Rare earth metal electrolytic reduction method based on high-frequency switching power supply

Abstract

The invention discloses a rare earth metal electrolytic reduction method based on a high-frequency switching power supply, and relates to the technical field of rare earth metal electrolytic reduction, and the method comprises the following steps: firstly, in a praseodymium-neodymium electro-deposition process, parameter data in the praseodymium-neodymium electro-deposition process is accurately obtained in real time by using an advanced sensor and detection equipment; and accurate, comprehensive and real-time data support is provided for subsequent intelligent analysis. According to the invention, the problems of stress accumulation, crack propagation and coating stripping caused by fixed preset intermittent closing time in high-current density electrodeposition are solved. Intelligent optimization is realized through real-time data acquisition, deep learning evaluation and dynamic pulse current regulation and control. And when the stress rise is detected, the system adaptively prolongs the intermittent closing time, and reduces the pulse peak current, so that the deposition rate is optimized, and the uniformity and compactness of a plating layer are improved. The method improves the mechanical strength and corrosion resistance of the coating, prolongs the service life of equipment, reduces the energy consumption, and improves the process stability and consistency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rare earth metal electrolytic reduction, specifically to a rare earth metal electrolytic reduction method based on high-frequency switching power supply. BACKGROUND

[0002] Rare earth metal electrolytic reduction based on high-frequency switching power supply is a process method that uses high-frequency switching power supply as the electrolysis energy source to reduce rare earth oxides or other rare earth compounds to metal through electrochemical reaction. In this process, high-frequency switching power supply can provide efficient, stable and adjustable direct current, so that the current density and voltage in the electrolytic cell can be optimized, thereby improving the electrolysis efficiency, reducing energy consumption, and improving the quality of metal deposition. Compared with traditional power frequency power supply, high-frequency switching power supply has higher power utilization rate, can reduce electrode polarization effect and local overheating phenomenon of electrolyte, improve current efficiency and reduce metal inclusion content, thereby improving the purity and yield of the final obtained rare earth metal. This technology is widely used in rare earth metal smelting industry, such as electrolytic extraction of praseodymium-neodymium, dysprosium, gadolinium and other metals, to meet the needs of high-end material manufacturing.

[0003] The prior art has the following disadvantages: High-frequency switching power supply can realize pulse current mode, that is, the current reaches a high set value in a short time and then is intermittently turned off. The prior art usually pre-sets the intermittent off time according to the electrolysis process requirements and system control strategy to perform power-off operation, so as to optimize the metal deposition quality, improve the current efficiency, and reduce the interference of side reactions. However, in the process of electrodeposition of praseodymium-neodymium, if the current density is too high and the deposition rate is too fast, it will cause the internal stress (including tensile stress and compressive stress) of the metal layer to be too large, thereby forming micro-cracks, affecting the mechanical properties and durability of the plated layer. If the power-off operation is continued using the fixed pre-set intermittent off time without dynamic adjustment according to the actual situation, under the conditions of high current density and fast deposition, serious quality problems may be caused, and even the entire plated layer may fail or the material may be scrapped. Since the stress cannot be released in time, the cracks will gradually expand and penetrate each other, eventually leading to large-area cracks or even peeling of the plated layer. This phenomenon not only seriously affects the appearance of the plated layer, but also greatly reduces the corrosion resistance and mechanical strength of the material, causing structural failure of the electroplated parts during actual use, and even causing equipment damage, resulting in irreversible economic losses.

[0004] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] The purpose of the present application is to provide a rare earth metal electrolytic reduction method based on high-frequency switching power supply, which solves the problems of stress accumulation, crack propagation and coating peeling caused by fixed preset intermittent closing time in high current density electrodeposition. Through real-time data acquisition, deep learning evaluation and dynamic regulation of pulse current, intelligent optimization is realized. When stress is detected to rise, the system adaptively prolongs the intermittent closing time and reduces the pulse peak current to optimize the deposition rate, improve the uniformity and density of the coating. This method improves the mechanical strength and corrosion resistance of the coating, prolongs the service life of the equipment, reduces energy consumption, and improves the process stability and consistency to solve the problems in the above background art.

[0006] To achieve the above purpose, the present application provides the following technical solution: a rare earth metal electrolytic reduction method based on high-frequency switching power supply, comprising the following steps: First, in the praseodymium-neodymium electrodeposition process, advanced sensors and detection equipment are used to accurately obtain the parameter data in the praseodymium-neodymium electrodeposition process in real time, providing accurate, comprehensive and real-time data support for subsequent intelligent analysis; The real-time collected parameter data are integrated to establish a data set, and the data set is preprocessed; Based on the high-quality data after preprocessing, feature engineering method is used to extract feature variables reflecting the internal stress state and change trend of the coating, and the extracted feature variables are analyzed in the detection window to quantify the change trend and severity of the internal stress of the praseodymium-neodymium layer in the current deposition process; The analyzed feature variables are input into the pre-trained deep learning model for forward reasoning to evaluate the internal stress state of the praseodymium-neodymium layer in real time; When the internal stress of praseodymium-neodymium is identified to rise in the electrodeposition process, the intermittent closing time is first dynamically prolonged to allow the re-adjustment of the atomic structure and the relaxation release of the lattice stress in the coating, reduce the accumulation degree of local stress in the coating, and at the same time, the pulse peak current is reduced to reduce the number of praseodymium-neodymium atoms deposited on the surface of the coating in a single deposition period.

[0007] Preferably, in the praseodymium-neodymium electrodeposition process, the parameter data in the deposition process are accurately obtained in real time as follows: First, multiple types of sensors are arranged inside and outside the electrolytic cell, including but not limited to current sensors, voltage sensors, temperature sensors, pH sensors, ion concentration sensors and laser interference sensors to comprehensively cover the key process parameters; Secondly, a high-frequency data acquisition system is used to synchronously collect the data of all sensors to ensure the time consistency of the key parameters, and an analog-to-digital converter is used to convert the analog signals into high-precision digital signals to ensure the accuracy and stability of the data; Then, the collected data is transmitted to a central data server in real time by using wireless communication technology, ensuring smooth and efficient data transmission. Finally, the stored data is used for subsequent intelligent analysis and deep learning model calculation, providing accurate, comprehensive, and real-time data support for dynamically optimizing the praseodymium-neodymium electrodeposition process, ensuring that the entire system can operate efficiently and stably under different process conditions.

[0008] Preferably, based on the pretreated high-quality data, feature engineering methods are used to extract feature variables reflecting the stress state and change trend of the coating. The extracted feature variables include the local charge distribution on the cathode surface and the formation and detachment of bubbles. The local charge distribution on the cathode surface and the formation and detachment of bubbles are analyzed under the detection window to generate a cathode surface charge accumulation reference value and a bubble residue reference value, respectively. The cathode surface charge accumulation reference value and the bubble residue reference value are used to quantify the change trend and severity of the internal stress of the coating during the praseodymium-neodymium deposition process.

[0009] Preferably, the specific steps for analyzing the local charge distribution on the cathode surface under the detection window to generate the cathode surface charge accumulation reference value are as follows: First, under the detection window, the local charge density distribution at different spatial positions on the cathode surface is obtained by using a high-precision current density sensor and an electric field measurement device. The charge distribution on the cathode surface is not uniform, but is affected by the electric field gradient, ion diffusion in the solution, local polarization, and the micro-geometric structure of the electrode surface. Therefore, the local charge gradient variation is defined to measure the spatial variation degree of the charge distribution, and the calculation expression is as follows: , where is the local charge density on the cathode surface, representing the local charge density on the cathode surface at coordinates , and are the second-order variation rates of the charge density on the cathode surface in the and directions, i.e., the local curvature of the charge density distribution, is the cathode surface potential, representing the local potential on the cathode surface at coordinates , and represent the gradient changes of the cathode surface potential along the and directions, is the weight factor, is the local charge gradient variation; The local charge gradient variation After that, the local charge accumulation potential energy function is introduced to measure the overall charge unevenness of the cathode surface, and finally the cathode surface charge accumulation reference value is defined, and the calculation expression is as follows: , wherein, is the cathode surface charge accumulation reference value, is the normalization coefficient, is the spatial range of the cathode surface for analysis, is the adjustment factor, is the potential decay factor, is the natural base, is the potential decay coefficient, is the infinitesimal area.

[0010] Preferably, the formation and detachment of bubbles are analyzed under the detection window to generate the specific steps of the bubble residue reference value as follows: First, the high-resolution optical information of the cathode surface is continuously collected, and the two-dimensional projection profile of all bubbles is identified through image processing technology, and then the radius information of each bubble is obtained, and the bubble size distribution function is defined to describe the distribution state of different bubble radii on the cathode surface, and the formula is as follows: , wherein, is the bubble size distribution function, is the continuous variable of the bubble radius, is the total number of bubbles, is the actual radius of the th bubble collected, is the local distribution width parameter around the th bubble radius, is the adjustment factor, is the scale factor for adjusting the normalization of bubble size, is a very small positive number; By analyzing the distribution of bubble positions on the electrode surface, a heterogeneity index reflecting the unevenness of the spatial distribution of bubbles is generated, and the spatial heterogeneity index generation formula is as follows: , wherein, is the spatial heterogeneity index, is the local density value of the th bubble on the cathode surface, is the local density value of the th bubble on the cathode surface, is the Euclidean distance between the th bubble and the th bubble, is the spatial distance adjustment factor, is the local distribution adjustment factor, is a bubble density nonlinear adjustment factor; comprehensive bubble size distribution function and spatial heterogeneity index , construct a bubble residual reference value to comprehensively characterize the degree of influence of bubbles on the metal deposition process, and generate the formula as follows: , wherein, is a bubble residual reference value, is a correlation factor of bubble size and spatial heterogeneity distribution, is the maximum value of bubble radius, is the minimum value of bubble radius, is an adjustment parameter, is a super-curvature adjustment factor, is a super-curvature sine function term.

[0011] Preferably, the analyzed cathode surface charge accumulation reference value and the bubble residual reference value are input into the pre-trained deep learning model to generate a stress change risk coefficient, and the stress change risk coefficient is used to evaluate the internal stress state of the praseodymium-neodymium layer in real time online.

[0012] Preferably, the stress change risk coefficient generated when the internal stress state of the praseodymium-neodymium layer is evaluated in real time online by the pre-trained deep learning model is compared and analyzed with a pre-set stress change risk coefficient reference threshold value, and the internal stress state of the praseodymium-neodymium layer is divided, and the division steps are as follows: If the stress change risk coefficient is greater than the pre-set stress change risk coefficient reference threshold value, the internal stress of the praseodymium-neodymium layer is divided into praseodymium-neodymium layer internal stress rise; if the stress change risk coefficient is less than or equal to the pre-set stress change risk coefficient reference threshold value, the internal stress of the praseodymium-neodymium layer is divided into praseodymium-neodymium layer internal stress normal and stable.

[0013] Preferably, when the internal stress of praseodymium-neodymium is identified to rise in the electrodeposition process, the intermittent closing time is first dynamically extended, and the specific steps of the pulse peak current are as follows: When the internal stress of praseodymium-neodymium is identified to rise in the electrodeposition process, the intermittent closing time is dynamically extended, and the extension range needs to be based on the current actual stress risk degree, and the calculation expression is as follows: , wherein, is the dynamically adjusted intermittent closing time, is the basic intermittent closing time, is a stress relaxation increment coefficient, is a stress change risk coefficient, is a stress change risk coefficient reference threshold value, is a nonlinear index, is a very small positive number, is an exponential decay term, is an exponential suppression factor, is the natural base number; At the same time, combined with the stress change risk coefficient The real-time change degree, the pulse peak current in the single deposition period is reduced in linkage, so as to reduce the number of praseodymium and neodymium atoms deposited on the plating layer surface in each period, and the specific pulse current adjustment amount calculation expression is as follows: , wherein, is the pulse peak current adjusted dynamically, is the standard pulse peak current set as the basis, is the current regulation sensitivity coefficient, is the logarithmic term in the function, is the function term, is the steepness factor, is an exponential suppression factor, used to balance the current adjustment amplitude, avoid excessive current reduction affecting the deposition efficiency, is an exponential balance factor.

[0014] In the above technical solution, the technical effects and advantages provided by the present application are: The present application effectively solves the quality problems of internal stress accumulation, crack propagation and plating layer peeling caused by traditional fixed preset intermittent closing time under high current density process conditions. Through a series of steps such as real-time data acquisition, intelligent data processing, feature extraction, deep learning evaluation and dynamic regulation of pulse current, the present application realizes intelligent and precise control of the electrodeposition process. When the internal stress of the plating layer is abnormally high, the system can adaptively prolong the intermittent closing time, so that the plating layer structure is fully adjusted to prevent excessive stress accumulation, and the pulse peak current is reduced in linkage to optimize the deposition rate and improve the uniformity and density of the plating layer. Compared with the traditional process, the present application not only significantly improves the mechanical strength and corrosion resistance of the plating layer, but also prolongs the service life of the equipment, reduces the electrode consumption, reduces the production energy consumption, and further improves the stability and consistency of the plating layer quality. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments or prior art of the present application, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0016] Figure 1 The method flow chart of the present application based on the high-frequency switching power supply rare earth metal electrolytic reduction method. DETAILED DESCRIPTION

[0017] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the inventive gist to those skilled in the art.

[0018] The present application provides a rare earth metal electrolytic reduction method based on a high-frequency switching power supply as shown in the figure, comprising the following steps: Figure 1 The rare earth metal electrolytic reduction method based on a high-frequency switching power supply as shown in the figure comprises the following steps: First, in the praseodymium-neodymium electrodeposition process, advanced sensors and detection equipment are used to accurately obtain the parameter data in the praseodymium-neodymium electrodeposition process in real time, providing accurate, comprehensive, and real-time data support for subsequent intelligent analysis; In the praseodymium-neodymium electrodeposition process, the specific steps of accurately obtaining the parameter data in the deposition process in real time are as follows: first, multiple types of sensors are arranged inside and outside the electrolytic cell, including but not limited to current sensors (monitoring current density changes), voltage sensors (measuring cathode potential), temperature sensors (monitoring electrolyte temperature), pH sensors (detecting electrolyte acidity and alkalinity), ion concentration sensors (monitoring praseodymium-neodymium ion concentration changes), and laser interference sensors (measuring coating thickness and growth rate) to comprehensively cover key process parameters; second, a high-frequency data acquisition system (DAQ, Data Acquisition System) is used to synchronously collect data from all sensors, ensuring the time consistency of key parameters, and an analog-to-digital converter (ADC) is used to convert analog signals into high-precision digital signals, ensuring the accuracy and stability of the data; then, wireless communication technology (such as 5G, Wi-Fi, LoRa) is used to transmit the collected data in real time to a central data server, ensuring smooth and efficient data transmission; finally, the stored data is used for subsequent intelligent analysis and deep learning model calculation, providing accurate, comprehensive, and real-time data support for dynamic optimization of the praseodymium-neodymium electrodeposition process, and ensuring that the entire system can operate efficiently and stably under different process conditions.

[0019] The parameter data includes but is not limited to cathode current density, pulse current waveform, voltage change, deposition rate, electrode surface potential, electrolyte temperature, metal ion concentration, electrolyte pH value, and coating thickness change trend data. These parameters are continuously collected by the high-frequency data acquisition system, and the original multi-modal data set is formed by real-time recording. The specific role of this step is to provide accurate, comprehensive, and real-time data support for subsequent intelligent analysis, ensuring the reliability and accuracy of the model analysis results.

[0020] The real-time collected parameter data is centrally integrated to establish a data set, and the data set is preprocessed; The data collected in real time often has noise, abnormal data or missing values, and the data set must be pre-processed, including data denoising (such as filtering, smoothing), abnormal value detection and removal, missing value interpolation (such as interpolation method or prediction model method), data normalization and standardization processing, etc. Data preprocessing can significantly improve data quality, reduce the difficulty of subsequent feature extraction, improve the accuracy and effectiveness of feature extraction, and lay a solid foundation for accurately identifying key features in the electrodeposition process. Based on the high-quality data after preprocessing, the feature engineering method is used to extract the feature variables reflecting the internal stress state and change trend of the coating, and the extracted feature variables are analyzed under the detection window to quantify the change trend and severity of the internal stress of the coating in the current praseodymium-neodymium deposition process. Based on the high-quality data after preprocessing, the feature engineering method is used to extract the feature variables reflecting the internal stress state and change trend of the coating, and the extracted feature variables include the local charge distribution on the cathode surface and the formation and detachment of bubbles. The local charge distribution on the cathode surface and the formation and detachment of bubbles are analyzed under the detection window to generate the cathode surface charge accumulation reference value and the bubble residue reference value, respectively. The cathode surface charge accumulation reference value and the bubble residue reference value are used to quantify the change trend and severity of the internal stress of the coating in the current praseodymium-neodymium deposition process.

[0021] The uneven local charge distribution on the cathode surface indicates that the internal stress of the metal layer is increasing, and the main reasons involve the microscopic process of metal deposition, the influence of electric field distribution on the structure of the coating, and the change of electrochemical polarization. When the charge distribution on the cathode surface is uneven, the electric field intensity in the local area will change, leading to accelerated deposition rate of metal ions in some areas, while slower deposition rate in other areas. This deposition unevenness will directly affect the microcrystalline grain orientation and growth pattern of the coating, causing mismatch of lattice arrangement in different regions, forming internal stress. In addition, uneven charge distribution may also lead to local overpotential increase, causing enhanced side reactions (such as hydrogen evolution reaction), resulting in bubbles or micropore defects in the coating, which become stress concentration points, causing local areas to bear greater tensile or compressive stress, thus exacerbating the accumulation of internal stress in the coating. In the long run, the continuous increase of internal stress may lead to the initiation and propagation of microcracks, reducing the mechanical strength and durability of the coating. Therefore, the unevenness of local charge distribution on the cathode surface is an important signal indicating that the internal stress of the metal layer is increasing, and appropriate current regulation or intermittent optimization must be performed to improve the coating quality and avoid coating failure.

[0022] The specific steps of analyzing the local charge distribution on the cathode surface under the detection window to generate the cathode surface charge accumulation reference value are as follows: Firstly, the local charge density distribution of different spatial positions on the cathode surface is obtained by high-precision current density sensors and electric field measurement devices (such as scanning Kelvin probe or electrochemical impedance spectroscopy) under the detection window. The cathode surface charge distribution is not uniform, but is affected by the electric field gradient, ion diffusion in the solution, local polarization, and the micro-geometric structure of the electrode surface. Therefore, the local charge gradient change is defined to measure the spatial variation degree of the charge distribution, and the calculation expression is as follows: , wherein, is the local charge density of the cathode surface, indicating the local charge density of the cathode surface at the coordinate and are the second-order change rates of the charge density of the cathode surface in the and directions, that is, the local curvature of the charge density distribution, is the local potential of the cathode surface, indicating the local potential (electric potential) of the cathode surface at the coordinate and indicate the gradient changes of the cathode surface potential along the and directions, is a weight factor, adjusting the relative contribution of the charge density gradient term and the potential gradient cross term in the overall gradient calculation, is the local charge gradient change, used to measure the variation degree of the charge distribution of the cathode surface at different spatial positions; The purpose of this step is to quantify the non-uniformity of the charge density change of the cathode surface. If the charge density change in a certain area is severe (high second-order derivative) or the local electric field gradient is large (high first-order product term), it means that there is a significant charge accumulation phenomenon in this area. Such unevenness of charge gradient is often related to abnormal local deposition rate and is an important precursor of stress increase in the metal layer.

[0023] After obtaining the local charge gradient change , the local charge accumulation potential function is introduced to measure the overall charge non-uniformity of the cathode surface, and the cathode surface charge accumulation reference value is finally defined, and the calculation expression is as follows: , wherein, is the cathode surface charge accumulation reference value, is a normalization coefficient, used to normalize the dimensionless coefficient of so that the calculated is in a reasonable numerical range for comparison under different working conditions, is the spatial range of the cathode surface for analysis, is an adjustment factor, used to adjust the overall​​ Calculate the impact, is the potential attenuation factor, which is used to describe the nonlinear weight of the local charge influence when the potential is large. is the natural base, Is the potential attenuation coefficient, which determines the larger the potential is. The decay rate of the contribution, Is the microelement area, representing the unit area element calculated during the integration process, used to calculate the entire detection window area Perform spatial integration.

[0024] The purpose of this step is to spatially integrate the local charge gradient change and calculate the overall charge accumulation intensity of the entire cathode surface. A greater weight is given to high potential areas because in these areas, charge accumulation is more likely to lead to increased local overpotential, which in turn affects the uniformity of the coating and the internal stress distribution. The larger it is, the more uneven the charge distribution is, the higher the stress inside the metal layer is, and the greater the risk of cracking or peeling; A smaller value indicates that the charge distribution is more uniform, the metal layer deposition is more stable, and the stress level is within the normal range.

[0025] The larger the cathode surface charge accumulation reference value generated after analyzing the local charge distribution on the cathode surface under the detection window, the higher the degree of charge accumulation in the local area of ​​the cathode surface and the more uneven the electric field distribution, which in turn leads to an imbalance in the deposition rate of metal ions in different areas. This deposition unevenness can cause a mismatch in the grain arrangement of the coating, leading to an increase in internal tensile or compressive stress, forming a stress concentration area, and making the coating more susceptible to microcracks or peeling. Therefore, a large cathode surface charge accumulation reference value indicates that the internal stress of the metal layer is increasing, and measures such as current distribution optimization or cathode surface potential balance control need to be taken to reduce the impact of uneven deposition. On the contrary, a small cathode surface charge accumulation reference value indicates that the cathode surface charge distribution is more uniform, the metal ion deposition rate is stable, the lattice structure of the coating is more regular, the internal stress is smaller, and the deposition quality is good.

[0026] The long residence time of bubbles on the coating surface can indicate an increase in internal stress in the metal layer, mainly due to its influence on the uniformity of metal deposition, crystal growth pattern, and stress distribution. During electrodeposition, the uniformity of the coating is crucial to the magnitude of internal stress, and the residence of bubbles can form a local shielding effect on the cathode surface, hindering the normal migration and deposition of metal ions, reducing the deposition rate in that area, and thus leading to uneven thickness of the deposited layer. When uneven thickness occurs during coating growth, the degree of metal shrinkage in different areas during deposition will differ, easily forming local residual stress concentration. In addition, bubble retention can also cause micropores or defects, reducing the local structural strength of the coating and increasing the risk of crack initiation. Under high current density or pulsed current mode, uneven distribution of bubbles can further interfere with crystal orientation, forming coarse grains or non-uniform grain arrangement, reducing the mechanical stability of the metal layer, and leading to uneven accumulation of tensile or compressive stress. With the continuous accumulation of stress, these defect areas can become the starting point for crack propagation, ultimately leading to failure of the entire coating structure. Therefore, long bubble residence time not only affects the deposition quality, but also serves as an important signal indicating an increase in internal stress in the metal layer.

[0027] The specific steps for analyzing the formation and detachment of bubbles under the detection window to generate a bubble retention reference value are as follows: First, high-resolution optical information is continuously collected on the cathode surface, and the two-dimensional projection contours of all bubbles are identified through image processing techniques, and then the radius information of each bubble is obtained. The bubble size distribution function is defined to describe the distribution of different bubble radii on the cathode surface, and the formula is as follows: , where is the bubble size distribution function, describing the distribution of different size bubbles on the cathode surface of the electrolytic cell, is a continuous variable of bubble radius, representing different sizes of bubbles, is the total number of bubbles, is the actual radius of the th bubble collected, is the local distribution width parameter around the th bubble radius, used to reflect the dispersion degree of bubble size distribution, is the adjustment factor, used to control the influence intensity of bubble size on the overall distribution, is the scale factor for adjusting bubble size normalization, used to normalize the influence of different bubble sizes to prevent excessive or insufficient weight deviation in the calculation process, is a very small positive number (such as ), to avoid division by zero in the formula; The above steps are to accurately depict the overall characteristics of the bubble size distribution in the electrodeposition process, and provide a basis for subsequent analysis of the bubble retention state.

[0028] By analyzing the distribution of bubble positions on the electrode surface, a heterogeneity index reflecting the degree of spatial distribution unevenness of the bubbles is generated. The spatial heterogeneity index generation formula is as follows: , wherein, is the spatial heterogeneity index, is the local density value of the first bubble on the cathode surface, which is represented by the number of bubbles per unit area around the bubble, is the local density value of the first bubble on the cathode surface, is the Euclidean distance between the first bubble and the second bubble, representing the relative distance of the bubbles on the cathode surface, is the spatial distance adjustment factor, which avoids zero abnormality and controls the influence intensity of spatial distance, is the local distribution adjustment factor, which is used to adjust the weight of index calculation and affect the calculation method of density distribution, is the bubble density nonlinear adjustment factor, which is used to control the nonlinear influence of local bubble density in calculation; is the bubble density nonlinear adjustment factor, which is used to control the nonlinear influence of local bubble density in calculation; The above steps are to accurately depict the spatial distribution unevenness of the bubbles on the cathode surface. The greater the spatial heterogeneity, the more concentrated the bubble distribution in the local area, which may cause uneven deposition of the plating layer and local stress concentration. The above steps are to accurately depict the spatial distribution unevenness of the bubbles on the cathode surface. The greater the spatial heterogeneity, the more concentrated the bubble distribution in the local area, which may cause uneven deposition of the plating layer and local stress concentration. The above steps are to accurately depict the spatial distribution unevenness of the bubbles on the cathode surface. The greater the spatial heterogeneity, the more concentrated the bubble distribution in the local area, which may cause uneven deposition of the plating layer and local stress concentration.

[0029] The above steps are to accurately depict the spatial distribution unevenness of the bubbles on the cathode surface. The greater the spatial heterogeneity, the more concentrated the bubble distribution in the local area, which may cause uneven deposition of the plating layer and local stress concentration. The above steps are to accurately depict the spatial distribution unevenness of the bubbles on the cathode surface. The greater the spatial heterogeneity, the more concentrated the bubble distribution in the local area, which may cause uneven deposition of the plating layer and local stress concentration. , wherein, is the bubble residual reference value, is the correlation factor of bubble size and spatial heterogeneity distribution, which is an adjustment coefficient used to control the influence degree of bubble size and spatial heterogeneity index on , is the maximum value of bubble radius, is the minimum value of bubble radius, is the adjustment parameter for optimizing the sensitivity of integration, is the super-curvature adjustment factor, is the super-sine function term.

[0030] ​​​The step serves to highly couple the bubble size distribution and spatial distribution characteristics, accurately quantify the influence degree of bubble retention on the coating stress through the nonlinear integral transformation and the type function relationship. The greater, the higher the bubble retention degree, the more likely the uneven growth of the coating and the abnormal grain structure, and then the formation of obvious internal stress accumulation; on the contrary, The smaller, the more stable the deposition environment and the normal stress state.

[0031] The greater the bubble retention reference value generated after analyzing the formation and detachment of bubbles under the detection window, the higher the internal stress of the metal layer, and vice versa. The fundamental reason for this phenomenon is that the formation, retention and detachment of bubbles directly affect the deposition uniformity and microstructure of the coating. In the process of electrodeposition, the uniform deposition of metal ions on the cathode surface is an important condition to ensure the quality of the coating, but when the bubble retention time is too long, it may lead to insufficient local metal deposition, resulting in a coating with uneven thickness in that area, and then causing stress concentration. In addition, the long-term retention of bubbles will interfere with the normal growth of metal grains, causing grain boundary dislocation or abnormal crystal orientation, making the structure of the coating uneven, and then inducing abnormal accumulation of residual tensile stress or compressive stress. Under high current density or pulse deposition conditions, the influence of bubbles is more significant, the longer the retention time, the more likely to cause instability in the deposition process, and even form micropore defects, aggravate stress concentration and reduce the mechanical strength of the coating.

[0032] The analyzed characteristic variables are input into the pre-trained deep learning model for forward reasoning to evaluate the internal stress state of the praseodymium-neodymium layer in real time online; The analyzed cathode surface charge accumulation reference value and bubble retention reference value are input into the pre-trained deep learning model to generate a stress change risk coefficient for forward reasoning, and the internal stress state of the praseodymium-neodymium layer is evaluated in real time online through the stress change risk coefficient.

[0033] A pre-trained deep learning model refers to a model that has been trained, optimized, and achieved stable generalization capability using a large amount of historical data before actual application, and can directly perform fast prediction and reasoning in actual production environment. In the process of praseodymium-neodymium electrodeposition, the deep learning model needs to identify the complex nonlinear relationship between the electrolytic parameters and the internal stress of the coating, so before being put into real-time application, it must be fully trained using a large amount of experimental data and historical production data to make it have accurate stress change prediction capability. During the training process, first of all, high-quality electrodeposition data of the coating is needed, including current density, electrode potential, cathode surface charge accumulation reference value, bubble residue reference value and other key parameters, and combined with the coating stress test data to establish a standardized data set. Then, a suitable deep learning architecture is selected, such as a convolutional neural network (CNN) to process electric field distribution image data, or a long short-term memory network (LSTM) to process time series data, to capture the trend of coating stress change over time. Through supervised learning, the model is trained using a data set with real stress labels, so that it learns the stress evolution law under different electrodeposition conditions. During the training process, the model is optimized using a loss function (such as mean square error MSE or cross-entropy loss) to gradually reduce the prediction error, and a gradient descent optimization algorithm (such as Adam, SGD) is used to adjust the model parameters to improve the generalization capability. In addition, cross-validation and model hyperparameter optimization are also needed to ensure that the model can not only accurately predict the stress change under known conditions, but also maintain good prediction ability in unknown conditions.

[0034] After being fully trained and optimized, the deep learning model can be used as a core component for real-time online reasoning and deployed in the praseodymium-neodymium electrodeposition control system to achieve intelligent evaluation of stress change. In actual application, the cathode surface charge accumulation reference value and bubble residue reference value parameters are input into the trained model in real time for forward reasoning to generate a stress change risk factor (StressRiskFactor, SRF), which reflects the dynamic change trend of the internal stress of the coating under the current electrodeposition process. Since the deep learning model has been trained with a large amount of data and can identify complex nonlinear relationships between multi-dimensional parameters, its calculation results can be more accurate and predictive than traditional mathematical models. When the stress change risk factor exceeds the set threshold, the system can automatically trigger intelligent adjustment strategies, such as dynamically extending the intermittent shutdown time, reducing the pulse peak current, optimizing the electric field distribution, etc., to prevent stress accumulation from causing cracks or coating peeling. In addition, the model can also be continuously optimized through online learning, i.e. collecting new production data during long-term operation and regularly updating model parameters to adapt to different production conditions and environmental changes, thereby achieving more efficient, stable and intelligent control of the praseodymium-neodymium electrodeposition process.

[0035] The deep learning model is not limited to the cathode surface charge accumulation reference value and bubble residual reference value comprehensive analysis to generate stress change risk coefficient The deep learning model can be used to implement the technical scheme of the present application, and the present application provides a specific implementation method. stress change risk coefficient The generation formula is as follows: , wherein and are the preset proportion coefficients of the cathode surface charge accumulation reference value and the bubble residual reference value , and and are both greater than 0.

[0036] The preset proportion coefficients and are weight factors used to weigh the relative influence degree of the cathode surface charge accumulation reference value and the bubble residual reference value on the stress change risk coefficient . In the formula, and determine the contribution proportion of and in the final calculated result. Its specific role is to adjust and optimize the influence of the two key parameters in stress evaluation, so that the SERF index can accurately reflect the risk of actual coating stress change.

[0037] Under different process conditions and application scenarios, and may have different influences on stress change. For example: Under high current density conditions, cathode surface charge accumulation may be the main influencing factor, and and may be larger to highlight the contribution of to stress change. In low flowability or high bubble generation rate electrolyte, bubble residual may be the dominant factor, and may be larger to strengthen the influence of in the calculation of . In addition, the square root normalization processing of the preset proportion coefficient is to ensure that is still comparable under different weight configurations, and to avoid the dominance of a single parameter in the calculation result. Usually, and obtained from experimental data or machine learning optimization, and its optimal value is adjusted through historical data training and testing to make The evaluated stress change trend is more in line with the physical law in the actual coating growth process.

[0038] From the stress change risk coefficient, the greater the cathode surface charge accumulation reference value generated after analyzing the local charge distribution existing on the cathode surface under the detection window, the greater the bubble residue reference value generated after analyzing the formation and detachment of bubbles under the detection window, the greater the stress change risk coefficient generated by the real-time online evaluation of the internal stress state of the praseodymium-neodymium layer through the pre-trained deep learning model, indicating that the electric field distribution is abnormal and the bubble retention is serious during electrodeposition, thereby causing the internal stress of the praseodymium-neodymium layer to increase, causing the stress change risk coefficient to rise. On the contrary, it indicates that the electric field distribution is uniform and the bubbles are effectively detached, and the coating structure is stable, and the internal stress is maintained within the normal range, and the stress change risk coefficient is relatively small. Therefore, the greater the stress change risk coefficient, the higher the internal stress of the metal layer, and there is a higher risk of cracking or peeling, and the smaller the stress change risk coefficient, indicating that the internal stress of the praseodymium-neodymium layer is in a stable state, and the coating quality is good.

[0039] The stress change risk coefficient generated by the real-time online evaluation of the internal stress state of the praseodymium-neodymium layer through the pre-trained deep learning model is compared and analyzed with the pre-set stress change risk coefficient reference threshold value, and the internal stress state of the praseodymium-neodymium layer is divided, and the division steps are as follows: If the stress change risk coefficient is greater than the pre-set stress change risk coefficient reference threshold value, the internal stress of the praseodymium-neodymium layer is divided into praseodymium-neodymium layer internal stress increase; if the stress change risk coefficient is less than or equal to the pre-set stress change risk coefficient reference threshold value, the internal stress of the praseodymium-neodymium layer is divided into praseodymium-neodymium layer internal stress normal and stable.

[0040] When the praseodymium-neodymium is identified to have internal stress increase during electrodeposition, the intermittent closing time is first dynamically extended to allow the re-adjustment of the atomic structure of the coating and the relaxation and release of the lattice stress, thereby reducing the accumulation degree of the local stress in the coating, and the pulse peak current is also reduced to reduce the number of praseodymium-neodymium atoms deposited on the surface of the coating in a single deposition period; During the electrodeposition of praseodymium-neodymium, the key control strategy adopted by the intelligent system when it identifies that the internal stress of the coating has increased is to dynamically extend the intermittent closing time and reduce the pulse peak current, and the core function of this strategy is to effectively relieve the internal residual stress, optimize the microstructure of the coating, and improve the uniformity and stability of the deposited layer.

[0041] Firstly, the role of dynamically extending the intermittent closing time is to provide sufficient time for the atomic structure of the coating to rearrange and relax. In the normal electrodeposition process, metal atoms will quickly deposit on the cathode surface with the application of current. When the current density is too high and the deposition rate is too fast, the residual stress inside the coating will continue to accumulate, which may exist in the form of tensile or compressive stress, causing lattice distortion and even micro-cracks. If no control measures are taken, the accumulated stress over a long period of time may cause the coating to become brittle, reduce adhesion, and even cause peeling or crack propagation during subsequent processing or service. By extending the intermittent closing time, the electric field on the electrode surface is weakened during the power-off stage, and the deposited praseodymium-neodymium atoms can rearrange and structure relax through thermal vibration, reducing lattice defects and improving the internal stability of the material. This adjustment process can release the residual stress of the coating, reduce the possibility of crack initiation, and thus enhance the overall performance of the deposited layer.

[0042] Secondly, the role of linkage reduction of pulse peak current is to control the deposition amount in a single deposition cycle, preventing stress concentration problems caused by rapid deposition. Under high pulse current conditions, a large number of praseodymium-neodymium atoms will be deposited on the cathode surface in a short period of time, which may cause grain structure disorder and thus generate high residual stress inside the coating. By reducing the pulse peak current, the deposition rate of metal atoms will slow down accordingly, making the crystal growth of the coating more uniform and dense, which helps to reduce stress accumulation caused by lattice dislocation. In addition, lower pulse current can also effectively reduce the amount of hydrogen gas evolution, avoiding damage to the coating structure caused by gas bubble retention, and improving the deposition quality.

[0043] Through the combined effect of these two measures, the accumulation of local stress in the coating can be significantly reduced, making the entire coating structure more uniform and stable, not only improving the mechanical properties, but also enhancing the corrosion resistance and wear resistance, ultimately ensuring the long-term reliability and durability of the praseodymium-neodymium coating in practical applications.

[0044] When it is identified that the internal stress of praseodymium-neodymium increases during electrodeposition, the specific steps of dynamically extending the intermittent closing time and linkage reducing the pulse peak current are as follows: When it is identified that the internal stress of praseodymium-neodymium increases during electrodeposition, the intermittent closing time is dynamically extended, and the extension amplitude needs to be based on the current actual stress risk level (i.e. the deviation between the reference threshold value), the calculation expression is as follows: , in the formula, is the dynamically adjusted intermittent closing time, is the basic intermittent closing time, i.e. the intermittent closing time setting value under stable process conditions, is the stress relaxation increment coefficient, controlling the time adjustment amplitude, is the stress change risk coefficient, is the stress change risk coefficient reference threshold, is the non-linear index, controlling the influence of the deviation degree on the time adjustment, making the response more sensitive, is a very small positive number, used to prevent the denominator from being zero in the calculation, is the exponential decay term, making when the threshold is greater than the threshold, the time adjustment tends to be gentle, avoiding excessive adjustment, is the exponential suppression factor, used to control the gradual convergence of the time extension, making the adjustment stable, is the natural base number; Through the above steps, the intermittent closing time can be intelligently and adaptively dynamically extended after the stress risk coefficient is greater than the safety threshold, effectively reducing the local residual stress of the deposited praseodymium and neodymium layer, avoiding further intensification of lattice distortion, preventing the generation and expansion of internal micro-cracks, providing sufficient adjustment time for atomic-level rearrangement and lattice structure stress release of the plated layer structure, and ensuring the stability and density of the plated layer.

[0045] At the same time, combined with the stress change risk coefficient the real-time variation degree, the pulse peak current in a single deposition cycle is reduced in linkage to reduce the number of praseodymium and neodymium atoms deposited on the plated layer surface in each cycle. The specific pulse current adjustment amount calculation expression is as follows: , wherein is the dynamically adjusted pulse peak current, is the standard pulse peak current set as the basis, is the current control sensitivity coefficient, controlling the adjustment amplitude, is the logarithmic term in the function, making the adjustment amplitude small when slightly exceeding the standard, and rapidly increasing when the exceeding is serious, is the function term, is the steepness factor, used to adjust slowly when the stress change risk coefficient is below the threshold, and to adjust rapidly when the stress change risk coefficient exceeds the threshold by a large margin, is the exponential suppression factor, used to balance the current adjustment amplitude to avoid excessive reduction of the current affecting the deposition efficiency, is the exponential balance factor, used to adjust the stability of the current reduction.

[0046] The above steps can intelligently reduce the current pulse peak value, slow down the deposition rate of praseodymium-neodymium atoms, effectively control the growth rate of the plating layer structure and thickness during the deposition process, reduce the problems of lattice distortion, grain orientation disorder and internal stress accumulation caused by local rapid deposition, and ensure that the deposited praseodymium-neodymium layer has fine, uniform and dense grain structure, thereby effectively improving the mechanical properties and durability of the plating layer and avoiding the risk of brittle fracture of the plating layer.

[0047] The present application effectively solves the quality problems of internal stress accumulation, crack propagation and plating layer peeling caused by the traditional fixed preset intermittent closing time under high current density process conditions. Through a series of steps such as real-time data acquisition, intelligent data processing, feature extraction, deep learning evaluation and dynamic control of pulse current, the present application realizes intelligent and precise control of the electrodeposition process. When the internal stress of the plating layer is abnormally high, the system can adaptively prolong the intermittent closing time, so that the plating layer structure is fully adjusted to prevent excessive stress accumulation, and at the same time, the pulse peak current is reduced to optimize the deposition rate and improve the uniformity and density of the plating layer. Compared with the traditional process, the present application not only significantly improves the mechanical strength and corrosion resistance of the plating layer, but also prolongs the service life of the equipment, reduces the electrode consumption, reduces the production energy consumption, and further improves the stability and consistency of the plating layer quality.

[0048] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the most recent real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0049] The above only describes certain exemplary embodiments of the present application by way of illustration, and it is self-evident that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above figures and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present application.

[0050] It should be noted that in this document, relationship terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, method, article or equipment including the element.

[0051] It should be understood that the magnitude of the sequence number of each process described above does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0052] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0053] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0054] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment of the present application according to actual needs.

[0055] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0056] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art cannot easily think of changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0057] The above only describes some exemplary embodiments of the present application by way of illustration, and it is self-evident that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above figures and descriptions are illustrative in nature and should not be understood as limiting the scope of the claims of the present application.

Claims

1. A method for the electrolytic reduction of rare earth metals based on high frequency switching power supplies, characterized in that, Comprising the following steps: First, in the process of praseodymium-neodymium electrodeposition, the advanced sensors and detection equipment are used to accurately obtain the parameter data in the process of praseodymium-neodymium electrodeposition in real time, providing accurate, comprehensive and real-time data support for subsequent intelligent analysis; The real-time collected parameter data is centrally integrated to establish a data set, and the data set is preprocessed; Based on the preprocessed high-quality data, the feature engineering method is used to extract the feature variables reflecting the internal stress state and change trend of the coating, and the extracted feature variables are analyzed under the detection window to quantify the change trend and severity of the internal stress of the coating in the current praseodymium-neodymium deposition process; The analyzed feature variables are input into the pre-trained deep learning model for forward reasoning to evaluate the internal stress state of the praseodymium-neodymium layer in real time. When the internal stress of praseodymium-neodymium is identified to increase in the electrodeposition process, first, the intermittent closing time is dynamically extended to allow the re-adjustment of the atomic structure and the relaxation of the lattice stress of the coating to reduce the accumulation of local stress in the coating, and the pulse peak current is also reduced to reduce the number of praseodymium-neodymium atoms deposited on the surface of the coating in a single deposition period.

2. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 1 wherein, In the process of praseodymium-neodymium electrodeposition, the parameter data in the deposition process is accurately obtained as follows: First, multiple types of sensors are arranged inside and outside the electrolytic cell, including but not limited to current sensors, voltage sensors, temperature sensors, pH sensors, ion concentration sensors, and laser interference sensors to comprehensively cover key process parameters; Second, use a high-frequency data acquisition system to synchronously collect data from all sensors to ensure the time consistency of key parameters, and convert analog signals to high-precision digital signals through an analog-to-digital converter to ensure the accuracy and stability of the data; Then, use wireless communication technology to transmit the collected data in real time to a central data server to ensure smooth and efficient data transmission; Finally, the stored data is used for subsequent intelligent analysis and deep learning model calculation to provide accurate, comprehensive and real-time data support for dynamically optimizing the praseodymium-neodymium electrodeposition process, ensuring that the entire system can operate efficiently and stably under different process conditions.

3. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 1 wherein, Based on the preprocessed high-quality data, the feature engineering method is used to extract feature variables reflecting the internal stress state and change trend of the coating, and the extracted feature variables include the local charge distribution on the cathode surface and the formation and detachment of bubbles. The local charge distribution on the cathode surface and the formation and detachment of bubbles are analyzed under the detection window to generate a cathode surface charge accumulation reference value and a bubble residue reference value, respectively, to quantify the change trend and severity of the internal stress of the coating in the current praseodymium-neodymium deposition process.

4. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 3 wherein, The specific steps for analyzing the local charge distribution on the cathode surface under the detection window to generate the cathode surface charge accumulation reference value are as follows: Firstly, under the detection window, the local charge density distribution of different spatial positions on the cathode surface is obtained by high-precision current density sensor and electric field measurement device. The charge distribution on the cathode surface is not uniform, but is affected by electric field gradient, ion diffusion in solution, local polarization and micro-geometric structure of electrode surface. Therefore, the local charge gradient change is defined to measure the spatial variation degree of charge distribution, and the calculation expression is as follows: wherein is the local charge density of the cathode surface, representing the local charge density of the cathode surface at the coordinate , and are the second order variation of the cathode surface charge density in the and directions, i.e. the local curvature of the charge density distribution, is the local potential of the cathode surface, representing the local potential of the cathode surface at the coordinate , and represent the gradient variation of the cathode surface potential along the and directions, is a weight factor, is the local charge gradient variation. In obtaining the local charge gradient variation After that, the local charge accumulation potential function is introduced to measure the overall charge non-uniformity of the cathode surface, and finally the cathode surface charge accumulation reference value is defined, and the calculation expression is as follows: wherein is a cathode surface charge accumulation reference value, is a normalization coefficient, is a spatial range of analysis of the cathode surface, is a regulation factor, is a potential decay factor, is a natural base, is a potential decay coefficient, is a microelement area.

5. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 3 wherein, The formation and detachment of bubbles are analyzed under the detection window to generate the specific steps of bubble residual reference value as follows: Firstly, the high-resolution optical information of the cathode surface is continuously collected, and the two-dimensional projection contour of all bubbles is identified through image processing technology, and then the radius information of each bubble is obtained. The bubble size distribution function is defined to describe the distribution state of different bubble radii on the cathode surface, and the formula is as follows: , where is the bubble size distribution function, is a continuous variable of bubble radius, is the total number of bubbles, It is the first The actual radius of the bubble, It is The local distribution width parameter around the bubble radius, is the regulating factor, is the scale factor that normalizes the bubble size, is a very small positive number; By analyzing the distribution of bubble position on the electrode surface, the heterogeneity index reflecting the uneven degree of spatial distribution of bubbles is generated, and the spatial heterogeneity index generation formula is as follows: , where is the spatial heterogeneity index, It is The local density of bubbles on the cathode surface is It is The local density of bubbles on the cathode surface is It is bubbles and The Euclidean distance between bubbles, is the spatial distance adjustment factor, is the local distribution adjustment factor, is the nonlinear adjustment factor of bubble density; Integrated bubble size distribution function With spatial heterogeneity index , construct bubble residual reference value to comprehensively characterize the degree of influence of bubbles on the metal deposition process, and the formula is as follows: wherein is a bubble residual reference value, is a correlation factor of bubble size and spatial heterogeneity distribution, is a maximum value of bubble radius, is a minimum value of bubble radius, is a tuning parameter, is a hyper-curvature tuning factor, is a hyper-curve sinusoidal function term.

6. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 3 wherein, The analyzed cathode surface charge accumulation reference value and bubble residual reference value are input into the pre-trained deep learning model to generate stress change risk coefficient, and the forward inference is carried out. The stress change risk coefficient is used to evaluate the internal stress state of praseodymium-neodymium layer in real time and online.

7. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 6 wherein, The stress change risk coefficient generated by the pre-trained deep learning model in real time and online evaluation of the internal stress state of praseodymium-neodymium layer is compared and analyzed with the pre-set stress change risk coefficient reference threshold value, and the internal stress state of praseodymium-neodymium layer is divided, and the division steps are as follows: If the stress change risk coefficient is greater than the pre-set stress change risk coefficient reference threshold value, the internal stress of praseodymium-neodymium layer is divided into praseodymium-neodymium layer internal stress rise; If the stress change risk coefficient is less than or equal to the pre-set stress change risk coefficient reference threshold value, the internal stress of praseodymium-neodymium layer is divided into praseodymium-neodymium layer internal stress normal and stable.

8. The high frequency switching power supply based process for electrolytic reduction of rare earth metals as claimed in claim 7 wherein, When the internal stress of praseodymium-neodymium is identified to rise in the electrodeposition process, the intermittent closing time is first dynamically extended, and the specific steps of the pulse peak current are as follows: When the internal stress of praseodymium-neodymium is identified to rise in the electrodeposition process, the intermittent closing time is dynamically extended, and the extension range needs to be based on the current actual stress risk degree, and the calculation expression is as follows: , wherein is the dynamically adjusted intermittent shutdown time, is the base intermittent shutdown time, is the stress relaxation increment coefficient, is the stress change risk coefficient, is the stress change risk coefficient reference threshold, is the non-linear exponent, is a very small positive number, is the exponential decay term, is the exponential dampening factor, is the natural base number; At the same time, combined with stress change risk coefficient Real-time change degree, linkage reduces the pulse peak current in a single deposition cycle, to reduce the number of praseodymium and neodymium atoms deposited on the surface of the plating layer in each cycle, the specific pulse current adjustment amount calculation expression is as follows: , wherein, is the dynamically adjusted pulse peak current, is the base set standard pulse peak current, is the current regulation sensitivity factor, is the logarithmic term in the function, is the function term, is the steepness factor, is the exponential dampening factor to balance the current adjustment amplitude and avoid excessive current reduction affecting the deposition efficiency, is the exponential balancing factor.

Images