A high-entropy alloy electrode production management method and system based on quality traceability

Abstract

The application discloses a high-entropy alloy electrode production management method and system based on quality tracing, and belongs to the technical field of production management, and specifically comprises the following steps: giving an identification to an electrode production batch and collecting a process parameter curve; dividing application scenarios and storing data in categories by associating the batch identification with terminal product models; obtaining the statistical failure life of the terminal product and extracting relevant process curves to form an analysis sample set; extracting the quantitative fluctuation characteristics of the process curve and performing correlation analysis with the corresponding batch life; defining the process feature value interval positively correlated with long life in a specific scenario and generating a corresponding production control parameter set; and matching the application scenario according to the order product model, calling the corresponding parameter set to perform directional production. Through the closed-loop tracing and correlation analysis between process data and terminal service performance, the application improves the performance adaptability of high-entropy alloy electrodes in subdivided applications and the accuracy of production management and control.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of production management, and in particular to a high-entropy alloy electrode production management method and system based on quality traceability. BACKGROUND

[0002] As a new type of electrode material, high-entropy alloy electrodes have shown great potential in specific energy, cycle stability and high-temperature resistance due to the high-entropy effect, lattice distortion and sluggish diffusion effect brought by the multi-component design. They are widely used in high-performance batteries, water electrolysis hydrogen production and other new energy fields. The superior performance of high-entropy alloy electrodes is highly dependent on the complex multi-component design and precise preparation process. However, the multi-component characteristics also bring challenges in quality control, such as composition segregation and complex phase transformation behavior. The stability of the whole process from smelting, heat treatment to forming directly determines the performance consistency of the final product. Therefore, establishing an effective quality traceability system is crucial to ensure the reliability of high-entropy alloy electrodes.

[0003] The existing technical solutions usually focus on the compliance of production parameters and the integrity of data chain. Specifically, key process parameters such as melting temperature, holding time and rolling rate are collected by sensors, and these parameters are bound to specific production batches or units to form production history files. When a product has quality problems, the files can be used for reverse traceability to locate the production link or raw material batch that may have deviated, thereby isolating and correcting the problem and providing basic data support for the standardization of the production process and problem attribution.

[0004] However, the above-mentioned existing technical solutions still have obvious limitations. The core concern is to ensure that the production parameters meet the preset general process window and to establish a one-way traceability chain, but it fails to consider the different service conditions and individualized performance requirements of high-entropy alloy electrodes in different terminal application scenarios. For example, electrodes used in electric vehicle fast-charging batteries need to withstand extremely high instantaneous current density, while electrodes used in power grid energy storage emphasize long-cycle cycle capacity retention. The core failure modes and performance optimization directions of the two are fundamentally different. The existing traceability method can only determine whether the product is "production qualified", but cannot correlate and analyze the vast amount of actual process parameter data with the long-term service performance of the electrode in a specific harsh scenario. Therefore, based on this traceability system, it is difficult to identify the most critical process characteristics for a specific application scenario and to optimize the production process accordingly, leading to the solidification of production process parameters and the difficulty in dynamically adapting to diversified high-end application requirements, which restricts the performance competitiveness of the product in the sub-market. SUMMARY

[0005] The application aims to provide a high-entropy alloy electrode production management method and system based on quality traceability, and solve the following technical problems:

[0006] The prior art does not correlate the actual process fluctuation data in the production process with the long-term service performance of the high-entropy alloy electrode in a specific terminal application scenario, resulting in a lack of scenario-specificity in production process optimization and difficulty in dynamically adapting to diversified downstream performance requirements.

[0007] The purpose of the application can be achieved by the following technical solutions:

[0008] A high-entropy alloy electrode production management method based on quality traceability, comprising the following steps:

[0009] S1: Assign a batch identifier to each production batch of high-entropy alloy electrodes, and collect process parameters during production to form a process parameter curve indexed by the batch identifier;

[0010] S2: Obtain the terminal product model assembled with the batch identifier, and correlate the batch identifier with the obtained terminal product model, divide the application scenario according to the terminal product model, and store the process parameter curve according to the application scenario;

[0011] S3: Obtain the statistical failure life of the electrode batch corresponding to the terminal product model, and extract the process parameter curve of the related batch from the corresponding scene according to the terminal product model to form an analysis sample set;

[0012] S4: Extract the quantitative fluctuation feature from the process parameter curve of the analysis sample set, and correlate the quantitative fluctuation feature with the statistical failure life of the corresponding batch;

[0013] S5: According to the correlation analysis, define the numerical interval in which the quantitative fluctuation feature and the statistical failure life in the application scenario are correlated as the optimized process feature, and generate the corresponding production control parameter set according to the optimized process feature;

[0014] S6: Determine the application scenario according to the terminal product model specified by the production order, and call the production control parameter set corresponding to the application scenario to execute production.

[0015] As a further scheme of the application: in S1, the specific process of forming the process parameter curve indexed by the batch identifier is:

[0016] Through the data acquisition interface deployed in each production unit, the original working condition signal of the production equipment during the corresponding batch production period is obtained in real time; the original working condition signal is subjected to analog-digital conversion and time stamp marking to generate discrete process parameter data points with time sequence;

[0017] The data points are equally spaced re-sampled and smoothed according to a preset sampling period to form a continuous process parameter data sequence; the data sequence is bound with a batch identifier assigned to a corresponding production batch to obtain a process parameter curve with the batch identifier as a primary key index.

[0018] As a further scheme of the present application: in S3, the specific process of statistical failure life is:

[0019] An inventory provided by a downstream manufacturer is received, which contains terminal product serial numbers and corresponding models; a production start date and an end-of-life date corresponding to the terminal product serial number are read; a difference between the end-of-life date and the production start date is calculated to obtain the failure life of the terminal product;

[0020] According to the assembly association between the terminal product serial number and the electrode batch identifier, the failure life of the terminal product is classified into the corresponding electrode batch identifier; the failure lives of all terminal products belonging to the same electrode batch identifier are arithmetically averaged to obtain the statistical failure life of the electrode batch identifier; an association record of the electrode batch identifier, the terminal product model and the statistical failure life is established.

[0021] As a further scheme of the present application: in S4, the specific acquisition process of quantified fluctuation characteristics is:

[0022] Each process parameter curve in the analysis sample set is read; the process parameter curve is divided into time windows with a preset time length as an interval; in each time window, a corresponding window-in process parameter data segment is intercepted, and the mean value, maximum value and minimum value of each window-in process parameter data segment are calculated; the time sequence identifier of the time window, the mean value, the maximum value and the minimum value are combined into a four-tuple data structure; all four-tuple data structures corresponding to one process parameter curve are sorted according to the time sequence identifier to form the quantified fluctuation characteristics of the process parameter curve.

[0023] As a further scheme of the present application: in S4, the specific process of associating and analyzing the quantified fluctuation characteristics with the statistical failure life of the corresponding batch is:

[0024] The quantified fluctuation characteristics of each electrode batch identifier are taken as input variables, and the statistical failure life of the batch identifier is taken as an output variable to form a data pair set for association analysis; the values of all quantified fluctuation characteristics in the data pair set are standardized.

[0025] The linear regression model is used to perform correlation analysis on the quantified fluctuation characteristics after the standardization and the statistical failure life, to obtain regression coefficients and p values corresponding to the mean value, maximum value and minimum value at each time sequence position in the quantified fluctuation characteristics; the mean value, maximum value or minimum value and the time sequence position thereof with a p value lower than a preset correlation threshold are screened out, and the time sequence positions are merged into continuous time sequence intervals according to the time axis continuity; and the regression coefficients corresponding to the screened mean value, maximum value or minimum value in the time sequence interval and the time sequence interval are recorded as an association list.

[0026] As a further scheme of the present application, in S5, the specific acquisition process of the optimized process characteristic is:

[0027] According to the association list, the time sequence intervals with positive regression coefficients of each electrode batch identifier and the corresponding mean value, maximum value and minimum value are screened out.

[0028] For each time point in each time sequence interval, electrode batch identifiers with positive correlation regression coefficients at the time point are screened out from the analysis sample set, and one with the maximum statistical failure life value is selected from the batch identifiers;

[0029] The mean value, maximum value and minimum value of the process parameter curve of the selected electrode batch identifier at the corresponding time point are extracted as the reference mean value, reference maximum value and reference minimum value of the time point, respectively; a reference mean value sequence is formed based on the reference mean values of all time points, a reference maximum value sequence is formed based on the reference maximum values of all time points, and a reference minimum value sequence is formed based on the reference minimum values of all time points;

[0030] The mean value optimization interval boundary value of each time point in the time sequence interval is determined according to the reference mean value sequence, the maximum value optimization upper limit reference value of each time point on the time axis is determined according to the reference maximum value sequence, and the minimum value optimization lower limit reference value of each time point on the time axis is determined according to the reference minimum value sequence;

[0031] The mean value optimization interval boundary value, the maximum value optimization upper limit reference value and the minimum value optimization lower limit reference value are summarized to obtain the optimized process characteristic.

[0032] As a further scheme of the present application, in S5, the specific process of the production control parameter set is:

[0033] The mean value optimization interval boundary value, the maximum value optimization upper limit reference value and the minimum value optimization lower limit reference value recorded in the optimized process characteristic are read; and the mean value optimization interval boundary value is calibrated as the target process parameter of the corresponding time point.

[0034] The maximum value optimization upper limit reference value is directly set as the upper boundary of the process parameter operation range corresponding to the time point; and the minimum value optimization lower limit reference value is directly set as the lower boundary of the process parameter operation range corresponding to the time point.

[0035] The target process parameter, the upper and lower boundaries of the process parameter operation range jointly form a closed-loop control parameter group at the corresponding time point; and the closed-loop control parameter groups at each time point on the time axis are arranged and packaged in time sequence to form the production control parameter set bound with the application scene.

[0036] A high-entropy alloy electrode production management system based on quality traceability, used for the high-entropy alloy electrode production management method based on quality traceability, comprising:

[0037] A data acquisition module is configured to assign a batch identifier to each production batch of the high-entropy alloy electrode, and collect process parameters in the production process to form a process parameter curve indexed by the batch identifier.

[0038] A scene classification module is configured to obtain an end product model assembled by the batch identifier, associate the batch identifier with the obtained end product model, divide the application scene according to the end product model, and store the process parameter curve according to the application scene.

[0039] A sample acquisition module is configured to obtain the statistical failure life of the electrode batch corresponding to the end product model, and extract the process parameter curve of the related batch from the corresponding scene according to the end product model to form an analysis sample set.

[0040] A data analysis module is configured to extract quantitative fluctuation features from the process parameter curve of the analysis sample set, and perform correlation analysis on the quantitative fluctuation features and the statistical failure life of the corresponding batch.

[0041] A parameter generation module is configured to define a numerical interval in which the quantitative fluctuation features and the statistical failure life in the application scene are correlated according to the correlation analysis and as an optimized process feature, and generate a corresponding production control parameter set according to the optimized process feature.

[0042] A production control module is configured to determine the application scene according to the end product model specified by the production order, and call the production control parameter set corresponding to the application scene to execute production.

[0043] The beneficial effects of the present application are as follows:

[0044] 1) The present application realizes the paradigm shift of quality traceability from process monitoring to performance attribution by establishing the full life cycle data association based on application scenario grouping. The present application first systematically associates the electrode production batch with the terminal product model, so that the process data is stored according to the specific application scenario; at the same time, by integrating the actual failure life data of the terminal product, a complete data link of "process parameter-scene label-service result" is constructed. Based on this link, massive historical process data and corresponding long-term service performance can be statistically associated and analyzed under the same application scenario dimension, so as to identify the key process characteristics and fluctuation patterns that determine the product life in a specific scene. This enables enterprises to convert production records that are only used to determine whether they are qualified into data assets for evaluating and predicting the actual durability of products in the target market, providing an empirical-based accurate direction for process optimization.

[0045] 2) The present application realizes the explicitization and standardization of implicit process knowledge by analyzing the mapping relationship between parameter fluctuation within the qualified process window and service performance. It can be understood that the final performance (such as fatigue life, electrical stability) of high-entropy alloy electrode is not only determined by the average value of process parameters, but also affected by the dynamic fluctuation characteristics of parameters around the target value. These fluctuations are directly related to the microscopic kinetic conditions in the melting, solidification and heat treatment processes, for example, the instantaneous fluctuation of temperature will affect the diffusion rate and phase change path of elements, and then change the key microstructure such as grain size, phase composition that determines the service performance. The present application converts the continuous process parameter curve into a quantifiable analysis of fluctuation characteristic sequence, and uses statistical models to establish the correlation model between these characteristics and batch failure life, which is essentially a data-driven reverse analysis of the causal chain of "process fluctuation-microstructure-macro performance". Through the correlation model, specific fluctuation value intervals that are significantly related to excellent service performance can be identified from numerous qualified productions. Further, by integrating the beneficial fluctuation performance verified by multiple long-life batches in each key process period, a "process characteristic template" representing the optimal microstructure in this application scenario is refined and generated, thereby determining the optimal process production parameters for different application scenarios.

[0046] 3) The application realizes the autonomous evolution of the production system from static execution to dynamic adaptation by constructing the automatic mapping and closed-loop updating mechanism of "scene demand-optimized process-control parameter". The application first reversely decomposes the "process feature template" into a closed-loop control parameter set driving specific production equipment, and forms a strong binding with the application scene. When a new production order indicates the product model, the corresponding control parameter set can be automatically called to execute production, thereby realizing scene-oriented targeted manufacturing. More importantly, by continuously collecting the service feedback data of the products produced according to the new parameters in the market, the associated model and process feature template can be periodically updated, and then the control parameter set is iteratively optimized. Thus, a complete data closed loop from "service feedback-data analysis-process optimization-production execution-new round of feedback" is formed, which drives the manufacturing system to be able to adapt to different high-end demands and continuously evolve to a higher scene adaptation degree. BRIEF DESCRIPTION OF DRAWINGS

[0047] The application will be further described below in conjunction with the accompanying drawings.

[0048] Figure 1 is a high-entropy alloy electrode production management method process schematic diagram based on quality traceability.

[0049] Figure 2 is a high-entropy alloy electrode production management system structure schematic diagram based on quality traceability. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the application will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0051] Please refer to Figure 1 The application is a high-entropy alloy electrode production management method based on quality traceability, which comprises the following steps:

[0052] S1: Give each production batch of high-entropy alloy electrode a batch identifier, and collect the process parameters in the production process to form a process parameter curve indexed by the batch identifier;

[0053] First, a globally unique batch identification code (e.g., "HEA-LOT-2023-08-001") is automatically generated by the production management system at the beginning of each batch of high-entropy alloy electrodes. During the production process, a high-precision sensor network deployed at key processes (such as vacuum melting furnaces, homogenization heat treatment furnaces, and rolling production lines) continuously collects process parameters in real time. For example, in the case of vacuum melting, temperature sensors record the melt temperature several times per second, pressure sensors record the vacuum degree in the furnace, and mass spectrometers record the key gas partial pressure. These heterogeneous time series data from different devices are received, time-synchronized, and standardized in format through a unified Internet of Things gateway. Subsequently, the system aligns and fuses multi-dimensional data (temperature, pressure, etc.) within the same batch and the same time window to form a complete "process parameter curve" with time as the horizontal axis and multiple parameter dimensions. Finally, this curve is strongly bound with the unique identification code of the batch, and stored in a time series database to form a traceable original data archive.

[0054] S2: Obtain the terminal product model assembled with the batch identification, and associate the batch identification with the obtained terminal product model, divide the application scenarios according to the terminal product model, and store the process parameter curve according to the application scenarios;

[0055] When the batch of electrodes is delivered to downstream customers (such as battery manufacturers), the customers will scan or record the batch identification code on their assembly lines. The customer's production execution system (MES) will associate and register the electrode batch identification code with the unique serial number of the terminal product being assembled (such as "Model NEX-P100 fast-charging power battery pack"), and synchronize this association relationship back to the electrode manufacturer's system through a secure interface. The electrode manufacturer's system automatically determines the application scenario to which the terminal product model belongs according to a predefined rule library (such as "all battery packs with model code containing 'P' and rated charge rate ≥ 3C are classified as'super-fast charging application scenario'"). After determination, the system moves or logically associates all process parameter curves under the batch identification code from the original database to the corresponding application scenario-specific data partition for storage. For example, all electrode batch curves for "super-fast charging scenario" battery packs are stored in the same analysis library, thereby laying the foundation for subsequent scenario-based data analysis.

[0056] S3: Obtain the electrode batch statistical failure life corresponding to the terminal product model, and extract the process parameter curves of the related batches from the corresponding scenario according to the terminal product model to form an analysis sample set;

[0057] After a certain market usage cycle, the electrode manufacturer obtains quality feedback data from downstream customers or after-sales channels. This data is usually provided in the form of a list, containing the terminal product model, serial number, and the production batch identification of the electrode installed in it, which is obtained through disassembly analysis or data statistical backtracking, and the average cycle life (i.e. statistical failure life, such as "Batch HEA-LOT-2023-08-001 electrode, in NEX-P100 battery pack, average cycle 1200 times, capacity decay to 80%") exhibited by the electrode in real vehicle or test. After receiving this list, first determine the application scenario of the terminal product (such as "ultra-fast charging scenario") according to the terminal product model. Then, from the exclusive data partition of this scenario, accurately retrieve and extract the complete process parameter curve corresponding to all electrode batch identifiers listed in the list. Finally, pair each batch's process parameter curve with its corresponding statistical failure life value to form a structured analysis sample set. Each sample in the sample set clearly shows the correspondence between "a certain fluctuation pattern of production process" and "its actual durability result in a certain scenario".

[0058] S4: Extract the quantitative fluctuation features from the process parameter curves in the analysis sample set, and perform correlation analysis on the quantitative fluctuation features and the statistical failure life of the corresponding batch;

[0059] Read each process parameter curve (such as smelting temperature curve) in the analysis sample set. First, pre-process the curve, then use a sliding time window algorithm to divide the continuous curve into a series of overlapping segments. Within each time window segment, calculate the mean value (representing the average level), maximum value and minimum value (representing the fluctuation amplitude) of the process parameter values in the segment, and bind these three statistics with the center time point of the window to form a feature vector that describes the fluctuation state at that time. After traversing the entire curve, a quantitative fluctuation feature sequence composed of time series feature vectors is obtained. Then, using multivariate statistical analysis methods (such as multiple linear regression), the fluctuation feature sequences of all samples are used as the independent variable matrix, and the corresponding statistical failure life is used as the dependent variable vector, to perform modeling analysis. This analysis aims to calculate the regression coefficients and their statistical significance (p-value) between the mean, maximum and minimum values at each time point in the fluctuation feature sequence and the life, so as to identify which specific process period and which fluctuation features of the parameters (such as "average temperature at 300 seconds" and "maximum vacuum at 450 seconds") have a stable and significant positive or negative correlation with the long life of the electrode.

[0060] S5: According to the correlation analysis, the numerical interval of the correlation between the quantized fluctuation characteristics and the statistical failure life in the application scenario is defined as an optimized process feature, and a corresponding production control parameter set is generated according to the optimized process feature;

[0061] Based on the results of the S4 correlation analysis (i.e. regression coefficient and p value), the feature points with high significance of positive correlation are screened out. For each such feature point (for example, "average temperature at 300 seconds in the melting stage"), all electrode batches that show a favorable value for long life at this feature point are found from the analysis sample set. By statistically analyzing the distribution of the actual parameter values of these "top student" batches at this feature point, an optimal numerical interval (for example, "1550°C-1565°C") is determined. All the screened out key feature points scattered in different process periods and their corresponding optimal numerical intervals are summarized, which form a set of optimized process feature templates for this application scenario. It can be understood that the optimized process feature template describes "in order to obtain long life in this scenario, the production curve should be guided to which numerical range at which key control points". Next, the process engineer converts this feature template into an executable file. For example, the requirement of "average temperature at 300 seconds in the range of 1550-1565°C" is converted into specific parameters for the PID control loop setting value, temperature rise rate limit and overshoot tolerance range of the melting furnace at the corresponding time period. All these device control parameters adjusted for the reproduction of the optimized features are packaged into a production control parameter set bound to this application scenario.

[0062] S6: According to the end product model specified by the production order, the application scenario is determined, and the production control parameter set corresponding to the application scenario is called to execute the production.

[0063] When a new production order is received, the order specifies explicitly the end product model to which the produced electrode will be used (such as "for NEX-P200 battery pack"). The production planning system automatically analyzes the model and matches its application scenario according to the predefined rules (such as matching to "super-fast charging scenario"). Then, the production instruction is issued to the manufacturing execution system (MES), and the production control parameter set bound to the scenario is automatically called from the parameter library. The MES downloads the parameter set to the production line controller (such as the melting furnace PLC and the heat treatment furnace DCS) of the corresponding workshop. The production equipment will run according to this set of optimized parameters, rather than the standard general parameters, so as to produce high-entropy alloy electrodes that are pre-adapted to the performance requirements of the target application scenario in terms of process genes.

[0064] It is worth noting that within the qualified process window, different batches of high-entropy alloy electrodes will inevitably have differentiated process parameter fluctuations, which are rooted in the subtle differences in the physical properties of raw materials, the time-varying characteristics of the dynamic response of production equipment, and the nonlinear coupling between multiple process parameters. These fluctuations are not meaningless noise, but rather key input signals that directly interfere with the flow of the melt, nucleation during solidification, atomic diffusion, and phase transformation kinetics. Different fluctuation patterns (such as the smoothness of the temperature curve and the stability of the key interval) will "write" different microstructure characteristics such as grain size, composition segregation degree, and second phase distribution, thereby fundamentally determining the differences in the final electrochemical performance and service life of the electrode. Based on this principle, the present application builds a data closed loop of "process parameter curve-application scenario-statistical failure life", systematically analyzes the statistical correlation between different fluctuation patterns within the qualified window and the terminal service performance, identifies "beneficial fluctuation characteristics" that positively contribute to long life in specific scenarios, and finally generates scenario-based production control parameter sets, transforming the traditionally passive constrained process window into an intelligent control target that can actively guide and reproduce optimized fluctuation patterns, thereby realizing a paradigm shift from producing "statistically qualified products" to manufacturing "scenario-adapted optimal products".

[0065] In a preferred embodiment of the present application, in S1, the specific process of forming the process parameter curve indexed by the batch identifier is:

[0066] Through the data acquisition interface deployed in each production unit, the original working condition signals of the production equipment during the corresponding batch production period are obtained in real time; the original working condition signals are subjected to analog-digital conversion and time stamp marking to generate discrete process parameter data points with time series;

[0067] The data points are subjected to equal-interval resampling and smoothing filter processing according to a preset sampling period to form a continuous process parameter data sequence; the data sequence is bound with the batch identifier assigned to the corresponding production batch to obtain a process parameter curve indexed by the batch identifier as the primary key.

[0068] In another preferred embodiment of the present application, in S3, the specific process of statistical failure life is:

[0069] An inventory provided by a downstream manufacturer containing terminal product serial numbers and their corresponding models is received; the production activation date and the final failure date corresponding to the terminal product serial number are read; the difference between the final failure date and the production activation date is calculated to obtain the failure life of the terminal product;

[0070] According to the assembly association of the terminal product serial number and the electrode batch identifier, the failure life of the terminal product is classified to the corresponding electrode batch identifier; the failure life of all terminal products belonging to the same electrode batch identifier is arithmetically averaged to obtain the statistical failure life of the electrode batch identifier; and the association record of the electrode batch identifier, the terminal product model and the statistical failure life is established.

[0071] Firstly, the system receives an electronic list provided by a downstream manufacturer (for example, an electric vehicle company) periodically or event-triggered, and the core content of the list is the unique serial number of the terminal product (such as a batch of power battery packs) that has been confirmed to be failed or retired and the corresponding product model; after reading the serial number, the system automatically initiates a query to the source, i.e. the product lifecycle management database of the downstream manufacturer, to obtain two key timestamps corresponding to each serial number: one is the “production start date” when the product is first put into use after leaving the production line, which marks the beginning of its life counting; the other is the “final failure date” when the product is withdrawn from service due to performance degradation to the threshold or failure, which marks the end of its life counting. The system calculates the difference between the two dates, and the principle is that the time interval directly corresponds to the actual length of time the product has experienced from being put into use to losing functionality, thereby obtaining the specific failure life data of the terminal product. Subsequently, the system needs to associate these scattered product life data with the upstream electrode production batch, and it is based on the “assembly association table” recorded and provided by the downstream manufacturer at the assembly link, which explicitly records the specific high-entropy alloy electrode batch identifier used by each terminal product serial number during production and assembly; by matching this table, the system can accurately collect the failure life value of each terminal product under the production batch identifier that provides the electrode raw material for it. Then, for all terminal product failure lives belonging to the same electrode batch identifier, the system performs an arithmetic average calculation, and the principle is to integrate the performance of the batch electrode in multiple terminal products by taking the average value, thereby obtaining a single, stable statistical characteristic value, i.e. “statistical failure life”, which can represent the overall life level of the batch. Finally, the system establishes and saves the statistical failure life result calculated together with the batch identifier and the terminal product model information as an association record, which builds a complete index foundation for subsequent data analysis according to products and scenarios.

[0072] The fragmented and individualized product failure information in the downstream market is converted into structured and batched performance characterization data in the upstream production analysis. By averaging the life of multiple terminal products, the data noise caused by extreme abnormal or accidental failure of individual products can be effectively smoothed out, thereby refining the statistical characteristics that more truly reflect the quality and durability level of the electrode batch itself. This step is a bridge connecting "market actual performance" and "internal process parameters", which converts the abstract "durable" or "early decay" market reputation into a numerical index that can be quantitatively correlated with the specific production process curve. Only after this conversion, the subsequent step of analyzing the impact of process fluctuations on life has a solid real basis, and the entire method finally realizes the purpose of "optimizing production process according to service performance", otherwise the optimization will lose direction and lack of empirical support.

[0073] In another preferred embodiment of the present application, the specific acquisition process of quantifying the fluctuation characteristics in S4 is:

[0074] Reading each process parameter curve in the analysis sample set; dividing the process parameter curve into time windows with a preset time interval; in each time window, intercepting the corresponding window process parameter data segment, calculating the mean, maximum and minimum of each window process parameter data segment; combining the time sequence identifier of the time window, the mean, the maximum and the minimum into a four-tuple data structure; sorting all four-tuple data structures corresponding to a process parameter curve according to the time sequence identifier to form the quantified fluctuation characteristics of the process parameter curve.

[0075] First, the analysis sample set is read, which means that it calls up the complete process parameter records of all relevant electrode batches in a specific application scenario from the classified stored data, for example, the continuous temperature change data of a batch of electrodes for super-fast charging batteries in the vacuum melting link; then, instead of directly processing the entire long curve, the system divides the continuous time curve into a series of time windows with a fixed time interval (for example, every thirty seconds), the principle of which is to discretize the continuous process into comparable standardized segments so that subsequent analysis can be located to specific process stages rather than the whole; in each defined time window, the system will intercept the process parameter data segment corresponding to the time period, such as a thirty-second temperature reading set, then calculate the arithmetic mean of all values in this data segment to represent the overall level of temperature in this period, and find the maximum (maximum) and minimum (minimum) values in this data segment to capture the extreme fluctuations of temperature in this period, the reason for calculating these three values is that the average value reflects the steady state, while the combination of maximum and minimum values clearly defines the range of fluctuations in this period, together describing the core statistical characteristics of the window period; after the calculation is completed, the system identifies the position of the time window in the overall production time sequence (i.e. time sequence identifier, such as representing the 300th to 330th second after the start of production), combines the four elements of the calculated mean, maximum and minimum values together to form a structured data unit (i.e. four-tuple data structure), the principle of this combination is to bind "when it happens" and "how the fluctuations are", ensuring that each feature point has a clear time attribute; finally, the system sorts and concatenates all such four-tuple data units corresponding to the same process parameter curve in chronological order, the principle of which is to restore the original occurrence sequence of these feature points on the real production time axis, thus forming a trajectory composed of discrete feature points that can represent the fluctuation pattern of the original continuous curve. This trajectory is the quantitative fluctuation characteristic of the process parameter curve, which converts a continuous and difficult to directly calculate curve into a series of time-ordered discrete point sets representing local statistical characteristics.

[0076] Through the division of fixed windows and the extraction of statistical quantities, the data volume is greatly compressed, the "dimension disaster" is overcome, the core information capable of representing the process stability and the fluctuation amplitude is effectively extracted and retained, and through the binding of the time sequence identifier, the time position information of the fluctuation characteristics in the production process is completely retained, which is crucial for identifying the key process window. This approach directly serves the ultimate purpose of the scheme, as it creates a common language that translates "physical fluctuations of the process" into "computable numerical features". Only by completing this translation can the system use statistical tools to discover the hidden and complex correlation patterns between these fluctuation numerical features and another key number (statistical failure life), thereby ultimately realizing the extraction of key process knowledge that determines the long-term performance of the product from massive production data, and providing clear and quantitative targets for actively controlling process fluctuations.

[0077] In another preferred embodiment of the present application, in S4, the specific process of correlating analysis of the quantified fluctuation features and the statistical failure life of the corresponding batch is:

[0078] The quantified fluctuation features identified by each electrode batch are used as input variables, and the statistical failure life identified by the batch is used as an output variable to form a data pair set for correlation analysis; the numerical values of all quantified fluctuation features in the data pair set are standardized;

[0079] Using a linear regression model, the quantified fluctuation features after standardization and the statistical failure life are subjected to correlation analysis, and the regression coefficients and p values corresponding to the mean, maximum and minimum values at each time sequence position in the feature sequence are obtained; the mean, maximum or minimum values and their time sequence positions whose p values are lower than a preset correlation threshold are screened out, and the time sequence positions are merged into continuous time sequence intervals according to the time axis continuity; the regression coefficients corresponding to the screened mean, maximum or minimum values in the time sequence interval and the time sequence interval are recorded together as an association relationship list.

[0080] Firstly, each batch of electrodes is treated as an independent analysis sample, and the complete feature sequence (i.e. all four-tuple data arranged in chronological order) converted from the process parameter curve of the batch is taken as a set of input variables, while the single statistical failure life value calculated through downstream feedback of the batch is taken as the corresponding output variable. The two are paired, and the principle is to establish a corresponding relationship between the "cause" (process feature) and the "effect" (life performance), thereby forming a data pair set for mathematical model analysis. Next, the system standardizes the values of all feature sequences in this set. The specific method is to calculate the overall mean and standard deviation of all values at each time sequence position (e.g. the mean of the "300th second window" of all batches), and then subtract the mean from each original value and divide by the standard deviation. The principle of this is to eliminate the incomparability between different process parameters (such as temperature and vacuum) due to different dimensions and absolute value sizes, and convert all feature data to a standard scale centered on zero with a standard deviation of 1, so that subsequent analysis can fairly measure the relative importance of each feature on life without mistakenly amplifying the influence of a parameter value simply because it is large. Subsequently, the system uses a linear regression model to analyze this standardized data set. The principle of this model is to try to find a set of weight coefficients (i.e. regression coefficients) that can best predict the output statistical failure life through linear weighted combination of the input feature sequence. By solving this model, the system calculates a regression coefficient and a corresponding p-value for each specific statistical quantity (such as "mean at 300th second" or "maximum at 300th second") at each specific time sequence position in the feature sequence. The sign and size of the regression coefficient indicate whether the feature is positively or negatively correlated with life and the strength of the correlation, while the p-value is used to assess whether this correlation is statistically significant and not just a coincidence. Based on the p-value, the system sets a threshold for judging significance (e.g. 0.05), and automatically screen out those feature points with p-value lower than this threshold, which means that these feature points are extremely unlikely to be random errors, and are "signals" worthy of attention; then, the system will check the continuity of these screened-out significant feature points scattered on the time axis according to their positions on the original production time axis, and merge the points with adjacent or close positions together to form several continuous time intervals (for example, "280th second to 320th second"), and the principle of doing so is that the influence of the production process is usually a physical and chemical process that lasts for a period of time, rather than an instantaneous action, and the merged intervals are more in line with engineering practice and are convenient for subsequent control strategy making; finally, the system will create a correlation list that not only records each identified significant time interval, but also explicitly lists which type of statistical quantity (mean value, maximum value or minimum value) in the interval is significant, and attaches the corresponding regression coefficient, thereby completely recording "which fluctuation trend of which process feature at what time has a statistically significant and directionally clear correlation with the life".

[0081] In another preferred embodiment of the present application, in the S5, the specific acquisition process of the optimized process feature is:

[0082] According to the correlation list, screen out the time intervals with positive regression coefficients of the electrode batch identifiers and the corresponding mean values, maximum values and minimum values;

[0083] For each time point in each time interval, screen out the electrode batch identifiers with positive correlation regression coefficients at the time point from the analysis sample set, and select one with the largest statistical failure life value from these batch identifiers;

[0084] Extract the mean value, maximum value and minimum value of the process parameter curve of the selected electrode batch identifier at the corresponding time point as the reference mean value, reference maximum value and reference minimum value of the time point respectively; form a reference mean value sequence based on the reference mean values of all time points, a reference maximum value sequence based on the reference maximum values of all time points, and a reference minimum value sequence based on the reference minimum values of all time points;

[0085] Determine the mean value optimization interval boundary value of each time point in the time interval according to the reference mean value sequence, determine the maximum value optimization upper limit reference value of each time point on the time axis according to the reference maximum value sequence, and determine the minimum value optimization lower limit reference value of each time point on the time axis according to the reference minimum value sequence;

[0086] Summarize the mean value optimization interval boundary value, the maximum value optimization upper limit reference value and the minimum value optimization lower limit reference value to obtain the optimized process feature.

[0087] First, according to the correlation list generated in the previous step, all entries with positive regression coefficients are filtered out, the principle being that a positive regression coefficient indicates that the process feature at that point (such as the average temperature at a certain time point) is positively correlated with the statistical failure life, that is, the larger the feature value, the longer the life tends to be. After filtering, a series of process feature points that positively contribute to life are obtained, which are scattered in different electrode batches and at different time points. Next, in order to construct a complete, continuous and optimal process feature template, an optimal reference value needs to be determined at each key time point. The specific method is as follows: for each specific time point (e.g. 305 seconds) in each positive correlation time interval identified in the list, the sample set is retrieved and analyzed to find all electrode batch identifiers with positive regression coefficients at that time point, which means that the process performance of these batches at that point is confirmed to be related to long life. Then, from these candidate batches, the batch identifier with the largest actual statistical failure life value is selected, the principle being that selecting the "longest" batch as a benchmark can be considered as the "best" among all "good students" and is the most representative practice. Subsequently, the system extracts the mean, maximum and minimum values calculated at the specific time point from the original process parameter curve corresponding to the selected optimal batch identifier, and records these three values as the reference mean value, reference maximum value and reference minimum value at that time point. The principle of this is to capture the typical level (mean) and fluctuation range (maximum and minimum) of the optimal batch at that moment, providing complete information for setting control targets. The system repeats the "filtering-selection-extraction" operation at all covered time points and connects the reference mean values obtained at each time point in chronological order to form a "reference mean value sequence" describing the ideal average level. Similarly, the reference maximum values and reference minimum values are connected to form a "reference maximum value sequence" and a "reference minimum value sequence". Then, based on these reference sequences, specific optimization targets are set: for the mean value, the system sets a reasonable floating range as the "mean value optimization interval boundary value" above and below the reference mean value sequence, the principle being to allow production to fluctuate within a small range around the optimal mean value to balance controllability; for the maximum and minimum values, the system directly uses the values of the reference maximum value sequence and the reference minimum value sequence as the "maximum value optimization upper limit reference value" and the "minimum value optimization lower limit reference value", the principle being to explicitly require that the instantaneous fluctuations in the production process should not exceed these upper and lower limit lines defined by the optimal batch practice. Finally, all time point mean value optimization interval boundary values, maximum value optimization upper limit reference values and minimum value optimization lower limit reference values on the entire production time axis are summarized and packaged, and the complete data set formed is the "optimized process feature", which defines an ideal process parameter fluctuation channel with a reasonable tolerance range.

[0088] The statistically found, fragmented "advantages" dispersed in different excellent batches are integrated and refined to construct a complete and executable "golden standard" process path. It avoids the risk of relying on the contingency or incomplete data of a single optimal batch, but selects the parameter value verified with long life and best in the long-life batch at each time point in a "selecting the best" way, thereby synthesizing a theoretically more robust and reliable optimization target. This approach plays a key role in realizing the ultimate purpose of the scheme: it is the key design link for converting "data insight" into "production instruction" that connects the objective laws found by correlation analysis and generates the control parameter set. By generating such an optimized process feature, a specific and quantitative chasing target is provided for the production line, so that "optimizing the process for a specific scenario" changes from an abstract concept to an engineering problem with clear numerical constraints, making subsequent precise control and production reproduction possible, and truly closing the intelligent cycle from service performance feedback to production process optimization.

[0089] In another preferred embodiment of the present application, the specific process of generating the production control parameter set in S5 is as follows:

[0090] reading the mean value optimization interval boundary value, the maximum value optimization upper limit reference value and the minimum value optimization lower limit reference value recorded in the optimized process feature; and setting the mean value optimization interval boundary value as the target process parameter of the corresponding time point;

[0091] setting the maximum value optimization upper limit reference value as the upper boundary of the process parameter operating range of the corresponding time point; and setting the minimum value optimization lower limit reference value as the lower boundary of the process parameter operating range of the corresponding time point;

[0092] the target process parameter, the upper and lower boundaries of the process parameter operating range jointly constitute the closed-loop control parameter group of the corresponding time point; and the closed-loop control parameter groups of the time points on the time axis are arranged and packaged in time sequence to form the production control parameter set bound with the application scenario.

[0093] Referring to Figure 2 The present application also includes a high-entropy alloy electrode production management system based on quality traceability, which is used to implement the above-mentioned high-entropy alloy electrode production management method based on quality traceability, and includes:

[0094] a data acquisition module, configured to assign a batch identifier to each production batch of the high-entropy alloy electrode, and collect process parameters in the production process to form a process parameter curve indexed by the batch identifier;

[0095] The scene classification module is configured to obtain a terminal product model assembled by the batch identifier, associate the batch identifier with the obtained terminal product model, divide application scenes according to the terminal product model, and store process parameter curves according to the application scenes;

[0096] The sample acquisition module is configured to obtain statistical failure life of an electrode batch corresponding to the terminal product model, extract process parameter curves of related batches from the corresponding scene according to the terminal product model, and form an analysis sample set;

[0097] The data analysis module is configured to extract quantified fluctuation features from the process parameter curves of the analysis sample set, and perform correlation analysis on the quantified fluctuation features and the statistical failure life of the corresponding batch;

[0098] The parameter generation module is configured to define a numerical interval in which the quantified fluctuation features and the statistical failure life are correlated in the application scene as an optimized process feature according to the correlation analysis, and generate a corresponding production control parameter set according to the optimized process feature;

[0099] The production control module is configured to determine an application scene according to a terminal product model specified by a production order, and execute production by calling a production control parameter set corresponding to the application scene.

[0100] The above describes one embodiment of the present application in detail, but the content is only a preferred embodiment of the present application and cannot be considered as limiting the scope of the present application. Any equivalent changes and improvements made according to the scope of the present application should still belong to the scope of the present application.

Claims

1. A high-entropy alloy electrode production management method based on quality traceability, characterized by, The method comprises the following steps: S1: Assign a batch identification to each production batch of high-entropy alloy electrodes, and collect process parameters in the production process to form a process parameter curve indexed by the batch identification; S2: Obtain the terminal product model assembled with the batch identification, and associate the batch identification with the obtained terminal product model, divide the application scenarios according to the terminal product model, and store the process parameter curve according to the application scenarios; S3: Obtain the statistical failure life of the electrode batch corresponding to the terminal product model, and extract the process parameter curve of the related batch from the corresponding scenario according to the terminal product model to form an analysis sample set; S4: Extract the quantized fluctuation feature from the process parameter curve in the analysis sample set, and perform correlation analysis on the quantized fluctuation feature and the statistical failure life of the corresponding batch; The specific acquisition process of the quantized fluctuation feature is as follows: Read each process parameter curve in the analysis sample set; divide the time window with a preset time interval for the process parameter curve; in each time window, intercept the corresponding window process parameter data segment, and calculate the mean, maximum and minimum of each window process parameter data segment; The time sequence identification of the time window, the mean, the maximum and the minimum are combined into a four-tuple data structure; Sort all four-tuple data structures corresponding to a process parameter curve according to the time sequence identification to form the quantized fluctuation feature of the process parameter curve; The specific process of correlation analysis between the quantized fluctuation feature and the statistical failure life of the corresponding batch is as follows: Take the quantized fluctuation feature of each electrode batch identification as the input variable, and take the statistical failure life of the batch identification as the output variable to form a data pair set for correlation analysis; standardize the values of all quantized fluctuation features in the data pair set; Using a linear regression model, the quantized fluctuation feature and the statistical failure life after standardization are correlated to obtain the regression coefficient and p value corresponding to each time sequence position in the quantized fluctuation feature; filter out the mean, maximum or minimum and its time sequence position whose p value is lower than the preset correlation threshold, and merge the time sequence positions into continuous time sequence intervals according to the time axis continuity; record the regression coefficient corresponding to the filtered mean, maximum or minimum in the time sequence interval and the time sequence interval as an association relationship list; S5: According to the correlation analysis, define the value interval in which the quantized fluctuation feature and the statistical failure life in the application scenario exist correlation as an optimized process feature, and generate a corresponding production control parameter set according to the optimized process feature; S6: Determine the application scenario according to the terminal product model specified by the production order, and call the production control parameter set corresponding to the application scenario to execute production.

2. The high-entropy alloy electrode production management method based on quality traceability according to claim 1, characterized by, In S1, the specific process of forming a process parameter curve indexed by a batch identification is as follows: The data acquisition interface deployed in each production unit acquires original working condition signals of the production equipment in the corresponding batch production period in real time; the original working condition signals are subjected to analog-digital conversion and time stamp marking to generate discrete process parameter data points with time sequence; The data points are subjected to equal-interval resampling and smoothing filtering processing according to a preset sampling period to form a continuous process parameter data sequence; The data sequence is bound with a batch identifier assigned to the corresponding production batch to obtain a process parameter curve with the batch identifier as a primary key index.

3. The high-entropy alloy electrode production management method based on quality traceability according to claim 1, characterized by, In S3, the specific process of statistical failure life is as follows: An inventory provided by a downstream manufacturer and containing terminal product serial numbers and corresponding models is received; the production start date and the final failure date corresponding to the terminal product serial number are read; the difference between the final failure date and the production start date is calculated to obtain the failure life of the terminal product; According to the assembly correlation between the terminal product serial number and the electrode batch identifier, the failure life of the terminal product is classified into the corresponding electrode batch identifier; the failure lives of all terminal products belonging to the same electrode batch identifier are subjected to arithmetic average calculation to obtain the statistical failure life of the electrode batch identifier; The correlation record of the electrode batch identifier, the terminal product model and the statistical failure life is established.

4. The high-entropy alloy electrode production management method based on quality traceability according to claim 1, characterized by, In S5, the specific acquisition process of the optimized process feature is as follows: According to the correlation list, the time sequence interval with positive regression coefficient and the corresponding mean value, maximum value and minimum value of each electrode batch identifier are screened; For each time point in each time sequence interval, the electrode batch identifiers with positive correlation regression coefficient at the time point are selected from the analysis sample set, and the electrode batch identifier with the maximum statistical failure life value is selected from the electrode batch identifiers; The mean value, maximum value and minimum value of the process parameter curve of the selected electrode batch identifier at the corresponding time point are extracted as the reference mean value, reference maximum value and reference minimum value of the time point, respectively; The reference mean value sequence is formed based on the reference mean values of all time points, the reference maximum value sequence is formed based on the reference maximum values of all time points, and the reference minimum value sequence is formed based on the reference minimum values of all time points; The mean value optimization interval boundary value of each time point in the time sequence interval is determined according to the reference mean value sequence, the maximum value optimization upper limit reference value of each time point on the time axis is determined according to the reference maximum value sequence, and the minimum value optimization lower limit reference value of each time point on the time axis is determined according to the reference minimum value sequence; The mean value optimization interval boundary value, the maximum value optimization upper limit reference value and the minimum value optimization lower limit reference value are summarized to obtain the optimized process feature.

5. The high-entropy alloy electrode production management method based on quality traceability according to claim 4, characterized by, In S5, the specific process of the production control parameter set is as follows: The mean value optimization interval boundary value, the maximum value optimization upper limit reference value and the minimum value optimization lower limit reference value recorded in the optimized process feature are read; the mean value optimization interval boundary value is calibrated as the target process parameter of the corresponding time point; The maximum value optimization upper limit reference value is directly set as the upper boundary of the process parameter operating range of the corresponding time point; The minimum optimization lower limit reference value is directly set as the lower boundary of the process parameter operation range at the corresponding time point; The target process parameter and the upper and lower boundaries of the process parameter operation range jointly form a closed-loop control parameter group at the corresponding time point; and the closed-loop control parameter groups at each time point on the time axis are arranged and packaged in time sequence to form the production control parameter set bound with the application scenario.

6. A high-entropy alloy electrode production management system based on quality traceability, for implementing a high-entropy alloy electrode production management method based on quality traceability according to any one of claims 1-5, characterized in that, Comprise: A data acquisition module is configured to assign a batch identifier to each production batch of high-entropy alloy electrodes and collect process parameters in the production process to form process parameter curves indexed by batch identifiers; A scenario classification module is configured to obtain the end product model equipped with the batch identifier, associate the batch identifier with the obtained end product model, divide the application scenarios according to the end product model, and store the process parameter curves according to the application scenarios; A sample acquisition module is configured to obtain the statistical failure life of the electrode batch corresponding to the end product model, extract the process parameter curves of the related batches from the corresponding scenario according to the end product model, and form an analysis sample set; A data analysis module is configured to extract quantitative fluctuation features from the process parameter curves in the analysis sample set, and perform correlation analysis on the quantitative fluctuation features and the statistical failure life of the corresponding batch; A parameter generation module is configured to define the numerical interval in which the quantitative fluctuation features and the statistical failure life are correlated in the application scenario as an optimized process feature according to the correlation analysis, and generate a corresponding production control parameter set according to the optimized process feature; A production control module is configured to determine the application scenario according to the end product model specified by the production order, and execute production by calling the production control parameter set corresponding to the application scenario.

Images