A fully automatic feedback piston cleaning device based on sensors and machine vision

Abstract

The present invention discloses a fully automatic feedback piston cleaning device based on sensors and machine vision. The device comprises a transmission system, a feeding device, an intelligent cleaning system, a cleaning result judgment system, a drying and air-washing system, and a storage system. The transmission system comprises a tray, on which the feeding device places the piston, which then rises to the intelligent cleaning system for cleaning. The cleaning result judgment system captures the piston image. If there is stain on the piston surface, the piston is re-cleaned; otherwise, the piston is sent to the drying and air-washing system for low-pressure air washing and air drying. The storage system comprises a robotic arm equipped with an image sensor. If it detects that there is no uncleaned stain on the piston surface, the robotic arm sends the piston to an external storage bin; otherwise, the robotic arm sends the piston to an external cleaning bin. The present invention facilitates the cleaning of large-scale contaminated pistons and is a fully intelligent, automated, efficient, and low-cost working device.

Description

Technical Field

[0001] The present invention belongs to the field of machine vision, and in particular relates to a full-automatic feedback piston cleaning device based on sensors and machine vision. Background Art

[0002] During the process of molding, transportation, and use, piston surfaces can become contaminated with oil, floating particles, and adhesives. This can significantly reduce piston efficiency and increase surface fatigue, shortening service life. This can also reduce printing accuracy and increase printing difficulty when printing on pistons. Furthermore, given the high demand for pistons in daily use, efficient piston cleaning is a challenge.

[0003] At present, the main cleaning methods include manual scrubbing, boiling cleaning, spraying cleaning, vibration cleaning and ultrasonic cleaning. Among them, manual scrubbing has the best effect, but the disadvantage is that it consumes a lot of human resources and has low cleaning efficiency; boiling cleaning and spraying cleaning are prone to the problem of incomplete cleaning; vibration cleaning and ultrasonic cleaning have good effects, but the cleaning time is long and targeted cleaning cannot be carried out, resulting in waste of resources. Summary of the Invention

[0004] To address the challenges of existing technologies, this invention proposes a fully automatic piston cleaning system based on sensors and machine vision. The main purpose of this invention is to combine data acquired by sensors with deep learning algorithms to address the shortcomings of traditional piston cleaning systems while also enabling highly efficient and intelligent cleaning through new technologies. This device can automatically identify different types of dirt, combine repeated cleaning with a feedback system, and intelligently provide cleaning methods.

[0005] The technical solution adopted in the present invention is as follows:

[0006] In a first aspect, the present invention discloses a fully automatic feedback piston cleaning device based on sensors and machine vision, comprising a transmission system, a feeding device, an intelligent cleaning system, a cleaning result judgment system, a drying and air washing system, and a storage system;

[0007] The transmission system includes a terminal processor, a servo motor, a lifting controller and a tray, the terminal processor is used to control the servo motor and the lifting controller, the servo motor is used to control the rotation of the tray, and the lifting controller is used to control the lifting of the tray;

[0008] The feeding device includes a piston clamp, which is used to clamp the piston. The piston clamp holding the piston will fall on the tray, and the lifting controller will control the tray with the piston to rise to the intelligent cleaning system for cleaning. The intelligent cleaning system is used to clean the piston;

[0009] The cleaning result judgment system includes an image sensor, which is used to capture an image of the piston surface after cleaning and input the captured image into a terminal processor. The terminal processor processes the captured image to determine whether the piston surface contains any uncleaned stains. If so, the tray is controlled to descend and the piston is cleaned again; otherwise, the piston is sent to a drying and air washing system.

[0010] The drying and air washing system is used to perform low-pressure air washing and air drying on the piston. After low-pressure air washing and air drying, the piston is sent to the storage system. The storage system includes a robotic arm with an image sensor. The robotic arm includes detection function and positioning and placement function. When it is detected that there is no uncleaned stain on the surface of the piston, the robotic arm will send the piston to an external storage bin; otherwise, the robotic arm will send the piston to an external cleaning bin.

[0011] Furthermore, the terminal processor processes the collected image, including:

[0012] First, the collected image is enhanced, and then the enhanced image is denoised to remove noise in the image and ensure image clarity; the threshold is adjusted using threshold segmentation technology to separate the stains and background in the denoised image; then the edge detection algorithm is used to identify the edge of the stain; finally, opening and closing operations are applied to remove water stains and highlight the stains.

[0013] Furthermore, the image sensor on the robotic arm captures an image of the piston surface and inputs the captured image to the terminal processor, which performs stain detection and damage detection on the captured image;

[0014] The stain detection process includes: performing image enhancement processing on the collected image by the terminal processor, and then performing denoising processing on the enhanced image to remove noise in the image and ensure image clarity; adjusting the threshold using threshold segmentation technology to separate the stain and background in the denoised image; then using an edge detection algorithm to identify the stain edge; and finally applying opening and closing operations to remove water stains and highlight the stain.

[0015] The damage detection comprises: performing grayscale conversion on the collected image by the terminal processor to obtain a grayscale image, performing denoising on the grayscale image, and then using histogram equalization to enhance the contrast of the denoised image to obtain a first image, extracting edge information in the first image using edge detection technology, and then using morphological image processing morphological operation technology to enhance features in the image and remove small noise points;

[0016] The denoised color image is subjected to color segmentation, and then the color segmented image is divided into damaged areas and non-damaged areas using threshold segmentation technology. At the same time, the edge information is used to extract the contour, detect the shape and size of the damaged area, and then the area, perimeter and roundness of the damaged area contour are analyzed to determine the type of damage and perform damage classification.

[0017] In a second aspect, the present invention discloses a piston cleaning method of the piston cleaning device, comprising the following steps:

[0018] The terminal processor sends an ascending signal to the lifting controller, which controls the pallet to ascend at a constant speed. At the same time, the lifting controller sends a transmitting signal to the conveying mechanism, which starts to convey the piston clamp. When the conveying mechanism conveys the piston clamp to the position of the pallet, the conveying mechanism releases the piston clamp, and the piston clamp falls onto the pallet.

[0019] At the same time, the tray with the piston continues to rise at a constant speed to the intelligent cleaning system. When the piston contacts the brush, the pressure sensor outputs a cleaning signal to the terminal processor, which then outputs a rotation signal to the servo motor. The servo motor controls the tray to start rotating, allowing the brush to clean the piston, and the tray with the piston continues to rise at a constant speed.

[0020] When the tray with the piston rises to the cleaning result judgment system, the image sensor captures an image of the cleaned piston surface and inputs the captured image to the terminal processor. The terminal processor processes the captured image to determine whether there are any uncleaned stains on the piston surface. If so, the tray is controlled to descend and the piston is cleaned again. Otherwise, the tray with the piston continues to rise at a constant speed, and the piston is sent to the drying and air washing system.

[0021] When the piston enters the drying and air washing system, the terminal processor drives the low-pressure nozzle and the fan to work, and performs low-pressure air washing and air drying on the piston. At the same time, the tray with the piston continues to rise at a uniform speed. The piston after low-pressure air washing and air drying is sent to the storage system. The image sensor on the robotic arm will capture the image of the piston surface and input the captured image to the terminal processor. The terminal processor performs stain detection on the captured image. When it is detected that there is no uncleaned stain on the piston surface, the robotic arm will send the piston to an external storage bin; otherwise, the robotic arm will send the piston to an external cleaning bin, and the terminal processor will perform damage detection on the image of the piston transported to the external cleaning bin. If the piston is damaged, it will no longer be cleaned; if the piston is not damaged, it will continue to be cleaned.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] (1) The cleaning process is fully automatic, which can greatly reduce human resources and reduce the accident rate;

[0024] (2) The present invention incorporates automatic feedback technology and machine vision detection technology, which can ensure that a qualified and clean piston is cleaned repeatedly, achieving high-quality cleaning results. It also has a function for detecting piston surface damage to avoid accidents caused by damage during later use.

[0025] (3) The present invention has multiple designs that achieve resource conservation and energy reuse, ensuring high efficiency while achieving low consumption.

[0026] (4) The present invention takes into account the requirements of later processing, transportation, assembly, cleaning, maintenance, and updating and upgrading, and designs a mosaic assembly structure to enhance its convenience. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 Schematic diagram of the fully automatic feedback piston cleaning device based on sensor and machine vision technology of the present invention;

[0028] Figure 2 This is a workflow diagram of a fully automatic feedback piston cleaning device based on sensor and machine vision technology;

[0029] Figure 3 Workflow diagram of pressure sensor;

[0030] Figure 4 Fully automatic feedback flow chart of the cleaning result judgment system. DETAILED DESCRIPTION

[0031] The present invention will be further described and illustrated below in conjunction with specific embodiments. The embodiments are merely illustrative of the present disclosure and do not limit its scope. The technical features of the various embodiments of the present invention may be combined accordingly, provided that there is no conflict between them.

[0032] The present invention uses a pressure sensor and an image sensor, which can perform cleaning work at corresponding times and methods for different degrees of stains on the piston surface through deep algorithm learning technology, and can perform repeated cleaning by detecting and feeding back to the terminal until the cleanest piston is cleaned in the shortest time. The pressure sensor can play two roles here. One is to send a signal to rotate the piston after sensing the pressure, and the other is to record the different dirt on the surface of each piston and calculate the corresponding cleaning mode through a deep learning algorithm. After the image sensor collects data, it uses image processing technology to judge the cleanliness of the piston and whether there is any damage. The present invention makes full use of the function of the sensor and then uses a fully automatic feedback system to clean the piston, which can achieve high-quality cleaning, high-efficiency cleaning, low power loss and save human resources.

[0033] like Figure 1 As shown, it is a schematic diagram of the fully automatic feedback piston cleaning device based on sensors and machine vision in this example, as shown Figure 2 As shown, the workflow of the device of the present invention, the specific operating steps are as follows:

[0034] First, the present invention's fully automatic feedback system for cleaning pistons, based on sensors and machine vision, includes a transmission system, a feeder, an intelligent cleaning system, a cleaning result determination system, a drying and air-washing system, and a storage system. The device also includes a sewage discharge device for collecting wastewater generated by the intelligent cleaning system. The device is easy to manufacture, transport, assemble, clean, and maintain, and is portable and simple to use for future equipment upgrades. The specific functions and operating methods of the piston cleaning transducer will be further explained below.

[0035] like Figure 1 As shown in the device, the bottom layer is the transmission system, which is the key output layer for piston movement. The transmission system includes a terminal processor, a servo motor, a lift controller, and a pallet. The terminal processor controls the servo motor and lift controller. The servo motor controls the rotation of the pallet, and the lift controller controls the pallet's elevation.

[0036] The pallet is provided with a piston fixture securing device for securing the piston fixture to the pallet. In one embodiment of the present invention, the pallet is provided with two opposing springs that interact to securely position the piston fixture on the pallet. The pallet is designed to be retractable to accommodate piston fixtures of varying sizes.

[0037] like Figure 1 As shown in the device, the second layer from the bottom to the top is the feeding device, which includes a piston clamp and a conveying mechanism for conveying the piston clamp. The piston clamp is used to clamp the piston, and the piston clamp clamping the piston will be placed on the pallet; the conveying mechanism clamps the piston clamp and conveys it. When the conveying mechanism conveys the piston clamp to the position of the pallet, the conveying mechanism releases the piston clamp and the piston clamp falls onto the pallet.

[0038] In a specific embodiment of the present invention, the piston clamp is clamped on the conveying mechanism, and a position sensor is also provided on the piston clamp. When the position sensor detects that the position of the pallet is in a suitable position, the position sensor sends a placement signal to the conveying mechanism, the conveying mechanism releases the piston clamp, and the piston clamp is placed on the pallet.

[0039] When the piston cleaning starts, the terminal processor sends an upward signal to the lifting controller, and the lifting controller controls the tray to rise at a uniform speed. At the same time, the lifting controller sends a transmission signal to the conveying mechanism, and the conveying mechanism starts to convey the piston clamp. When the conveying mechanism conveys the piston clamp to the position of the tray, the conveying mechanism releases the piston clamp, and the piston clamp falls onto the tray; at the same time, the tray with the piston continues to rise at a uniform speed to the intelligent cleaning system.

[0040] like Figure 1 As shown in the device in FIG, the third layer from the bottom to the top is an intelligent cleaning system, which is used to clean the piston; the intelligent cleaning system includes a pressure sensor, a brush fixed to the pressure sensor, multiple nozzles capable of spraying detergent, and multiple nozzles capable of spraying clean water, wherein one nozzle sprays one detergent, and different nozzles spray different detergents;

[0041] like Figure 3 As shown, when the piston contacts the brush, the pressure sensor outputs a scrubbing signal to the terminal processor, and the terminal processor outputs a rotation signal to the servo motor. The servo motor controls the tray to start rotating and can control the rotation speed of the tray to reduce energy consumption.

[0042] When the brush contacts the rotating piston, the pressure sensor generates an electronic signal indicating the stain on the piston surface and inputs the electronic signal to the terminal processor. The terminal processor compares the received electronic signal with the preset electronic signal and determines the type of stain on the current piston surface through deep learning algorithm. The terminal processor controls the nozzle capable of spraying detergent to clean the corresponding stain to spray detergent to the piston. At the same time, the terminal processor controls the speed of rotation of the tray (the terminal processor determines the type of stain and proposes a cleaning plan, which includes issuing instructions to the servo motor to control the speed of rotation of the servo motor). When the detergent is washed, the terminal processor controls the nozzle capable of spraying clean water to spray clean water to the piston, completing the cleaning of the piston.

[0043] And when the piston is no longer in contact with the brush, the pressure sensor will output a first signal to the terminal processor, and the terminal processor will output a rotation stop signal to the servo motor, and the servo motor will control the tray to stop rotating.

[0044] The piston completes cleaning in the intelligent cleaning system, and the tray with the piston continues to rise at a constant speed to the cleaning result judgment system.

[0045] like Figure 1 As shown in the device, the fourth layer from the bottom to the top is a cleaning result judgment system, which includes an image sensor, such as Figure 4As shown, the image sensor is used to capture images of the piston surface after cleaning and input the captured images into the terminal processor. The terminal processor processes the captured images to determine whether the piston surface contains uncleaned stains. If so, the tray is controlled to descend and the piston is cleaned again; otherwise, the piston will be sent to the drying and air washing system.

[0046] In a specific embodiment of the present invention, image data is collected through a high-resolution CMOS sensor to ensure that tiny dirt can be identified, avoiding the problem of poor cleaning effect due to some small dirt not being detected. The terminal processor uses image enhancement technology to perform basic processing on the collected image, including adjusting brightness, contrast, sharpness, etc. Filters are then applied to perform denoising to remove noise in the image and ensure clarity. At the same time, stain detection is performed using image processing algorithms. Among them, the key technologies involved are: using threshold segmentation technology to adjust the threshold to separate stains and background. Using Sobel and Canny algorithms for edge detection to identify the edges of stains. Finally, using morphological image processing technology, opening and closing operations are applied to remove water stains and highlight stains.

[0047] In a specific embodiment of the present invention, if there is uncleaned dirt on the surface of the piston, the terminal processor sends a re-cleaning signal to the lifting controller. The lifting controller first controls the tray to descend to a height where the piston and the brush can make initial contact, and then controls the tray to rise at a constant speed. The intelligent cleaning system cleans the piston again.

[0048] When the tray with the piston rises to the cleaning result judgment system again, the cleaning result judgment system judges whether there is any uncleaned stain on the surface of the piston. If so, the piston is cleaned again.

[0049] If the same piston still has stains after being cleaned three times, the piston will no longer be cleaned, but will be sent to the drying and air washing system for low-pressure air washing and air drying.

[0050] like Figure 1 As shown in the device, the fifth layer from the bottom up is the drying air washing system, which is used to perform low-pressure air washing and air drying on the piston. The drying air washing system includes multiple fans for air drying the piston and multiple low-pressure nozzles and compressors for low-pressure air washing on the piston. The low-pressure nozzles can perform low-pressure air washing over a large area, and together with the fans, they can process water vapor, small particles and dust on the surface of the piston to complete the final cleaning of the piston.

[0051] In a specific embodiment of the present invention, the low-pressure nozzle of the present invention is duckbill-shaped, so that low-pressure air washing can be carried out over a large area, wherein the ejected gas is the heat produced by the compressor motor itself (to fully utilize energy), and the main purpose of the low-pressure nozzle ejecting the low-pressure hot air flow is to dry the moisture on the piston surface.

[0052] When the piston enters the drying and air washing system, the terminal processor drives the low-pressure nozzle and fan to work, performing low-pressure air washing and air drying on the piston. At the same time, the tray with the piston continues to rise at a uniform speed, and the piston after low-pressure air washing and air drying is sent to the storage system.

[0053] like Figure 1 As shown in the device, the sixth layer from the bottom to the top is the storage system, which includes a robotic arm with an image sensor. The robotic arm includes detection function and positioning and placement function. When it is detected that there is no uncleaned stain on the surface of the piston, the robotic arm will send the piston to an external storage bin; otherwise, the robotic arm will send the piston to an external cleaning bin.

[0054] In a specific embodiment of the present invention, the image sensor on the robotic arm captures an image of the piston surface and inputs the captured image to a terminal processor, which performs stain detection and damage detection on the captured image;

[0055] The stain detection is implemented in the same manner as the stain detection performed by the image sensor in the cleaning result judgment layer.

[0056] The damage detection is:

[0057] First, image preprocessing involves grayscale conversion of color images. Denoising is then performed, using filters (such as Gaussian and median filters) to reduce image noise. Finally, histogram equalization is used to enhance image contrast. Feature extraction is then performed, using edge detection techniques based on the application scenario, using algorithms such as Sobel and Canny, to extract edge information from the image. Morphological operations are then performed, including dilation, erosion, and opening and closing, to enhance image features and remove small noise points.

[0058] Finally, the denoised color image is subjected to color segmentation. Within the color image, color segmentation techniques are used to detect damage based on its color characteristics. Next, threshold segmentation techniques are used to set an appropriate threshold to separate the image into damaged and non-damaged areas. Global or adaptive thresholding can be used. Contour detection is also used to extract contours from the image using edge information and determine the shape and size of the damaged areas. To ensure data integrity, shape analysis can be used to determine the damage type by analyzing contour features such as area, perimeter, and roundness. Damage classification is then performed by fusing the aforementioned data, performing a simple classification based on characteristics such as size, shape, and location. Machine learning or deep learning is then used to classify damage using a trained model. For example, convolutional neural networks (CNNs) can be used for complex damage detection tasks. Finally, data is output and a report is generated to the terminal for operator review and statistics on damaged pistons. Damage markers are used to mark damaged areas on the image for easy visual analysis. The data is statistically analyzed to count the number, size, location and other data of the detected damage to provide a basis for quality control and improvement. Finally, a report is generated and the test results are output in the form of a report to record the test process, results and analysis, thereby realizing damage detection and damage marking functions.

[0059] This embodiment also provides a piston cleaning method of the device, comprising the following steps:

[0060] The terminal processor sends an ascending signal to the lifting controller, which controls the pallet to ascend at a constant speed. At the same time, the lifting controller sends a transmitting signal to the conveying mechanism, which starts to convey the piston clamp. When the conveying mechanism conveys the piston clamp to the position of the pallet, the conveying mechanism releases the piston clamp, and the piston clamp falls onto the pallet.

[0061] At the same time, the tray with the piston continues to rise at a constant speed to the intelligent cleaning system. When the piston contacts the brush, the pressure sensor outputs a cleaning signal to the terminal processor, which then outputs a rotation signal to the servo motor. The servo motor controls the tray to start rotating, allowing the brush to clean the piston, and the tray with the piston continues to rise at a constant speed.

[0062] When the tray with the piston rises to the cleaning result judgment system, the image sensor captures an image of the cleaned piston surface and inputs the captured image to the terminal processor. The terminal processor processes the captured image to determine whether there are any uncleaned stains on the piston surface. If so, the tray is controlled to descend and the piston is cleaned again. Otherwise, the tray with the piston continues to rise at a constant speed, and the piston is sent to the drying and air washing system.

[0063] When the piston enters the drying and air washing system, the terminal processor drives the low-pressure nozzle and the fan to work, and performs low-pressure air washing and air drying on the piston. At the same time, the tray with the piston continues to rise at a uniform speed. The piston after low-pressure air washing and air drying is sent to the storage system. The image sensor on the robotic arm will capture the image of the piston surface and input the captured image to the terminal processor. The terminal processor performs stain detection on the captured image. When it is detected that there is no uncleaned stain on the piston surface, the robotic arm will send the piston to an external storage bin; otherwise, the robotic arm will send the piston to an external cleaning bin, and the terminal processor will perform damage detection on the image of the piston transported to the external cleaning bin. If the piston is damaged, it will no longer be cleaned; if the piston is not damaged, it will continue to be cleaned.

[0064] Based on the large amount of damage data information obtained during damage detection, a specialized damage detection model can be trained to facilitate subsequent damage detection and classification. A quantitative damage detection report will also be output to prepare for subsequent manual inspection.

[0065] The above-described embodiments merely illustrate several implementations of the present invention, and while the descriptions are relatively specific and detailed, they are not to be construed as limiting the scope of the present invention. Persons skilled in the art will readily appreciate that variations and modifications may be made without departing from the spirit of the present invention, and all such variations and modifications fall within the scope of protection of the present invention.

Claims

1. A fully automatic feedback piston cleaning device based on sensors and machine vision, characterized in that: It includes a transmission system, a feeding device, an intelligent cleaning system, a cleaning result judgment system, a drying and air washing system and a storage system arranged from bottom to top; The transmission system includes a terminal processor, a servo motor, a lifting controller and a tray, the terminal processor is used to control the servo motor and the lifting controller, the servo motor is used to control the rotation of the tray, and the lifting controller is used to control the lifting of the tray; The feeding device includes a piston clamp, which is used to clamp the piston. The piston clamp holding the piston will fall on the tray, and the lifting controller will control the tray with the piston to rise to the intelligent cleaning system for cleaning. The intelligent cleaning system is used to clean the piston; The intelligent cleaning system includes a pressure sensor, a brush fixed on the pressure sensor, a plurality of nozzles capable of spraying detergent, and a plurality of nozzles capable of spraying clean water, wherein one nozzle sprays one detergent, and different nozzles spray different detergents; when the piston contacts the brush, the pressure sensor outputs a washing signal to the terminal processor, the terminal processor outputs a rotation signal to the servo motor, the servo motor controls the tray to start rotating, and controls the rotation speed of the tray to reduce energy consumption; when the brush contacts the rotating piston, the terminal processor controls the nozzle capable of spraying detergent for cleaning the corresponding stains to spray the detergent to the piston, and when the detergent washing is completed, the terminal processor controls the nozzle capable of spraying clean water to spray clean water to the piston to complete a cleaning of the piston; and when the piston is no longer in contact with the brush, the pressure sensor outputs a first signal to the terminal processor, the terminal processor outputs a rotation stop signal to the servo motor, and the servo motor controls the tray to stop rotating; The cleaning result judgment system includes an image sensor, which is used to capture an image of the piston surface after cleaning and input the captured image into a terminal processor. The terminal processor processes the captured image to determine whether the piston surface contains any uncleaned stains. If so, the tray is controlled to descend and the piston is cleaned again; otherwise, the piston is sent to a drying and air washing system. The drying and air washing system is used to perform low-pressure air washing and air drying on the piston. After low-pressure air washing and air drying, the piston is sent to the storage system. The storage system includes a robotic arm with an image sensor. The robotic arm includes detection function and positioning and placement function. When it is detected that there is no uncleaned stain on the surface of the piston, the robotic arm will send the piston to an external storage bin; otherwise, the robotic arm will send the piston to an external cleaning bin.

2. The device according to claim 1, characterized in that A piston clamp fixing device is provided on the pallet, and the piston clamp fixing device is used to fix the piston clamp on the pallet.

3. The device according to claim 1, characterized in that The feeding device also includes a conveying mechanism for conveying the piston clamp, which clamps the piston clamp and transports it. When the conveying mechanism conveys the piston clamp to the position of the tray, the conveying mechanism releases the piston clamp and the piston clamp falls onto the tray.

4. The device according to claim 1, characterized in that The terminal processor processes the collected image, including: First, the collected image is enhanced, and then the enhanced image is denoised to remove noise in the image and ensure image clarity; the threshold is adjusted using threshold segmentation technology to separate the stains and background in the denoised image; then the edge detection algorithm is used to identify the edge of the stain; finally, opening and closing operations are applied to remove water stains and highlight the stains.

5. The device according to claim 4, characterized in that The method of judging whether there are uncleaned stains on the piston surface and, if so, controlling the tray to descend and re-cleaning the piston comprises: If there are stains on the surface of the piston that have not been cleaned, the terminal processor sends a signal for cleaning again to the lifting controller. The lifting controller first controls the tray to descend to the height where the piston and the brush can make initial contact, and then controls the tray to rise at a constant speed. The intelligent cleaning system cleans the piston again. When the tray with the piston rises to the cleaning result judgment system again, the cleaning result judgment system judges whether there is any uncleaned stain on the surface of the piston. If so, the piston is cleaned again. If the same piston still has stains after being cleaned three times, the piston will no longer be cleaned, but will be sent to the drying and air washing system for low-pressure air washing and air drying.

6. The device according to claim 1, characterized in that The drying and air washing system includes multiple fans for air drying the piston and multiple low-pressure nozzles for low-pressure air washing the piston. The low-pressure nozzles can perform low-pressure air washing over a large area, and together with the fans, they process water vapor, small particles and dust on the surface of the piston, completing the final cleaning of the piston.

7. The device according to claim 1, characterized in that The image sensor on the robotic arm captures an image of the piston surface and inputs the captured image to the terminal processor, which performs stain detection and damage detection on the captured image; The stain detection process includes: performing image enhancement processing on the collected image by the terminal processor, and then performing denoising processing on the enhanced image to remove noise in the image and ensure image clarity; adjusting the threshold using threshold segmentation technology to separate the stain and background in the denoised image; then using an edge detection algorithm to identify the stain edge; and finally applying opening and closing operations to remove water stains and highlight the stain. The damage detection comprises: performing grayscale conversion on the collected image by the terminal processor to obtain a grayscale image, performing denoising on the grayscale image, and then using histogram equalization to enhance the contrast of the denoised image to obtain a first image, extracting edge information in the first image using edge detection technology, and then using morphological image processing morphological operation technology to enhance features in the image and remove small noise points; The denoised color image is subjected to color segmentation, and then the color segmented image is divided into damaged areas and non-damaged areas using threshold segmentation technology. At the same time, the edge information is used to extract the contour, detect the shape and size of the damaged area, and then the area, perimeter and roundness of the damaged area contour are analyzed to determine the type of damage and perform damage classification.

8. A method for cleaning the piston of the device according to claim 7, characterized in that: The following steps are involved: The terminal processor sends an ascending signal to the lifting controller, which controls the pallet to ascend at a constant speed. At the same time, the lifting controller sends a transmitting signal to the conveying mechanism, which starts to convey the piston clamp. When the conveying mechanism conveys the piston clamp to the position of the pallet, the conveying mechanism releases the piston clamp, and the piston clamp falls onto the pallet. At the same time, the tray with the piston continues to rise at a constant speed to the intelligent cleaning system. When the piston contacts the brush, the pressure sensor outputs a cleaning signal to the terminal processor, which then outputs a rotation signal to the servo motor. The servo motor controls the tray to start rotating, allowing the brush to clean the piston, and the tray with the piston continues to rise at a constant speed. When the tray with the piston rises to the cleaning result judgment system, the image sensor captures an image of the cleaned piston surface and inputs the captured image to the terminal processor. The terminal processor processes the captured image to determine whether there are any uncleaned stains on the piston surface. If so, the tray is controlled to descend and the piston is cleaned again. Otherwise, the tray with the piston continues to rise at a constant speed, and the piston is sent to the drying and air washing system. When the piston enters the drying and air washing system, the terminal processor drives the low-pressure nozzle and the fan to work, and performs low-pressure air washing and air drying on the piston. At the same time, the tray with the piston continues to rise at a uniform speed. The piston after low-pressure air washing and air drying is sent to the storage system. The image sensor on the robotic arm will capture the image of the piston surface and input the captured image to the terminal processor. The terminal processor performs stain detection on the captured image. When it is detected that there is no uncleaned stain on the piston surface, the robotic arm will send the piston to an external storage bin; otherwise, the robotic arm will send the piston to an external cleaning bin, and the terminal processor will perform damage detection on the image of the piston transported to the external cleaning bin. If the piston is damaged, it will no longer be cleaned; if the piston is not damaged, it will continue to be cleaned.

Images