How Machine Vision Can Help In EV Battery Inspection?

Machine Vision (MV) or Computer Vision is a compound solution of utilizing algorithm, machine learning and customized cameras in order to receive and analyze images. It is a technology widely used in autopilot, product inspection, security and medical appliances.

By combining machine learning and highly specialized cameras, MV is vastly superior than traditional methods in both precision and efficiency. It is one of the fastest developing technologies thanks to Artificial Intelligence boom.

What can machine vision do for EV battery inspection?

One of the most prominent machine vision adoptions in manufacturing is EV battery inspection, due to the massively growing popularity of electric vehicles.
So, in what ways can MV help in EV battery inspection?

Here are some application examples:

Seal screw defect detection

Many defects such as spots, frying or missing may occur during welding process; these can be hard to detect not only because they are often microscopic but also come in various areas or shapes.

Machine vision powered by deep learning algorithms, however, can easily detect such defects.

Square case battery laser welding defect detection

Square cell casing inspection often includes checks for explosion points, welding shorts, missing welds, broken welds and more.

Usually camera cannot take a full short due to the length of the product and thus requiring a complex imaging solution divided into three different angles.

Lithium coating inspection

Machine vision can also be utilized in checking for lithium coating deficits.

Density test

With algorithms that have a clear definition of area lithium density, machine vision can detect areas with irregular lithium density as quickly as 30ms.

Insulation film defect detection

Detection of insulation film dents or breaks, while omitting natural occurrences such as light reflections or dusts.

Cylindrical cell sealing defect detection

Due to welding process, quality defects such as deviation, penetration, breakage and more are a common occurrence when sealing lithium batteries and can be hard to detect via traditional methods; but they can be accurately and efficiently detected thanks to machine vision with customized algorithm.

Scanning EV batteries is costly, in terms of space

As mentioned before, machine vision solutions for EV battery inspection usually require customized cameras due to the sheer size/length or shape of the product. It typically needs two or three different imaging devices to inspect at unique angles simutanously. A compound MV solution such as this may end up taking a lot of space at the end of the production line.

However, a unique design of line scan cameras may solve this issue – the LTCM/MFA series; with the narrow head design, this lens allows cameras to reach narrow and complex spacing inbetween production/inspection lines and thus making sure the total inspection solution to be more compact and space-saving.

Read more: LTCM Series

Read more: MFA Series

What is a line scan camera?

Line scan camera captures images by a single thin line rather than an area by traditional cameras. It has a linear sensor array which receives a data stream when scanning. The data would then be used to create a high-resolution image.

EV battery image by a line scan camera

These “line scanning” work in rapid session and creates an advantage in manufacturing, especially the product inspection process for its unparalleled precision and resolution, making sure that no dust, dents or other defects would pass over.

Advantages of Line Scan Camera Lenses

Line scan cameras offer significant benefits over traditional models. They excel at capturing rapid movements with high-speed and high-resolution imaging capabilities. Moreover, they perform exceptionally well in low-light settings, delivering top-notch image quality that traditional cameras struggle to match. Additionally, these cameras excel at capturing detailed and crisp images of expansive surfaces, making them an excellent choice for product inspection and analysis.

Types of Line Scan Camera Lens

Line scan camera lenses come in two main categories: telecentric and non-telecentric. Telecentric lenses produce distortion-free images, making them perfect for meticulous product inspection tasks. On the other hand, non-telecentric lenses are adept at capturing images of irregular surfaces, offering versatility in imaging freeform objects.

How to choose a compatible line scan lens?

The usual process of selecting a compatible line scan machine vision solution starts with getting the camera and its capture card, then move on to lens and light source. Picking a line scan system is similar.
First, we need to determine the resolution and line scanning speed of the CCD camera according to the requirements of detection accuracy and speed, and in the meantime settle on the corresponding capture card; then pick the lens based on the interface of the line array camera and the field of view requirements of the inspection workpiece. Lastly, we choose the right light source based on the inspection environment and the workpiece size/shape.

Picking the right line scan industrial camera

Step one: Get the right resolution – the width of the inspection object (perpendicular to the direction of scan movement) divided by the minimum detection accuracy equals the pixel value required for each line.
Step two: Calculate the right camera accuracy. Pick a camera based on the resolution you need calculated from step one. Then divide the width of the inspection object by the number of pixels needed to get the actual detection accuracy you need.
For step three: You need to determine the number of scan lines needed per second, which is what we call the line scan rate. You can calculate that by dividing your inspection object’s motion length per second by accuracy per second.

Now you are ready to pick the right line scan camera based on the three essential data in your hands. For instance, if the inspection object has a width of 160mm with an accuracy requirement of 0.1mm and a motion speed of 2200mm/s, and your camera’s resolution would be 160/0.1=1600, so the 2k camera would do.
The practical accuracy would end up as 160mm/2048=0.08mm
The actual line scan rate would be 2200mm/0.08mm=27.5KHz
So your typical line scan camera for this job would be something of a 29kHz 2k camera.

There are more things to consider when picking your line scan camera

Resolution

The common resolutions for line scan camera would be 0.5k, 1k, 2k, 4k, 6k, 8k, 12k, 16k; their pixel size would be something like 5μm, 7μm, 10μm, 14μm and so on, and thus making the sensor size ranging from 168mm (512×14μm) to 81.920mm (16kx5μm).

Focal length

The focal length of a lens, measured in millimeters, determines your shot’s width or closeness. To capture subjects effectively, calculate the focal length by considering distances from the subject to the lens and from the lens to the camera sensor. Understanding if your lens is concave or convex is crucial in determining the focal length’s impact on your image.

Aperture

The aperture determines the amount of light that enters the camera, affecting the exposure of the image.

β,Magnification

Once you have camera resolution and pixel size, you can calculate your sensor size; and once you have those two, you can get the magnification you need by dividing your sensor size with FOV. β=CCD/FOV

Mount

Mounts play a crucial role in camera setups. On the market, we primarily encounter C, CS, F, M42x1, T2, and M72x0.75 mounts. It’s important to note that the C mount, for instance, is limited to supporting sensors up to 22mm, approximately 1.3 inches. This limitation often leads to the preference for F mounts in many cameras. Each mount comes with its own unique flange distances, influencing the working distances in distinct ways.

Flange distance

Flange distance is the length from the flange to the imaging sensor. It is an extremely important number that is usually determined by the optical designs of camera manufacturers. Different camera manufacturers may have different flange distances even if they are of the same camera mounts.

Conclusion

Line scan cameras can reliably, accurately, and efficiently assist in EV battery inspection; however, thorough preparation is necessary. This includes determining the right specifics, the quantity and types of cameras needed, and selecting appropriate light sources—all crucial for achieving optimal results.

Getting these specifics right is not an easy task. Moreover, it usually requires a properly equipped lab to conduct sufficient experiments before you can gain a clear understanding.

VICO Imaging® offers a full team of experts and optical labs dedicated to creating customized lenses. Since 2008, we have catered to over 5600 clients with 8500 unique design lenses. Contact us to discover how we can assist you in developing your industrial lenses specifically.