An electrolyte preparation control method and system based on electrochemical fingerprint feature extraction and trajectory correction

Abstract

The application discloses an electrolyte preparation control method and system based on electrochemical fingerprint feature extraction and trajectory correction, relates to the technical fields of electrolyte preparation process control, electrochemical detection and intelligent manufacturing, and comprises the following steps: obtaining electrochemical response data sequences and process event sequences of a target batch, and dividing stage response segments according to process stage identification; extracting electrochemical fingerprint state vectors of the stage response segments, and writing preparation state trajectories in the order of process stages; performing phase correction on the preparation state trajectories, generating real-time control parameters based on deviations, and updating a correction mapping relationship. The method disclosed by the application achieves better effects in terms of stage identification accuracy, state representation integrity, real-time control parameter generation, and preparation process consistency.

Description

Technical Field

[0001] This invention relates to the fields of electrolyte preparation process control, electrochemical detection and intelligent manufacturing technology, specifically to an electrolyte preparation control method and system based on electrochemical fingerprint feature extraction and trajectory correction. Background Technology

[0002] Current electrolyte preparation processes largely rely on pre-set process formulations, fixed stage durations, and human experience for control. Although some preparation systems can collect process data such as temperature, pressure, flow rate, conductivity, or electrochemical response, these data are usually only used as single-point monitoring indicators and are difficult to further form comprehensive state characteristics that can characterize reaction progress, mass transfer hysteresis, and stability changes.

[0003] Especially during the stage switching process, the start and end times of the stage recorded in the process event record are often not completely consistent with the actual abrupt change points in the electrochemical response curve. If the data segments are still directly divided according to the event time, it is easy to cause the boundary of the stage response segment to shift, which may lead to misjudgment in subsequent state identification and parameter adjustment.

[0004] Existing control methods typically lack a dynamic correction mechanism for the prepared state trajectory. When the actual response process is ahead, behind, or drifts between stages relative to the target trajectory, the system struggles to distinguish in a timely manner whether the deviation originates from phase misalignment, abnormal reaction progress, mass transfer lag, or stability degradation, resulting in unclear basis for generating real-time control parameters.

[0005] While some methods can adjust feeding, mixing, or temperature parameters based on test results, their adjustment rules are mostly fixed thresholds or manually set rules, making it difficult to continuously update and correct the mapping relationship by combining feedback response data. As a result, the adaptability and traceability of the control results are insufficient. Summary of the Invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by this invention is that existing electrolyte preparation control methods usually rely on process event time, manual experience parameters, or single process detection data for control. These methods suffer from problems such as difficulty in accurately corresponding the electrochemical response state with the actual process stage, inaccurate stage boundary division, low dimension of preparation process state characterization, lag in control parameter generation, and insufficient feedback correction capability. Furthermore, when the stage switching event is inconsistent with the actual electrochemical response abrupt change, it is difficult to identify the phase shift in time and correct the preparation state trajectory, resulting in a low matching degree between real-time control parameters and the current preparation state, which affects the consistency and stability of the electrolyte preparation process.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction, comprising acquiring the electrochemical response data sequence and process event sequence of the target batch, and dividing the stage response segments according to the process stage identifier.

[0009] The electrochemical fingerprint state vectors of the response fragments in the extraction stage are written into the preparation state trajectory in the order of the process stages.

[0010] Phase correction is performed on the prepared state trajectory, and real-time control parameters are generated based on the deviation and the correction mapping relationship is updated.

[0011] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the step of dividing the stage response segments according to the process stage identifier includes reading the stage start event, stage end event and stage switching event corresponding to the target batch from the process event sequence.

[0012] Electrochemical response data located between the start and end events of the same phase are written into the same phase response fragment.

[0013] When the stage switching event is inconsistent with the response mutation point in the electrochemical response data, a stage boundary offset is generated.

[0014] The start and end positions of the stage response segments are corrected based on the stage boundary offset, and the corrected stage response segments are associated with the process stage identifier.

[0015] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the electrochemical fingerprint state vector includes a reaction progress component, a mass transfer hysteresis component, and a stability component.

[0016] The reaction progress component is determined by the offset of the peak position, peak amplitude, or plateau interval in the stage response segment relative to the target stage response baseline.

[0017] The mass transfer hysteresis component is determined by the correspondence between the change in response slope, the amount of phase change, and the duration of the stage.

[0018] The stability components are determined by the response fluctuation, fragment state identifier, and stage boundary offset.

[0019] The reaction progress component, mass transfer hysteresis component, and stability component are all written into the electrochemical fingerprint state vector.

[0020] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the step of writing the preparation state trajectory in the order of process stages includes determining the writing position of the electrochemical fingerprint state vector in the preparation state trajectory according to the process stage identifier.

[0021] The process event sequence determines the stage event anchor point, and the response morphology anchor point is determined by the peak inflection point, plateau start point, or slope change point in the stage response segment.

[0022] Each electrochemical fingerprint state vector, along with stage event anchors and response morphology anchors, is written into the preparation state trajectory.

[0023] When the offset between the stage event anchor point and the response morphology anchor point exceeds the preset anchor point offset range, the corresponding electrochemical fingerprint state vector is marked as a state vector to be corrected.

[0024] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the step of performing phase correction on the preparation state trajectory includes reading stage event anchor points and response morphology anchor points under the same process stage from the preparation state trajectory.

[0025] The first phase offset is determined based on the stage event anchor point, and the second phase offset is determined based on the response pattern anchor point.

[0026] When the difference between the first phase offset and the second phase offset is within the allowable difference range, the target phase correction amount is determined based on the weighted result of the first phase offset and the second phase offset.

[0027] When the difference between the first phase offset and the second phase offset exceeds the allowable difference range, the corresponding process stage is marked as the anchor point conflict stage, and it is prohibited to directly input the deviation of the process stage into the real-time control parameter generation step.

[0028] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the deviations in the process stages include phase deviation components, reaction progress deviation components, mass transfer hysteresis deviation components, and stability degradation components.

[0029] The phase deviation component is determined by the position difference of the preparation state trajectory before and after phase correction, and the reaction progress deviation component is determined by the difference between the phase-corrected reaction progress component and the corresponding reaction progress benchmark in the target trajectory.

[0030] The mass transfer hysteresis deviation component is determined by the difference between the phase-corrected mass transfer hysteresis component and the corresponding mass transfer hysteresis reference in the target trajectory.

[0031] The stability degradation component is determined by the difference between the phase-corrected stability component and the corresponding stability benchmark in the target trajectory.

[0032] The phase deviation component, reaction progress deviation component, mass transfer hysteresis deviation component, and stability degradation component are written into the trajectory deviation object.

[0033] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the real-time control parameters based on deviation generation include constructing a control correction vector based on the phase deviation component, reaction progress deviation component, mass transfer hysteresis deviation component, and stability degradation component in the trajectory deviation object.

[0034] Invoke the parameter coupling constraint relationship corresponding to the current process stage, solve the constraint on the control correction vector, and obtain at least two types of candidate control parameters from the following: feeding ratio parameter, mixing time parameter, temperature setting parameter, settling time parameter, or filtering timing parameter.

[0035] When at least two types of candidate control parameters satisfy the parameter coupling constraint relationship, the corresponding real-time control parameters are output.

[0036] When at least two types of candidate control parameters do not satisfy the parameter coupling constraint relationship, the control correction vector is split into the control parameters of the current stage and the compensation parameters of the next stage according to the priority of the process stage.

[0037] As a preferred embodiment of the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, the updated and corrected mapping relationship includes: acquiring feedback response data after executing real-time control parameters, and generating a feedback electrochemical fingerprint state vector based on the feedback response data.

[0038] The type of feedback residual is determined based on the residual between the feedback electrochemical fingerprint state vector and the target trajectory.

[0039] When the feedback residual type is a mapped valid residual, the corresponding deviation, real-time control parameters and residual are written into the available sample set for correcting the mapping relationship.

[0040] When the feedback residual type is phase model failure residual, update the stage event anchor or response morphology anchor used for phase correction.

[0041] When the feedback residual type is fingerprint drift residual, update the baseline of the reaction progress component, mass transfer hysteresis component, or stability component in the electrochemical fingerprint state vector.

[0042] When the feedback residual type is insufficient control response residual, adjust the parameter coupling constraint relationship or generate a mapping verification identifier.

[0043] Another objective of this invention is to provide an electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction. This system, through collaborative processing among a data acquisition unit, a segmentation unit, a fingerprint extraction unit, a trajectory correction unit, a parameter generation unit, and a mapping update unit, achieves the following: correlation between electrochemical response data and process event sequences; boundary correction of stage response segments; extraction of electrochemical fingerprint state vectors; phase correction of the preparation state trajectory; generation of real-time control parameters; and feedback updating of the corrected mapping relationship. This solves the problems in current electrolyte preparation control technologies, such as reliance on human experience for state recognition, lack of dynamic correction basis for control parameters, and difficulty in updating the control mapping relationship using feedback data.

[0044] As a preferred embodiment of the electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, it includes: a data acquisition unit, a fragment segmentation unit, a fingerprint extraction unit, a trajectory correction unit, a parameter generation unit, and a mapping update unit.

[0045] The data acquisition unit is used to acquire the electrochemical response data sequence and process event sequence of the target batch.

[0046] The segmentation unit is used to divide the stage response segments according to the process stage identifier and to correct the stage response segments according to the stage boundary offset.

[0047] The fingerprint extraction unit is used to extract the electrochemical fingerprint state vector of the stage response fragment and write it into the preparation state trajectory in the order of process stages.

[0048] The trajectory correction unit is used to perform phase correction on the prepared state trajectory based on the stage event anchor point and the response morphology anchor point.

[0049] The parameter generation unit is used to generate real-time control parameters based on the trajectory deviation object and the parameter coupling constraint relationship.

[0050] The mapping update unit is used to update and correct the mapping relationship based on the feedback residual type.

[0051] As a preferred embodiment of the electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction described in this invention, it further includes: a data acquisition unit connected to the preparation process data interface, a fragment segmentation unit connected to the data acquisition unit, a fingerprint extraction unit connected to the fragment segmentation unit, a trajectory correction unit connected to both the fingerprint extraction unit and the target trajectory library, a parameter generation unit connected to both the trajectory correction unit and the correction mapping relationship library, and a mapping update unit connected to both the feedback data interface, the target trajectory library, and the correction mapping relationship library.

[0052] The parameter generation unit writes the real-time control parameters into the control parameter output interface, and the mapping update unit updates the phase correction parameters, electrochemical fingerprint state vector reference, parameter coupling constraint relationship or mapping verification identifier according to the feedback residual type.

[0053] This invention provides an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction. By dividing the response segments according to process stage identifiers and correcting them using stage boundary offsets, the matching degree between process stage division and actual electrochemical response changes is improved. By extracting an electrochemical fingerprint state vector containing reaction progress, mass transfer hysteresis, and stability components, the quantitative characterization capability of the electrolyte preparation process state is enhanced. Phase correction of the preparation state trajectory is performed using stage event anchors and response morphology anchors, reducing the impact of asynchronous stage event times and actual response states on control judgment. Real-time control parameters are generated through trajectory deviation objects, enabling dynamic adjustment of control parameters such as feeding ratio, mixing time, temperature setting, settling time, or filtration sequence based on the current preparation state. The mapping relationship is updated and corrected through feedback residual type updates, giving the system the ability to continuously correct control rules based on actual execution results. This invention achieves better results in terms of stage identification accuracy, state characterization completeness, real-time control parameter generation, and consistency of the preparation process. Attached Figure Description

[0054] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0055] Figure 1 This is a schematic diagram of the stage response segment division and electrochemical fingerprint state vector extraction of an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction provided in Embodiment 1 of the present invention.

[0056] Figure 2 This is a schematic diagram of the preparation state trajectory phase correction and trajectory deviation object generation of an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction provided in Embodiment 2 of the present invention.

[0057] Figure 3 This is a schematic diagram illustrating the real-time control parameter generation and feedback update based on the trajectory deviation object in an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction provided in Embodiment 3 of the present invention.

[0058] Figure 4The above is a system flowchart of an electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction provided in Embodiment 4 of the present invention. Detailed Implementation

[0059] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0060] It should be noted that in the formula content of the embodiment, When used as a superscript or subscript, it is only used to distinguish the corresponding complete symbol and is not interpreted as an independent variable.

[0061] Example 1

[0062] Reference Figure 1 This is an embodiment of the present invention, which provides a specific implementation method for stage response segment division and electrochemical fingerprint state vector extraction in an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction.

[0063] First, obtain the electrochemical response data sequence and process event sequence for the target batch.

[0064] The electrochemical response data sequence includes at least the sampling time and the response amplitude, and preferably also includes the phase quantity.

[0065] The process event sequence includes stage start events, stage end events, and stage switching events, and each event has an event time and a process stage identifier.

[0066] If the electrochemical response data and the process event sequence come from different time sources, the electrochemical response data are time-aligned based on the batch time of the process event sequence.

[0067] If the time interval between two adjacent sampling points is greater than the normal sampling period but does not exceed 30 seconds, then adjacent valid sampling values ​​are used for linear padding. If it exceeds 30 seconds, then it is marked as a data shortage interval.

[0068] Furthermore, the electrochemical response data is smoothed.

[0069] Specifically, five consecutive sampling points are used as a short-time window, and the median of the response amplitude within the window is used as the smoothed response value of the current point.

[0070] The median, rather than the simple average, is used to reduce the impact of occasional spike noise on the peak position, slope variation, and plateau interval judgment.

[0071] The smoothed response data serves as input for subsequent stage segmentation and fingerprint extraction.

[0072] Furthermore, the start and end events of the same process stage are read from the process event sequence.

[0073] Electrochemical response data located between the start and end events of the same phase are written into the same initial phase response fragment.

[0074] If a segment has fewer than 30 valid sampling points or its effective duration is less than 20% of the target duration for that segment, then that segment is marked as a data-insufficient segment.

[0075] Data segments with insufficient data are not directly involved in the generation of control parameters, but their stage identifiers and time ranges are retained for subsequent trajectory integrity assessment.

[0076] Furthermore, the stage boundaries are corrected based on the stage switching events and response mutation points in the electrochemical response data.

[0077] Specifically, the search range is 180 seconds before and 180 seconds after the stage switching event time.

[0078] Within this search range, the mean difference of the response five sampling points before and after each sampling point is calculated. If the mean difference before and after a sampling point is more than three times the standard deviation of the response fluctuation in the current stage, then the sampling point is considered a candidate response mutation point.

[0079] Among all candidate response mutation points, the sampling point with the largest mean difference is selected as the actual response mutation point.

[0080] Furthermore, the difference between the actual response mutation point time and the stage switching event time is used as the stage boundary offset.

[0081] If the offset does not exceed 300 seconds, the start and end positions of the phase response segment are corrected according to the offset.

[0082] If the offset exceeds 300 seconds, the boundary limit correction is performed only for 300 seconds, and the segment in that phase is marked as a boundary anomalous state.

[0083] If no response mutation point that meets the conditions is found within the search range, the stage boundary offset is set to 0, and the stage segment is marked as a low mutation confidence state.

[0084] Extract the reaction progress component, specifically:

[0085] In the corrected stage response segment, identify the peak location, peak amplitude, and plateau response. The peak location is the moment when the response amplitude within the stage segment reaches a local maximum and is higher than the segment mean by a certain margin; the peak amplitude is the difference between the peak response and the stage start response; the plateau response is the average of the interval where the response rate of change is continuously lower than the maximum response rate of change by 10% for a duration of not less than 30 seconds.

[0086] Subsequently, the peak position is compared with the target peak position, the peak amplitude is compared with the target peak amplitude, and the platform response is compared with the target platform response.

[0087] The three comparison results were normalized and then used in the calculation of the reaction progress component, with the peak position deviation accounting for 40%, the peak amplitude deviation accounting for 40%, and the platform response deviation accounting for 20%.

[0088] If there is no obvious peak in a certain stage, the peak correlation result is not forced to be used. Instead, the weights of the peak position and peak amplitude are assigned to the platform response and the end response of the stage to ensure that the calculation process is not interrupted.

[0089] Extracting the mass transfer hysteresis component, specifically:

[0090] First, calculate the response change trend between adjacent sampling points within the stage segment, and then statistically analyze the fluctuation of this trend throughout the entire stage; next, read the phase change amount between the stage start interval and the stage end interval; and simultaneously calculate the difference between the actual duration of the stage and the target stage duration.

[0091] The mass transfer hysteresis component is determined by three types of results: the degree of fluctuation in the response change trend accounts for 40%, the phase change deviation accounts for 40%, and the stage duration deviation accounts for 20%.

[0092] If the phase quantity is missing, the difference between the time required for the response to reach 90% of the target platform response and the target arrival time is used to replace the phase change quantity; if both the phase quantity and the arrival time are unavailable, this stage is marked as a low confidence state of the mass transfer component.

[0093] Extracting the stability component, specifically:

[0094] First, calculate the degree of fluctuation in the response amplitude within the segment. This degree of fluctuation can be represented by the ratio of the response standard deviation to the response mean.

[0095] Then, the fragment status is converted into anomaly level: normal fragments are assigned the lowest anomaly level, low mutation confidence states are assigned the medium anomaly level, and boundary anomaly states are assigned the highest anomaly level.

[0096] Finally, the stability components are determined by combining the magnitude of the stage boundary offset, with response fluctuation accounting for 60%, segment state identifier accounting for 20%, and stage boundary offset accounting for 20%.

[0097] If the stability component is not higher than 0.08, the response at this stage is considered stable. If it is greater than 0.08 but not higher than 0.2, acceptable fluctuations are considered to exist. If it is greater than 0.2, stability is considered to have deteriorated. This result is subsequently used to limit the adjustment range of the control parameters.

[0098] Furthermore, the reaction progress component, mass transfer hysteresis component, and stability component are all incorporated into the electrochemical fingerprint state vector. A preferred scheme for incorporating these components into the electrochemical fingerprint state vector is as follows:

[0099]

[0100] in, Indicates the process stage number. Indicates the first Electrochemical fingerprint state vectors for each process stage Indicates the reaction progress component. Indicates the mass transfer lag component. This represents the stability component.

[0101] The electrochemical fingerprint state vector is output together with the process stage identifier, the corrected stage response fragment, the stage boundary offset, and the fragment state identifier.

[0102] Example 2

[0103] Reference Figure 2 This is an embodiment of the present invention, which provides a specific implementation method for writing the preparation state trajectory, determining the anchor point, and correcting the phase in an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction.

[0104] First, read the electrochemical fingerprint state vector, process stage identifier, corrected stage response fragment, stage boundary offset, and fragment state identifier.

[0105] Simultaneously, the target stage sequence, target event anchor point, target response form anchor point, and target stage duration are read from the target trajectory library corresponding to the process stage.

[0106] Furthermore, the writing position of the electrochemical fingerprint state vector in the preparation state trajectory is determined according to the process stage identifier.

[0107] Specifically, the target trajectory library pre-stores the standard sequence of process stages, such as the feeding stage, mixing stage, settling stage, and filtering stage.

[0108] If the current stage identifier matches a stage in the standard sequence, then the current electrochemical fingerprint state vector is written to the trajectory position corresponding to that stage.

[0109] If the same stage identifier appears repeatedly in the target batch, the sub-trajectory positions of that stage are written in the order of the stage start event times.

[0110] If a stage identifier is missing, the candidate position is inferred based on the event times of the previous and next stages; if it cannot be inferred, the segment is not used for automatic control and is only retained as an abnormal trajectory record.

[0111] Furthermore, identify the anchor points for each stage of the event.

[0112] Stage event anchors are used to represent the stage time base in the process record. Stage switching events are preferred as stage event anchors.

[0113] If a phase transition event is missing, the phase start event is selected as the phase event anchor point.

[0114] If both are missing, the event anchor point is marked as missing, and the confidence level of the trajectory node for that stage is reduced. The output of the stage event anchor point is used for subsequent calculation of the first phase offset.

[0115] Furthermore, determine the response morphology anchor point.

[0116] Response morphology anchors are used to represent key moments in the electrochemical response curve that reflect changes in the process state.

[0117] The response morphology anchor points are determined sequentially according to the slope abrupt change point, the plateau start point, and the peak inflection point.

[0118] Specifically, the first step is to determine whether there is a significant abrupt change in slope in the stage response segment. The criterion for this is that the difference in the response trend before and after a certain sampling point is more than three times the normal fluctuation level within the stage.

[0119] If it exists, the sampling point with the largest difference is selected as the response pattern anchor point.

[0120] If there is no obvious abrupt change in slope, then find the plateau starting point. The plateau starting point is the first sampling moment when the response rate of change is continuously lower than the maximum response rate of change by 10% for a duration of not less than 30 seconds.

[0121] If the platform starting point does not exist, then look for the peak inflection point. The peak inflection point is the moment when the response within a segment reaches a local extreme and the amplitude is significantly higher than the segment mean.

[0122] If none of the three types of points can be determined, the response pattern anchor point will be marked as missing.

[0123] The electrochemical fingerprint state vector, stage event anchors, and response morphology anchors are all written into the preparation state trajectory.

[0124] Each trajectory node includes a process stage identifier, an electrochemical fingerprint state vector, a stage event anchor point, a response morphology anchor point, a stage boundary offset, and a segment state identifier.

[0125] Trajectory nodes are used to represent the actual preparation status of the target batch at the corresponding process stage.

[0126] To determine whether trajectory nodes need correction, the specific steps are as follows:

[0127] When both the stage event anchor point and the response pattern anchor point exist, calculate the time offset between them.

[0128] If the time offset does not exceed 20% of the target stage duration and does not exceed 120 seconds, then the consistency of the two anchor points is considered to meet the requirements.

[0129] If the value exceeds this range, the corresponding electrochemical fingerprint state vector will be marked as a state vector to be corrected.

[0130] If any anchor point is missing, the trajectory node is marked as low confidence and will not directly block subsequent processes, but the adjustment range needs to be reduced when generating subsequent control parameters.

[0131] Calculate the first phase offset and the second phase offset. The first phase offset is determined by the time offset of the stage event anchor point relative to the target event anchor point and normalized by the target stage duration.

[0132] The second phase offset is determined by the time offset of the response pattern anchor point relative to the target response pattern anchor point and normalized by the target stage duration.

[0133] The normalized phase offset is convenient for comparison between different process stages.

[0134] Furthermore, the target phase correction amount is determined based on the first phase offset and the second phase offset.

[0135] A preferred method for determining the target phase correction amount is:

[0136]

[0137] in, Indicates the target phase correction amount. This indicates the first phase offset. This indicates the second phase offset. Indicates the anchor point weight of the stage event. This indicates the anchor weight of the response pattern.

[0138] The aforementioned target phase correction amount aims to simultaneously reduce the deviation between the correction result and the first phase offset and the second phase offset.

[0139] Specifically, the differences between the desired phase correction and the first and second phase offsets are determined. The squares of these two differences are multiplied by their respective weights and summed. The minimum value of this weighted sum of squares is used to solve for the desired phase correction. When the sum of the weights of the stage event anchor points and the response morphology anchor points is 1, the solution is the weighted sum of the first and second phase offsets, which yields the target phase correction. Therefore, anchor points with larger weights have a greater constraining effect on the trajectory correction position.

[0140] The sum of the two weights is 1. Under normal circumstances, both weights are 0.5.

[0141] If the fragment state is a low mutation confidence state, the weight of the response morphology anchor is reduced to 0.3, and the weight of the stage event anchor is increased to 0.7.

[0142] If the event anchor is missing but the response morphology anchor is valid, then only the response morphology anchor is used to determine the phase correction amount, and the correction result is marked as a low confidence result.

[0143] To determine whether there is an anchor point conflict, if the difference between the first phase offset and the second phase offset is not greater than 0.1, it is considered that the two anchor points can be used together for phase correction, and the trajectory position in the prepared state trajectory is corrected according to the target phase correction amount.

[0144] When the difference between the two is greater than 0.1, the corresponding process stage will be marked as the anchor point conflict stage. It is prohibited to directly input the deviation of this process stage into the real-time control parameter generation process to avoid erroneous control due to time base errors.

[0145] Furthermore, the calculation results of the target phase correction amount were verified.

[0146] Specifically, three stage response trajectories with different target stage durations are read from the target trajectory library. A predetermined time offset is applied to each stage response trajectory, and a response perturbation with a mean of 0 and an amplitude not exceeding 1.5% of the corresponding stage response range is superimposed to obtain three sets of verification sequences. The pre-applied time offset is used as the true value of the phase offset for each verification sequence, and then each verification sequence is processed according to the aforementioned method for determining stage event anchor points and response morphology anchor points.

[0147] The stage event anchors and response morphology anchors in all three sets of verification sequences can be determined, and the segment state identifiers are all normal. Therefore, the weights of both the stage event anchors and response morphology anchors are set to 0.5. The target phase correction amount is multiplied by the corresponding target stage duration to obtain the time correction amount used to correct the preparation of the state trajectory, and this time correction amount is used as the prediction result of the phase correction step.

[0148] The target phase duration of the first verification sequence is 1735 seconds, and the pre-applied true phase offset is 67.9 seconds. The time offset determined by the phase event anchor is 61.8 seconds, and the time offset determined by the response morphology anchor is 75.6 seconds. After normalizing both by the target phase duration, the difference between the first and second phase offsets is 0.00795, which is less than 0.1. Based on weights of 0.5, a time correction of 68.70 seconds is obtained, and the absolute difference between this time correction and the true phase offset is 0.80 seconds.

[0149] The target phase duration of the second verification sequence is 1480 seconds, and the pre-applied true phase offset is -42.6 seconds, where a negative value indicates that the verification sequence is ahead of the target phase response trajectory. The time offset determined by the phase event anchor point is -38.7 seconds, and the time offset determined by the response morphology anchor point is -47.9 seconds. After normalizing both by the target phase duration, the difference between the first and second phase offsets is 0.00622, which is less than 0.1. Based on weights of 0.5, a time correction of -43.30 seconds is obtained, and the absolute difference between this time correction and the true phase offset is 0.70 seconds.

[0150] The target phase duration of the third verification sequence is 2160 seconds, and the pre-applied true phase offset is 112.4 seconds. The time offset determined by the phase event anchor is 103.5 seconds, and the time offset determined by the response morphology anchor is 119.8 seconds. After normalizing both by the target phase duration, the difference between the first and second phase offsets is 0.00755, which is less than 0.1. Based on weights of 0.5, a time correction of 111.65 seconds is obtained, and the absolute difference between this time correction and the true phase offset is 0.75 seconds.

[0151] The mean absolute difference between the time correction and the true phase shift of the three sets of verification sequences was 0.75 seconds, the maximum absolute difference was 0.80 seconds, and the maximum relative deviation was 1.64%.

[0152] Furthermore, the corresponding time correction amount is subtracted from the stage event anchor point offset and response morphology anchor point offset in each verification sequence to obtain the phase-corrected anchor point deviation.

[0153] The phase event anchor point deviation and response morphology anchor point deviation of the first verification sequence changed from 61.8 seconds and 75.6 seconds to -6.9 seconds and 6.9 seconds, respectively, and the average absolute deviation of the two anchor points decreased from 68.70 seconds to 6.90 seconds.

[0154] The stage event anchor point deviation and response morphology anchor point deviation of the second verification sequence changed from -38.7 seconds and -47.9 seconds to 4.6 seconds and -4.6 seconds, respectively, and the average absolute deviation of the two anchor points decreased from 43.30 seconds to 4.60 seconds.

[0155] The phase event anchor point deviation and response morphology anchor point deviation of the third verification sequence changed from 103.5 seconds and 119.8 seconds to -8.15 seconds and 8.15 seconds, respectively, and the average absolute deviation of the two anchor points decreased from 111.65 seconds to 8.15 seconds.

[0156] Therefore, the time correction corresponding to the target phase correction is consistent with the pre-applied true value of the phase offset, and it can reduce the overall deviation of the stage event anchor and the response morphology anchor relative to the corresponding target anchor. The phase correction process does not change the original sequential relationship between the two anchors, nor does it change the reaction progress component, mass transfer hysteresis component, and stability component in the electrochemical fingerprint state vector.

[0157] A trajectory deviation object is generated based on the phase-corrected trajectory position.

[0158] The trajectory deviations include phase deviation components, reaction progress deviation components, mass transfer hysteresis deviation components, and stability degradation components.

[0159] The phase deviation component is determined by the difference in trajectory position before and after phase correction; the reaction progress deviation component is determined by the difference between the phase-corrected reaction progress component and the corresponding reaction progress benchmark in the target trajectory; the mass transfer hysteresis deviation component is determined by the difference between the phase-corrected mass transfer hysteresis component and the corresponding mass transfer hysteresis benchmark in the target trajectory; and the stability degradation component is determined by the difference between the phase-corrected stability component and the corresponding stability benchmark in the target trajectory.

[0160] Finally, the trajectory deviation object is output. If the status of the trajectory deviation object is normal or low confidence, the real-time control parameter generation process is allowed; if the status is anchor point conflict, only the verification mark is output and the automatic control parameter generation process is not entered.

[0161] Example 3

[0162] Reference Figure 3 This is an embodiment of the present invention, which provides a specific implementation method for generating real-time control parameters based on trajectory deviation objects and updating and correcting mapping relationships in an electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction.

[0163] First, read the trajectory deviation object, which includes the process stage identifier, phase deviation component, reaction progress deviation component, mass transfer lag deviation component, stability degradation component, trajectory position and status identifier after phase correction.

[0164] If the status flag indicates an anchor point conflict, automatic real-time control parameters will not be generated. Instead, the previously verified set of basic control parameters for the current stage will be retained, and a mapping verification flag will be output.

[0165] If the status is marked as normal or low confidence, the control correction process will begin.

[0166] Furthermore, the phase deviation component, reaction progress deviation component, mass transfer hysteresis deviation component, and stability degradation component are combined into a control correction vector.

[0167] If the overall deviation of the four types of deviations is less than 0.03, the current deviation is considered to be in the control dead zone, no parameter adjustment is made, and only the zero correction result is output.

[0168] The control dead zone is set to avoid process fluctuations caused by frequent fine-tuning. If the overall deviation is not less than 0.03, candidate control parameters are generated.

[0169] Furthermore, the corresponding parameter coupling constraint relationship is invoked according to the current process stage.

[0170] The control parameters used in different process stages are different, but they include at least two of the following: feeding ratio parameters, mixing time parameters, temperature setting parameters, settling time parameters, or filtration sequence parameters.

[0171] For mixing-related stages, prioritize using parameters such as feed ratio, mixing duration, and temperature setting.

[0172] For the static or post-processing stages, prioritize the use of static duration parameters and filtering timing parameters. Parameters not involved in the current stage control remain unchanged to avoid introducing unnecessary parameters into the control process.

[0173] Furthermore, a preferred approach for generating candidate control parameters by modifying the mapping relationship and calculating the changes in the candidate control parameters is as follows:

[0174]

[0175] in, This indicates the amount of change in the candidate control parameter. This represents the control correction vector. This indicates the corrected mapping relationship corresponding to the current process stage.

[0176] The corrected mapping relationship is implemented using an interpretable coefficient table, rather than a black-box model. This coefficient table is used to represent the direction and intensity of the influence of different deviation components on various control parameters.

[0177] Specifically, when the reaction progress deviation indicates a reaction lag, prioritize increasing the mixing time or fine-tuning the feeding ratio.

[0178] When the mass transfer hysteresis deviation increases, the mixing time should be extended first, and the temperature setting should be fine-tuned according to the safe temperature range.

[0179] When the stability degradation component increases, prioritize increasing the resting time or delaying the filtering timing. When the phase deviation display stage progresses ahead of schedule or behind schedule, prioritize adjusting timing parameters.

[0180] The single adjustment range of the feeding ratio is preferably no more than ±2%, the single adjustment range of the mixing time is preferably no more than ±15 minutes, the single adjustment range of the temperature setting is preferably no more than ±3℃, the single adjustment range of the settling time is preferably no more than ±30 minutes, and the single adjustment range of the filtration sequence is preferably no more than ±10 minutes.

[0181] If the candidate parameter exceeds the above range, the boundary value is taken and the parameter saturation state is recorded.

[0182] Perform parameter coupling constraint judgment on at least two types of candidate control parameters.

[0183] Specifically, the adjusted feeding ratio should be kept within the allowable ratio range of the process, the adjusted temperature setting should be kept between 15℃ and 45℃, the total increase in mixing time and settling time should not exceed the allowable delay time of the batch, preferably not exceeding 60 minutes, and when the temperature rises by more than 1℃, the mixing time should not be shortened by more than 3 minutes simultaneously to avoid unstable response caused by both temperature rise and insufficient mixing.

[0184] When all coupling conditions are met, the corresponding real-time control parameters are output.

[0185] When at least two types of candidate control parameters do not satisfy the parameter coupling constraint relationship, all candidate control parameters are not directly output, but are split according to the priority of the process stage.

[0186] Specifically: if the dominant deviation is the reaction progress deviation, then the feeding ratio parameter and mixing time parameter should be retained first, and other adjustments should be written into the next stage of compensation.

[0187] If the dominant deviation is mass transfer hysteresis deviation, then the mixing time parameter and temperature setting parameter should be retained first.

[0188] If the dominant deviation is a stability degradation component, then the static time parameter or the filtering timing parameter should be retained first.

[0189] If the dominant deviation is phase deviation, then time-series parameters are retained first. Adjustments that cannot be performed in the current stage are used as compensation parameters for the next stage and participate in the judgment together with the new trajectory deviation object in the next stage.

[0190] After executing the real-time control parameters, obtain the feedback response data.

[0191] The preferred window for collecting feedback response data is 10 minutes after the execution of real-time control parameters and continues until the end of the current stage.

[0192] If less than 10 minutes remain in the current phase, the time will be extended to the first 5 minutes after the start of the next phase.

[0193] The feedback electrochemical fingerprint state vector was re-extracted from the feedback response data according to the rules in Example 1.

[0194] The feedback electrochemical fingerprint state vector is compared with the target fingerprint state at the corresponding position in the target trajectory to obtain the feedback residual.

[0195] The type of feedback residual is determined based on the feedback residual.

[0196] If the normalized offset difference between the stage event anchor point and the response pattern anchor point in the feedback data is still greater than 0.1, it is determined to be a phase model failure residual.

[0197] If the phase anchor point is normal, but more than three consecutive batches in the same process stage show residuals in the same direction, and the average residual value is greater than 0.05, then it is determined to be a fingerprint drift residual.

[0198] If the candidate control parameter has reached the parameter adjustment boundary, but the improvement of the feedback residual is less than 20%, it is determined to be an insufficient control response residual.

[0199] If the improvement of the feedback residual reaches more than 40%, and the overall feedback residual does not exceed 0.08, it is determined to be a mapping effective residual.

[0200] If any of the above conditions are not met, it will be treated as a normal observation residual, and only recorded without updating the core mapping relationship.

[0201] Furthermore, the mapping relationship is updated and corrected based on the feedback residual type.

[0202] When the feedback residual type is a mapped effective residual, the current deviation, real-time control parameters, and feedback residual are written into the available sample set and used to fine-tune the correction mapping relationship in the current stage.

[0203] The fine-tuning range is preferably 5% to 20% of the historical mapping coefficients, and the fine-tuning range for low-confidence samples is halved.

[0204] When the feedback residual type is phase model failure residual, the control mapping relationship is not updated; instead, the stage event anchor or response morphology anchor used for phase correction is updated.

[0205] When the feedback residual type is fingerprint drift residual, the target benchmark for updating the reaction progress component, mass transfer lag component, or stability component is updated, and the update magnitude is preferably 10% of the mean of continuous effective feedback.

[0206] When the feedback residual type is insufficient control response residual, first determine whether the candidate control parameters have been pruned due to coupling constraints.

[0207] If the trimming does not reach the process safety boundary, the corresponding single adjustment limit will be increased by 10%.

[0208] If the process safety boundary has been reached, the adjustment scope will not be expanded, but a mapping verification identifier will be generated instead.

[0209] Finally, the updated corrected mapping relationships, parameter coupling constraint relationships, target trajectory benchmarks or mapping verification identifiers are saved for use in subsequent process stages or subsequent target batches.

[0210] Example 4

[0211] Reference Figure 4 As an embodiment of the present invention, an electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction is provided, comprising: a data acquisition unit 100, a fragment division unit 200, a fingerprint extraction unit 300, a trajectory correction unit 400, a parameter generation unit 500, and a mapping update unit 600.

[0212] The data acquisition unit 100 is used to acquire the electrochemical response data sequence and process event sequence of the target batch.

[0213] The segmentation unit 200 is used to divide the stage response segments according to the process stage identifier and to correct the stage response segments according to the stage boundary offset.

[0214] The fingerprint extraction unit 300 is used to extract the electrochemical fingerprint state vector of the stage response fragment and write it into the preparation state trajectory in the order of the process stages.

[0215] The trajectory correction unit 400 is used to perform phase correction on the prepared state trajectory based on the stage event anchor point and the response morphology anchor point.

[0216] The parameter generation unit 500 is used to generate real-time control parameters based on the trajectory deviation object and the parameter coupling constraint relationship.

[0217] The mapping update unit 600 is used to update and correct the mapping relationship according to the feedback residual type.

[0218] The data acquisition unit 100 is connected to the preparation process data interface, the fragment segmentation unit 200 is connected to the data acquisition unit 100, the fingerprint extraction unit 300 is connected to the fragment segmentation unit 200, the trajectory correction unit 400 is connected to the fingerprint extraction unit 300 and the target trajectory library respectively, the parameter generation unit 500 is connected to the trajectory correction unit 400 and the correction mapping relation library respectively, and the mapping update unit 600 is connected to the feedback data interface, the target trajectory library and the correction mapping relation library respectively.

[0219] The parameter generation unit 500 writes the real-time control parameters into the control parameter output interface, and the mapping update unit 600 updates the phase correction parameters, electrochemical fingerprint state vector reference, parameter coupling constraint relationship or mapping verification identifier according to the feedback residual type.

[0220] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0221] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for controlling electrolyte preparation based on electrochemical fingerprint feature extraction and trajectory correction, characterized in that, include: Obtain the electrochemical response data sequence and process event sequence of the target batch, and divide the response segments into stages based on the process stage identifier; The electrochemical fingerprint state vectors of the response fragments in the extraction stage are written into the preparation state trajectory in the order of the process stages; Phase correction is performed on the prepared state trajectory, and real-time control parameters are generated based on the deviation and the correction mapping relationship is updated. The process stage response segment division based on process stage identifiers includes: Read the phase start event, phase end event, and phase switch event corresponding to the target batch from the process event sequence; Write the electrochemical response data located between the start event and the end event of the same stage into the same stage response fragment; When the stage switching event is inconsistent with the response mutation point in the electrochemical response data, a stage boundary offset is generated. The start and end positions of the stage response segments are corrected based on the stage boundary offset, and the corrected stage response segments are associated with the process stage identifier.

2. The electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in claim 1, characterized in that: The electrochemical fingerprint state vector includes a reaction progress component, a mass transfer hysteresis component, and a stability component. The reaction progress component is determined by the offset of the peak position, peak amplitude, or plateau interval in the stage response segment relative to the target stage response baseline. The mass transfer hysteresis component is determined by the correspondence between the response slope change, the phase change, and the stage duration. Stability components are determined by response fluctuation, segment status identifier, and stage boundary offset. The reaction progress component, mass transfer hysteresis component, and stability component are all written into the electrochemical fingerprint state vector.

3. The electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in claim 2, characterized in that: The process of writing the preparation status trajectory in sequence according to the process stages includes... The writing position of the electrochemical fingerprint state vector in the preparation state trajectory is determined according to the process stage identifier; The process event sequence determines the stage event anchor point, and the response morphology anchor point is determined by the peak inflection point, plateau start point or slope change point in the stage response segment. Each electrochemical fingerprint state vector, along with stage event anchors and response morphology anchors, is written into the fabrication state trajectory. When the offset between the stage event anchor point and the response morphology anchor point exceeds the preset anchor point offset range, the corresponding electrochemical fingerprint state vector is marked as a state vector to be corrected.

4. The electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in claim 3, characterized in that: The phase correction of the prepared state trajectory includes, Read the stage event anchor points and response morphology anchor points under the same process stage from the preparation state trajectory; The first phase offset is determined based on the stage event anchor point, and the second phase offset is determined based on the response pattern anchor point. When the difference between the first phase offset and the second phase offset is within the allowable difference range, the target phase correction amount is determined based on the weighted result of the first phase offset and the second phase offset. When the difference between the first phase offset and the second phase offset exceeds the allowable difference range, the corresponding process stage is marked as the anchor point conflict stage, and it is prohibited to directly input the deviation of the process stage into the real-time control parameter generation step.

5. The electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in claim 4, characterized in that: The deviations in the process stages include phase deviation components, reaction progress deviation components, mass transfer lag deviation components, and stability degradation components. The phase deviation component is determined by the position difference of the preparation state trajectory before and after phase correction; The reaction progress deviation component is determined by the difference between the phase-corrected reaction progress component and the corresponding reaction progress benchmark in the target trajectory. The mass transfer hysteresis component is determined by the difference between the phase-corrected mass transfer hysteresis component and the corresponding mass transfer hysteresis reference in the target trajectory. The stability degradation component is determined by the difference between the phase-corrected stability component and the corresponding stability benchmark in the target trajectory. The phase deviation component, reaction progress deviation component, mass transfer hysteresis deviation component, and stability degradation component are written into the trajectory deviation object.

6. The electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in claim 5, characterized in that: The generation of real-time control parameters based on deviation includes, A control correction vector is constructed based on the phase deviation component, reaction progress deviation component, mass transfer hysteresis deviation component, and stability degradation component in the trajectory deviation object. Call the parameter coupling constraint relationship corresponding to the current process stage, solve the constraint on the control correction vector, and obtain at least two types of candidate control parameters from the following: feeding ratio parameter, mixing time parameter, temperature setting parameter, settling time parameter, or filtering timing parameter. When at least two types of candidate control parameters satisfy the parameter coupling constraint relationship, the corresponding real-time control parameters are output. When at least two types of candidate control parameters do not satisfy the parameter coupling constraint relationship, the control correction vector is split into the control parameters of the current stage and the compensation parameters of the next stage according to the priority of the process stage.

7. The electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in claim 6, characterized in that: The updated and corrected mapping relationship includes, Obtain feedback response data after executing real-time control parameters, and generate a feedback electrochemical fingerprint state vector based on the feedback response data; The type of feedback residual is determined based on the residual between the feedback electrochemical fingerprint state vector and the target trajectory; When the feedback residual type is a mapped effective residual, the corresponding deviation, real-time control parameters and residual are written into the available sample set for correcting the mapping relationship; When the feedback residual type is phase model failure residual, update the stage event anchor or response morphology anchor used for phase correction. When the feedback residual type is fingerprint drift residual, update the baseline of the reaction progress component, mass transfer hysteresis component or stability component in the electrochemical fingerprint state vector. When the feedback residual type is insufficient control response residual, adjust the parameter coupling constraint relationship or generate a mapping verification identifier.

8. An electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction, employing the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in any one of claims 1 to 7, characterized in that: It includes a data acquisition unit (100), a segmentation unit (200), a fingerprint extraction unit (300), a trajectory correction unit (400), a parameter generation unit (500), and a mapping update unit (600). The data acquisition unit (100) is used to acquire the electrochemical response data sequence and process event sequence of the target batch; The segmentation unit (200) is used to divide the stage response segments according to the process stage identifier and to correct the stage response segments according to the stage boundary offset. The fingerprint extraction unit (300) is used to extract the electrochemical fingerprint state vector of the stage response fragment and write it into the preparation state trajectory in the order of the process stages; A trajectory correction unit (400) is used to perform phase correction on the prepared state trajectory based on stage event anchor points and response morphology anchor points; The parameter generation unit (500) is used to generate real-time control parameters based on the trajectory deviation object and the parameter coupling constraint relationship; The mapping update unit (600) is used to update the correction mapping relationship according to the feedback residual type.

9. An electrolyte preparation control system based on electrochemical fingerprint feature extraction and trajectory correction, employing the electrolyte preparation control method based on electrochemical fingerprint feature extraction and trajectory correction as described in any one of claims 1 to 7, characterized in that: It also includes a data acquisition unit (100) connected to the preparation process data interface, a fragment division unit (200) connected to the data acquisition unit (100), a fingerprint extraction unit (300) connected to the fragment division unit (200), a trajectory correction unit (400) connected to the fingerprint extraction unit (300) and the target trajectory library respectively, a parameter generation unit (500) connected to the trajectory correction unit (400) and the correction mapping relation library respectively, and a mapping update unit (600) connected to the feedback data interface, the target trajectory library and the correction mapping relation library respectively; The parameter generation unit (500) writes the real-time control parameters into the control parameter output interface, and the mapping update unit (600) updates the phase correction parameters, electrochemical fingerprint state vector reference, parameter coupling constraint relationship or mapping verification identifier according to the feedback residual type.

Images