Crystallinity detection method and related equipment

Abstract

The present application discloses a method for detecting crystallinity and related equipment, and relates to a technical field related to crystallinity. The method for detecting crystallinity includes: determining relevant parameters of a laser ultrasonic detection device, and the laser ultrasonic detection device is used to detect the crystallinity of a polymer to be detected; according to the relevant parameters, emitting a laser to the current detection position of the polymer to be detected, and determining the original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries the intrinsic information of the polymer to be detected; based on the original ultrasonic signal and each ultrasonic detection parameter, determining the effective data after fusion, and determining the crystallinity of the polymer based on the effective data after fusion. The present application realizes online and accurate detection of polymer crystallinity.

Description

Technical Field

[0001] The present application relates to the technical field of crystallinity, and in particular to a crystallinity detection method and related equipment. Background Art

[0002] The ratio of the weight of the crystalline part of a polymer to the total weight of the crystalline polymer is called crystallinity. Crystallinity is important information for understanding the mechanical and chemical properties of crystalline polymers. In current technology, polymer crystallinity is usually detected using detection methods including thermal analysis, X-ray diffraction, and ultrasound. However, the above methods require offline sampling and analysis, or require the sample to be placed in water or in a mold for offline measurement, and generally use a single detection parameter to analyze and detect crystallinity, which results in low accuracy in polymer crystallinity detection.

[0003] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are related technologies. Summary of the invention

[0004] The main purpose of this application is to provide a crystallinity detection method and related equipment, aiming to solve the technical problems in the current technology that a lossy detection mode is required to measure the crystallinity and the detection accuracy is low.

[0005] To achieve the above objectives, the present application proposes a method for detecting crystallinity, the method comprising:

[0006] Determining relevant parameters of a laser ultrasonic testing device, wherein the laser ultrasonic testing device is used to detect the crystallinity of a polymer to be tested;

[0007] According to the relevant parameters, emitting laser light to the current detection position of the polymer to be detected, and determining the original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries the intrinsic information of the polymer to be detected;

[0008] Based on the original ultrasonic signal and each ultrasonic detection parameter, the fused effective data is determined, and based on the fused effective data, the crystallinity of the polymer is determined.

[0009] In one embodiment, the effective ultrasonic detection parameters include ultrasonic detection parameters having a correlation with the polymer crystallinity greater than a preset correlation value, the effective data include effective features, and the step of determining the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter includes at least one of the following:

[0010] Extracting effective features corresponding to each effective ultrasonic detection parameter from the original ultrasonic signal, and performing feature fusion on the effective features of each effective ultrasonic detection parameter to obtain fused effective data;

[0011] The effective ultrasonic detection parameters are combined, and fusion features corresponding to the combined parameters are extracted from the original ultrasonic signal, and the fusion features are used as the effective features.

[0012] In one embodiment, the step of determining the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter, and determining the crystallinity of the polymer based on the fused effective data, further includes:

[0013] Based on the original ultrasonic signal and each ultrasonic detection parameter, a corresponding new input data set is generated, and the new input data set is used as the fused valid data;

[0014] Inputting the new input data set into a preset multimodal fusion deep learning model;

[0015] Processing the new input data set based on a preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity;

[0016] The preset multimodal fusion deep learning model is obtained by performing supervised and semi-supervised combined training on a preset basic model based on first training data with preset labels.

[0017] In one embodiment, the preset overall loss function includes a preset binary cross entropy loss function and a preset mean square error loss function;

[0018] Before the step of processing the new input data set based on the preset multimodal fusion deep learning model to obtain the output result of the polymer crystallinity, the step further includes:

[0019] Inputting the second training data into a preset pseudo-label generation model to obtain second training data with pseudo-labels;

[0020] Combining first training data with preset labels and second training data with pseudo labels to obtain a training data set, performing feature extraction based on each ultrasonic detection parameter on the training data set to obtain an initial feature set;

[0021] Inputting the initial feature set into the classification branch and the fusion data regression branch of the preset basic model respectively;

[0022] Based on the initial feature set and the preset binary cross entropy loss function, determine a first loss result corresponding to the classification branch; and based on the initial feature set and the preset mean square error loss function, determine a second loss result;

[0023] Based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, return to the step of redetermining the first loss result and the second loss result until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

[0024] In one embodiment, the step of determining the second loss result based on the initial feature set and the preset mean square error loss function further includes:

[0025] Determine the key features to focus on based on the preset attention mechanism;

[0026] Based on the initial feature set, the key feature, and a preset mean square error loss function, a second loss result is determined.

[0027] In one embodiment, the step of determining relevant parameters of the laser ultrasonic detection equipment further includes:

[0028] According to the currently set detection accuracy requirements and / or the size-related information of the polymer to be detected, the relevant parameters of the laser ultrasonic detection equipment are determined, and the relevant parameters include one or more of the scanning range, scanning speed, scanning step and average number of times.

[0029] In one embodiment, the step of determining the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter, and determining the crystallinity of the polymer based on the fused effective data, further includes any one of the following:

[0030] After the original ultrasonic signals of each detection position of the polymer to be detected are acquired, based on each original ultrasonic signal and each corresponding ultrasonic detection parameter, the fused effective data is determined, and the crystallinity of the polymer is determined based on the fused effective data;

[0031] After the original ultrasonic signal of any detection position of the polymer to be detected is acquired, the fused effective data is determined based on the original ultrasonic signal and each ultrasonic detection parameter, and the crystallinity of the polymer is determined based on the fused effective data.

[0032] In addition, to achieve the above-mentioned purpose, the present application also proposes a crystallinity detection device, which comprises:

[0033] A first determination module is used to determine relevant parameters of a laser ultrasonic testing device, wherein the laser ultrasonic testing device is used to detect the crystallinity of a polymer to be tested;

[0034] A second determination module is used to emit a laser to a current detection position of the polymer to be detected according to the relevant parameters, and determine an original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries intrinsic information of the polymer to be detected;

[0035] The third determination module is used to determine the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter, and determine the crystallinity of the polymer based on the fused effective data.

[0036] In one embodiment, the effective ultrasonic detection parameter includes an ultrasonic detection parameter with a correlation with the polymer crystallinity greater than a preset correlation value, and the crystallinity detection device is used to achieve at least one of the following:

[0037] Extracting effective features corresponding to each effective ultrasonic detection parameter from the original ultrasonic signal, and performing feature fusion on the effective features of each effective ultrasonic detection parameter to obtain fused effective data;

[0038] The effective ultrasonic detection parameters are combined, and fusion features corresponding to the combined parameters are extracted from the original ultrasonic signal, and the fusion features are used as the effective features.

[0039] In one embodiment, the crystallinity detection device is also used to achieve:

[0040] Based on the original ultrasonic signal and each ultrasonic detection parameter, a corresponding new input data set is generated, and the new input data set is used as the fused valid data;

[0041] Inputting the new input data set into a preset multimodal fusion deep learning model;

[0042] Processing the new input data set based on a preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity;

[0043] The preset multimodal fusion deep learning model is obtained by performing supervised and semi-supervised combined training on a preset basic model based on first training data with preset labels.

[0044] In one embodiment, the preset overall loss function includes a preset binary cross entropy loss function and a preset mean square error loss function;

[0045] The crystallinity detection device is also used to achieve:

[0046] Inputting the second training data into a preset pseudo-label generation model to obtain second training data with pseudo-labels;

[0047] Combining first training data with preset labels and second training data with pseudo labels to obtain a training data set, performing feature extraction based on each ultrasonic detection parameter on the training data set to obtain an initial feature set;

[0048] Inputting the initial feature set into the classification branch and the fusion data regression branch of the preset basic model respectively;

[0049] Based on the initial feature set and the preset binary cross entropy loss function, determine a first loss result corresponding to the classification branch; and based on the initial feature set and the preset mean square error loss function, determine a second loss result;

[0050] Based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, return to the step of redetermining the first loss result and the second loss result until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

[0051] In one embodiment, the crystallinity detection device is also used to achieve:

[0052] Determine the key features to focus on based on the preset attention mechanism;

[0053] Based on the initial feature set, the key feature, and a preset mean square error loss function, a second loss result is determined.

[0054] In one embodiment, the crystallinity detection device is also used to achieve:

[0055] According to the currently set detection accuracy requirements and / or the size-related information of the polymer to be detected, the relevant parameters of the laser ultrasonic detection equipment are determined, and the relevant parameters include one or more of the scanning range, scanning speed, scanning step and average number of times.

[0056] In one embodiment, the crystallinity detection device is further used to achieve any of the following:

[0057] After the original ultrasonic signals of each detection position of the polymer to be detected are acquired, based on each original ultrasonic signal and each corresponding ultrasonic detection parameter, the fused effective data is determined, and the crystallinity of the polymer is determined based on the fused effective data;

[0058] After the original ultrasonic signal of any detection position of the polymer to be detected is acquired, the fused effective data is determined based on the original ultrasonic signal and each ultrasonic detection parameter, and the crystallinity of the polymer is determined based on the fused effective data.

[0059] In addition, to achieve the above-mentioned purpose, the present application also proposes a crystallinity detection device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the crystallinity detection method described above.

[0060] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the crystallinity detection method described above are implemented.

[0061] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the crystallinity detection method described above are implemented.

[0062] One or more technical solutions proposed in this application have at least the following technical effects:

[0063] The present application proposes a crystallinity detection method and related equipment. Compared with the related art in which the crystallinity of a polymer is analyzed and detected by detection methods such as thermal analysis, X-ray diffraction and ultrasound, which results in loss in the detection process and low detection accuracy, in the present application, relevant parameters of a laser ultrasonic detection equipment are determined, and the laser ultrasonic detection equipment is used to detect the crystallinity of the polymer to be detected; according to the relevant parameters, a laser is emitted to the current detection position of the polymer to be detected, and the original ultrasonic signal corresponding to the current detection position is determined, wherein the original ultrasonic signal carries the intrinsic information of the polymer to be detected; based on the original ultrasonic signal and each ultrasonic detection parameter, the fused valid data is determined, and the crystallinity of the polymer is determined based on the fused valid data. It can be understood that in the present application, a laser is emitted to the current detection position of the polymer to be detected by a laser ultrasonic detection device to obtain an original ultrasonic signal reflecting the intrinsic information of the polymer to be detected, that is, the present application can obtain the ultrasonic characteristic information of the polymer by non-contact laser detection means (and the means is non-destructive), and then based on the original ultrasonic signal and each ultrasonic detection parameter, the fused effective data is determined, and the crystallinity of the polymer is determined based on the fused effective data, that is, in the present application, effective fusion (based on the signal fusion of the original ultrasonic signal and each ultrasonic detection parameter) is performed to realize the detection of the crystallinity of the polymer, that is, in the present application, the online detection of the crystallinity of the polymer is realized only after the data corresponding to multiple ultrasonic detection parameters are effectively fused instead of the data corresponding to one ultrasonic detection parameter (avoiding the poor anti-interference ability caused by one and causing low accuracy), thereby solving the technical problem in the related art that offline sampling or even lossy detection mode is required to realize the measurement of crystallinity and the measurement accuracy is low. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0066] Figure 1 A schematic diagram of a crystallinity detection method provided in Example 1 of the present application;

[0067] Figure 2 A schematic diagram of a crystallinity detection method provided in Example 2 of the present application;

[0068] Figure 3 A schematic diagram of the overall process of the embodiment provided in the first embodiment of the present application;

[0069] Figure 4 A schematic diagram of a scene of a laser ultrasonic detection device provided in Example 1 of the present application;

[0070] Figure 5 A schematic diagram of the first fusion method provided in Example 1 of the present application;

[0071] Figure 6 A schematic diagram of the second fusion method provided in Example 1 of the present application;

[0072] Figure 7 Schematic diagram of the equipment structure of the hardware operating environment involved in the crystallinity detection method in the embodiment of the present application;

[0073] Figure 8 Schematic diagram of the model involved in the crystallinity detection method in the embodiment of the present application.

[0074] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0075] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0076] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0077] The present application embodiment provides a method for detecting crystallinity, referring to Figure 1 , Figure 1 This is a schematic flow chart of the first embodiment of the crystallinity detection method of the present application.

[0078] The crystallinity detection method comprises steps S100 to S300:

[0079] Step S100, determining relevant parameters of a laser ultrasonic testing device, wherein the laser ultrasonic testing device is used to test the crystallinity of a polymer to be tested;

[0080] In this embodiment, the execution subject is a crystallinity detection device, and the main components of the crystallinity detection device include: a laser, an ultrasonic receiver, a motion controller, a signal acquisition and processing system, and a storage and display system, etc., wherein:

[0081] Laser: used to generate high-energy laser beams;

[0082] Ultrasonic receiver: used to receive ultrasonic signals reflected from the surface of the polymer being tested (polymer to be tested) (which will be converted into electrical signals later);

[0083] Motion controller: used to adjust the laser scanning mode, etc.

[0084] Signal acquisition and processing system: used to amplify, filter, digitize, and process received electrical signals for subsequent data analysis and processing (the signal acquisition and processing system may be a computing service device with data processing, network communication, and program running functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device capable of achieving the above functions, etc.);

[0085] Storage and display system: used to display and record the internal structure and performance information of the tested polymer (polymer to be tested).

[0086] In this embodiment, the specific application scenario may be:

[0087] It is necessary to detect the crystallinity of the polymer to be tested, such as a film, in order to obtain important information on the mechanical and chemical properties of the polymer. However, the currently commonly used polymer crystallinity detection methods, such as thermal analysis, X-ray diffraction, and ultrasound, are difficult to meet the requirements. This is because: the thermal analysis method determines the crystallinity by measuring the change in thermal properties of the polymer to be tested during the heating or cooling process. This method requires offline sampling and analysis, which will damage the sample; while the X-ray diffraction method infers the crystallinity by analyzing the diffraction spectrum of the polymer to be tested (this method also requires offline sampling and analysis. Although it will not damage the sample, there is a risk of radiation exposure during the detection process). Both methods are relatively complex to operate and require professional personnel to operate and analyze, which is time-consuming and labor-intensive. In addition, the related ultrasonic detection method of polymer crystallinity is only based on the principle that the ultrasonic velocity will change with the different crystallinity (but during the measurement of its sound velocity, the sample needs to be placed in water or in a mold for measurement, and online monitoring cannot be achieved). In addition, the above methods generally analyze and detect crystallinity through a single detection parameter, which cannot provide comprehensive information on crystallinity and may not be able to accurately describe the overall situation of polymer crystallinity, which will lead to limitations in the results (accuracy does not meet the requirements).

[0088] Based on this background, the embodiment of the present application adopts a laser ultrasonic method to realize non-contact online detection of the polymer to be detected, and adopts a multi-ultrasonic parameter feature fusion technology to fuse multiple ultrasonic parameter features reflecting the crystallinity of the polymer, ultimately improving the accuracy and anti-interference ability of the polymer crystallinity measurement.

[0089] Specifically, the laser ultrasonic testing device is used to detect the crystallinity of the polymer to be tested. Therefore, before the test, it is necessary to determine the relevant parameters of the laser ultrasonic testing device (through the control adjustment of the galvanometer, such as Figure 4 As shown), the relevant parameters may specifically be one or more of the scanning range, scanning speed, scanning step, scanning path and average number of times.

[0090] The step of determining the relevant parameters of the laser ultrasonic detection equipment also includes:

[0091] Step S101, determining relevant parameters of the laser ultrasonic detection equipment according to the currently set detection accuracy requirements and / or size-related information of the polymer to be detected, wherein the relevant parameters include one or more of the scanning range, scanning speed, scanning step and average number of times.

[0092] In this embodiment, the relevant parameters of the laser ultrasonic detection equipment can be determined according to the position of the polymer to be detected and the size of the polymer sample to be detected, or the relevant parameters of the laser ultrasonic detection equipment can be determined according to the currently set detection accuracy requirements, or the relevant parameters of the laser ultrasonic detection equipment can be determined according to the currently set detection accuracy requirements and information related to the size of the polymer to be detected.

[0093] Among them, according to the size-related information of the polymer to be detected, the relevant parameters of the laser ultrasonic detection equipment are determined to include:

[0094] Determine the scanning range of the laser ultrasonic testing equipment based on the size-related information of the polymer to be tested;

[0095] According to the current detection accuracy requirements, the relevant parameters of the laser ultrasonic detection equipment are determined to include:

[0096] According to the current detection accuracy requirements, determine the scanning speed, scanning step and average number of times of the laser ultrasonic detection equipment.

[0097] Among them, the higher the detection accuracy requirement, the higher the average number of scans, the smaller the scanning step, and the slower it can be.

[0098] Step S200, emitting laser light to the current detection position of the polymer to be detected according to the relevant parameters, and determining the original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries the intrinsic information of the polymer to be detected;

[0099] In this embodiment, it should be noted that the laser ultrasonic detection equipment uses a laser to generate a corresponding laser beam according to the relevant parameters, and then irradiates the surface of the polymer to be detected (the surface includes many positions, and the current laser irradiation position is used as the current detection position) according to the setting requirements of the relevant parameters. After the surface of the polymer to be detected absorbs the laser energy, it will produce instantaneous expansion and contraction, thereby generating an ultrasonic signal, which is the original ultrasonic signal. It should be noted that the original ultrasonic signal carries the intrinsic information of the polymer to be detected.

[0100] In this embodiment, it should be noted that after obtaining the original ultrasonic signal of the current detection position, the coordinate information of the current detection position and its corresponding original ultrasonic signal are stored, and the two are stored correspondingly. After the detection of the current detection position is completed, the next position to be detected is detected according to parameters such as the scanning step to obtain another original ultrasonic signal. It should be noted that the next position to be detected is determined by a preset scanning path, wherein the preset scanning path can be scanned in a zigzag shape, and the zigzag scanning is the current optimal path planning, which can be orderly and without repeated scanning.

[0101] Step S300, determining fused valid data based on the original ultrasonic signal and each ultrasonic detection parameter, and determining the crystallinity of the polymer based on the fused valid data.

[0102] In this embodiment, based on the original ultrasonic signal and various ultrasonic detection parameters (different fusion methods), the fused effective data is determined, and the effective data can be an effective feature. Then, the crystallinity of the polymer is determined based on the fused effective data.

[0103] Among them, the crystallinity of the polymer can be determined based on the fused effective data and a preset multimodal fusion deep learning model.

[0104] In this embodiment, before determining the valid data after fusion, the original ultrasonic signal collected at each position can also be preprocessed to obtain an ultrasonic signal with a higher signal-to-noise ratio, and then based on the ultrasonic signal with a high signal-to-noise ratio (greater than a preset signal-to-noise ratio) and various ultrasonic detection parameters, the valid data after fusion is determined.

[0105] Specifically, the preprocessing process includes but is not limited to the following methods: preprocessing method, time gain compensation, filtering, and Hilbert transform.

[0106] In this embodiment, the effective ultrasonic detection parameters include ultrasonic detection parameters (which can be measured in advance) having a correlation with the polymer crystallinity greater than a preset correlation value, the effective data include effective features, and the step of determining the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter includes at least one of the following:

[0107] Step S301, extracting effective features corresponding to each effective ultrasonic detection parameter from the original ultrasonic signal, and performing feature fusion on the effective features of each effective ultrasonic detection parameter to obtain fused effective data;

[0108] In this embodiment, the ultrasonic detection parameters reflecting the crystallinity of the polymer include: ultrasonic velocity, ultrasonic attenuation coefficient, ultrasonic attenuation rate, ultrasonic backscatter signal, etc. The effective ultrasonic detection parameters can be determined first, such as ultrasonic velocity, ultrasonic attenuation coefficient, and ultrasonic attenuation rate are effective ultrasonic detection parameters, while ultrasonic backscatter signal is not an effective ultrasonic detection parameter (this is just an example, and ultrasonic backscatter signal can also be an effective ultrasonic detection parameter).

[0109] In this embodiment, a method of feature fusion is provided: Figure 5 As shown, effective features corresponding to each effective ultrasonic detection parameter are extracted from the original ultrasonic signal, such as extracting ultrasonic velocity features from the original ultrasonic signal, or extracting ultrasonic attenuation coefficient features from the original ultrasonic signal, or extracting ultrasonic attenuation rate features from the original ultrasonic signal, etc. After extracting the effective features corresponding to the effective ultrasonic detection parameters, the ultrasonic velocity feature and the ultrasonic attenuation coefficient feature are fused, or the ultrasonic velocity feature and the ultrasonic attenuation rate feature are fused, or the ultrasonic attenuation coefficient feature and the ultrasonic attenuation rate feature are fused, and so on.

[0110] Step S302, combining the effective ultrasonic detection parameters, and extracting the fusion features corresponding to the combined parameters from the original ultrasonic signal, and using the fusion features as the effective features.

[0111] In this embodiment, another method of feature fusion is also provided, such as Figure 6 As shown: firstly, each effective ultrasonic detection parameter is combined or fused, and then the fusion feature corresponding to the combined parameter is extracted from the original ultrasonic signal, and the fusion feature is used as the effective feature.

[0112] For example, first determine each effective ultrasonic detection parameter, such as determining that the ultrasonic velocity, ultrasonic attenuation coefficient, and ultrasonic attenuation rate are effective ultrasonic detection parameters (ultrasonic backscattered signal is not an effective ultrasonic detection parameter), then determine that the ultrasonic velocity and ultrasonic attenuation coefficient are fused into a first combination parameter, the ultrasonic velocity and ultrasonic attenuation rate parameters are fused into a second combination parameter, and the ultrasonic attenuation coefficient and ultrasonic attenuation rate are fused into a third combination parameter, and then extract the fusion feature corresponding to the first combination parameter from the original ultrasonic signal, and then extract the fusion feature corresponding to the second combination parameter from the original ultrasonic signal, and extract the fusion feature corresponding to the third combination parameter from the original ultrasonic signal (in the current embodiment, focus on establishing the connection between ultrasonic detection parameters to enhance the understanding of the characteristics of the polymer material to be detected).

[0113] In this embodiment, the complementarity between different features can be utilized to provide more comprehensive and accurate information to obtain a more accurate crystallinity.

[0114] In this embodiment, the step of determining the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter, and determining the crystallinity of the polymer based on the fused effective data, further includes any one of the following:

[0115] Step S303, after the original ultrasonic signals of each detection position of the polymer to be detected are acquired, based on each original ultrasonic signal and each corresponding ultrasonic detection parameter, the fused valid data is determined, and the crystallinity of the polymer is determined based on the fused valid data;

[0116] Step S304, after the original ultrasonic signal of any detection position of the polymer to be detected is acquired, the fused effective data is determined based on the original ultrasonic signal and each ultrasonic detection parameter, and the crystallinity of the polymer is determined based on the fused effective data.

[0117] In this embodiment, each original ultrasonic signal can be processed after all the original ultrasonic signals at each detection position of the polymer to be detected are acquired, or each original ultrasonic signal can be processed after all the original ultrasonic signals at any detection position of the polymer to be detected are acquired. Specifically, the method to be adopted can be determined according to the busyness of the current crystallinity detection device. For example, if the current crystallinity detection device is too busy, after all the original ultrasonic signals at each detection position of the polymer to be detected are acquired, the fused valid data can be determined based on each original ultrasonic signal and the corresponding ultrasonic detection parameters.

[0118] In this application, an overall embodiment is also provided, such as Figure 3 As shown, the specific steps are as follows:

[0119] Step 1: First, according to the size-related information (including position) of the polymer to be tested, configure the relevant parameters of the laser ultrasonic testing equipment such as the laser scanning mode, wherein the laser scanning mode includes the scanning range, scanning speed, scanning step and average number of times;

[0120] Step 2: According to the set scanning parameters, the first detection position of the polymer to be detected is obtained, the laser is emitted to the first detection position of the polymer to be detected by the laser, and the ultrasonic A-scan signal (original ultrasonic signal) containing the intrinsic information of the polymer is collected and excited by the laser vibrometer;

[0121] Step 3: Obtain the second detection position, and transmit laser and receive ultrasonic signals through the system;

[0122] Step 4: Iterate step 3 until the laser scans to the last position within the scanning range, and emits the laser and receives the corresponding ultrasonic A-scan signal (original ultrasonic signal);

[0123] Step 5: Preprocess the original ultrasonic signal collected at each position to obtain an ultrasonic signal with a high signal-to-noise ratio;

[0124] Step 6: Analyze the ultrasonic signal preprocessed in step 5 to obtain a variety of ultrasonic detection parameter characteristics that can reflect the crystallinity of the polymer at each position;

[0125] Step 7: Perform multimodal feature fusion on multiple ultrasonic detection parameter features that can reflect the crystallinity of the polymer resolved at each position;

[0126] Step 8: Based on the fused multimodal features and the constructed and trained deep learning model, accurate measurement of polymer crystallinity is achieved.

[0127] The present application proposes a crystallinity detection method and related equipment. Compared with the related art in which the crystallinity of a polymer is analyzed and detected by detection methods such as thermal analysis, X-ray diffraction and ultrasound, which results in loss in the detection process and low detection accuracy, in the present application, relevant parameters of a laser ultrasonic detection equipment are determined, and the laser ultrasonic detection equipment is used to detect the crystallinity of the polymer to be detected; according to the relevant parameters, a laser is emitted to the current detection position of the polymer to be detected, and the original ultrasonic signal corresponding to the current detection position is determined, wherein the original ultrasonic signal carries the intrinsic information of the polymer to be detected; based on the original ultrasonic signal and each ultrasonic detection parameter, the fused valid data is determined, and the crystallinity of the polymer is determined based on the fused valid data. It can be understood that in the present application, a laser is emitted to the current detection position of the polymer to be detected by a laser ultrasonic detection device to obtain an original ultrasonic signal reflecting the intrinsic information of the polymer to be detected, that is, the present application can obtain the ultrasonic characteristic information of the polymer by non-contact laser detection means (and the means is non-destructive), and then based on the original ultrasonic signal and each ultrasonic detection parameter, the fused effective data is determined, and the crystallinity of the polymer is determined based on the fused effective data, that is, in the present application, effective fusion (based on the signal fusion of the original ultrasonic signal and each ultrasonic detection parameter) is performed to realize the detection of the crystallinity of the polymer, that is, in the present application, the online detection of the crystallinity of the polymer is realized only after the data corresponding to multiple ultrasonic detection parameters are effectively fused instead of the data corresponding to one ultrasonic detection parameter (avoiding the poor anti-interference ability caused by one and causing low accuracy), thereby solving the technical problem in the related art that offline sampling or even lossy detection mode is required to realize the measurement of crystallinity and the measurement accuracy is low.

[0128] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 The step of determining the effective data after fusion based on the original ultrasonic signal and each ultrasonic detection parameter, and determining the crystallinity of the polymer based on the effective data after fusion, further includes:

[0129] Step A100, generating a corresponding new input data set based on the original ultrasonic signal and each ultrasonic detection parameter, and using the new input data set as fused valid data;

[0130] Step A200, inputting the new input data set into a preset multimodal fusion deep learning model;

[0131] Step A300, processing the new input data set based on a preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity;

[0132] The preset multimodal fusion deep learning model is obtained by performing supervised and semi-supervised combined training on a preset basic model based on first training data with preset labels.

[0133] In this embodiment, it is explained how to process corresponding data in parallel based on a trained preset multimodal fusion deep learning model.

[0134] It should be noted that traditional deep learning models rely only on labeled ultrasound data and are trained by calculating the mean square error (MSE) loss through regression. Under this single supervised learning framework, the model can only learn using limited labeled data, which limits its generalization ability when facing new data.

[0135] In the present embodiment, the preset multimodal fusion deep learning model is obtained by performing supervised and semi-supervised combined training on the preset basic model based on the first training data with preset labels. It can be understood that in the present embodiment, in addition to the first training data based on limited labels, the present embodiment also introduces the training of unlabeled ultrasonic detection parameter data to achieve supervised and semi-supervised combined training of the preset basic model (in the present embodiment, there are both supervised training based on preset labels (true labels) and a semi-supervised mode for relative training based on the true and false labels of the classification branch and the regression branch after the fusion of data), thereby increasing data diversity and improving the accuracy of model training to improve the accuracy of prediction.

[0136] In this embodiment, the preset overall loss function includes a preset binary cross entropy loss function and a preset mean square error loss function; Figure 8 shown.

[0137] Before the step of processing the new input data set based on the preset multimodal fusion deep learning model to obtain the output result of the polymer crystallinity, the step further includes:

[0138] Step B100, inputting the second training data into a preset pseudo-label generation model to obtain the second training data with the pseudo-label;

[0139] In this embodiment, the training process of the preset multimodal fusion deep learning model is specifically described. First, the second training data, i.e., the unlabeled ultrasonic detection parameter data, is input into the preset pseudo-label generation model to obtain the second training data with a pseudo-label, wherein the preset pseudo-label generation model may be a trained model.

[0140] Step B200, combining the first training data with preset labels and the second training data with pseudo labels to obtain a training data set, performing feature extraction based on each ultrasonic detection parameter on the training data set to obtain an initial feature set;

[0141] After obtaining the second training data with the pseudo-label, the first training data with the preset label and the second training data with the pseudo-label are combined as a whole training data set, and then the training data set is subjected to feature extraction based on each ultrasonic detection parameter to obtain an initial feature set (the initial feature set may be a data set consisting of partially fused features, or may be a data set consisting of features before fusion, that is, the initial feature set is obtained based on the original ultrasonic signal).

[0142] Step B300, inputting the initial feature set into the classification branch and the fusion data regression branch of the preset basic model respectively;

[0143] Specifically, the classification branch and the fused data regression branch share the corresponding network front end. For the classification branch (classifier), the corresponding network front end is connected to the classification-specific network layer (such as the fully connected layer and the softmax layer). For the fused data regression branch, the corresponding network front end is connected to the regression-specific network layer (such as the fully connected layer and the linear activation layer).

[0144] Based on different network layers, the initial feature set is respectively input into the classification branch and the fusion data regression branch of the preset basic model.

[0145] Step B400, determining a first loss result corresponding to a classification branch based on the initial feature set and a preset binary cross entropy loss function; and determining a second loss result based on the initial feature set and a preset mean square error loss function;

[0146] In this embodiment, a preset binary cross entropy loss function (BCE, BinaryCrossEntropy function) is set in the classification branch, and a preset mean square error loss function (MSE, MeanSquared Error function) is set in the fusion data regression branch. Figure 8 shown.

[0147] Specifically, in this embodiment, based on the initial feature set (determining each corresponding feature set (fused or non-fused) and the corresponding first weight information) and the preset binary cross entropy loss function, the first loss result corresponding to the classification branch is determined, and based on the initial feature set (determining each corresponding feature set (fused or non-fused) and the corresponding second weight information) and the preset mean square error loss function, the second loss result is determined.

[0148] Wherein, the step of determining the second loss result based on the initial feature set and the preset mean square error loss function further includes:

[0149] Determine the key features to focus on based on the preset attention mechanism;

[0150] Based on the initial feature set, the key feature, and a preset mean square error loss function, a second loss result is determined.

[0151] In this embodiment, key features focused on by a preset attention mechanism are also provided. Furthermore, when determining the second loss result, the second loss result is determined based on the initial feature set, the key features, and the preset mean square error loss function to meet the actual application needs of the user.

[0152] Step B500, based on the first loss result and the second loss result, determine whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, return to the step of redetermining the first loss result and the second loss result until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

[0153] In this embodiment, based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. Specifically, the two losses (preset overall loss function) can be combined in the form of a weighted sum: ( is a weight parameter used to balance the two losses, can be changed), where The preset binary cross entropy loss function is the corresponding first loss result. It is the second loss result corresponding to the preset mean square error loss function.

[0154] In this embodiment, based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, back propagation and optimization are performed based on the first loss result and the second loss result. Specifically, based on the first loss result and the second loss result (total loss), the gradient of the total loss relative to the model parameters is calculated, and the model weights of the preset basic model are updated using an optimization algorithm (such as Adam or SGD optimization algorithm), and then the step of redetermining the first loss result and the second loss result is returned until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

[0155] It should be noted that in the above process, for the classification branch, in addition to being associated with the BCE indicator, it can also be associated with indicators such as accuracy, recall, and F1 score. For the fused data regression branch, in addition to being associated with the MSE indicator, it can also be associated with other regression evaluation indicators, etc., which will not be explained in detail.

[0156] In this embodiment, the corresponding preset multimodal fusion deep learning model can simultaneously learn the classification features corresponding to the ultrasonic detection parameters and the regression features of the fused data, thereby improving the accuracy and robustness of the polymer crystallinity estimation.

[0157] Among them, after the preset multimodal fusion deep learning model is trained, the new input data set is input into the preset multimodal fusion deep learning model, and the new input data set is processed based on the preset multimodal fusion deep learning model to obtain the output result of the polymer crystallinity.

[0158] Specifically, the preset multimodal fusion deep learning model first extracts features in a new input data set through a feature extractor (including a multi-channel input matrix, where a channel represents a different ultrasonic detection parameter data set or ultrasonic detection parameter features). The features in the extracted new input data set may be unfused features of each channel, and then the unfused features are input into a classification branch, which determines the classification weights of the unfused features of each channel (or in the classification branch, the unfused features of each channel are fused, and then the classification weights corresponding to each fused channel after the fusion are determined).

[0159] Similarly, the unfused features are input into the fused data regression branch, which determines the regression weights of the unfused features of each channel).

[0160] In this embodiment, after the classification weight and the regression weight are determined, the output result of the polymer crystallinity is determined based on the classification weight and the regression weight.

[0161] Specifically,

[0162] In this embodiment, a corresponding new input data set is generated based on the original ultrasonic signal and each ultrasonic detection parameter, and the new input data set is used as the valid data after fusion; the new input data set is input into a preset multimodal fusion deep learning model; the new input data set is processed based on the preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity; wherein the preset multimodal fusion deep learning model is obtained by combining supervised and semi-supervised training of a preset basic model based on first training data with preset labels. In this embodiment, the preset multimodal fusion deep learning model obtained based on the supervised and semi-supervised training realizes accurate processing of the new input data set.

[0163] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the crystallinity detection method of the present application. More simple transformations based on this technical concept are all within the protection scope of the present application.

[0164] This application also provides a crystallinity detection device, please refer to Figure 6 The present application also proposes a crystallinity detection device, which comprises:

[0165] A first determination module is used to determine relevant parameters of a laser ultrasonic testing device, wherein the laser ultrasonic testing device is used to detect the crystallinity of a polymer to be tested;

[0166] A second determination module is used to emit a laser to a current detection position of the polymer to be detected according to the relevant parameters, and determine an original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries intrinsic information of the polymer to be detected;

[0167] The third determination module is used to determine the fused effective data based on the original ultrasonic signal and each ultrasonic detection parameter, and determine the crystallinity of the polymer based on the fused effective data.

[0168] In one embodiment, the effective ultrasonic detection parameter includes an ultrasonic detection parameter with a correlation with the polymer crystallinity greater than a preset correlation value, and the crystallinity detection device is used to achieve at least one of the following:

[0169] Extracting effective features corresponding to each effective ultrasonic detection parameter from the original ultrasonic signal, and performing feature fusion on the effective features of each effective ultrasonic detection parameter to obtain fused effective data;

[0170] The effective ultrasonic detection parameters are combined, and fusion features corresponding to the combined parameters are extracted from the original ultrasonic signal, and the fusion features are used as the effective features.

[0171] In one embodiment, the crystallinity detection device is also used to achieve:

[0172] Based on the original ultrasonic signal and each ultrasonic detection parameter, a corresponding new input data set is generated, and the new input data set is used as the fused valid data;

[0173] Inputting the new input data set into a preset multimodal fusion deep learning model;

[0174] Processing the new input data set based on a preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity;

[0175] The preset multimodal fusion deep learning model is obtained by performing supervised and semi-supervised combined training on a preset basic model based on first training data with preset labels.

[0176] In one embodiment, the preset overall loss function includes a preset binary cross entropy loss function and a preset mean square error loss function;

[0177] The crystallinity detection device is also used to achieve:

[0178] Inputting the second training data into a preset pseudo-label generation model to obtain second training data with pseudo-labels;

[0179] Combining first training data with preset labels and second training data with pseudo labels to obtain a training data set, performing feature extraction based on each ultrasonic detection parameter on the training data set to obtain an initial feature set;

[0180] Inputting the initial feature set into the classification branch and the fusion data regression branch of the preset basic model respectively;

[0181] Based on the initial feature set and the preset binary cross entropy loss function, determine a first loss result corresponding to the classification branch; and based on the initial feature set and the preset mean square error loss function, determine a second loss result;

[0182] Based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, return to the step of redetermining the first loss result and the second loss result until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

[0183] In one embodiment, the crystallinity detection device is also used to achieve:

[0184] Determine the key features to focus on based on the preset attention mechanism;

[0185] Based on the initial feature set, the key feature, and a preset mean square error loss function, a second loss result is determined.

[0186] In one embodiment, the crystallinity detection device is also used to achieve:

[0187] According to the currently set detection accuracy requirements and / or the size-related information of the polymer to be detected, the relevant parameters of the laser ultrasonic detection equipment are determined, and the relevant parameters include one or more of the scanning range, scanning speed, scanning step and average number of times.

[0188] In one embodiment, the crystallinity detection device is further used to achieve any of the following:

[0189] After the original ultrasonic signals of each detection position of the polymer to be detected are acquired, based on each original ultrasonic signal and each corresponding ultrasonic detection parameter, the fused effective data is determined, and the crystallinity of the polymer is determined based on the fused effective data;

[0190] After the original ultrasonic signal of any detection position of the polymer to be detected is acquired, the fused effective data is determined based on the original ultrasonic signal and each ultrasonic detection parameter, and the crystallinity of the polymer is determined based on the fused effective data.

[0191] The crystallinity detection device provided by the present application adopts the crystallinity detection method in the above embodiment, which can solve the technical problem of crystallinity detection. Compared with the related art, the beneficial effects of the crystallinity detection device provided by the present application are the same as the beneficial effects of the crystallinity detection method provided by the above embodiment, and the other technical features in the crystallinity detection device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0192] The present application provides a crystallinity detection device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the crystallinity detection method in the above-mentioned embodiment one.

[0193] Reference below Figure 7 , which shows a schematic diagram of the structure of a crystallinity detection device suitable for implementing the embodiment of the present application. The crystallinity detection device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The crystallinity detection device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0194] like Figure 7As shown, the crystallinity detection device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the crystallinity detection device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the crystallinity detection device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a crystallinity detection device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0195] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0196] The crystallinity detection device provided by the present application adopts the crystallinity detection method in the above embodiment, which can solve the technical problem of crystallinity detection. Compared with the related art, the beneficial effects of the crystallinity detection device provided by the present application are the same as the beneficial effects of the crystallinity detection method provided by the above embodiment, and the other technical features in the crystallinity detection device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0197] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0198] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which 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.

[0199] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the crystallinity detection method in the above-mentioned embodiment.

[0200] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0201] The computer-readable storage medium may be included in the crystallinity detection device; or may exist independently without being assembled into the crystallinity detection device.

[0202] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the crystallinity detection device, the crystallinity detection device:

[0203] Receive the current frame image collected by the visual sensor;

[0204] Performing semantic segmentation processing on the current frame image through a preset semantic segmentation model to obtain a semantic segmentation result of the current frame image;

[0205] Determine pixels corresponding to the non-road element category in the semantic segmentation result of the current frame image as occluded pixels, and replace the occluded pixels with target pixels, where the target pixels are pixels corresponding to the road element that appears the most times at the same position as the occluded pixels in the semantic segmentation results of the previous K consecutive frames;

[0206] Based on the replaced current frame image, a structured analysis is performed on the road surface elements in the scene to obtain structured road surface element information.

[0207] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0208] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0209] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0210] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned crystallinity detection method, and can solve the technical problem of crystallinity detection. Compared with the related art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the crystallinity detection method provided in the above-mentioned embodiment, and will not be described in detail here.

[0211] The present application also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the steps of the above-mentioned crystallinity detection method are implemented.

[0212] The computer program product provided in this application can solve the technical problem of crystallinity detection. Compared with the related art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the crystallinity detection method provided in the above embodiment, which will not be repeated here.

[0213] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A method for detecting crystallinity, characterized in that: The method comprises: Determining relevant parameters of a laser ultrasonic testing device, wherein the laser ultrasonic testing device is used to detect the crystallinity of a polymer to be tested; According to the relevant parameters, emitting laser light to the current detection position of the polymer to be detected, and determining the original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries the intrinsic information of the polymer to be detected; Based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, determine the effective data after the fusion of multiple detection parameters, and determine the crystallinity of the polymer based on the fused effective data; The step of determining effective data after fusion of multiple detection parameters based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, and determining the crystallinity of the polymer based on the fused effective data, further includes: Based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, a corresponding new input data set is generated, and the new input data set is used as the fused valid data; Inputting the new input data set into a preset multimodal fusion deep learning model; Processing the new input data set based on a preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity; The preset multimodal fusion deep learning model is obtained by combining supervised and semi-supervised training on a preset basic model based on first training data with preset labels; The preset overall loss function includes a preset binary cross entropy loss function and a preset mean square error loss function; Before the step of processing the new input data set based on the preset multimodal fusion deep learning model to obtain the output result of the polymer crystallinity, the step further includes: Inputting the second training data into a preset pseudo-label generation model to obtain second training data with pseudo-labels; Combining first training data with preset labels and second training data with pseudo labels to obtain a training data set, performing feature extraction based on ultrasonic detection parameters of crystallinity of each reactive polymer on the training data set to obtain an initial feature set; The initial feature set is input into the classification branch and the fusion data regression branch of the preset basic model respectively. ， Wherein, the classification branch and the fusion data regression branch share a network front end; Based on the initial feature set and the preset binary cross entropy loss function, determine a first loss result corresponding to the classification branch; and based on the initial feature set and the preset mean square error loss function, determine a second loss result; Based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, return to the step of redetermining the first loss result and the second loss result until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

2. The method for detecting crystallinity according to claim 1, wherein: The effective ultrasonic detection parameters include ultrasonic detection parameters whose correlation with the polymer crystallinity is greater than a preset correlation value, the effective data include effective features, and the step of determining the effective data after the fusion of multiple detection parameters based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer includes at least one of the following: Extracting effective features corresponding to effective ultrasonic detection parameters of the crystallinity of each reactive polymer from the original ultrasonic signal, and performing feature fusion on the effective features of each effective ultrasonic detection parameter to obtain effective data after fusion of multiple detection parameters; The effective ultrasonic detection parameters are combined, and fusion features corresponding to the combined parameters are extracted from the original ultrasonic signal, and the fusion features are used as the effective features.

3. The method for detecting crystallinity according to claim 1, wherein: The step of determining the second loss result based on the initial feature set and the preset mean square error loss function also includes: Determine the key features to focus on based on the preset attention mechanism; Based on the initial feature set, the key feature, and a preset mean square error loss function, a second loss result is determined.

4. The method for detecting crystallinity according to claim 1, wherein: The step of determining the relevant parameters of the laser ultrasonic detection equipment also includes: According to the currently set detection accuracy requirements and / or the size-related information of the polymer to be detected, the relevant parameters of the laser ultrasonic detection equipment are determined, and the relevant parameters include one or more of the scanning range, scanning speed, scanning step and average number of times.

5. The method for detecting crystallinity according to claim 1, wherein: The step of determining effective data after fusion of multiple detection parameters based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, and determining the crystallinity of the polymer based on the fused effective data, further includes any one of the following: After the original ultrasonic signals of each detection position of the polymer to be detected are acquired, based on each original ultrasonic signal and the ultrasonic detection parameters corresponding to the crystallinity of each reactive polymer, effective data after fusion of multiple detection parameters are determined, and the crystallinity of the polymer is determined based on the fused effective data; After the original ultrasonic signal of any detection position of the polymer to be detected is obtained, the effective data after the fusion of multiple detection parameters is determined based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reactive polymer, and the crystallinity of the polymer is determined based on the fused effective data.

6. A crystallinity detection device, characterized in that: The device comprises: A first determination module is used to determine relevant parameters of a laser ultrasonic testing device, wherein the laser ultrasonic testing device is used to detect the crystallinity of a polymer to be tested; A second determination module is used to emit a laser to a current detection position of the polymer to be detected according to the relevant parameters, and determine an original ultrasonic signal corresponding to the current detection position, wherein the original ultrasonic signal carries intrinsic information of the polymer to be detected; A third determination module is used to determine effective data after fusion of multiple detection parameters based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, and determine the crystallinity of the polymer based on the fused effective data; The step of determining effective data after fusion of multiple detection parameters based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, and determining the crystallinity of the polymer based on the fused effective data, further includes: Based on the original ultrasonic signal and the ultrasonic detection parameters of the crystallinity of each reaction polymer, a corresponding new input data set is generated, and the new input data set is used as the fused valid data; Inputting the new input data set into a preset multimodal fusion deep learning model; Processing the new input data set based on a preset multimodal fusion deep learning model to obtain an output result of polymer crystallinity; The preset multimodal fusion deep learning model is obtained by combining supervised and semi-supervised training on a preset basic model based on first training data with preset labels; The preset overall loss function includes a preset binary cross entropy loss function and a preset mean square error loss function; Before the step of processing the new input data set based on the preset multimodal fusion deep learning model to obtain the output result of the polymer crystallinity, the step further includes: Inputting the second training data into a preset pseudo-label generation model to obtain second training data with pseudo-labels; Combining first training data with preset labels and second training data with pseudo labels to obtain a training data set, performing feature extraction based on ultrasonic detection parameters of crystallinity of each reactive polymer on the training data set to obtain an initial feature set; The initial feature set is input into the classification branch and the fusion data regression branch of the preset basic model respectively. ， Wherein, the classification branch and the fusion data regression branch share a network front end; Based on the initial feature set and the preset binary cross entropy loss function, determine a first loss result corresponding to the classification branch; and based on the initial feature set and the preset mean square error loss function, determine a second loss result; Based on the first loss result and the second loss result, it is determined whether the preset overall loss function is the minimum. If it is determined that the preset overall loss function is not the minimum, return to the step of redetermining the first loss result and the second loss result until the number of training times reaches the preset number of times or the preset overall loss function is minimized, so as to train and obtain the preset multimodal fusion deep learning model.

7. A crystallinity detection device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the crystallinity detection method according to any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the crystallinity detection method according to any one of claims 1 to 5 are implemented.

9. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the crystallinity detection method according to any one of claims 1 to 5 are implemented.

Images