CN108732507A - A kind of lithium battery defect detecting device based on battery temperature field and visible images - Google Patents

Abstract

The invention discloses a kind of lithium battery defect detecting device based on battery temperature field and visible images, it is by measuring lithium battery in different conditions (charging, electric discharge, processing transport), (side of the positive electrode regards different direction, negative side regards, shell is overlooked), the lithium battery appearance information of infrared thermal imagery and thermo parameters method and visible spectrum image under different time (persistently being observed on production line) and different complex backgrounds (there are more complex thermal noise environment for production line), use two waveband defects detection algorithm, while detecting defect, production process can be corrected in real time, influence of the production line environment to detection, the sensitivity of raising system, defective lithium battery is finally removed into production line.The present invention combines outer packaging defects detection and internal material defect inspection method, can be under the premise of ensureing defects detection integrality with accuracy, the confidence level of significant increase lithium battery defects detection, the degree of automation of efficiency and lithium battery production line.

Description

A lithium battery defect detection device based on battery temperature field and visible light image

technical field

The invention relates to the technical field of electric vehicle lithium battery production and detection, in particular to a lithium battery defect detection device based on the battery temperature field and visible light images.

Background technique

The traditional lithium battery assembly line is based on manual inspection. In order to improve production efficiency, it is necessary to replace the manual defect detection link. Existing lithium battery detection devices have a single method, whether it is extracting defect features from the appearance of the shape, or extracting defect features from the lithium battery material and working mechanism, there is a problem that it cannot cover all defect features, and the probability of missed detection is relatively high.

One of the difficulties of the detection device is that it is equally important to increase the detection probability and suppress the false detection probability, and the two are difficult to be effectively compatible. In order to strengthen the extraction of typical defect features, the traditional defect detection device will form a helpless state where the probability of missed detection decreases and the probability of false detection increases correspondingly, and the probability of false detection is suppressed and the probability of missed detection increases.

No matter what kind of detection device, there is a problem of environmental adaptability, that is, the detection sensitivity and environmental adaptability of the detection device cannot be improved together. Generally speaking, traditional inspection devices perform well in detecting certain defect features in certain environments, but perform poorly in detecting some defect features in other environments.

Contents of the invention

The purpose of the present invention is to provide a lithium battery defect detection device based on the battery temperature field and visible light images, so as to realize the automatic assembly line production of lithium battery defect detection.

The technical solution adopted in the present invention is: a lithium battery defect detection device based on the battery temperature field and visible light images, including a visual information sensor, a defect sensing computer, and a defect actuating device; the visual information sensor is used to obtain the lithium battery visible light band , short-wave infrared band images; the defect-aware computer extracts lithium battery defect information through lithium battery temperature computer vision technology; when a defective lithium battery is found, it sends a push-out signal to the defect actuating device;

The visual information sensor is composed of 3 to 4 industrial-grade high-definition pixel visible light cameras, 3 to 4 short-wave infrared cameras, and LED light sources. The installation positions of the two cameras can be crossed to form a unified field of view as much as possible. Each type Install two units in series on the top of the conveying surface, keep 20% overlap of the field of view, and install one on each side of the conveyor belt. According to the lighting environment of the production line, select the appropriate shape of the light source and installation position to make the lithium battery transport Areas can be evenly illuminated;

The defect-aware computer extracts the omnidirectional image of the lithium battery to be detected from the visible light image through computer vision technology; uses the appearance feature extraction module to extract the appearance defect information of the lithium battery; extracts the target and the surrounding environment temperature field data from the infrared image, and fuses these two types of data , based on the temperature-time response relationship of different behavioral stages of lithium battery, the relationship between lithium battery temperature and background environment, detection and discovery of possible defect parts, and fusion of defect targets to reduce the missed detection rate and improve the accuracy of defect detection. When a defective lithium battery is found in the detection, a push-out signal is sent to the defective actuator;

The defect actuating device is composed of a drive motor, a cam, and a push rod. When receiving the push signal, the drive motor starts to work, drives the cam to rotate, and pushes the push rod to push the defective lithium battery at the target position out of the conveyor belt. The push rod is reset and cut off. The drive motor works.

Among them, the combination of visible light band and short-wave infrared band dual-band image information is used to detect lithium battery defects;

The first step is image-based defect extraction. According to the principle, in actual measurement, the lithium battery is always in a certain environment. In addition to the heat radiation emitted by itself, the lithium battery also needs to exchange heat with the surrounding environment, that is, absorb and reflect the surrounding environment. Radiant heat energy emitted to the target,

W _{image radiation flux density} = W _{lithium battery radiation flux density} + W _{lithium battery reflection environment radiation flux density}

Making full use of the high-resolution advantages of visible light images, it can realize the appearance outline and production environment recognition of lithium batteries with high precision; thus establish a lithium battery appearance radiation characteristic model,

_{Radiation flux density of reflected environment of W lithium battery} = (1–ελ) (H _{Δ production workshop environment} + H _{Δ production line process environment} + H _{Δ various system backgrounds} )

Among them, ελ is the emissivity coefficient of lithium battery. By comparing the high-sensitivity temperature field information provided by infrared images, the target defects are detected and identified;

The second step is based on the defect extraction based on the thermal characteristics of the material, according to the temperature-time characteristic response curve at different stages (charging and discharging), and the thermal image profile and brightness difference of lithium batteries containing defects have different distributions compared with normal lithium batteries ;

Since the heat transfer rate of heat radiation is proportional to the 4th power of the temperature, and the lithium battery has a large change rate in the temperature response during the charging and discharging stages of the production process, so by improving the signal-to-noise ratio of the data, a better Defect detection results;

Different areas of lithium batteries have different infrared radiation characteristics due to their different materials or working mechanisms, as well as different appearance shapes; It is caused by the different materials of the positive and negative electrodes of the battery; there is a protruding part in the geometric shape of the top cover, but there is no bottom cover, which will also affect its thermal characteristics;

Using the high-precision positioning information of the high-resolution image, it is possible to accurately determine the abnormal temperature area on the lithium battery, detect and identify the target defect;

The third step integrates the defect information detected in the first two steps according to different principles to obtain a unified defect feature information description framework; expands the shortcomings of defect detection carried out by a single detection method and data theory, and describes it relatively more comprehensively and objectively Defect mechanism of lithium battery.

Among them, a unified defect feature description framework is formed on the basis of the defect detection method combined with material thermal characteristic detection and shape detection. On top of this, the cross-validation method based on historical detection data improves the self-adaptive ability of the defect detection algorithm;

In order to adapt to many inconsistent constraints such as different specific production environments, different appearances of lithium batteries, and thermal characteristics of different lithium battery materials, it is difficult to use uniform defect characteristics and static feature descriptions to adapt in all cases;

In order to solve the above-mentioned contradiction, the present invention proposes to conduct cross-validation and testing based on historical data, so as to relatively objectively judge the conformity of these defect detection results. The source of these historical data is to conduct manual supervision and review tests on the defect information generated during the detection process, and add the detection results and image data to the defect detection database system to enrich the test data and provide data guarantee for the subsequent verification of new defect information.

Advantage of the present invention and positive effect are:

(1) The present invention uses computer vision technology. Vision is the most direct and least condition-dependent means to obtain target feature information. Image sensors and computers are used to replace human eyes to identify, track and measure targets. Machine vision based on lithium batteries Defect feature description framework, further corresponding graphic data processing, improve the signal-to-noise ratio of the image, and process it into an image that is more suitable for defect feature detection;

(2) The optical flow method image target capture technology of the present invention, scale-invariant feature transformation matching and video splicing algorithm, can quickly identify the lithium battery from the background environment of the complex production line, and splice the images of multiple detection intervals into a panoramic image, In order to obtain dynamic, real-time and more informative lithium battery panoramic information;

(3) The present invention utilizes the image edge detection module and the image feature extraction module, which can greatly improve the efficiency of lithium battery defect detection and the automation of lithium battery production lines under the premise of ensuring the correct rate.

Description of drawings

1 is a schematic diagram of a lithium battery defect detection device based on the battery temperature field and visible light images of the present invention;

Fig. 2 is a structural block diagram of a lithium battery defect detection device based on the battery temperature field and visible light images of the present invention;

Fig. 3 is a schematic diagram of the application of a lithium battery defect detection device based on the battery temperature field and visible light images of the present invention;

Figure 4 is a schematic diagram of the installation position of the lithium battery defect detection device;

Figure 5 is a schematic diagram of the composition of the lithium battery detection interval;

Fig. 6 is a working principle diagram of the defect-aware computing subsystem;

Figure 7 is a time-temperature characteristic curve of a certain lithium battery, where Figure 7(a) is the time-temperature characteristic curve of a lithium battery in different behavior modes, and Figure 7(b) is the time-temperature characteristic curve of a lithium battery in different time stages temperature characteristic curve;

Figure 8 is the framework of lithium battery defect feature extraction algorithm;

Figure 9 shows the appearance defect detection process of the lithium battery bottom cover in the production scenario, where Figure 9(a) is the original image, Figure 9(b) is noise suppression and feature enhancement, and Figure 9(c) is feature extraction.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

As shown in Figures 1-2, the present invention is a lithium battery defect detection device based on the battery temperature field and visible light images, including a visual information sensor, a defect sensing computer, and a defect actuating device.

The visual information sensor consists of 3 to 4 industrial-grade high-definition pixel cameras and LED light sources. Among them, two sets are installed in series on the top of the conveying surface to keep 20% of the field of view overlapping, and one set is installed on each side of the conveyor belt. According to the lighting environment of the production line, select the appropriate shape of the light source and the installation position of the LED light source, so that the lithium battery delivery area can be evenly illuminated.

The defect-aware computer is an industrial control computer; the omnidirectional image of the lithium battery to be detected is extracted from the visual information through computer vision technology; the defect information of the lithium battery is extracted by using the feature extraction algorithm; Signal the exit.

The fault actuating device is composed of a driving motor, a cam and a push rod. When the push-out signal is received, the drive motor starts to work, drives the cam to rotate, pushes the push rod to push the defective lithium battery at the target position out of the conveyor belt, and cuts off the drive motor after the push rod resets.

As shown in Figure 2, the lithium battery defect detection device transforms the traditional lithium battery assembly line and replaces the manual defect detection link. Improve the current continuous production line of lithium batteries, add a defect detection link at the end of the existing production line, and realize real-time defect detection of lithium battery production. Real-time image acquisition based on computer vision technology, image denoising and automatic white balance, target recognition, extraction and enhancement, target omnidirectional image stitching, defect detection, and action control of defective lithium batteries. Applicable lithium battery defects include scratches, skin damage, bulging, extrusion deformation, etc.

As shown in FIG. 3 , the defect detection device supports two installation modes of serial and parallel. The defect detection link can dynamically increase or decrease the detection interval to match the production speed of the lithium battery production line. According to the installation and deployment needs of the pipeline, each detection section supports parallel and serial arrangement.

As shown in Figure 4, the system composition of the lithium battery defect detection section includes four subsystems. Each detection section is composed of four parts: rolling conveying subsystem, visual information sensor subsystem, defect perception computing subsystem, and defect actuator subsystem.

The rolling conveying subsystem is composed of 2-5 power rollers selected according to the size of the lithium battery; it drives the lithium battery forward and keeps rotating.

As shown in Figure 5, the working principle of the lithium battery defect sensing subsystem. The visual information sensor subsystem consists of 3 to 4 industrial-grade high-definition pixel cameras and LED light sources. Among them, two sets are installed in series on the top of the conveying surface to keep 20% of the field of view overlapping, and one set is installed on each side of the conveyor belt. According to the lighting environment of the production line, select the appropriate shape of the light source and the installation position of the LED light source, so that the lithium battery delivery area can be evenly illuminated.

The defect-aware computing subsystem consists of an industrial control computer; the omnidirectional image of the lithium battery to be detected is extracted from the visual information through computer vision technology; the defect information of the lithium battery is extracted by using the feature extraction algorithm; The actuator sends an eject signal.

Defect actuator subsystem is composed of drive motor, cam and push rod. When the push-out signal is received, the drive motor starts to work, drives the cam to rotate, pushes the push rod to push the defective lithium battery at the target position out of the conveyor belt, and cuts off the drive motor after the push rod resets.

As shown in Figure 9, the detection process of appearance defect feature extraction: Figure 9(a) acquires the original real-time image of the lithium battery production line through the image sensor; Figure 9(b) suppresses noise and enhances features, and removes the original image according to the light environment Noise, white balance and other operations, perform projection transformation according to the relative relationship between the lens and the target to eliminate image distortion, and strengthen the corners, edges, spots and other features in the image according to the detection of defect features; Figure 9 (c) through defect feature extraction Algorithms extract features.

Claims (3)

1. A lithium battery defect detection device based on battery temperature field and visible light image, characterized in that: it includes a visual information sensor, a defect perception computer, and a defect actuating device; the visual information sensor is used to obtain lithium battery visible light band, short-wave Images in the infrared band; the defect-aware computer extracts lithium battery defect information through lithium battery temperature computer vision technology; when a defective lithium battery is found, it sends a push-out signal to the defect actuating device; The visual information sensor is composed of 3 to 4 industrial-grade high-definition pixel visible light cameras, 3 to 4 short-wave infrared cameras, and LED light sources. The installation positions of the two cameras can be crossed to form a unified field of view as much as possible. Each type Install two units in series on the top of the conveying surface, keep 20% overlap of the field of view, and install one on each side of the conveyor belt. According to the lighting environment of the production line, select the appropriate shape of the light source and installation position to make the lithium battery transport Areas can be evenly illuminated; The defect-aware computer extracts the omnidirectional image of the lithium battery to be detected from the visible light image through computer vision technology; uses the appearance feature extraction module to extract the appearance defect information of the lithium battery; extracts the target and the surrounding environment temperature field data from the infrared image, and fuses these two types of data , based on the temperature-time response relationship of different behavioral stages of lithium battery, the relationship between lithium battery temperature and background environment, detection and discovery of possible defect parts, and fusion of defect targets to reduce the missed detection rate and improve the accuracy of defect detection. When a defective lithium battery is found in the detection, a push-out signal is sent to the defective actuating device; The defect actuating device is composed of a drive motor, a cam, and a push rod. When receiving the push signal, the drive motor starts to work, drives the cam to rotate, and pushes the push rod to push the defective lithium battery at the target position out of the conveyor belt. The push rod is reset and cut off. The drive motor works. 2. A lithium battery defect detection device based on battery temperature field and visible light image according to claim 1, characterized in that: lithium battery defect detection is performed by combining dual-band image information of visible light band and shortwave infrared band; The first step is image-based defect extraction. According to the principle, in actual measurement, the lithium battery is always in a certain environment. In addition to the heat radiation emitted by itself, the lithium battery also needs to exchange heat with the surrounding environment, that is, absorb and reflect the surrounding environment. Radiant heat energy emitted to the target, W _{image radiation flux density} = W _{lithium battery radiation flux density} + W _{lithium battery reflection environment radiation flux density} Making full use of the high-resolution advantages of visible light images, it can realize the appearance outline and production environment recognition of lithium batteries with high precision; thus establish a lithium battery appearance radiation characteristic model, _{Radiation flux density of reflected environment of W lithium battery} = (1–ελ) (H _{Δ production workshop environment} + H _{Δ production line process environment} + H _{Δ various system backgrounds} ) Among them, ελ is the emissivity coefficient of lithium battery. By comparing the high-sensitivity temperature field information provided by infrared images, the target defects are detected and identified; The second step is based on the defect extraction based on the thermal characteristics of the material, according to the temperature-time characteristic response curve at different stages (charging and discharging), and the thermal image profile and brightness difference of lithium batteries containing defects have different distributions compared with normal lithium batteries ; Since the heat transfer rate of heat radiation is proportional to the 4th power of the temperature, and the lithium battery has a large change rate in the temperature response during the charging and discharging stages of the production process, so by improving the signal-to-noise ratio of the data, a better Defect detection results; Different areas of lithium batteries have different infrared radiation characteristics due to their different materials or working mechanisms, as well as different appearance shapes; It is caused by the different materials of the positive and negative electrodes of the battery; there is a protruding part in the geometric shape of the top cover, but there is no bottom cover, which will also affect its thermal characteristics; Using the high-precision positioning information of the high-resolution image, it is possible to accurately determine the abnormal temperature area on the lithium battery, detect and identify the target defect; The third step integrates the defect information detected in the first two steps according to different principles to obtain a unified defect feature information description framework; expands the shortcomings of defect detection carried out by a single detection method and data theory, and describes it relatively more comprehensively and objectively Defect mechanism of lithium battery. 3. A lithium battery defect detection device based on battery temperature field and visible light image according to claim 1, characterized in that: a unified defect feature description is formed on the basis of defect detection combined with material thermal characteristic detection and shape detection The framework, on top of which, based on the cross-validation method of historical inspection data, improves the adaptive ability of the defect detection algorithm; In order to adapt to many inconsistent constraints such as different specific production environments, different appearances of lithium batteries, and thermal characteristics of different lithium battery materials, it is difficult to use uniform defect characteristics and static feature descriptions to adapt all situations.