CN115389955A - Differential pressure prediction method and device for total pressure test inside and outside battery - Google Patents
Differential pressure prediction method and device for total pressure test inside and outside battery Download PDFInfo
- Publication number
- CN115389955A CN115389955A CN202211017933.2A CN202211017933A CN115389955A CN 115389955 A CN115389955 A CN 115389955A CN 202211017933 A CN202211017933 A CN 202211017933A CN 115389955 A CN115389955 A CN 115389955A
- Authority
- CN
- China
- Prior art keywords
- internal
- external
- battery
- submodel
- regression
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/385—Arrangements for measuring battery or accumulator variables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/389—Measuring internal impedance, internal conductance or related variables
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
The invention discloses a differential pressure prediction method and a differential pressure prediction device for testing the internal and external total pressures of a battery, wherein the internal and external total pressures, the internal and external impedances of a positive electrode and the internal and external impedances of a negative electrode of the battery are tested to form a test data set; predicting a predicted value of the battery pressure difference by adopting sub models of various regression algorithms; comparing the pressure difference predicted value obtained by each submodel with the tested pressure difference actual value to obtain the submodel with the smallest error between the predicted value and the actual value, and using the submodel as a label; bringing each pressure difference predicted value and a label into secondary prediction based on support vector regression to obtain a probability vector corresponding to each sub-model; and finally, calculating a weighted average value as a final prediction result according to the prediction value of each sub-model and the corresponding probability vector. According to the invention, through testing data sets of the total internal and external pressures of the battery, the internal and external impedances of the anode and the cathode, and through an integrated learning method, several traditional machine learning regression algorithms are used as submodels, secondary prediction based on support vector regression is provided, and the prediction accuracy is improved.
Description
Technical Field
The invention relates to the field of battery testing, in particular to a method and a device for predicting differential pressure of total pressure testing inside and outside a battery.
Background
The battery pack has the advantages that after the battery cells are connected in series and in parallel, the total positive pole and the total negative pole (called as inner total pressure) can be output, the inner total pressure is output to a high-voltage connector outside the battery (called as outer total pressure) through a battery internal control system or a switching loop, when voltage is acquired, an internal clamp and an external matched connector are generally adopted to connect and acquire the voltage, the general phase difference between the inner total pressure and the outer total pressure is about 1-2V, and the specific phase difference value needs to be manually verified in repeatability under different testing conditions.
In the test of the internal total pressure and the external total pressure of the battery production line, the voltage acquisition of the internal total pressure and the external total pressure is respectively carried out at the present stage, after repeated acquisition is carried out for multiple times, the stable voltage values are subtracted, the allowable range of the internal and external differential pressure is calculated, and the acquisition measurement corresponding to different battery packs or different process flows needs to be repeatedly confirmed, which is particularly inconvenient. When a prediction method is used, the prediction precision of a single regression model is limited.
Disclosure of Invention
The invention aims at: the method and the device for predicting the pressure difference of the total internal and external pressure test of the battery are provided, the voltage and the resistance of a test loop are collected, and a large amount of test data is combined, so that the relation of the total internal and external pressure difference can be found and predicted.
The technical scheme of the invention is as follows:
a differential pressure prediction method for a total pressure test inside and outside a battery comprises the following steps:
s1, testing the internal total pressure and the external total pressure of a battery, the internal and external impedance of a positive electrode and the internal and external impedance of a negative electrode to form a test data set D;
s2, respectively predicting new battery data according to the test data set D by adopting sub models of N regression algorithms to obtain differential pressure predicted values V1, V2 \8230, \8230andVN;
s3, comparing the predicted values V1 and V2 \8230 \ 8230and VN obtained by the submodels with the actual value VT of the tested pressure difference to obtain the submodel with the minimum error between the predicted value and the actual value, and using the submodel as a label;
s4, carrying out secondary prediction based on support vector regression with the pressure difference predicted values V1 and V2 \8230 \ 8230: (VN) and the label obtained from each sub-model to obtain N probabilities P1 and P2 \8230 \ 8230: (PN);
s5, finally, calculating weighted average values according to predicted values V1 and V2 of each submodel V1, V2 \8230velocity \8230VNand corresponding probability vectors P1 and P2 \8230velocity \8230andPN, and taking the weighted average values as final prediction results VF:
VF=（V1* P1+V2* P2……+VN* PN）/（P1+ P2……+ PN）。
preferably, the submodels of the N regression algorithms at least include a random forest regression submodel, a support vector regression submodel and a bayesian ridge regression submodel.
Preferably, the test data set D further includes a relationship between the voltage difference and the impedance between the positive electrode and the negative electrode.
Preferably, the internal total pressure and the external total pressure data of the battery are acquired through a digital multimeter and a switching module, and meanwhile, the impedance data of an internal loop and an external loop of the battery are acquired through the digital multimeter; all data are read through computer communication and stored in a database.
The utility model provides a pressure differential prediction device of total pressure test inside and outside battery, includes digital multimeter and computer, the digital multimeter gathers the interior total pressure of battery, total pressure outside, positive pole internal and external impedance, negative pole internal and external impedance data, and the computer reads digital multimeter's collection data through gathering software to store in the database.
Preferably, the digital multimeter tests total internal pressure, total external pressure, internal and external positive impedance and internal and external negative impedance of the plurality of batteries to form a test data set D;
the computer adopts sub models of N regression algorithms, and respectively predicts new battery data according to the test data set D to obtain predicted values V1 and V2 \8230; \8230andVN;
comparing the predicted values V1 and V2 of the differential pressure V8230A V8230VN and VN obtained by each submodel with the actual value VT of the differential pressure tested to obtain the submodel with the minimum error between the predicted value and the actual value as a label;
the pressure difference predicted values V1 and V2 \8230 \ 8230 \ VN and the label obtained by each sub-model are brought into secondary prediction based on support vector regression to obtain N probabilities P1 and P2 \8230 \ 8230and PN corresponding to each sub-model;
and finally, calculating a weighted average value according to predicted values V1 and V2 (8230) \8230; VN and corresponding probability vectors P1 and P2 (8230) \8230and PN of each sub-model N, wherein the weighted average value is used as a final prediction result VF:
VF=（V1* P1+V2* P2……+VN* PN）/（P1+ P2……+ PN）。
preferably, the submodels of the N regression algorithms at least include a random forest regression submodel, a support vector regression submodel and a bayesian ridge regression submodel.
Preferably, the test data set D further includes data of internal and external impedances of the positive electrode and the negative electrode of the battery.
Preferably, the database is stored by a computer hard disk, a server or a cloud.
The invention has the advantages that:
according to the invention, the internal and external total pressures of the battery, the internal and external impedances of the anode and the cathode are tested through the digital multimeter data, the differential pressure range of the internal and external total pressures corresponding to the new battery can be predicted through an integrated learning method, several traditional machine learning regression algorithms are used as submodels, the quadratic prediction based on support vector regression is provided, and the prediction accuracy is improved.
Drawings
The invention is further described below with reference to the following figures and examples:
FIG. 1 is a schematic diagram of a differential pressure predicting device for testing total pressure inside and outside a battery according to the present invention;
FIG. 2 is a flow chart of the prediction of the differential pressure in the total pressure test inside and outside the battery.
Detailed Description
As shown in fig. 1, the differential pressure prediction apparatus for testing total internal and external voltages of a battery of the present invention includes a digital multimeter and a computer, wherein the digital multimeter and a switching module collect data of the total internal and external voltages, the internal and external impedances of a positive electrode, and the internal and external impedances of a negative electrode of the battery. The computer reads the collected data of the digital multimeter through the collection software and stores the data in the database. And inputting data into test data sets of the internal and external total pressures, the internal and external impedances of the anode and the negative internal and external impedances, which correspond to the test, and predicting the differential pressure range of the internal and external total pressures corresponding to the new battery by an integrated learning method.
The digital multimeter and the switching module test the total pressure and the total external pressure data in the battery and the internal and external impedance and internal and external impedance data of the anode and the cathode of the battery, input the data into a computer and store the data in a database to form a test data set D, wherein the test data set D also comprises the relation between the pressure difference and the impedance between the anode and the cathode; the database is stored by a computer hard disk, a server or a cloud.
The computer integrated learning adopts N sub-models of regression algorithms, and new battery data are respectively predicted according to the test data set D to obtain predicted values V1 and V2 of 8230, 8230and VN; the submodels of the N regression algorithms at least comprise a random forest regression submodel, a support vector regression submodel and a Bayesian ridge regression submodel;
comparing the predicted values V1 and V2 of the differential pressure V8230A V8230VN and VN obtained by each submodel with the actual value VT of the differential pressure tested to obtain the submodel with the minimum error between the predicted value and the actual value as a label;
carrying out secondary prediction based on support vector regression on VN and labels to obtain N probabilities P1 and P2 \8230; \ 8230and PN corresponding to each submodel;
and finally, calculating a weighted average value according to predicted values V1 and V2 (8230) \8230; VN and corresponding probability vectors P1 and P2 (8230) \8230and PN of each sub-model N, wherein the weighted average value is used as a final prediction result VF:
VF=（V1* P1+V2* P2……+VN* PN）/（P1+ P2……+ PN）。
in specific implementation, as shown in fig. 2, the method for predicting a pressure difference according to an embodiment of the present invention includes the steps of:
s1, acquiring internal and external total pressure data of a battery through a digital multimeter and a switching module, and acquiring internal and external impedance data of a positive electrode and internal and external impedance data of a negative electrode of the battery through the digital multimeter; inputting the voltage difference into a computer and storing the voltage difference in a database to form a test data set D, wherein the test data set D also comprises the relation between the voltage difference and the impedance between the anode and the cathode; all data are read by computer communication.
S2, integrating the advantages of each sub-model, and training each sub-model firstly when the model is trained. Three traditional machine learning regression algorithms are used as submodels: random forest regression, support vector regression and Bayesian ridge regression, and respectively predicting new battery data according to the test data set D to obtain predicted values V1, V2 and V3 of the pressure difference;
1. random forest regression is a classifier that trains and predicts samples using multiple decision trees, and the class of its output is determined by the mode of the class output by the individual trees. The random forest construction method has two aspects: random selection of data and random selection of features to be selected.
(1) Random selection of data
First, a set of sub-data is constructed from the original data set using the replaced samples, the sub-data set having the same data size as the original data set. Elements of different sub data sets may be repeated, as may elements in the same sub data set. Second, the sub-decision trees are constructed using the sub-data sets, and this data is placed into each sub-decision tree, which outputs a result. And finally, if new data is needed to obtain a classification result through the random forest, the output result of the random forest can be obtained through voting on the judgment result of the sub-decision tree.
(2) Random selection of features to be selected
Similar to the random selection of the data set, each splitting process of the subtrees in the random forest does not use all the features to be selected, but randomly selects a certain feature from all the features to be selected, and then selects the optimal feature from the randomly selected features. Therefore, decision trees in the random forest can be different from each other, the diversity of the system is improved, and the classification performance is improved.
2. Support vector regression is a supervised learning model with associated learning algorithms for analyzing data for classification and regression analysis in machine learning. In support vector regression, the straight line required to fit the data is called the hyperplane. The goal of the support vector machine algorithm is to find a hyperplane in n-dimensional space that unambiguously classifies data points. The data points on both sides of the hyperplane that are closest to the hyperplane are called support vectors. These affect the location and orientation of the hyperplane and thus help in building the SVM.
The SVM regression algorithm is called support vector regression or SVR. Support vector regression is a supervised learning algorithm used to predict discrete values. Support vector regression uses the same principles as SVM. The basic idea behind SVR is to find the best fit line. In SVR, the best fit line is the hyperplane with the greatest number of points. One of the main advantages of SVR is that its computational complexity is independent of the dimensions of the input space. In addition, it has excellent generalization ability and high prediction precision.
3. Bayesian linear regression (Bayesian linear regression) is a linear regression (linear regression) model solved using the Bayesian inference (Bayesian inference) method in statistics.
Bayesian linear regression considers the parameters of a linear model as random variables and calculates its posterior (posterior) by a priori (prior) of the model parameters (weighting coefficients). Bayesian linear regression can be solved by using a numerical method, and under certain conditions, the posterior of an analytic form or relevant statistics of the analytic form can be obtained. The Bayesian linear regression has the basic property of a Bayesian statistical model, and can solve the probability density function of the weight coefficient, perform online learning and model hypothesis test based on Bayesian factors (Bayes factors). If the Bayesian linear regression uses the normal prior, the estimation result of the MAP is equivalent to the ridge regression.
And S3, comparing the pressure difference predicted values V1, V2 and V3 obtained by the random forest regression, support vector regression and Bayesian ridge regression submodels with a tested pressure difference true value VT to obtain a submodel with the minimum error between the predicted value and the true value, and using the submodel as a label.
When the prediction method is used, due to the fact that the prediction precision of a single regression model is limited, the prediction accuracy is improved by using ensemble learning. The model takes three conventional machine learning regression algorithms which are common in machine learning as submodels. The different regression prediction algorithms may have different prediction accuracies for different battery data, and quadratic prediction based on support vector regression is proposed on the basis of the three submodels.
S4, bringing the pressure difference predicted values V1, V2 and V3 obtained by each sub-model and the labels into secondary prediction based on support vector regression to obtain 3 probabilities P1, P2 and P3 corresponding to each sub-model to form a probability vector;
s5, finally, calculating a weighted average value according to the predicted values V1, V2 and V3 of the three sub models and the corresponding probability vectors P1, P2 and P3 to serve as a final prediction result VF:
VF=（V1* P1+V2* P2+V3* P3）/（P1+ P2+ P3）。
the above embodiments are only for illustrating the technical idea and features of the present invention, and the purpose of the embodiments is to enable those skilled in the art to understand the content of the present invention and implement the present invention, and not to limit the protection scope of the present invention. All modifications made according to the spirit of the main technical scheme of the invention are covered in the protection scope of the invention.
Claims (9)
1. A differential pressure prediction method for a total pressure test inside and outside a battery is characterized by comprising the following steps:
s1, testing the internal total pressure and the external total pressure of a battery, the internal and external impedance of a positive electrode and the internal and external impedance of a negative electrode to form a test data set D;
s2, respectively predicting new battery data according to the test data set D by adopting sub models of N regression algorithms to obtain predicted values V1 and V2 \8230; VN;
s3, comparing the differential pressure predicted values V1 and V2 \8230, \8230VNand VN obtained by each submodel with the tested differential pressure true value VT to obtain the submodel with the smallest error between the predicted value and the true value as a label;
s4, carrying out secondary prediction based on support vector regression with the pressure difference predicted values V1 and V2 \8230 \ 8230: (VN) and the label obtained from each sub-model to obtain N probabilities P1 and P2 \8230 \ 8230: (PN);
s5, finally, calculating weighted average values according to predicted values V1 and V2 of each submodel V1, V2 \8230velocity \8230VNand corresponding probability vectors P1 and P2 \8230velocity \8230andPN, and taking the weighted average values as final prediction results VF:
VF=（V1* P1+V2* P2……+VN* PN）/（P1+ P2……+ PN）。
2. the method for predicting the differential pressure in the total pressure test inside and outside the battery according to claim 1, wherein the submodels of the N regression algorithms at least comprise a random forest regression submodel, a support vector regression submodel and a Bayesian ridge regression submodel.
3. The method of predicting differential pressure for total pressure tests inside and outside a battery of claim 1, wherein the test data set D further comprises a relationship between differential pressure and impedance between a positive electrode and a negative electrode.
4. The differential pressure prediction method for the internal and external total pressure test of the battery according to claim 3, characterized in that the internal and external total pressure data of the battery are collected through a digital multimeter and a measurement switching module, and the internal and external impedance data of the anode and the external and internal impedance data of the cathode of the battery are collected through the digital multimeter and the measurement switching module; all data are read by computer communication and stored in a database.
5. The pressure difference prediction device for the internal and external total pressure test of the battery is characterized by comprising a digital multimeter and a computer, wherein the digital multimeter is used for acquiring data of the internal total pressure, the external total pressure, the internal and external impedance of a positive electrode and the internal and external impedance of a negative electrode of the battery, and the computer is used for reading the acquired data of the digital multimeter through acquisition software and storing the data in a database.
6. The differential pressure prediction device for the total internal and external battery pressure test according to claim 5, wherein the digital multimeter tests total internal and external battery pressures, total external pressures, internal and external positive electrode impedances and internal and external negative electrode impedances of a plurality of batteries to form a test data set D;
the computer adopts sub models of N regression algorithms, and respectively predicts new battery data according to the test data set D to obtain predicted values V1 and V2 \8230; \8230andVN;
comparing the predicted values V1 and V2 of the differential pressure V8230A V8230VN and VN obtained by each submodel with the actual value VT of the differential pressure tested to obtain the submodel with the minimum error between the predicted value and the actual value as a label;
the pressure difference predicted values V1 and V2 \8230 \ 8230 \ VN and the label obtained by each sub-model are brought into secondary prediction based on support vector regression to obtain N probabilities P1 and P2 \8230 \ 8230and PN corresponding to each sub-model;
and finally, calculating a weighted average value according to predicted values V1 and V2 (8230) \8230; VN and corresponding probability vectors P1 and P2 (8230) \8230and PN of each sub-model N, wherein the weighted average value is used as a final prediction result VF:
VF=（V1* P1+V2* P2……+VN* PN）/（P1+ P2……+ PN）。
7. the device for predicting the differential pressure of the total pressure test inside and outside the battery according to claim 6, wherein the submodels of the N regression algorithms at least comprise a random forest regression submodel, a support vector regression submodel and a Bayesian ridge regression submodel.
8. The device for predicting the differential pressure of the total pressure inside and outside the battery according to claim 7, wherein the test data set D further comprises the internal and external impedance of the positive electrode and the internal and external impedance of the negative electrode of the battery.
9. The device for predicting the pressure difference of the total pressure test inside and outside the battery according to claim 7, wherein the database is stored by a computer hard disk, a server or a cloud.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211017933.2A CN115389955A (en) | 2022-08-24 | 2022-08-24 | Differential pressure prediction method and device for total pressure test inside and outside battery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211017933.2A CN115389955A (en) | 2022-08-24 | 2022-08-24 | Differential pressure prediction method and device for total pressure test inside and outside battery |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115389955A true CN115389955A (en) | 2022-11-25 |
Family
ID=84121222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211017933.2A Pending CN115389955A (en) | 2022-08-24 | 2022-08-24 | Differential pressure prediction method and device for total pressure test inside and outside battery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115389955A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024133594A1 (en) * | 2022-12-20 | 2024-06-27 | Aequilliving Inc Limited | Device and method for capturing a digital signature of a reference product |
-
2022
- 2022-08-24 CN CN202211017933.2A patent/CN115389955A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024133594A1 (en) * | 2022-12-20 | 2024-06-27 | Aequilliving Inc Limited | Device and method for capturing a digital signature of a reference product |
| WO2024133595A1 (en) * | 2022-12-20 | 2024-06-27 | Aequilliving Inc Limited | Device and method for imprinting an active effect of at least one reference product into a treatment product |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112858919B (en) | Battery system online fault diagnosis method and system based on cluster analysis | |
| CN101464964B (en) | Pattern recognition method capable of holding vectorial machine for equipment fault diagnosis | |
| CN112699913A (en) | Transformer area household variable relation abnormity diagnosis method and device | |
| CN111007417A (en) | Battery pack SOH and RUL prediction method and system based on inconsistency evaluation | |
| CN107422266B (en) | Fault diagnosis method and device for high-capacity battery energy storage system | |
| KR20210116801A (en) | Method and apparatus for predicting state of battery health based on neural network | |
| US20120116722A1 (en) | Detection of Defects in an Electrochemical Device | |
| CN109633448B (en) | Method and device for identifying battery health state and terminal equipment | |
| JP2023520970A (en) | Lithium battery SOC estimation method, apparatus, and computer-readable storage medium | |
| CN113010504B (en) | Electric power data anomaly detection method and system based on LSTM and improved K-means algorithm | |
| CN116842459B (en) | Electric energy metering fault diagnosis method and diagnosis terminal based on small sample learning | |
| CN114024328A (en) | Comprehensive evaluation method and device for energy storage power station system and readable medium | |
| CN116522153B (en) | Lithium battery capacity prediction method, lithium battery capacity prediction device, computer equipment and storage medium | |
| CN113554200B (en) | Method, system and equipment for predicting voltage inconsistency of power battery | |
| CN116449218A (en) | Lithium battery health state estimation method | |
| CN115389955A (en) | Differential pressure prediction method and device for total pressure test inside and outside battery | |
| CN117289167A (en) | Battery remaining life prediction method, device and medium based on multiple neural network | |
| CN117458480A (en) | Photovoltaic power generation power short-term prediction method and system based on improved LOF | |
| CN115149123B (en) | Lithium battery module consistency analysis method and system and storage medium | |
| CN111626324A (en) | Seabed observation network data heterogeneous analysis integration method based on edge calculation | |
| CN113466718B (en) | SOC correction method for mobile phone battery | |
| CN115409317A (en) | Transformer area line loss detection method and device based on feature selection and machine learning | |
| CN112485694B (en) | Battery pack detection method and device | |
| CN115598459A (en) | Power failure prediction method for 10kV feeder line fault of power distribution network | |
| Zhang et al. | Design automation for fast, lightweight, and effective deep learning models: A survey |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |