A method and system for detecting the production and assembly quality of a vehicle-mounted starting power supply

Abstract

The present application relates to the technical field of battery manufacturing and automatic detection, and particularly relates to a method and system for detecting the production and assembly quality of a vehicle-mounted starting power supply. The method comprises the following steps: calculating a material anomaly factor of a pixel point according to the gradient amplitude of the pixel point emissivity, the coefficient of variation in the eight-neighborhood of the pixel point, and the difference between the pixel point emissivity and the material reference emissivity; calculating a spatial anomaly factor of the pixel point according to the gradient amplitude of the pixel point temperature value, the temperature weighted variance in the local range of the pixel point, and the difference between the pixel point temperature and the minimum temperature in the power supply production thermal image; obtaining a target pixel point in response to the material anomaly factor and the spatial anomaly factor of the pixel point being greater than the corresponding threshold values; and processing the target pixel point based on the LOF algorithm to obtain the production and assembly quality detection result of the vehicle-mounted starting power supply, thereby effectively improving the efficiency of the production and assembly quality detection result of the vehicle-mounted starting power supply.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery manufacturing and automatic detection, and particularly relates to a method and system for detecting the production and assembly quality of a vehicle starting power supply. BACKGROUND

[0002] As a core component for ensuring the stable operation of the electrical system of a vehicle, the production and assembly quality of the vehicle starting power supply is directly related to the safety and reliability of the vehicle. In the process of manufacturing a vehicle, defects in the assembly quality of the power supply may cause problems such as virtual welding of the electrode lug of the battery, insulation failure of the module, or parameter drift of the protection plate. Therefore, a precise and efficient quality detection technology is crucial in the production process.

[0003] At present, when detecting the production and assembly quality of the vehicle starting power supply, a detection scheme based on a high-precision thermal imager and a local outlier factor (LOF) algorithm is usually used. In this scheme, the high-precision thermal imager is used to capture the thermal image of the key area of the product in a very short time after the pulse ends, and then the LOF algorithm is used to identify abnormal temperature data, thereby judging the production quality of the vehicle starting power supply. The principle of this scheme is that the power supply normally assembled has consistency in heat conduction and heat radiation characteristics, while quality defects such as virtual welding and uneven application of thermal grease will cause abnormal local temperature distribution. The LOF algorithm can effectively locate the temperature abnormal points by calculating the distance relationship between the pixel points.

[0004] However, the traditional LOF algorithm has a very high computational complexity. In the execution process, the distance between each pixel point in the thermal image and other points needs to be calculated to evaluate the degree of local density deviation of the pixel point. The number of pixel points included in the thermal image is large, and complete LOF calculation for each point will generate a large amount of computation and increase the detection energy consumption. If the pixel sampling is greatly reduced or the calculation process is simplified in order to shorten the detection time, small defects may be missed or normal areas may be misjudged, which will reduce the accuracy and reliability of the detection. Therefore, how to maintain high accuracy and reliable detection results of the production and assembly quality of the vehicle starting power supply while ensuring the detection efficiency is a problem to be solved at present. SUMMARY

[0005] To solve the technical problem of how to maintain high accuracy and reliable detection results of the production and assembly quality of the vehicle starting power supply while ensuring the detection efficiency, the present application provides a method and system for detecting the production and assembly quality of a vehicle starting power supply.

[0006] In a first aspect, the present application provides a method for detecting the production and assembly quality of a vehicle starting power supply, which adopts the following technical scheme:

[0007] A method for detecting the production and assembly quality of a vehicle starting power supply, comprising the following steps:

[0008] The temperature value and the emissivity of each pixel point in the power production thermal image are acquired, the material anomaly factor of the pixel point is calculated according to the gradient amplitude of the emissivity of the pixel point, the coefficient of variation in the eight-neighbor domain of the pixel point, and the difference between the emissivity of the pixel point and the material reference emissivity, the local range of the pixel point is constructed with the pixel point as the center, the weight coefficient of other pixel points in the local range to the pixel point is calculated according to the absolute value of the temperature difference between the pixel point and other pixel points in the local range and the included angle of the gradient direction, the temperature weighted variance in the local range of the pixel point is acquired, the spatial anomaly factor of the pixel point is calculated according to the gradient amplitude of the temperature value of the pixel point, the temperature weighted variance in the local range of the pixel point, and the difference between the temperature of the pixel point and the minimum temperature in the power production thermal image, the target pixel point is obtained in response to the material anomaly factor and the spatial anomaly factor of the pixel point being greater than the corresponding threshold value, and the on-board starting power production and assembly quality detection result is obtained by processing the target pixel point based on the LOF algorithm.

[0009] The present application can accurately obtain the on-board starting power production and assembly quality detection result by processing the temperature value of each pixel point in the power production thermal image through the LOF. In the process of obtaining the detection result, the present application considers that the number of pixel points in the power production thermal image is huge, and the time consumed by directly performing the LOF detection is long, so the present application screens out the pixel points with higher abnormality degree for abnormality detection by analyzing the possibility of each pixel point being an abnormal temperature point, which can reduce the data processing amount while ensuring the accuracy of the detection result, and effectively improve the efficiency of obtaining the detection result. When obtaining the possibility of each pixel point being an abnormal temperature point, the present application obtains the material anomaly factor of the pixel point by analyzing the change of the emissivity of the pixel point at the same position, so as to capture the thermal diffusion anomaly, and obtains the spatial anomaly factor of the pixel point by analyzing the change of the temperature gradient amplitude and the confusion degree, so as to capture the material interface defect, so that the LOF detection can be accurately focused on the defect point with higher risk by combining the thermal diffusion anomaly and the material interface defect, and the efficiency of obtaining the on-board starting power production and assembly quality detection result is effectively improved.

[0010] According to the on-board starting power production and assembly quality detection method provided by the present application, the temperature value and the emissivity of each pixel point in the power production thermal image are acquired, including: shooting the temperature data set and the emissivity data set in the thermal image in the power production process after a standard transient large current pulse is applied to the starting power, and preprocessing the temperature data set and the emissivity data set in the thermal image in the power production process to obtain the temperature value and the emissivity of each pixel point in the power production thermal image, wherein the preprocessing mode at least includes normalizing the temperature data set and the emissivity data set.

[0011] The present application considers that image noise, non-uniform data dimensions and the like can affect the accuracy of subsequent data processing, and therefore provides a partial preprocessing means for processing data, which can effectively improve data quality and prepare for subsequent data analysis.

[0012] According to the vehicle starting power supply production and assembly quality detection method provided by the application, the material abnormal factor of the pixel point is calculated, including: obtaining the material reference emissivity data of the pixel point for normalization processing to obtain the material reference emissivity of the pixel point; calculating the material abnormal factor of the first pixel point :

[0013] ;

[0014] The gradient amplitude of the emissivity of the first pixel point, , , The emissivity standard deviation, emissivity mean value and coefficient of variation in the eight-neighborhood of the first pixel point, , The emissivity and material reference emissivity of the first pixel point, is the logarithmic function with e as the base, is the absolute value symbol, is a linear normalization function.

[0015] The present application considers that when the material interface has a defect mutation, the thermal diffusion continuity will be destroyed, and the emissivity of the pixel point on the material will change significantly, so that the pixel point emissivity gradient amplitude is obtained to capture the material interface mutation, the coefficient of variation is used to capture the possibility of false welding, and the emissivity reference deviation degree is used to assist in identifying material pollution, so that the material abnormal factor of the pixel point can be accurately obtained.

[0016] According to the vehicle starting power supply production and assembly quality detection method provided by the application, the local range of the pixel point is constructed with the pixel point as the center, including: the local range size of the preset pixel point is , the local range of the pixel point is constructed by acquiring pixel points around the pixel point with the pixel point as the center.

[0017] According to the vehicle starting power supply production and assembly quality detection method provided by the application, the weight coefficient of other pixel points in the local range to the pixel point is calculated, including:

[0018] ;

[0019] For the first In the local area of ​​the nth pixel The pixel point is the first The weight coefficients of each pixel For the first The pixel and the In the local area of ​​the nth pixel The cosine of the angle between the gradient directions of each pixel. For the first The pixel and the In the local area of ​​the nth pixel The absolute value of the temperature difference between each pixel. , These represent the maximum and minimum temperatures in the thermal image of the power source, respectively.

[0020] This invention takes into account the continuity of thermal diffusion; for a temperature point experiencing an abnormal change, its local area will also exhibit an abnormal response. Therefore, before obtaining the spatial anomaly factor of a pixel, this invention obtains the gradient direction angle and temperature difference between the pixel and other pixels in its local area. The larger the value, the more likely the direction of other pixels in the local area is the abnormal thermal diffusion direction of this pixel. Therefore, this invention assigns higher weight coefficients to other pixels that differ significantly from the current pixel to increase the contribution of these other pixels to the local disorder of the current pixel when calculating the spatial anomaly factor, thereby accurately obtaining the spatial anomaly factor of the current pixel.

[0021] According to the present invention, a method for quality inspection of vehicle-mounted starting power supply production assembly includes calculating the spatial anomaly factor of the pixel, comprising:

[0022] ;

[0023] For the first Spatial anomaly factor of each pixel, For the first Normalized gradient magnitude of temperature values ​​for each pixel For the first Temperature-weighted variance in a local area of ​​each pixel For the first Temperature value of each pixel , These represent the maximum and minimum temperatures in the thermal image of the power source, respectively. It is an exponential function with base e.

[0024] The application provides a vehicle-mounted starting power supply production and assembly quality detection method, and the target pixel point is processed based on a LOF algorithm to obtain a vehicle-mounted starting power supply production and assembly quality detection result, which comprises the following steps of: obtaining a comprehensive score of the target pixel point according to a material abnormality factor and a space abnormality factor of the target pixel point; obtaining comprehensive scores of all target pixel points in descending order, and obtaining a preset number of target pixel points in the comprehensive scores in descending order as candidate points; calculating a LOF value of the candidate points, calculating an abnormal threshold value according to a mean value and a standard deviation of the LOF values of all candidate points, and obtaining an abnormal temperature point in the vehicle-mounted starting power supply production and assembly quality detection result according to a comparison result of the LOF value of the candidate points and the abnormal threshold value.

[0025] The application provides a vehicle-mounted starting power supply production and assembly quality detection method, and the calculation of the LOF value of the candidate points comprises the following steps of: presetting neighbor points in a K-distance neighborhood of the candidate points to obtain a local density of the candidate points, and taking a mean value of a ratio of the neighbor points to the local density of the candidate points as the LOF value of the candidate points.

[0026] The application provides a vehicle-mounted starting power supply production and assembly quality detection method, and the obtaining of the vehicle-mounted starting power supply production and assembly quality detection result further comprises the following steps of: clustering the abnormal temperature points to obtain a plurality of clustering clusters, obtaining a material abnormality factor mean value and a space abnormality factor mean value of the abnormal temperature points in the clustering clusters; if the material abnormality factor mean value of the clustering cluster is greater than the space abnormality factor mean value, the abnormality of the clustering cluster is a material interface defect; if the material abnormality factor mean value of the clustering cluster is less than the space abnormality factor mean value, the abnormality of the clustering cluster is a thermal diffusion abnormality; and if the material abnormality factor mean value of the clustering cluster is equal to the space abnormality factor mean value, the abnormality of the clustering cluster is a material interface defect and a thermal diffusion abnormality.

[0027] The application considers that the abnormal temperature points in the thermal image of the power supply production may be caused by a material interface defect or a thermal diffusion abnormality, and thus the dominant abnormality factor in the clustering cluster is obtained through clustering analysis, so that the abnormal source of the abnormal temperature points can be accurately distinguished, and subsequent analysis and processing of workers are facilitated.

[0028] In the second aspect, the application provides a vehicle-mounted starting power supply production and assembly quality detection system, which adopts the following technical scheme.

[0029] The vehicle-mounted starting power supply production and assembly quality detection system comprises a processor and a memory, and the memory stores computer program instructions, and the computer program instructions are executed by the processor to realize the vehicle-mounted starting power supply production and assembly quality detection method.

[0030] By adopting the technical scheme, the computer program of the vehicle-mounted starting power supply production and assembly quality detection method is generated and stored in the memory to be loaded and executed by the processor, so that the terminal device is manufactured according to the memory and the processor, and use is facilitated.

[0031] The present application has the following technical effects:

[0032] Based on the above technical scheme, the present application provides a vehicle-mounted starting power supply production and assembly quality detection method and system, which can accurately obtain the vehicle-mounted starting power supply production and assembly quality detection result by processing the temperature values of each pixel point in the power supply production thermal image through LOF. In the process of obtaining the detection result, the present application considers that the number of pixel points in the power supply production thermal image is huge, and the time consumed by direct LOF detection is long, therefore, the present application screens out the pixel points with higher abnormality degree for abnormality detection by analyzing the possibility of each pixel point being an abnormal temperature point, which can reduce the data processing amount while ensuring the accuracy of the detection result, and effectively improve the efficiency of obtaining the detection result. In the process of obtaining the possibility of each pixel point being an abnormal temperature point, the present application obtains the material abnormality factor of the pixel point at the same position by analyzing the change of the emissivity of the pixel point, so as to capture thermal diffusion abnormality, and obtains the spatial abnormality factor by analyzing the change of the temperature gradient amplitude and the degree of confusion, so as to capture material interface defects, so that by combining the thermal diffusion abnormality and the material interface defects, the LOF detection can be accurately focused on the defect points with higher risk, and the efficiency of obtaining the vehicle-mounted starting power supply production and assembly quality detection result is effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 A flowchart of a vehicle-mounted starting power supply production and assembly quality detection method provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments.

[0035] The embodiment of the present application discloses a vehicle-mounted starting power supply production and assembly quality detection method, which captures thermal diffusion abnormality by calculating the spatial abnormality factor in the power supply production thermal image, and identifies material interface defects by the material abnormality factor. The high-risk target pixel points are focused by double screening of the material abnormality factor and the spatial abnormality factor, and only the target pixel points screened out are subjected to LOF abnormality detection, so that high-precision and rapid detection is finally realized, and the vehicle-mounted starting power supply production and assembly quality detection efficiency is effectively improved.

[0036] For details, please refer to Figure 1 as shown, Figure 1A flowchart of a vehicle-mounted starting power supply production assembly quality detection method provided by the embodiment of the present application is shown in the figure, and the method specifically comprises the following steps:

[0037] S1: Obtain the temperature value and emissivity corresponding to each pixel point in the power supply production thermal image.

[0038] It should be noted that in the power supply production thermal image, the instantaneous large current pulse is an extreme situation that the power supply may encounter in actual use. By applying a standard pulse, the stability of the power supply under short-time high load can be tested. Defects such as poor welding inside the power supply may not be obvious under normal current, but under a large current pulse, the power loss will increase rapidly and form a temperature anomaly point. Therefore, the embodiment of the present application obtains the temperature to intuitively reflect the state of the power supply production thermal image. In addition, material interface anomalies are usually manifested as fluctuations in emissivity.

[0039] Therefore, the embodiment of the present application can analyze the temperature value of each pixel point in the power supply production thermal image to capture thermal diffusion anomalies, and identify material interface defects through emissivity, thereby accurately locating the target pixel point with high risk.

[0040] For example, in the embodiment of the present application, obtaining the temperature value and emissivity corresponding to each pixel point in the power supply production thermal image comprises: shooting the temperature data set and the emissivity data set in the thermal image during the power supply production process after applying a standard instantaneous large current pulse to the starting power supply, and preprocessing the temperature data set and the emissivity data set in the thermal image during the power supply production process to obtain the temperature value and the emissivity corresponding to each pixel point in the power supply production thermal image, wherein the preprocessing method at least includes normalizing the temperature data set and the emissivity data set.

[0041] The standard instantaneous large current pulse can be set to (500A, 100ms), and can be set according to actual needs.

[0042] Specifically, the key areas of the product to be detected (such as the connection between the battery cells, the PCB mainboard and other areas) can be shot within a very short time (for example, 50ms) after the end of the pulse, the thermal image during the power supply production process can be shot by a high-precision thermal imager, each pixel point corresponds to a temperature data, and the temperature data of all pixel points form a temperature data set.

[0043] A high-precision contact infrared temperature measurement array is laid out to ensure that it can cover the key areas of the product. According to the high-precision contact infrared temperature measurement array, the contact actual temperature measured by each contact point is obtained, the radiation energy at each pixel point is collected by a thermal imager, and the contact actual temperature and the radiation energy are substituted into the Stefan-Boltzmann law formula to obtain the emissivity data of each pixel point, and the emissivity data of all pixel points form an emissivity data set.

[0044] The specific steps for obtaining the emissivity data of pixels can be implemented using existing technologies, and will not be elaborated upon here in this embodiment of the invention. After obtaining the temperature dataset and emissivity dataset according to the above steps, the temperature dataset and emissivity dataset can be preprocessed.

[0045] Specifically, adaptive filtering is performed on the temperature dataset and emissivity dataset to remove high-frequency noise. Then, Z-score standard normalization is performed on the temperature dataset and Min-Max standard normalization is performed on the emissivity dataset to finally obtain the temperature value and emissivity corresponding to each pixel.

[0046] After obtaining the temperature value and emissivity of each pixel in the thermal image of power generation based on the above steps, the material anomaly factor and spatial anomaly factor of the pixel can be calculated based on this, i.e., the following steps are performed.

[0047] S2: Calculate the material anomaly factor of the pixel based on the gradient magnitude of the pixel's emissivity, the coefficient of variation in the pixel's eight neighborhoods, and the difference between the pixel's emissivity and the material reference emissivity.

[0048] It should be noted that emissivity is used to characterize the heat dissipation capacity of a material surface through radiation. Abrupt changes at the material interface can disrupt the continuity of heat diffusion. Through increased thermal resistance, temperature field distortion, and microstructure scattering effects, local abnormal fluctuations in emissivity can occur. Therefore, by analyzing the fluctuations in emissivity, the material anomaly factor of the pixel can be obtained.

[0049] It is understood that emissivity is a dimensionless scalar field, while gradient can describe the trend of scalar field changes in space. The gradient of emissivity can reflect the intensity and direction of its spatial abrupt change. The emissivity gradient amplitude of defects such as dummy solder joints is much higher than that of normal material interfaces, and can be used as a feature parameter for defect identification. Therefore, embodiments of the present invention can obtain the gradient amplitude of emissivity to analyze the degree of abrupt change at the material interface where the pixel is located.

[0050] For example, in an embodiment of the present invention, obtaining the gradient magnitude of the emissivity of a pixel includes:

[0051] ;

[0052] For the first The gradient magnitude of the emissivity of each pixel. For the first Partial derivatives of emissivity in the x-direction of each pixel. For the first 1 pixel Partial derivative of emissivity in the direction, It is a linear normalization function.

[0053] wherein the partial derivative of the emissivity of the first pixel point in the x direction and the partial derivative of the emissivity of the first pixel point in the y direction can be calculated by the central difference method, and the present embodiment will not be described herein. The partial derivative of the emissivity of the first pixel point in the x direction and the partial derivative of the emissivity of the first pixel point in the y direction can be calculated by the central difference method, and the present embodiment will not be described herein.

[0054] Based on the above steps, the gradient amplitude of the emissivity of each pixel point can be obtained. In the production of the thermal image of the power supply, the material abnormality can include virtual welding or material pollution. The material in the region where each pixel point is located has a corresponding material reference emissivity. If the material is polluted or oxidized, the gradient amplitude can not change, but the heat conduction can be obviously abnormal, resulting in a significant change in the emissivity value.

[0055] Based on this, when the material abnormality factor of the pixel point is obtained, the material abnormality factor of the pixel point can be calculated by combining the gradient amplitude of the pixel point, the fluctuation of the emissivity, and the deviation degree between the emissivity and the reference value, so as to accurately obtain the possibility that the material in the region where the pixel point is located is abnormal.

[0056] For example, before calculating the material abnormality factor of the pixel point, the material reference emissivity data of the pixel point can be obtained and normalized to obtain the material reference emissivity of the pixel point.

[0057] The material reference emissivity data of the pixel point can be obtained from the material corresponding to the position of the pixel point. Since the emissivity data of the material is usually in a small fluctuation interval, the present embodiment can take the middle value of the interval of the emissivity data as the material reference emissivity of the pixel point. The present embodiment will not be limited too much in this regard.

[0058] For example, in the present embodiment, the material abnormality factor of the pixel point is calculated, including:

[0059] ;

[0060] The material abnormality factor of the first pixel point, The gradient amplitude of the emissivity of the first pixel point, The standard deviation of the emissivity in the eight-neighborhood of the first pixel point, The mean value of the emissivity in the eight-neighborhood of the first pixel point, The emissivity of the first pixel point, The material reference emissivity of the first pixel point, The material reference emissivity of the first pixel point, The material reference emissivity of the first pixel point, The material reference emissivity of the first pixel point, The material reference emissivity of the first pixel point, The material reference emissivity of the first pixel point, The material reference emissivity of the first pixel point, The logarithmic function with e as the base,​ It is the absolute value symbol. It is a linear normalization function.

[0061] In this calculation method, For the first The coefficient of variation in the eight neighborhood of each pixel.

[0062] No. The larger the gradient magnitude of the emissivity of the i-th pixel, the stronger the i-th pixel's emissivity gradient. The greater the likelihood of an abrupt change in the material interface at a given pixel, the higher the probability of such a change.

[0063] In thermal imaging of power supply production, materials typically have low emissivity. If a pixel on the material has a poor solder joint, the number of low-emissivity pixels will increase due to the exposed surrounding metal, leading to a decrease in the average pixel emissivity. Furthermore, the breakage of the solder joint caused by the poor solder joint will generate significant fluctuations, increasing the standard deviation of the pixel's emissivity. Therefore, the larger the coefficient of variation, the greater the likelihood of an anomaly in the corresponding material. If the pixel is not a poor solder joint, its coefficient of variation will approach 0.

[0064] This is a penalty term for the degree of deviation from the emissivity benchmark. Combined with the gradient magnitude, this term helps identify material contamination or oxidation. The greater the deviation of a pixel's emissivity from the corresponding material benchmark emissivity, the greater the likelihood of an abnormal emissivity change. The emissivity difference is quantified using a logarithmic function, so that when... When it approaches 0, the value of this term is 0; when When the value is small, this term grows approximately linearly; when... When the value is large, the increase rate is relatively slow to prevent extreme values ​​from dominating the calculation of material anomaly factors.

[0065] In summary, the larger the gradient amplitude of a pixel, the greater the possibility of an abnormal mutation. If the coefficient of variation of the pixel is larger, the greater the possibility that the anomaly is caused by a poor solder joint defect. Furthermore, if the emissivity of the pixel deviates more from the reference emissivity, the higher the credibility of the pixel having an abnormal mutation such as a poor solder joint or material oxidation, and the greater the corresponding material anomaly factor.

[0066] The material anomaly factor of each pixel in the thermal image of power generation can be obtained by following the above steps.

[0067] S3: Calculate the spatial anomaly factor of the pixel based on the gradient magnitude of the pixel temperature value, the temperature-weighted variance in the local area of ​​the pixel, and the difference between the pixel temperature and the minimum temperature in the power generation thermal image.

[0068] The gradient amplitude of the pixel point temperature value can be calculated by a Scharr operator, which is not described herein again.

[0069] It should be noted that, due to thermal diffusion continuity, the pixel points in the abnormal area have two characteristics in space, which are temperature gradient amplitude mutation and local disorder degree anomaly. The temperature gradient amplitude mutation specifically represents that the defect point causes local temperature change to be violent due to heat flow obstruction, so that the temperature gradient amplitude increases. The local disorder degree anomaly specifically represents that the temperature distribution in the heat diffusion direction presents non-uniformity, and the variance is large.

[0070] Based on this, the gradient direction angle and the temperature of the pixel points in the local range of the current pixel point can be obtained, so as to obtain the weight of each pixel point in the local range of the current pixel point in calculating the temperature weighted variance of the current pixel point. If the temperature difference between the pixel points in the local range and the current pixel point is larger and the gradient direction angle is larger, it is indicated that the pixel point can represent the heat diffusion direction of the current pixel point anomaly, and the weight needs to be increased.

[0071] The gradient direction angle range is [0, 180].

[0072] For example, in the embodiment of the present application, the local range of the pixel point is constructed with the pixel point as the center, including: the local range size of the preset pixel point is The local range of the pixel point is constructed with the pixel point as the center, and pixels around the pixel point are obtained.

[0073] The gradient direction angle is calculated according to the gradient of the pixel point and the gradient of the other pixel points in the local range. It can be set to 5; the local range size of the current pixel point is the number of pixel points contained in the local range, including the current pixel point itself, and the local range size can be set according to actual needs.

[0074] For example, the weight coefficient of the other pixel points in the local range to the pixel point can be calculated according to the absolute value of the temperature difference between the pixel point and the other pixel points in the local range and the gradient direction angle.

[0075] It can be understood that the other pixel points in the local range of the current pixel point are each pixel point in the local range, and also include the current pixel point itself.

[0076] For example, in the embodiment of the present application, the weight coefficient of the other pixel points in the local range to the pixel point is calculated, including:

[0077] ;

[0078] The weight coefficient of the first pixel point in the local range of the first pixel point to the first pixel point is calculated. The weight coefficient of the first pixel point in the local range of the first pixel point to the first pixel point is calculated. The pixel point is the first The weight coefficients of each pixel For the first The pixel and the In the local area of ​​the nth pixel The cosine of the angle between the gradient directions of each pixel. It is a cosine function. For the first The pixel and the In the local area of ​​the nth pixel The absolute value of the temperature difference between each pixel. The maximum temperature in the thermal image of the power source. The minimum temperature in the thermal image of the power source.

[0079] In this calculation method, if the gradient direction angle between the current pixel and other pixels in its neighborhood is larger and the temperature difference is greater, then the weight of the other pixels in calculating the local disorder of the current pixel will be greater.

[0080] For example, when obtaining the temperature-weighted variance in the local range of the current pixel, you can first use the product of the temperature value of the pixel in the local range of the current pixel and the weight coefficient as the target temperature value of the pixel, and then use the variance of the target temperature value in the local range of the current pixel as the temperature-weighted variance in the local range of the current pixel.

[0081] For example, in an embodiment of the present invention, calculating the spatial anomaly factor of the pixel includes:

[0082] ;

[0083] For the first Spatial anomaly factor of each pixel, For the first Normalized gradient magnitude of temperature values ​​for each pixel For the first Temperature-weighted variance in a local area of ​​each pixel For the first Temperature value of each pixel The maximum temperature in the thermal image of the power source. The minimum temperature in the thermal image of the power source. It is an exponential function with base e.

[0084] In this calculation method, Characterizing the first The larger the value of the temperature gradient amplitude changes within a local area of ​​a pixel, the greater the change in the local temperature gradient amplitude.

[0085] characterize the local chaos degree in the local range of the i-th pixel point, and the greater the value, the greater the local chaos degree.

[0086] is a high-temperature enhancement term. In general, the temperature of an abnormal pixel point is higher than that of a normal pixel point. Therefore, the high-temperature enhancement term is used to give a higher weight to a high-temperature region. The greater the difference between the temperature value of the i-th pixel point and the minimum temperature value and the difference between the maximum temperature value and the minimum temperature value, the closer the ratio is to 1, and the greater the temperature value of the i-th pixel point.

[0087] In summary, the greater the gradient amplitude of a pixel point, the greater the temperature weighted variance in the local range, and the greater the high-temperature enhancement term, the higher the temperature of the pixel point and the greater the temperature gradient amplitude mutation and local chaos degree, and therefore the greater the spatial anomaly factor of the pixel point.

[0088] After obtaining the material anomaly factor and the spatial anomaly factor of the pixel points according to the above steps, the target pixel points with a higher abnormal degree can be accurately screened out.

[0089] S4: obtaining the target pixel point in response to the material anomaly factor and the spatial anomaly factor of the pixel point being greater than the corresponding threshold value; processing the target pixel point based on the LOF algorithm to obtain the production and assembly quality detection result of the vehicle-mounted starting power supply.

[0090] The threshold value can be set according to actual needs. As an example, the target pixel point is obtained according to the comparison result of the material anomaly factor and the spatial anomaly factor of the pixel point with the corresponding threshold value, including: sorting all pixel points in descending order according to the material anomaly factor and the spatial anomaly factor of the pixel point to obtain a material anomaly sequence and a spatial anomaly sequence; taking the value at the 75% position in the material anomaly sequence and the spatial anomaly sequence as the threshold value of the material anomaly factor and the threshold value of the spatial anomaly factor, respectively; if the material anomaly factor and the spatial anomaly factor of the pixel point are greater than the corresponding threshold value, the pixel point is the target pixel point.

[0091] ​​​For example, in the embodiment of the present application, the target pixel point is processed based on the LOF algorithm to obtain the production and assembly quality detection result of the vehicle starting power supply, which comprises: obtaining the comprehensive score of the target pixel point according to the material anomaly factor and the spatial anomaly factor of the target pixel point; obtaining the comprehensive scores of all target pixel points in descending order, and obtaining a preset number of target pixel points in the descending order of the comprehensive scores as candidate points; calculating the LOF value of the candidate points, calculating the abnormal threshold value according to the mean and standard deviation of the LOF values of all candidate points, and obtaining the abnormal temperature point in the production and assembly quality detection result of the vehicle starting power supply according to the comparison result of the LOF value of the candidate point and the abnormal threshold value.

[0092] The preset number of target pixel points can be set to the number of target pixel points in the top 30% of the descending order of the comprehensive scores.

[0093] For example, when the comprehensive score of the target pixel point is obtained according to the material anomaly factor and the spatial anomaly factor of the target pixel point, the ratio of the material anomaly factor of the pixel point to the maximum value of the material anomaly factor in the power production thermal image can be obtained first, which is recorded as the material anomaly coefficient of the pixel point; the ratio of the spatial anomaly factor of the current pixel point to the maximum value of the spatial anomaly factor in the power production thermal image is obtained, which is recorded as the spatial anomaly coefficient of the pixel point; the weight of the material anomaly coefficient is set to 0.6, the weight of the spatial anomaly coefficient is set to 0.4, and the weighted sum of the material anomaly coefficient and the spatial anomaly coefficient of the pixel point is taken as the comprehensive score of the pixel point.

[0094] The weight can be set according to actual needs, and the embodiment of the present application does not make too many limitations here.

[0095] For example, in the embodiment of the present application, the LOF value of the candidate point is calculated, which comprises: obtaining the neighbor points in the K-distance neighborhood of the preset candidate point to obtain the local density of the candidate point, and taking the mean of the ratio of the neighbor points to the local density of the candidate point as the LOF value of the candidate point.

[0096] The K can be set to 20, and can be set according to actual needs. The specific steps of calculating the LOF value of the candidate point can be realized by the prior art, and the embodiment of the present application does not make too many limitations here.

[0097] For example, when the neighbor points in the K-distance neighborhood of the candidate point are obtained, the distances between the candidate point and other target pixel points can be arranged in ascending order from left to right, and the target pixel points corresponding to the K distances from the left end are obtained as the neighbor points of the candidate point.

[0098] It can be understood that the neighbor points obtained in this way are the K nearest target pixel points to the candidate point. The larger the LOF value of the candidate point is, the more abnormal it is.

[0099] For example, the abnormal threshold is calculated according to the mean and standard deviation of the LOF values of all candidate points, and the abnormal temperature points in the quality detection result of the vehicle starting power production and assembly are obtained according to the comparison result of the LOF values of the candidate points and the abnormal threshold.

[0100] Specifically, when the abnormal threshold is calculated according to the mean and standard deviation of the LOF values of all candidate points, the sum of the mean of the LOF values of all candidate points and twice the standard deviation can be recorded as the abnormal threshold. The abnormal threshold can be set according to actual needs.

[0101] According to the above steps, the abnormal temperature points in the power production thermal image can be obtained, and the abnormal temperature points are the position points in the power production thermal image that may have quality abnormalities. After obtaining the abnormal temperature points, the abnormal types that the abnormal temperature points may correspond to can be obtained, so as to facilitate the processing of the staff.

[0102] For example, in the embodiment of the present application, after obtaining the quality detection result of the vehicle starting power production and assembly, the method further includes: clustering the abnormal temperature points to obtain a plurality of clustering clusters, obtaining the material abnormal factor mean and the space abnormal factor mean of the abnormal temperature points in the clustering cluster; if the material abnormal factor mean of the clustering cluster is greater than the space abnormal factor mean, the abnormal of the clustering cluster is a material interface defect; if the material abnormal factor mean of the clustering cluster is less than the space abnormal factor mean, the abnormal of the clustering cluster is a thermal diffusion abnormality; and if the material abnormal factor mean of the clustering cluster is equal to the space abnormal factor mean, the abnormal of the clustering cluster is a material interface defect and a thermal diffusion abnormality.

[0103] After clustering the abnormal temperature points to obtain a plurality of clustering clusters, the clustering clusters can be screened first, and the larger clustering clusters can be analyzed.

[0104] For example, the clustering cluster in which the number of abnormal temperature points is less than 50 can be excluded.

[0105] It can be understood that for the clustering cluster in which the number of abnormal temperature points is not less than 50, it means that there may be a larger defect, which needs to be processed as soon as possible.

[0106] It can be seen that in the embodiment of the application, when the vehicle starting power supply production and assembly quality detection result is acquired, the temperature value and the emissivity corresponding to each pixel point in the power supply production thermal image can be acquired; the material anomaly factor of the pixel point is calculated according to the gradient amplitude of the pixel point emissivity, the variation coefficient in the eight-neighbor domain of the pixel point, and the difference between the emissivity of the pixel point and the material reference emissivity; the local range of the pixel point is constructed with the pixel point as the center, the weight coefficient of other pixel points in the local range to the pixel point is calculated according to the absolute value of the temperature difference between the pixel point and other pixel points in the local range and the included angle of the gradient direction, the temperature weighted variance in the local range of the pixel point is acquired; the spatial anomaly factor of the pixel point is calculated according to the gradient amplitude of the temperature value of the pixel point, the temperature weighted variance in the local range of the pixel point, and the difference between the temperature of the pixel point and the minimum temperature in the power supply production thermal image; the target pixel point is obtained in response to the material anomaly factor and the spatial anomaly factor of the pixel point being greater than the corresponding threshold value; the target pixel point is processed based on the LOF algorithm, and the vehicle starting power supply production and assembly quality detection result is obtained, thereby effectively improving the efficiency of the vehicle starting power supply production and assembly quality detection result.

[0107] The embodiment of the application further discloses a vehicle starting power supply production and assembly quality detection system, comprising a processor and a memory, and the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a vehicle starting power supply production and assembly quality detection method provided by the application is realized.

[0108] The above system further comprises a communication bus and a communication interface and other components familiar to those skilled in the art, and the settings and functions thereof are known in the art, and thus will not be described here.

[0109] In the application, the aforementioned memory can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, device or apparatus.

[0110] The above are preferred embodiments of the application, and do not limit the protection scope of the application, therefore: any equivalent changes made on the structure, shape, principle of the application should be covered within the protection scope of the application.

Claims

1. A method for quality inspection during the production and assembly of vehicle-mounted starting power supplies, characterized in that, include: Obtain the temperature value and emissivity of each pixel in the thermal image of the power source production; The material anomaly factor of the pixel is calculated based on the gradient magnitude of the pixel's emissivity, the coefficient of variation in the pixel's eight neighborhoods, and the difference between the pixel's emissivity and the material reference emissivity. Construct a local range for a pixel with the pixel as the center. Calculate the weighting coefficients of other pixels in the local range for the pixel based on the absolute value of the temperature difference between the pixel and other pixels in the local range and the gradient direction angle. Obtain the temperature-weighted variance in the local range of the pixel. The spatial anomaly factor of a pixel is calculated based on the gradient magnitude of the pixel temperature value, the temperature-weighted variance in the local area of ​​the pixel, and the difference between the pixel temperature and the minimum temperature in the power generation thermal image. The target pixel is obtained when both the material anomaly factor and the spatial anomaly factor of the pixel are greater than the corresponding threshold. The target pixel is then processed using the LOF algorithm to obtain the production assembly quality inspection result of the vehicle starting power supply.

2. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 1, characterized in that, The acquisition of the temperature value and emissivity corresponding to each pixel in the thermal image of the power source includes: After applying a standard instantaneous high-current pulse to the power supply, the temperature dataset and emissivity dataset in the thermal image of the power supply production process are obtained. The temperature dataset and emissivity dataset in the thermal image of the power supply production process are preprocessed to obtain the temperature value and emissivity corresponding to each pixel in the thermal image of the power supply production process. The preprocessing method includes at least normalizing the temperature dataset and emissivity dataset.

3. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 1, characterized in that, The calculation of the material anomaly factor for this pixel includes: Obtain the material reference emissivity data of the pixel, perform normalization processing to obtain the material reference emissivity of the pixel; calculate the first... Material anomalous factor per pixel : ； For the first The gradient magnitude of the emissivity of each pixel. , , The first Standard deviation of emissivity, mean emissivity, and coefficient of variation in the eight neighborhood of each pixel. , The first Emissivity of each pixel, material reference emissivity Let e ​​be the base-e logarithmic function. It is the absolute value symbol. It is a linear normalization function.

4. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 1, characterized in that, The construction of the local range of a pixel centered on the pixel includes: The local area size of the preset pixel is Using a pixel as the center, obtain the data around that pixel. The local extent of a pixel is constructed using individual pixels.

5. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 1, characterized in that, The calculation of the weight coefficients of other pixels in the local area for that pixel includes: ； For the first In the local area of ​​the nth pixel The pixel point is the first The weight coefficients of each pixel For the first The pixel and the In the local area of ​​the nth pixel The cosine of the angle between the gradient directions of each pixel. For the first The pixel and the In the local area of ​​the nth pixel The absolute value of the temperature difference between each pixel. , These represent the maximum and minimum temperatures in the thermal image of the power source, respectively.

6. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 1, characterized in that, The calculation of the spatial anomaly factor of the pixel includes: ； For the first Spatial anomaly factor of each pixel, For the first Normalized gradient magnitude of temperature values ​​for each pixel For the first Temperature-weighted variance in a local area of ​​each pixel For the first Temperature value of each pixel , These represent the maximum and minimum temperatures in the thermal image of the power source, respectively. It is an exponential function with base e.

7. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 1, characterized in that, The process of processing target pixels based on the LOF algorithm to obtain the production assembly quality inspection results of the vehicle-mounted starting power supply includes: The comprehensive score of the target pixel is obtained based on the material anomaly factor and spatial anomaly factor of the target pixel; the comprehensive scores of all target pixels are sorted in descending order, and a preset number of target pixels are selected as candidate points from the comprehensive score descending order; the LOF value of the candidate points is calculated, and the anomaly threshold is calculated based on the mean and standard deviation of the LOF values ​​of all candidate points; based on the comparison result of the LOF value of the candidate points and the anomaly threshold, the abnormal temperature points in the production assembly quality inspection results of the vehicle starting power supply are obtained.

8. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 7, characterized in that, The calculation of the LOF value of the candidate point includes: The candidate point is located at a distance of K from its neighboring points in the neighborhood. The local density of the candidate point is obtained, and the average of the ratios of the local densities of the neighboring points to that of the candidate point is recorded as the LOF value of the candidate point.

9. The method for quality inspection of vehicle-mounted starting power supply production and assembly according to claim 7, characterized in that, After obtaining the production and assembly quality inspection results of the vehicle-mounted starting power supply, the process also includes: Multiple clusters are obtained by clustering the anomalous temperature points. The mean material anomaly factor and the mean spatial anomaly factor of the anomalous temperature points in the clusters are obtained. If the mean material anomaly factor of a cluster is greater than the mean spatial anomaly factor, the anomaly of the cluster is a material interface defect. If the mean material anomaly factor of a cluster is less than the mean spatial anomaly factor, the anomaly of the cluster is a thermal diffusion anomaly. If the mean material anomaly factor of a cluster is equal to the mean spatial anomaly factor, the anomaly of the cluster is both a material interface defect and a thermal diffusion anomaly.

10. A production assembly quality inspection system for vehicle-mounted starting power supplies, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a method for quality inspection of vehicle-mounted starting power supply production assembly according to any one of claims 1-9.

Images