Electronic product defect classification method based on heterogeneous sensor data fusion

Abstract

The present application relates to the technical field of electronic product defect classification, and particularly relates to an electronic product defect classification method based on heterogeneous sensor data fusion. The working temperature collected by the temperature sensor arranged in the electronic product is combined with the CPU occupancy rate of the electronic product to determine the working abnormality classification of the electronic product and determine the image collection time length ratio. The frequency spectrum image and the oscilloscope image of the electronic product are collected based on the image collection time length ratio. The fault classification of the electronic product is determined based on the vibration frequency and the clock frequency. The defect classification of the electronic product is determined by combining the working abnormality classification of the electronic product and the fault classification of the electronic product, and the image collection time length ratio corresponding to the defect classification is adjusted. The frequency spectrum image and the oscilloscope image of the electronic product collected based on the adjusted image collection time length ratio are adjusted to adjust the defect classification. The present application improves the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion.

Description

[0001] This invention relates to the field of electronic product defect classification technology, and in particular to an electronic product defect classification method based on heterogeneous sensor data fusion. Background Technology

[0002] Traditional methods often rely on data from single-type sensors for defect identification, which has significant limitations in complex application scenarios, necessitating the development of heterogeneous data fusion technology. Data from a single sensor alone cannot accurately reflect and classify defect characteristics. The operating environment of electronic products is subject to electromagnetic interference and noise superposition, making single-sensor data susceptible to interference and resulting in defect classification bias. When defects occur in electronic products, single-source data struggles to achieve accurate defect location and classification, leading to insufficient targeting of subsequent processing solutions. As electronic products evolve towards higher integration, higher power density, and higher intelligence, their internal structures and operating mechanisms become more complex, placing higher demands on defect classification technology.

[0003] For example, Chinese Patent Publication No. CN111213045A provides a method for automatic defect classification, the method comprising: (i) acquiring at least one first image of at least one region of an object via a first camera; (ii) processing the at least one first image to detect a group of suspected defects within the at least one region; (iii) performing a first classification process to initially classify the group of suspected defects; (iii) determining whether the first subgroup of suspected defects requires additional information from a second camera to complete the classification; (iv) when it is determined that the first subgroup of suspected defects requires additional information from the second camera, then: (a) acquiring a second image of the first subgroup of suspected defects via the second camera; and (b) performing a second classification process to classify the first subgroup of suspected defects. The prior art also has the following problems:

[0004] Existing technologies that classify product defects solely by detecting a single data source, such as image detection and classification methods, are imperfect and fail to combine multiple heterogeneous sensor data for comprehensive analysis to improve the accuracy of product defect classification. Summary of the Invention

[0005] Therefore, this invention provides a method for classifying electronic product defects based on heterogeneous sensor data fusion, which overcomes the problem in the prior art that classifies products by detecting only a single data source and fails to combine multiple heterogeneous sensor data for comprehensive analysis to improve the accuracy of product defect classification.

[0006] To achieve the above objectives, the present invention provides a method for classifying defects in electronic products based on heterogeneous sensor data fusion, comprising:

[0007] The working temperature is collected by a temperature sensor installed inside the electronic product, and the CPU utilization rate of the electronic product is combined to calculate the working characterization value of the electronic product. Based on the working characterization value, the working abnormality classification of the electronic product is determined and the proportion of image acquisition time is determined.

[0008] The spectrum image and oscilloscope image of the electronic product are acquired based on the image acquisition time ratio. The vibration frequency of the electronic product is obtained from the spectrum image, and the clock frequency of the electronic product is obtained from the oscilloscope image.

[0009] Frequency feature values ​​are calculated and generated based on the vibration frequency and the clock frequency, and the fault classification of the electronic product is determined based on the frequency feature values.

[0010] The defect classification of the electronic product is determined by combining the abnormal operation classification and fault classification of the electronic product, and the proportion of the image acquisition duration corresponding to the defect classification is adjusted.

[0011] Based on the spectrum image and oscilloscope image of the electronic product acquired according to the adjusted image acquisition time ratio, an adjusted frequency feature value is generated. The feature deviation value between the adjusted frequency feature value and the frequency feature value is calculated, and the defect classification is adjusted based on the feature deviation value.

[0012] Furthermore, the process of calculating the operating temperature of the electronic product by combining the operating temperature collected by the temperature sensor installed inside the electronic product with the CPU utilization rate of the electronic product includes,

[0013] The ratio of the operating temperature to the rated temperature is determined as the temperature factor;

[0014] The ratio of CPU utilization to CPU workload utilization is defined as the CPU utilization factor.

[0015] The weighted sum of the temperature factor and the CPU utilization factor is determined to be the working characterization value.

[0016] Further, the step of determining the operational anomaly classification of the electronic product and determining the image acquisition duration ratio based on the operational characterization value, wherein,

[0017] If the working characterization value is greater than the first preset characterization value and less than the second preset characterization value, then the working abnormality of the electronic product is determined to be an abnormality of the external associated system of the circuit board, and the spectrum image and oscilloscope image are acquired according to the first image acquisition duration ratio.

[0018] If the working characteristic value is greater than the second preset characteristic value, the working abnormality of the electronic product is determined to be an internal system abnormality of the circuit board, and a spectrum image and an oscilloscope image are acquired according to the second image acquisition duration ratio.

[0019] Furthermore, the process of calculating and generating frequency characteristic values ​​based on the vibration frequency and the clock frequency includes,

[0020] The deviation value between the vibration frequency and the rated vibration frequency is determined to be the first vibration influence factor;

[0021] The deviation value between the clock frequency and the rated clock frequency is determined to be the first clock influence factor;

[0022] The weighted sum of the first vibration influence factor and the first clock influence factor is determined to be the frequency characteristic value.

[0023] Furthermore, the step of determining the fault classification of the electronic product based on the frequency characteristic values, wherein,

[0024] If the frequency characteristic value is greater than the first preset characteristic value and less than the second preset characteristic value, then the fault of the electronic product is determined to be a fault of the circuit board related components.

[0025] If the frequency characteristic value is greater than the second preset characteristic value, the fault of the electronic product is determined to be a fault of internal components of the circuit board.

[0026] Furthermore, the defect classification of the electronic product is determined by combining the operational anomaly classification and fault classification of the electronic product, wherein,

[0027] If the malfunction of the electronic product is classified as an abnormality of the external associated system of the circuit board and the fault is classified as a fault of the associated component of the circuit board, then the defect is classified as a heat dissipation defect of the circuit board.

[0028] If the malfunction of the electronic product is classified as an abnormality of the external associated system of the circuit board and the fault is classified as a fault of the internal components of the circuit board, then the defect is determined to be an abnormal defect of the circuit board drive circuit.

[0029] If the malfunction of the electronic product is classified as an internal system malfunction of the circuit board and the fault is classified as a fault of the associated components of the circuit board, then the defect is classified as a mismatch defect of the associated components of the circuit board.

[0030] If the malfunction of the electronic product is classified as an internal system malfunction of the circuit board and the fault is classified as a fault of an internal component of the circuit board, then the defect is classified as a circuit board damage defect.

[0031] Furthermore, the adjustment of the image acquisition duration ratio corresponding to the defect classification, wherein,

[0032] If the defect is determined to be an abnormal defect in the circuit board driver circuit, then the ratio of the first image acquisition time is adjusted to the ratio of the second image acquisition time.

[0033] If the defect is classified as a mismatch defect of associated components on the circuit board, the ratio of the second image acquisition time is adjusted to the ratio of the first image acquisition time.

[0034] Furthermore, the process of calculating and generating adjusted frequency characteristic values ​​based on the spectral image of the electronic product and the oscilloscope image acquired based on the adjusted image acquisition duration ratio includes,

[0035] The deviation value between the adjusted vibration frequency and the rated vibration frequency is determined to be the second vibration influence factor;

[0036] The deviation value between the adjusted clock frequency and the rated clock frequency is determined to be the second clock influence factor;

[0037] The weighted sum of the second vibration influence factor and the second clock influence factor is determined to be the adjusted frequency characteristic value.

[0038] Further, the process of calculating the adjusted frequency characteristic value and the characteristic deviation value of the frequency characteristic value includes,

[0039] Several adjusted frequency feature values ​​are acquired based on the adjusted image acquisition duration ratio for a preset duration.

[0040] The average of the differences between several of the adjusted frequency characteristic values ​​and the frequency characteristic values ​​is calculated as the characteristic deviation value.

[0041] Further, the defect classification is adjusted based on the feature deviation value, wherein,

[0042] If the feature deviation value is greater than or equal to the preset feature deviation value, and the first image acquisition duration ratio is adjusted to the second image acquisition duration ratio, then the defect classification is determined to be a circuit board heat dissipation defect.

[0043] If the feature deviation value is greater than or equal to the preset feature deviation value, and the second image acquisition duration ratio is adjusted to the first image acquisition duration ratio, then the defect classification is determined to be a circuit board damage defect.

[0044] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention determines the classification of operational anomalies by collecting the operating temperature of the temperature sensor inside the electronic product and combining it with the CPU utilization rate of the electronic product to generate operational characterization values. It utilizes the existing and essential sensors and system monitoring data of the equipment without increasing new hardware costs. By analyzing the operating temperature and CPU utilization rate of the electronic product, the problem can be quickly located to anomalies in external related systems or internal systems of the circuit board. At the same time, the image acquisition time ratio is determined according to the operating temperature and CPU utilization rate of the electronic product, avoiding the accumulation of invalid data caused by equal acquisition across all dimensions. It rationally allocates limited detection resources, ensuring that the acquired features are highly correlated with the current anomaly, improving the efficiency and accuracy of subsequent defect classification, and further enhancing the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion.

[0045] Furthermore, this invention acquires spectral images and oscilloscope images of electronic products based on the image acquisition duration ratio. It obtains the vibration frequency of the electronic product from the spectral images and the clock frequency from the oscilloscope images. Combining the vibration frequency and clock frequency, it calculates and generates frequency characteristic values. By analyzing these frequency characteristic values, it determines whether the electronic product's fault is classified as a fault in a circuit board-related component or a fault in an internal component of the circuit board. Based on a preliminary judgment of the electronic product's defects using its operating temperature and CPU utilization, the defect classification is further verified, improving diagnostic efficiency and providing more reliable input for subsequent defect classification based on multi-source data. This reduces misclassification due to data quality issues and further improves the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion.

[0046] Furthermore, this invention combines the classification of electronic product malfunctions and faults to determine the defect classification of electronic products, analyzing and judging defects from multiple dimensions, avoiding the limitations of a single classification method, and thus more accurately determining the defect classification. At the same time, it adjusts the image acquisition time ratio corresponding to the defect classification according to the defect type, improving the overall accuracy of classification judgment, and further improving the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion.

[0047] Furthermore, this invention re-acquires spectral images and oscilloscope images of electronic products based on the adjusted image acquisition duration ratio, rationally allocates acquisition resources, and calculates and generates adjusted frequency feature values ​​based on the vibration frequency and clock frequency obtained from the secondary acquisition images. By calculating and analyzing the adjusted frequency feature values ​​and the feature deviation values ​​of the frequency feature values, the defect classification is readjusted, enabling the defect classification to be dynamically adjusted according to the actual situation. Through continuous optimization and adjustment of the defect classification, it can provide strong support for the quality control of electronic products and further improve the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the steps of the electronic product defect classification method based on heterogeneous sensor data fusion according to an embodiment of the present invention.

[0049] Figure 2 A logic diagram for determining the classification of operational anomalies of electronic products in an embodiment of the present invention;

[0050] Figure 3 A logic diagram for determining fault classification of electronic products in an embodiment of the present invention;

[0051] Figure 4 A logic diagram for determining the defect classification of electronic products in an embodiment of the present invention. Detailed Implementation

[0052] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0053] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0054] It should be noted that in the description of this invention, the terms "upper," "lower," "inner," "outer," etc., which indicate the direction or positional relationship, are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0055] Please see Figure 1 The diagram illustrates the steps of an electronic product defect classification method based on heterogeneous sensor data fusion according to an embodiment of the present invention. The electronic product defect classification method based on heterogeneous sensor data fusion according to an embodiment of the present invention includes:

[0056] Step S1: The working temperature of the electronic product is collected by the temperature sensor inside the electronic product and the CPU utilization rate of the electronic product is combined to calculate the working characterization value of the electronic product. Based on the working characterization value, the working abnormality classification of the electronic product is determined and the image acquisition time ratio is determined.

[0057] Step S2: Acquire the spectrum image and oscilloscope image of the electronic product based on the image acquisition time ratio, obtain the vibration frequency of the electronic product based on the spectrum image, and obtain the clock frequency of the electronic product based on the oscilloscope image.

[0058] Step S3: Calculate and generate frequency characteristic values ​​based on vibration frequency and clock frequency, and determine the fault classification of electronic products based on frequency characteristic values;

[0059] Step S4: Combine the abnormal operation classification and fault classification of electronic products to determine the defect classification of electronic products, and adjust the image acquisition time ratio corresponding to the defect classification.

[0060] Step S5: Calculate and generate adjusted frequency feature values ​​based on the spectrum image and oscilloscope image of the electronic product acquired according to the adjusted image acquisition time ratio, calculate the adjusted frequency feature values ​​and the feature deviation value of the frequency feature values, and adjust the defect classification based on the feature deviation value.

[0061] Understandably, the core function of temperature sensors in electronic products is to accurately monitor the thermal state of core components that play a decisive role in the stability and lifespan of the equipment, providing data for heat dissipation control, fault warning, and safety protection. The core principle of CPU utilization is that the operating system inside the electronic product will count the percentage of CPU task execution time in real time and make the data available to users or programs through standardized interfaces, without the need for manual calculation.

[0062] Understandably, a small amount of thermally conductive silicone is used to attach a temperature sensor to the edge of the circuit board to collect the operating temperature of the electronic product. The CPU utilization rate can be directly collected through the operating system of the electronic product. The operating system kernel records the cumulative working time of the CPU in various states since the system started. The spectrum image of the electronic product is collected by attaching a vibration sensor to the vibrating part of the electronic product, such as the fan or the casing, to capture the time-acceleration time-domain signal. The oscilloscope image of the electronic product is collected by clamping the oscilloscope's grounding clip to the ground solder point of the integrated circuit of the electronic product, touching the signal point to be measured with the probe of the voltage probe, powering on the electronic product and running it normally, and observing the waveform on the screen.

[0063] Specifically, in step S1, the process of calculating and generating the operating characteristic value of the electronic product by combining the operating temperature collected by the temperature sensor installed inside the electronic product with the CPU utilization rate of the electronic product includes the following steps:

[0064] The ratio of the operating temperature to the rated temperature is determined as the temperature factor.

[0065] The ratio of CPU utilization to CPU workload utilization is defined as the CPU utilization factor.

[0066] The weighted sum of the temperature factor and the CPU utilization factor is determined to be the working characterization value.

[0067] Specifically, in practice, the rated temperature is the average temperature collected within a predetermined time period when the electronic product component is in normal working condition, and the CPU utilization rate is the average CPU utilization rate collected within a predetermined time period when the electronic product component is in normal working condition.

[0068] Specifically, the sum of the weighting coefficients for rated temperature and CPU utilization is 1. Since the temperature rises rapidly when the circuit board malfunctions, the weighting coefficient for the temperature factor is generally taken as 0.6, and the weighting coefficient for the CPU utilization factor is 0.4. Only the numerical values ​​are used for calculation when calculating the working performance values.

[0069] In one specific embodiment, the rated temperature is 65°C, the measured operating temperature is 63°C, so the temperature factor is 0.95; the CPU utilization rate is 90%, the measured CPU utilization rate is 84%, so the CPU utilization factor is 0.93; and the operating performance value is 0.94.

[0070] Please see Figure 2 This is a logic diagram for determining the classification of operational anomalies of electronic products according to an embodiment of the present invention. In step S1, the classification of operational anomalies of electronic products is determined based on operational characterization values, and the proportion of image acquisition time is determined.

[0071] If the working characterization value is less than the first preset characterization value, the electronic product is determined to be in normal working condition and no image is collected.

[0072] If the working characterization value is greater than the first preset characterization value and less than the second preset characterization value, the working abnormality of the electronic product is determined to be an abnormality of the external associated system of the circuit board, and the spectrum image and oscilloscope image are acquired according to the first image acquisition duration ratio.

[0073] If the working characterization value is greater than the second preset characterization value, the abnormal operation of the electronic product is classified as an internal system abnormality of the circuit board, and the spectrum image and oscilloscope image are acquired according to the second image acquisition duration ratio.

[0074] Specifically, in practice, the actual temperature of the circuit board and the CPU utilization rate are two of the most critical intrinsic parameters for measuring the working status of the circuit board and its core components. When the actual temperature and actual CPU utilization rate of the circuit board deviate little from the data under normal working conditions, problems such as CPU failure, abnormal power supply, failure of the heat dissipation system, short circuit, and overload of the circuit board can be ruled out. Therefore, it is initially determined that the external system failure factor of the circuit board has a greater impact. Hence, the acquisition time ratio of the spectrum image to the oscilloscope image in the first image is 2:3. There is a strong coupling relationship between the actual temperature of the circuit board and the CPU utilization rate. Any internal failure will disrupt this balance and be directly reflected in these two parameters. When the actual temperature and actual CPU utilization rate of the circuit board deviate significantly from the data under normal working conditions, it is unlikely that the external failure of the circuit board will simultaneously cause a large deviation in these two core parameters. Therefore, it is initially determined that the internal system failure factor of the circuit board has a greater impact. Hence, the acquisition time ratio of the spectrum image to the oscilloscope image in the second image is 3:2.

[0075] Specifically, in one specific embodiment, the first preset characterization value is set to 1.13, the second preset characterization value is set to 1.25, and if the working characterization value of 1.09 is less than the first preset characterization value, the electronic product is determined to be in normal working condition.

[0076] If the working characterization value is 1.18, which is greater than the first preset characterization value and less than the second preset characterization value, then the working abnormality of the electronic product is determined to be an abnormality of the external associated system of the circuit board, and the spectrum image and oscilloscope image are acquired according to the first image acquisition duration ratio.

[0077] If the working characterization value is 1.28, which is greater than the second preset characterization value, then the working abnormality of the electronic product is determined to be an internal system abnormality of the circuit board, and the spectrum image and oscilloscope image are acquired according to the second image acquisition duration ratio.

[0078] It is understandable that the higher the operating temperature and CPU utilization of electronic products, the more serious the abnormality of electronic products. Therefore, the range of the first preset characterization value is 1.1 to 1.15, and the range of the second preset characterization value is 1.2 to 1.4.

[0079] Understandably, a spectrum image measures the overall effect of all vibration sources in an electronic product. This method is cost-effective and can detect malfunctions in electronic products. The oscilloscope image acquired is of the electronic product as a whole. This method is direct and efficient, requiring only one or a few probes to make a preliminary judgment on the system status.

[0080] Specifically, this invention uses a temperature sensor located inside an electronic product to collect the operating temperature and combines it with the product's CPU utilization rate to calculate the product's operating characteristics. Based on these characteristics, it determines the product's operational anomalies and the appropriate image acquisition duration. This solves the problems of resource waste and data redundancy caused by traditional indiscriminate acquisition. For example, if the characteristics indicate an anomaly in an external system connected to the circuit board, a spectrum image and an oscilloscope image are acquired using a first image acquisition duration. If the characteristics indicate an anomaly in an internal system connected to the circuit board, a second image acquisition duration is used to acquire the spectrum image and an oscilloscope image. This targeted allocation of acquisition resources transforms the acquisition behavior from full-coverage to precise targeting, achieving a synergistic optimization that minimizes resource consumption and maximizes diagnostic accuracy.

[0081] Specifically, in step S2, the process of calculating and generating frequency characteristic values ​​based on the vibration frequency and clock frequency includes,

[0082] The deviation of the vibration frequency from the rated vibration frequency is determined as the first vibration influence factor;

[0083] The deviation of the clock frequency from the rated clock frequency is determined as the first clock influence factor;

[0084] The weighted sum of the first vibration influence factor and the first clock influence factor is determined to be the frequency characteristic value.

[0085] It is understandable that vibration frequency refers to the number of times that mechanical components and the entire structure of an electronic product complete periodic reciprocating motion per unit time, measured in Hertz. It mainly originates from moving parts in electronic products, including cooling fans, hard drives, and built-in motors. Vibration frequency is a direct reflection of the mechanical health status. Clock frequency refers to the number of times that a reference signal, like a metronome, changes periodically per unit time in a digital circuit, also measured in Hertz. It mainly originates from the clock source and transmission path inside the electronic product. Clock frequency is the life cycle for the normal operation of a digital system.

[0086] Specifically, in practice, the rated vibration frequency is the average vibration frequency in the spectrum image of the electronic product collected within a predetermined time, and the rated clock frequency is the average clock frequency in the oscilloscope image of the electronic product collected within a predetermined time.

[0087] Specifically, the first vibration influence factor is the absolute value of the difference between the vibration frequency and the rated vibration frequency divided by the rated vibration frequency, and the first clock influence factor is the absolute value of the difference between the clock frequency and the rated clock frequency divided by the rated clock frequency multiplied by 10.

[0088] Specifically, the sum of the weighting coefficients of the rated vibration frequency and the rated clock frequency is 1. Since the vibration frequency and clock frequency have different degrees of influence on the working state of electronic products, the clock frequency is the foundation for the operation of core functions and affects whether the equipment can normally perform core tasks such as calculation and control. It usually has a higher priority. The vibration frequency is the guarantee of hardware stability and mainly affects the long-term reliability of the equipment. Therefore, the weighting coefficient of the first vibration influence factor is generally taken as 0.4 and the weighting coefficient of the first clock influence factor is 0.6. When calculating the first vibration influence factor, the first clock influence factor and the frequency characteristic value, only the numerical values ​​are used.

[0089] In one specific embodiment, the rated vibration frequency is 30Hz, the measured vibration frequency is 29.9Hz, so the first vibration influence factor is 0.1; the rated clock frequency is 2.9GHz, the measured clock frequency is 2.89GHz, so the first clock influence factor is 0.1; and the frequency characteristic value is 0.1.

[0090] Please see Figure 3 As shown, this is a logic diagram for determining the fault classification of electronic products according to an embodiment of the present invention. In step S3, the fault classification of the electronic product is determined based on the frequency characteristic value, wherein...

[0091] If the frequency characteristic value is less than the first preset characteristic value, the electronic product is determined to be in normal working condition.

[0092] If the frequency characteristic value is greater than the first preset characteristic value and less than the second preset characteristic value, the fault of the electronic product is determined to be a fault of the circuit board related components.

[0093] If the frequency characteristic value is greater than the second preset characteristic value, the fault of the electronic product is determined to be a fault of the internal components of the circuit board.

[0094] In one specific embodiment, the first preset feature value is set to 0.08 and the second preset feature value is set to 0.12. If the frequency feature value of 0.03 is less than the first preset deviation, the electronic product is determined to be in normal working condition.

[0095] If the frequency characteristic value is 0.11, which is greater than the first preset characteristic value and less than the second preset characteristic value, then the fault of the electronic product is determined to be a fault of the circuit board related components.

[0096] If the frequency characteristic value is 0.15, which is greater than the second preset characteristic value, then the fault of the electronic product is determined to be a fault of the internal components of the circuit board.

[0097] It is understandable that the larger the vibration influence factor and clock influence factor are, the greater the probability of abnormal operation of electronic products. Therefore, the range of the first preset characteristic value is 0.05 to 0.1, and the range of the second preset characteristic value is 0.11 to 0.13.

[0098] Please see Figure 4 As shown, this is a logic diagram for determining the defect classification of electronic products according to an embodiment of the present invention. In step S4, the defect classification of the electronic product is determined by combining the malfunction classification and the fault classification of the electronic product.

[0099] If the malfunction of an electronic product is classified as an abnormality of the external associated system of the circuit board and the fault is classified as a fault of the associated components of the circuit board, then the defect is classified as a heat dissipation defect of the circuit board.

[0100] If the malfunction of an electronic product is classified as an abnormality of the external associated system of the circuit board and the fault is classified as a fault of the internal components of the circuit board, then the defect is classified as an abnormal defect of the circuit board drive circuit.

[0101] If the malfunction of an electronic product is classified as an internal system malfunction of the circuit board and the fault is classified as a fault of the circuit board-related components, then the defect is classified as a mismatch defect of the circuit board-related components.

[0102] If the malfunction of an electronic product is classified as an internal system malfunction of the circuit board and the fault is classified as a fault of an internal component of the circuit board, then the defect is classified as a circuit board damage defect.

[0103] Understandably, determining the defect classification type involves a detailed analysis based on the pre-defined operational anomaly classification of electronic products, combined with the product's fault classification.

[0104] Specifically, this invention combines the classification of operational anomalies with fault classification of electronic products to automatically determine the final defect classification. The operational anomaly classification identifies the system location of the problem. External system anomalies refer to faults in external systems connected to and supporting the circuit board, such as cooling systems, power supply systems, and mechanical structures. Internal system anomalies refer to faults in the core functional systems of the circuit board itself, such as signal timing, power management, and logic operations. Fault classification identifies the specific damaged physical components. Circuit board-related component faults refer to functional, separable components installed on or directly connected to the circuit board, such as cooling fans, connectors, sensors, and speakers. Internal component faults refer to integrated chips or passive components constituting the core circuitry of the circuit board, such as CPUs, memory, power supplies, resistors, capacitors, and crystal oscillators. This further improves the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion.

[0105] Specifically, in step S4, the image acquisition time ratio corresponding to the defect classification is adjusted, wherein...

[0106] If the defect is classified as an abnormal defect in the circuit board driver circuit, the ratio of the first image acquisition time is adjusted to the ratio of the second image acquisition time.

[0107] If the defect is classified as a mismatch defect of related components on the circuit board, the proportion of the second image acquisition time is adjusted to the proportion of the first image acquisition time.

[0108] It is understood that the premise of adjusting the corresponding image acquisition time ratio according to defect classification in this invention is that the abnormal operation classification and the fault classification of electronic products are inconsistent. Therefore, it is necessary to divide the image acquisition time ratio a second time according to defect classification.

[0109] Specifically, the present invention determines the defect classification of electronic products by combining the malfunction classification and fault classification of electronic products. When the defect classification is determined to be an abnormal defect of the circuit board drive circuit, the first image acquisition time ratio is adjusted to the second image acquisition time ratio. When the defect classification is determined to be a mismatch defect of the circuit board associated components, the second image acquisition time ratio is adjusted to the first image acquisition time ratio.

[0110] Specifically, in step S5, the process of calculating and generating adjusted frequency characteristic values ​​based on the spectrum image of the electronic product and the oscilloscope image acquired according to the adjusted image acquisition duration ratio includes the following:

[0111] The deviation value between the adjusted vibration frequency and the rated vibration frequency is determined to be the second vibration influence factor;

[0112] The deviation value between the adjusted clock frequency and the rated clock frequency is determined to be the second clock influence factor;

[0113] The weighted sum of the second vibration influence factor and the second clock influence factor is determined to be the adjusted frequency characteristic value.

[0114] Specifically, the second vibration influence factor is the absolute value of the difference between the vibration frequency and the rated vibration frequency divided by the rated vibration frequency, and the second clock influence factor is the absolute value of the difference between the clock frequency and the rated clock frequency divided by the rated clock frequency multiplied by 10.

[0115] Specifically, in step S5, the process of calculating the adjusted frequency characteristic value and the characteristic deviation value of the frequency characteristic value includes,

[0116] Several adjusted frequency feature values ​​are acquired based on the adjusted image acquisition duration ratio for a preset duration.

[0117] The mean of the differences between several adjustment frequency characteristic values ​​and frequency characteristic values ​​is calculated as the characteristic deviation value.

[0118] Specifically, this invention collects several adjustment frequency feature values ​​for a preset duration based on the adjusted image acquisition duration ratio, and then calculates the average of the differences between the several adjustment frequency feature values ​​and the frequency feature values ​​as the feature deviation value. This avoids data errors caused by collecting a single adjustment frequency feature value, and collecting several adjustment frequency feature values ​​makes the calculation result of the feature deviation value more comprehensive, reliable and accurate.

[0119] Specifically, in step S5, the defect classification is adjusted based on the feature deviation value, wherein...

[0120] If the feature deviation value is greater than or equal to the preset feature deviation value, and the first image acquisition time ratio is adjusted to the second image acquisition time ratio, then the defect classification is adjusted to circuit board heat dissipation defect.

[0121] If the feature deviation value is greater than or equal to the preset feature deviation value, and the feature acquisition duration ratio of the second image is adjusted to the first image acquisition duration ratio, the defect classification is adjusted to circuit board damage defect.

[0122] In one specific embodiment, a preset feature deviation value of 0.1 is set. Under the condition that the ratio of the first image acquisition time is adjusted to the ratio of the second image acquisition time, if the feature deviation value of 0.14 is greater than the preset feature deviation value, the defect classification is determined to be a circuit board heat dissipation defect.

[0123] If the feature deviation value is 0.16, which is greater than the preset feature deviation value, and the defect classification is adjusted to circuit board damage defect, then the defect classification is adjusted to circuit board damage defect.

[0124] It is understandable that the purpose of setting a preset feature deviation value is to characterize the consistency between the adjusted frequency feature value and the frequency feature value. The preset feature deviation value is selected within the range [0.1, 0.2].

[0125] Specifically, this invention reclassifies electronic product defects by adjusting the image acquisition duration ratio. If the initial defect classification is an abnormal defect in the circuit board drive circuit or a mismatch defect in related components of the circuit board, it indicates that there are problems both inside and outside the circuit board of the electronic product. By adjusting the acquisition duration ratio, vibration frequency and clock frequency can be obtained more comprehensively, reducing misjudgments caused by insufficient information. At the same time, the response of internal circuits and external components is analyzed, and the acquisition strategy is adjusted. Instead of simply increasing the acquisition duration, the ratio is dynamically adjusted according to the defect type. This ensures the quality of analysis while avoiding unnecessary data redundancy, further improving the accuracy of the electronic product defect classification method based on heterogeneous sensor data fusion.

[0126] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0127] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for classifying defects in electronic products based on heterogeneous sensor data fusion, characterized in that, include: The working temperature is collected by a temperature sensor installed inside the electronic product, and the CPU utilization rate of the electronic product is combined to calculate the working characterization value of the electronic product. Based on the working characterization value, the working abnormality classification of the electronic product is determined and the proportion of image acquisition time is determined. The spectrum image and oscilloscope image of the electronic product are acquired based on the image acquisition time ratio. The vibration frequency of the electronic product is obtained from the spectrum image, and the clock frequency of the electronic product is obtained from the oscilloscope image. Frequency feature values ​​are calculated and generated based on the vibration frequency and the clock frequency, and the fault classification of the electronic product is determined based on the frequency feature values. The defect classification of the electronic product is determined by combining the abnormal operation classification and fault classification of the electronic product, and the proportion of the image acquisition duration corresponding to the defect classification is adjusted. Based on the spectrum image and oscilloscope image of the electronic product acquired according to the adjusted image acquisition time ratio, an adjusted frequency feature value is generated. The feature deviation value between the adjusted frequency feature value and the frequency feature value is calculated, and the defect classification is adjusted based on the feature deviation value.

2. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 1, characterized in that, The process of calculating the operating temperature of the electronic product by combining the operating temperature collected by the temperature sensor installed inside the electronic product with the CPU utilization rate of the electronic product includes the following steps: The ratio of the operating temperature to the rated temperature is determined as the temperature factor; The ratio of CPU utilization to CPU workload utilization is defined as the CPU utilization factor. The weighted sum of the temperature factor and the CPU utilization factor is determined to be the working characterization value.

3. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 2, characterized in that, The step involves determining the operational anomaly classification of the electronic product based on the operational characterization value and determining the image acquisition duration ratio, wherein... If the working characterization value is greater than the first preset characterization value and less than the second preset characterization value, then the working abnormality of the electronic product is determined to be an abnormality of the external associated system of the circuit board, and the spectrum image and oscilloscope image are acquired according to the first image acquisition duration ratio. If the working characteristic value is greater than the second preset characteristic value, the working abnormality of the electronic product is determined to be an internal system abnormality of the circuit board, and a spectrum image and an oscilloscope image are acquired according to the second image acquisition duration ratio.

4. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 3, characterized in that, The process of calculating and generating frequency characteristic values ​​based on the vibration frequency and the clock frequency includes: The deviation value between the vibration frequency and the rated vibration frequency is determined to be the first vibration influence factor; The deviation value between the clock frequency and the rated clock frequency is determined to be the first clock influence factor; The weighted sum of the first vibration influence factor and the first clock influence factor is determined to be the frequency characteristic value.

5. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 4, characterized in that, The fault classification of the electronic product is determined based on the frequency characteristic values, wherein... If the frequency characteristic value is greater than the first preset characteristic value and less than the second preset characteristic value, then the fault of the electronic product is determined to be a fault of the circuit board related components. If the frequency characteristic value is greater than the second preset characteristic value, the fault of the electronic product is determined to be a fault of internal components of the circuit board.

6. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 5, characterized in that, The defect classification of the electronic product is determined by combining the malfunction classification and fault classification of the electronic product, wherein... If the malfunction of the electronic product is classified as an abnormality of the external associated system of the circuit board and the fault is classified as a fault of the associated component of the circuit board, then the defect is classified as a heat dissipation defect of the circuit board. If the malfunction of the electronic product is classified as an abnormality of the external associated system of the circuit board and the fault is classified as a fault of the internal components of the circuit board, then the defect is determined to be an abnormal defect of the circuit board drive circuit. If the malfunction of the electronic product is classified as an internal system malfunction of the circuit board and the fault is classified as a fault of the associated components of the circuit board, then the defect is classified as a mismatch defect of the associated components of the circuit board. If the malfunction of the electronic product is classified as an internal system malfunction of the circuit board and the fault is classified as a fault of an internal component of the circuit board, then the defect is classified as a circuit board damage defect.

7. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 6, characterized in that, The adjustment of the image acquisition duration ratio corresponding to the defect classification, wherein... If the defect is determined to be an abnormal defect in the circuit board driver circuit, then the ratio of the first image acquisition time is adjusted to the ratio of the second image acquisition time. If the defect is classified as a mismatch defect of associated components on the circuit board, the ratio of the second image acquisition time is adjusted to the ratio of the first image acquisition time.

8. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 7, characterized in that, The process of calculating and generating adjusted frequency characteristic values ​​based on the spectrum image and oscilloscope image of the electronic product acquired according to the adjusted image acquisition duration ratio includes the following steps: The deviation value between the adjusted vibration frequency and the rated vibration frequency is determined to be the second vibration influence factor; The deviation value between the adjusted clock frequency and the rated clock frequency is determined to be the second clock influence factor; The weighted sum of the second vibration influence factor and the second clock influence factor is determined to be the adjusted frequency characteristic value.

9. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 8, characterized in that, The process of calculating the adjusted frequency characteristic value and the characteristic deviation value of the frequency characteristic value includes: Several adjusted frequency feature values ​​are acquired based on the adjusted image acquisition duration ratio for a preset duration. The average of the differences between several of the adjusted frequency characteristic values ​​and the frequency characteristic values ​​is calculated as the characteristic deviation value.

10. The electronic product defect classification method based on heterogeneous sensor data fusion according to claim 9, characterized in that, The defect classification is adjusted based on the feature deviation value, wherein... If the feature deviation value is greater than or equal to the preset feature deviation value, and the first image acquisition duration ratio is adjusted to the second image acquisition duration ratio, then the defect classification is determined to be a circuit board heat dissipation defect. If the feature deviation value is greater than or equal to the preset feature deviation value, and the second image acquisition duration ratio is adjusted to the first image acquisition duration ratio, then the defect classification is determined to be a circuit board damage defect.

Images