An intelligent identification and sorting system and method for multi-target medical waste classification

Abstract

This invention discloses an intelligent identification and sorting system and method for multi-target medical waste classification. The system adopts a closed enclosure with air curtain isolation units at the inlet and outlet to block airflow exchange, and a built-in ultraviolet disinfection unit for continuous disinfection. After the adaptive rake spreads and flattens the accumulated waste, the image acquisition unit, combined with a CNN deep learning model, identifies the waste category and size. The sorting execution unit grabs targets according to the large-size priority rule, shaking and separating small adhering waste into a temporary storage bin during grabbing. After secondary sorting via return flow, fragments are automatically collected into the final collection bin. This invention achieves fully unmanned operation, effectively blocking cross-infection paths, significantly improving classification accuracy and sorting thoroughness, and meeting the requirements for compliant disposal of medical waste.

Description

Technical Field

[0001] This invention relates to the field of medical waste treatment technology, specifically to an intelligent identification and sorting system and method for multi-objective medical waste classification. Background Technology

[0002] Medical waste is classified as high-risk waste in the "National Hazardous Waste List," and its classification, collection, and disposal are directly related to public health safety and environmental protection. According to the "Regulations on the Management of Medical Waste," medical waste must be classified and treated according to infectious waste, sharps waste, pathological waste, chemical waste, and pharmaceutical waste, and mixed collection is strictly prohibited.

[0003] Current medical waste sorting methods primarily rely on manual sorting or simple mechanical screening. In manual sorting, operators directly handle waste contaminated with pathogens, posing a high risk of cross-infection. Furthermore, manual sorting is inefficient, failing to meet the daily processing needs of large-scale medical institutions, and its accuracy is compromised by subjective factors. While some mechanical screening equipment reduces human contact to some extent, it typically only roughly separates waste by size through mesh screens, unable to identify specific categories (such as distinguishing between infectious swabs and abrasive needles), resulting in insufficient sorting accuracy. Extensive manual verification is still required, failing to fundamentally address the issues of efficiency and safety.

[0004] Furthermore, existing sorting equipment lacks a continuous disinfection mechanism for the internal environment during operation, allowing pathogens carried by the waste to easily proliferate and accumulate inside the equipment, creating a secondary source of pollution. The lack of effective airflow barriers at the inlet and outlet allows for free airflow inside and outside the container, increasing the risk of pathogens spreading outwards. For small pieces of waste adhering to the surface of large pieces of waste, existing equipment also lacks an effective separation and re-sorting mechanism, easily leading to incomplete sorting and affecting the compliance of subsequent processing.

[0005] Therefore, there is an urgent need for a medical waste sorting system and method that can achieve automation, high-precision identification, closed-loop disinfection throughout the entire process, and intelligent return sorting, in order to solve the problems of high risk of cross-infection, low sorting accuracy, incomplete sorting, and insufficient efficiency in existing technologies. Summary of the Invention

[0006] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0007] Therefore, the purpose of this invention is to provide an intelligent identification and sorting system and method for multi-objective medical waste classification, so as to solve the problems mentioned in the background art.

[0008] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution:

[0009] An intelligent identification and sorting system for multi-objective medical waste classification, comprising:

[0010] The box body has a feeding unit and a discharging unit on opposite sides, and an air curtain isolation unit on opposite sides to block the exchange of airflow inside and outside the box to prevent cross-infection.

[0011] A conveying assembly, located inside the container, is used to carry and convey medical waste;

[0012] A disinfection unit is installed on the inner wall of the top of the container to disinfect the environment inside the container and the medical waste being transported.

[0013] A rake assembly, disposed on the conveying path of the conveying assembly, is used to spread out accumulated medical waste;

[0014] An image acquisition unit is located downstream of the paving unit in the conveying path and is used to acquire image information of the medical waste after paving.

[0015] The intelligent identification unit is communicatively connected to the image acquisition unit and is used to analyze the image information based on deep learning algorithms to identify the type and size of medical waste.

[0016] The sorting execution unit is located downstream of the image acquisition unit in the conveying path. It is used to sort medical waste according to a preset priority based on the recognition result of the intelligent recognition module. The sorting execution unit can move back and forth along the direction of the conveying path.

[0017] The temporary storage bin is used to receive small-sized waste separated from large-sized waste during the sorting process, and the sorting execution unit returns this waste to the starting end of the conveying assembly for secondary sorting.

[0018] Waste storage bins are used to store the sorted waste.

[0019] The control unit is communicatively connected to the conveying assembly, image acquisition unit, intelligent recognition module, sorting execution unit, disinfection unit, air curtain isolation unit, sorting execution unit, feeding unit, and discharging unit, respectively, and is used to coordinate the operation sequence of each unit.

[0020] As a preferred embodiment of the intelligent identification and sorting system for multi-objective medical waste classification described in this invention, the rake assembly includes multiple liftable columns evenly arranged along the conveying direction and a rake rod with one end rotatably connected to the top of the columns.

[0021] As a preferred embodiment of the intelligent identification and sorting system for multi-target medical waste classification described in this invention, the disinfection unit consists of multiple ultraviolet lamps distributed on the top of the container. The ultraviolet lamps are continuously turned on or turned off according to a preset cycle during the operation of the system to fully disinfect the air, inner walls, and passing medical waste inside the container, preventing the accumulation of bacteria and viruses inside the container.

[0022] As a preferred embodiment of the intelligent identification and sorting system for multi-objective medical waste classification described in this invention, the intelligent identification module incorporates a deep learning model based on a convolutional neural network. The model pre-stores various categories of medical waste, size thresholds, and corresponding sorting rules, with larger waste having a higher sorting priority than smaller waste. The identification results simultaneously output the waste's category information and size level.

[0023] As a preferred embodiment of the intelligent identification and sorting system for multi-target medical waste classification described in this invention, the sorting execution unit is a mobile robot. The mobile robot has a multi-degree-of-freedom mechanical claw. The mechanical claw is used to shake off small pieces of waste adhering to large-sized waste into the collection bin when grabbing large-sized waste, and then put the large-sized waste into the corresponding category of waste storage bin.

[0024] As a preferred embodiment of the intelligent identification and sorting system for multi-objective medical waste classification described in this invention, the conveying component, feeding unit, and discharging unit are all conveyor belts.

[0025] As a preferred embodiment of the intelligent identification and sorting system for multi-target medical waste classification described in this invention, it further includes a fragment collection bin, which is located at the end of the conveying component and is used to collect fragmented waste that has not been picked up by the sorting execution unit.

[0026] An intelligent identification and sorting method for multi-objective medical waste classification includes the following steps:

[0027] S1. Activate the air curtain isolation unit to form an air curtain barrier on both sides of the box, blocking the exchange of airflow between the inside and outside of the box; at the same time, activate the disinfection unit to disinfect the environment inside the box.

[0028] S2. Medical waste is placed on the feeding unit and transported to the conveying component inside the container. The conveying component moves the waste forward along the conveying path.

[0029] S3. When the garbage passes through the rake assembly, the rake spreads the accumulated garbage evenly, making each piece of garbage dispersed and exposed.

[0030] S4. After the spread waste reaches the acquisition position of the image acquisition unit, the image acquisition unit acquires the image information of the waste and sends it to the intelligent recognition unit;

[0031] S5. The intelligent identification unit analyzes image information based on a convolutional neural network model, identifies the category label and size level of waste, and generates sorting instructions according to the preset rule that large-sized waste takes precedence over small-sized waste.

[0032] S6. The sorting execution unit moves to the target position along the conveying path and grabs the large-sized waste of the corresponding category according to the sorting instructions using a multi-degree-of-freedom mechanical claw. After grabbing, the small pieces of waste attached to it are shaken off to the temporary storage bin by high-frequency small-amplitude shaking.

[0033] S7. The sorting unit will dispose of the large-sized waste that has been shaken off into the corresponding waste storage bin;

[0034] S8. After a single mainline sorting process is completed, the sorting execution unit will grab the small-sized waste in the temporary storage bin and put it back to the starting end of the conveying component, and return to step S3 for secondary sorting until all the waste in this batch has been sorted.

[0035] S9. During the conveying process, small, fragmented waste that is too small to be gripped by the mechanical claw moves to the end with the conveying component and automatically falls into the fragment collection bin.

[0036] S10. When garbage needs to be discharged, the sorting execution unit will grab the sorted garbage storage box or slag collection bin and transfer it to the discharge unit, which will then send it out of the box.

[0037] As a preferred embodiment of the intelligent identification and sorting method for multi-objective medical waste classification described in this invention, in step S5, the intelligent identification unit divides the waste into three levels according to size: large-sized waste with a size greater than a first preset threshold, medium-sized waste with a size between the first preset threshold and a second preset threshold, and fragments with a size less than the second preset threshold.

[0038] The sorting priority decreases sequentially. Large-sized waste is sorted first, followed by medium-sized waste. Small pieces are directly conveyed into the small piece collection bin, achieving step-by-step sorting according to size gradient.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] 1. By setting up air curtain isolation units at the inlet and outlet of the container, a non-contact air curtain barrier is formed, which keeps the airflow inside and outside the container isolated while the garbage continuously enters and exits the container, thus physically blocking the channels for pathogens to spread outward. In conjunction with the ultraviolet disinfection unit arranged on the inner wall of the top of the container, the air, inner wall and garbage being transported inside the container are continuously or periodically disinfected, inhibiting the growth and accumulation of bacteria inside the equipment and greatly reducing the risk of cross-infection between operators and the external environment.

[0041] 2. The convolutional neural network model is used to analyze the collected garbage images in real time, which can simultaneously identify the category attributes and size grades of garbage. This breaks through the limitation of traditional mechanical screening that only relies on aperture to distinguish size, and realizes dual determination of category and size. The classification accuracy is much higher than that of manual sorting and traditional screening equipment, reducing the workload of subsequent manual verification.

[0042] 3. By setting a sorting rule that prioritizes large-sized waste over small-sized waste, and in conjunction with the shaking function of the mechanical claw, small pieces of waste adhering to the surface of large pieces of waste can be actively shaken off into a temporary storage bin when large pieces of waste are grabbed. Then, the small pieces of waste in the temporary storage bin are returned to the conveying start end by the sorting execution unit for secondary sorting, forming a closed-loop return mechanism. This effectively solves the problem of incomplete sorting caused by large pieces of waste carrying out small pieces of waste, and ensures that all types of waste can be accurately classified into the corresponding storage bins.

[0043] 4. The rake arm assembly achieves height adjustment through a liftable column and has an adaptive avoidance function that swings around the axis. When encountering large pieces of garbage, it can automatically lift to avoid them and prevent blockage. After resetting, it continues to spread and flatten the subsequent garbage, making the stacked garbage dispersed and exposed, providing clear and unobstructed shooting conditions for the image acquisition unit, while providing a reliable gripping posture for the mechanical claw, thus improving the recognition accuracy and gripping success rate from the source. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0045] Figure 1 This is a three-dimensional structural diagram of an intelligent identification and sorting system for multi-objective medical waste classification according to the present invention;

[0046] Figure 2 This is a front view of an intelligent identification and sorting system for multi-objective medical waste classification according to the present invention.

[0047] Figure 3This is a flowchart of an intelligent identification and sorting method for multi-objective medical waste classification according to the present invention. Detailed Implementation

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0049] Example 1

[0050] like Figure 1 and Figure 2 As shown in the figure, this embodiment discloses an intelligent identification and sorting system for multi-objective medical waste classification. The system adopts a closed box structure, and each functional unit is integrated inside the box 100 or carried by the box 100, realizing the fully automated and closed treatment of medical waste from feeding to discharging.

[0051] The housing 100 is a rectangular sealed cavity made of stainless steel with an antibacterial coating on the inner wall. An inlet and an outlet are respectively located on opposite sides of the housing 100. An inlet unit 110 is installed at the inlet, and an outlet unit 120 is installed at the outlet. Both inlet and outlet units 110 employ a conveyor belt structure. The inlet unit 110 connects to the conveying assembly 200 inside the housing 100, forming a continuous conveying channel. Air curtain isolation units 130 are installed above the outer sides of the inlet and outlet, respectively. These air curtain isolation units 130 use a high-speed fan to generate a flat airflow curtain from top to bottom, forming an invisible air curtain barrier between the conveying surfaces of the inlet and outlet units 110 and the external environment. When waste is fed into the container 100 via the feeding unit 110, the air curtain operates continuously to prevent air containing pathogens from escaping from inside the container 100. Similarly, the air curtain at the discharge unit 120 prevents unfiltered air from flowing back into the container 100, thus achieving bidirectional airflow isolation.

[0052] The conveying assembly 200 is arranged along the length of the housing 100 and adopts a belt conveyor structure. The surface of the conveyor belt is provided with anti-slip texture to prevent the waste from sliding and deviating during the conveying process. The drive motor of the conveying assembly 200 is controlled by the control module, which can adjust the conveying speed according to the sorting cycle, such as slowing down or pausing briefly when the image acquisition unit 500 is taking pictures, to ensure the clarity of the image.

[0053] The disinfection unit 300 is located at the top inner part of the housing 100, with multiple ultraviolet germicidal lamps evenly arranged along the extension direction of the conveying path. The wavelength range is 200nm-275nm, covering the area above the conveying assembly 200. After the system is powered on, the disinfection unit 300 enters the working state and can be set to continuous illumination mode or intermittent working mode to disinfect the air flowing through the housing 100, the surface of the conveyor belt, and the waste itself in real time. The ultraviolet lamps are surrounded by a quartz glass protective cover to prevent moisture and corrosive gases from corroding the lamp tubes, extending their service life.

[0054] The rake assembly 400 is installed upstream of the conveying path of the conveying assembly 200, located upstream of the image acquisition unit 500. The rake assembly 400 includes multiple liftable columns 410 evenly spaced along the conveying direction. The bottom of each column 410 is fixed to the bottom plate of the housing 100, and its top is rotatably connected to one end of the rake 420 via a pivot. The rake 420 is a long, strip-shaped rod made of lightweight alloy material, with its free end extending downwards above the bearing surface of the conveying assembly 200, maintaining a gap of 5mm-20mm with the conveyor belt surface. The column 410 has a built-in electric push rod or screw lifting mechanism, which can drive the rake 420 to rise and fall as a whole, thereby adjusting the gap between the rake 420 and the conveyor belt to accommodate waste of different thicknesses. When large pieces of waste pass under the rake 420, the rake 420 is lifted upwards by force around its axis, and after passing the large pieces of waste, it automatically returns to its original position under gravity, restoring the paved state. This adaptive oscillating structure avoids jamming or conveyor belt overload shutdown caused by rigid blocking.

[0055] The image acquisition unit 500 is located downstream of the rake assembly 400 and includes a high-definition industrial camera and a matching light source assembly. The industrial camera is mounted on the top inner side of the housing 100, with its lens vertically downwards and aimed at the center area of ​​the bearing surface of the conveyor assembly 200, covering the entire width of the conveyor belt. The light source assembly uses a ring-shaped LED white light, arranged around the camera lens or providing supplemental lighting from the side to eliminate shadows on the waste surface and ensure image quality. The image data acquired by the camera is transmitted to the intelligent recognition unit for processing via gigabit Ethernet.

[0056] The intelligent identification unit is an embedded industrial control computer or GPU edge computing module, with a built-in deep learning model trained based on a convolutional neural network (CNN). This model is pre-trained using a labeled medical waste image dataset, which includes annotation information for various types of samples, such as infectious waste (e.g., blood-stained swabs, gauze), sharps waste (e.g., needles, scalpel blades), and plastic containers (e.g., IV bottles, medicine bottles). Each sample image is labeled with the waste category, bounding box size, and center point coordinates. After deployment, the model receives image frames from the image acquisition unit 500 in real time and outputs the category label (e.g., infectious, sharps, plastic) and size class (e.g., large, medium, small) for each waste target in each image. The intelligent identification unit also has a built-in sorting rule table, specifying the target storage bin number corresponding to different types of waste, as well as size priority rules (large size prioritizes medium size, medium size prioritizes small size).

[0057] The sorting execution unit 600 is a track-mounted mobile robot, arranged on one side of the conveying assembly 200. The sorting execution unit 600 includes a transverse moving guide rail, a slide, and a multi-degree-of-freedom mechanical gripper 610 mounted on the slide. The transverse moving guide rail extends along the conveying direction, and the slide is driven by a servo motor to reciprocate along the guide rail, allowing the mechanical gripper 610 to reach directly above any piece of waste on the conveying path. The mechanical gripper 610 uses a pneumatic or electric gripper with a three-finger or two-finger structure and adjustable gripping force, capable of grasping both large infusion bottles and small syringes. The mechanical gripper 610 also has a shaking function: after grasping large pieces of waste, the gripper can perform high-frequency, small-amplitude shaking at a preset frequency and amplitude, dislodging small pieces of waste adhering to the surface of the large waste. The dislodged small pieces of waste fall into a temporary storage bin 700 located below the mechanical gripper 610's path.

[0058] The temporary storage bin 700 is located on the side of the conveyor assembly 200 near the tail end, with its opening facing upwards, to receive small-sized waste shaken off by the sorting execution unit 600. After a single main line sorting process—from the entry of the first batch of waste until all large and medium-sized waste that can be grabbed is sorted—is completed, the control module instructs the sorting execution unit 600 to move above the temporary storage bin 700. The mechanical gripper 610 clamps the temporary storage bin 700 and flips it over, emptying the accumulated small-sized waste into the starting end of the conveyor assembly 200. The conveyor assembly 200 then starts, performing a secondary sorting of this batch of small-sized waste. This return process can be repeated multiple times until no new waste falls into the temporary storage bin 700 or all waste has been sorted and categorized.

[0059] There are multiple waste storage bins 800, each corresponding to a different category of medical waste, such as infectious waste bins, sharps waste bins, and plastic recycling bins. Each storage bin 800 is arranged side-by-side on the other side of the conveying assembly 200, opposite to the guide rail of the sorting execution unit 600. The position coordinates of each storage bin 800 are pre-stored in the control module. The sorting execution unit 600, based on the category label output by the intelligent identification unit, places the collected waste into the corresponding numbered storage bin 800. The storage bins 800 feature a lid design, opening only briefly when the robotic gripper 610 disposes of waste, remaining closed at other times to further reduce contact between contaminants inside and the outside environment.

[0060] The debris collection bin is located below the end of the conveying assembly 200 to collect debris that is too small to be gripped by the mechanical claw 610 during the conveying process. Since this type of debris does not have a complete category or is too small to be effectively grasped, the system classifies it as unsortable debris.

[0061] The control module serves as the central hub of the system, employing a PLC or industrial PC as the main controller. It connects to the drivers, sensors, and actuators of each unit via I / O interfaces and a communication bus. The control module incorporates runtime logic to coordinate the start-up, shutdown, and actions of each unit. For example, it activates the air curtain isolation unit 130 and the disinfection unit 300 before starting the feeding unit 110; it sends pulse signals to briefly stop or slow down the conveyor assembly 200 when the image acquisition unit 500 is capturing images; it plans the movement path and gripping sequence of the robotic gripper 610 based on the output of the intelligent recognition unit; and it triggers the flipping and reflux action of the temporary storage bin 700 after the main line sorting is completed. The control module also features a human-machine interface (such as a touchscreen) for operators to view the system's operating status, manually intervene, or set parameters.

[0062] Example 2

[0063] Based on the system architecture of Embodiment 1 above, this embodiment discloses an intelligent identification and sorting method for multi-objective medical waste classification, the process of which is as follows: Figure 3 As shown, the specific steps include:

[0064] S1. When the operator presses the start button, the control module first instructs the air curtain isolation unit 130 to start at the feeding unit 110 and the discharging unit 120 to form an air curtain barrier; at the same time, it instructs the ultraviolet lamps of the disinfection unit 300 to light up and continuously disinfect the internal space of the container 100. After ensuring that the environment inside the container reaches a safe state, it enters standby mode to wait for the garbage to be put in.

[0065] S2. Medical waste is placed on the conveyor belt of the feeding unit 110 by operators or upstream equipment. The feeding unit 110 feeds the waste into the conveying assembly 200 inside the container 100. The conveying assembly 200 runs at a preset first speed, moving the waste along the conveying path towards the discharge end. At this time, the air curtain isolation unit 130 continues to operate, preventing the polluted air brought in by the waste from spreading outward.

[0066] S3. The waste moves with the conveyor assembly 200 to the location of the rake assembly 400. The rake 420 maintains a downward posture under its own weight, with its free end maintaining an appropriate gap with the surface of the conveyor belt. When the waste passes under the rake 420, the clumps of waste are dispersed and flattened by the rake 420, and the waste that was originally piled up at the bottom is turned over to the top, so that each piece of waste is basically distributed in a single layer. If a large piece of waste with a thickness exceeding the gap of the rake 420 is encountered, the rake 420 swings upward around its rotation axis to allow the large piece of waste to pass smoothly and avoid obstruction; after the large piece of waste passes, the rake 420 returns to its original position under the action of gravity and continues to spread the subsequent waste. The spread waste continues to be conveyed downstream in a relatively dispersed and stable state.

[0067] S4. The spread-out waste enters the field of view of the image acquisition unit 500. At this time, the control module sends a trigger signal, and the conveying component 200 decelerates to the second speed or pauses briefly. Simultaneously, the industrial camera of the image acquisition unit 500 captures one or more high-definition images under supplementary lighting conditions. After the images are captured, the conveying component 200 resumes its first speed and continues operating. The acquired image data is transmitted to the intelligent recognition unit via Ethernet.

[0068] S5. After receiving the image data, the intelligent recognition unit calls the pre-trained convolutional neural network model to perform inference analysis on the image. The model first performs target detection on the image, delineates the bounding rectangle of each waste target, extracts its visual features, and then the classification head outputs the category probability distribution of each target, while the size regression head outputs the bounding dimensions of the target. The intelligent recognition unit classifies the size into three levels according to preset thresholds: those with a length or width greater than 150mm are large sizes, those between 50mm and 150mm are medium sizes, and those less than 50mm are small sizes (i.e., fragments). The system generates a sorting instruction sequence according to the rule of prioritizing large sizes, followed by medium sizes, and then small sizes, and sends the category label, size level, and position information of each target in the conveyor belt coordinate system to the control module.

[0069] S6. According to the sorting instruction sequence, the control module drives the slide of the sorting execution unit 600 to move along the guide rail to directly above the first target to be sorted. The robotic gripper 610 descends to the gripping height and clamps the large-sized waste. After clamping, the robotic gripper 610 performs a high-frequency, small-amplitude shaking motion to shake off small pieces of waste adhering to the surface of the large-sized waste. The shaken-off small pieces of waste fall into the temporary storage bin 700 below. Then, the robotic gripper 610 lifts up, carrying the large-sized waste to above the waste storage bin 800 corresponding to the waste category, releases the gripper, and puts the waste into the bin. After disposal, the robotic gripper 610 returns to the top of the conveyor path, ready to grab the next target.

[0070] S7. Repeat step S6, picking up and sorting various types of waste along the conveyor path one by one, prioritizing large-sized waste and then medium-sized waste. During this process, all small pieces of waste that fall off from large and medium-sized waste are collected in the temporary storage bin 700. When all the waste in the current batch on the conveyor component 200 has passed through the working range of the sorting execution unit 600, and there are no new large or medium-sized targets available for picking, the main line sorting process ends.

[0071] S8. After the main line sorting is completed, the control module determines whether there is any accumulated waste in the temporary storage bin 700. If so, it instructs the sorting execution unit 600 to move above the temporary storage bin 700, and the mechanical gripper 610 clamps the temporary storage bin 700 and flips it over, dumping the small-sized waste accumulated in the bin to the starting end of the conveying component 200. After dumping, the temporary storage bin 700 is reset. The control module restarts the conveying component 200, using the small-sized waste as the new round of sorting objects, and repeats the spreading, identification, and sorting process of steps S3-S7. Since the small-sized waste was not prioritized during the initial main line sorting, this secondary sorting is equivalent to supplementing its identification and classification, improving the overall sorting completeness. This return process can be repeated multiple times depending on the accumulation status of the temporary storage bin 700, until the temporary storage bin 700 is empty after one main line sorting, or the preset maximum number of return times is reached. After this, the system determines that the batch of waste has completed all sorting operations.

[0072] S9. During the sorting process described above, small, fragmented waste that is too small and not successfully gripped by the mechanical claw 610 continues to move towards the end with the conveying component 200. When the conveying component 200 reaches the end, these fragments fall naturally under the influence of gravity and fall into the fragment collection bin 900 located at the end, completing the automatic collection of fragments.

[0073] S10. When a waste storage bin 800 is nearing full capacity, the system issues a warning signal. Upon receiving the warning, the operator waits outside the discharge unit 120. The control module instructs the sorting execution unit 600 to move to the side of the full waste storage bin 800. The robotic gripper 610 clamps the storage bin 800 and transports it to the conveyor belt of the discharge unit 120. The discharge unit 120 starts, sending the storage bin 800 out of the container 100. The operator removes the full bin and replaces it with an empty one. The discharge unit 120 then sends the empty bin back into the container 100, and the sorting execution unit 600 returns the empty bin to its original storage position. The replacement process for the slag collection bin 900 is similar; the sorting execution unit 600 transfers it to the discharge unit 120 for delivery. Throughout the entire discharge process, the air curtain isolation unit 130 remains operational, ensuring airflow isolation between the inside of the container 100 and the external environment.

[0074] Through the above steps, the system achieves fully automated processing of medical waste from feeding, spreading, identification, sorting, re-sorting and discharge, minimizing direct contact between humans and pollutants while ensuring classification accuracy.

[0075] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. An intelligent identification and sorting system for multi-objective medical waste classification, characterized in that, include: The box (100) has a feeding unit (110) and a discharging unit (120) on opposite sides, and an air curtain isolation unit (130) on opposite sides to block the exchange of airflow inside and outside the box (100) to prevent cross-infection. A conveying assembly (200), disposed inside the container (100), is used to carry and convey medical waste; A disinfection unit (300) is installed on the inner wall of the top of the box (100) for disinfecting the environment inside the box (100) and the medical waste being transported. A rake assembly (400) is disposed on the conveying path of the conveying assembly (200) for spreading out accumulated medical waste; An image acquisition unit (500) is located downstream of the paving unit in the conveying path and is used to acquire image information of the medical waste after paving. The intelligent identification unit is communicatively connected to the image acquisition unit (500) and is used to analyze the image information based on a deep learning algorithm to identify the type and size of medical waste; The sorting execution unit (600) is located downstream of the image acquisition unit (500) on the conveying path. It is used to sort medical waste according to a preset priority based on the recognition result of the intelligent recognition module. The sorting execution unit (600) can move back and forth along the direction of the conveying path. The temporary storage bin (700) is used to receive small-sized waste separated from large-sized waste during the sorting process, and the sorting execution unit (600) returns this part of the waste to the starting end of the conveying assembly (200) for secondary sorting; Waste storage bin (800) is used to store the sorted waste; The control unit is communicatively connected to the conveying assembly (200), image acquisition unit (500), intelligent identification module, sorting execution unit, disinfection unit (300), air curtain isolation unit (130), sorting execution unit (600), feeding unit (110), and discharging unit (120), respectively, and is used to coordinate the operation sequence of each unit.

2. The intelligent identification and sorting system for multi-objective medical waste classification according to claim 1, characterized in that, The rake assembly (400) includes a plurality of liftable columns (410) evenly arranged along the conveying direction and a rake (420) with one end rotatably connected to the top of the column (410).

3. The intelligent identification and sorting system for multi-objective medical waste classification according to claim 1, characterized in that, The disinfection unit (300) consists of multiple ultraviolet lamps distributed on the top of the box (100). The ultraviolet lamps are continuously turned on or turned off according to a preset cycle during the operation of the system to fully disinfect the air, inner walls and passing medical waste inside the box (100) and prevent bacteria and viruses from accumulating inside the box (100).

4. The intelligent identification and sorting system for multi-objective medical waste classification according to claim 1, characterized in that, The intelligent recognition module has a built-in deep learning model based on convolutional neural networks. The model has pre-stored various categories of medical waste, size thresholds, and corresponding sorting rules. Larger waste is sorted with higher priority than smaller waste. The recognition results simultaneously output the waste's category information and size level.

5. The intelligent identification and sorting system for multi-objective medical waste classification according to claim 1, characterized in that, The sorting execution unit (600) is a mobile robot with a multi-degree-of-freedom mechanical claw (610). The mechanical claw (610) is used to shake off small pieces of garbage adhering to large-sized garbage into the collection bin (700) when grabbing large-sized garbage, and then put the large-sized garbage into the corresponding type of garbage storage bin (800).

6. The intelligent identification and sorting system for multi-target medical waste classification according to claim 1, wherein the conveying component (200), the feeding unit (110) and the discharging unit (120) are all conveyor belts.

7. The intelligent identification and sorting system for multi-target medical waste classification according to claim 1 further includes a fragment collection bin, which is disposed at the end of the conveying component (200) for collecting fragmented waste that has not been picked up by the sorting execution unit (600).

8. An intelligent identification and sorting method for multi-objective medical waste classification, implemented based on the system described in any one of claims 1-8, characterized in that, Includes the following steps: S1. Activate the air curtain isolation unit (130) to form an air curtain barrier on both sides of the box (100) to block the exchange of airflow inside and outside the box (100); at the same time, activate the disinfection unit (300) to disinfect the environment inside the box (100); S2. Medical waste is placed on the feeding unit (110) and transported to the conveying component (200) inside the container (100) via the feeding unit (110). The conveying component (200) drives the waste to move forward along the conveying path. S3. When the garbage passes through the rake assembly (400), the rake (420) spreads the accumulated garbage flat, so that each piece of garbage is dispersed and exposed. S4. After the spread waste reaches the acquisition position of the image acquisition unit (500), the image acquisition unit (500) acquires the image information of the waste and sends it to the intelligent recognition unit; S5. The intelligent identification unit analyzes image information based on a convolutional neural network model, identifies the category label and size level of waste, and generates sorting instructions according to the preset rule that large-sized waste takes precedence over small-sized waste. S6. The sorting execution unit (600) moves to the target position along the conveying path and grabs the large-sized waste of the corresponding category according to the sorting instructions by the multi-degree-of-freedom mechanical claw (610). After grabbing, the small pieces of waste attached to it are shaken off to the temporary storage bin (700) by high-frequency small-amplitude shaking. S7. The sorting unit (600) places the large-sized waste that has been shaken off into the corresponding waste storage bin (800); S8. After a single main line sorting process is completed, the sorting execution unit (600) grabs the small-sized garbage in the temporary storage bin (700) and puts it back to the starting end of the conveying component (200), and returns to step S3 for secondary sorting until all the garbage in this batch is sorted. S9. During the conveying process, small pieces of waste that are too small to be gripped by the mechanical claw (610) move to the end with the conveying assembly (200) and automatically fall into the waste collection bin. S10. When garbage needs to be discharged, the sorting execution unit (600) grabs the sorted garbage storage box (800) or the slag collection bucket and transfers it to the discharge unit (120), which then sends it out of the box (100).

9. The intelligent identification and sorting method for multi-objective medical waste classification according to claim 8, wherein in step S5, the intelligent identification unit divides the waste into three levels according to size: large-sized waste with a size greater than a first preset threshold, medium-sized waste with a size between the first preset threshold and a second preset threshold, and fragments with a size less than the second preset threshold. The sorting priority decreases sequentially. Large-sized waste is sorted first, followed by medium-sized waste. Small pieces are directly conveyed into the small piece collection bin, achieving step-by-step sorting according to size gradient.

Images