Acoustic composite membrane detection method and system

Abstract

The present application relates to an acoustic composite membrane detection method and system, and the method includes the following steps: based on the model information of the composite membrane to be detected, the corresponding detection requirement information is obtained, and the detection scheme is generated based on the detection requirement information matching; according to the detection scheme, the pretreatment equipment is controlled to clean and pre-treat the composite membrane according to the appearance detection process, and the cleaned composite membrane is dried; according to the detection scheme, the appearance detection equipment is controlled to perform defect detection on the composite membrane according to the appearance detection process, and a defect detection result is generated; based on the detection scheme, the composite membrane is transported to each performance detection equipment in turn, and each performance detection equipment is controlled to perform performance detection on the composite membrane in turn according to the performance detection process, and the performance detection result is integrated and generated; after the defect detection result and the performance detection result are data sorted, a visualization processing is performed to generate a visualization detection report. The present application has the effect of effectively improving the detection accuracy and detection efficiency of the composite membrane.

Description

Technical Field

[0001] The present application relates to the field of composite membrane detection, and in particular to an acoustic composite membrane detection method and system. Background Art

[0002] Acoustic composite membrane is a composite membrane structure composed of multiple materials, with unique acoustic properties, and is widely used in various acoustic devices and products. Acoustic composite membranes are usually composed of multiple materials to achieve excellent acoustic performance. Common materials include silicone rubber, metal mesh, TPEE (thermoplastic elastomer), PEEK (polytetrafluoroethylene), PTFE (polytetrafluoroethylene), PET (polyester), graphene, etc.

[0003] With the development of my country's industry and the continuous improvement of its technological level, the requirements for the structure and performance of acoustic composite membranes in acoustic product lines are gradually increasing. Against this background, non-destructive testing technology has entered the field of vision of many researchers. Non-destructive testing technology refers to the use of changes in heat, sound, light, electricity, magnetism and other reactions caused by abnormal structure or defects of materials, combined with advanced processing methods and special instruments and equipment for detection, without affecting the performance and organizational structure of the object being inspected. However, with the continuous application and development of composite membrane technology, composite membranes have different testing standards and methods due to their different material types, even for composite membrane components of the same type. Therefore, how to quickly and accurately detect composite membranes of various materials is a problem to be solved. Summary of the invention

[0004] In order to solve the above problems, the present application provides an acoustic composite membrane detection method and system.

[0005] In the first aspect, the present application provides an acoustic composite membrane detection method, which adopts the following technical solution:

[0006] An acoustic composite film detection method comprises the following steps:

[0007] Based on the model information of the composite film to be tested, corresponding testing requirement information is obtained, and a testing scheme is generated based on the matching of the testing requirement information, wherein the testing scheme includes at least one testing consumables information, at least one testing equipment parameter and testing process information, wherein the testing process information includes appearance testing process information and performance testing process information, wherein the performance testing process information includes one or more of an air permeability testing process, a water pressure resistance testing process, an acoustic loss testing process, a temperature resistance testing process and a salt resistance testing process;

[0008] According to the inspection plan, the pretreatment equipment is controlled to perform cleaning pretreatment on the composite film according to the appearance inspection process, and the cleaned composite film is dried;

[0009] According to the inspection plan, control the appearance inspection equipment to perform defect inspection on the composite film according to the appearance inspection process and generate defect inspection results;

[0010] Based on the testing scheme, the composite film is sequentially transported to various performance testing equipment, and each performance testing equipment is controlled to sequentially test the performance of the composite film according to the performance testing process, and the performance testing results are integrated and generated;

[0011] After the defect detection results and performance test results are sorted, they are visualized and processed to generate a visual inspection report.

[0012] Preferably, the generation of a detection scheme based on the matching of detection requirement information specifically includes: generating a detection scheme by matching the detection requirement information with a preset scheme matching model, wherein the scheme matching model is a machine learning model trained with historical composite membrane detection data.

[0013] Preferably, the controlling the appearance inspection device according to the inspection scheme to perform defect inspection on the composite film according to the appearance inspection process specifically includes the following steps:

[0014] Control the appearance inspection equipment to collect image information of the composite film to be inspected according to the inspection plan;

[0015] Controlling the appearance inspection device to preprocess the image information according to the appearance inspection process, wherein the preprocessing includes one or more of image denoising, image enhancement, grayscale processing, image correction, image filtering, image threshold separation, image morphological operation, image fusion and image sharpening;

[0016] The preprocessed image information is transmitted to a preset defect detection model for defect detection. The defect detection model is a convolutional neural network model that uses standard composite film images as samples for deep learning iterative training.

[0017] Preferably, controlling each performance testing device to sequentially test the performance of the composite membrane according to the performance testing process includes controlling the acoustic loss testing device to test the acoustic loss performance of the composite membrane according to the acoustic loss testing process, which specifically includes the following steps:

[0018] Obtain historical abnormal detection data of the same type of composite membrane to be detected, and divide the detection frequency bands to generate a primary detection frequency band set and a secondary detection frequency band set;

[0019] Calculate and obtain the proportion of the primary detection frequency band set and the secondary detection frequency band set in the detection frequency band, take the inverse ratio of the primary detection frequency band set and the secondary detection frequency band set as the ratio of the number of extractions of the primary detection frequency band set and the secondary detection frequency band set, and determine the number of extractions of the primary detection frequency band set and the number of extractions of the secondary detection frequency band set;

[0020] The sound loss detection device is controlled to randomly select a corresponding number of detection frequencies from the primary detection frequency band set and the secondary detection frequency band set according to the sound loss detection process, thereby performing sound loss detection.

[0021] Preferably, the acquisition of historical abnormal detection data of the same type of composite membrane to be detected and the division of the detection frequency bands to generate a primary detection frequency band set and a secondary detection frequency band set specifically include the following steps:

[0022] Acquire historical abnormal detection data of the same model of the composite film to be detected, wherein the historical abnormal detection data includes acoustic loss abnormal data and appearance defect data;

[0023] A recurrent neural network model is established, and historical anomaly detection data is used as learning samples for deep learning iterative training to obtain an abnormal frequency band prediction model;

[0024] The defect detection results are input into the abnormal frequency band prediction model to divide the detection frequency band into a primary detection frequency band set and a secondary detection frequency band set.

[0025] Preferably, after data collation of the defect detection results and the performance detection results, visualization processing is performed to generate a visualization detection report, which specifically includes the following steps:

[0026] Perform data cleaning and normalization on defect detection results and performance test results;

[0027] Import defect detection results and performance test results into the pre-set report template to form a standard report;

[0028] Extract abnormal data, draw a visual table and insert it into the standard version report to form a visual detection report. The visual table includes one or more of a curve chart, a line chart, a bar chart, a pie chart, a heat map, a Sanchi chart, a scatter plot, a rose chart and a radar chart.

[0029] In a second aspect, the present application provides an acoustic composite membrane detection system, which adopts the following technical solution:

[0030] An acoustic composite membrane detection system, comprising:

[0031] A scheme decision module, used for acquiring corresponding detection requirement information based on the model information of the composite membrane to be detected, and generating a detection scheme based on the detection requirement information matching, wherein the detection scheme includes at least one detection consumables information, at least one detection equipment parameter and detection process information, wherein the detection process information includes appearance detection process information and performance detection process information, wherein the performance detection process information includes one or more of air permeability detection process, water pressure resistance detection process, sound loss detection process, temperature resistance detection process and salt resistance detection process;

[0032] The pretreatment module is used to control the pretreatment equipment to clean and pretreat the composite film according to the appearance inspection process according to the inspection plan, and to dry the cleaned composite film;

[0033] An appearance inspection module is used to control the appearance inspection equipment to perform defect inspection on the composite film according to the inspection plan and the appearance inspection process, and generate defect inspection results;

[0034] The performance testing module is used to sequentially transport the composite film to various performance testing devices based on the testing scheme, control various performance testing devices to sequentially perform performance testing on the composite film according to the performance testing process, and integrate and generate performance testing results;

[0035] The report arrangement module is used to arrange the defect detection results and performance detection results, and then perform visualization processing to generate a visual detection report.

[0036] Preferably, the appearance detection module includes:

[0037] An image acquisition unit, used to control the appearance inspection device to collect image information of the composite film to be inspected according to the inspection plan;

[0038] An image preprocessing unit, used to control the appearance inspection device to preprocess the image information according to the appearance inspection process, wherein the preprocessing includes one or more of image denoising, image enhancement, grayscale processing, image correction, image filtering, image threshold separation, image morphological operation, image fusion and image sharpening;

[0039] The defect detection unit is used to transmit the preprocessed image information to a preset defect detection model for defect detection. The defect detection model is a convolutional neural network model that uses standard composite film images as samples for deep learning iterative training.

[0040] Preferably, the performance detection module includes a sound loss detection submodule, which is used to control the sound loss detection device to detect the sound loss performance of the composite membrane according to the sound loss detection process, specifically including:

[0041] A frequency band division unit is used to obtain historical abnormal detection data of the same type of composite membrane to be detected, and divide the detection frequency band to generate a main detection frequency band set and a secondary detection frequency band set;

[0042] An extraction calculation unit, used to calculate and obtain the proportion of the main detection frequency band set and the secondary detection frequency band set in the detection frequency band, take the inverse ratio of the main detection frequency band set and the secondary detection frequency band set as the ratio of the number of extractions of the main detection frequency band set and the secondary detection frequency band set, and determine the number of extractions of the main detection frequency band set and the number of extractions of the secondary detection frequency band set;

[0043] The sound loss test unit is used to control the sound loss detection device to randomly select a corresponding number of detection frequencies from the main detection frequency band set and the secondary detection frequency band set according to the sound loss detection process, thereby performing sound loss detection.

[0044] Preferably, the report collating module includes:

[0045] A data processing unit, used for data cleaning and normalization of defect detection results and performance detection results;

[0046] Report making unit, used to import defect detection results and performance test results into a preset report template to form a standard report;

[0047] A visualization collating unit is used to extract abnormal data, draw a visualization table, and insert it into a standard version report to form a visualization detection report. The visualization table includes one or more of a curve chart, a line chart, a bar chart, a pie chart, a heat map, a Sanchi chart, a scatter plot, a rose chart, and a radar chart.

[0048] In summary, the present application includes at least one of the following beneficial technical effects:

[0049] 1. According to the testing demand information of the composite film to be tested, the testing plan is generated through intelligent decision-making, and the testing process is customized to achieve accurate and efficient testing of various types of composite films; the appearance defect detection is carried out after the composite film to be tested is pre-treated and cleaned. The pre-treatment can effectively remove the pollutants attached to the surface of the composite film due to various unexpected factors during the production and processing process, ensuring the accuracy of subsequent test data; while screening out composite films with serious appearance defects through appearance defect detection, targeted testing can be carried out in the subsequent performance testing process based on the defect detection results of the composite film. While improving the accuracy of the test data, the hidden small structural defects and performance defects of the composite film can be more accurately detected, achieving the effect of effectively improving the detection accuracy and efficiency of the composite film;

[0050] 2. By obtaining historical abnormal detection data of the same type of composite membrane, the interval where the composite membrane is prone to acoustic loss performance defects due to process defects, design defects and other unexpected factors can be accurately extracted, and the number of detections can be intelligently allocated, which helps to improve the accuracy and efficiency of acoustic loss performance detection;

[0051] 3. Through the historical abnormal detection data, the recurrent neural network model is trained iteratively from two aspects: abnormal sound loss data and corresponding defect performance. When facing the demand for composite membrane sound loss performance detection, the appearance defect and the composite membrane sound loss performance, which are two conditions with strong correlation, are correlated based on the historical abnormal defect generation data. The prediction results are corrected based on the defect detection results of the composite membrane to determine the main detection frequency band set and the secondary detection frequency band set, which helps to improve the accuracy of sound loss performance detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a method flow chart of an acoustic composite membrane detection method in an embodiment of the present application;

[0053] Figure 2 is a flow chart of a method for performing appearance inspection on a composite film in an embodiment of the present application;

[0054] Figure 3 is a flow chart of a method for performing acoustic damage detection on a composite membrane in an embodiment of the present application;

[0055] Figure 4 is a flow chart of a method for dividing a detection frequency band in an embodiment of the present application;

[0056] Figure 5 is a flow chart of a method for visualizing a report in an embodiment of the present application;

[0057] Figure 6 It is a system block diagram of an acoustic composite membrane detection system in an embodiment of the present application.

[0058] Explanation of the accompanying drawings: 1. Scheme decision module; 2. Preprocessing module; 3. Appearance detection module; 31. Image acquisition unit; 32. Image preprocessing unit; 33. Defect detection unit; 4. Performance detection module; 41. Frequency band division unit; 411. Data acquisition subunit; 412. Abnormal prediction subunit; 413. Frequency band division subunit; 42. Extraction calculation unit; 43. Sound loss test unit; 5. Report compilation module; 51. Data processing unit; 52. Report preparation unit; 53. Visualization compilation unit. DETAILED DESCRIPTION

[0059] The following is combined with Figure 1-6 This application is described in further detail.

[0060] The present application embodiment discloses an acoustic composite membrane detection method. Figure 1 , an acoustic composite membrane detection method, comprising the following steps:

[0061] S1. Generate a solution: Based on the model information of the composite film to be tested, obtain the corresponding test requirement information, and generate a test solution based on the test requirement information. The test solution includes at least one test consumables information, at least one test equipment parameter and test process information. The test process information includes appearance test process information and performance test process information. The performance test process information includes one or more of air permeability test process, water pressure resistance test process, sound loss test process, temperature resistance test process and salt resistance test process.

[0062] S2. Pretreatment: According to the inspection plan, the pretreatment equipment is controlled to perform cleaning pretreatment on the composite film according to the appearance inspection process, and the cleaned composite film is dried;

[0063] S3. Defect detection: according to the detection plan, the appearance detection equipment is controlled to perform defect detection on the composite film according to the appearance detection process, and the defect detection results are generated;

[0064] S4, performance testing: based on the testing scheme, the composite film is sequentially transported to each performance testing device, each performance testing device is controlled to sequentially perform performance testing on the composite film according to the performance testing process, and the performance testing results are integrated and generated;

[0065] S5. Generate report: After sorting the defect detection results and performance test results, perform visualization processing to generate a visual test report. According to the test demand information of the composite film to be tested, intelligent decision-making is made to generate a test plan, and the test process is customized to achieve accurate and efficient testing of various types of composite films; after pre-treatment and cleaning of the composite film to be tested, appearance defect detection is performed. Pre-treatment can effectively remove pollutants attached to the surface of the composite film due to various unexpected factors during the production and processing process, ensuring the accuracy of subsequent test data; while screening out composite films with serious appearance defects through appearance defect detection, targeted testing can be performed in the subsequent performance testing process based on the defect detection results of the composite film. While improving the accuracy of the test data, the hidden small structural defects and performance defects of the composite film can be more accurately detected, achieving the effect of effectively improving the detection accuracy and efficiency of the composite film.

[0066] The generation of a detection scheme based on the matching of detection requirement information specifically includes: generating a detection scheme by matching the detection requirement information with a preset scheme matching model, and the scheme matching model is a machine learning model trained with historical data of composite membrane detection. It should be noted that the specific training steps of the machine learning model are prior art and will not be repeated here. Through the setting of the scheme matching model, it is possible to intelligently match and generate a detection scheme according to the actual situation of the composite membrane to be detected and the detection requirements, which helps to realize intelligent automation of composite membrane detection and helps to improve the detection accuracy and efficiency of composite membranes.

[0067] Reference Figure 2 According to the inspection plan, the appearance inspection equipment is controlled to perform defect inspection on the composite film according to the appearance inspection process, which specifically includes the following steps:

[0068] A1. Image acquisition: Control the appearance inspection equipment to collect image information of the composite film to be inspected according to the inspection plan;

[0069] A2. Image preprocessing: Control the appearance inspection equipment to preprocess the image information according to the appearance inspection process. The preprocessing includes one or more of image denoising, image enhancement, grayscale processing, image correction, image filtering, image threshold separation, image morphological operation, image fusion and image sharpening;

[0070] A3. Appearance defect detection: The preprocessed image information is transmitted to a preset defect detection model for defect detection. The defect detection model is a convolutional neural network model that uses standard composite film images as samples for deep learning iterative training. The training steps of the convolutional neural network model as a machine learning model are prior art and will not be repeated here. Through the setting of the defect detection model, the appearance defects on the composite film can be accurately and efficiently identified. While screening out composite films with serious appearance defects, based on the defect detection results of the composite film, for composite films that have appearance defects but meet the appearance inspection standards, these appearance defects can reveal possible structural defects and performance defects in the composite film, guide the detection bias, and help improve the subsequent performance detection efficiency and detection accuracy.

[0071] Reference Figure 3 , controlling each performance testing device to perform performance testing on the composite membrane in turn according to the performance testing process, including controlling the sound loss testing device to test the sound loss performance of the composite membrane according to the sound loss testing process, specifically including the following steps:

[0072] B1. Frequency band division: Obtain historical abnormal detection data of the same type of composite membrane to be detected, divide the detection frequency band into a main detection frequency band set and a secondary detection frequency band set;

[0073] B2. Calculation times: Calculate and obtain the proportion of the primary detection frequency band set and the secondary detection frequency band set in the detection frequency band, take the inverse ratio of the primary detection frequency band set and the secondary detection frequency band set as the ratio of the number of extractions of the primary detection frequency band set and the secondary detection frequency band set, and determine the number of extractions of the primary detection frequency band set and the number of extractions of the secondary detection frequency band set;

[0074] B3. Sound loss detection: Control the sound loss detection equipment to randomly select a corresponding number of detection frequencies from the primary detection frequency band set and the secondary detection frequency band set according to the sound loss detection process, and thus perform sound loss detection. By obtaining the historical abnormal detection data of the same model of composite membrane, the intervals where the composite membrane is prone to acoustic loss performance defects due to process defects, design defects and other unexpected factors are accurately extracted, and the main detection frequency band set and the secondary detection frequency band set are divided. Then, the proportion of the number of extractions in the two sets is determined according to the proportion of the two sets in the test frequency band. The proportion of the number and the proportion of the set frequency band are inversely proportional. The smaller the proportion of the main detection frequency band set, the better the historical detection results of the composite membrane are, the more stable the quality is, and the acoustic loss performance is excellent. Carrying out more random inspections in the areas with abnormal records will help to more accurately and quickly determine the possible defects of the composite membrane; if the proportion of the main detection frequency band set is large, it means that the historical detection results of the composite membrane are excellent but not stable, and abnormal points may appear in each frequency band. The possibility that the secondary detection frequency band set also has defects is greatly increased. At this time, the secondary detection frequency band set and the main detection frequency band set are close to 1:1. On the basis of leaning towards the main detection frequency band set, a considerable part of the detection times are invested in the secondary detection frequency band set, which helps to improve the accuracy and efficiency of acoustic loss performance detection.

[0075] It should also be emphasized that in the present application, the proportion of the main detection frequency band set in the entire detection frequency band of the composite membrane is definitely smaller than the proportion of the secondary detection frequency band set in the entire detection frequency band of the composite membrane.

[0076] Reference Figure 4 , obtaining historical abnormal detection data of the same type of composite membrane to be detected, dividing the detection frequency band to generate a main detection frequency band set and a secondary detection frequency band set specifically includes the following steps:

[0077] C1. Obtain abnormal data: Obtain historical abnormal detection data of the same model of the composite film to be detected, including abnormal sound loss data and appearance defect data;

[0078] C2. Model building: Establish a recurrent neural network model, use historical anomaly detection data as learning samples to perform deep learning iterative training to obtain an abnormal frequency band prediction model;

[0079] C3. Frequency band division: Input the defect detection results into the abnormal frequency band prediction model to divide the detection frequency bands to generate the main detection frequency band set and the secondary detection frequency band set. Through the historical abnormal detection data, the recurrent neural network model is multi-directionally associated and iteratively trained from two aspects: abnormal sound loss data and corresponding defect performance. When facing the demand for composite membrane sound loss performance detection, based on the calculation of abnormal high probability frequency bands based on its historical abnormal defect generation data, the appearance defects and composite membrane sound loss performance, which are two conditions that have a strong correlation, are associated. Based on the defect detection results of the composite membrane, the prediction results are corrected to determine the main detection frequency band set and the secondary detection frequency band set, which helps to improve the accuracy of sound loss performance detection.

[0080] Reference Figure 5 After sorting the defect detection results and performance test results, the visualization process is performed to generate a visualization test report, which specifically includes the following steps:

[0081] D1. Data preprocessing: Data cleaning and normalization of defect detection results and performance test results;

[0082] D2. Generate report: Import defect detection results and performance test results into the preset report template to form a standard report;

[0083] D3. Visualization: Extract abnormal data, draw a visualization table, insert it into the standard report to form a visualization test report. The visualization table includes one or more of the following: curve chart, line chart, bar chart, pie chart, heat map, Sanchi chart, scatter plot, rose chart and radar chart. Through visualization, the test report can more intuitively reflect the test results of the composite membrane, which helps to improve the readability of the test report.

[0084] The present application also discloses an acoustic composite membrane detection system. Figure 6 , an acoustic composite membrane detection system, comprising:

[0085] A scheme decision module 1 is used to obtain corresponding detection requirement information based on the model information of the composite membrane to be detected, and to generate a detection scheme based on the detection requirement information matching. The detection scheme includes at least one detection consumables information, at least one detection equipment parameter and detection process information. The detection process information includes appearance detection process information and performance detection process information. The performance detection process information includes one or more of air permeability detection process, water pressure resistance detection process, sound loss detection process, temperature resistance detection process and salt resistance detection process;

[0086] Pretreatment module 2, used to control the pretreatment equipment to clean and pretreat the composite film according to the appearance inspection process according to the inspection plan, and to dry the cleaned composite film;

[0087] Appearance inspection module 3, used to control the appearance inspection equipment to perform defect inspection on the composite film according to the inspection plan and the appearance inspection process, and generate defect inspection results;

[0088] The performance testing module 4 is used to sequentially transport the composite film to various performance testing devices based on the testing scheme, control various performance testing devices to sequentially perform performance testing on the composite film according to the performance testing process, and integrate and generate performance testing results;

[0089] The report arrangement module 5 is used to organize the defect detection results and performance test results, and then perform visualization processing to generate a visualization test report. According to the test demand information of the composite film to be tested, the test plan is generated by intelligent decision-making, and the test process is customized to achieve accurate and efficient testing of various types of composite films; the appearance defect test is performed after the composite film to be tested is pre-treated and cleaned. The pre-treatment can effectively remove the pollutants attached to the surface of the composite film due to various unexpected factors during the production and processing process, ensuring the accuracy of subsequent test data; while screening out composite films with serious appearance defects through appearance defect detection, targeted testing can be performed in the subsequent performance test process based on the defect detection results of the composite film. While improving the accuracy of the test data, the hidden small structural defects and performance defects of the composite film can be more accurately detected, achieving the effect of effectively improving the detection accuracy and efficiency of the composite film.

[0090] Reference Figure 6 , the appearance detection module 3 includes:

[0091] An image acquisition unit 31 is used to control the appearance inspection device to collect image information of the composite film to be inspected according to the inspection scheme;

[0092] An image preprocessing unit 32, used to control the appearance inspection device to preprocess the image information according to the appearance inspection process, wherein the preprocessing includes one or more of image denoising, image enhancement, grayscale processing, image correction, image filtering, image threshold separation, image morphological operation, image fusion and image sharpening;

[0093] The defect detection unit 33 is used to transmit the pre-processed image information to a preset defect detection model for defect detection. The defect detection model is a convolutional neural network model that uses a standard composite film image as a sample for deep learning iterative training. Through the setting of the defect detection model, the appearance defects on the composite film can be accurately and efficiently identified. While screening out composite films with serious appearance defects, based on the defect detection results of the composite film, for composite films that have appearance defects but meet the appearance detection standards, these appearance defects can reveal the possible structural defects and performance defects of the composite film, guide the detection bias, and help improve the subsequent performance detection efficiency and detection accuracy.

[0094] Reference Figure 6 The performance detection module 4 includes a sound loss detection submodule, which is used to control the sound loss detection device to detect the sound loss performance of the composite membrane according to the sound loss detection process, specifically including:

[0095] The frequency band division unit 41 is used to obtain historical abnormal detection data of the same type of composite membrane to be detected, and divide the detection frequency band to generate a main detection frequency band set and a secondary detection frequency band set;

[0096] The extraction calculation unit 42 is used to calculate the proportion of the primary detection frequency band set and the secondary detection frequency band set in the detection frequency band, take the inverse ratio of the primary detection frequency band set and the secondary detection frequency band set as the ratio of the number of extractions of the primary detection frequency band set and the secondary detection frequency band set, and determine the number of extractions of the primary detection frequency band set and the number of extractions of the secondary detection frequency band set;

[0097] The sound loss testing unit 43 is used to control the sound loss detection device to randomly select a corresponding number of detection frequencies from the primary detection frequency band set and the secondary detection frequency band set according to the sound loss detection process to perform sound loss detection. By obtaining the historical abnormal detection data of the same model of composite membrane, the intervals where the composite membrane is prone to acoustic loss performance defects due to process defects, design defects and other unexpected factors are accurately extracted, and the main detection frequency band set and the secondary detection frequency band set are divided. Then, the proportion of the number of extractions in the two sets is determined according to the proportion of the two sets in the test frequency band. The proportion of the number and the proportion of the set frequency band are inversely proportional. The smaller the proportion of the main detection frequency band set, the better the historical detection results of the composite membrane are, the more stable the quality is, and the acoustic loss performance is excellent. Carrying out more random inspections in the areas with abnormal records will help to more accurately and quickly determine the possible defects of the composite membrane; if the proportion of the main detection frequency band set is large, it means that the historical detection results of the composite membrane are excellent but not stable, and abnormal points may appear in each frequency band. The possibility that the secondary detection frequency band set also has defects is greatly increased. At this time, the secondary detection frequency band set and the main detection frequency band set are close to 1:1. On the basis of leaning towards the main detection frequency band set, a considerable part of the detection times are invested in the secondary detection frequency band set, which helps to improve the accuracy and efficiency of acoustic loss performance detection.

[0098] Reference Figure 6 , the frequency band division unit 41 includes:

[0099] The data acquisition subunit 411 is used to obtain historical abnormal detection data of the same type of composite film to be detected, and the historical abnormal detection data includes sound loss abnormal data and appearance defect data;

[0100] The abnormality prediction subunit 412 is used to establish a recurrent neural network model, and use the historical abnormality detection data as learning samples to perform deep learning iterative training to obtain an abnormal frequency band prediction model;

[0101] The frequency band division subunit 413 is used to input the defect detection results into the abnormal frequency band prediction model to divide the detection frequency band to generate a main detection frequency band set and a secondary detection frequency band set. Through the historical abnormal detection data, the recurrent neural network model is multi-directionally associated and iteratively trained from two aspects: abnormal sound loss data and corresponding defect performance. When facing the demand for composite membrane sound loss performance detection, based on the calculation of the abnormal high probability frequency band based on its historical abnormal defect generation data, the appearance defect and the composite membrane sound loss performance, which are two conditions that have a strong correlation, are associated. Based on the defect detection results of the composite membrane, the prediction results are corrected to determine the main detection frequency band set and the secondary detection frequency band set, which is helpful to improve the accuracy of sound loss performance detection.

[0102] Reference Figure 6 , the report arrangement module 5 includes:

[0103] A data processing unit 51 is used to perform data cleaning and normalization processing on defect detection results and performance detection results;

[0104] A report making unit 52, used for importing defect detection results and performance detection results into a preset report template to form a standard version report;

[0105] The visualization collating unit 53 is used to extract abnormal data, draw a visualization table, insert it into the standard version report to form a visualization test report, and the visualization table includes one or more of a curve chart, a line chart, a bar chart, a pie chart, a heat map, a Sanchi chart, a scatter plot, a rose chart, and a radar chart. Through visualization processing, the test report can more intuitively reflect the test results of the composite membrane, which helps to improve the readability of the test report.

[0106] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in this field can still combine, add, delete or make other adjustments to the features in the various embodiments of the present invention according to the circumstances without conflict, without making creative work, so as to obtain different other technical solutions that do not deviate from the concept of the present invention in essence, and these technical solutions also belong to the scope of protection of the present invention.

Claims

1. An acoustic composite membrane detection method, characterized in that: The following steps are involved: Based on the model information of the composite film to be tested, corresponding testing requirement information is obtained, and a testing scheme is generated based on the matching of the testing requirement information, wherein the testing scheme includes at least one testing consumables information, at least one testing equipment parameter and testing process information, wherein the testing process information includes appearance testing process information and performance testing process information, wherein the performance testing process information includes one or more of an air permeability testing process, a water pressure resistance testing process, an acoustic loss testing process, a temperature resistance testing process and a salt resistance testing process; According to the inspection plan, the pretreatment equipment is controlled to perform cleaning pretreatment on the composite film according to the appearance inspection process, and the cleaned composite film is dried; According to the inspection plan, control the appearance inspection equipment to perform defect inspection on the composite film according to the appearance inspection process and generate defect inspection results; Based on the testing scheme, the composite film is sequentially transported to various performance testing equipment, and each performance testing equipment is controlled to sequentially test the performance of the composite film according to the performance testing process, and the performance testing results are integrated and generated; After sorting the defect detection results and performance test results, perform visualization processing to generate a visual inspection report; The controlling each performance testing device to sequentially test the performance of the composite membrane according to the performance testing process includes controlling the acoustic loss testing device to test the acoustic loss performance of the composite membrane according to the acoustic loss testing process, specifically including the following steps: Obtain historical abnormal detection data of the same type of composite membrane to be detected, and divide the detection frequency bands to generate a primary detection frequency band set and a secondary detection frequency band set; Calculate and obtain the proportion of the primary detection frequency band set and the secondary detection frequency band set in the detection frequency band, take the inverse ratio of the primary detection frequency band set and the secondary detection frequency band set as the ratio of the number of extractions of the primary detection frequency band set and the secondary detection frequency band set, and determine the number of extractions of the primary detection frequency band set and the number of extractions of the secondary detection frequency band set; Control the sound loss detection device to randomly select a corresponding number of detection frequencies from the primary detection frequency band set and the secondary detection frequency band set to perform sound loss detection according to the sound loss detection process; The method of obtaining historical abnormal detection data of the same type of composite membrane to be detected and dividing the detection frequency bands to generate a primary detection frequency band set and a secondary detection frequency band set specifically includes the following steps: Acquire historical abnormal detection data of the same model of the composite film to be detected, wherein the historical abnormal detection data includes acoustic loss abnormal data and appearance defect data; A recurrent neural network model is established, and historical anomaly detection data is used as learning samples for deep learning iterative training to obtain an abnormal frequency band prediction model; The defect detection results are input into the abnormal frequency band prediction model to divide the detection frequency band into a primary detection frequency band set and a secondary detection frequency band set.

2. The acoustic composite membrane detection method according to claim 1, characterized in that: The generation of a detection scheme based on the matching of detection requirement information specifically includes: generating a detection scheme by matching the detection requirement information with a preset scheme matching model, wherein the scheme matching model is a machine learning model trained with historical composite membrane detection data.

3. The acoustic composite membrane detection method according to claim 1, characterized in that: The control of the appearance inspection equipment according to the inspection scheme to perform defect inspection on the composite film according to the appearance inspection process specifically includes the following steps: Control the appearance inspection equipment to collect image information of the composite film to be inspected according to the inspection plan; Controlling the appearance inspection device to preprocess the image information according to the appearance inspection process, wherein the preprocessing includes one or more of image denoising, image enhancement, grayscale processing, image correction, image filtering, image threshold separation, image morphological operation, image fusion and image sharpening; The preprocessed image information is transmitted to a preset defect detection model for defect detection. The defect detection model is a convolutional neural network model that uses standard composite film images as samples for deep learning iterative training.

4. The acoustic composite membrane detection method according to claim 1, characterized in that: After the defect detection results and the performance detection results are sorted, the visualization processing is performed to generate a visualization detection report, which specifically includes the following steps: Perform data cleaning and normalization on defect detection results and performance test results; Import defect detection results and performance test results into the pre-set report template to form a standard report; Extract abnormal data, draw a visual table and insert it into the standard version report to form a visual detection report. The visual table includes one or more of a curve chart, a line chart, a bar chart, a pie chart, a heat map, a Sanchi chart, a scatter plot, a rose chart and a radar chart.

5. An acoustic composite membrane detection system, characterized in that: include: A scheme decision module (1) is used to obtain corresponding detection requirement information based on the model information of the composite membrane to be detected, and generate a detection scheme based on the detection requirement information matching, wherein the detection scheme includes at least one detection consumable information, at least one detection equipment parameter and detection process information, wherein the detection process information includes appearance detection process information and performance detection process information, and the performance detection process information includes one or more of an air permeability detection process, a water pressure resistance detection process, a sound damage detection process, a temperature resistance detection process and a salt resistance detection process; A pretreatment module (2) is used to control the pretreatment equipment according to the detection scheme to perform cleaning pretreatment on the composite film according to the appearance detection process, and to dry the cleaned composite film; An appearance inspection module (3) is used to control the appearance inspection equipment to perform defect inspection on the composite film according to the inspection plan and the appearance inspection process, and generate defect inspection results; A performance testing module (4) is used to sequentially transport the composite film to various performance testing devices based on the testing scheme, control each performance testing device to sequentially perform performance testing on the composite film according to the performance testing process, and integrate and generate performance testing results; A report arrangement module (5) is used to arrange the defect detection results and the performance detection results, and then perform visualization processing to generate a visualization detection report; The performance detection module (4) includes a sound loss detection submodule, which is used to control the sound loss detection equipment to detect the sound loss performance of the composite film according to the sound loss detection process, specifically including: A frequency band division unit (41) is used to obtain historical abnormality detection data of the same type of composite membrane to be detected, and divide the detection frequency band to generate a main detection frequency band set and a secondary detection frequency band set; An extraction calculation unit (42) is used to calculate and obtain the proportion of the main detection frequency band set and the secondary detection frequency band set in the detection frequency band, take the inverse ratio of the main detection frequency band set and the secondary detection frequency band set as the ratio of the number of extractions of the main detection frequency band set and the secondary detection frequency band set, and determine the number of extractions of the main detection frequency band set and the number of extractions of the secondary detection frequency band set; A sound loss testing unit (43) is used to control the sound loss detection device to randomly select a corresponding number of detection frequencies from the primary detection frequency band set and the secondary detection frequency band set according to the sound loss detection process to perform sound loss detection; The frequency band division unit (41) comprises: A data acquisition subunit (411) is used to obtain historical abnormal detection data of the same type of composite film to be detected, wherein the historical abnormal detection data includes sound loss abnormal data and appearance defect data; The abnormality prediction subunit (412) is used to establish a recurrent neural network model, and use the historical abnormality detection data as learning samples to perform deep learning iterative training to obtain an abnormal frequency band prediction model; The frequency band division subunit (413) is used to input the defect detection result into the abnormal frequency band prediction model to divide the detection frequency band into a main detection frequency band set and a secondary detection frequency band set.

6. The acoustic composite membrane detection system according to claim 5, characterized in that: The appearance detection module (3) comprises: An image acquisition unit (31) is used to control the appearance inspection device to collect image information of the composite film to be inspected according to the inspection scheme; An image preprocessing unit (32) is used to control the appearance detection device to preprocess the image information according to the appearance detection process, wherein the preprocessing includes one or more of image denoising, image enhancement, grayscale processing, image correction, image filtering, image threshold separation, image morphological operation, image fusion and image sharpening; The defect detection unit (33) is used to transmit the preprocessed image information to a preset defect detection model for defect detection, wherein the defect detection model is a convolutional neural network model obtained by deep learning iterative training using a standard composite film image as a sample.

7. The acoustic composite membrane detection system according to claim 5, characterized in that: The report collating module (5) comprises: A data processing unit (51) is used to perform data cleaning and normalization processing on defect detection results and performance detection results; A report making unit (52), used for importing defect detection results and performance detection results into a preset report template to form a standard version report; The visualization arrangement unit (53) is used to extract abnormal data, draw a visualization table, and insert it into a standard version report to form a visualization detection report. The visualization table includes one or more of a curve chart, a line chart, a bar chart, a pie chart, a heat map, a Sanchi map, a scatter plot, a rose chart, and a radar chart.

Images