Prediction-based solid electrolyte electrolysis efficiency cooperative control method and system

Abstract

The invention discloses a prediction-based solid electrolyte electrolysis efficiency cooperative control method and system, and relates to the technical field of battery control, and the method comprises the steps: responding to the selection information of a control end on a multi-source sensing unit in a solid electrolyte electrolysis process, and taking multi-dimensional state sensing data output by the selected sensing unit as an input data source of a prediction algorithm; accessing a prediction algorithm to the input data source, and generating an electrolytic efficiency disturbance prediction component based on the multi-dimensional state sensing data; matching the target control strategy from a preset cooperative control strategy library, and calculating to generate a multi-dimensional cooperative control instruction; and according to the multi-dimensional cooperative control instruction, a multi-physical field execution unit is controlled to execute cooperative control operation, an electrolytic efficiency cooperative control result is obtained through calculation based on state data collected after the cooperative control operation is executed, and the electrolytic efficiency cooperative control result is fed back to the control end. According to the invention, high-efficiency operation of the solid electrolyte electrolysis process is effectively maintained.

Description

Technical Field

[0001] This application relates to the field of battery control technology, and in particular to a method and system for predictive-based synergistic control of solid electrolyte electrolysis efficiency. Background Technology

[0002] In the process of hydrogen production or metal extraction by solid electrolyte electrolysis, constant current or constant voltage methods are usually used for electrolysis. Once the power supply output parameters are set, they remain unchanged, and thermal management often employs fixed heating or cooling strategies. The interface state is only detected through manual inspection or after an efficiency drop. This control method can maintain basic operation in a stable laboratory environment, but in actual industrial scenarios, electrolysis efficiency often exhibits inexplicable fluctuations or even sudden declines.

[0003] Conventional electrolysis control methods only collect macroscopic parameters such as cell voltage, total current, and temperature. Based on these parameters, the current efficiency is calculated and used as a feedback signal to fine-tune the power supply output. Because there is no direct sensing of the interfacial contact state between the solid electrolyte and the electrodes, changes in the lattice structure within the electrolyte material, or the bubble coverage at the reaction interface, the control system cannot identify the physical causes of efficiency degradation in advance. When the interface experiences micro-motion due to vibration, the material undergoes valence state migration due to electrochemical polarization, or bubbles firmly adhere under liquid-free conditions, these disturbances have already substantially affected the electrolysis process, yet the control unit is still compensating with lagging efficiency data. Summary of the Invention

[0004] In view of the aforementioned problems, this application is hereby filed.

[0005] Therefore, this application provides a predictive method and system for synergistic control of solid electrolyte electrolysis efficiency, which can solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, this application provides the following technical solution: In a first aspect, this application provides a method for synergistic control of solid electrolyte electrolysis efficiency based on prediction, including: Preferably, the response control terminal uses the selection information of the multi-source sensing units during the solid electrolyte electrolysis process, and uses the multi-dimensional state sensing data output by the selected sensing units as the input data source for the prediction algorithm, including: The receiving control terminal sends a sensing unit selection instruction, which specifies the multi-source sensing units participating in this electrolysis efficiency collaborative control. According to the instruction selected by the sensing unit, the corresponding mechanical vibration sensor, rare earth valence state monitoring sensor and interface bubble imaging sensor are activated. Mechanical vibration spectrum data is acquired from the activated mechanical vibration sensor, valence characteristic data is acquired from the rare earth valence state monitoring sensor, and bubble coverage state data is acquired from the interface bubble imaging sensor. The mechanical vibration spectrum data, the valence state characteristic data, and the bubble coverage state data are integrated into multidimensional state perception data, which is then used as the input data source for the prediction algorithm.

[0007] Preferably, the step of connecting the prediction algorithm to the input data source and generating an electrolysis efficiency perturbation prediction component based on the multidimensional state-aware data includes: The multidimensional state-aware data is preprocessed, including time alignment, noise filtering, and feature normalization, to obtain a standardized input sequence; The standardized input sequence is input into the prediction algorithm, which is a multi-task temporal prediction network that includes a shared bottom-level feature extraction layer and three independent output branches. The coupling features in the multidimensional state-aware data are extracted through the shared bottom-level feature extraction layer, and the interface contact stability change trend, lattice channel reconstruction trend and three-phase interface effective reaction area decay trend are generated through the first output branch, the second output branch and the third output branch, respectively. The trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of effective reaction area of ​​the three-phase interface are combined into the electrolysis efficiency disturbance prediction component and output to the collaborative control strategy library.

[0008] Preferably, the step of retrieving the collaborative control strategy library and generating pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions, and interface stripping trigger instructions based on the electrolysis efficiency disturbance prediction components includes: The pulse electrolysis control rules corresponding to the interface contact stability change trend are retrieved from the collaborative control strategy library, and pulse electrolysis parameter adjustment instructions are generated according to the amplitude and rate of change of the interface contact stability change trend. The electrochemical potential control rules corresponding to the lattice channel reconstruction trend are retrieved from the collaborative control strategy library, and an electrochemical potential window control instruction is generated based on the openness index of the lattice channel reconstruction trend and its evolution direction. The interface stripping trigger rule corresponding to the decay trend of the effective reaction area of ​​the three-phase interface is retrieved from the collaborative control strategy library, and an interface stripping trigger command is generated based on the decay rate of the effective reaction area decay trend and the current coverage status.

[0009] Preferably, the step of synchronously coordinating power output operation, thermoelectric coupling operation, and ultrasonic stripping operation according to the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command to obtain a coordinated control result for electrolysis efficiency, and feeding the coordinated control result back to the control terminal for updating the prediction algorithm parameters, includes: According to the pulse electrolysis parameter adjustment command, configure the current waveform parameters of the power output unit and execute the pulse electrolysis operation; According to the electrochemical potential window control command, the heating or cooling power of the thermoelectric coupling unit is adjusted so that the solid electrolyte is maintained within the temperature range corresponding to the electrochemical potential window. According to the interface peeling trigger command, the ultrasonic peeling unit is started and the interface air film removal operation is performed according to the specified power and duration. The actual execution parameters and corresponding electrolysis efficiency feedback data of pulse electrolysis operation, thermoelectric coupling operation and ultrasonic stripping operation are collected to form a collaborative control result, which is then sent to the control terminal to update the prediction algorithm parameters.

[0010] Preferably, the step of extracting coupling features from the multidimensional state-aware data through the shared bottom-level feature extraction layer, and generating the interface contact stability change trend, lattice channel reconstruction trend, and three-phase interface effective reaction area decay trend via the first output branch, the second output branch, and the third output branch, respectively, includes: Based on the vibration feature sub-vectors in the standardized input sequence, the dominant frequency band of mechanical disturbance is identified and mapped to the interface contact impedance response control to generate the interface contact stability change trend. Based on the valence state feature sub-vectors in the standardized input sequence, rare earth element redox activity indicators are extracted and mapped to the lattice channel conduction capability evolution model to generate lattice channel reconstruction trends. Based on the interface feature sub-vectors in the standardized input sequence, the spatiotemporal distribution entropy of bubble coverage is quantified and mapped to the effective reaction area decay kinetic model to generate the three-phase interface effective reaction area decay trend.

[0011] Preferably, the step of identifying the dominant frequency band of mechanical disturbance based on the vibration feature sub-vectors in the standardized input sequence and mapping it to the interface contact impedance response control to generate the interface contact stability change trend includes: Spectral energy density analysis is performed on the vibration characteristic sub-vectors to calculate the cumulative energy percentage in each frequency range; The lowest frequency range where the cumulative energy percentage exceeds a preset threshold is selected as the dominant frequency band for mechanical disturbance; Substituting the center frequency and amplitude of the dominant frequency band of the mechanical disturbance into the interface contact impedance response control, the output of the rate of change sequence of interface contact impedance within the future time window is used as the trend of interface contact stability change.

[0012] Preferably, the step of extracting rare earth element redox activity indices based on the valence state feature sub-vectors in the standardized input sequence and mapping them to a lattice channel conductivity evolution model to generate a lattice channel reconstruction trend includes: Extract the L3 side absorption peak shift and full width at half maximum (FWHM) of a specific rare earth element from the valence state eigenvectors; The redox activity index is calculated based on the absorption peak shift and the full width at half maximum (FWHM), and the redox activity index characterizes the potential for changes in the concentration of mobile vacancies in the crystal lattice. The redox activity index is input into the lattice channel conductivity evolution model, and the evolution curve of the lattice channel openness index within the future time window is output as the lattice channel reconstruction trend.

[0013] Preferably, the step of quantifying the spatiotemporal distribution entropy of bubble coverage based on the interface feature sub-vectors in the standardized input sequence and mapping it to the effective reaction area decay kinetic model to generate the three-phase interface effective reaction area decay trend includes: Connected component labeling is performed on the bubble mask image in the interface feature sub-vector to obtain the set of pixel coordinates for each bubble region; Based on the spatial location and area of ​​each bubble region, the spatial entropy and temporal rate of change of bubble coverage distribution are calculated to form the spatiotemporal distribution entropy of bubble coverage. Substituting the spatiotemporal distribution entropy of the bubble coverage into the effective reaction area decay kinetic model, the decay rate of the effective reaction area ratio within the future time window is output as the decay trend of the effective reaction area of ​​the three-phase interface.

[0014] Secondly, this application also provides a prediction-based solid electrolyte electrolysis efficiency collaborative control system, including: a sensing data scheduling module, which responds to the selection information of multi-source sensing units in the solid electrolyte electrolysis process by the control end, and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source of the prediction algorithm; The efficiency disturbance prediction module connects to the input data source and uses a prediction algorithm to generate electrolysis efficiency disturbance prediction components based on the multidimensional state perception data. The electrolysis efficiency disturbance prediction components include the trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of effective reaction area at the three-phase interface. The collaborative control instruction module retrieves the collaborative control strategy library and generates pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions, and interface stripping trigger instructions based on the electrolysis efficiency disturbance prediction components. The execution coordination feedback module synchronously coordinates the power output operation, thermoelectric coupling operation, and ultrasonic stripping operation according to the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command to obtain the electrolysis efficiency coordinated control result, and feeds the coordinated control result back to the control terminal to update the prediction algorithm parameters.

[0015] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps: The response control terminal selects the multi-source sensing unit during the solid electrolyte electrolysis process and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source for the prediction algorithm. The prediction algorithm is connected to the input data source to generate an electrolysis efficiency perturbation prediction component based on the multidimensional state perception data. The electrolysis efficiency perturbation prediction component includes the trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of the effective reaction area of ​​the three-phase interface. The collaborative control strategy library is retrieved, and pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions, and interface stripping trigger instructions are generated according to the electrolysis efficiency disturbance prediction components. Based on the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command, the power output operation, thermoelectric coupling operation, and ultrasonic stripping operation are synchronously coordinated to obtain the electrolysis efficiency collaborative control result, and the collaborative control result is fed back to the control terminal to update the prediction algorithm parameters.

[0016] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps: The response control terminal selects the multi-source sensing unit during the solid electrolyte electrolysis process and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source for the prediction algorithm. The prediction algorithm is connected to the input data source to generate an electrolysis efficiency perturbation prediction component based on the multidimensional state perception data. The electrolysis efficiency perturbation prediction component includes the trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of the effective reaction area of ​​the three-phase interface. The collaborative control strategy library is retrieved, and pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions, and interface stripping trigger instructions are generated according to the electrolysis efficiency disturbance prediction components. Based on the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command, the power output operation, thermoelectric coupling operation, and ultrasonic stripping operation are synchronously coordinated to obtain the electrolysis efficiency collaborative control result, and the collaborative control result is fed back to the control terminal to update the prediction algorithm parameters.

[0017] Implementing this application will have the following beneficial effects: This application provides a method and system for synergistic control of solid electrolyte electrolysis efficiency based on prediction. 1. This application generates perturbation prediction components corresponding to interface contact stability, lattice channel reconstruction, and effective reaction area by sensing three types of multidimensional state data: mechanical vibration, material valence state, and interface bubbles. Based on these components, it synchronously coordinates three operations: pulse electrolysis, thermoelectric coupling, and ultrasonic ablation. In contrast, existing technologies only perform single feedback adjustment of the total current or voltage after efficiency declines. This multi-source sensing-multi-dimensional prediction-multi-execution coordinated control mechanism enables the system to actively intervene in physical perturbation sources before efficiency has significantly decreased, thereby effectively maintaining the high-efficiency operation of the solid electrolyte electrolysis process.

[0018] 2. When generating control commands, this application maps different predicted disturbance components to dedicated control rules: interface contact degradation triggers pulse current adjustment, lattice channel closure guides electrochemical potential window reset, and bubble coverage initiates ultrasonic stripping. These three operations are independent yet time-aligned, avoiding compensation conflicts or action cancellations caused by using the same control method to address multiple problems in traditional methods. This collaborative logic of "one disturbance, one policy; simultaneous execution" significantly improves the targeting of the control response and the stability of overall efficiency. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is an overall flowchart of the predictive-based synergistic control method for solid electrolyte electrolysis efficiency involved in this application; Figure 2 This is a schematic diagram of the overall structure of the predictive solid electrolyte electrolysis efficiency synergistic control method involved in this application; Figure 3 This is a computer device diagram of the predictive solid electrolyte electrolysis efficiency synergistic control method involved in this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] In one exemplary embodiment, such as Figure 1 As shown, a predictive-based method for synergistic control of solid electrolyte electrolysis efficiency is provided, including: S100: The response control terminal selects the multi-source sensing unit during the solid electrolyte electrolysis process and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source for the prediction algorithm. It should be noted that in the scenario of coordinated control of solid electrolyte electrolysis efficiency, the control end refers to the control host deployed locally or remotely at the electrolysis station, and the multi-source sensing unit includes dedicated sensing devices for monitoring the mechanical, material, and interface states. Multi-dimensional state sensing data specifically refers to the raw or pre-processed data output by these sensing units that reflects the sources of disturbance in the electrolysis process. Traditional electrolysis efficiency control schemes typically rely solely on feedback adjustment based on basic electrochemical parameters such as voltage and current, failing to identify in advance efficiency decay caused by external vibrations, material valence state evolution, or interface bubble accumulation, resulting in control lag and difficulty in maintaining high-efficiency operation. This application uses the selection information of the multi-source sensing unit by the control end as the trigger condition, dynamically specifying the data source participating in the prediction before or during the electrolysis process starts, and directly using the multi-dimensional state sensing data output by the selected sensing unit as the input to the prediction algorithm. This enables the control system to generate forward-looking control commands before significant efficiency disturbances occur, thereby achieving proactive coordinated optimization of solid electrolyte electrolysis efficiency under complex or extreme conditions.

[0023] In some embodiments, step S100 includes steps S110, S120, S130, and S140, as follows: Step S110: Receive the sensing unit selection instruction sent by the control terminal, wherein the sensing unit selection instruction specifies the multi-source sensing units participating in this electrolysis efficiency collaborative control.

[0024] It should be noted that traditional electrolysis control schemes typically use all sensors at a fixed rate, making it impossible to dynamically adjust the sensing dimensions according to operating conditions, resulting in data redundancy and computational delays. This application addresses this by actively issuing sensing unit selection commands from the control terminal, enabling the system to activate only sensing units relevant to the current operating environment. For example, mechanical vibration sensors can be turned off in a vibration-free environment, thereby reducing resource consumption and improving predictive accuracy.

[0025] Step S120: Activate the corresponding mechanical vibration sensor, rare earth valence state monitoring sensor and interface bubble imaging sensor according to the selection instruction of the sensing unit.

[0026] It should be noted that the mechanical vibration sensor is a triaxial accelerometer, installed on the electrolytic cell support structure; the rare earth valence state monitoring sensor is based on in-situ X-ray absorption spectroscopy or electrochemical impedance spectroscopy to invert valence state information; and the interface bubble imaging sensor is a high-resolution infrared or optical imaging device, aimed at the contact interface between the solid electrolyte and the electrode. Each sensor is in a low-power sleep state when not selected, and only outputs valid data after being activated.

[0027] Step S130: Obtain mechanical vibration spectrum data from the activated mechanical vibration sensor, obtain valence characteristic data from the rare earth valence state monitoring sensor, and obtain bubble coverage state data from the interface bubble imaging sensor.

[0028] It should be noted that the mechanical vibration spectrum data is obtained by performing a fast Fourier transform on the time-domain acceleration signal; the valence state characteristic data is characterized by the L3 edge absorption peak shift or equivalent redox potential of a specific rare earth element; and the bubble coverage state data is obtained by calculating the proportion of bubble pixels in the interface area using an image segmentation algorithm. All three types of data are timestamped to ensure spatiotemporal consistency.

[0029] Step S140: Integrate the mechanical vibration spectrum data, the valence state characteristic data, and the bubble coverage state data into multidimensional state perception data, and use it as the input data source for the prediction algorithm.

[0030] Preferably, steps S110 to S140 together construct a scene-adaptive perception input mechanism: step S110 enables the control end to configure the perception dimensions on demand, step S120 ensures that only relevant physical quantities are collected, step S130 provides structured and quantifiable raw features, and step S140 completes the standardized fusion of multi-source heterogeneous data, providing a high-quality and low-redundancy input data source for subsequent prediction algorithms and avoiding prediction deviations caused by irrelevant data interference.

[0031] S200: Connect the prediction algorithm to the input data source and generate an electrolysis efficiency perturbation prediction component based on the multidimensional state perception data. The electrolysis efficiency perturbation prediction component includes the interface contact stability change trend, the lattice channel reconstruction trend, and the three-phase interface effective reaction area decay trend. It should be noted that in the scenario of coordinated control of solid electrolyte electrolysis efficiency, the prediction algorithm refers to a multi-output regression or sequence prediction program deployed in the edge controller. The input data source is the multi-dimensional state-sensing data generated in step S100. The electrolysis efficiency disturbance prediction components are the three structured trend signals output by the algorithm, corresponding to three types of efficiency decay mechanisms: interface, material, and reaction surface. Traditional electrolysis control methods usually only perform hysteresis feedback adjustment based on the current current-voltage efficiency, which cannot predict the efficiency decline caused by external disturbances or internal state evolution, resulting in the control action always lagging behind performance degradation. This application directly inputs the multi-dimensional state-sensing data into the prediction algorithm. By establishing the mapping relationship between mechanical vibration spectrum and interface contact stability, valence state characteristics and lattice channel conductivity, and bubble coverage state and effective reaction area, the three disturbance prediction components are output before the efficiency drops significantly, enabling the control system to have a forward intervention capability, thereby maintaining the stability of electrolysis efficiency under conditions of mechanical disturbance, material state changes, or interface bubble accumulation.

[0032] In some embodiments, step S200 includes steps S210 to S240, as follows: Step S210: Preprocess the multidimensional state-aware data, including time alignment, noise filtering and feature normalization, to obtain a standardized input sequence.

[0033] Among them, time alignment refers to interpolating or resampling data from different sensing units according to a unified timestamp; noise filtering uses moving average or wavelet denoising methods; feature normalization maps data of each dimension to the [0,1] interval.

[0034] Understandably, the mechanical vibration spectrum data, valence state characteristic data, and bubble coverage state data are collected by heterogeneous sensors, resulting in differences in sampling frequency, dimensions, and dynamic range. Directly inputting these into the prediction algorithm would cause the model to become overly sensitive to high-amplitude or high-frequency signals. By ensuring strict temporal correspondence among the three types of data through time alignment, suppressing random sensor interference through noise filtering, and eliminating the influence of dimensions through feature normalization, the prediction algorithm can fairly learn the contribution weights of each dimension to efficiency perturbations.

[0035] Step S220: Input the standardized input sequence into the prediction algorithm, which is a multi-task temporal prediction network that includes a shared bottom-level feature extraction layer and three independent output branches.

[0036] The prediction algorithm receives multidimensional state-aware data within a time window of length T as input and outputs three perturbation trend sequences corresponding to the next K time steps. Each sequence represents the evolution direction and intensity of the corresponding physical perturbation in continuous numerical form.

[0037] In one specific embodiment, the prediction algorithm combines a multi-task learning mechanism with physical perturbation decoupling modeling. The mapping relationship of the prediction algorithm can be established in advance by analyzing historical operating data of the solid electrolyte electrolysis process. For example, multi-source sensing data and corresponding efficiency decay records of more than 1000 hours of continuous operation of the electrolysis equipment under different operating conditions are collected. The temporal correlation patterns between mechanical vibration spectrum, rare earth valence state characteristics, bubble coverage state and interface impedance change rate, lattice channel openness, and effective reaction area decay rate are extracted, and these correlation patterns are solidified into shared feature extraction rules and branch output mapping relationships.

[0038] The prediction algorithm structurally includes: a standardized input layer for receiving multidimensional state-aware data sequences after time alignment, noise filtering, and feature normalization; a shared bottom-level feature extraction layer that uses stacked long short-term memory network units to perform temporal modeling on the input sequence, extract cross-dimensional coupling features, and output a high-dimensional hidden state vector; and three independent output branches, namely a first output branch, a second output branch, and a third output branch, wherein the first output branch generates the interface contact stability change trend based on the high-dimensional hidden state vector, the second output branch generates the lattice channel reconstruction trend, and the third output branch generates the three-phase interface effective reaction area decay trend.

[0039] Step S230: Extract coupling features from the multidimensional state-aware data through the shared bottom-level feature extraction layer, and generate the interface contact stability change trend, lattice channel reconstruction trend and three-phase interface effective reaction area decay trend through the first output branch, the second output branch and the third output branch respectively.

[0040] Understandably, the first output branch consists of a fully connected layer with a linear activation function, and its output is a continuous numerical sequence representing the expected rate of change of the interfacial contact impedance. The second output branch outputs the lattice channel openness index caused by the valence state migration of rare earth elements. The third output branch outputs the predicted curve of the proportion of the bubble-covered area. All three outputs are generated at the same time resolution, which facilitates the timing alignment of subsequent collaborative control strategies.

[0041] Step S240: Combine the trend of interface contact stability change, the trend of lattice channel reconstruction and the trend of decay of effective reaction area of ​​three-phase interface into the electrolysis efficiency disturbance prediction component, and output it to the collaborative control strategy library.

[0042] Preferably, steps S210 to S240 together construct a structured prediction process for multi-source heterogeneous sensing data: step S210 ensures the quality and temporal consistency of input data, step S220 provides a scalable multi-task prediction architecture, step S230 achieves decoupling output of three types of physical disturbance mechanisms, and step S240 completes the format encapsulation of prediction results, providing standardized and resolvable disturbance prediction components for the collaborative control strategy library, so that the subsequent control command generation has clear physical basis and temporal foresight.

[0043] Step S210: Preprocess the multidimensional state-aware data, including time alignment, noise filtering and feature normalization, to obtain a standardized input sequence.

[0044] Among them, time alignment refers to interpolating or resampling data from different sensing units according to a unified timestamp; noise filtering uses moving average or wavelet denoising methods; feature normalization maps data of each dimension to the [0,1] interval.

[0045] Understandably, the mechanical vibration spectrum data, valence state characteristic data, and bubble coverage state data are collected by heterogeneous sensors, resulting in differences in sampling frequency, dimensions, and dynamic range. Directly inputting these into the prediction algorithm would cause the model to become overly sensitive to high-amplitude or high-frequency signals. By ensuring strict temporal correspondence among the three types of data through time alignment, suppressing random sensor interference through noise filtering, and eliminating the influence of dimensions through feature normalization, the prediction algorithm can fairly learn the contribution weights of each dimension to efficiency perturbations.

[0046] Step S220: Input the standardized input sequence into the prediction algorithm, which is a multi-task temporal prediction network that includes a shared bottom-level feature extraction layer and three independent output branches.

[0047] It should be noted that the multi-task temporal prediction network uses a long short-term memory network (LSTM) or a Transformer encoder as a shared bottom feature extraction layer to capture cross-dimensional coupling relationships and temporal dependencies in multi-dimensional state-aware data; the three independent output branches correspond to three types of perturbation mechanisms, namely interface contact stability, lattice channel reconstruction and effective reaction area, to avoid interference between tasks.

[0048] Understandably, the shared underlying feature extraction layer first performs high-dimensional feature mapping on the standardized input sequence to extract implicit patterns such as "vibration-valence coupling features" or "bubble-temperature interaction features". Subsequently, the first output branch receives the high-dimensional feature and outputs the trend of interface contact stability change over the next N time steps, the second output branch outputs the trend of lattice channel reconstruction, and the third output branch outputs the trend of the decay of the effective reaction area of ​​the three-phase interface.

[0049] Step S230: Extract coupling features from the multidimensional state-aware data through the shared bottom-level feature extraction layer, and generate the interface contact stability change trend, lattice channel reconstruction trend and three-phase interface effective reaction area decay trend through the first output branch, the second output branch and the third output branch respectively.

[0050] In some embodiments, generating the electrolysis efficiency perturbation prediction component in step S230 includes steps S231 to S233, as follows: In some embodiments, step S231 includes steps S2311 to S2313, as follows: Step S2311: Perform spectral energy density analysis on the vibration characteristic sub-vectors and calculate the cumulative energy percentage in each frequency range.

[0051] Among them, the vibration feature sub-vector is composed of the energy value sequence divided into fixed frequency intervals after the acceleration signal output by the mechanical vibration sensor is transformed by spectrum transformation.

[0052] Understandably, the entire vibration spectrum is divided into multiple consecutive frequency intervals. Starting from the lowest frequency, the energy value of each interval is accumulated sequentially, and the current accumulated sum is compared with the total energy of the entire frequency band to obtain the cumulative energy percentage corresponding to each frequency point, forming a monotonically rising curve.

[0053] Step S2312: Select the lowest frequency range where the cumulative energy percentage exceeds a preset threshold as the dominant frequency band for mechanical disturbance.

[0054] The preset threshold is 85%.

[0055] Understandably, on the cumulative energy percentage curve, the frequency range corresponding to the position where it first exceeds 85% is the frequency band dominated by mechanical disturbances; this frequency band concentrates the main energy of system vibration and has the greatest impact on the stability of the solid electrolyte and electrode interface.

[0056] Step S2313: Substitute the center frequency and amplitude of the dominant frequency band of the mechanical disturbance into the interface contact impedance response control, and output the rate of change sequence of interface contact impedance within the future time window as the trend of interface contact stability change.

[0057] Among them, the interface contact impedance response control is a time-series response relationship pre-calibrated based on the structural stiffness of the electrolytic cell.

[0058] Understandably, by inputting the center frequency of the dominant frequency band and its corresponding amplitude into the model, the model extrapolates the rate of change of the interface contact impedance over the next few seconds based on historical calibration data, and outputs a continuous trend sequence to characterize the deterioration tendency of the interface contact state.

[0059] In some embodiments, step S232 includes steps S2321 to S2323, as follows: Step S2321: Extract the L3 edge absorption peak shift and full width at half maximum (FWHM) of a specific rare earth element from the valence state eigenvector.

[0060] Among them, the valence state characteristic vector is generated by fitting the peak shape of the spectral data collected in real time by the in-situ X-ray absorption spectrometer.

[0061] It is understandable that for europium-containing solid electrolytes, the X-ray absorption spectrum exhibits a characteristic absorption peak at the L3 edge; the center position and width of this peak are determined by Gaussian fitting; the offset of the center position relative to the standard trivalent europium reference peak reflects the degree of valence reduction, and the width of the peak reflects the degree of dispersion of the valence distribution.

[0062] Step S2322: Calculate the redox activity index based on the absorption peak shift and full width at half maximum (FWHM), wherein the redox activity index characterizes the potential for change in the concentration of mobile vacancies in the crystal lattice.

[0063] Among them, the redox activity index takes into account both the magnitude of valence state shift and the degree of peak concentration.

[0064] It is understandable that if the absorption peak shifts significantly to lower energies and has a sharp peak shape, it indicates that a large number of europium ions in the material are undergoing synchronous reduction and the crystal structure is in an active reconstruction state, at which point the redox activity index is high; conversely, if the shift is weak or the peak shape is diffuse, the activity index is low.

[0065] Step S2323: Input the redox activity index into the lattice channel conductivity evolution model and output the evolution curve of the lattice channel openness index within the future time window as the lattice channel reconstruction trend.

[0066] The lattice channel conductivity evolution model is a time-series mapping relationship established based on material experimental data.

[0067] Understandably, the model records the degree to which lattice channels open over time under different redox activity levels; when a high activity index is input, the model outputs a curve that rises rapidly and then stabilizes, indicating that the channel is being effectively opened; this curve is used to guide the dynamic adjustment of the electrochemical potential window.

[0068] In some embodiments, step S233 includes steps S2331 to S2333, as follows: Step S2331: Perform connected component labeling on the bubble mask image in the interface feature sub-vector to obtain the set of pixel coordinates for each bubble region.

[0069] The bubble mask image is a binary image after image segmentation, and the white area represents the location covered by the bubble.

[0070] Understandably, the eight-neighbor connectivity criterion is used to label regions in the bubble mask image, grouping interconnected white pixels into the same bubble region, ultimately resulting in multiple independent bubble regions, each containing the spatial location information of all its pixels.

[0071] Step S2332: Based on the spatial location and area of ​​each bubble region, calculate the spatial entropy and temporal change rate of the bubble coverage distribution to form the spatiotemporal distribution entropy of bubble coverage.

[0072] Among them, spatial entropy reflects the degree of dispersion or aggregation of bubbles at the interface, and temporal change rate reflects how fast the degree of dispersion changes over time.

[0073] It is understandable that when bubbles are evenly distributed across the entire interface, the spatial entropy is high; when bubbles gradually gather into a large area, the spatial entropy decreases; if the spatial entropy decreases significantly in two consecutive frames, the temporal change rate is negative and the absolute value is large; by weighting and fusing the spatial entropy and the temporal change rate, the spatiotemporal distribution entropy of bubble coverage is obtained.

[0074] Step S2333: Substitute the spatiotemporal distribution entropy of the bubble coverage into the effective reaction area decay kinetic model, and output the decay rate of the effective reaction area ratio within the future time window as the decay trend of the effective reaction area of ​​the three-phase interface.

[0075] The effective reaction area decay kinetic model is based on the response relationship calibrated from bubble evolution experiments.

[0076] Understandably, the model indicates that the lower the spatiotemporal distribution entropy of the bubble coverage and the faster the descent rate, the higher the decay rate of the effective reaction area. Based on this, the model outputs a prediction sequence indicating the rate at which the interface area available for electrochemical reaction decreases in the next few seconds, which is used to determine whether to trigger the ultrasonic ablation operation.

[0077] It should be noted that steps S231 to S233 address three independent physical mechanisms—mechanical disturbance, material state, and interface morphology—each employing its own dedicated feature extraction path and physical mapping model to avoid prediction distortion caused by the aliasing of multi-source signals. The trend signals output by the three sub-steps are strictly aligned in the time dimension and all represent the future efficiency disturbance intensity in the form of continuous numerical sequences, providing a unified benchmark for the timing coordination of subsequent collaborative control commands.

[0078] Preferably, step S231 achieves precise capture of harmful vibrations through dominant frequency band identification; step S232 establishes a bridge between the material's microstructure and macroscopic properties through redox activity indicators; and step S233 quantifies the critical state of interface failure through distribution entropy. These three steps together constitute a multi-dimensional disturbance decoupling prediction mechanism for the unique behavior of solid-state electrolytes, enabling the control system to distinguish "why efficiency decreases" and thus implement "targeted intervention," significantly different from traditional adjustment methods based solely on current-voltage efficiency feedback.

[0079] Step S240: Combine the trend of interface contact stability change, the trend of lattice channel reconstruction and the trend of decay of effective reaction area of ​​three-phase interface into the electrolysis efficiency disturbance prediction component, and output it to the collaborative control strategy library.

[0080] Preferably, steps S210 to S240 together construct a structured prediction process for multi-source heterogeneous sensing data: step S210 ensures the quality and temporal consistency of input data, step S220 provides a scalable multi-task prediction architecture, step S230 achieves decoupling output of three types of physical disturbance mechanisms, and step S240 completes the format encapsulation of prediction results, providing standardized and resolvable disturbance prediction components for the collaborative control strategy library, so that the subsequent control command generation has clear physical basis and temporal foresight.

[0081] S300: Retrieve the collaborative control strategy library and generate pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions and interface stripping trigger instructions according to the electrolysis efficiency disturbance prediction components. It should be noted that in the scenario of coordinated control of solid electrolyte electrolysis efficiency, the coordinated control strategy library refers to a set of multiple mapping rules pre-stored in the controller. Each set of rules defines a deterministic relationship between a specific disturbance prediction component and the corresponding control command. The pulse electrolysis parameter adjustment command is used to regulate the current waveform characteristics of the power supply output, the electrochemical potential window control command is used to limit the working potential range of the electrolysis process, and the interface stripping trigger command is used to activate the physical stripping device to remove the interface gas film. Traditional electrolysis control methods usually use a single feedback loop to adjust the overall efficiency, which cannot implement differentiated intervention for efficiency disturbances from different physical sources, resulting in the control effects canceling each other out or insufficient response under the condition of multiple disturbances coexisting. After obtaining the three electrolysis efficiency disturbance prediction components, this application calls the corresponding mapping rules in the coordinated control strategy library to independently generate three types of control commands, enabling the control system to simultaneously address three types of problems: interface contact degradation, lattice channel closure, and bubble coverage, thereby achieving multi-dimensional coordinated optimization of solid electrolyte efficiency under complex operating conditions.

[0082] In some embodiments, step S300 includes steps S310 to S330, as follows: Step S310: Retrieve the pulse electrolysis control rule corresponding to the interface contact stability change trend from the collaborative control strategy library, and generate a pulse electrolysis parameter adjustment command based on the amplitude and rate of change of the interface contact stability change trend.

[0083] The pulse electrolysis parameters include pulse frequency, pulse width, and current amplitude.

[0084] Understandably, the collaborative control strategy library contains multiple pre-stored mapping relationships: when the interface contact stability trend shows a rapid increase in impedance with a large amplitude, a high-frequency, short-pulse-width strong disturbance pulse rule is invoked; when the trend change is gradual, a low-frequency, long-pulse-width sustaining pulse rule is invoked. The controller matches the closest rule entry based on the specific value of the current trend and outputs the corresponding pulse electrolysis parameter adjustment command to drive the power supply to output a current with a specific waveform, thereby suppressing contact degradation caused by interface micro-movements.

[0085] Step S320: Retrieve the electrochemical potential control rule corresponding to the lattice channel reconstruction trend from the collaborative control strategy library, and generate an electrochemical potential window control instruction based on the openness index and evolution direction of the lattice channel reconstruction trend.

[0086] The electrochemical potential window refers to the range of the lowest and highest potentials that can be applied during electrolysis.

[0087] Understandably, if the lattice channel reconstruction trend shows a continuous increase in the openness index, indicating a positive shift in the valence state of rare earth elements, the "widening the upper limit of electrochemical potential" rule is invoked to allow the application of higher potentials to accelerate ion transport. Conversely, if the openness index tends towards saturation or declines, the "narrowing the electrochemical potential window" rule is invoked to prevent excessive polarization from triggering side reactions. Based on the trend's evolution direction and the current openness level, the controller selects appropriate rules from the strategy library to generate electrochemical potential window control instructions, which are used to limit the potential boundaries of subsequent electrolysis operations.

[0088] Step S330: Retrieve the interface stripping trigger rule corresponding to the decay trend of the effective reaction area of ​​the three-phase interface from the collaborative control strategy library, and generate an interface stripping trigger command based on the decay rate of the effective reaction area decay trend and the current coverage status.

[0089] The interface peeling trigger command includes whether to start the ultrasonic peeling device and the ultrasonic power level.

[0090] Understandably, the collaborative control strategy library defines multi-level triggering conditions: when the attenuation rate exceeds the first threshold and the current bubble coverage area is less than 50%, a low-power ultrasonic trigger command is generated; when the attenuation rate exceeds the second threshold (higher than the first threshold) or the coverage area has exceeded 70%, a high-power ultrasonic trigger command is generated; if the attenuation rate is below the threshold, no trigger command is generated. The controller compares the current attenuation trend and coverage status with the rule conditions and outputs the corresponding interface stripping trigger command to activate the ultrasonic transducer to perform the air film removal operation.

[0091] It should be noted that steps S310, S320, and S330, respectively targeting three independent disturbance mechanisms, retrieve their respective exclusive control rules from the same collaborative control strategy library and generate differentiated control commands based on the specific dynamic characteristics of the predicted components. These three commands can be generated in parallel in time and executed synchronously, together forming a collaborative intervention system for multidimensional efficiency disturbances in solid-state electrolytes, avoiding insufficient response or conflicting actions of traditional single feedback control under complex operating conditions.

[0092] S400: Based on the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command, the power output operation, thermoelectric coupling operation, and ultrasonic stripping operation are synchronously coordinated to obtain the electrolysis efficiency collaborative control result, and the collaborative control result is fed back to the control terminal to update the prediction algorithm parameters.

[0093] In some embodiments, step S400 includes steps S410 to S440, as follows: Step S410: Configure the current waveform parameters of the power output unit according to the pulse electrolysis parameter adjustment command, and perform pulse electrolysis operation.

[0094] The current waveform parameters include pulse frequency, pulse width, and peak current amplitude.

[0095] Understandably, after receiving the pulse electrolysis parameter adjustment command, the controller writes the parameter value in the command into the waveform generator register of the programmable power supply; the power supply output unit then periodically outputs square wave current according to the set pulse frequency, the duration of each pulse is the pulse width specified by the command, and the current amplitude is the peak current specified by the command; this operation is used to apply micro-perturbation in the early stage of the interface contact impedance rise, and suppress the contact degradation caused by solid-solid interface micro-movement.

[0096] Step S420: Adjust the heating or cooling power of the thermoelectric coupling unit according to the electrochemical potential window control command, so that the solid electrolyte is maintained within the temperature range corresponding to the electrochemical potential window.

[0097] Among them, the thermoelectric coupling unit is a Peltier thermoelectric module, and its working mode is determined by the upper and lower limits of the electrochemical potential window.

[0098] Understandably, if the electrochemical potential window control command indicates an expansion of the upper potential limit, the controller increases the heating power of the thermoelectric coupling unit to raise the electrolyte temperature to a higher set point, thereby improving ionic conductivity and supporting high-potential operation; if the command indicates a narrowing of the window, the heating power is reduced or cooling is activated to prevent the side reactions from intensifying at high temperatures; the temperature control target corresponds one-to-one with the electrochemical potential window, ensuring that the material operates in a safe and efficient state.

[0099] Step S430: According to the interface peeling trigger command, start the ultrasonic peeling unit and perform the interface air film removal operation at the specified power and duration.

[0100] The ultrasonic stripping unit is a piezoelectric ultrasonic transducer, which is installed on the outer wall of the electrolytic cell and coupled to the back of the solid electrolyte.

[0101] Understandably, when the interface stripping trigger command includes a "high power" flag, the controller drives the ultrasonic stripping unit to operate at its maximum rated power for 500 milliseconds; if it is a "low power" flag, it operates at 50% of the rated power for 200 milliseconds. The ultrasonic vibration is transmitted through the solid to the electrolyte-electrode interface, causing the attached bubbles to resonate and detach, restoring the effective reaction area. After the operation is completed, the ultrasonic unit is automatically turned off to avoid continuous vibration interfering with the electrolysis process.

[0102] Step S440: Collect the actual execution parameters and corresponding electrolysis efficiency feedback data of pulse electrolysis operation, thermoelectric coupling operation and ultrasonic stripping operation, form a collaborative control result, and send the collaborative control result to the control terminal to update the prediction algorithm parameters.

[0103] The actual execution parameters include the actual output pulse current waveform, the real-time temperature of the thermoelectric module, the ultrasonic power and the duration of action; the electrolysis efficiency feedback data includes the real-time hydrogen yield, the cell voltage fluctuation amplitude and the current efficiency.

[0104] Understandably, the built-in sensors in each execution unit transmit actual parameters back to the controller; simultaneously, the efficiency monitoring module collects data from the hydrogen flow meter and the electricity meter to calculate the current electrolysis efficiency; the controller packages the execution parameters and efficiency data into a collaborative control result and uploads it to the control terminal through the communication interface; the control terminal uses this result to perform online correction of the mapping relationship in the prediction algorithm, such as adjusting the correlation weight between the dominant frequency band and the impedance change rate, or correcting the correspondence between bubble distribution entropy and decay rate, thereby improving the accuracy of subsequent predictions.

[0105] It should be noted that steps S410, S420, and S430 respectively implement three types of physical interventions: electrical, thermal, and mechanical. These three interventions are initiated synchronously based on the same control cycle, forming a multi-dimensional synergistic effect. Step S440 completes the closed-loop feedback, enabling the system to continuously learn and optimize. These four sub-steps together constitute a complete control chain from instruction to execution to feedback, ensuring that the solid-state electrolyte electrolysis efficiency remains stable under complex disturbance environments.

[0106] It should be noted that in the scenario of coordinated control of solid electrolyte electrolysis efficiency, the pulse electrolysis parameter adjustment command is used to regulate the current waveform characteristics of the power supply output, the electrochemical potential window control command is used to limit the working potential boundary of the electrolysis process, and the interface stripping trigger command is used to activate the ultrasonic device to perform physical cleaning operations. The power supply output operation refers to driving the electrolysis current according to the specified pulse parameters, the thermoelectric coupling operation refers to dynamically adjusting the thermal management unit according to the electrochemical potential window to maintain the electrolyte working temperature, and the ultrasonic stripping operation refers to applying ultrasonic vibration at a specified power and duration to strip the interfacial gas film. Traditional electrolysis control methods usually treat each execution action in isolation, or only perform single compensation after the efficiency has dropped significantly, which cannot achieve the timing coordination and target alignment of multiple actuators. After receiving three control commands, this application coordinates the three operations synchronously through a unified scheduling mechanism: the power output operation adjusts the current waveform in real time according to the pulse parameters; the thermoelectric coupling operation ensures temperature stability within the electrochemical potential window; and the ultrasonic stripping operation actively intervenes in the early stage of gas film formation. The three are aligned on the time axis and focused on maintaining high electrolysis efficiency, thus forming a closed-loop coordinated control result. This result includes actual execution parameters and efficiency feedback data, which are transmitted back to the control end for online updating of the mapping relationship or strategy library rules in the prediction algorithm, thereby achieving continuous optimization of control performance.

[0107] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0108] Based on the same inventive concept, this application also provides a prediction-based solid electrolyte electrolysis efficiency synergistic control system. The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the prediction-based solid electrolyte electrolysis efficiency synergistic control system provided below can be found in the limitations of the prediction-based solid electrolyte electrolysis efficiency synergistic control method described above, and will not be repeated here.

[0109] In one exemplary embodiment, refer to Figure 2 A prediction-based synergistic control system for solid electrolyte electrolysis efficiency is provided, including: The sensing data scheduling module responds to the control terminal's selection information for multi-source sensing units during solid electrolyte electrolysis, and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source for the prediction algorithm. The efficiency disturbance prediction module connects to the input data source and uses a prediction algorithm to generate electrolysis efficiency disturbance prediction components based on the multidimensional state perception data. The electrolysis efficiency disturbance prediction components include the trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of effective reaction area of ​​three-phase interface. The collaborative control instruction module retrieves the collaborative control strategy library and generates pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions, and interface stripping trigger instructions based on the electrolysis efficiency disturbance prediction components. The coordination feedback module executes the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command to synchronously coordinate the power output operation, thermoelectric coupling operation, and ultrasonic stripping operation, thereby obtaining the electrolysis efficiency coordinated control result. The coordinated control result is then fed back to the control terminal to update the prediction algorithm parameters. The modules in the aforementioned predictive solid-state electrolyte electrolysis efficiency collaborative control system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0110] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 3As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a predictive method for coordinated control of solid-state electrolyte electrolysis efficiency. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0111] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0112] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0113] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0114] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0115] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0116] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0117] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0118] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for coordinated control of solid electrolyte electrolysis efficiency based on prediction, characterized in that, include: The response control terminal selects the multi-source sensing unit during the solid electrolyte electrolysis process and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source for the prediction algorithm. A prediction algorithm is connected to the input data source to generate an electrolysis efficiency perturbation prediction component based on the multidimensional state-aware data. Based on the electrolysis efficiency disturbance prediction component, a target control strategy is matched from a preset collaborative control strategy library, and a multi-dimensional collaborative control command is calculated and generated based on the target control strategy and the electrolysis efficiency disturbance prediction component. The multi-physics field execution unit is controlled to perform collaborative control operations according to the multi-dimensional collaborative control command. The electrolysis efficiency collaborative control result is calculated based on the state data collected after the collaborative control operation is executed, and the electrolysis efficiency collaborative control result is fed back to the control terminal.

2. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 1, characterized in that: The response control terminal selects multi-source sensing units during the solid electrolyte electrolysis process, and uses the multi-dimensional state sensing data output by the selected sensing units as the input data source for the prediction algorithm, including: Receive the sensing unit selection command sent by the control terminal; According to the instruction selected by the sensing unit, the corresponding mechanical vibration sensor, rare earth valence state monitoring sensor and interface bubble imaging sensor are activated. Mechanical vibration spectrum data is acquired from the activated mechanical vibration sensor, valence characteristic data is acquired from the rare earth valence state monitoring sensor, and bubble coverage state data is acquired from the interface bubble imaging sensor. The mechanical vibration spectrum data, the valence state characteristic data, and the bubble coverage state data are integrated into multidimensional state perception data, which is then used as the input data source for the prediction algorithm.

3. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 1, characterized in that: The step of connecting the prediction algorithm to the input data source and generating electrolysis efficiency perturbation prediction components based on the multidimensional state-aware data includes: The multidimensional state-aware data is preprocessed to obtain a standardized input sequence; The standardized input sequence is input into a prediction algorithm, which includes a standardized input layer and a shared bottom-level feature extraction layer. The shared bottom feature extraction layer extracts coupling features from the preprocessed multidimensional state-aware data, and generates the interface contact stability change trend, lattice channel reconstruction trend and three-phase interface effective reaction area decay trend through the first output branch, the second output branch and the third output branch. The trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of effective reaction area of ​​the three-phase interface are combined into the electrolysis efficiency disturbance prediction component and output to the collaborative control strategy library.

4. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 1, characterized in that: The step of matching a target control strategy from a preset collaborative control strategy library based on the electrolysis efficiency disturbance prediction component, and calculating and generating a multi-dimensional collaborative control command based on the target control strategy and the electrolysis efficiency disturbance prediction component, includes: The pulse electrolysis control rules corresponding to the interface contact stability change trend are retrieved from the collaborative control strategy library, and pulse electrolysis parameter adjustment instructions are generated according to the amplitude and rate of change of the interface contact stability change trend. The electrochemical potential control rules corresponding to the lattice channel reconstruction trend are retrieved from the collaborative control strategy library, and an electrochemical potential window control instruction is generated based on the openness index and evolution direction of the lattice channel reconstruction trend. The interface stripping trigger rule corresponding to the decay trend of the effective reaction area of ​​the three-phase interface is retrieved from the collaborative control strategy library, and an interface stripping trigger command is generated based on the decay rate of the effective reaction area decay trend and the current coverage status.

5. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 1, characterized in that: The process of regulating the multi-physics execution unit to perform coordinated control operations according to the multi-dimensional coordinated control command, calculating the electrolysis efficiency coordinated control result based on the state data collected after the coordinated control operation is executed, and feeding back the electrolysis efficiency coordinated control result to the control terminal includes: According to the pulse electrolysis parameter adjustment command, configure the current waveform parameters of the power output unit and execute the pulse electrolysis operation; According to the electrochemical potential window control command, the heating or cooling power of the thermoelectric coupling unit is adjusted to control the solid electrolyte within the temperature range corresponding to the electrochemical potential window. According to the interface peeling trigger command, the ultrasonic peeling unit is started and the interface air film removal operation is performed according to the specified power and duration. The actual execution parameters and corresponding electrolysis efficiency feedback data of pulse electrolysis operation, thermoelectric coupling operation and ultrasonic stripping operation are collected to form a collaborative control result, which is then sent to the control terminal to update the prediction algorithm parameters.

6. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 3, characterized in that: The process of extracting coupling features from multidimensional state-aware data through the shared bottom-level feature extraction layer, and generating trends in interface contact stability, lattice channel reconstruction, and the decay of the effective reaction area of ​​the three-phase interface via the first, second, and third output branches respectively, includes: Based on the vibration feature sub-vectors in the standardized input sequence, the dominant frequency band of mechanical disturbance is identified and mapped to the interface contact impedance response control to generate the interface contact stability change trend. Based on the valence state feature sub-vectors in the standardized input sequence, rare earth element redox activity indicators are extracted and mapped to the lattice channel conduction capability evolution model to generate lattice channel reconstruction trends. Based on the interface feature sub-vectors in the standardized input sequence, the spatiotemporal distribution entropy of bubble coverage is quantified and mapped to the effective reaction area decay kinetic model to generate the three-phase interface effective reaction area decay trend.

7. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 6, characterized in that: The process of identifying the dominant frequency band of mechanical disturbance based on the vibration feature sub-vectors in the standardized input sequence and mapping it to interface contact impedance response control to generate the interface contact stability change trend includes: Spectral energy density analysis is performed on the vibration characteristic sub-vectors to calculate the cumulative energy percentage in each frequency range; The lowest frequency range where the cumulative energy percentage exceeds a preset threshold is selected as the dominant frequency band for mechanical disturbance; Substituting the center frequency and amplitude of the dominant frequency band of the mechanical disturbance into the interface contact impedance response control, the output of the rate of change sequence of interface contact impedance within the future time window is used as the trend of interface contact stability change.

8. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 6, characterized in that: The step of extracting rare earth element redox activity indices based on valence state feature sub-vectors in the standardized input sequence and mapping them to a lattice channel conductivity evolution model to generate lattice channel reconstruction trends includes: Extract the L3 side absorption peak shift and full width at half maximum (FWHM) of a specific rare earth element from the valence state eigenvectors; The redox activity index is calculated based on the absorption peak shift and the full width at half maximum (FWHM), and the redox activity index characterizes the potential for changes in the concentration of mobile vacancies in the crystal lattice. The redox activity index is input into the lattice channel conductivity evolution model, and the evolution curve of the lattice channel openness index within the future time window is output as the lattice channel reconstruction trend.

9. The method for coordinated control of solid electrolyte electrolysis efficiency based on prediction as described in claim 6, characterized in that: The step of quantifying the spatiotemporal distribution entropy of bubble coverage based on the interface feature sub-vectors in the standardized input sequence and mapping it to the effective reaction area decay kinetic model to generate the three-phase interface effective reaction area decay trend includes: Connected component labeling is performed on the bubble mask image in the interface feature sub-vector to obtain the set of pixel coordinates for each bubble region; Based on the spatial location and area of ​​each bubble region, the spatial entropy and temporal rate of change of bubble coverage distribution are calculated to form the spatiotemporal distribution entropy of bubble coverage. Substituting the spatiotemporal distribution entropy of the bubble coverage into the effective reaction area decay kinetic model, the decay rate of the effective reaction area ratio within the future time window is output as the decay trend of the effective reaction area of ​​the three-phase interface.

10. A predictive solid-state electrolyte electrolysis efficiency collaborative control system, employing the predictive solid-state electrolyte electrolysis efficiency collaborative control method as described in any one of claims 1 to 9, characterized in that, include: The sensing data scheduling module responds to the control terminal's selection information for multi-source sensing units during solid electrolyte electrolysis, and uses the multi-dimensional state sensing data output by the selected sensing unit as the input data source for the prediction algorithm. The efficiency disturbance prediction module connects to the input data source and uses a prediction algorithm to generate electrolysis efficiency disturbance prediction components based on the multidimensional state perception data. The electrolysis efficiency disturbance prediction components include the trend of interface contact stability change, the trend of lattice channel reconstruction, and the trend of decay of effective reaction area at the three-phase interface. The collaborative control instruction module retrieves the collaborative control strategy library and generates pulse electrolysis parameter adjustment instructions, electrochemical potential window control instructions, and interface stripping trigger instructions based on the electrolysis efficiency disturbance prediction components. The execution coordination feedback module synchronously coordinates the power output operation, thermoelectric coupling operation, and ultrasonic stripping operation according to the pulse electrolysis parameter adjustment command, the electrochemical potential window control command, and the interface stripping trigger command to obtain the electrolysis efficiency coordinated control result, and feeds the coordinated control result back to the control terminal to update the prediction algorithm parameters.

Images