Liquid content analysis method, device, medium, host computer and drying line system

Abstract

The application relates to a liquid content analysis method, device, medium, upper computer and drying line system. The liquid content analysis method comprises the following steps: collecting current production process parameters of various types of process equipment of a to-be-tested object; based on the production process parameters, a liquid content prediction model corresponding to the process equipment is used for prediction to obtain a liquid content prediction result of the to-be-tested object; and qualified analysis is performed according to the liquid content prediction result to obtain a liquid content qualified analysis result of the to-be-tested object. By using the application, the liquid content analysis efficiency can be improved, and thus the production line production efficiency can be improved.

Description

Technical Field

[0001] The present application relates to the field of process detection technology, and in particular to a liquid content analysis method, device, medium, host computer and drying line system. Background Art

[0002] In process production involving contact with water or other liquids, it is usually necessary to monitor the liquid content in products or equipment, such as monitoring the water content of products after baking in an oven in a drying line system, or monitoring the water content in the coating process of battery cells.

[0003] Most of the existing liquid content monitoring methods rely on manual detection. Taking the water content monitoring of the battery cells in the drying line system oven as an example, the existing water content monitoring scheme of the drying line system is mainly offline monitoring, such as Figure 1 As shown, after the battery cells are assembled and baked once in the oven, the RGV (Rail Guided Vehicle) will transport the pallet containing the battery cells to the depalletizing position. On-site engineers will manually select fake battery cells (fake battery cells are unqualified products in the early production process, which are used for depalletizing to determine the drying effect of the entire pallet of battery cells). Then, they will use special testing equipment to measure the water content in them. During the testing period, the RGV will return the other battery cells in the pallet to the original oven. If the water content test fails, the battery cells will begin a second baking in the oven. If the test passes, the RGV will remove the battery cells and send them to the depalletizing position for depalletizing.

[0004] The manual detection and analysis method of liquid content, although the liquid content test results are accurate, is labor-intensive and time-consuming, and has low analysis efficiency, which leads to low production line efficiency. Summary of the Invention

[0005] Based on this, it is necessary to provide a liquid content analysis method, device, medium, host computer and drying line system that can improve the efficiency of liquid content analysis and thus improve the production efficiency of the production line in response to the above technical problems.

[0006] A method for analyzing liquid content, comprising:

[0007] Collect multiple types of current production process parameters of the process equipment of the object to be tested;

[0008] Based on the production process parameters, a liquid content prediction model corresponding to the process equipment is used to perform prediction to obtain a liquid content prediction result of the object to be tested;

[0009] A qualification analysis is performed based on the liquid content prediction result to obtain a qualification analysis result of the liquid content of the object to be tested.

[0010] In one of the embodiments, the process equipment includes an oven for baking the object to be measured, and the production process parameters include at least two of the following: vacuum degree in the oven, temperature in the oven, temperature and humidity outside the oven, aging characterization parameter of the sealing rubber strip of the oven, working current, working voltage, and power consumption.

[0011] In one of the embodiments, before the step of predicting the liquid content of the object to be measured based on the production process parameters by using the liquid content prediction model corresponding to the process equipment, the method further includes:

[0012] obtaining a plurality of types of historical production process parameters and historical liquid content actual values corresponding to historical production of the process equipment;

[0013] determining multi-dimensional feature data according to the plurality of types of historical production process parameters;

[0014] training an initial model by using the multi-dimensional feature data and the corresponding historical liquid content actual values to obtain a liquid content prediction model representing the corresponding relationship between the production process parameters and the liquid content data.

[0015] In one of the embodiments, after the step of training an initial model by using the multi-dimensional feature data and the corresponding historical liquid content actual values to obtain a liquid content prediction model representing the corresponding relationship between the production process parameters and the liquid content data, the method further includes:

[0016] selecting a sample from a plurality of groups of data obtained from a plurality of times of historical production, with the plurality of types of historical production process parameters and the historical liquid content actual values of one time of historical production as one group of data;

[0017] inputting the plurality of types of historical production process parameters in the sample into the liquid content prediction model to obtain a model output, comparing the model output with the historical liquid content actual values in the verification sample, and calculating a deviation;

[0018] if the deviation is within a preset allowable deviation, taking the liquid content prediction model as the liquid content prediction model corresponding to each process equipment;

[0019] if the deviation exceeds the preset allowable deviation, adjusting the historical production process parameters to determine the multi-dimensional feature data again, and returning to the step of training an initial model by using the multi-dimensional feature data and the corresponding historical liquid content actual values.

[0020] In one of the embodiments, after the step of predicting the liquid content of the object to be measured based on the production process parameters by using the liquid content prediction model corresponding to the process equipment, the method further includes:

[0021] obtaining actual liquid content values of the object to be measured corresponding to each process equipment.

[0022] The prediction accuracy of the liquid content prediction model for each process equipment is obtained based on the actual liquid content value of each process equipment and the liquid content prediction result; the liquid content prediction model is constructed based on multidimensional feature data determined by historical production process parameters of multiple process equipment.

[0023] In one embodiment, after obtaining the prediction accuracy of the liquid content prediction model for each process equipment, the method further includes:

[0024] If the prediction accuracy does not meet the preset accuracy requirement, the liquid content prediction model of the corresponding process equipment is adjusted.

[0025] In one embodiment, adjusting the liquid content prediction model of the corresponding process equipment includes:

[0026] Acquire multiple types of historical production process parameters of the corresponding process equipment;

[0027] Determining multidimensional feature data of the corresponding process equipment according to multiple types of historical production process parameters of the corresponding process equipment;

[0028] Assigning a first preset weight to the multidimensional feature data of the corresponding process equipment, assigning a second preset weight to the multidimensional feature data determined by historical production process parameters of the multiple process equipment, and performing weighted summation to obtain updated multidimensional feature data;

[0029] The liquid content prediction model of the corresponding process equipment is obtained by retraining based on the updated multi-dimensional feature data.

[0030] A liquid content analysis device, comprising:

[0031] Parameter acquisition module, used to collect multiple types of current production process parameters of the process equipment of the object to be tested;

[0032] A content prediction module, configured to perform prediction based on the production process parameters using a liquid content prediction model corresponding to the process equipment to obtain a liquid content prediction result of the object to be tested;

[0033] The qualification analysis module is used to perform qualification analysis based on the liquid content prediction result to obtain the liquid content qualification analysis result of the object to be tested.

[0034] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0035] Collect multiple types of current production process parameters of the process equipment of the object to be tested;

[0036] Based on the production process parameters, a liquid content prediction model corresponding to the process equipment is used to perform prediction to obtain a liquid content prediction result of the object to be tested;

[0037] A qualification analysis is performed based on the liquid content prediction result to obtain a qualification analysis result of the liquid content of the object to be tested.

[0038] A host computer includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0039] Collect multiple types of current production process parameters of the process equipment of the object to be tested;

[0040] Based on the production process parameters, a liquid content prediction model corresponding to the process equipment is used to perform prediction to obtain a liquid content prediction result of the object to be tested;

[0041] A qualification analysis is performed based on the liquid content prediction result to obtain a qualification analysis result of the liquid content of the object to be tested.

[0042] A drying line system comprises an oven and the above-mentioned host computer, wherein the oven is connected to the host computer; after the baking operation is completed, the oven sends multiple types of production process parameters to the computer equipment.

[0043] In one embodiment, the drying line system further comprises an operating vehicle, and the operating vehicle is connected to the host computer;

[0044] When the upper computer obtains a qualified liquid content analysis result of the test object, the upper computer sends a test object removal instruction signal to the operating vehicle, and the operating vehicle removes the test object from the oven;

[0045] When the upper computer obtains a liquid content qualification analysis result of the test object that is unqualified, it sends a re-bake instruction signal to the oven, and the oven starts a baking operation in response to the re-bake instruction signal.

[0046] The aforementioned liquid content analysis method, device, medium, host computer, and drying line system predict liquid content based on multiple production process parameters of the process equipment of the object to be tested and employ a liquid content prediction model corresponding to the process equipment. A qualification analysis is then performed based on the predicted liquid content results. This allows for automatic prediction and qualification analysis of liquid content, eliminating the need for manual testing and analysis, saving process time and improving both analysis and production line efficiency. Furthermore, predictions based on multiple production process parameters provide a more comprehensive consideration and yield better prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the conventional technology, the following briefly introduces the drawings required for use in the embodiments or the conventional technology descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1 This is a flow chart of the water content detection of battery cells in a traditional drying line system;

[0049] Figure 2 Schematic diagram of a liquid content analysis method according to one embodiment;

[0050] Figure 3 A flowchart of the construction and deployment of a liquid content prediction model in one embodiment;

[0051] Figure 4 Build a flow chart for the model in one embodiment;

[0052] Figure 5 A flowchart of reliability evaluation in one embodiment;

[0053] Figure 6 A schematic diagram of a process for predicting and analyzing the water content of a drying line system in one embodiment;

[0054] Figure 7 FIG. 4 is a structural block diagram of a liquid content analysis device in one embodiment. DETAILED DESCRIPTION

[0055] To facilitate understanding of the present application, the present application will be described more fully below with reference to the accompanying drawings. The accompanying drawings provide embodiments of the present application. However, the present application may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the present application more thorough and comprehensive.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application pertains. The terms used herein in the specification of this application are for the purpose of describing specific embodiments only and are not intended to limit this application.

[0057] It should be noted that when an element is considered to be "connected" to another element, it can be directly connected to the other element or connected to the other element through an intervening element. In addition, the "connection" in the following embodiments should be understood as "electrical connection", "communication connection", etc., if there is transmission of electrical signals or data between the connected objects.

[0058] In one embodiment, a liquid content analysis method is provided, which can be applied to a host computer, such as a computer, a notebook, or other portable terminal. Figure 2 As shown, the method includes the following steps:

[0059] S110: Collecting multiple types of current production process parameters of the process equipment of the object to be tested.

[0060] The DUT refers to an item whose liquid content requires prediction and analysis; the process equipment for the DUT refers to the process equipment used in the production process of the DUT that requires liquid content analysis. For example, the DUT could be a battery cell, and the liquid content analysis for the battery cell during baking would require an oven. Production process parameters are the operating parameters used by the process equipment to process the DUT. Specifically, various production process parameters can be collected by the process equipment and / or other data acquisition devices and then transmitted to a host computer.

[0061] S130: Based on the production process parameters, a liquid content prediction model corresponding to the process equipment is used to perform prediction to obtain a liquid content prediction result of the object to be tested.

[0062] A liquid content prediction model corresponding to the process equipment is pre-established. Specifically, the liquid content prediction model characterizes the correspondence between production process parameters and liquid content data. Based on the production process parameters, the liquid content prediction model processes the liquid content data, which is then used as the liquid content prediction result. Specifically, based on the production process parameters of a single process, the liquid content prediction model corresponding to the process equipment is used to perform a prediction, resulting in the predicted liquid content of all analytes processed by that process equipment in that batch.

[0063] S150: Performing a qualification analysis based on the liquid content prediction result to obtain a qualification analysis result of the liquid content of the object to be tested.

[0064] The liquid content prediction result reflects the amount of liquid content; specifically, the liquid content prediction result can be compared with the preset content qualification conditions to determine whether the liquid content of the object to be tested is qualified. Based on the liquid content prediction result, the liquid content qualification analysis of the object to be tested can be used for subsequent process operations. For example, in the case of battery cell baking, the liquid content qualification analysis result can be used to control the oven to bake again, or to control the baking to end and remove the battery cell.

[0065] The liquid content prediction result can be a specific content value, and the preset content qualification condition can be that the liquid content value is less than or equal to a set threshold. If the liquid content prediction result is less than or equal to the set threshold, it indicates that the liquid content of the test object is qualified; otherwise, it is unqualified. For another example, the liquid content prediction result can be a content classification result, such as one of high-quality, qualified, and unqualified. The preset content qualification condition can be that the prediction result is high-quality or qualified. If the liquid content prediction result is high-quality or qualified, it indicates that the liquid content of the test object is qualified. If the liquid content prediction result is unqualified, it indicates that the liquid content of the test object exceeds the standard or is unqualified.

[0066] The above-mentioned liquid content analysis method predicts liquid content based on multiple production process parameters of the process equipment of the object to be tested and uses the corresponding liquid content prediction model of the process equipment to perform a qualification analysis based on the predicted liquid content. This method can automatically predict and automatically perform qualification analysis of liquid content, eliminating the need for manual inspection and analysis, saving process time, thereby improving analysis efficiency and production line efficiency. Moreover, prediction based on multiple production process parameters takes into account a more comprehensive range of factors and achieves better prediction results.

[0067] In one embodiment, the process equipment includes an oven for baking the test object, and the production process parameters include at least two of the following: the vacuum degree inside the oven, the temperature inside the oven, the temperature and humidity outside the oven, the aging characterization parameter of the oven sealing strip, the operating current, the operating voltage, and the power consumption. Correspondingly, the liquid content can be the water content of the battery cell. The temperature and humidity outside the oven include the temperature outside the oven and the humidity outside the oven; the aging characterization parameter of the oven sealing strip is a parameter used to characterize the degree of aging of the oven sealing strip, such as the length of use.

[0068] Liquid content is generally influenced by many factors. For example, the liquid content of battery cells in an oven is affected by the cell material, oven consistency, oven heating curve, and the condition of the sealing strips. A single operating parameter cannot accurately predict water content. The oven's operating parameters include the vacuum level, oven temperature, external temperature and humidity, parameters characterizing aging of the oven's sealing strips, operating current, operating voltage, and power consumption. Using at least two of these parameters in the liquid content prediction model provides a more comprehensive consideration than using only the vacuum level or any other parameter alone, thus optimizing the prediction results.

[0069] Preferably, production process parameters can include oven vacuum, oven temperature, oven external temperature and humidity, oven sealant aging parameters, operating current, operating voltage, and power consumption. By collecting comprehensive data and analyzing the complete baking process, the final moisture content can be fully predicted.

[0070] In one embodiment, a model building step is further included before step S130, and the model building step includes steps (a1) to (a3). Preferably, the model building step can be performed before step S110.

[0071] Step (a1): obtaining multiple types of historical production process parameters of multiple process equipment and actual values ​​of historical liquid contents corresponding to the historical production.

[0072] Multiple process equipment can be the same type of equipment used in the same process flow, such as multiple ovens in a drying system. For a piece of process equipment, a historical production corresponds to multiple historical production process parameters and historical actual liquid content values ​​for that production.

[0073] Step (a2): Determine multidimensional feature data based on multiple categories of historical production process parameters.

[0074] A feature analysis is performed on at least two historical production process parameters among the multiple categories of historical production process parameters to obtain feature values; the multidimensional feature data includes the feature values ​​corresponding to at least two historical production process parameters among the multiple categories of historical production process parameters.

[0075] Step (a3): Use the multi-dimensional feature data and the corresponding historical actual values ​​of liquid content to train the initial model to obtain a liquid content prediction model that characterizes the corresponding relationship between production process parameters and liquid content.

[0076] The initial model includes any one of a classification model and a regression model; wherein the classification model can be SVM (Support Vector Machines), CNN (Convolutional Neural Networks), etc., and the regression model can be XGBOOST (eXtreme Gradient Boosting), LSTM (Long short-term memory), etc. It is understood that in other embodiments, the initial model can also adopt other types of models as long as they can achieve numerical prediction.

[0077] Based on the historical production of multiple process equipment, sample data is accumulated. Model training is performed after the process is stable and a certain number of sample libraries are accumulated. This can ensure the prediction accuracy of the trained liquid content prediction model.

[0078] In one embodiment, after step (a1) and before step (a2), the method further includes performing data cleaning on historical production process parameters and historical actual liquid content values. Accordingly, step (a2) determines multidimensional feature data based on the cleaned historical production process parameters, and step (a3) ​​trains an initial model based on the cleaned historical actual liquid content values. Data cleaning filters out abnormal data, improves data accuracy, and thus improves modeling accuracy.

[0079] In one embodiment, after step (a3), steps (a4) to (a7) are further included.

[0080] Step (a4): Taking multiple types of historical production process parameters and actual values ​​of historical liquid content of one historical production as a set of data, samples are selected from multiple sets of data obtained from multiple historical productions.

[0081] Step (a5): Input multiple categories of historical production process parameters in the sample into the liquid content prediction model to obtain model output, compare the model output with the actual value of the historical liquid content in the sample, and calculate the deviation.

[0082] A set of sample data includes historical production process parameters and historical actual liquid content values. The model output is compared with the historical actual liquid content values ​​of the corresponding sample. Specifically, the calculated error is a numerical value that represents the difference between the model output and the historical actual liquid content values. For example, it can be an error value or prediction accuracy. Furthermore, there can be multiple samples. In this case, the deviation can be calculated for each sample, and then the overall deviation is calculated based on the deviations of each sample, such as by averaging, to obtain the final deviation.

[0083] Step (a6): If the deviation is within the preset allowable deviation, the liquid content prediction model is used as the liquid content prediction model corresponding to each process equipment.

[0084] Among them, the preset allowable deviation is set according to actual needs.

[0085] Step (a7): If the deviation exceeds the preset allowable deviation, adjust the historical production process parameters to re-determine the multi-dimensional feature data, and return to step (a3).

[0086] After the liquid content prediction model is obtained through training, the liquid content prediction model is verified to ensure the prediction accuracy of the liquid content prediction model applied to each process equipment.

[0087] For example, take the prediction of water content of battery cells in the drying oven of the drying line system as an example. Figure 3 As shown, the model building operations can be as follows:

[0088] 1. Data Collection

[0089] The oven sends the time stamp of each baking process, the vacuum degree in the oven, the temperature in the oven, the working voltage, the working current and the power consumption to the upper computer, the temperature and humidity sensor outside the oven collects the temperature and humidity outside the oven and sends it to the upper computer, and the upper computer saves it to the local file. All production process parameters of each oven in each baking will be stored as a new file. For example, the production process parameters collected each time include 5 kinds, each baking lasts about 11 hours, and data is collected every half minute. After the batch is baked, a total of 2*60*11*5=6600 process parameter points are generated in the local file.

[0090] 2. Artificial inspection of water content

[0091] When the battery cell is baked in the oven, the RGV will take out the battery cell tray from the oven and then transport it to the water content detection position (usually at the tray disassembly position). The on-site inspection engineer takes out the dummy battery cell on the tray after the baking is completed, and then puts the un-baked dummy battery cell (for detection after the second baking is completed) in the empty place, and then sends it to the experimental device for water content detection. At the same time, the RGV transports the entire tray of battery cells back to the original oven position, waits for the water content detection result of the dummy battery cell, and if it is unqualified (water content detection value greater than 500 micrograms / gram), the oven door is closed for re-baking. If it is qualified (water content detection value less than 500 micrograms / gram), the RGV takes out the battery cell tray and sends it to the tray disassembly position for the next process.

[0092] 3. Accumulation of sample library data

[0093] After each baking of each oven is completed, 6600 process parameter points and a final water content detection value are generated as a group of samples. Assuming that there are 72 ovens in a drying line on site, and the baking time is 11 hours each time, 144 samples can be accumulated in a day. If production does not stop, 1008 samples can be accumulated in a week, providing sufficient sample library data for subsequent model establishment.

[0094] 4. Model building

[0095] When the sample library data is sufficient, the model building process can be started, as shown in Figure 4

[0096] a) Data cleaning is performed on the sample library data to filter out some abnormal data, including missing sample data (represented by discontinuous time axis), communication abnormal data, process unstable data, and battery product quality abnormal data (represented by obviously abnormal data).

[0097] ​b) Feature processing: In combination with common signal features and process features, the feature values of the process parameter points in each group of samples are analyzed to obtain feature values, such as the slope and extreme value of the vacuum degree under different baking time periods, the extreme value and root mean square value of the temperature in the oven, the stability of the current and voltage, and the total power consumption, as well as the use time of the oven adhesive tape, the temperature and humidity outside the oven, etc. A large number of raw data are refined into a multi-dimensional feature space to obtain multi-dimensional feature data. For example, the multi-dimensional feature data can be a matrix composed of multiple feature values wherein X1 represents the first feature value, such as the slope of the vacuum degree, and Xn represents the nth feature value.

[0098] c) Model building: A classification model or a regression model is used to model the training data in the sample data, which can be the compressed multi-dimensional feature data and the water content detection value as two variables of the model, such as one as an input variable and one as an output variable, to train the corresponding classification model or regression model.

[0099] d) Model verification: Part of the data in the sample database is randomly extracted as a verification set, which is input into the trained model, and the model output is compared with the water content detection value. If the prediction accuracy meets the standard, the model is determined, otherwise the model parameters are adjusted and retrained.

[0100] e) Model output: After the model verification is completed, the model file is output and exported to be applied to the actual scene for inference.

[0101] 5. Model inference:

[0102] After the model verification is completed, the model file is issued to the production line computing resource layer, which is pre-installed with a real-time data interface and a model execution engine. Through the collection of actual field data, the execution engine performs real-time model inference, and the prediction result of the water content can be quickly calculated after each baking is completed.

[0103] 6. Prediction of water content:

[0104] After the model inference is completed, the prediction value of the water content is output, and different modes of water content prediction results are output according to different model selections. If the user selects a classification model, the output results are I class (high quality, water content 0-200 micrograms per gram), II class (qualified, water content 200-500 micrograms per gram), III class (unqualified, water content > 500 micrograms per gram), etc. If the user selects a regression model, the output is the specific prediction value of the water content, such as 159 micrograms per gram or 354 micrograms per gram, etc.

[0105] 7. Reliability evaluation

[0106] When the trained model is loaded locally and can predict the water content in real time, the model reliability assessment phase can be started. A large amount of actual data is used to test whether the model can maintain sufficient robustness to fluctuations in field signals, communication interference, fine-tuning of work steps, and even process instability. The reliability assessment process is as follows: Figure 5 As shown. After entering the model reliability assessment phase, the predicted moisture content value is recorded for each bake. The user selects the frequency of manual spot checks, records and calculates the prediction accuracy in real time, and regularly issues an assessment report. Taking into account the actual differences between ovens on site, the report content can be divided into overall prediction accuracy and individual oven prediction accuracy. If the accuracy of an individual oven is low, this deviation is fed back into the trained model, and independent parameter fine-tuning is performed to adapt the model to the actual conditions of that oven. Finally, the user or expert group reviews the assessment report. If the model reliability is deemed to meet the standard, the model is officially deployed.

[0107] In one embodiment, after step S130, the method further includes: obtaining the actual liquid content value of the test object corresponding to each process equipment; and obtaining the prediction accuracy of the liquid content prediction model for each process equipment based on the actual liquid content value and the liquid content prediction result of each process equipment. The liquid content prediction model is constructed based on multidimensional feature data determined by historical production process parameters of multiple process equipment. Specifically, the liquid content prediction model is obtained by training an initial model using the multidimensional feature data determined by the historical actual liquid content values ​​and historical production process parameters of multiple process equipment.

[0108] For each process equipment, the prediction accuracy is obtained based on the actual value and the predicted result after the prediction, and the prediction accuracy of each process equipment is monitored to regularly evaluate the reliability of the liquid content prediction model of each process equipment. For example, if the liquid content prediction result is a specific content value, the prediction error can be calculated based on the actual liquid content value and the liquid content prediction result. The prediction error is calculated by dividing the actual liquid content value to obtain the relative error, and the prediction accuracy is calculated by subtracting the relative error from 1. If the liquid content prediction result is a classification result, such as one of high-quality, qualified, and unqualified, it can be first classified based on the actual liquid content value and judged as high-quality, qualified, or unqualified. Then, the classification result of the actual liquid content value is compared with the predicted classification result to see if they are consistent. If they are inconsistent, the prediction accuracy is low. If they are consistent, the prediction accuracy is high.

[0109] In one embodiment, after obtaining the prediction accuracy of the liquid content prediction model for each process equipment, the method further includes: if the prediction accuracy does not meet the preset accuracy requirement, adjusting the liquid content prediction model of the corresponding process equipment.

[0110] The preset accuracy requirement can be set according to actual conditions. If the prediction accuracy does not meet the preset accuracy requirement, it indicates that the reliability of the liquid content prediction model of the process equipment is low. By adjusting the liquid content prediction model of the process equipment, the original old model is replaced by the adjusted liquid content prediction model, so as to ensure the prediction accuracy of the liquid content.

[0111] In one of the embodiments, the step of adjusting the liquid content prediction model of the corresponding process equipment comprises:

[0112] Step (b1): Obtain a plurality of types of historical production process parameters of historical production of the corresponding process equipment.

[0113] Specifically, a plurality of types of historical production process parameters of historical production of the process equipment are collected from the process equipment to be measured.

[0114] Step (b2): Determine the multi-dimensional feature data of the process equipment according to the corresponding multi-dimensional feature data of the corresponding process equipment.

[0115] The determination method of the multi-dimensional feature data of the process equipment can be the same as the method of obtaining the multi-dimensional feature data in the aforementioned step (a2), which will not be repeated here.

[0116] Step (b3): Assign a first preset weight to the multi-dimensional feature data of the corresponding process equipment, and assign a second preset weight to the multi-dimensional feature data determined by the historical production process parameters of the plurality of process equipment, and perform weighted summation to obtain updated multi-dimensional feature data.

[0117] The first preset weight and the second preset weight can be set according to actual conditions. Specifically, the first preset weight is greater than the second preset weight, so as to increase the weight of the sample of the process equipment that needs to be adjusted, for example, the second preset weight is 20%, and the first preset weight is 80%.

[0118] For example, the multi-dimensional feature data determined by the historical production process parameters of the plurality of process equipment can be a matrix including X1, X2, X3, …, Xn, and the multi-dimensional feature data of the process equipment can be a matrix including Y1, Y2, Y3, …, Yn. The value of X1*first preset weight+Y1*second preset weight can be used as the first characteristic value in the updated multi-dimensional feature data, the value of X2*first preset weight+Y2*second preset weight can be used as the second characteristic value in the updated multi-dimensional feature data, and so on, to obtain each characteristic value in the updated multi-dimensional feature data.

[0119] Step (b4): Re-train the liquid content prediction model of the corresponding process equipment based on the updated multi-dimensional feature data.

[0120] Specifically, the updated multi-dimensional feature data is used to retrain the initial model to obtain an adjusted liquid content prediction model for the process equipment that needs to be adjusted.

[0121] After the initial model training based on the historical production process parameters of multiple process equipment is completed, the resulting liquid content prediction model is used as the liquid content prediction model for all process equipment. Considering the consistency of various process equipment on site, after the actual application of the model, the corresponding liquid content prediction model is adjusted according to the historical data of each process equipment to form a unique model dedicated to each process equipment, further improving the accuracy of content prediction.

[0122] Taking the above liquid content analysis method applied to the upper computer of the drying line system as an example, the water content of the battery cells baked in the oven is predicted, such as Figure 6 As shown, when the upper computer obtains the qualified result of the liquid content of the test object, it sends the test object removal instruction signal to RGV, and RGV takes the test object out of the oven; when the upper computer obtains the qualified result of the liquid content of the test object, it sends the re-bake instruction signal to the oven, and the oven responds to the re-bake instruction signal to start the baking operation. The liquid content prediction model of this application can accurately predict the interval value of water content; comparison Figure 6 and Figure 1 It can be seen that compared with the traditional process, the implementation of automated testing using this application saves the workload of on-site inspection engineers, improves the utilization rate of RGV and oven, and overall can greatly improve the production efficiency of the production line.

[0123] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0124] In one embodiment, Figure 7 As shown, a liquid content analysis device is provided, including a parameter acquisition module 710 , a content prediction module 730 and a qualification analysis module 750 .

[0125] The parameter acquisition module 710 is used to collect the current multiple types of production process parameters of the process equipment of the object to be tested; the content prediction module 730 is used to predict the liquid content prediction model corresponding to the process equipment based on the production process parameters to obtain the liquid content prediction result of the object to be tested; the qualification analysis module 750 is used to perform qualification analysis based on the liquid content prediction result to obtain the liquid content qualification analysis result of the object to be tested.

[0126] The liquid content analysis device predicts liquid content based on multiple production process parameters of the process equipment used for the object being tested and employs a corresponding liquid content prediction model. The device then performs a qualification analysis based on the predicted liquid content. This allows for automatic prediction and qualification analysis of liquid content, eliminating the need for manual testing and analysis, saving process time and improving both analysis and production line efficiency. Furthermore, predictions based on multiple production process parameters provide a more comprehensive approach and yield better prediction results.

[0127] In one embodiment, the process equipment includes an oven for baking the object to be tested, and the production process parameters include at least two of the vacuum degree in the oven, the temperature in the oven, the temperature and humidity outside the oven, the aging characterization parameters of the oven sealing strip, the operating current, the operating voltage and the power consumption.

[0128] In one embodiment, the liquid content analysis device further includes a model building module (not shown), which is used to obtain multiple categories of historical production process parameters of multiple process equipment and actual values ​​of historical liquid content corresponding to the historical production before the content prediction module 730 performs the corresponding function; determine multidimensional feature data based on the multiple categories of historical production process parameters; and use the multidimensional feature data and the corresponding actual values ​​of historical liquid content to train the initial model to obtain a liquid content prediction model that characterizes the correspondence between the production process parameters and the liquid content.

[0129] In one embodiment, the model building module is also used to, after obtaining the liquid content prediction model that characterizes the correspondence between the production process parameters and the liquid content, use multiple categories of historical production process parameters and historical liquid content actual values ​​of a historical production as a group of data, and select samples from multiple groups of data obtained from multiple historical productions; input the multiple categories of historical production process parameters in the sample into the liquid content prediction model to obtain a model output, compare the model output with the historical liquid content actual value in the sample, and calculate the deviation; if the deviation is within the preset allowable deviation, use the liquid content prediction model as the liquid content prediction model corresponding to each process equipment; if the deviation exceeds the preset allowable deviation, adjust the historical production process parameters to re-determine the multidimensional feature data, and return to re-use the multidimensional feature data and the corresponding historical liquid content actual value to train the initial model to obtain a liquid content prediction model that characterizes the correspondence between the production process parameters and the liquid content.

[0130] In one embodiment, the liquid content analysis device further includes an accuracy analysis module (not shown) configured to obtain the actual liquid content value of the analyte corresponding to each process device after the content prediction module 730 performs its corresponding functions. The prediction accuracy of the liquid content prediction model for each process device is determined based on the actual liquid content value and the liquid content prediction result for each process device. The liquid content prediction model is constructed based on multidimensional feature data determined from historical production process parameters of multiple process devices.

[0131] In one embodiment, the liquid content analysis device further includes a model adjustment module (not shown), which is used to adjust the liquid content prediction model of the corresponding process equipment when the prediction accuracy does not meet the preset accuracy requirements after the accuracy analysis module obtains the prediction accuracy of the liquid content prediction model for each process equipment.

[0132] In one embodiment, the model adjustment module obtains multiple categories of historical production process parameters of the corresponding process equipment; determines multidimensional feature data of the process equipment based on the multiple categories of historical production process parameters of the corresponding process equipment; assigns a first preset weight to the multidimensional feature data of the corresponding process equipment, and assigns a second preset weight to the multidimensional feature data determined by the historical production process parameters of multiple process equipment, and performs weighted summation to obtain updated multidimensional feature data; and retrains the liquid content prediction model of the corresponding process equipment based on the updated multidimensional feature data.

[0133] The specific limitations of the liquid content analysis device can refer to the limitations of the liquid content analysis method described above, which will not be repeated here. Each module in the above liquid content analysis device can be implemented by software, hardware, and a combination thereof. The above modules can be embedded in or independent of the processor in the host computer in hardware form, or can be stored in the memory in the host computer in software form, so that the processor can call and execute the operations corresponding to each module. It should be noted that the division of modules in the embodiments of the present application is illustrative, and is only a logical functional division. In actual implementation, there can be another division method.

[0134] In one embodiment, a host computer is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in the above method embodiments.

[0135] The above host computer can implement the steps in the above method embodiments, and for the same reason, can improve the liquid content prediction efficiency, thereby improving the production line production efficiency.

[0136] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps in the above method embodiments.

[0137] The above computer readable storage medium can implement the steps in the above method embodiments, and for the same reason, can improve the liquid content prediction efficiency, thereby improving the production line production efficiency.

[0138] In one embodiment, a drying line system is provided, including an oven and the host computer in the above embodiment, the oven being connected to the host computer; the oven sends multiple types of production process parameters to the computer device after the roasting operation is completed.

[0139] The above drying line system, since the above host computer is used, for the same reason, the liquid content analysis efficiency is high, and the production line production efficiency is high.

[0140] In one of the embodiments, the drying line system further includes an operation vehicle connected to the host computer; the host computer sends a measured object taking-out instruction signal to the operation vehicle when the liquid content eligibility analysis result of the measured object is qualified, and the operation vehicle takes out the measured object from the oven; the host computer sends a re-roasting instruction signal to the oven when the liquid content eligibility analysis result of the measured object is unqualified, and the oven starts the roasting operation in response to the re-roasting instruction signal.

[0141] The operating vehicle, such as an RGV, is used to load and unload cells from the drying oven in the drying line system. Compared to traditional processes, the drying line system in this application implements automated liquid content detection, reducing the workload of on-site inspection engineers and improving the utilization rate of the RGV and the drying oven. Overall, it can significantly improve production line efficiency.

[0142] Throughout this specification, references to terms such as "some embodiments," "other embodiments," and "desired embodiments" indicate that a particular feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present application. The schematic descriptions of these terms throughout this specification do not necessarily refer to the same embodiment or example.

[0143] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0144] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for analyzing liquid content, characterized in that: include: Collect multiple types of current production process parameters of the process equipment of the object to be tested; Based on the production process parameters, a liquid content prediction model corresponding to the process equipment is used to perform prediction to obtain a liquid content prediction result of the object to be tested; Performing a qualification analysis based on the liquid content prediction result to obtain a qualification analysis result of the liquid content of the object to be tested; After obtaining the predicted result of the liquid content of the object to be measured by using the liquid content prediction model corresponding to the process equipment based on the production process parameters, the method further includes: Obtain the actual liquid content value of the object to be tested corresponding to each process equipment; Obtaining the prediction accuracy of the liquid content prediction model for each process equipment based on the actual liquid content value of each process equipment and the liquid content prediction result; the liquid content prediction model is constructed based on multidimensional feature data determined by historical production process parameters of multiple process equipment; If the prediction accuracy does not meet the preset accuracy requirement, adjusting the liquid content prediction model of the corresponding process equipment, wherein adjusting the liquid content prediction model of the corresponding process equipment includes: Acquire multiple types of historical production process parameters of the corresponding process equipment; Determining multidimensional feature data of the corresponding process equipment according to multiple types of historical production process parameters of the corresponding process equipment; Assigning a first preset weight to the multidimensional feature data of the corresponding process equipment, assigning a second preset weight to the multidimensional feature data determined by historical production process parameters of the multiple process equipment, and performing weighted summation to obtain updated multidimensional feature data; The liquid content prediction model of the corresponding process equipment is obtained by retraining based on the updated multi-dimensional feature data.

2. The method according to claim 1, characterized in that The process equipment includes an oven for baking the object to be tested, and the production process parameters include at least two of the vacuum degree in the oven, the temperature in the oven, the temperature and humidity outside the oven, the aging characterization parameters of the oven sealing strip, the working current, the working voltage and the power consumption.

3. The method according to claim 1, characterized in that Before obtaining the liquid content prediction result of the object to be tested by using the liquid content prediction model corresponding to the process equipment to perform prediction based on the production process parameters, the method further includes: Obtain multiple types of historical production process parameters of multiple process equipment and actual values ​​of historical liquid contents corresponding to the historical production; determining multidimensional feature data based on multiple categories of historical production process parameters; The multi-dimensional feature data and the corresponding historical actual values ​​of liquid content are used to train an initial model to obtain a liquid content prediction model that characterizes the corresponding relationship between the production process parameters and the liquid content data.

4. The method according to claim 3, characterized in that After the multi-dimensional feature data and the corresponding historical actual values ​​of liquid content are used to train the initial model to obtain a liquid content prediction model that characterizes the corresponding relationship between the production process parameters and the liquid content data, the method further includes: Taking multiple categories of historical production process parameters and historical liquid content actual values ​​of one historical production as a set of data, samples are selected from multiple sets of data obtained from multiple historical productions; Inputting multiple types of historical production process parameters in the sample into the liquid content prediction model to obtain a model output, comparing the model output with the actual value of the historical liquid content in the sample, and calculating the deviation; If the deviation is within the preset allowable deviation, the liquid content prediction model is used as the liquid content prediction model corresponding to each process equipment; If the deviation exceeds the preset allowable deviation, the historical production process parameters are adjusted to re-determine the multidimensional feature data, and the process returns to the step of training the initial model using the multidimensional feature data and the corresponding historical actual liquid content value.

5. A liquid content analysis device, characterized in that: include: Parameter acquisition module, used to collect multiple types of current production process parameters of the process equipment of the object to be tested; A content prediction module, configured to perform prediction based on the production process parameters using a liquid content prediction model corresponding to the process equipment to obtain a liquid content prediction result of the object to be tested; A qualification analysis module is used to perform qualification analysis based on the liquid content prediction result to obtain a qualification analysis result of the liquid content of the object to be tested; An accuracy analysis module is configured to obtain the actual liquid content value of the test object corresponding to each process equipment; obtain the prediction accuracy of the liquid content prediction model for each process equipment based on the actual liquid content value of each process equipment and the liquid content prediction result; the liquid content prediction model is constructed based on multidimensional feature data determined by historical production process parameters of multiple process equipment; A model adjustment module is used to adjust the liquid content prediction model of the corresponding process equipment when the prediction accuracy does not meet the preset accuracy requirements, including: obtaining multiple categories of historical production process parameters of the corresponding process equipment; determining the multidimensional feature data of the process equipment based on the multiple categories of historical production process parameters of the corresponding process equipment; assigning a first preset weight to the multidimensional feature data of the corresponding process equipment, and assigning a second preset weight to the multidimensional feature data determined by the historical production process parameters of the multiple process equipment, and performing weighted summation to obtain updated multidimensional feature data; and retraining based on the updated multidimensional feature data to obtain the liquid content prediction model of the corresponding process equipment.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A host computer, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

8. A drying line system, characterized in that: The invention comprises an oven and the host computer according to claim 7, wherein the oven is connected to the host computer; after the baking operation is completed, the oven sends multiple types of production process parameters to the host computer.

9. The drying line system according to claim 8, characterized in that Also included is an operating vehicle connected to the host computer; When the upper computer obtains a qualified liquid content analysis result of the test object, the upper computer sends a test object removal instruction signal to the operating vehicle, and the operating vehicle removes the test object from the oven; When the upper computer obtains a liquid content qualification analysis result of the test object that is unqualified, it sends a re-bake instruction signal to the oven, and the oven starts a baking operation in response to the re-bake instruction signal.

Images