Intelligent detection system for cleanliness of automobile parts

Abstract

The invention discloses an intelligent detection system for the cleanliness of automobile parts, and aims to efficiently and accurately evaluate the cleanliness of the automobile parts. The system comprises a data acquisition module, an image processing and analysis module, a cleanliness evaluation module, an automatic calibration module, a database and model training module and a report generation and feedback module, through cooperative work of the multiple modules, the intelligence and automation level of cleanliness detection is improved, and the system has wide industrial application value.

Description

Technical Field

[0001] The present invention relates to the field of automobile manufacturing and maintenance, and in particular to an intelligent detection system for the cleanliness of automobile parts, and is particularly suitable for cleanliness evaluation technology based on multi-sensor data acquisition, image processing and intelligent analysis to meet the needs of high-precision and automated industrial detection. Background Art

[0002] With the rapid development of the automotive industry, the precision requirements for parts manufacturing and assembly have gradually increased, and the cleanliness of parts surfaces has become a key factor affecting product performance and reliability. At present, traditional cleanliness detection methods mainly rely on manual detection or single sensor equipment, which has the following shortcomings:

[0003] Low detection accuracy: Manual detection is subject to subjective judgment and is difficult to identify micron-sized particles or slight oil residues. A single sensor device is difficult to cover the comprehensive analysis of multi-dimensional cleanliness indicators.

[0004] Poor environmental adaptability: Traditional detection methods are sensitive to environmental factors such as temperature, humidity, and light, resulting in large data errors and making it difficult to meet industrial site needs.

[0005] Lack of intelligent analysis: The existing detection system fails to effectively combine machine learning and data analysis technologies, and is unable to achieve dynamic optimization of detection data and adaptive updating of models.

[0006] Insufficient efficiency and automation: Traditional testing is mostly done offline, with complex and time-consuming processes, which cannot meet the requirements of modern production lines for real-time testing.

[0007] Therefore, we made improvements to this and proposed an intelligent detection system for the cleanliness of automotive parts. Summary of the invention

[0008] In view of the shortcomings of the prior art, the present invention provides the following technical solution, an intelligent detection system for the cleanliness of automobile parts, comprising the following modules:

[0009] (1) Data acquisition module: Through the coordinated work of multiple sensors, multi-dimensional data collection of the cleanliness of automotive parts is achieved, including:

[0010] Infrared spectrum sensor: Utilizes the infrared light penetration and absorption characteristics of a specific band to detect the chemical composition and distribution of residual oil on the surface of automotive parts. By analyzing the spectral signals of different bands, it can identify the type of oil (such as lubricating oil, cleaning fluid residue) and the concentration gradient of the distribution.

[0011] Micro laser particle analyzer: Through the principle of laser scattering and absorption, it can perform high-precision detection of particles on the surface of parts and components, accurately measure the size, quantity and density of particles, and infer the size distribution range of particles by analyzing the scattered signals. At the same time, it can also filter out background noise signals to effectively avoid the occurrence of false detection;

[0012] Temperature and humidity sensor: used to monitor the temperature and humidity parameters of the detection environment so as to make real-time corrections to environmental variables during data collection.

[0013] (2) Image processing and analysis module: It uses advanced multi-scale feature extraction algorithms and dirt recognition models to achieve high-precision detection and analysis of dirt on the surface of parts. In the multi-scale feature extraction algorithm, image information of different scales is used to comprehensively analyze the dirt characteristics on the surface of parts. Multi-scale feature extraction can adapt to the different sizes and distribution states of dirt particles, especially for the detection of tiny particles (0.01 mm particles), which significantly improves the detection accuracy. The algorithm formula is as follows:

[0014]

[0015] Among them, F(x,y) is the weighted composite value of the dirt feature, G i (x, y) is the image feature response value at different scales, ω i is the feature weight, which is obtained through convolutional neural network training optimization. In the dirt recognition model, the features are trained through deep learning of convolutional neural networks to optimize the distribution of feature weights (i.e., ω i ), thereby achieving accurate classification and detection of dirt.

[0016] (3) Cleanliness evaluation module: Introduce fuzzy comprehensive evaluation method to assign weights and comprehensively evaluate multi-dimensional cleanliness indicators to ensure the objectivity and accuracy of the evaluation results. The fuzzy comprehensive evaluation method is to analyze and score multiple key indicators that affect the cleanliness of parts, including particle concentration (detecting the number and density of particles on the surface of parts through a particle analyzer to evaluate the degree of particle contamination), oil coverage (combining infrared spectrum sensor data to evaluate the distribution area and coverage ratio of surface oil), surface smoothness (calculating the surface smoothness based on the data of the image processing module and identifying scratches or irregular surfaces that may affect cleanliness), and calculate the comprehensive score through the following scoring formula:

[0017]

[0018] Among them, S is the comprehensive cleanliness score, R j is the evaluation value of the jth indicator, P j is the corresponding weight value.

[0019] (4) Automatic calibration module: It uses the closed-loop control principle and can dynamically adjust the equipment parameters based on the following four key processes:

[0020] Real-time detection data acquisition: Continuously collect detection data from each sensor, including particle concentration, oil coverage, temperature and humidity, etc., as basic data for subsequent calibration;

[0021] Calculate the detection deviation value: By comparing the real-time detection data with the pre-set standard value or historical reference data, the detection deviation value is calculated to quantify the possible deviation of the sensor;

[0022] Generate calibration signal based on deviation value: Generate calibration signal based on deviation value to guide dynamic adjustment of equipment parameters. The strength and direction of calibration signal are proportional to the deviation value, ensuring the calibration process is accurate and efficient;

[0023] Adjust sensor sensitivity and filter parameters: The calibration signal optimizes the data acquisition process by adjusting the sensor sensitivity and filter parameters, making the detection results closer to the actual situation.

[0024] (5) Database and model training module: Through cluster analysis and neural network training, the detection data is efficiently processed and optimized to improve the system's recognition efficiency for different types of dirt. First, the original data collected during the operation of the system (including particle concentration, oil composition, surface smoothness, etc.) is stored to form a comprehensive data set. Then, the accumulated data is analyzed through a clustering algorithm to classify the dirt types and distribution characteristics on the surface of the parts. Cluster analysis helps to extract common features from the data, reduce redundant information and improve the generalization ability of the model. Finally, based on neural network technology, the accumulated data is used to train the dirt recognition model. The training goal of the neural network is to minimize the error between the model prediction value and the actual value by optimizing the loss function L(θ). The model formula is as follows:

[0025]

[0026] Among them, L(θ) is the model loss function, N is the number of data samples, and y i is the actual label of the sample, p(y i |θ) is the sample prediction probability.

[0027] (6) Report generation and feedback module: quickly output test results and provide users with data visualization, cleanliness assessment reports, optimization suggestions and archive management functions. After the test is completed, the module quickly processes the data and generates reports, including test data visualization (displaying information such as particle concentration, oil coverage and surface smoothness in the form of bar graphs, line graphs, etc., to make the test results clear), cleanliness assessment reports (based on the calculation results of the cleanliness model, providing comprehensive scores and detailed evaluations of various indicators), optimization suggestions (providing cleaning optimization suggestions based on the test results, recommending appropriate cleaning methods or equipment adjustment parameters), operation log records (detailed records of the test process, including test time, equipment parameters and operation records, for user reference), and can generate PDF documents for easy storage and sharing. At the same time, the module sends the report to the user through remote communication means, supporting instant notification (SMS, email or App push) and cloud sharing modes.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] High-precision cleanliness detection: The infrared spectrum sensor and micro laser particle analyzer in the data acquisition module can accurately detect oil and particulate matter on the surface of components, and the temperature and humidity sensors can be used to correct environmental influences to ensure the accuracy and stability of the test results.

[0030] Multi-dimensional cleanliness assessment: The fuzzy comprehensive evaluation method is used to assign weights to particle concentration, oil coverage and surface smoothness, and the cleanliness score is comprehensively calculated through the cleanliness assessment model to provide users with scientific and intuitive cleaning status assessment results.

[0031] Intelligent data processing and optimization: The image processing and analysis module uses a multi-scale feature extraction algorithm to achieve accurate identification of tiny particles (0.01 mm level); the database and model training module optimizes the model based on a neural network to continuously improve the system's recognition efficiency for different types of dirt.

[0032] Real-time feedback and remote management: The report generation and feedback module can quickly generate visual charts and cleanliness assessment reports, send test results in real time through the remote communication module, and support PDF document generation and archive management, greatly improving information transmission and management efficiency.

[0033] Dynamic adjustment and adaptive calibration: The automatic calibration module is based on the closed-loop control principle, dynamically adjusts the sensor sensitivity and filter parameters, and corrects the operating status of the detection equipment in real time to ensure that the detection results are consistent with the actual situation, thereby improving the system's adaptive ability.

[0034] Full-process closed-loop management: The system integrates detection, analysis, evaluation, calibration, feedback and other functions to form an efficient closed-loop cleanliness management model that covers the entire process from data collection to result feedback, reducing the need for manual intervention while improving detection efficiency and consistency. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the data acquisition module flow provided for this application;

[0036] Figure 2 A schematic diagram of the image processing and analysis module flow provided for this application;

[0037] Figure 3 A schematic diagram of the cleanliness assessment module process provided for this application;

[0038] Figure 4 A schematic diagram of the automatic calibration module process provided for this application;

[0039] Figure 5 Schematic diagram of the database and model training module flow provided for this application;

[0040] Figure 6 Schematic diagram of the report generation and feedback module flow provided for this application;

[0041] Figure 7 A schematic diagram summarizing the module processes provided for this application. DETAILED DESCRIPTION

[0042] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0043] An intelligent detection system for the cleanliness of automotive parts, see Figure 1-6 , including the following modules:

[0044] (1) Data acquisition module: Through the coordinated work of multiple sensors, multi-dimensional data collection of the cleanliness of automotive parts is achieved, including:

[0045] Infrared spectrum sensor: Utilizes the infrared light penetration and absorption characteristics of a specific band to detect the chemical composition and distribution of residual oil on the surface of automotive parts. By analyzing the spectral signals of different bands, it can identify the type of oil (such as lubricating oil, cleaning fluid residue) and the concentration gradient of the distribution.

[0046] Micro laser particle analyzer: Through the principle of laser scattering and absorption, it can perform high-precision detection of particles on the surface of parts and components, accurately measure the size, quantity and density of particles, and infer the size distribution range of particles by analyzing the scattered signals. At the same time, it can also filter out background noise signals to effectively avoid the occurrence of false detection;

[0047] Temperature and humidity sensor: used to monitor the temperature and humidity parameters of the detection environment so as to make real-time corrections to environmental variables during data collection.

[0048] (2) Image processing and analysis module: It uses advanced multi-scale feature extraction algorithms and dirt recognition models to achieve high-precision detection and analysis of dirt on the surface of parts. In the multi-scale feature extraction algorithm, image information of different scales is used to comprehensively analyze the dirt characteristics on the surface of parts. Multi-scale feature extraction can adapt to the different sizes and distribution states of dirt particles, especially for the detection of tiny particles (0.01 mm particles), which significantly improves the detection accuracy. The algorithm formula is as follows:

[0049]

[0050] Among them, F(x,y) is the weighted composite value of the dirt feature, G i (x, y) is the image feature response value at different scales, ω i is the feature weight, which is obtained through convolutional neural network training optimization. In the dirt recognition model, the features are trained through deep learning of convolutional neural networks to optimize the distribution of feature weights (i.e., ω i ), thereby achieving accurate classification and detection of dirt.

[0051] (3) Cleanliness evaluation module: Introduce fuzzy comprehensive evaluation method to assign weights and comprehensively evaluate multi-dimensional cleanliness indicators to ensure the objectivity and accuracy of the evaluation results. The fuzzy comprehensive evaluation method is to analyze and score multiple key indicators that affect the cleanliness of parts, including particle concentration (detecting the number and density of particles on the surface of parts through a particle analyzer to evaluate the degree of particle contamination), oil coverage (combining infrared spectrum sensor data to evaluate the distribution area and coverage ratio of surface oil), surface smoothness (calculating the surface smoothness based on the data of the image processing module and identifying scratches or irregular surfaces that may affect cleanliness), and calculate the comprehensive score through the following scoring formula:

[0052]

[0053] Among them, S is the comprehensive cleanliness score, R j is the evaluation value of the jth indicator, Pj is the corresponding weight value.

[0054] (4) Automatic calibration module: It uses the closed-loop control principle and can dynamically adjust the equipment parameters based on the following four key processes:

[0055] Real-time detection data acquisition: Continuously collect detection data from each sensor, including particle concentration, oil coverage, temperature and humidity, etc., as basic data for subsequent calibration;

[0056] Calculate the detection deviation value: By comparing the real-time detection data with the pre-set standard value or historical reference data, the detection deviation value is calculated to quantify the possible deviation of the sensor;

[0057] Generate calibration signal based on deviation value: Generate calibration signal based on deviation value to guide dynamic adjustment of equipment parameters. The strength and direction of calibration signal are proportional to the deviation value, ensuring the calibration process is accurate and efficient;

[0058] Adjust sensor sensitivity and filter parameters: The calibration signal optimizes the data acquisition process by adjusting the sensor sensitivity and filter parameters, making the detection results closer to the actual situation.

[0059] (5) Database and model training module: Through cluster analysis and neural network training, the detection data is efficiently processed and optimized to improve the system's recognition efficiency for different types of dirt. First, the original data collected during the operation of the system (including particle concentration, oil composition, surface smoothness, etc.) is stored to form a comprehensive data set. Then, the accumulated data is analyzed through a clustering algorithm to classify the dirt types and distribution characteristics on the surface of the parts. Cluster analysis helps to extract common features from the data, reduce redundant information and improve the generalization ability of the model. Finally, based on neural network technology, the accumulated data is used to train the dirt recognition model. The training goal of the neural network is to minimize the error between the model prediction value and the actual value by optimizing the loss function L(θ). The model formula is as follows:

[0060]

[0061] Among them, L(θ) is the model loss function, N is the number of data samples, and y i is the actual label of the sample, p(y i |θ) is the sample prediction probability.

[0062] (6) Report generation and feedback module: quickly output test results and provide users with data visualization, cleanliness assessment reports, optimization suggestions and archive management functions. After the test is completed, the module quickly processes the data and generates reports, including test data visualization (displaying information such as particle concentration, oil coverage and surface smoothness in the form of bar graphs, line graphs, etc., to make the test results clear), cleanliness assessment reports (based on the calculation results of the cleanliness model, providing comprehensive scores and detailed evaluations of various indicators), optimization suggestions (providing cleaning optimization suggestions based on the test results, recommending appropriate cleaning methods or equipment adjustment parameters), operation log records (detailed records of the test process, including test time, equipment parameters and operation records, for user reference), and can generate PDF documents for easy storage and sharing. At the same time, the module sends the report to the user through remote communication means, supporting instant notification (SMS, email or App push) and cloud sharing modes.

[0063] Obviously, the embodiments described above are only some embodiments of the present invention, rather than all embodiments. The preferred embodiments of the present invention are given in the accompanying drawings, but they do not limit the patent scope of the present invention. The present invention can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive. Although the present invention has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific embodiments, or to perform equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of the present invention, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present invention.

Claims

1. An intelligent detection system for the cleanliness of automobile parts, characterized in that: Includes the following modules: The data acquisition module is used to collect physical data related to surface particles, oil residue and cleanliness of automotive parts through a combination of multiple sensors; Image processing and analysis module, which analyzes dirt distribution and characteristics through convolutional neural network model; Cleanliness assessment module, used to conduct multi-dimensional quantitative assessment of component cleanliness based on preset cleanliness standards and collected data; Automatic calibration module, used to dynamically adjust the sensitivity of the detection equipment and environmental interference parameters according to the real-time detection data of the sensor; Database and model training module, used to store detection data and optimize and self-learn the detection algorithm model based on historical data; The report generation and feedback module is used to generate test result reports and provide cleaning improvement suggestions and decision support information.

2. The intelligent detection system for cleanliness of automobile parts according to claim 1, characterized in that: The data acquisition module includes an infrared spectrum sensor for detecting the chemical composition and distribution of oil stains on the surface of parts, a micro laser particle analyzer for accurately measuring the size and density of particles, and a temperature and humidity sensor for monitoring the temperature and humidity of the detection environment to correct the impact of the environment on data acquisition and ensure the accuracy of the detection data.

3. The intelligent detection system for cleanliness of automobile parts according to claim 1, characterized in that: The image processing and analysis module includes calculating dirt features using a multi-scale feature extraction algorithm and a dirt recognition model, which can achieve accurate detection of 0.01 mm particles. The formula is: Among them, F(x,y) is the weighted composite value of the dirt feature, G i (x, y) is the image feature response value at different scales, ω i is the feature weight, which is obtained through convolutional neural network training optimization.

4. The intelligent detection system for cleanliness of automobile parts according to claim 1, characterized in that: The cleanliness evaluation module includes weighting cleanliness indicators, including particle concentration, oil coverage and surface smoothness, through a fuzzy comprehensive evaluation method, and calculating a comprehensive score using the following scoring formula: Among them, S is the comprehensive cleanliness score, R j is the evaluation value of the jth indicator, P j is the corresponding weight value.

5. The intelligent detection system for cleanliness of automobile parts according to claim 1, characterized in that: The automatic calibration module includes, based on the closed-loop control principle, four processes to dynamically adjust the equipment parameters: acquiring real-time detection data, calculating detection deviation values, generating calibration signals according to the deviation values, and adjusting sensor sensitivity and filter parameters to ensure that the detection data is consistent with the actual situation.

6. The intelligent detection system for cleanliness of automobile parts according to claim 1, characterized in that: The database and model training module includes improving the recognition efficiency of the model for different types of dirt through cluster analysis of accumulated data and neural network training, and optimizing the model through the following formula: Among them, L(θ) is the model loss function, N is the number of data samples, and y i is the actual label of the sample, p(y i |θ) is the sample prediction probability.

7. The intelligent detection system for cleanliness of automobile parts according to claim 1, characterized in that: The report generation and feedback module includes the ability to send test results to users in real time via a remote communication module, including visual charts of test data, cleanliness assessment reports, improvement suggestions, and operation logs, and can automatically generate PDF documents for archiving and management.

Images