Method and system for controlling electrolysis of electronic grade perfluoroheptyl ether

Abstract

The application discloses an electrolysis process control method and system for electronic-grade perfluoroheptyl ether, and relates to the technical field of intelligent control. The method comprises the following steps: performing ultra-short-term electrolysis state tracking to obtain an electrolysis state frame sequence; S1: locating a real-time reaction stage; S2: performing control parameter matching to obtain a target control parameter set; S3: performing side reaction impurity prediction to obtain a parameter allowable fluctuation baseline; S4: performing product quality trend prediction to obtain a product qualified interval; and S5: correcting parameter deviation until the parameter deviation meets the parameter allowable fluctuation baseline and falls within the product qualified interval, and then triggering a cyclic iteration step S1 to S5 to perform full-process dynamic closed-loop control until a preset reaction endpoint determination condition is met, and electrolysis power supply and process control are stopped. The application solves the technical problems of electrolysis process regulation lag, increased side reactions and unstable product quality in the prior art, realizes dynamic regulation of the electrolysis process, and improves the technical effect of product quality stability.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to a method and system for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether. Background Technology

[0002] Perfluoroheptyl ethyl ether (PFEE), a fluorinated fine chemical with excellent chemical stability and insulating properties, has high application value in the electronics industry and semiconductor manufacturing. Currently, the preparation of electronic-grade PFEE typically employs electrochemical synthesis, controlling process parameters such as electrolysis current, voltage, temperature, and material concentration to achieve the target product. However, in actual electrolysis reactions, the reaction system exhibits significant dynamic changes. Different reaction stages show variations in electrochemical state, material conversion rates, and the tendency for side reactions. Traditional control methods based on fixed process parameters or manual experience are insufficient to adapt to these changes in a timely manner, easily leading to parameter adjustment lags and deviations from optimal reaction conditions. Furthermore, the formation of side reaction products during electrolysis affects the purity of the target product, and changes in product quality usually require subsequent testing for confirmation, resulting in a time delay between process control and quality feedback, making real-time optimization of the electrolysis process difficult. Summary of the Invention

[0003] This application provides a method and system for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether, which solves the technical problems of delayed electrolysis process control in the prior art, leading to increased side reactions and unstable product quality.

[0004] The first aspect of this application provides a method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether, the method comprising: After the electrolytic synthesis reaction is started, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain an electrolytic state frame sequence; S1: Multi-scale process features are extracted from the electrolytic state frame sequence, and reaction stage clustering is performed to locate the real-time reaction stage; S2: The latest state frame is extracted from the electrolytic state frame sequence, and control parameters are matched with the real-time reaction stage to obtain a target control parameter set; S3: Side reaction impurities are predicted based on historical state trend features, the latest state frame, the real-time reaction stage, and the target control parameter set to obtain a parameter allowable fluctuation baseline; S4: Product quality trend is predicted based on the latest state frame and the target control parameter set to obtain a product qualified range; S5: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, the parameter deviation is corrected based on the parameter allowable fluctuation baseline until the parameter deviation meets the parameter allowable fluctuation baseline and the product quality trend falls within the product qualified range, triggering iterative steps S1 to S5 to perform full-process dynamic closed-loop control until the preset reaction endpoint judgment condition is met, and electrolytic power supply and process control are stopped.

[0005] A second aspect of this application provides a control system for the electrolysis process of electronic-grade perfluoroheptyl ethyl ether, the system comprising: The state acquisition module is used to perform ultra-short-term electrolytic state tracking through multi-source online sensing nodes after the electrolytic synthesis reaction is started, and obtain an electrolytic state frame sequence. The stage identification module is used to perform S1, extract multi-scale process features from the electrolytic state frame sequence, perform reaction stage clustering identification, and locate the real-time reaction stage. The parameter matching module is used to perform S2, extract the latest state frame from the electrolytic state frame sequence, and perform control parameter matching in combination with the real-time reaction stage to obtain the target control parameter set. The impurity prediction module is used to perform S3, perform secondary impurity prediction based on historical state trend features, the latest state frame, the real-time reaction stage, and the target control parameter set. The reaction impurity prediction process yields the baseline for allowable parameter fluctuations. The quality prediction module executes step S4, predicting the product quality trend based on the latest state frame and the target control parameter set to obtain the product qualification range. The closed-loop control module executes step S5, adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, and correcting the parameter deviation based on the baseline for allowable parameter fluctuations until the parameter deviation meets the baseline for allowable parameter fluctuations and the product quality trend falls within the product qualification range. This triggers iterative steps S1 to S5, performing dynamic closed-loop control throughout the entire process until the preset reaction endpoint determination condition is met, at which point the electrolysis power supply and process control are stopped.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: After the electrolytic synthesis reaction is initiated, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain an electrolytic state frame sequence. Next, S1: Multi-scale process features are extracted from the electrolytic state frame sequence, and reaction stage clustering is performed to locate the real-time reaction stage. Simultaneously, S2: The latest state frame is extracted from the electrolytic state frame sequence and combined with the real-time reaction stage for control parameter matching to obtain a target control parameter set. Then, S3: Based on historical state trend characteristics, the latest state frame, the real-time reaction stage, and the target control parameter set, side reaction impurities are predicted to obtain a baseline for allowable parameter fluctuations. S4: Based on the latest state frame and the target control parameter set, product quality trend is predicted to obtain a product qualification range. Finally, S5: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, parameter deviations are corrected based on the baseline for allowable parameter fluctuations until the parameter deviations meet the baseline for allowable parameter fluctuations and the product quality trend falls within the product qualification range. This triggers iterative steps S1 to S5, performing dynamic closed-loop control throughout the entire process until the preset reaction endpoint determination conditions are met, at which point electrolytic power supply and process control are stopped. This invention solves the technical problems of lagging electrolysis process control in existing technologies, which leads to increased side reactions and unstable product quality. It achieves dynamic control of the electrolysis process and improves the stability of product quality. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 A schematic diagram of the electrolysis process control method for electronic-grade perfluoroheptyl ethyl ether provided in the embodiments of this application; Figure 2 A comparison chart of impurity content changes before and after parameter fluctuation control is provided for embodiments of this application; Figure 3 A schematic diagram of the control system structure for the electrolysis process of electronic-grade perfluoroheptyl ethyl ether provided in this application embodiment.

[0009] Figure labeling: 11 Status acquisition module, 12 Stage identification module, 13 Parameter matching module, 14 Impurity prediction module, 15 Quality prediction module, 16 Closed-loop control module. Detailed Implementation

[0010] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0011] Example 1, as Figure 1 As shown, this application provides a method for controlling the electrolysis process of electronic-grade perfluoroheptyl ether, wherein the method includes: After the electrolytic synthesis reaction is started, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain the electrolytic state frame sequence.

[0012] Furthermore, after the electrolytic synthesis reaction is initiated, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain an electrolytic state frame sequence, including: The multi-source online sensing nodes are pre-deployed in the electrolytic cell. These nodes include voltage and current acquisition nodes located at the electrode leads, raw material concentration sensing nodes located in the feed circuit, online component detection nodes located at the tail gas outlet of the electrolytic cell, and temperature and moisture sensing nodes located in the inner cavity of the cell. Using 1 / K of the total electrolysis reaction time as an ultra-short-term sliding acquisition window, the multi-source online sensing nodes are synchronously controlled to perform data acquisition on the electrolytic cell at a preset sampling frequency, and output the electrolysis status frame sequence.

[0013] The multi-source online sensing nodes are pre-deployed in the electrolytic cell. These nodes include voltage and current acquisition nodes located at the electrode leads, raw material concentration sensing nodes located in the feed circuit, online component detection nodes located at the electrolytic cell tail gas outlet, and temperature and moisture sensing nodes located within the cell cavity. The voltage and current acquisition nodes acquire the instantaneous cell voltage and operating current during the electrolysis power supply process using current and voltage sampling modules, respectively, and calculate the current density based on the effective reaction area of ​​the electrodes. The raw material concentration sensing nodes acquire data on the concentration changes of the raw material precursors entering the electrolytic cell using online concentration detection. The online component detection nodes continuously detect the tail gas generated during electrolysis using a gas component analysis unit, obtaining the molar percentage of hydrogen and the molar percentage of low-fluorine byproducts. The temperature and moisture sensing nodes acquire the reaction temperature and system moisture content inside the electrolytic cell, respectively, to obtain environmental parameters affecting the electrolysis reaction rate and the generation of byproducts.

[0014] After clock synchronization of all multi-source online sensing nodes is completed, 1 / K of the total electrolysis reaction time is used as an ultra-short-term sliding acquisition window, where K is a preset time segmentation coefficient used to limit the time length corresponding to a single state tracking. Within each ultra-short-term sliding acquisition window, each sensing node is controlled to synchronously perform data acquisition according to a preset sampling frequency, and the acquired data from different sources are time-aligned based on a unified timestamp. The voltage and current data, raw material concentration data, exhaust gas component data, temperature data, and moisture data corresponding to the same sampling moment are fused to generate a single electrolysis state frame. ,in, This represents the electrolysis state frame at time t; Indicates instantaneous slot voltage; Indicates instantaneous operating current; Indicates the current current density; Indicates the concentration of the raw material precursor; Indicates the molar percentage of hydrogen; Indicates the molar percentage of low-fluorine byproducts; Indicates the instantaneous reaction temperature; This indicates the instantaneous moisture content.

[0015] Furthermore, the multiple electrolysis state frames acquired continuously are arranged in chronological order to form an electrolysis state frame sequence: ,in, This represents the electrolysis state frame sequence corresponding to the current moment; This indicates the number of state frames contained within the ultra-short-term sliding acquisition window. By continuously updating the ultra-short-term sliding acquisition window, the electrolysis state frame sequence reflects in real time the changes in electrochemical parameters, material states, and side reaction trends during the electrolytic synthesis process, providing continuous dynamic state data for subsequent reaction stage clustering and identification, control parameter matching, side reaction impurity prediction, and product quality trend prediction.

[0016] Furthermore, the latest status frame includes the instantaneous tank voltage, instantaneous operating current, and current density collected and fed back by the voltage and current acquisition node; the raw material precursor concentration collected and fed back by the raw material concentration sensing node; the molar ratio of hydrogen and the molar ratio of low-fluorine by-products collected and fed back by the online component detection node; the instantaneous reaction temperature collected and fed back by the temperature sensing node; and the instantaneous moisture content collected and fed back by the moisture sensing node.

[0017] Specifically, the voltage and current acquisition node acquires the instantaneous cell voltage during the current electrolysis process in real time through a voltage detection module and a current detection module located at the electrode lead-out end. and instantaneous operating current And combined with the effective reaction area of ​​the pre-stored electrode The current current density is calculated. ,in: ,in, Indicates the current current density; Indicates instantaneous operating current; This indicates the effective reaction area of ​​the electrode.

[0018] The raw material concentration sensing node, through an online concentration detection unit installed in the feed circuit, continuously samples and detects the raw material precursor entering the electrolyzer to obtain the current raw material precursor concentration. The online component detection node is used to reflect the supply status of reactants; it uses a gas component analysis unit located at the tail gas outlet of the electrolyzer to detect the tail gas generated during the electrolysis reaction in real time and obtain the hydrogen molar ratio. molar percentage of low-fluorine byproducts It is used to characterize the trend of side reactions during the electrolysis process; the temperature sensing node collects the instantaneous reaction temperature of the current reaction zone in the electrolytic cell. The moisture sensing node collects the instantaneous moisture content in the electrolysis system. This reflects the environmental conditions that affect the selectivity of the electrolysis reaction and the purity of the products.

[0019] The data fed back from the above collection nodes at the same timestamp are synchronized and fused to form the latest status frame at the current moment: ,in, Indicates the latest status frame; Indicates instantaneous slot voltage; Indicates instantaneous operating current; Indicates the current current density; Indicates the concentration of the raw material precursor; Indicates the molar percentage of hydrogen; Indicates the molar percentage of low-fluorine byproducts; Indicates the instantaneous reaction temperature; This represents the instantaneous moisture content. By constructing the latest state frame, a multi-dimensional quantitative description of the current electrolysis reaction state is achieved, providing current state input for subsequent control parameter matching in conjunction with real-time reaction stages, as well as for predicting side reaction impurities and product quality trends.

[0020] S1: Extract multi-scale process features from the electrolysis state frame sequence, perform reaction stage clustering identification, and locate the real-time reaction stage.

[0021] Furthermore, multi-scale process features are extracted from the electrolysis state frame sequence, and reaction stage clustering is performed to locate the real-time reaction stage, including: The electrolysis state frame sequence is subjected to full-length single-index slope calculation to obtain global change parameters; the electrolysis state frame sequence is subjected to full-length single-index fluctuation variance calculation to obtain global fluctuation parameters; the global change parameters and global fluctuation parameters are fused to construct global scale trend features; taking the timestamp of the latest state frame as the endpoint, N frames are pushed forward from the electrolysis state frame sequence, and the electrolysis frame subsequence is segmented to calculate single-index slope and single-index fluctuation variance to construct local scale trend features; the global scale trend features and local scale trend features are fused and spliced ​​to obtain the multi-scale process features, and nearest neighbor clustering is performed in the preset reaction stage feature library to locate the real-time reaction stage.

[0022] First, the slope of each single index is calculated across the entire electrolysis state frame sequence to obtain the overall trend of each electrolysis state index throughout the reaction process. Specifically, a time series is established for each state index in the electrolysis state frame sequence, and the slope of index change is calculated using the ratio of the difference between adjacent state frames to the time interval. ,in, This represents the slope of change of the i-th state index; This represents the state index value at time t. The value of the status index corresponding to the previous sampling time is represented by t; t represents the current sampling time. This indicates the previous sampling time. By calculating the slope of change for indicators such as current density, raw material concentration, tail gas component ratio, reaction temperature, and moisture content, global change parameters characterizing the direction of change throughout the entire reaction cycle are obtained.

[0023] Furthermore, the full-length single-index fluctuation variance is calculated on the electrolysis state frame sequence to obtain the stability change characteristics of each state index throughout the reaction process. Specifically, for any state index sequence... Calculate its variance: ,in, The variance of the global fluctuation of the i-th state indicator is represented by ; n represents the number of state frames. This represents the i-th index value in the j-th state frame; This represents the average value of the index sequence. The global fluctuation parameter reflects the magnitude and stability of electrolytic state changes at different reaction stages.

[0024] Subsequently, the global change parameters and global fluctuation parameters are concatenated according to a preset feature fusion rule to construct global scale trend features: Where G represents the global scale trend feature; m represents the number of state indicators involved in the feature calculation.

[0025] Furthermore, using the timestamp corresponding to the latest state frame as the current analysis endpoint, a predetermined number of N state frames are traced back in the electrolysis state frame sequence. A subsequence of electrolysis frames within the current time window is extracted. Single-index slope calculation and single-index fluctuation variance calculation are performed on each state index in the subsequence to obtain local change parameters and local fluctuation parameters reflecting the current short-term reaction change trend. These local change parameters and local fluctuation parameters are then fused to construct local-scale trend features. Where L represents the local scale trend feature; This represents the slope of the i-th indicator change within a local time window; This represents the variance of the i-th indicator fluctuation within the local time window.

[0026] Subsequently, the global-scale trend feature G and the local-scale trend feature L are concatenated to form a multi-scale process feature vector: Where M represents multi-scale process characteristics. By simultaneously integrating the overall change patterns of the reaction process and the current short-term timescale change trends, the multi-scale process characteristics can simultaneously characterize the evolution trend of the electrolysis reaction stages and the current state change characteristics.

[0027] Finally, the multi-scale process features are input into a pre-established reaction stage feature library, which stores standard multi-scale process feature vectors and stage labels corresponding to different historical reaction stages; the distance between the current multi-scale process features and the features of each historical stage is calculated using nearest neighbor clustering. ,in, This represents the distance between the current multi-scale process features and the features of the j-th historical stage; This represents the r-th eigenvalue of the current eigenvector; This indicates that the feature vector of the j-th stage corresponds to the r-th feature value; q represents the number of feature dimensions. The stage label corresponding to the historical stage feature with the smallest distance is selected as the output of the current real-time reaction stage, achieving dynamic positioning of the current electrolysis reaction stage.

[0028] The reaction stage feature library is constructed using historical qualified batch electrolysis operation data. Specifically, it includes: First, collecting complete state frame sequences from multiple batches of electronic-grade perfluoroheptyl ethyl ether electrolytic synthesis processes, and arranging the state frames for each batch in chronological order. For each state frame, current density, cell voltage, reaction temperature, feedstock precursor concentration, hydrogen molar ratio, and low-fluorine by-hydrogen molar ratio are extracted as stage identification state indicators. Furthermore, the rate of change of each state indicator between adjacent state frames is calculated. ,in, This represents the rate of change of the i-th state indicator within the current time window. This represents the detection value of the i-th state indicator at the current time. Indicates the sampling time of the previous time step. Each status indicator detection value, and These represent the corresponding sampling times. The change rates of multiple state indicators are normalized and then fused according to preset weights to obtain the stage change evaluation value: Where E represents the evaluation value of the stage change, Indicates the first Each state indicator corresponds to a weight. When the evaluation value of the stage change within multiple consecutive sampling windows exceeds the stage change threshold... When the current state transitions from the original reaction stage to the next reaction stage, the sampling time at which the continuous determination condition is first met is taken as the stage boundary. The stage change threshold is... This data was obtained based on statistical analysis of the stage change evaluation values ​​during the stable phase of historical qualified batches' operation. Specifically, it was formed by combining the mean and standard deviation of the stable phase evaluation values. ,in, This represents the average evaluation value during the stable phase. represents the standard deviation of the evaluation value in the stable stage, and k represents the empirical correction coefficient. Based on the above stage division results, the multi-scale process features within the corresponding stages in different batches are aggregated, the mean vector of the features corresponding to each stage is calculated as the standard stage feature template, and the feature fluctuation range is calculated as the stage identification tolerance range. These are then stored in association with the stage labels to form the reaction stage feature library.

[0029] S2: Extract the latest state frame from the electrolysis state frame sequence, and perform control parameter matching in conjunction with the real-time reaction stage to obtain the target control parameter set.

[0030] Furthermore, the latest state frame is extracted from the electrolysis state frame sequence, and control parameters are matched in conjunction with the real-time reaction stage to obtain the target control parameter set, including: Based on the real-time response stage, the target parameter interval template is retrieved from the stage matching parameter mapping library, wherein the stage matching parameter mapping library stores multiple sample parameter interval templates for multiple sample response stages; the latest state frame is used as the matching input, and a deviation range matching with a preset precision is performed within the target parameter interval template to capture the optimal target parameter set, thereby constituting the target control parameter set.

[0031] Specifically, historical reaction state data during the electrolytic synthesis of electronic-grade perfluoroheptyl ethyl ether from different batches were collected in advance. Based on the above-mentioned reaction stage division results, parameters for the corresponding reaction stages of each historical batch were statistically analyzed. For any historical reaction stage, valid samples whose electronic-grade product detection results met the preset quality requirements after the corresponding stage were first screened, retaining only the operational data where product purity, impurity content, and target product content all met the requirements. The current density, cell voltage, reaction temperature, raw material precursor concentration, and side reaction inhibition parameters in the valid samples were statistically analyzed, and the mean and standard deviation of each parameter were calculated. , ,in, This represents the average value of the parameters for the corresponding stage. Indicates the standard deviation of the parameter. This represents the parameter value of the i-th valid sample. The target parameter interval is generated based on the mean and standard deviation of these parameters. , ,in, The interval adjustment coefficient is determined based on the historical sample coverage ratio. Target parameter intervals corresponding to different reaction stages are associated and stored to form a stage-matching parameter mapping library, indexed by reaction stage tags and containing target parameter interval templates. Specifically, the stage-matching parameter mapping library uses reaction stage identifiers as index items and stores sample parameter interval templates as associated data items. Each sample parameter interval template includes a target current density interval, a target cell voltage interval, a target reaction temperature interval, a target feedstock concentration interval, and a side reaction suppression parameter interval.

[0032] After step S1 outputs the real-time reaction stage, an index query is performed in the stage matching parameter mapping library according to the stage identifier corresponding to the real-time reaction stage to obtain the target parameter range template corresponding to the current reaction stage: ,in, Represents the template for the target parameter range; , These represent the lower and upper limits of the target current density, respectively. , These represent the lower and upper limits of the target slot voltage, respectively. , These represent the lower and upper limits of the target reaction temperature, respectively. , These represent the lower and upper limits of the target raw material concentration, respectively.

[0033] Furthermore, using the latest state frame as matching input, the deviation range of each state parameter in the latest state frame is matched with the target parameter interval template. Specifically, for each state parameter in the latest state frame, it is determined whether it is within the corresponding target parameter interval; when multiple candidate parameter combinations all satisfy the target parameter interval constraints, the comprehensive deviation value between each candidate parameter combination and the current state parameter is calculated: ,in, This represents the overall deviation value of the k-th candidate parameter combination; This represents the i-th state parameter in the latest state frame; This represents the corresponding control parameter in the k-th candidate parameter combination; represents the weight coefficient corresponding to the i-th state parameter; m represents the number of state parameters involved in the matching.

[0034] Based on the comprehensive deviation value, the parameter combination with the smallest deviation value is selected from the candidate parameter combinations that satisfy the template constraint of the target parameter range, and this combination is taken as the optimal target parameter set. ,in, This represents the optimal target parameter set. Subsequently, this optimal target parameter set is output as the target control parameter set corresponding to the current electrolysis reaction stage: ,in, Represents the target control parameter set; This indicates the target current density adjustment parameter; Indicates the target tank voltage adjustment parameters; Indicates the target temperature control parameter; This indicates the target raw material concentration adjustment parameter; This represents the adjustment parameters related to the suppression of side reactions. Through the above matching process, the target control parameter set can be dynamically adjusted according to changes in the real-time reaction stage and the current electrolysis state, providing a basis for subsequent prediction of side reaction impurities and closed-loop control of the electrolysis process.

[0035] S3: Based on the historical state trend characteristics, the latest state frame, the real-time reaction stage, and the target control parameter set, side reaction impurities are predicted to obtain the baseline for allowable parameter fluctuations.

[0036] Furthermore, based on historical state trend characteristics, the latest state frame, the real-time reaction stage, and the target control parameter set, side reaction impurities are predicted to obtain a baseline for allowable parameter fluctuations, including: Based on the real-time reaction stage, the side reaction correlation index is retrieved; based on the side reaction correlation index, the historical state trend feature is extracted from the global scale trend feature; based on the latest state frame, the target control parameter set, and the historical state trend feature, the impurity increment is predicted to obtain the predicted total impurity content; an upper limit for allowable impurities in electronic-grade products is introduced, and the predicted total impurity content is deduced by reverse deviation to construct the parameter allowable fluctuation baseline.

[0037] Specifically, based on the real-time reaction stage determined in step S1, stage index matching is performed in a preset side reaction correlation index library. This library stores a set of key state indicators affecting the generation of side reaction impurities corresponding to different reaction stages. Since the formation mechanisms and sensitive parameters of side reactions differ at different reaction stages during the electrolytic synthesis of electronic-grade perfluoroheptyl ethyl ether, corresponding side reaction correlation indicators are selected for the current real-time reaction stage. These indicators include at least changes in current density, cell voltage, feedstock precursor concentration, reaction temperature, hydrogen molar ratio, and low-fluorine by-hydrogen molar ratio. Subsequently, historical state trend features are extracted from the global-scale trend features based on the side reaction correlation indicators. Specifically, using the side reaction correlation indicators as feature indexes, the global-scale trend features generated in step S1 are filtered, retaining only historical change features related to the generation of side reactions in the current real-time reaction stage, resulting in historical state trend features: Where H represents the historical trend characteristics; This represents the historical trend characteristics corresponding to the m-th side reaction correlation index.

[0038] Furthermore, the difference between the current state parameters in the latest state frame and the target adjustment parameters in the target control parameter set is calculated to obtain the current state deviation characteristics: ,in, This represents the current deviation corresponding to the i-th state indicator; This represents the current state parameters in the latest state frame; This represents the target control parameter corresponding to the target control parameter set. Simultaneously, the changing trend of the side reaction index is calculated based on the historical state trend characteristics: ,in, This represents the historical rate of change of the i-th side reaction-related index; This represents the historical trend characteristic value corresponding to the current moment; This represents the historical trend characteristic value corresponding to the previous time window; This represents a time interval. The current deviation characteristics, historical change rates, and stage-related weights corresponding to the real-time response stage are fused to calculate the predicted impurity increment: ,in, This indicates a predicted increase in impurities; This represents the stage influence weight corresponding to the i-th side reaction correlation index; This represents the correction factor for historical trend changes.

[0039] Furthermore, the currently detected cumulative impurity content is added to the predicted impurity increment to obtain the predicted total impurity content: ,in, This indicates the predicted total impurity content; This indicates the current cumulative impurity content; This indicates a predicted increase in impurities.

[0040] Finally, an upper limit for permissible impurities in electronic-grade products is introduced, and the predicted total impurity content is derived by reverse deviation to construct the baseline for permissible parameter fluctuations. Specifically, this is based on the upper limit for permissible impurities corresponding to electronic-grade perfluoroheptyl ether. Calculate the impurity margin under the current predicted state: ,in, Indicates the remaining impurity tolerance; Indicates the upper limit of permissible impurities in electronic-grade products; This indicates the predicted total impurity content. Based on the remaining impurity tolerance and the sensitivity of each control parameter to side reaction formation, the allowable offset range of each control parameter is calculated in reverse: ,in, This represents the allowable fluctuation range corresponding to the i-th control parameter; This represents the sensitivity coefficient of the i-th control parameter to impurity generation. The sensitivity coefficient is obtained through offline calibration using historical experimental perturbation data. Specifically, while keeping other control parameters constant in the current reaction stage, the sensitivity coefficient is adjusted for a single control parameter. Multiple levels of perturbation were applied, including positive and negative perturbations, and the changes in the content of side reaction impurities before and after the perturbations were collected. By performing linear fitting based on the changes in impurities corresponding to different disturbance amplitudes, the slope of the influence of the i-th control parameter on impurity generation is obtained: ,in, Let represent the sensitivity coefficient of the i-th control parameter. The sensitivity coefficients obtained from multiple disturbance experiments are normalized to allow for comparison of the degree of influence between different control parameters, ultimately yielding a set of sensitivity coefficients in the range of 0 to 1.

[0041] By combining the allowable fluctuation ranges corresponding to each control parameter, the baseline for allowable parameter fluctuations is obtained: Where B represents the baseline for parameter fluctuation; This represents the allowable fluctuation range corresponding to the nth control parameter.

[0042] Through the above process, side reaction impurities can be predicted based on real-time reaction stage, historical change trend and current control state. The prediction results are then converted into parameter allowable fluctuation constraints, so that the subsequent target control parameter correction process can be dynamically adjusted within the range of ensuring the quality requirements of electronic-grade products.

[0043] Furthermore, based on the latest state frame, the target control parameter set, and historical state trend characteristics, impurity increment prediction is performed to obtain the predicted total impurity content, including: The latest state frame and the target control parameter set are mapped and compared to locate the intersection index and calculate the multi-source transient deviation. The intersection index deviation is calculated for the target control parameter set and historical state trend characteristics to obtain the historical deviation change rate. After calling the real-time stage correlation function according to the real-time response stage, the increment of the multi-source transient deviation and the historical deviation change rate is substituted to obtain the predicted impurity increment. The historical accumulated impurity mass and the predicted impurity increment are summed to output the predicted total impurity content.

[0044] Specifically, the set of state parameters in the latest state frame is mapped to the set of control parameters in the target control parameter set. Intersection indices with direct correlation are determined based on parameter names, parameter types, and corresponding physical interactions. These intersection indices include parameters that can simultaneously characterize the current reaction state and the target control state, such as current density, cell voltage, reaction temperature, raw material precursor concentration, and the proportion of side reaction gas components.

[0045] For each intersection index, the difference between the real-time detection value in the latest state frame and the target set value in the target control parameter set is calculated to obtain the corresponding multi-source transient deviation: ,in, This represents the multi-source transient deviation corresponding to the i-th intersection index; This represents the current detection value of the i-th indicator in the latest state frame; This represents the target parameter value corresponding to the i-th indicator in the target control parameter set.

[0046] Further, the intersection index deviation of the target control parameter set and the historical state trend features is calculated to obtain the historical deviation change rate. Specifically, based on the intersection index, the index change data corresponding to the historical time moment is extracted from the historical state trend features, and the actual change trajectory in the historical state trend is compared with the target change trend corresponding to the target control parameters to calculate the historical deviation change rate. ,in, This represents the historical deviation change rate corresponding to the i-th intersection index; The indicator value represents the current historical trend moment; This indicates the indicator value corresponding to the previous historical time window; This indicates the target value corresponding to the target control parameter; This indicates the time interval for sampling historical states.

[0047] Furthermore, after retrieving the real-time stage correlation function based on the real-time reaction stage, the multi-source transient deviation and historical deviation change rate are substituted into the real-time stage correlation function for increment calculation to obtain the predicted impurity increment. Specifically, different real-time reaction stages correspond to different side reaction generation characteristics; therefore, real-time stage correlation functions corresponding to each reaction stage are pre-established. ,in, This indicates a predicted increase in impurities; This represents the stage association function corresponding to the current real-time response stage; This represents the multi-source transient deviation corresponding to the i-th intersection index; represents the historical deviation change rate corresponding to the i-th intersection index; m represents the number of intersection indexes.

[0048] In the specific calculation, the corresponding stage association function is called according to the current real-time response stage, and the influence of different indicators is adjusted by stage association weights: ,in, This represents the weight of the impurity influence of the i-th indicator at the current reaction stage; This represents the historical trend correction coefficient. The stage-related weights are also mentioned. Historical trend correction coefficient It is obtained through training and calibration using historical operational samples. Specifically, for each reaction stage, the deviation of each intersection index during the historical operation is obtained. Historical deviation rate of change and the actual increase in impurities within the corresponding time window Establish a sample of the influence of indicators: Based on multiple historical samples, the correlation between the deviation of each indicator and the increase in impurities is calculated, and the correlation is normalized to obtain the initial stage correlation weight. Further, the predicted increase in impurities is then... Compared with actual impurity increment Using the error between them as the optimization objective, the final stage correlation weights are obtained by adjusting the weights of each indicator to minimize the prediction error. .in, satisfy: The correction coefficient The value is dynamically determined based on historical prediction errors; it increases when historical prediction errors decrease. When historical prediction errors increase, reduce and limited The range of values ​​is .

[0049] Finally, by summing the historical cumulative impurity mass and the predicted impurity increment, the predicted total impurity content is output: ,in, This indicates the predicted total impurity content; This indicates the current historical cumulative impurity mass; This indicates a predicted increase in impurities.

[0050] Through the above process, the transient effects of the current electrolysis state deviating from the target control state and the changes in historical operating trends are jointly introduced into the impurity generation prediction process, so that the predicted total impurity content can reflect the impact of the current control parameter changes on the evolution of side reactions, and provide data basis for subsequent calculation of the baseline of allowable parameter fluctuations.

[0051] like Figure 2The figure shows a comparison of impurity content changes before and after parameter fluctuation control. The horizontal axis represents the electrolysis reaction time, and the vertical axis represents the changes in key parameter deviations and side reaction impurity content, respectively. Without parameter fluctuation control, key process parameters such as current density, cell voltage, reaction temperature, and raw material concentration are affected by changes in the reaction state during electrolysis, resulting in significant fluctuations in parameter deviations. Furthermore, as the reaction time increases, side reactions accumulate, leading to a gradual increase in impurity content. After using the method described in this application to generate an allowable parameter fluctuation baseline and perform deviation correction, by limiting the range of key parameter deviations, the process parameters during electrolysis are kept within the allowable fluctuation range. This suppresses the occurrence of side reactions and significantly reduces the growth trend of impurity content.

[0052] S4: Based on the latest state frame and the target control parameter set, predict the product quality trend to obtain the qualified product range.

[0053] Furthermore, based on the latest state frame and the target control parameter set, product quality trend prediction is performed to obtain the product qualification range, including: Based on the stage-specific empirical correlation between the main product quality and control parameters in the real-time reaction phase, the instantaneous parameters of the latest state frame and the target control parameter set are input into the empirical correlation to perform time-series prediction of the main product quality and output the quality prediction fluctuation range; an electronic-grade product qualification standard is introduced, and the product qualification interval is intercepted within the quality prediction fluctuation range.

[0054] Specifically, based on the real-time reaction stage determined in S1, stage index matching is performed in a preset quality correlation database. This database stores the correspondence between the main product quality changes and electrolysis control parameters at different reaction stages. These control parameters include current density, cell voltage, reaction temperature, feed concentration, and side reaction suppression parameters. The main product quality parameters include the purity of electronic-grade perfluoroheptyl ether, the content of the target product, and key quality indicators. Furthermore, corresponding stage-specific empirical correlations are established for different reaction stages. ,in, This indicates the correlation of the main product quality corresponding to the s-th reaction stage; Indicates current density; Indicates the slot voltage; Indicates the reaction temperature; Indicates the concentration of the raw material precursor; This indicates parameters related to side reaction control; This represents the mass correlation function corresponding to the current reaction stage.

[0055] Furthermore, the instantaneous parameters and target control parameter set in the latest state frame are input into the empirical correlation to perform time-series prediction of the main product quality, and the predicted fluctuation range of quality is output. Specifically, the current state parameters in the latest state frame are used as real-time input, and the target control parameter set is used as future control constraints. The trend of main product quality change over time under the current control strategy is predicted through the quality correlation corresponding to the real-time reaction stage. ,in, This represents the predicted quality of the main product at the predicted time. Indicates the latest status frame; Represents the target control parameter set; This represents the mass correlation function corresponding to the current reaction stage.

[0056] Furthermore, based on multiple predicted quality values ​​within a continuous prediction time window, the range of quality prediction fluctuations is calculated: ,in, Indicates the range of predicted quality fluctuations; This indicates the minimum predicted quality of the main product within the prediction window. This represents the maximum predicted quality of the main product within the prediction window. Specifically, by statistically analyzing the quality prediction sequence within a preset future time window, the predicted quality change trend and fluctuation range are obtained, which are used to determine the product quality stability corresponding to the current target control parameter set.

[0057] Finally, an electronic-grade product qualification standard is introduced, and the qualified range of the product is extracted from the predicted quality fluctuation range. Specifically, the quality qualification constraint range corresponding to electronic-grade perfluoroheptyl ether is obtained as follows: ,in, Indicates the quality standard range for electronic-grade products; Indicates the lower limit of quality requirements; This indicates the upper limit requirement for quality. The predicted range of quality fluctuations... With respect to the quality standard range of the electronic-grade products Perform intersection calculation: ,in, This indicates the acceptable range for the product.

[0058] When the predicted fluctuation range of quality intersects with the quality standard range of electronic-grade products, the intersecting region is taken as the qualified range of products under the current control conditions; when there is no intersection, it is determined that the current target control parameter set cannot meet the quality requirements of electronic-grade products, and the result is fed back to the subsequent parameter adjustment process to correct the target control parameter set.

[0059] Through the above process, the product quality trend can be predicted based on the current electrolysis state and target control parameters, and the quality variation range that meets the electronic grade product standards can be determined in advance, providing quality constraints for subsequent dynamic closed-loop control.

[0060] S5: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, the parameter deviation is corrected according to the parameter allowable fluctuation baseline until the parameter deviation meets the parameter allowable fluctuation baseline and the product quality trend falls within the product qualified range. Then, the iterative steps S1 to S5 are triggered to perform dynamic closed-loop control of the entire process until the preset reaction endpoint determination condition is met, and the electrolysis power supply and process control are stopped.

[0061] The target control parameter set obtained in step S2 is sent to the electrolytic cell control execution unit, and the electrolysis process is controlled by adjusting the operating parameters of the electrochemical unit of the electrolytic cell material. The electrochemical unit of the material includes an electrode power supply unit, a material circulation adjustment unit, and a reaction environment adjustment unit. The electrode power supply unit adjusts the output current and voltage of the electrolytic power supply according to the target current density and target cell voltage in the target control parameter set. The material circulation adjustment unit adjusts the raw material supply amount and circulation flow rate according to the target raw material concentration parameter. The reaction environment adjustment unit adjusts the reaction temperature and system moisture state according to the target temperature parameter and moisture control parameter, so that the electrolysis reaction state changes towards the target control state.

[0062] After the target control parameter set is issued, the latest state frame in the electrolysis state frame sequence is re-acquired. The real-time operating parameters in the latest state frame are compared with the target control parameter set, and the current parameter deviation is calculated. ,in, This represents the parameter deviation corresponding to the i-th control parameter; This represents the real-time detected value of the corresponding parameter in the latest status frame; This represents the target parameter value corresponding to the target control parameter set.

[0063] Furthermore, the parameter deviation is compared with the parameter allowable fluctuation baseline generated in step S3: ,in, This represents the allowable fluctuation range corresponding to the i-th control parameter.

[0064] When the parameter deviation meets the baseline constraint for the allowable fluctuation of the parameter, it is determined that the current process parameter adjustment result meets the stability control requirements; when any parameter deviation exceeds the corresponding allowable fluctuation range, the target control parameter set is corrected according to the deviation direction and deviation amplitude, and then applied to the electrochemical unit of the electrolytic cell material.

[0065] Specifically, the parameter correction is calculated based on the difference between the parameter deviation and the allowable fluctuation baseline: ,in, This represents the correction amount for the i-th control parameter; This represents the adjustment coefficient corresponding to the i-th control parameter; Indicates the current parameter deviation; This indicates the allowable fluctuation range. The target control parameter set is updated based on the parameter correction amount. ,in, This indicates the updated target control parameters; Indicates the original target control parameter; This indicates the parameter correction amount.

[0066] Subsequently, the parameter deviation detection and correction process is continuously executed until all control parameter deviations meet the baseline constraint for allowable parameter fluctuations. Simultaneously, the product quality trend prediction result output from step S4 is obtained, and the current quality prediction range is matched with the product qualification range. ,in, Indicates the predicted range of quality fluctuations; This indicates the acceptable range for the product.

[0067] When the parameter deviation meets the baseline of the parameter allowable fluctuation and the product quality trend falls into the qualified product range, it is determined that the current control state meets the closed-loop optimization conditions, and the iterative steps S1 to S5 are triggered to re-execute the electrolysis state acquisition, reaction stage identification, parameter matching, side reaction prediction and quality trend prediction, so as to realize continuous dynamic closed-loop control of the entire electrolytic synthesis process.

[0068] During the cyclic control process, the electrolysis state is monitored based on preset reaction endpoint determination conditions. These preset conditions include the target product quality meeting preset requirements, the target product quality trend remaining stable, the remaining raw material concentration being below a preset threshold, and the reaction time reaching a set range. When all endpoint determination conditions are simultaneously met, a reaction termination command is output, controlling the electrolysis power supply unit to stop outputting power, and stopping material adjustment and process control, thus completing the electrolytic synthesis process of electronic-grade perfluoroheptyl ethyl ether.

[0069] Furthermore, after adjusting the electrochemical unit of the electrolyzer material using the target control parameter set, the parameter deviation is corrected according to the parameter allowable fluctuation baseline until the parameter deviation meets the parameter allowable fluctuation baseline and the product quality trend falls within the product qualified range. Then, iterative steps S1 to S5 are triggered to perform full-process dynamic closed-loop control until the preset reaction endpoint determination condition is met, at which point the electrolysis power supply and process control are stopped, including: a1: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, trigger ultra-short-term electrolysis state tracking to capture incremental state frame sequences; a2: Extract the final state frame from the incremental state frame sequence, match and compare it with the target control parameter set, and locate the intersection index deviation set; a3: Compare the intersection index deviation set with the parameter allowable fluctuation baseline to obtain the over-deviation index set; a4: Based on the over-deviation index set, after the target control parameter set is corrected to the parameter allowable fluctuation baseline in the material electrochemical unit, cyclically trigger steps a1 to a4 to perform near-closed-loop deviation correction; a5: After the product quality trend falls within the product qualified range, trigger steps a1 to a5 to execute full-process dynamic closed-loop control until the preset reaction endpoint determination condition is met, and then stop electrolysis power supply and process control.

[0070] Specifically, after the target control parameter set is sent to the electrochemical unit of the electrolytic cell material, the multi-source online sensing nodes are kept in continuous acquisition mode, and the current parameter adjustment completion time is used as the new state tracking starting point. The electrolytic state change data after parameter adjustment is collected according to the ultra-short-term sliding acquisition window and the preset sampling frequency. The continuously acquired state data is arranged in chronological order to form an incremental state frame sequence: ,in, This represents an incremental state frame sequence; Indicates the first parameter after adjustment One incremental state frame.

[0071] Further, the final state frame is extracted from the incremental state frame sequence, and matched against the target control parameter set to locate the intersection index deviation set. Specifically, the latest final state frame in the incremental state frame sequence is selected. The set of state parameters in the terminal state frame is mapped and matched with the set of control parameters in the target control parameter set to determine the intersection index that exists simultaneously in the state detection parameters and the control target parameters.

[0072] For each intersection index, calculate the parameter deviation between the current actual state value and the target control value: ,in, This represents the parameter deviation corresponding to the i-th intersection index; This represents the actual detected value of the i-th indicator in the end-state frame; This represents the target parameter value corresponding to the target control parameter set.

[0073] The parameter deviations corresponding to each intersection index are aggregated to obtain the intersection index deviation set: , where E represents the intersection index deviation set; This represents the deviation corresponding to the nth intersection index.

[0074] Further, by comparing the intersection index deviation set with the parameter allowable fluctuation baseline, a set of deviation indicators is obtained. Specifically, the absolute value of the deviation of each intersection indicator is compared with the parameter allowable fluctuation baseline generated in step S3: ,in, This represents the allowable fluctuation range corresponding to the i-th indicator.

[0075] When a certain intersection indicator meets the above conditions, the indicator is determined to be an out-of-deviation indicator, and all indicators that exceed the allowable fluctuation range are summarized to form an out-of-deviation indicator set: ,in, This represents the set of deviation indicators; Indicates the first One deviation indicator.

[0076] Further, after the target control parameter set is corrected to the allowable fluctuation baseline in the material electrochemical unit based on the set of out-of-deviation indicators, steps a1 to a4 are triggered cyclically to perform near-closed-loop deviation correction. Specifically, the parameter correction direction is determined according to the deviation direction corresponding to the out-of-deviation indicator, and the corresponding control parameter correction amount is calculated based on the deviation amplitude. ,in, This represents the correction amount for the i-th control parameter; This represents the adjustment coefficient of the corresponding control parameter; This indicates the deviation of the corresponding indicator.

[0077] Update the target control parameter set based on the control parameter correction amount: ,in, This indicates the corrected target control parameter; This represents the target control parameters before correction. Subsequently, the corrected target control parameters are applied again to the electrochemical unit of the electrolyzer material, and the incremental state frame acquisition, deviation location, and over-deviation judgment processes are re-executed until all intersection index deviations meet the following requirements: This means that the current electrolysis operation is within the baseline range of allowable parameter fluctuations, achieving near-closed-loop deviation correction of the target control parameters.

[0078] Furthermore, after the product quality trend falls within the product qualification range, steps a1 to a5 are triggered to execute full-process dynamic closed-loop control. Specifically, when the product quality prediction result output in step S4 satisfies: ,in, Indicates the predicted fluctuation range of product quality; This indicates the acceptable range for the product.

[0079] If the current control parameters are determined to meet the quality requirements of electronic-grade products, the system will enter the full-process dynamic closed-loop control stage, and will cyclically execute electrolysis state tracking, reaction stage identification, target parameter matching, side reaction prediction, quality trend prediction, and parameter adjustment.

[0080] During the cyclic control process, the reaction endpoint status is continuously monitored. When the preset reaction endpoint determination conditions are met, a stop control command is output to shut down the electrolysis power supply unit and stop the material conditioning process, thus completing the control of the electrolytic synthesis process of electronic-grade perfluoroheptyl ether.

[0081] In summary, the embodiments of this application have at least the following technical effects: After the electrolytic synthesis reaction is initiated, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain an electrolytic state frame sequence. Next, S1: Multi-scale process features are extracted from the electrolytic state frame sequence, and reaction stage clustering is performed to locate the real-time reaction stage. Simultaneously, S2: The latest state frame is extracted from the electrolytic state frame sequence and combined with the real-time reaction stage for control parameter matching to obtain a target control parameter set. Then, S3: Based on historical state trend characteristics, the latest state frame, the real-time reaction stage, and the target control parameter set, side reaction impurities are predicted to obtain a baseline for allowable parameter fluctuations. S4: Based on the latest state frame and the target control parameter set, product quality trend is predicted to obtain a product qualification range. Finally, S5: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, parameter deviations are corrected based on the baseline for allowable parameter fluctuations until the parameter deviations meet the baseline for allowable parameter fluctuations and the product quality trend falls within the product qualification range. This triggers iterative steps S1 to S5, performing dynamic closed-loop control throughout the entire process until the preset reaction endpoint determination conditions are met, at which point electrolytic power supply and process control are stopped. This invention solves the technical problems of lagging electrolysis process control in existing technologies, which leads to increased side reactions and unstable product quality. It achieves dynamic control of the electrolysis process and improves the stability of product quality.

[0082] Example 2, based on the same inventive concept as the electrolysis process control method for electronic-grade perfluoroheptyl ethyl ether in the foregoing examples, such as... Figure 3 As shown, this application provides an electrolysis process control system for electronic-grade perfluoroheptyl ether, wherein the system includes: The state acquisition module 11 is used to perform ultra-short-term electrolysis state tracking through multi-source online sensing nodes after the electrolytic synthesis reaction is started, and obtain an electrolysis state frame sequence; the stage identification module 12 is used to perform S1, extract multi-scale process features from the electrolysis state frame sequence, perform reaction stage clustering identification, and locate the real-time reaction stage; the parameter matching module 13 is used to perform S2, extract the latest state frame from the electrolysis state frame sequence, and perform control parameter matching in combination with the real-time reaction stage to obtain the target control parameter set; the impurity prediction module 14 is used to perform S3, based on historical state trend features, the latest state frame, the real-time reaction stage, and the target control parameter set, to... The side reaction impurities are predicted to obtain the baseline for allowable parameter fluctuations. The quality prediction module 15 is used to execute S4, which predicts the product quality trend based on the latest state frame and the target control parameter set to obtain the product qualified range. The closed-loop control module 16 is used to execute S5, which adjusts the electrochemical unit of the electrolyzer material using the target control parameter set, and corrects the parameter deviation based on the baseline for allowable parameter fluctuations until the parameter deviation meets the baseline for allowable parameter fluctuations and the product quality trend falls within the product qualified range. Then, the iterative steps S1 to S5 are triggered to perform dynamic closed-loop control of the entire process until the preset reaction endpoint determination condition is met, and the electrolysis power supply and process control are stopped.

[0083] Furthermore, the closed-loop control module 16 is used to execute the following method: a1: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, trigger ultra-short-term electrolysis state tracking to capture incremental state frame sequences; a2: Extract the final state frame from the incremental state frame sequence, match and compare it with the target control parameter set, and locate the intersection index deviation set; a3: Compare the intersection index deviation set with the parameter allowable fluctuation baseline to obtain the over-deviation index set; a4: Based on the over-deviation index set, after the target control parameter set is corrected to the parameter allowable fluctuation baseline in the material electrochemical unit, cyclically trigger steps a1 to a4 to perform near-closed-loop deviation correction; a5: After the product quality trend falls within the product qualified range, trigger steps a1 to a5 to execute full-process dynamic closed-loop control until the preset reaction endpoint determination condition is met, and then stop electrolysis power supply and process control.

[0084] Furthermore, the stage identification module 12 is used to perform the following method: The electrolysis state frame sequence is subjected to full-length single-index slope calculation to obtain global change parameters; the electrolysis state frame sequence is subjected to full-length single-index fluctuation variance calculation to obtain global fluctuation parameters; the global change parameters and global fluctuation parameters are fused to construct global scale trend features; taking the timestamp of the latest state frame as the endpoint, N frames are pushed forward from the electrolysis state frame sequence, and the electrolysis frame subsequence is segmented to calculate single-index slope and single-index fluctuation variance to construct local scale trend features; the global scale trend features and local scale trend features are fused and spliced ​​to obtain the multi-scale process features, and nearest neighbor clustering is performed in the preset reaction stage feature library to locate the real-time reaction stage.

[0085] Furthermore, the parameter matching module 13 is used to perform the following method: Based on the real-time response stage, the target parameter interval template is retrieved from the stage matching parameter mapping library, wherein the stage matching parameter mapping library stores multiple sample parameter interval templates for multiple sample response stages; the latest state frame is used as the matching input, and a deviation range matching with a preset precision is performed within the target parameter interval template to capture the optimal target parameter set, thereby constituting the target control parameter set.

[0086] Furthermore, the impurity prediction module 14 is used to perform the following method: Based on the real-time reaction stage, the side reaction correlation index is retrieved; based on the side reaction correlation index, the historical state trend feature is extracted from the global scale trend feature; based on the latest state frame, the target control parameter set, and the historical state trend feature, the impurity increment is predicted to obtain the predicted total impurity content; an upper limit for allowable impurities in electronic-grade products is introduced, and the predicted total impurity content is deduced by reverse deviation to construct the parameter allowable fluctuation baseline.

[0087] The latest state frame and the target control parameter set are mapped and compared to locate the intersection index and calculate the multi-source transient deviation. The intersection index deviation is calculated for the target control parameter set and historical state trend characteristics to obtain the historical deviation change rate. After calling the real-time stage correlation function according to the real-time response stage, the increment of the multi-source transient deviation and the historical deviation change rate is substituted to obtain the predicted impurity increment. The historical accumulated impurity mass and the predicted impurity increment are summed to output the predicted total impurity content.

[0088] Furthermore, the quality prediction module 15 is used to perform the following method: Based on the stage-specific empirical correlation between the main product quality and control parameters in the real-time reaction phase, the instantaneous parameters of the latest state frame and the target control parameter set are input into the empirical correlation to perform time-series prediction of the main product quality and output the quality prediction fluctuation range; an electronic-grade product qualification standard is introduced, and the product qualification interval is intercepted within the quality prediction fluctuation range.

[0089] Furthermore, the status acquisition module 11 is used to perform the following method: The multi-source online sensing nodes are pre-deployed in the electrolytic cell. These nodes include voltage and current acquisition nodes located at the electrode leads, raw material concentration sensing nodes located in the feed circuit, online component detection nodes located at the tail gas outlet of the electrolytic cell, and temperature and moisture sensing nodes located in the inner cavity of the cell. Using 1 / K of the total electrolysis reaction time as an ultra-short-term sliding acquisition window, the multi-source online sensing nodes are synchronously controlled to perform data acquisition on the electrolytic cell at a preset sampling frequency, and output the electrolysis status frame sequence.

[0090] The latest status frame includes the instantaneous tank voltage, instantaneous operating current, and current density collected and fed back by the voltage and current acquisition node; the raw material precursor concentration collected and fed back by the raw material concentration sensing node; the molar ratio of hydrogen and the molar ratio of low-fluorine by-products collected and fed back by the online component detection node; the instantaneous reaction temperature collected and fed back by the temperature sensing node; and the instantaneous moisture content collected and fed back by the moisture sensing node.

[0091] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether, characterized in that, The method includes: After the electrolytic synthesis reaction is started, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain the electrolytic state frame sequence; S1: Extract multi-scale process features from the electrolysis state frame sequence, perform reaction stage clustering identification, and locate the real-time reaction stage; S2: Extract the latest state frame from the electrolysis state frame sequence, and perform control parameter matching in combination with the real-time reaction stage to obtain the target control parameter set; S3: Based on the historical state trend characteristics, the latest state frame, the real-time reaction stage, and the target control parameter set, predict the side reaction impurities to obtain the baseline for allowable parameter fluctuations; S4: Based on the latest status frame and the target control parameter set, predict the product quality trend to obtain the qualified product range; S5: After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, the parameter deviation is corrected according to the parameter allowable fluctuation baseline until the parameter deviation meets the parameter allowable fluctuation baseline and the product quality trend falls within the product qualified range. Then, the iterative steps S1 to S5 are triggered to perform dynamic closed-loop control of the entire process until the preset reaction endpoint determination condition is met, and the electrolysis power supply and process control are stopped.

2. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 1, characterized in that, After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, the parameter deviation is corrected according to the parameter allowable fluctuation baseline until the parameter deviation meets the parameter allowable fluctuation baseline and the product quality trend falls within the product qualified range. Then, iterative steps S1 to S5 are triggered to perform full-process dynamic closed-loop control until the preset reaction endpoint determination condition is met, at which point the electrolysis power supply and process control are stopped. The method includes: a1: After adjusting the electrochemical unit of the electrolyzer material using the target control parameter set, ultra-short-term electrolysis state tracking is triggered to capture incremental state frame sequences; a2: Extract the final state frame from the incremental state frame sequence, match and compare it with the target control parameter set, and locate the intersection index deviation set; a3: By comparing the intersection index deviation set with the parameter allowable fluctuation baseline, the set of out-of-deviation indices is obtained; a4: After the target control parameter set is corrected to the parameter allowable fluctuation baseline in the material electrochemical unit according to the set of out-of-deviation indices, steps a1 to a4 are triggered cyclically to perform near-closed-loop deviation correction. a5: After the product quality trend falls within the product qualified range, steps a1 to a5 are triggered to execute full-process dynamic closed-loop control until the preset reaction endpoint determination condition is met, at which point the electrolysis power supply and process control are stopped.

3. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 1, characterized in that, The method includes extracting multi-scale process features from the electrolysis state frame sequence, performing reaction stage clustering identification, and locating the real-time reaction stage. The global variation parameters are obtained by performing a full-length single-index slope calculation on the electrolysis state frame sequence. The global fluctuation parameters are obtained by calculating the full-length single-index fluctuation variance of the electrolysis state frame sequence. By integrating the global change parameters and global fluctuation parameters, a global-scale trend feature is constructed. Using the timestamp of the latest state frame as the endpoint, push forward N frames from the electrolysis state frame sequence, divide the electrolysis frame subsequence, perform single index slope calculation and single index fluctuation variance calculation, and construct local scale trend features; By fusing and splicing the global-scale trend features and local-scale trend features, the multi-scale process features are obtained. Nearest neighbor clustering is then performed on the preset reaction stage feature library to locate the real-time reaction stage.

4. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 3, characterized in that, The method involves extracting the latest state frame from the electrolysis state frame sequence, matching the control parameters with the real-time reaction stage, and obtaining the target control parameter set. Based on the real-time response stage, the target parameter range template is retrieved from the stage matching parameter mapping library, wherein the stage matching parameter mapping library stores multiple sample parameter range templates for multiple sample response stages. Using the latest state frame as the matching input, a deviation range matching with a preset precision is performed within the target parameter interval template to capture the optimal target parameter set, thus forming the target control parameter set.

5. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 3, characterized in that, Based on historical state trend characteristics, the latest state frame, real-time reaction stage, and target control parameter set, side reaction impurities are predicted to obtain a baseline for allowable parameter fluctuations. The method includes: Retrieve the related indicators of the side reactions based on the real-time reaction stage; Based on the side reaction correlation index, the historical state trend features are extracted from the global-scale trend features; Based on the latest state frame, the target control parameter set, and the historical state trend characteristics, the impurity increment is predicted to obtain the predicted total impurity content. An upper limit for permissible impurities in electronic-grade products is introduced, and the predicted total impurity content is derived by reverse deviation to construct the baseline for permissible fluctuation of the parameters.

6. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 5, characterized in that, Based on the latest state frame, the target control parameter set, and historical state trend characteristics, the method for predicting the total impurity content is as follows: The latest state frame and the target control parameter set are mapped and compared, and the intersection index is used to calculate the multi-source transient deviation. The intersection index deviation is calculated for the target control parameter set and historical state trend characteristics to obtain the historical deviation change rate; Based on the real-time reaction stage, the correlation function of the real-time stage is retrieved, and the increments of the multi-source transient deviation and the historical deviation change rate are substituted into the calculation to obtain the predicted impurity increment. The predicted total impurity content is output by summing the historical cumulative impurity mass and the predicted impurity increment.

7. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 1, characterized in that, Based on the latest state frame and the target control parameter set, product quality trend prediction is performed to obtain the product qualification range. The method includes: Based on the stage-specific empirical correlation between the main product quality and control parameters retrieved during the real-time reaction phase; The instantaneous parameters of the latest state frame and the target control parameter set are input into the empirical correlation to perform time-series prediction of the main product quality and output the predicted fluctuation range of the quality. An electronic-grade product qualification standard is introduced, and the qualified range of the product is selected from the predicted fluctuation range of the quality.

8. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 1, characterized in that, After initiating the electrolytic synthesis reaction, ultra-short-term electrolytic state tracking is performed through multi-source online sensing nodes to obtain an electrolytic state frame sequence. The method includes: The multi-source online sensing nodes are pre-deployed in the electrolytic cell, wherein the multi-source online sensing nodes include voltage and current acquisition nodes arranged at the electrode lead-out end, raw material concentration sensing nodes arranged in the feed circuit, online component detection nodes arranged at the tail gas outlet of the electrolytic cell, and temperature sensing nodes and moisture sensing nodes arranged in the inner cavity of the cell. Using 1 / K of the total electrolysis reaction time as an ultra-short-term sliding acquisition window, the multi-source online sensing nodes are synchronously controlled to perform data acquisition on the electrolytic cell at a preset sampling frequency, and the electrolysis state frame sequence is output.

9. The method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether as described in claim 8, characterized in that, The latest status frame includes the instantaneous tank voltage, instantaneous operating current, and current density collected and fed back by the voltage and current acquisition node; the raw material precursor concentration collected and fed back by the raw material concentration sensing node; the molar ratio of hydrogen and the molar ratio of low-fluorine by-products collected and fed back by the online component detection node; the instantaneous reaction temperature collected and fed back by the temperature sensing node; and the instantaneous moisture content collected and fed back by the moisture sensing node.

10. A control system for the electrolysis process of electronic-grade perfluoroheptyl ethyl ether, characterized in that, A method for controlling the electrolysis process of electronic-grade perfluoroheptyl ethyl ether according to any one of claims 1-9, the system comprising: The state acquisition module is used to perform ultra-short-term electrolysis state tracking through multi-source online sensing nodes after the electrolytic synthesis reaction is started, and obtain the electrolysis state frame sequence. The stage identification module is used to execute S1, extract multi-scale process features from the electrolysis state frame sequence, perform reaction stage clustering identification, and locate the real-time reaction stage; The parameter matching module is used to execute S2, extract the latest state frame from the electrolysis state frame sequence, and perform control parameter matching in combination with the real-time reaction stage to obtain the target control parameter set; The impurity prediction module is used to execute S3, which predicts side reaction impurities based on historical state trend characteristics, the latest state frame, real-time reaction stage and target control parameter set, and obtains the baseline for allowable parameter fluctuations. The quality prediction module is used to execute S4, which predicts the product quality trend based on the latest state frame and the target control parameter set, and obtains the product qualified range. The closed-loop control module is used to execute S5. After adjusting the electrochemical unit of the electrolytic cell material using the target control parameter set, it corrects the parameter deviation according to the parameter allowable fluctuation baseline until the parameter deviation meets the parameter allowable fluctuation baseline and the product quality trend falls within the product qualified range. Then, it triggers the cyclic iteration steps S1 to S5 to perform full-process dynamic closed-loop control until the preset reaction endpoint judgment condition is met, and stops the electrolysis power supply and process control.

Images