A binding detection method of an intelligent ring sensor, a medium and an electronic device

Abstract

The present application relates to the technical field of intelligent ring detection, in particular to a kind of binding detection method, medium and electronic equipment of intelligent ring sensor, it includes the following steps: step a, based on the sensor binding detection of electrical signal, through simulating wearing scene acquisition and analysis electrical signal data;Step b, the electrical signal data of each channel is carried out single-channel stability detection and multi-channel consistency detection;Step c, using machine vision combined with deep learning algorithm, defect detection is carried out to binding area;Step d, whether binding is qualified is comprehensively judged according to electrical signal detection result and machine vision detection result.This application judges whether binding meets the requirements by analyzing and comparing the differences between the electrical signals of individual channels and the standard electrical signals and the electrical signal differences between multiple channels;Using machine vision and deep learning algorithm, binding defects that are difficult to distinguish visually are detected, which greatly reduces the demand for manpower and improves detection efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of smart ring detection technology, and in particular to a smart ring sensor binding detection method, medium, and electronic device. Background Technology

[0002] In recent years, smart rings, as a highly anticipated smart wearable device, have successfully integrated the elegant aesthetics of traditional rings with the powerful functionality of modern technology. As China's aging population becomes increasingly prominent, with the number of empty-nest and elderly people living alone continuing to rise, people are paying more and more attention to health and lifestyle. The smart ring industry has made significant progress in health monitoring technology, continuously driving innovation within the industry. Continuous upgrades in sensing technology have made smart rings increasingly powerful. Sensors, as the core component of smart rings, can accurately monitor and analyze users' physiological data. Currently, smart rings have implemented multiple functions such as heart rate monitoring, sleep quality monitoring, and step counting. However, the sensors in smart rings are thin, light, and precise, making them highly susceptible to damage during the soldering or bonding process.

[0003] Existing smart ring sensor-based detection methods have the following drawbacks:

[0004] 1. Slow detection speed: Traditional detection methods require manual visual inspection of the sensor binding area under a microscope, which is time-consuming and labor-intensive, slow, and prone to missed detections, and cannot meet the needs of the production line.

[0005] 2. Poor reliability: Since visual inspection can only detect and judge the bonding effect from the surface features of the bonding area, it is difficult to accurately distinguish defects such as ACF attachment marks and sensor circuit tearing, which poses a risk of misjudgment.

[0006] 3. Lack of comprehensive judgment based on multiple factors: Traditional testing methods rely primarily on electrical signal detection as the main detection and judgment criterion. If the electrical signal is normal, the product is judged as good. This method cannot detect defects in sensors after they are bonded, leading to defective products in later stages. Summary of the Invention

[0007] This invention addresses the problems of existing technologies by providing a binding detection method, medium, and electronic device for a smart ring sensor. It determines whether the binding meets the requirements by analyzing and comparing the differences between the electrical signals of a single channel and the standard electrical signals, as well as the differences between the electrical signals of multiple channels. It uses machine vision and deep learning algorithms to detect binding defects that are difficult to detect visually, which greatly reduces the need for manpower and improves detection efficiency and accuracy.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0009] This invention provides a method for detecting the binding of a smart ring sensor, comprising the following steps:

[0010] Step a: Sensor binding detection based on electrical signals, which involves collecting and analyzing electrical signal data by simulating wearable scenarios;

[0011] Step b: Perform single-channel stability testing and multi-channel consistency testing on the electrical signal data of each channel;

[0012] Step c: Use machine vision combined with deep learning algorithms to detect defects in the binding area; Step d: Determine whether the binding is qualified based on the combined results of electrical signal detection and machine vision detection.

[0013] In step a, the electrical signal acquisition is carried out under specific lighting conditions, and the transmitted electrical signal data is continuously acquired and recorded. The electrical signal data is saved in array form for subsequent analysis.

[0014] In step b, the stability of the electrical signal is evaluated by calculating the standard deviation of the electrical signal data collected by the same channel, and a preset threshold is used to determine whether there is signal instability in the bound sensor channel.

[0015] In step b, potential bonding problems are identified by comparing the difference ratio between multiple channel electrical signals with the theoretical signal value.

[0016] In step c, the YOLOv10-S model is used to detect sensor bonding defects. The detection of sensor bonding defects using the YOLOv10-S model includes the following steps:

[0017] Step c1: Collection and annotation of the defect dataset;

[0018] Step c2: Model selection for YOLOv10.

[0019] Step c includes the process of training and evaluating the YOLOv10-S model, which is used to optimize the selection of hyperparameters.

[0020] In step d, the specific defect type and its confidence level are output. Defects with a confidence level of less than 70% are ignored; defects with a confidence level between 70% and 85% are manually reviewed; and defects with a confidence level higher than 85% are directly detected by the algorithm model.

[0021] The present invention also provides a computer storage medium storing computer instructions, which, when invoked, are used to execute the binding detection method of the smart ring sensor.

[0022] The present invention also provides an electronic device comprising: a processor; and a memory arranged to store computer-executable instructions, which, when executed, cause the processor to perform the binding detection method of the smart ring sensor.

[0023] The beneficial effects of this invention are:

[0024] This invention significantly improves the efficiency and accuracy of sensor-based smart ring inspection after bonding by combining electrical signal detection technology and machine vision defect detection algorithms. It introduces an automated inspection process based on electrical signals and machine vision, replacing traditional manual visual inspection, greatly reducing human intervention and increasing inspection speed. It enables real-time detection and immediate feedback of results, meeting the production line's need for rapid response. Beyond electrical signal detection, it incorporates machine vision inspection, enabling the identification of subtle defects difficult to detect with the naked eye, such as ACF adhesion marks and sensor circuit tears, improving the comprehensiveness and accuracy of inspection. The use of machine learning models (such as YOLOv10) allows for more precise identification and classification of defects, reducing the possibility of misjudgment. The integration of electrical signal stability and consistency detection with machine vision defect analysis enables a comprehensive evaluation of bonding effectiveness, ensuring that only high-quality products enter the market. It effectively detects products with physical damage or bonding problems even if the electrical signals are normal, preventing defective products from entering subsequent stages. It reduces reliance on human inspectors, lowering long-term labor costs. Early detection and interception of defective products reduces potential rework and repair costs later on. Finally, it improves the automation level and efficiency of production. Attached Figure Description

[0025] Figure 1 This is a flowchart of a binding detection method for a smart ring sensor according to the present invention.

[0026] Figure 2 This is a flowchart of the electrical signal detection method of the present invention.

[0027] Figure 3 This is a flowchart of the sensor binding detection method based on YOLOv10 of the present invention. Detailed Implementation

[0028] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention. The present invention will be described in detail below with reference to the accompanying drawings.

[0029] Example 1

[0030] In Embodiment 1 of this application, a binding detection method for a smart ring sensor is provided, which includes the following steps:

[0031] Step a: Sensor binding detection based on electrical signals, which involves collecting and analyzing electrical signal data by simulating wearable scenarios;

[0032] Step b: Perform single-channel stability testing and multi-channel consistency testing on the electrical signal data of each channel;

[0033] Step c: Use machine vision combined with deep learning algorithms to detect defects in the binding area; Step d: Determine whether the binding is qualified based on the combined results of electrical signal detection and machine vision detection.

[0034] This application significantly improves the efficiency and accuracy of sensor-based smart ring inspection by combining electrical signal detection technology and machine vision defect detection algorithms. It introduces an automated inspection process based on electrical signals and machine vision, replacing traditional manual visual inspection, greatly reducing human intervention and increasing inspection speed. It enables real-time detection and immediate feedback of results, meeting the production line's need for rapid response. Beyond electrical signal detection, it incorporates machine vision inspection, enabling the identification of subtle defects difficult to detect with the naked eye, such as ACF adhesion marks and sensor circuit tears, improving the comprehensiveness and accuracy of inspection. The use of machine learning models (such as YOLOv10) allows for more precise identification and classification of defects, reducing the possibility of misjudgment. The integration of electrical signal stability and consistency detection with machine vision defect analysis enables a comprehensive evaluation of the bonding effect, ensuring that only high-quality products enter the market. It effectively detects products with physical damage or bonding problems even if the electrical signals are normal, preventing defective products from entering subsequent stages. It reduces reliance on human inspectors, lowering long-term labor costs. Early detection and interception of defective products reduces potential rework and repair costs later on. Finally, it improves the automation level and efficiency of production.

[0035] In this embodiment, step a involves acquiring electrical signals under specific lighting conditions, continuously acquiring and recording the transmitted electrical signal data, and saving the data as an array for subsequent analysis. Under specific lighting conditions, a scenario simulating the wearing of a smart ring is simulated, and photoelectric data received by the sensor terminals is acquired through the PCBA board on the smart ring. Within a continuous 2000 milliseconds, the transmitted electrical signal data for each channel is acquired and recorded every 100 milliseconds, and finally saved as an array. The acquired data is as follows:

[0036]

[0037] In this embodiment, step b assesses the stability of the electrical signal by calculating the standard deviation of the electrical signal data collected from the same channel, and determines whether there is signal instability in the bound sensor channel based on a preset threshold. Specifically, after sensor binding, analyzing and comparing the electrical signal data of each channel through single-channel electrical signal detection provides a better understanding of the stability and reliability of the sensor's signal acquisition. Theoretically, each individual channel, under the same conditions, will collect 20 identical electrical signal data. Therefore, the stability of the electrical signal can be assessed by calculating the standard deviation of the electrical signal data collected from each channel, as shown in Equation 1. When S exceeds the set threshold H... S If the signal is unstable, the channel to which the sensor is attached has an unstable signal; otherwise, the test passes.

[0038]

[0039] Where, x i This represents electrical signal data collected from the same channel. It represents the average value of electrical signal data collected from the same channel.

[0040] In this embodiment, step b involves comparing the difference ratio between multiple channel electrical signals with the theoretical signal value to identify potential bonding problems. Specifically, the main detection principle of multi-channel electrical signal detection is that, theoretically, identical photoelectric sensors transmit identical electrical signal values ​​under the same conditions. Based on this characteristic, corresponding algorithms can be designed to intercept bonded products where the signal value difference exceeds a threshold. This embodiment utilizes the difference ratio between signal channels and the difference between the actual and theoretical signals to design a corresponding detection method based on sensor characteristics.

[0041] The difference ratio between signal channels is determined by the fact that, theoretically, the electrical signal data acquired by the photoelectric sensor built into the same sensor under the same optical signal input conditions should be identical each time. Considering interference from factors such as resistance, a difference ratio threshold H can be set for the difference ratio between signal channels. φ When the difference ratio φ between signal channels exceeds H φ If the signal difference between the signal channels is too large, it indicates a problem with the sensor bonding connection; conversely, this step can be used to detect it. Wherein, H... φ The calculation method is shown in equation (2):

[0042] right x1≠x2, therefore Among them, I standard It is the theoretical electrical signal value output under the current optical signal conditions.

[0043] In this embodiment, step c involves using the YOLOv10-S model to detect sensor-attached defects. Step c includes training and evaluating the YOLOv10-S model to optimize hyperparameter selection. Step d outputs the specific defect type and its confidence level. Defects with a confidence level less than 70% are ignored; defects with a confidence level between 70% and 85% are manually reviewed; and defects with a confidence level higher than 85% are directly detected using the algorithm model. Specifically, under the above settings, sensor tethering detection based on YOLOv10 is a detection technology that integrates deep learning and machine vision to achieve accurate detection of sensor tethers. YOLOv10 is a highly efficient target detection algorithm that can quickly and accurately identify multiple targets in an image, and it has a good balance between speed and accuracy. By applying the YOLOv10 algorithm to the field of sensor tethering detection, the accuracy and efficiency of sensor tethering can be improved, thereby better meeting the needs of sensor tethering in fields such as industrial production and smart devices. The sensor tethering detection method based on YOLOv10 is as follows:

[0044] Defect Dataset Collection and Labeling: Defective data is collected, organized into a defect dataset, and labeled. Deep learning plays a crucial role in training sensor-attachment defect detection models, particularly in detecting sensor tearing, attachment misalignment, and insufficient attachment particle counts. These issues can lead to quality problems in smart ring production, affecting product performance and reliability. Therefore, collecting a defect detection dataset is essential, requiring consideration of dataset quality and diversity to ensure the model accurately captures the characteristics of various defect types.

[0045] YOLOv10 Model Selection: For the problem of sensor binding defect detection in smart rings, the selection of the YOLOv10 model needs to comprehensively consider the actual application scenario. First, YOLOv10, as an object detection model, has six different sizes: N / S / M / B / L / X. Therefore, selection needs to be based on the specific application. For the specific task of sensor binding defect detection, detailed detection and analysis of sensor binding conditions are required, thus necessitating consideration of model accuracy and robustness. Furthermore, considering that sensor binding defects are relatively typical and limited types, the model's generalization ability is low, and high model efficiency is required. In addition, considering the data characteristics and actual needs of the specific application scenario, factors such as model training and deployment costs, as well as hardware requirements, also need to be comprehensively considered. Therefore, this embodiment, based on the characteristics of sensor binding defects in smart rings and considering the specific detection situation, selects YOLOv10-S as the YOLOv10 model for sensor binding detection.

[0046] Model Training and Evaluation: The selection of appropriate hyperparameters also significantly impacts model performance. Continuously optimizing the model training process can improve its effectiveness and accuracy in detecting sensor tearing, binding misalignment, and insufficient bound particles. Training a deep learning defect detection model is a complex and crucial task, requiring adjustments to the training and validation set ratios based on the actual dataset, and continuous adjustments and optimizations to adapt to changing production environments and demands. Furthermore, the use of a validation set is essential for evaluating the model during training. Using evaluation metrics such as accuracy, recall, and F1 score effectively measures model performance. Analyzing the evaluation results allows for targeted adjustments to model parameters and training strategies, further optimizing model performance and improving accuracy and generalization ability. The selection of hyperparameters for model training is shown in Table 1.

[0047] Table 1 Hyperparameters for model training

[0048]

[0049]

[0050] Output Defect Type: Based on the inspection results, output "Pass" or "Fail" information. If the product passes the inspection, the output will be "No defect detected," meaning the product meets quality standards and has no defects. However, if the product fails the inspection, the specific defect type detected and its confidence level μ will be output in detail. If the output defect confidence level is less than 70%, the defect is ignored. Defects with a confidence level between 70% and 85% are manually reviewed to reduce the possibility of misjudgment. Defects with a confidence level higher than 85% are directly evaluated using the algorithm model's detection results.

[0051]

[0052] In summary, this application's embodiments possess three key points: Key Point 1: Addressing the difficulties and low efficiency of detecting sensor bonding after smart rings are completed, a novel sensor bonding detection method based on electrical signals is proposed. The method determines whether the bonding meets requirements by analyzing and comparing the differences between the electrical signals of a single channel and standard electrical signals, as well as the differences between electrical signals of multiple channels. Key Point 2: Utilizing machine vision and deep learning algorithms to detect bonding defects that are difficult to discern visually, significantly reduces manpower requirements and improves detection efficiency and accuracy. Key Point 3: Based on the above two points, this method can detect products that pass electrical signal detection but have defects, such as torn sensor bonding lines or insufficient bonding particle counts.

[0053] Compared with existing technologies, the embodiments of this application propose a novel electrical signal detection method based on the characteristics and principles of sensors. This method can efficiently and accurately detect electrical signals from the bonded sensors. Furthermore, based on electrical signal detection, the improved machine vision and deep learning algorithms of this application are used to detect bonding defects in the bonding area. This allows for the detection of hidden defects that electrical signal detection methods cannot detect, such as sensor wire damage, partial tearing, and insufficient bonding particle count. The embodiments of this application can be applied to the actual production of different models of smart rings. Only flexible adjustments are needed to the hardware devices such as cameras, light sources, fixing bases, and electrical signal connection lines. This enables fast and real-time smart ring sensor bonding detection while ensuring the accuracy of the judgment, greatly improving the production efficiency of smart ring bonding detection.

[0054] Example 2

[0055] In Embodiment 2 of this application, a computer storage medium is provided, which stores computer instructions. When the computer instructions are invoked, they are used to execute the binding detection method of the smart ring sensor.

[0056] Example 3

[0057] In a third embodiment of this application, an electronic device is provided, comprising: a processor; and a memory arranged to store computer-executable instructions, which, when executed, cause the processor to perform the binding detection method of the smart ring sensor.

[0058] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0059] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0060] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0062] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0063] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting the binding of a smart ring sensor, characterized in that: Includes the following steps: Step a) Sensor binding detection based on electrical signals: Electrical signal data is collected and analyzed by simulating a wearing scenario. In step a), electrical signal acquisition is performed under specific lighting conditions. The transmitted electrical signal data is continuously collected and recorded, and the data is saved as an array for subsequent analysis. Under specific lighting conditions, a smart ring wearing scenario is simulated, and photoelectric data received by the sensor terminals is collected through the PCBA board on the smart ring. Within a continuous 2000 milliseconds, the transmitted electrical signal data of each channel is collected and recorded every 100 milliseconds, and finally saved as an array. The electrical signals of the photoelectric sensors of the four channels of the smart ring are collected simultaneously, with 20 sets of data collected from each channel and saved as an independent array. The collected data is as follows: ； Step b: Perform single-channel stability testing and multi-channel consistency testing on the electrical signal data of each channel; In step b, the stability of the electrical signal is evaluated by calculating the standard deviation of the electrical signal data collected from the same channel, and a preset threshold is used to determine whether there is signal instability in the bound sensor channel. The stability of the electrical signal is evaluated by calculating the standard deviation of the electrical signal data collected from each channel, as shown in the following formula: ; Represents electrical signal data collected from the same channel. It is the average value of electrical signal data collected from the same channel; when S Exceeding the set threshold If the signal is unstable, the channel to which the sensor is attached has an unstable signal; otherwise, the test passes. In step b, potential bonding problems are identified by comparing the difference ratio between multiple channel electrical signals with the theoretical signal value; a difference ratio threshold is then set. H φ ; in, H φ The calculation method is as follows: ; in, I standard It is the theoretical electrical signal value output under the current optical signal conditions; When the difference between signal channels is φ Exceed H φ If the difference between the signal channels is too large, it indicates a problem with the sensor bonding connection; otherwise, this step can be used to detect it. Step c: Use machine vision combined with deep learning algorithms to detect defects in the bonding region; in step c, the YOLOv10-S model is used to detect sensor bonding defects, and the detection of sensor bonding defects using the YOLOv10-S model includes the following steps: Step c1: Collection and annotation of the defect dataset; Step c2, YOLOv10 model selection; Step d: Determine whether the binding is qualified based on the combined results of electrical signal detection and machine vision detection. In step d, the specific defect type and its confidence level are output. Defects with a confidence level of less than 70% are ignored. For defects with a confidence level between 70% and 85%, manual re-judgment is performed. For defects with a confidence level higher than 85%, the detection results of the algorithm model are directly adopted.

2. The binding detection method for a smart ring sensor according to claim 1, characterized in that, Step c includes the process of training and evaluating the YOLOv10-S model, which is used to optimize the selection of hyperparameters.

3. A computer storage medium storing computer instructions, wherein the computer storage medium, when invoked, is used to execute the binding detection method of the smart ring sensor as described in any one of claims 1-2.

4. An electronic device, wherein, The electronic device includes: a processor; and a memory arranged to store computer-executable instructions, which, when executed, cause the processor to perform the binding detection method of the smart ring sensor as described in any one of claims 1-2.

Images