenishment control amount is determined based on the type of pump. Mapped to one of the following two control signals: one is duty cycle control mode: suitable for electromagnetic diaphragm metering pumps, with a fixed control cycle of 30 seconds set. The linear mapping is the on-time within that cycle, for example, if If the value is 50%, the pump is controlled to start for 15 seconds and stop for 15 seconds within the cycle; another is frequency control mode: suitable for metering pumps driven by variable frequency motors, which... The linear mapping is to the pump's stroke frequency. For example, if the maximum frequency of the pump is set to 120 strokes per minute, then... If the value is 50%, the output control command adjusts the pump to run at a frequency of 60 times / minute. Based on the control signal, the flow rate of the dosing metering pump is controlled to be the same as the replenishment amount, so as to realize the replenishment operation of the additive.

[0075] In summary, this control logic based on physical state reshaping overcomes the specific defect of PID control algorithms in time-varying component deterioration baths, which leads to excessive replenishment due to response lag, thus ensuring the uniformity and accuracy of electroplating.

[0076] To more intuitively illustrate the effects of the present invention, such as Figure 2 As shown, this paper compares the dynamic control response of the reconstructed PID control algorithm of this invention with that of the existing PID control algorithm under the typical condition of component deterioration due to long-term use of electroplating solution. The horizontal axis represents the control time, and the vertical axis represents the normalized relative concentration of additive. The target value is set to 1. In the initial stage of control, due to the aging of the plating solution, the mass transfer lag is caused by the aging of the plating solution, and the change in conductivity fed back by the sensor is not obvious. For the existing technology, the traditional PID control algorithm cannot identify the physical lag and misjudges it as insufficient dosage, which leads to the continuous accumulation of false error signals in the integral term and integral saturation. When the reagent finally diffuses and takes effect, the accumulated excessive control command is released instantly, causing the concentration to rise sharply to above 1.4 and fall into the high-risk zone. However, the present invention, through the reconstruction algorithm, automatically reduces the integral confidence and actively suppresses the blind accumulation of the integral term during the physical lag period. When the lag ends (about 30 cycles later), the system smoothly transitions to steady-state control and maintains reasonable robustness in following high-frequency sensor noise and small environmental disturbances. The comparative results show that the present invention effectively solves the problem of integral saturation caused by the deterioration of electroplating solution components, eliminates overshoot phenomenon, and ensures that the additive concentration is always within the safe process window.

[0077] The above merely provides the preferred embodiment of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. An automatic replenishment control method for electroplating bath additives based on conductivity monitoring data, characterized in that, include: The temperature, initial conductivity, and carbonate concentration of the electroplating solution are obtained. The activation energy coefficient of the electroplating solution is determined based on the changing trends of the initial conductivity and temperature, so as to reflect the sensitivity of ion migration in the electroplating solution to temperature changes. The original conductivity is subjected to dynamic thermodynamic decoupling treatment to eliminate the influence of temperature fluctuations on conductivity, thereby generating the net conductivity of the electroplating solution. Based on the cumulative change in net conductivity over a preset historical period and the total amount of additives during that historical period, combined with the carbonate concentration, the transport inertia factor of the electroplating solution is calculated to reflect the degree of lag in the electroplating solution's response to additive replenishment. The electroplating solution is controlled by PID control algorithm to replenish additives. Based on the transport inertia factor, the resistance risk index of the electroplating solution is calculated by combining the evolution trend of the transport inertia factor over time. The integral term of the PID control algorithm is reconstructed in real time using the aforementioned resistance risk index, and the amount of additive replenishment is obtained based on the reconstructed PID control algorithm. The electroplating solution is then automatically replenished based on the replenishment amount.

2. The automatic replenishment control method according to claim 1, characterized in that, The activation energy coefficient of the electroplating solution is determined based on the following method: At each chemical addition point in the electroplating process, the natural logarithm of the original conductivity and the reciprocal of the temperature at all historical chemical addition points within a preset historical time period are obtained. The natural logarithm and reciprocal of the temperature at all historical chemical addition points are then fitted using the least squares method to obtain the slope of the fitted line. Based on the slope, the activation energy coefficient at the chemical addition point is determined using the Arrhenius law.

3. The automatic replenishment control method according to claim 1, characterized in that, The net conductivity of the electroplating solution is determined based on the following method: ;in, In preparation for administering the medication, The net conductivity of the electroplating solution at the moment before chemical addition is applied. The original conductivity of the electroplating solution at the moment before chemical addition. It is a natural exponential function. The activation energy coefficient of the electroplating solution at the moment of chemical addition is given. Let be the ideal gas constant. The temperature of the electroplating solution at the moment the chemical is to be added. This is the preset standard temperature.

4. The automatic replenishment control method according to claim 1, characterized in that, The transport inertia factor of the electroplating solution is determined based on the following method: ； in, To provide an inert factor for the electroplating solution at the moment of chemical addition. It is the hyperbolic tangent function. This is a preset dimensional balance coefficient used to convert the calculation terms within parentheses into dimensionless values. The length of the historical time period during which medication is to be administered. For integration variables, and They are respectively The rate of change of drug dosage and net conductivity at any given time. This is a preset measure to prevent positive numbers with a denominator of 0. This represents the relative growth rate of carbonate concentration in the electroplating solution at the time of chemical addition. The weighting coefficient is the effect of carbonate concentration on the viscosity of the electroplating solution, which was determined in advance through experiments.

5. The automatic replenishment control method according to claim 4, characterized in that, The carbonate concentration at the time of drug addition is determined by obtaining it from an online ion concentration analyzer or by receiving periodic test data input through a human-computer interaction interface.

6. The automatic replenishment control method according to claim 4, characterized in that, The relative growth rate of carbonate concentration in the electroplating solution at the time of chemical addition is determined by first obtaining the carbonate concentration of the electroplating solution at the initial tank preparation, calculating the relative increment of carbonate concentration at the time of chemical addition and carbonate concentration at the initial tank preparation, and taking the ratio of the relative increment to carbonate concentration at the initial tank preparation as the relative growth rate of carbonate concentration in the electroplating solution at the time of chemical addition.

7. The automatic replenishment control method according to claim 1, characterized in that, The resistance risk index of electroplating solutions is determined based on the following method: The transport inertia factor for each historical dosing moment within the historical time period to be dosing is obtained, and a linear regression is performed to obtain a fitting slope. This reflects the evolution gradient of the transmission inertia factor; Calculate the risk index of obstruction: ;in, This represents the resistance risk index of the electroplating solution at the moment of chemical addition. To provide an inert factor for the electroplating solution at the moment of chemical addition. For symbolic functions, It is a nonlinear normalization function. It is the absolute value symbol.

8. The automatic replenishment control method according to claim 7, characterized in that, The integral term of the PID control algorithm is reconstructed in real time using the aforementioned stall risk index, including: The stagnation risk index is mapped to a value between 0 and 1 using a nonlinear decay function, which serves as an integral confidence level. This integral confidence level is then used to weight the integral term of the PID control algorithm to obtain the reconstructed PID control algorithm.

9. The automatic replenishment control method according to claim 8, characterized in that, The amount of additive to be supplied, obtained from the real-time reconstructed PID control algorithm, is determined based on the following relationship: ;in, This refers to the amount of additives to be replenished in the electroplating solution at the time of chemical addition. These are the proportional gain constant, integral gain constant, and derivative gain constant of the pre-set PID control algorithm. for The deviation between the net conductivity at a given time and the preset standard conductivity. for The integral confidence level at time step. This is the derivative of the deviation with respect to time. It involves integrating over time.

10. The automatic replenishment control method according to claim 1, characterized in that, Automatic replenishment control of the electroplating solution based on the replenishment amount includes: when the chemical is to be added, driving the additive replenishment device to release the additive into the electroplating solution according to the replenishment amount.

Images