Ultrasonic guided wave measurement system and intelligent prediction method for electrolyte content of soft package battery

Abstract

The invention discloses an ultrasonic guided wave measurement system and intelligent prediction method for the electrolyte content of a soft package battery, and the intelligent prediction method comprises the steps: measuring the soft package battery with the unknown electrolyte content through the ultrasonic guided wave measurement system, and obtaining a time domain signal; the time domain signals are input into a prediction model, the electrolyte content of the soft package battery is obtained, the prediction model is obtained through training of a training set, the training set is a typical feature set-electrolyte content data set, and the prediction model is constructed through a machine learning model or a deep learning model. By means of the trained machine learning model or deep learning model, accurate prediction of the electrolyte content is achieved for the soft package battery with the unknown electrolyte content.

Description

Technical Field

[0001] The present invention belongs to the field of power battery manufacturing and quality inspection, and in particular relates to an ultrasonic guided wave measurement system and an intelligent prediction method for the electrolyte content of a soft-pack battery. Background Art

[0002] Lithium-ion batteries are secondary batteries with high voltage, high energy density, long cycle performance and no memory effect. They are mainly composed of positive electrode, negative electrode, electrolyte and diaphragm. The electrolyte is the carrier of ion transmission in lithium-ion batteries. When the electrolyte content is too low, the electrodes and diaphragm are not completely impregnated, the internal resistance increases and the battery capacity decreases. For batteries with too high electrolyte content, the production cost increases, the energy density decreases and the sealing becomes worse. In addition, a large amount of gas is formed during the charging and discharging process, which increases the internal pressure of the battery and even causes the shell to rupture and the electrolyte to leak. Therefore, the appropriate electrolyte content in the manufacturing process is the key to ensuring the safe operation of the battery, and a matching prediction method is urgently needed.

[0003] The evaluation of soft-pack batteries after encapsulation is somewhat difficult. Non-destructive testing methods based on ultrasound, radiation, neutron diffraction (Neutronradiography) and electrochemical impedance spectroscopy (Electrochemical impedance spectroscopy) can detect the content of the electrolyte without destroying the battery structure, and have been widely used. Among them, radiation is extremely harmful to the human body and requires more complex equipment and higher costs. In addition, the inconvenience of use and high cost limit the popularity of the neutron diffraction method. Furthermore, non-destructive testing based on electrochemical impedance spectroscopy is only applicable to the evaluation of the degree of electrolyte infiltration of soft-pack batteries that have been formed. Ultrasound is sensitive to defects or damage inside soft-pack batteries. Combined with artificial intelligence algorithms, it is expected to accurately predict the electrolyte content of soft-pack batteries.

[0004] At present, nondestructive testing methods based on ultrasonic body waves have attracted the attention of scholars. When the electrolyte content reaches a certain value, the C-scan based on ultrasonic body waves can intuitively present the distribution of the electrolyte, but there are also problems such as insensitivity to soft-pack batteries with low electrolyte content, slow detection speed and complex equipment. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes an ultrasonic guided wave measurement system and an intelligent prediction method for the electrolyte content of a soft-pack battery, which can accurately predict the electrolyte content of the soft-pack battery.

[0006] The present invention also provides an ultrasonic guided wave measurement system for the electrolyte content of a soft-pack battery, comprising: an arbitrary function generator, a power amplifier, a clamping device, an excitation transducer, a receiving transducer and an oscilloscope;

[0007] The arbitrary function generator is used to send an excitation signal;

[0008] The power amplifier is used to amplify the excitation signal;

[0009] The clamping device is used to clamp the excitation transducer and the receiving transducer;

[0010] The excitation transducer is used to convert the amplified excitation signal into a mechanical vibration signal;

[0011] The receiving transducer is used to convert the mechanical vibration signal into an electrical signal;

[0012] The oscilloscope is used to visualize the electrical signal.

[0013] Optionally, the arbitrary function generator is connected to the power amplifier, the power amplifier is respectively connected to the oscilloscope and the excitation transducer, the excitation transducer is mounted on the clamping device, and the receiving transducer is mounted on the clamping device and connected to the oscilloscope.

[0014] Optionally, the excitation transducer is composed of a first piezoelectric ceramic, a first matching layer, a first shielding layer, a first backing layer, and a first BNC connecting wire, wherein the first piezoelectric ceramic is placed on the first matching layer and connected to the first BNC connecting wire, the first shielding layer is placed on the first piezoelectric ceramic, and the first backing layer is placed on the first shielding layer;

[0015] The receiving transducer is composed of a second piezoelectric ceramic, a second matching layer, a second shielding layer, a second backing layer, and a second BNC connecting line, wherein the second piezoelectric ceramic is placed on the second matching layer and connected to the second BNC connecting line, the second shielding layer is placed on the second piezoelectric ceramic, and the second backing layer is placed on the second shielding layer.

[0016] The present invention also provides an ultrasonic guided wave intelligent prediction method for the electrolyte content of a soft-pack battery, comprising:

[0017] Use ultrasonic guided wave measurement system to measure soft-pack batteries with unknown electrolyte content and obtain time domain signals;

[0018] The time domain signal is input into a prediction model to obtain the electrolyte content of the soft-pack battery, wherein the prediction model is obtained by training with a training set, the training set is a data set of a typical feature set - electrolyte content, and the prediction model is constructed using a machine learning model or a deep learning model.

[0019] Optionally, using the ultrasonic guided wave measurement system to measure a soft-pack battery with unknown electrolyte content, obtaining a time domain signal includes:

[0020] Obtaining an incentive signal;

[0021] The mechanical vibration signal is input into a receiving transducer to obtain a time domain signal.

[0022] Optionally, obtaining the training set includes:

[0023] Using the ultrasonic guided wave measurement system to measure a soft-pack battery with a known electrolyte content to obtain an original data set;

[0024] Extracting typical features from the original data set, combining the electrolyte content corresponding to the typical features, and obtaining an initial typical feature set - electrolyte content data set;

[0025] The initial typical feature set-electrolyte content data set is preprocessed to obtain a typical feature set-electrolyte content data set.

[0026] Optionally, the typical features include at least total energy, the number of wave packets, the amplitude of the first wave packet, the arrival time of the first wave packet, the maximum amplitude, the amplitude of the last wave packet and the arrival time of the last wave packet.

[0027] Optionally, preprocessing the data set of the initial typical feature set - electrolyte content includes:

[0028] The data set of the initial typical feature set - electrolyte content is subjected to missing value filling, feature scaling and encoding processing.

[0029] Optionally, inputting the time domain signal into a prediction model to obtain the electrolyte content of the soft pack battery includes:

[0030] Extracting typical features of the time domain signal to obtain a feature set;

[0031] The feature set is input into the prediction model to predict the electrolyte content of the soft-pack battery.

[0032] Compared with the prior art, the present invention has the following advantages and technical effects:

[0033] The present invention performs ultrasonic guided wave detection on soft-pack batteries to form a typical feature set - a data set of electrolyte content, which is divided into a training set and a validation set. With the help of methods such as five-fold cross-validation, a suitable machine learning model or deep learning model and its parameters are screened out. With the help of the trained machine learning model or deep learning model, the electrolyte content of soft-pack batteries with unknown electrolyte content is accurately predicted, and the quality inspection of power batteries is realized in the manufacturing process, ensuring the consistency and reliability of the power batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0035] Figure 1 It is a flow chart of the ultrasonic guided wave intelligent prediction method of the electrolyte content of the soft-pack battery according to an embodiment of the present invention;

[0036] Figure 2 is a schematic diagram of conducting ultrasonic guided wave detection of soft-pack batteries by means of an ultrasonic guided wave measurement system according to an embodiment of the present invention;

[0037] Figure 3 is a schematic diagram of typical time domain signals of soft-pack batteries with different electrolyte contents according to an embodiment of the present invention;

[0038] Figure 4 is a schematic diagram of a confusion matrix for electrolyte content prediction based on a machine learning model according to an embodiment of the present invention;

[0039] Figure 5 It is a schematic diagram of the importance ranking of typical features of the embodiments of the present invention. DETAILED DESCRIPTION

[0040] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0041] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0042] The present invention provides an ultrasonic guided wave measurement system for the electrolyte content of a soft-pack battery, comprising: an arbitrary function generator, a power amplifier, a clamping device, an excitation transducer, a receiving transducer and an oscilloscope;

[0043] An arbitrary function generator, used to generate an excitation signal;

[0044] A power amplifier, used for amplifying the excitation signal;

[0045] A clamping device, used for clamping the excitation transducer and the receiving transducer;

[0046] An excitation transducer, used for converting the amplified excitation signal into a mechanical vibration signal;

[0047] A receiving transducer, used for converting a mechanical vibration signal into an electrical signal;

[0048] Oscilloscope, used to visualize electrical signals.

[0049] Furthermore, the arbitrary function generator is connected to the power amplifier, the power amplifier is respectively connected to the oscilloscope and the excitation transducer, the excitation transducer is mounted on the clamping device, and the receiving transducer is mounted on the clamping device and connected to the oscilloscope.

[0050] Furthermore, the excitation transducer is composed of a first piezoelectric ceramic, a first matching layer, a first shielding layer, a first backing layer, and a first BNC connecting line, wherein the first piezoelectric ceramic is placed on the first matching layer and connected to the first BNC connecting line, the first shielding layer is placed on the first piezoelectric ceramic, and the first backing layer is placed on the first shielding layer;

[0051] The receiving transducer is composed of a second piezoelectric ceramic, a second matching layer, a second shielding layer, a second backing layer, and a second BNC connecting line, wherein the second piezoelectric ceramic is placed on the second matching layer and connected to the second BNC connecting line, the second shielding layer is placed on the second piezoelectric ceramic, and the second backing layer is placed on the second shielding layer.

[0052] Specifically, the ultrasonic guided wave measurement system is composed of an arbitrary function generator, a pulse amplifier, an oscilloscope, a clamping device, an excitation transducer and a receiving transducer, and is used to excite and receive ultrasonic guided waves. Step S1 includes the following features.

[0053] Step S11, the excitation signal is generated by an arbitrary function generator, which is a sine tone burst with a Hanning window, n cycles and a frequency of f. The energy of the signal is enhanced by a pulse amplifier and enters the excitation transducer.

[0054] Step S12, the present invention designs and manufactures matching layers, backing layers and shielding layers of the excitation transducer and the receiving transducer, wherein the center frequencies of the excitation transducer and the receiving transducer are both f.

[0055] Step S13, the clamping device is used to fix the excitation transducer and the receiving transducer. Through different wiring methods, it can adapt to complex working conditions and realize ultrasonic guided wave detection of the battery to be tested. When the A line and the C line are working, the present invention excites and receives ultrasonic guided waves on the upper surface of the soft-pack battery; when the B line and the D line are working, the present invention excites and receives ultrasonic guided waves on the lower surface of the soft-pack battery; when the A line and the D line are working, the present invention excites and receives ultrasonic guided waves on the upper and lower surfaces of the soft-pack battery respectively; when the B line and the C line are working, the present invention excites and receives ultrasonic guided waves on the lower and upper surfaces of the soft-pack battery respectively.

[0056] Step S14, the present invention adjusts the height of the ultrasonic transducer with the help of a lead screw to meet the detection requirements of soft-pack batteries of different thicknesses. The present invention adjusts the position of the right side structure of the clamping device to change the distance between the excitation transducer and the receiving transducer to meet the detection requirements of soft-pack batteries of different lengths. The pressure sensor is used to record the force between the excitation transducer or the receiving transducer and the soft-pack battery, and the force between the two is maintained at F 1 and F 2 between.

[0057] In step S15, the excitation transducer converts the electrical signal into mechanical vibration, and the ultrasonic guided wave is formed in the soft-pack battery and interacts with the defect. The ultrasonic guided wave carrying the defect information enters the receiving transducer and is converted into an electrical signal. The signal is displayed on the oscilloscope and saved in the USB flash drive.

[0058] The present invention also provides an ultrasonic guided wave intelligent prediction method for the electrolyte content of a soft-pack battery, comprising:

[0059] Use ultrasonic guided wave measurement system to measure soft-pack batteries with unknown electrolyte content and obtain time domain signals;

[0060] The time domain signal is input into the prediction model to obtain the electrolyte content of the soft-pack battery, wherein the prediction model is trained by a training set, which is a data set of a typical feature set - electrolyte content, and the prediction model is constructed using a machine learning model or a deep learning model.

[0061] Specifically, step S1, designing an ultrasonic measurement system, a clamping device and an ultrasonic transducer, building an ultrasonic guided wave measurement system, and exciting and receiving ultrasonic guided waves;

[0062] Step S2, prepare x soft-pack batteries with different electrolyte contents, where the electrolyte increment is y ml. Conduct ultrasonic guided wave measurement of the soft-pack battery, measure the same battery z times, and form an original data set with a data volume of x×z;

[0063] Step S3, extracting features such as amplitude and arrival time from the original data set to form a data set consisting of typical features and electrolyte content;

[0064] Step S4, dividing the data set into a training set and a validation set, adjusting the parameters of the model, and completing the training of the machine learning model or the deep learning model;

[0065] Step S5: For soft-pack batteries with unknown electrolyte content, ultrasonic guided wave detection is carried out to extract typical features, and the electrolyte content is predicted with the help of a trained machine learning model or a deep learning model.

[0066] Furthermore, the ultrasonic guided wave measurement system is used to measure the soft-pack battery with unknown electrolyte content, and the time domain signals obtained include:

[0067] Obtaining an incentive signal;

[0068] The mechanical vibration signal is input into the receiving transducer to obtain the time domain signal.

[0069] Furthermore, obtaining a training set includes:

[0070] The ultrasonic guided wave measurement system is used to measure the soft-pack battery with known electrolyte content to obtain the original data set;

[0071] Extract typical features from the original data set, combine the electrolyte content corresponding to the typical features, and obtain an initial typical feature set - electrolyte content data set;

[0072] The initial typical feature set-electrolyte content data set is preprocessed to obtain the typical feature set-electrolyte content data set.

[0073] Specifically, the soft-pack battery is composed of a positive electrode sheet, a negative electrode sheet, a separator, an electrolyte, an aluminum-plastic film, etc. The present invention prepares soft-pack batteries with different electrolyte contents, and performs ultrasonic guided wave measurement with the aid of an ultrasonic guided wave measurement system, a clamping device, and an ultrasonic transducer, etc., to form an original data set. Step S2 includes the following features.

[0074] Step S21, alternately stacking positive electrode sheets, separators, negative electrode sheets, and separators until the required number of layers is reached.

[0075] Step S22, selecting a suitable welding method and adjusting welding parameters to weld the tabs to the positive electrode sheet and the negative electrode sheet of the battery.

[0076] Step S23, according to the size and shape of the dry battery cell, cut out a suitable aluminum-plastic film, seal the three sides of the aluminum-plastic film with a heat sealer, put the welded dry battery cell into the packaged aluminum-plastic film, inject different contents of electrolyte, evacuate and seal with the help of a vacuum pre-sealer, where the increment of the electrolyte content is yml.

[0077] In step S24, the electrolyte gradually infiltrates the dry battery cell due to capillary action, and the time required is generally between 24 hours and 48 hours.

[0078] Step S25, using an ultrasonic guided wave measurement system to measure x soft-pack batteries with different electrolyte contents, each battery is measured z times, and the data volume of the original data set is x×z.

[0079] Specifically, features related to the battery electrolyte content are extracted to form a typical feature set - a data set of electrolyte content. Step S3 includes the following features.

[0080] Step S31, intercepting the required part of the original time domain signal.

[0081] Step S32, extracting typical features of the time domain signal, wherein the typical features include at least total energy, number of wave packets, amplitude of the first wave packet, arrival time of the first wave packet, maximum amplitude, amplitude of the last wave packet, arrival time of the last wave packet, etc.

[0082] In step S33, all typical features are synthesized into a feature set, together with the corresponding electrolyte content, to form a data set of typical feature set-electrolyte content for subsequent model training and evaluation.

[0083] Furthermore, the typical features include at least total energy, the number of wave packets, the amplitude of the first wave packet, the arrival time of the first wave packet, the maximum amplitude, the amplitude of the last wave packet and the arrival time of the last wave packet.

[0084] Furthermore, the data set of the initial typical feature set - electrolyte content is preprocessed including:

[0085] The dataset of the initial typical feature set - electrolyte content - was processed with missing value filling, feature scaling and encoding.

[0086] Specifically, the machine learning model or deep learning model is trained according to the typical feature set - electrolyte content data set. Step S4 includes the following features.

[0087] Step S41, preprocessing the data set of the typical feature set - electrolyte content, the preprocessing process includes operations such as missing value filling, feature scaling and encoding to ensure the quality and consistency of the data set.

[0088] Step S42, divide the preprocessed data set into a training set and a validation set. The training set generally accounts for 80% of the data set of the typical feature set-electrolyte content, and the validation set generally accounts for 20% of the data set of the typical feature set-electrolyte content. The specific division ratio can be adjusted according to actual conditions.

[0089] Step S43, using methods such as five-fold cross validation to evaluate the performance of the machine learning model or the deep learning model and optimize the parameters of the machine learning model or the deep learning model, select a suitable machine learning model or deep learning model and its parameters to ensure the accuracy of the prediction of the electrolyte content of the soft-pack battery.

[0090] Furthermore, the time domain signal is input into the prediction model to obtain the electrolyte content of the soft pack battery including:

[0091] Extract typical features of time domain signals and obtain feature sets;

[0092] The feature set is input into the prediction model to predict the electrolyte content of the soft-pack battery.

[0093] Specifically, a waveguide test is performed on a soft-pack battery with unknown electrolyte content, a feature set of a time domain signal is extracted, and the electrolyte content of the soft-pack battery under test is predicted with the help of a trained machine learning model or a deep learning model. Step S5 includes the following features.

[0094] Step S51, using the ultrasonic guided wave measurement system, conduct guided wave detection on the soft-pack battery with unknown electrolyte content to obtain a complete time domain signal.

[0095] Step S52: extract typical features from the measured time domain signal and form a feature set.

[0096] Step S53, inputting the feature set into a trained machine learning model or a deep learning model to predict the electrolyte content of the tested soft-pack battery.

[0097] The present embodiment is described in detail below with reference to the accompanying drawings:

[0098] Step S1, designing an ultrasonic measurement system, a clamping device and an ultrasonic transducer, building an ultrasonic guided wave measurement system, and exciting and receiving ultrasonic guided waves.

[0099] Furthermore, this embodiment is constructed as follows Figure 2 The ultrasonic measurement system shown in the figure has an excitation signal of a 150kHz short sine pure tone with a Hanning window. The excitation transducer and the receiving transducer are self-made ultrasonic transducers, wherein the matching layer, backing layer and shielding layer of the excitation transducer and the receiving transducer are designed and manufactured with the aid of aluminum blocks, copper meshes, tungsten powders, epoxy resins, etc.

[0100] Step S2, this embodiment prepares 5 soft-pack batteries with different electrolyte contents, wherein the electrolyte increment is 0.5 ml. This embodiment carries out ultrasonic guided wave measurement of soft-pack batteries, and measures the same battery 100 times to form an original data set with a data volume of 500.

[0101] Furthermore, in this embodiment, five ternary lithium-ion batteries with a capacity of 0.5 Ah (including three positive electrode sheets and four negative electrode sheets) were prepared, and ultrasonic guided wave testing was performed on the above batteries. Figure 3 As shown, the contents of the electrolyte are 0.25 ml, 0.75 ml, 1.25 ml, 1.75 ml, and 2.25 ml, respectively, and the increment of the electrolyte is 0.5 ml. In this embodiment, each battery is measured 100 times by means of an ultrasonic guided wave measurement system, and the amount of data in the original data set is 500.

[0102] Step S3, extracting features such as amplitude and arrival time from the original data set to form a data set consisting of typical features and electrolyte content.

[0103] Furthermore, this embodiment selects 7 typical features, including total energy, number of wave packets, maximum amplitude, arrival time of the first wave packet, amplitude of the first wave packet, arrival time of the last wave packet and amplitude of the last wave packet. The arrival time of the first wave packet and the arrival time of the last wave packet reflect the propagation characteristics of ultrasonic guided waves in soft-pack batteries with different electrolyte contents, the maximum amplitude, the amplitude of the first wave packet and the amplitude of the last wave packet reflect the attenuation of ultrasonic guided waves in soft-pack batteries with different electrolyte contents, the number of wave packets reflects the multi-mode and dispersion of ultrasonic guided waves in soft-pack batteries with different electrolyte contents, and the total energy reflects the energy change of ultrasonic guided waves in soft-pack batteries with different electrolyte contents.

[0104] Step S4, divide the data set of the typical feature set - electrolyte content into a training set and a validation set, adjust the parameters of the model, and complete the training of the machine learning model.

[0105] Furthermore, this embodiment predicts the electrolyte content in the soft-pack battery with the help of a random forest model. Random forest is an ensemble learning method based on decision trees, which performs label classification by constructing multiple decision trees. This ensemble learning method is suitable for multi-feature and multi-category classification problems and can effectively process high-dimensional features and large-scale data sets.

[0106] This embodiment uses the five-fold cross-validation method to train and validate the machine learning model, and divides the data set into 5 subsets. In the first fold, the first subset, the second subset, the third subset, and the fourth subset are used as training sets, and the fifth subset is used as a validation set; in the second fold, the first subset, the second subset, the third subset, and the fifth subset are used as training sets, and the fourth subset is used as a validation set; this cycle repeats, and each subset is used as a validation set. Five-fold cross-validation ensures the diversity of the validation set and the training set, and effectively avoids the occurrence of overfitting.

[0107] In this embodiment, 80% of the data set of the typical feature set - electrolyte content is used as the training set, and 20% of the data set of the typical feature set - electrolyte content is used as the validation set. The validation set is not used in the model training process, which ensures the objectivity of the results and the generalization ability of the model. The training set includes 400 data, of which 240 are related to unwetted batteries (the content of the electrolyte is 0.25ml, 0.75ml and 1.25ml), 80 are related to poorly wetted batteries (the content of the electrolyte is 1.75ml), and 80 are related to well-wetted batteries (the content of the electrolyte is 2.25ml).

[0108] For the parameter adjustment of the random forest model, select appropriate hyperparameters (such as the number of decision trees and the maximum depth of the decision tree, etc.). For the selection of the machine learning model, compare the predicted value with the true value, calculate the recognition rate of the electrolyte content of the soft-pack battery, and select the machine learning model with the highest recognition rate.

[0109] like Figure 4 As shown, this embodiment evaluates the random forest model with the help of confusion matrix, and the percentage of different cells reflects the relationship between the predicted value and the true value. The results show that the random forest model misidentified 2 samples with an average accuracy of 98%. Among them, the recognition rate of soft-pack batteries with injection volumes of 0.25ml, 1.25ml, 1.75ml and 2.25ml can reach 100%; for 3 types of batteries with different electrolyte contents, the recognition rates of unwetted batteries (electrolyte content of 0.25ml, 0.75ml and 1.25ml), poorly wetted batteries (batteries with an electrolyte content of 1.75ml) and well-wetted batteries (batteries with an electrolyte content of 2.25ml) have all reached 100%.

[0110] like Figure 5 As shown, this embodiment analyzes the above typical features, among which the amplitude of the last wave packet, the maximum amplitude and the arrival time of the last wave packet are the key features in the ultrasonic guided wave intelligent prediction method of the electrolyte content of the soft-pack battery. The above features may effectively reflect the distribution of the electrolyte inside the soft-pack battery.

[0111] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. Ultrasonic guided wave measurement system for soft pack battery electrolyte content, characterized in that: include: Arbitrary function generator, power amplifier, clamping device, excitation transducer, receiving transducer and oscilloscope; The arbitrary function generator is used to send an excitation signal; The power amplifier is used to amplify the excitation signal; The clamping device is used to clamp the excitation transducer and the receiving transducer; The excitation transducer is used to convert the amplified excitation signal into a mechanical vibration signal; The receiving transducer is used to convert the mechanical vibration signal into an electrical signal; The oscilloscope is used to visualize the electrical signal.

2. The ultrasonic guided wave measurement system for the electrolyte content of a soft-pack battery according to claim 1, characterized in that: The arbitrary function generator is connected to the power amplifier, the power amplifier is respectively connected to the oscilloscope and the excitation transducer, the excitation transducer is mounted on the clamping device, and the receiving transducer is mounted on the clamping device and connected to the oscilloscope.

3. The ultrasonic guided wave measurement system for the electrolyte content of a soft-pack battery according to claim 2, characterized in that: The excitation transducer is composed of a first piezoelectric ceramic, a first matching layer, a first shielding layer, a first backing layer, and a first BNC connecting line, wherein the first piezoelectric ceramic is placed on the first matching layer and connected to the first BNC connecting line, the first shielding layer is placed on the first piezoelectric ceramic, and the first backing layer is placed on the first shielding layer; The receiving transducer is composed of a second piezoelectric ceramic, a second matching layer, a second shielding layer, a second backing layer, and a second BNC connecting line, wherein the second piezoelectric ceramic is placed on the second matching layer and connected to the second BNC connecting line, the second shielding layer is placed on the second piezoelectric ceramic, and the second backing layer is placed on the second shielding layer.

4. An ultrasonic guided wave intelligent prediction method for the electrolyte content of soft-pack batteries, characterized in that: include: Using the ultrasonic guided wave measurement system as described in any one of claims 1 to 3 to measure a soft-pack battery with unknown electrolyte content to obtain a time domain signal; The time domain signal is input into a prediction model to obtain the electrolyte content of the soft-pack battery, wherein the prediction model is obtained by training with a training set, the training set is a data set of a typical feature set - electrolyte content, and the prediction model is constructed using a machine learning model or a deep learning model.

5. The ultrasonic guided wave intelligent prediction method for the electrolyte content of soft-pack batteries according to claim 4, characterized in that: The ultrasonic guided wave measurement system is used to measure a soft-pack battery with unknown electrolyte content, and the time domain signal is obtained including: Obtaining an incentive signal; The mechanical vibration signal is input into a receiving transducer to obtain a time domain signal.

6. The ultrasonic guided wave intelligent prediction method for the electrolyte content of soft-pack batteries according to claim 4, characterized in that: Acquiring the training set includes: Using the ultrasonic guided wave measurement system to measure a soft-pack battery with a known electrolyte content to obtain an original data set; Extracting typical features from the original data set, combining the electrolyte content corresponding to the typical features, and obtaining an initial typical feature set-electrolyte content data set; The initial typical feature set-electrolyte content data set is preprocessed to obtain a typical feature set-electrolyte content data set.

7. The ultrasonic guided wave intelligent prediction method for the electrolyte content of a soft-pack battery according to claim 6, characterized in that: The typical features include at least total energy, the number of wave packets, the amplitude of the first wave packet, the arrival time of the first wave packet, the maximum amplitude, the amplitude of the last wave packet and the arrival time of the last wave packet.

8. The ultrasonic guided wave intelligent prediction method for the electrolyte content of a soft-pack battery according to claim 6, characterized in that: Preprocessing the data set of the initial typical feature set - electrolyte content includes: The data set of the initial typical feature set - electrolyte content is subjected to missing value filling, feature scaling and encoding processing.

9. The ultrasonic guided wave intelligent prediction method for the electrolyte content of a soft-pack battery according to claim 4, characterized in that: Inputting the time domain signal into the prediction model to obtain the electrolyte content of the soft pack battery includes: Extracting typical features of the time domain signal to obtain a feature set; The feature set is input into the prediction model to predict the electrolyte content of the soft-pack battery.

Images