System and method for detecting apparent defects of automobile engine cylinder cover

Abstract

The invention discloses an apparent defect detection system for an automobile engine cylinder cover, which comprises a multi-modal data acquisition module, a data fusion and processing module, a defect identification and classification module and a result output and feedback module, comprise but not limited to optical image data, ultrasonic detection data, thermal imaging data and electromagnetic induction data. According to the apparent defect detection system and method for the automobile engine cylinder cover, various apparent defects, including surface visible defects, internal hidden defects and tiny defects, of the engine cylinder cover can be comprehensively detected by fusing multiple detection technologies of optics, ultrasound, thermal imaging and electromagnetic induction, the detection precision is remarkably improved, and the missing detection rate and the false detection rate are reduced; the whole detection process is automatic, the detection time is greatly shortened through multi-modal data acquisition and parallel processing, the rapid detection requirement on a large-scale production line is met, and the production efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of automotive manufacturing quality inspection technology, specifically to a system and method for detecting apparent defects in automotive engine cylinder heads. Background Technology

[0002] In the production and manufacturing of automobile engines, the cylinder head is a key component, and its quality directly affects the performance and reliability of the engine. The working environment of the cylinder head is harsh, and it must withstand high temperature, high pressure and complex mechanical stress. Therefore, its surface quality is of paramount importance. If the cylinder head has surface defects such as sand holes, cracks, air holes, shrinkage, etc., it will lead to air leakage, water leakage, and oil leakage, reduce engine power, increase fuel consumption, and even cause serious safety accidents.

[0003] Currently, traditional methods for detecting surface defects in engine cylinder heads mainly include manual visual inspection and simple automated inspection equipment. Manual visual inspection relies on the experience and eyesight of the inspectors, and its accuracy and efficiency are greatly affected by human factors, easily leading to missed or false detections. Simple automated inspection equipment is mostly based on a single inspection technology, such as relying solely on optical imaging to detect visible surface defects, and cannot detect internal hidden defects. Its ability to detect complex surface textures and minute defects is limited, making it difficult to meet the high-precision, high-efficiency inspection requirements of the modern automotive manufacturing industry. Therefore, we propose a surface defect detection system for automotive engine cylinder heads. Summary of the Invention

[0004] The purpose of this invention is to provide a system and method for detecting apparent defects in automobile engine cylinder heads, in order to solve the problems mentioned in the background art, such as low detection accuracy, low efficiency, and inability to comprehensively detect various defects in existing automobile engine cylinder head appearance defect detection.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a surface defect detection system for automobile engine cylinder heads, comprising a multimodal data acquisition module, a data fusion and processing module, a defect identification and classification module, and a result output and feedback module. The multimodal data acquisition module is used to acquire multimodal data of the engine cylinder head, including but not limited to optical image data, ultrasonic testing data, thermal imaging data, and electromagnetic induction data. The data fusion and processing module is connected to the multimodal data acquisition module, receives and fuses the multimodal data, and preprocesses the fused data, including noise reduction, enhancement, and feature extraction operations. The defect identification and classification module is connected to the data fusion and processing module, analyzes the preprocessed data based on a deep learning algorithm, identifies surface defects of the cylinder head, and classifies the defects. The result output and feedback module is connected to the defect identification and classification module, outputs defect detection results, including the location, type, and severity information of the defects, and feeds the detection results back to the production system for quality control and adjustment during the production process.

[0006] Preferably, the multimodal data acquisition module includes: a high-resolution optical imaging unit equipped with multiple high-resolution cameras at different angles to acquire high-definition optical images of the engine cylinder head surface to detect visible defects such as scratches, pits, and pinholes; an ultrasonic detection unit using a phased array ultrasonic probe capable of emitting and receiving ultrasonic waves of different frequencies, and detecting internal cracks, porosity, and shrinkage defects by analyzing the propagation characteristics of ultrasonic waves inside the cylinder head; a thermal imaging unit using infrared thermal imaging technology to monitor the temperature distribution on the cylinder head surface in real time, identifying potential defects such as abnormal heat conduction caused by internal defects through temperature anomalies; and an electromagnetic induction detection unit that detects electromagnetic induction changes caused by defects by applying an alternating magnetic field to the cylinder head, used to detect surface and near-surface defects such as cracks and inclusions.

[0007] Preferably, the data fusion and processing module adopts a deep learning-based multimodal fusion network, which includes multiple sub-networks that extract features from data of different modalities. Then, the features extracted by each sub-network are fused through a fusion layer to obtain a more comprehensive and accurate representation of defect features.

[0008] Preferably, the defect identification and classification module adopts a deep learning model based on a combination of convolutional neural networks (CNN) and recurrent neural networks (RNN). CNN is used to extract spatial features of defects, and RNN is used to process time series features (such as data changes during dynamic detection). Through the collaborative work of the two, the accuracy of defect identification and classification is improved.

[0009] Preferably, the result output and feedback module includes: a visualization interface: displaying the test results in an intuitive graphical interface, including a 3D model of the cylinder head, with the location, type, and severity of defects marked on the model for easy viewing and analysis by operators; a report generation system: automatically generating a detailed test report, including defect statistics, key data from the testing process, and quality analysis suggestions, with the report stored in multiple formats for easy subsequent review and traceability; and a report generation system that automatically generates a detailed test report, including defect statistics, key data from the testing process, and quality analysis suggestions, with the report stored in multiple formats for easy subsequent review and traceability.

[0010] A method for detecting apparent defects in an automobile engine cylinder head includes the following steps: Step 1: Place the engine cylinder head on a testing platform with automatic positioning and calibration functions, ensuring the cylinder head is in the correct testing position and orientation. Then, sequentially activate the high-resolution optical imaging unit, ultrasonic testing unit, thermal imaging unit, and electromagnetic induction testing unit. The high-resolution camera captures high-definition optical images of the cylinder head surface from multiple angles; the phased-array ultrasonic probe emits ultrasonic waves towards the cylinder head and receives the reflected wave signals; the thermal imaging unit monitors the temperature distribution on the cylinder head surface in real time, generating thermal imaging data; and the electromagnetic induction testing unit applies an alternating magnetic field to collect electromagnetic induction data. The data collected by each unit is transmitted in real time to the data fusion and processing module via a high-speed data transmission interface.

[0011] Step Two: After receiving the multimodal data, the data fusion and processing module first preprocesses each modality of data, such as denoising and contrast enhancement for optical images, filtering for ultrasonic data, and temperature calibration for thermal imaging data. A deep learning-based multimodal fusion network is then used to input the preprocessed modal data into the corresponding sub-networks for feature extraction. Each sub-network learns the feature representations of different modalities, and then these features are fused through a fusion layer to generate a comprehensive defect feature vector.

[0012] Step 3: Input the fused defect feature vector into a deep learning model based on a combination of CNN and RNN. The CNN layer extracts spatial features from the feature vector, capturing the shape and edge features of the defect; the RNN layer processes the time-series features in the feature vector, learning the dynamic changes of the defect. Through the multi-layer processing of the model, the final output is the defect identification and classification results, determining whether there is a defect in the cylinder head and the specific type of defect, such as cracks, pinholes, or air holes.

[0013] Step Four: The results output and feedback module receives the output results from the defect identification and classification module. It displays the cylinder head as a 3D model on the visualization interface, accurately marking the location, type, and severity of defects on the model. The report generation system automatically generates a detailed inspection report, including defect statistical charts, key data from the inspection process, and quality analysis suggestions for different defect types. The report is stored in common formats such as PDF and Excel for easy access and traceability by production and quality control personnel. The feedback control system relays defect information to the production system in real time based on the inspection results. For cylinder heads with severe defects, the system automatically sends a signal to pause the relevant production line, prompting production personnel to inspect and adjust the equipment. For general defects, the system provides optimization suggestions to the production system based on the defect type and location, such as adjusting casting process parameters and machining process parameters, to achieve closed-loop quality control of the production process.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This automotive engine cylinder head surface defect detection system and method integrates multiple detection technologies, including optics, ultrasound, thermal imaging, and electromagnetic induction, to comprehensively detect various surface defects in engine cylinder heads, including visible surface defects, internal hidden defects, and minute defects. This significantly improves detection accuracy and reduces missed and false detection rates. The entire detection process is automated, and multimodal data acquisition and parallel processing greatly shorten detection time, meeting the rapid detection needs of large-scale production lines and improving production efficiency. A deep learning-based defect identification and classification module accurately identifies defect types and assesses their severity, providing strong data support for quality control and product improvement during production. Detection results are fed back to the production system in real time, automatically triggering corresponding production adjustment measures based on defect conditions. This achieves closed-loop quality control in the production process, reducing defect generation at the source and improving product quality and reliability. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall structure of the system of the present invention; Figure 2 This is a flowchart of the detection method of the present invention; Figure 3 This is a schematic diagram of the multimodal data acquisition module of the present invention; Figure 4 This is a schematic diagram of the data fusion and processing module structure of the present invention; Figure 5 This is a schematic diagram of the defect identification and classification module of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1-5This invention provides a technical solution: a surface defect detection system for automotive engine cylinder heads, comprising a multimodal data acquisition module, a data fusion and processing module, a defect identification and classification module, and a result output and feedback module. The multimodal data acquisition module is used to acquire multimodal data from the engine cylinder head, including but not limited to optical image data, ultrasonic testing data, thermal imaging data, and electromagnetic induction data. The multimodal data acquisition module uses a high-resolution camera with a resolution of 20 megapixels and is equipped with lenses of different focal lengths to adapt to the imaging needs of different parts. The phased array ultrasonic probe uses a product with a frequency range of 1-10MHz, capable of detecting internal defects at different depths. The thermal imager has a temperature resolution of 0.05℃, accurately capturing temperature changes. The excitation frequency of the electromagnetic induction detection probe is adjustable within the range of 10-100kHz to adapt to the detection of surface and near-surface defects of different types and depths. Each detection unit is mounted on the detection platform via a high-precision bracket and robotic arm to ensure the accuracy and stability of the detection position. The data fusion and processing module is connected to the multimodal data acquisition module, receiving and fusing multimodal data, and preprocessing the fused data, including noise reduction, enhancement, and feature extraction. The data fusion and processing module uses a high-performance server as the data processing platform and installs multimodal fusion network software based on deep learning frameworks such as TensorFlow or PyTorch. According to the characteristics of different modal data, the network parameters are optimized, such as adjusting the size and number of convolutional kernels in the convolutional layers and the fusion weights in the fusion layers. The defect identification and classification module is connected to the data fusion and processing module. Based on deep learning algorithms, it analyzes the preprocessed data, identifies the apparent defects of the cylinder head, and classifies the defects. The defect identification and classification module deploys a deep learning model based on a combination of CNN and RNN on the server. The model is trained on a large amount of defect sample data to improve the accuracy of identification and classification. The training data is regularly updated to adapt to emerging defect types and changes in production processes. The result output and feedback module is connected to the defect identification and classification module, outputting defect detection results, including the location, type, and severity of defects, and feeding these results back to the production system for quality control and adjustment during production. The result output and feedback module utilizes user-friendly visual interface software and employs 3D modeling technology to display the cylinder head's 3D model and annotate defects. The report generation system uses a template-based design to automatically generate customized inspection reports based on the detection results. The feedback control system is connected to the production system via industrial Ethernet to ensure that detection results are transmitted to the production system in real time and stably.

[0018] The multimodal data acquisition module includes: a high-resolution optical imaging unit equipped with multiple high-resolution cameras at different angles to acquire high-definition optical images of the engine cylinder head surface to detect visible defects such as scratches, pits, and pinholes; an ultrasonic detection unit using a phased array ultrasonic probe capable of emitting and receiving ultrasonic waves of different frequencies, and detecting internal cracks, porosity, and shrinkage defects by analyzing the propagation characteristics of ultrasonic waves inside the cylinder head; a thermal imaging unit using infrared thermal imaging technology to monitor the temperature distribution on the cylinder head surface in real time, identifying potential defects such as abnormal heat conduction caused by internal defects through temperature anomalies; and an electromagnetic induction detection unit that detects electromagnetic induction changes caused by defects by applying an alternating magnetic field to the cylinder head, used to detect surface and near-surface defects such as cracks and inclusions.

[0019] The data fusion and processing module adopts a deep learning-based multimodal fusion network, which includes multiple sub-networks that extract features from data of different modalities. Then, the features extracted by each sub-network are fused through a fusion layer to obtain a more comprehensive and accurate representation of defect features.

[0020] The defect identification and classification module adopts a deep learning model based on the combination of convolutional neural networks (CNN) and recurrent neural networks (RNN). CNN is used to extract spatial features of defects, while RNN is used to process time series features (such as data changes during dynamic detection). Through the collaborative work of the two, the accuracy of defect identification and classification is improved.

[0021] The result output and feedback module includes: a visualization interface: displaying the test results in an intuitive graphical interface, including a 3D model of the cylinder head, with the location, type, and severity of defects marked on the model for easy viewing and analysis by operators; a report generation system: automatically generating detailed test reports, including defect statistics, key data from the testing process, and quality analysis suggestions. The reports are stored in multiple formats for easy review and traceability.

[0022] A method for detecting apparent defects in an automobile engine cylinder head includes the following steps: Step 1: Place the engine cylinder head on a testing platform with automatic positioning and calibration functions, ensuring the cylinder head is in the correct testing position and orientation. Then, sequentially activate the high-resolution optical imaging unit, ultrasonic testing unit, thermal imaging unit, and electromagnetic induction testing unit. The high-resolution camera captures high-definition optical images of the cylinder head surface from multiple angles; the phased-array ultrasonic probe emits ultrasonic waves towards the cylinder head and receives the reflected wave signals; the thermal imaging unit monitors the temperature distribution on the cylinder head surface in real time, generating thermal imaging data; and the electromagnetic induction testing unit applies an alternating magnetic field to collect electromagnetic induction data. The data collected by each unit is transmitted in real time to the data fusion and processing module via a high-speed data transmission interface.

[0023] Step Two: After receiving the multimodal data, the data fusion and processing module first preprocesses each modality of data, such as denoising and contrast enhancement for optical images, filtering for ultrasonic data, and temperature calibration for thermal imaging data. A deep learning-based multimodal fusion network is then used to input the preprocessed modal data into the corresponding sub-networks for feature extraction. Each sub-network learns the feature representations of different modalities, and then these features are fused through a fusion layer to generate a comprehensive defect feature vector.

[0024] Step 3: Input the fused defect feature vector into a deep learning model based on a combination of CNN and RNN. The CNN layer extracts spatial features from the feature vector, capturing the shape and edge features of the defect; the RNN layer processes the time-series features in the feature vector, learning the dynamic changes of the defect. Through the multi-layer processing of the model, the final output is the defect identification and classification results, determining whether there is a defect in the cylinder head and the specific type of defect, such as cracks, pinholes, or air holes.

[0025] Step Four: The results output and feedback module receives the output results from the defect identification and classification module. It displays the cylinder head as a 3D model on the visualization interface, accurately marking the location, type, and severity of defects on the model. The report generation system automatically generates a detailed inspection report, including defect statistical charts, key data from the inspection process, and quality analysis suggestions for different defect types. The report is stored in common formats such as PDF and Excel for easy access and traceability by production and quality control personnel. The feedback control system relays defect information to the production system in real time based on the inspection results. For cylinder heads with severe defects, the system automatically sends a signal to pause the relevant production line, prompting production personnel to inspect and adjust the equipment. For general defects, the system provides optimization suggestions to the production system based on the defect type and location, such as adjusting casting process parameters and machining process parameters, to achieve closed-loop quality control of the production process.

[0026] Inspection Process: The engine cylinder head is placed on the inspection platform. The platform automatically identifies the cylinder head model and location, adjusting the parameters and positions of each inspection unit to accommodate the inspection requirements of different cylinder head models. The multimodal data acquisition module is activated, and each inspection unit acquires data according to a preset time sequence and parameters. The acquired data is transmitted in real-time to the data fusion and processing module via a high-speed data transmission line. The data fusion and processing module preprocesses the acquired multimodal data, removing noise and interference signals. Then, the preprocessed data is input into the multimodal fusion network. The network automatically learns the correlation and feature representation between different modalities, generating a fused feature vector. This fused feature vector is input into the deep learning model of the defect identification and classification module. The model analyzes the feature vector layer by layer to identify whether defects exist in the cylinder head and determine the type and severity of the defects. The identification results are stored in the server's database. The result output and feedback module reads the inspection results from the database, displays the 3D model of the cylinder head and defect information on a visual interface, and generates an inspection report. The feedback control system sends the inspection results to the production system. The production system automatically adjusts production parameters or suspends the production line based on the defect situation for quality control and improvement.

[0027] In summary, this automotive engine cylinder head surface defect detection system and method, by integrating multiple detection technologies including optics, ultrasound, thermal imaging, and electromagnetic induction, can comprehensively detect various surface defects in engine cylinder heads, including visible surface defects, internal hidden defects, and minute defects, significantly improving detection accuracy and reducing missed and false detection rates. The entire detection process is automated, and multimodal data acquisition and parallel processing greatly shorten the detection time, meeting the rapid detection needs of large-scale production lines and improving production efficiency. The deep learning-based defect identification and classification module can accurately identify defect types and assess the severity of defects, providing strong data support for quality control and product improvement during the production process. The detection results can be fed back to the production system in real time, automatically triggering corresponding production adjustment measures based on the defect situation, realizing closed-loop quality control in the production process, reducing the generation of defects from the source, and improving product quality and reliability.

[0028] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0029] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A system for detecting apparent defects in automobile engine cylinder heads, characterized in that: The system includes a multimodal data acquisition module, a data fusion and processing module, a defect identification and classification module, and a result output and feedback module. The multimodal data acquisition module is used to acquire multimodal data of the engine cylinder head, including but not limited to optical image data, ultrasonic detection data, thermal imaging data, and electromagnetic induction data. The data fusion and processing module is connected to the multimodal data acquisition module, receives and fuses the multimodal data, and preprocesses the fused data, including noise reduction, enhancement, and feature extraction operations. The defect identification and classification module is connected to the data fusion and processing module, analyzes the preprocessed data based on deep learning algorithms, identifies apparent defects in the cylinder head, and classifies the defects. The result output and feedback module is connected to the defect identification and classification module, outputs defect detection results, including the location, type, and severity information of the defects, and feeds the detection results back to the production system for quality control and adjustment during the production process.

2. The apparent defect detection system for an automobile engine cylinder head according to claim 1, characterized in that: The multimodal data acquisition module includes: a high-resolution optical imaging unit equipped with multiple high-resolution cameras at different angles to acquire high-definition optical images of the engine cylinder head surface to detect visible defects such as scratches, pits, and pinholes; an ultrasonic detection unit using a phased array ultrasonic probe capable of emitting and receiving ultrasonic waves of different frequencies, and detecting internal cracks, porosity, and shrinkage defects by analyzing the propagation characteristics of ultrasonic waves inside the cylinder head; a thermal imaging unit using infrared thermal imaging technology to monitor the temperature distribution on the cylinder head surface in real time, identifying potential defects such as abnormal heat conduction caused by internal defects through temperature anomalies; and an electromagnetic induction detection unit that detects electromagnetic induction changes caused by defects by applying an alternating magnetic field to the cylinder head, used to detect surface and near-surface defects such as cracks and inclusions.

3. The apparent defect detection system for an automobile engine cylinder head according to claim 1, characterized in that: The data fusion and processing module adopts a deep learning-based multimodal fusion network, which includes multiple sub-networks that extract features from data of different modalities. Then, the features extracted by each sub-network are fused through a fusion layer to obtain a more comprehensive and accurate representation of defect features.

4. The apparent defect detection system for an automobile engine cylinder head according to claim 1, characterized in that: The defect identification and classification module adopts a deep learning model based on a combination of convolutional neural networks and recurrent neural networks.

5. The apparent defect detection system for an automobile engine cylinder head according to claim 1, characterized in that: The result output and feedback module includes: a visualization interface: displaying the test results in an intuitive graphical interface, including a 3D model of the cylinder head, with the location, type, and severity of defects marked on the model for easy viewing and analysis by operators; and a report generation system: automatically generating detailed test reports, including defect statistics, key data from the testing process, and quality analysis suggestions. The reports are stored in multiple formats for easy access and traceability.

6. A method for detecting apparent defects in an automobile engine cylinder head according to any one of claims 1-5, characterized in that, Includes the following steps: Step 1: Place the engine cylinder head on a testing platform with automatic positioning and calibration functions, ensuring the cylinder head is in the correct testing position and orientation. Then, sequentially activate the high-resolution optical imaging unit, ultrasonic testing unit, thermal imaging unit, and electromagnetic induction testing unit. The high-resolution camera captures high-definition optical images of the cylinder head surface from multiple angles; the phased array ultrasonic probe emits ultrasonic waves towards the cylinder head and receives reflected wave signals; the thermal imaging unit monitors the temperature distribution on the cylinder head surface in real time and generates thermal imaging data; the electromagnetic induction testing unit applies an alternating magnetic field and collects electromagnetic induction data. The data collected by each unit is transmitted in real time to the data fusion and processing module via a high-speed data transmission interface. Step 2: After receiving the multimodal data, the data fusion and processing module first preprocesses each modal data, such as denoising and contrast enhancement for optical images, filtering for ultrasonic data, and temperature calibration for thermal imaging data. A deep learning-based multimodal fusion network is used to input the preprocessed modal data into the corresponding sub-network for feature extraction. Each sub-network learns the feature representation of different modal data, and then the fusion layer fuses these features to generate a comprehensive defect feature vector. Step 3: Input the fused defect feature vector into a deep learning model based on a combination of CNN and RNN. The CNN layer extracts spatial features from the feature vector, capturing the shape and edge features of the defect; the RNN layer processes the time series features in the feature vector, learning the dynamic change features of the defect. Through the multi-layer processing of the model, the final output is the defect identification and classification results, determining whether there is a defect in the cylinder head and the specific type of defect, such as cracks, sand holes, and air holes. Step 4: Result Output and Feedback Module receives the output results from the Defect Identification and Classification Module. It displays the cylinder head as a 3D model on the visualization interface, accurately marking the location, type, and severity of defects on the model. The report generation system automatically generates a detailed inspection report, including defect statistics charts, key data from the inspection process, and quality analysis suggestions for different defect types. The report is stored in common formats such as PDF and Excel for easy access and traceability by production and quality control personnel. The feedback control system feeds defect information back to the production system in real time based on the inspection results. For cylinder heads with severe defects, the system automatically sends a signal to pause the relevant production line, prompting production personnel to inspect and adjust the equipment. For general defects, the system provides optimization suggestions to the production system based on the defect type and location, such as adjusting casting process parameters and machining process parameters, to achieve closed-loop quality control of the production process.

Images