An ultrasonic dust removal method
By combining ultrasonic vibration and suction devices with image recognition technology, efficient and precise dust removal from the surface of semiconductor chips is achieved, overcoming the shortcomings of traditional cleaning methods and making it suitable for non-destructive cleaning of high-precision components.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHONGKEJIANWEI INTELLIGENT EQUIPMENT (SUZHOU) CO LTD
- Filing Date
- 2024-08-31
- Publication Date
- 2026-07-31
AI Technical Summary
In semiconductor chip manufacturing, existing technologies often fail to meet the requirements for high-precision cleaning. Liquid cleaning may leave residual moisture that causes corrosion, while adhesive cleaning methods are limited in their effectiveness against fine dust and are costly.
The system uses ultrasonic vibration to decompose dust and then removes it through an air suction device. By combining image recognition and model training, ultrasonic parameters and the position of the air suction device are determined, thus achieving controllable decomposition and removal of dust.
It provides non-contact cleaning, reaching deep into tiny crevices to precisely locate dust, avoiding energy waste and secondary pollution, and is suitable for non-destructive cleaning of high-precision components.
Smart Images
Figure CN118874959B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of semiconductor manufacturing technology, specifically to an ultrasonic dust removal method. Background Technology
[0002] In the semiconductor industry, especially in chip manufacturing, a cleanroom environment is indispensable for ensuring product quality. Chip manufacturing is an extremely delicate and complex process, and even the smallest dust particle can significantly affect the electrical performance of the chip, leading to short circuits, signal interference, or other unpredictable problems, thereby reducing the overall chip quality and reliability.
[0003] In existing technologies, chips are typically cleaned using liquid cleaning or adhesive methods. Traditional liquid cleaning involves soaking or spraying to remove oil, dust, and other contaminants from the chip surface. Adhesive methods use sticky rollers or adhesive paper to remove tiny particles.
[0004] Liquid cleaning methods may leave residual moisture, which is detrimental to humidity-sensitive components and may cause corrosion or deformation of certain materials. Furthermore, the cleaning solution may be harmful to the environment. Adhesive cleaning methods are better suited for removing large surface particles, have limited effectiveness against fine dust, cannot provide deep cleaning, and require frequent replacement of adhesive materials, resulting in high costs. Therefore, a more efficient and precise cleaning method is needed to meet the extremely high cleanliness requirements of chip cleaning scenarios. Summary of the Invention
[0005] This invention provides an ultrasonic dust removal method that can decompose dust adhering to the surface of product parts through ultrasonic vibration, and then control the suction force and position of the suction device to remove the dust, thereby achieving controllable decomposition and removal of dust.
[0006] This invention provides an ultrasonic dust removal method, comprising the following steps:
[0007] Obtain the parts of the product to be cleaned;
[0008] Acquire an image of the surface of the product to be cleaned and identify the dust characteristics on the surface of the components of the product to be cleaned;
[0009] The cleaning test results are determined based on the dust characteristics on the surface of the product components to be cleaned;
[0010] Based on the detection results to be cleaned, the corresponding ultrasonic frequency, ultrasonic intensity, and cleaning duration are generated.
[0011] Using the generated ultrasonic frequency, ultrasonic intensity, and cleaning duration, the dust on the surface of the product parts to be cleaned is decomposed by ultrasonic oscillation.
[0012] Based on the particle characteristics of the decomposed dust, the suction force of the suction device and the position of the suction device are determined.
[0013] It should be further noted that acquiring the image of the surface of the product to be cleaned and identifying the dust features on the surface of the product components to be cleaned includes:
[0014] Histogram equalization is performed on the image to be processed to improve image contrast;
[0015] The image to be processed is improved by using the dark channel prior method;
[0016] Perform grayscale conversion on the image to be processed;
[0017] The image to be processed is divided into a dust part and a background part by threshold segmentation;
[0018] The dust shape is separated and optimized to obtain the optimized dust shape;
[0019] Extract the shape, size, texture, area, and location features of the dust, input them into the dust feature judgment model, and obtain the dust feature set of the image to be processed.
[0020] It should be further noted that training the dust feature judgment model includes:
[0021] Acquire a first image to be processed containing different dust conditions and a corresponding standard image without dust, and create a corresponding label for each of the first images to be processed to describe the characteristics of the dust;
[0022] The first image to be processed is compared pixel by pixel with the standard image to obtain the first comparison result;
[0023] Substitute the first comparison result into the loss function to calculate the loss result.
[0024] It should be further noted that training the dust feature judgment model also includes:
[0025] The gradient descent optimization algorithm is used to compare the loss result with the expected result of the standard image to obtain a second comparison result;
[0026] The second comparison result is substituted into the loss function for iterative calculation. A dust feature judgment model is constructed based on the loss function whose iterative calculation result is less than the threshold.
[0027] It should be further noted that the dust feature judgment model is used to output the predicted dust features.
[0028] It should be further noted that the suction power of the suction device and the position of the suction device are determined based on the particle diameter and density of the decomposed dust, including:
[0029] Based on the particle diameter and density of the dust after vibration, estimate the settling velocity of the dust particles in the static fluid.
[0030] The suction force of the air intake device is calculated based on the pressure difference between the inside of the air intake device and the surrounding environment and the area of the air intake port of the air intake device.
[0031] The position of the suction device is determined based on the suction force of the suction device and the settling velocity of the dust particles in the still fluid.
[0032] It should be further noted that determining the position of the suction device means determining the angle and distance between the suction device and the product component to be cleaned.
[0033] Beneficial effects:
[0034] First, the ultrasonic dust removal method provided in this application is a non-contact cleaning method, which avoids damage to fragile or precision parts, and can penetrate into tiny gaps and complex structures to remove dirt that is difficult to reach by traditional methods. It has a good removal effect on tiny particles (such as nanoscale particles).
[0035] Secondly, by accurately locating dust through image recognition, it guides the optimization of ultrasonic parameters, enabling targeted cleaning and avoiding energy waste.
[0036] Third, dynamically adjusting the ultrasonic frequency, intensity, and duration improves the cleaning effect.
[0037] Finally, the suction power and position are intelligently adjusted based on the characteristics of the dust to ensure thorough removal of residual particles and avoid secondary pollution.
[0038] In summary, the ultrasonic dust removal method provided in this application not only significantly improves cleaning efficiency but also reduces manual intervention, making it suitable for non-destructive cleaning of high-precision components and possessing broad industrial application prospects. Attached Figure Description
[0039] To more clearly illustrate the technical solutions in this invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0040] Figure 1 This is a first specific flowchart of the ultrasonic dust removal method provided in an embodiment of the present invention.
[0041] Figure 2 This is a second specific flowchart of the ultrasonic dust removal method provided in the embodiments of the present invention.
[0042] Figure 3 This is a flowchart illustrating the dust feature judgment model provided in an embodiment of the present invention.
[0043] Figure 4 This is a third specific flowchart of the ultrasonic dust removal method provided in the embodiments of the present invention.
[0044] Figure 5 This is a schematic diagram of the ultrasonic dust removal device provided in an embodiment of the present invention.
[0045] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Specific Implementation
[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0047] The following sections provide detailed descriptions. It should be noted that the order of the following embodiments is not intended to limit the preferred order of the embodiments. Figure 1 As shown, the specific process of this ultrasonic dust removal method can be as follows:
[0048] This embodiment can be applied to various scenarios such as removing tiny particulate contaminants during chip manufacturing and circuit board assembly, and removing tiny metal shavings in medical device production.
[0049] 110. Obtain the product parts to be cleaned.
[0050] It is understandable that before obtaining the product parts to be cleaned, clean air is used to blow and clean the product parts to remove large particles of dust from the surface.
[0051] Among them, product components can be chips, precision parts, optical components, medical devices, etc.
[0052] In another specific embodiment, a gentle brush can also be used to perform preliminary cleaning of the product parts to remove large particles of dust and loose dirt from the surface.
[0053] 120. Obtain an image of the surface of the product to be cleaned and identify the dust features on the surface of the product components to be cleaned.
[0054] The dust characteristics on the surface of the product components to be cleaned include shape characteristics, size characteristics, texture characteristics, area characteristics, and positional characteristics.
[0055] Understandably, these dust characteristics can be captured using optical sensors, such as 3D imaging sensors and spectral analysis sensors. Alternatively, combining image processing software with machine vision sensors can automatically analyze and identify dust in images, allowing for real-time detection and marking of areas requiring further cleaning.
[0056] 130. Determine the cleaning test result based on the dust characteristics on the surface of the product component to be cleaned.
[0057] The results of the cleaning test can include dust type, adhesion level, product component material, cleanliness requirements, and object geometry. For example, the dust type can be inorganic dust, organic dust, mixed dust, or dust from a specific industry, such as the electronics industry or the chemical industry.
[0058] Suppose there is a circuit board of an electronic device with dust contamination on its surface. The dust characteristics include: the dust particles are spherical, with a diameter between 0.1-0.5 mm; the surface texture is rough; the dust covers 10% of the total circuit board area; the dust is mainly concentrated around the integrated circuits on the circuit board. Based on these characteristics, the following cleaning test results can be derived: Based on the shape and texture, it is judged to be particles possibly generated by metal fragments or wear of electronic components; due to the large and concentrated dust area and high degree of adhesion, deep cleaning is required; the circuit board material is likely a PCB (Printed Circuit Board), as this is a common material in electronic devices; considering the sensitivity of the circuit board, the cleanliness requirement is extremely high, requiring the use of specialized cleaning agents and methods that will not damage the circuit board; the object is a rectangular circuit board with a complex layout of circuits and components.
[0059] 140. Based on the cleaning test results, generate the corresponding ultrasonic frequency, ultrasonic intensity, and cleaning duration.
[0060] Understandably, let's assume a cleaning test result is: "The object surface contains mixed dust, the product components are made of stainless steel, optical-grade cleanliness is required, and the object surface has micropores." Based on this result, a frequency of 40kHz or higher can be selected to penetrate the micropores. An intensity of 0.3-0.5 W / cm² should be chosen. 2The process effectively cleans without damaging stainless steel. The duration is set to 10-15 minutes to ensure thorough removal of mixed dust. The above examples are limited to understanding the inventive concept of this application. Those skilled in the art can adjust these parameters to achieve optimal cleaning results based on actual cleaning performance. Therefore, by generating corresponding ultrasonic frequencies, ultrasonic intensities, and cleaning durations based on the cleaning test results, it is possible to ensure that the energy is just right to knock down dust without damaging product components.
[0061] 150. Using the generated ultrasonic frequency, ultrasonic intensity, and cleaning duration, the dust particles on the surface of the product parts to be cleaned are decomposed by ultrasonic vibration.
[0062] It is understandable that, when choosing a frequency, low frequencies (around 20kHz) are suitable for large particles or hard contaminants because they can more effectively impact and remove larger or more stubborn dirt; high frequencies (above 40kHz) are suitable for cleaning small particles or fine structures, as high-frequency ultrasound can penetrate into tiny crevices to clean finer contaminants.
[0063] 160. Based on the particle characteristics of the decomposed dust, determine the suction force of the suction device and the position of the suction device.
[0064] The particulate characteristics of dust include the diameter and density of the dust particles.
[0065] Understandably, larger particles settle more quickly due to gravity, requiring lower suction power and a suitable location to avoid being sucked in too quickly and causing blockages. Smaller particles, on the other hand, may remain suspended in the air for longer periods, requiring higher suction power and a suitable location to ensure effective capture. This ensures that the dust particle capture efficiency of the suction device remains within a preset range, guaranteeing that dust is effectively sucked in rather than being re-scattered back into the area to be cleaned.
[0066] In some embodiments, reference Figure 2 120, acquiring an image of the surface of the product to be cleaned and identifying dust features on the surface of the product components to be cleaned, includes the following sub-steps:
[0067] 121. Perform histogram equalization on the image to be processed to improve image contrast;
[0068] 122. The image to be processed is processed using the dark channel prior method to improve image clarity;
[0069] 123. Perform grayscale conversion on the image to be processed;
[0070] 124. The image to be processed is divided into a dust part and a background part by threshold segmentation;
[0071] 125. Separate and optimize the dust shape to obtain the optimized dust shape;
[0072] 126. Extract the shape, size, texture, area, and location features of the dust, input them into the dust feature judgment model, and obtain the dust feature set of the image to be processed.
[0073] Histogram equalization, used to enhance image contrast, can be achieved by calculating the cumulative distribution function (CDF) of each pixel and mapping it to a new grayscale value. The expression is as follows:
[0074] G(x)=min{L-1,max{0,F^-1(x)+(L-1)×(x-0.5)}}
[0075] Where G(x) is the new gray value, F^-1(x) is the inverse cumulative distribution function of the original image gray value, and L is the gray level.
[0076] The grayscale conversion of the image to be processed involves converting a color image to a grayscale image to simplify subsequent processing. The conversion formula is as follows:
[0077] gray=0.2989×R+0.5870×G+0.1140×B
[0078] Wherein, gray represents the gray value that reflects the visual effect of the human eye, R represents the intensity of the red component, G represents the intensity of the green component, and B represents the intensity of the blue component.
[0079] The process of dividing the image to be processed into a dust portion and a background portion through threshold segmentation can be represented as follows:
[0080] G(x,y)={255ifI(x,y)>T,0otherwise}
[0081] Where G(x,y) is the gray value of pixel (x,y) in the binarized image, i.e., the pixel value of the output image, I(x,y) is the gray value of the corresponding pixel (x,y) in the original image, i.e., the pixel value of the input image, and T is the threshold, which is a pre-set gray value used to distinguish between black and white parts.
[0082] In some embodiments, reference Figure 3 126, training the dust feature judgment model includes the following steps:
[0083] 1261. Obtain a first image to be processed containing different dust conditions and a corresponding standard image without dust, and create a corresponding label for each of the first images to be processed to describe the characteristics of the dust;
[0084] 1262. Compare the first image to be processed with the standard image pixel by pixel to obtain the first comparison result;
[0085] 1263. Substitute the first comparison result into the loss function to calculate the loss result;
[0086] 1264. Use the gradient descent optimization algorithm to compare the loss result with the expected result of the standard image to obtain a second comparison result;
[0087] 1265. Substitute the second comparison result into the loss function for iterative calculation, and construct a dust feature judgment model based on the loss function whose iterative calculation result is less than the threshold.
[0088] The dust feature judgment model is used to output the predicted dust features.
[0089] The characteristics of dust can include type, quantity, and location.
[0090] This involves comparing each first image to be processed, I1, with its corresponding standard dust-free image, I0, pixel by pixel. This can be achieved through subtraction, yielding the difference image D.
[0091] D = |I1 - I0|
[0092] The mean squared error (MSE) can be used to calculate the first comparison result, which measures the difference between the predicted result (i.e., the comparison result) and the expected value (standard image).
[0093] The loss function can be minimized using the gradient descent algorithm. Updating the standard image I0 to reduce the loss can be expressed as:
[0094]
[0095] Here, [i] represents the i-th element of the vector, and α is the learning rate. It is the gradient of the loss function. By applying the same update rule to each element, we obtain an updated I0, which makes the loss function smaller than before.
[0096] Therefore, through iteration and optimization, the model can better capture the effects of dust and improve recognition accuracy. Moreover, the trained model has good generalization ability and can be applied to new and unseen images containing dust for effective recognition and analysis.
[0097] In some embodiments, reference Figure 4 In step 160, determining the suction power of the suction device and the position of the suction device based on the particle diameter and density of the decomposed dust includes the following sub-steps:
[0098] 161. Based on the particle diameter and density of the dust after vibration, estimate the settling velocity of the dust particles in a static fluid.
[0099] 162. Calculate the suction force of the air intake device based on the pressure difference between the inside of the air intake device and the surrounding environment and the area of the air intake port of the air intake device;
[0100] 163. Determine the position of the suction device based on the suction force of the suction device and the settling velocity of the dust particles in the still fluid.
[0101] The determination of the position of the suction device involves determining the angle and distance between the suction device and the product component to be cleaned.
[0102] Understandably, heavy particles, due to their large mass, tend to settle easily even if small in size. Therefore, the adsorption device should be set with appropriate suction power and angle to prevent excessively rapid inhalation. Lighter particles, floating in the air, require greater suction power and more precise angles and distances to ensure capture. Furthermore, the suction power of the adsorption device should be adjusted according to the size and weight of the dust particles. For example, if most particles are larger than 50μm in diameter and heavier, the suction power can be set to medium; if the particles are small and light, the suction power should be set to high. In addition, the angle between the adsorption device and the part of the product being cleaned also affects the path of the particles into the adsorption port. For example, for larger or heavier particles, maintain a vertical or near-vertical angle so that the particles fall directly; for smaller or lighter particles, it may be necessary to adjust to an angle of 45° or less to increase the chance of capture. The distance should be close enough to capture the particles, but not too close to avoid interfering with the vibration process or causing excessively rapid inhalation. For large particles, the distance can be slightly greater, while for small particles, they should be as close as possible.
[0103] Based on the particle diameter and density of the dust particles after vibration, the settling velocity of the dust particles in the static fluid can be estimated using Stokes' law, expressed as follows:
[0104] v_s=frac{2g×rho_p(rho_p-rho_f)d_p^2}{18×mu_f}
[0105] Where v_s is the settling velocity of the particle, g is the gravitational acceleration, rho_p is the particle density, rho_f is the fluid density, d_p is the particle diameter, and mu_f is the fluid viscosity.
[0106] The suction force of the intake device is primarily determined by the pressure difference between the device's interior and the surrounding environment. Ideally, this pressure difference can be calculated using Bernoulli's theorem, expressed as:
[0107] F = Delta P × A
[0108] Where Delta P is the pressure difference between the inlet of the inhalation device and the environment, and A is the area of the inlet of the inhalation device.
[0109] The location of the suction device is determined by the angle and distance between the suction device and the part of the product to be cleaned. Assuming the suction force is greater than the settling velocity of the particles (i.e., sufficient to overcome the particle's gravity), the suction device should be placed near the dust source to prevent particles from falling to the ground during settling. The specific location needs to be adjusted by those skilled in the art based on the actual environment and equipment characteristics. For example, if the suction force is insufficient to completely overcome particle settling, it needs to be placed closer to the source or the suction force increased.
[0110] This invention also provides an ultrasonic dust removal device; please refer to [link / reference]. Figure 5 , Figure 5 This is a schematic diagram of the ultrasonic dust removal device provided in the embodiments of this application. The ultrasonic dust removal device 200 may include: a pretreatment module 210, a dust feature recognition module 220, a cleaning result determination module 230, an ultrasonic wave generation module 240, an ultrasonic wave processing module 250, and an air suction device determination module 260.
[0111] Pre-processing module 210 acquires the product parts to be cleaned;
[0112] The dust feature recognition module 220 is used to acquire a processing image of the surface of the product to be cleaned and to identify the dust features on the surface of the product component to be cleaned.
[0113] The cleaning result determination module 230 is used to determine the cleaning detection result based on the dust characteristics of the surface of the product component to be cleaned;
[0114] The ultrasonic generation module 240 is used to generate corresponding ultrasonic frequency, ultrasonic intensity and cleaning duration based on the detection results to be cleaned;
[0115] The ultrasonic processing module 250 is used to decompose dust particles on the surface of the product parts to be cleaned by means of ultrasonic vibration using the generated ultrasonic frequency, ultrasonic intensity and cleaning duration.
[0116] The suction device determination module 260 is used to determine the suction force of the suction device and the position of the suction device based on the particle characteristics of the decomposed dust.
[0117] In the ultrasonic dust removal device provided in this embodiment, the image processing device can obtain the product component to be cleaned through the preprocessing module 210, then identify the dust features on the surface of the product component to be cleaned through the dust feature recognition module 220, and then determine the cleaning detection result based on the dust features on the surface of the product component to be cleaned through the cleaning result determination module 230. Next, the ultrasonic generation module 240 generates the corresponding ultrasonic frequency, ultrasonic intensity, and cleaning duration based on the cleaning detection result. The ultrasonic processing module 250 then uses the generated ultrasonic frequency, ultrasonic intensity, and cleaning duration to decompose the dust particles on the surface of the product component to be cleaned through ultrasonic vibration. Finally, the suction device determination module 260 determines the suction force of the suction device and the position of the suction device based on the particle diameter and density of the decomposed dust. Thus, the ultrasonic dust removal device can achieve controllable decomposition and removal of dust, achieving a more efficient, accurate, and precise cleaning effect.
[0118] refer to Figure 6 This application also provides an electronic device 300. The electronic device can be a smartphone, tablet computer, gaming device, AR (Augmented Reality) device, automobile, video playback device, laptop computer, desktop computing device, wearable device such as an electronic helmet, electronic clothing, etc. The electronic device 300 includes a processor 301 and a memory 302. The processor 301 and memory 302 are electrically connected. The processor 301 is the control center of the electronic device 300, connecting various parts of the electronic device 300 through various interfaces and lines. By running or calling computer programs stored in the memory 302, and by calling data stored in the memory 302, it executes various functions of the electronic device 300 and processes data, thereby performing overall monitoring of the electronic device 300.
[0119] In this embodiment, the processor 301 in the electronic device 300 loads the instructions corresponding to the processes of one or more computer programs into the memory 302 according to the following steps, and the processor 301 runs the computer programs stored in the memory 302 to perform the following steps:
[0120] Obtain the parts of the product to be cleaned;
[0121] Acquire an image of the surface of the product to be cleaned and identify the dust characteristics on the surface of the components of the product to be cleaned;
[0122] The cleaning test results are determined based on the dust characteristics on the surface of the product components to be cleaned;
[0123] Based on the detection results to be cleaned, the corresponding ultrasonic frequency, ultrasonic intensity, and cleaning duration are generated.
[0124] Using the generated ultrasonic frequency, ultrasonic intensity, and cleaning duration, the dust on the surface of the product parts to be cleaned is decomposed by ultrasonic oscillation.
[0125] Based on the particle characteristics of the decomposed dust, the suction force of the suction device and the position of the suction device are determined.
[0126] This application also provides a storage medium storing a computer program. When the computer program is run on a computer, the computer executes the ultrasonic dust removal method described in any of the above embodiments.
[0127] It should be noted that those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, which may include, but is not limited to, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.
[0128] In the description of this application, it should be understood that terms such as “first” and “second” are used only to distinguish similar objects and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.
[0129] The ultrasonic dust removal method, apparatus, and electronic equipment provided in the embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and its core ideas. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. An ultrasonic dust removal method, characterized in that, Includes the following steps: Obtain the parts of the product to be cleaned; Acquire an image of the surface of the product to be cleaned and identify dust features on the surface of the product components; wherein, acquiring the image of the surface of the product to be cleaned and identifying dust features on the surface of the product components includes: Histogram equalization is performed on the image to be processed to improve image contrast; The image to be processed is improved by using the dark channel prior method; Perform grayscale conversion on the image to be processed; The image to be processed is divided into a dust part and a background part by threshold segmentation; The dust shape is separated and optimized to obtain the optimized dust shape; Extract the shape, size, texture, area, and location features of the dust, input them into the dust feature judgment model, and obtain the dust feature set of the image to be processed; The cleaning test results are determined based on the dust characteristics on the surface of the product components to be cleaned; Based on the detection results to be cleaned, the corresponding ultrasonic frequency, ultrasonic intensity, and cleaning duration are generated. Using the generated ultrasonic frequency, ultrasonic intensity, and cleaning duration, the dust on the surface of the product parts to be cleaned is decomposed by ultrasonic oscillation. Based on the characteristics of the decomposed dust particles, the suction power of the suction device and the position of the suction device are determined; wherein, determining the suction power of the suction device and the position of the suction device based on the particle diameter and density of the decomposed dust includes: Based on the particle diameter and density of the dust after vibration, estimate the settling velocity of the dust particles in the static fluid. The suction force of the air intake device is calculated based on the pressure difference between the inside of the air intake device and the surrounding environment and the area of the air intake port of the air intake device. The position of the suction device is determined based on the suction force of the suction device and the settling velocity of the dust particles in the still fluid; determining the position of the suction device involves determining the angle and distance between the suction device and the part of the product to be cleaned.
2. The ultrasonic dust removal method according to claim 1, characterized in that, Training the dust feature judgment model includes: Acquire a first image to be processed containing different dust conditions and a corresponding standard image without dust, and create a corresponding label for each of the first images to be processed to describe the characteristics of the dust; The first image to be processed is compared pixel by pixel with the standard image to obtain the first comparison result.
3. The ultrasonic dust removal method according to claim 2, characterized in that, Training the dust feature judgment model further includes: Substitute the first comparison result into the loss function to calculate the loss result; The gradient descent optimization algorithm is used to compare the loss result with the expected result of the standard image to obtain a second comparison result; The second comparison result is substituted into the loss function for iterative calculation. A dust feature judgment model is constructed based on the loss function whose iterative calculation result is less than the threshold.
4. The ultrasonic dust removal method according to claim 3, characterized in that, The dust feature judgment model is used to output predicted dust features.