Intelligent grading system for fried prepared food

Abstract

The invention relates to the field of agricultural product processing equipment, in particular to an intelligent deep-fried prepared food grading system which comprises a supporting frame, a conveying mechanism, a detecting mechanism, a grading executing mechanism, a control mechanism and a collecting mechanism. The supporting frame is constructed by aluminum profiles to form a mounting foundation; the conveying mechanism comprises a stainless steel mesh conveying belt, a motor and an encoder; the detection mechanism comprises an industrial camera and an edge computing device, a deep learning model embedded in the edge computing device classifies products in real time, grabbing position coordinates and grabbing time intervals of the unqualified products are sent to a control mechanism end Arduino development board, and the Arduino development board determines the grabbing position coordinates of the unqualified products according to the two pieces of received information and the speed of a conveying belt. Task scheduling, movement control of a Delta robot and opening and closing grabbing of a fin line claw are further conducted, unqualified products above the conveying belt are safely grabbed to a defective product collecting area, qualified products and the unqualified products fall into corresponding collecting frames after reaching the tail end of the conveying belt, and the consistency and the percent of pass of fried prepared food are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of agricultural product processing equipment, and in particular to an intelligent grading system for fried and conditioned food. BACKGROUND

[0002] Cephalopod aquatic products (such as squid) are a common aquatic product, rich in protein, and after conditioning, powder coating and frying processing, they have unique color and crisp taste, and are favored by consumers. The important process after sorting and frying is currently mainly dependent on manual sorting, which has high labor intensity, high production cost and difficult to guarantee product consistency.

[0003] Therefore, it is necessary to provide an intelligent grading system for fried and conditioned food to solve the above technical problems. SUMMARY

[0004] To solve the above technical problems, the present application provides an intelligent grading system for fried and conditioned food.

[0005] The intelligent grading system for fried and conditioned food provided by the present application comprises a support frame, a conveying mechanism, a detection mechanism, a grading execution mechanism, a control mechanism and a collection mechanism; the support frame is formed by aluminum profiles to form an installation base; the conveying mechanism comprises a stainless steel mesh conveyor belt, a motor and an encoder, the motor is drivingly connected with the stainless steel mesh conveyor belt to drive it to convey the fried and conditioned food, and the encoder is used to collect the running speed of the stainless steel mesh conveyor belt; the grading execution mechanism comprises a Delta robot, an air compressor and a pressure regulating valve, the Delta robot comprises a stepper motor, a fixed base plate, a driving arm, a driven arm, a spherical hinge joint, a static platform and a pneumatic flexible mechanical claw, the air compressor is connected with the pneumatic flexible mechanical claw in gas circuit through the pressure regulating valve, the stepper motor is installed on the fixed base plate, one end of the driving arm is connected with the output end of the stepper motor, the other end is connected with one end of the driven arm through the spherical hinge joint, the other end of the driven arm is connected with the static platform through the spherical hinge joint, and the static platform is connected with the pneumatic flexible mechanical claw; the control mechanism comprises an electric control box; the detection mechanism comprises an industrial camera installed on the support frame and an edge computing device installed in the electric control box, a deep learning model runs on the edge computing device, and the industrial camera is used to collect image information of the fried and conditioned food on the stainless steel mesh conveyor belt in real time. The electric control box is provided with a 24V DC power supply, an edge computing device, a gas solenoid valve, an Arduino development board, a DC voltage reduction module and a 57-step motor closed-loop driving module. The edge computing device is embedded with an improved CBAM_YOLO v8 deep learning model. An industrial camera, an encoder and the edge computing device are signal connected. The edge computing device is in communication connection with the Arduino development board through a serial port. The Arduino development board is electrically connected with the gas solenoid valve and the 57-step motor closed-loop driving module. The 57-step motor closed-loop driving module is electrically connected with the stepping motor of the Delta robot. The gas solenoid valve is in series connection with a pressure regulating valve in a gas circuit. Preferably, the sizes of the driving arm and the driven arm of the Delta robot are determined by a robot kinematics equation and the size of a hierarchical plane region. The three stepping motors of the Delta robot are driven in parallel. The pneumatic flexible mechanical gripper is moved to an arbitrary position in the workspace of the robot by changing the stepping angle. The position is calculated by a robot kinematics inverse solution formula.

[0006] Preferably, the improved CBAM_YOLO v8 deep learning model of the edge computing device performs frame-by-frame detection and grading on the images collected by the industrial camera. When the number of times that the same ID product is judged as a "defective" category is greater than a set threshold value, the product is locked as a "defective" category, and the current plane coordinates and time value of the product are output.

[0007] Preferably, the edge computing device sends the output plane coordinates to the task queue of the Arduino development board after coordinate conversion. The Arduino development board controls the Delta robot to move to the grabbing position according to the time value of the locked "defective" category received by task scheduling, and controls the gas solenoid valve to open, so that the pneumatic flexible mechanical gripper performs a grabbing action.

[0008] Preferably, the DC voltage reduction module is used to convert the voltage output by the 24V DC power supply into a working voltage suitable for the Arduino development board.

[0009] Preferably, the stainless steel mesh conveyor belt is arranged below the oil-frying air-cooling end discharge port. The shooting area of the industrial camera covers the to-be-detected section of the stainless steel mesh conveyor belt.

[0010] Preferably, the collecting mechanism includes a defective product collecting frame and a qualified product collecting frame, which are arranged in different hierarchical regions at the end of the stainless steel mesh conveyor belt.

[0011] Preferably, the defective product collecting frame and the qualified product collecting frame are arranged side by side along the width direction of the stainless steel mesh conveyor belt, and correspond to the grabbing and placing areas of the Delta robot, respectively.

[0012] The pneumatic flexible mechanical claw is preferably a profiled fin line claw structure, which is used for flexible grabbing of the fried prepared food.

[0013] Compared with the related art, the application has the following advantages: The fried prepared food intelligent grading system has the advantages of low cost, simple structure, flexible and safe grading, high sorting efficiency and accuracy, and the like. On the structure, a simple high-frame-rate camera, a self-made Delta parallel robot and an edge computing device are added on the mesh conveyor belt, and the like. After real-time detection of the production line image frame signal, the queue task scheduling is performed, the deep learning model embedded in the edge computing device performs real-time grading on the product, the grabbing position coordinates and grabbing time interval of the unqualified product are sent to the control mechanism end Arduino development board, the Arduino development board further performs task scheduling and movement control of the Delta robot and opening and closing grabbing of the fin line claw according to the received two information and the conveyor belt speed, the unqualified product above the conveyor belt is safely grabbed to the defective product collection area, and the qualified product and the unqualified product fall into the corresponding collection frame at the end of the conveyor belt, thereby ensuring the consistency and qualified rate of the fried prepared food. The system can effectively replace the traditional manual grading, is reliable in work, effectively reduces the production cost and improves the economic benefit. BRIEF DESCRIPTION OF DRAWINGS

[0014] Fig. 1 FIG. 1 is a schematic diagram of the overall structure of the application; Fig. 2 FIG. 2 is a schematic diagram of the Delta robot structure of the application; Fig. 3 FIG. 3 is a main program grading flowchart of the application.

[0015] In the figure, 1 is a support frame, 2 is a conveying mechanism, 3 is a detection mechanism, 4 is a grading execution mechanism, 5 is a control mechanism, 6 is a collection mechanism, 7 is a stainless steel mesh conveyor belt, 8 is a motor, 9 is an encoder, 10 is an industrial camera, 11 is a Delta robot, 12 is an air compressor, 13 is a stepping motor, 14 is a fixed base plate, 15 is a driving arm, 16 is a driven arm, 17 is a spherical hinge joint, 18 is a pneumatic flexible mechanical claw, 19 is a pressure regulating valve, 20 is an electric control box, 21 is a defective product collection frame, 22 is a qualified product collection frame, and 23 is a static platform. DETAILED DESCRIPTION

[0016] The application will be further described below in combination with the drawings and embodiments.

[0017] Please refer to Figs. 1-3The application discloses an intelligent grading system for fried and conditioned food, which is composed of a support frame 1, a conveying mechanism 2, a detection mechanism 3, a grading execution mechanism 4, a control mechanism 5 and a collecting mechanism 6, and realizes automatic intelligent grading of the fried and conditioned food through the cooperation of the six parts. The support frame 1 is made of aluminum profiles as core building materials, and the profiles are spliced and fixed to form the mounting base of the whole system, thereby providing a stable mounting platform and support for the conveying mechanism 2, the detection mechanism 3, the grading execution mechanism 4 and other components. The conveying mechanism 2 is a core component for food conveying, and specifically comprises a stainless steel mesh conveyor belt 7, a motor 8 and an encoder 9. The control mechanism 5 is the control center of the system and mainly comprises an electric control box 20. The detection mechanism 3 comprises an industrial camera 10 installed on the support frame 1 and an edge computing device installed in the electric control box 20. The grading execution mechanism 4 is responsible for the grabbing and grading actions of substandard food, and comprises a Delta robot 11, an air compressor 12 and a pressure regulating valve 19. The Delta robot 11 is the execution core and is composed of a stepping motor 13, a fixed base plate 14, a driving arm 15, a driven arm 16, a spherical hinge joint 17, a static platform 23 and a pneumatic flexible mechanical gripper 18. The fixed base plate 14 is the mounting base of the Delta robot 11, the stepping motor 13 is fixedly installed on the preset mounting position of the fixed base plate 14 through bolts, one end of the driving arm 15 is fixedly connected with the output shaft of the stepping motor 13 through a coupling, the other end of the driving arm 15 is movably connected with one end of the driven arm 16 through the spherical hinge joint 17, the other end of the driven arm 16 is also connected with the static platform 23 through the spherical hinge joint 17, the pneumatic flexible mechanical gripper 18 is installed on the static platform 23, thereby forming a flexible grabbing mechanism. Meanwhile, the air outlet of the air compressor 12 is connected with the air inlet of the pressure regulating valve 19 through an air pipe, the air outlet of the pressure regulating valve 19 is connected with the air path interface of the pneumatic flexible mechanical gripper 18 through an air pipe, thereby providing stable air pressure for the opening and closing of the pneumatic flexible mechanical gripper 18. The electric control box 20 is internally provided with a 24V DC power supply, an edge computing device, a gas solenoid valve, an Arduino development board, a DC voltage reduction module and a 57-step motor closed-loop driving module; the 24V DC power supply provides basic power for all electrical components inside the electric control box 20, and the DC voltage reduction module converts the 24V DC voltage into a voltage suitable for the operation of the Arduino development board; the edge computing device internally embeds an improved CBAM_YOLO v8 deep learning model, the image signal output end of the industrial camera 10 and the signal output end of the encoder 9 are connected to the signal input end of the edge computing device through data lines, realizing real-time transmission of image data and speed data; the edge computing device is connected to the serial communication interface of the Arduino development board through a serial line, and can transmit control instructions and data to the Arduino development board; the control signal output end of the Arduino development board is electrically connected to the control end of the gas solenoid valve and the control end of the 57-step motor closed-loop driving module, and the output end of the 57-step motor closed-loop driving module is electrically connected to the stepper motor 13 of the Delta robot 11, and the gas solenoid valve is connected in series in the gas circuit between the pressure regulating valve 19 and the pneumatic flexible mechanical claw 18, realizing the control of the on-off of the gas pressure. Preferably, the specific size parameters of the driving arm 15 and the driven arm 16 of the Delta robot 11 need to be accurately calculated and determined through the robot kinematics equation combined with the size of the actual hierarchical plane area, so as to ensure that the working range of the Delta robot 11 can completely cover the hierarchical area of the stainless steel mesh conveyor belt 7; and the Delta robot 11 adopts a parallel driving mode of three stepper motors 13, and when working, by changing the step angle of the three stepper motors 13, the driving arm 15 and the driven arm 16 are linked to move the pneumatic flexible mechanical claw 18 to any position in the robot working space, which is calculated by the inverse solution formula of the robot kinematics, to ensure the accuracy of the grabbing position; Preferably, the improved CBAM_YOLO v8 deep learning model in the edge computing device can detect and grade the food images collected by the industrial camera 10 frame by frame, the system will assign a unique ID to each food entering the detection area to track its motion trajectory, when the same ID food is judged by the model as the "defective" category more than a preset threshold, the system will lock the product as the "defective" category, and output the current plane coordinates and time value of the product; Preferably, after outputting the planar coordinates of the "defective" food, the edge computing device first converts the coordinates from the image coordinate system to the spatial coordinate system recognizable by the Delta robot 11, and then sends the converted coordinate data and the control command for "grabbing the defective product" to the task queue of the Arduino development board through the serial port; the Arduino development board will determine the optimal grabbing time window according to the real-time feedback of the conveyor belt speed from the encoder 9 and the current position of the "defective" food, to avoid execution conflicts when multiple foods enter the detection area at the same time. Preferably, the core function of the DC voltage reduction module is to convert the voltage output by the 24V DC power supply in the electric control box 20 to a voltage value compatible with the working voltage of the Arduino development board, to ensure that the Arduino development board can be powered stably and work normally, and to avoid equipment damage or functional failure caused by voltage mismatch. Preferably, the installation position of the stainless steel mesh conveyor belt 7 needs to be set below the end discharge port of the oil-frying and air-cooling equipment, so that the food after oil-frying and air-cooling can directly fall onto the stainless steel mesh conveyor belt 7 for conveying; at the same time, the shooting area of the industrial camera 10 needs to completely cover the detection section of the stainless steel mesh conveyor belt 7, to ensure that all foods passing through this area can be captured by the industrial camera 10 to obtain clear images. Preferably, the collecting mechanism 6 specifically includes two independent collecting components, i.e., a defective product collecting frame 21 and a qualified product collecting frame 22, both of which are arranged at the end position of the stainless steel mesh conveyor belt 7 and correspond to different classification areas at the end of the stainless steel mesh conveyor belt 7, respectively, for collecting defective products and qualified products. Preferably, the defective product collecting frame 21 and the qualified product collecting frame 22 are arranged in parallel along the width direction of the stainless steel mesh conveyor belt 7, wherein the position of the defective product collecting frame 21 corresponds to the front belt grabbing and placing area of the Delta robot 11, to facilitate the Delta robot 11 to directly drop the grabbed defective products; the qualified product collecting frame 22 corresponds to the natural conveying area of the stainless steel mesh conveyor belt 7, so that the qualified products not determined as defective products can directly fall into the conveying belt. Preferably, the pneumatic flexible mechanical claw 18 adopts a profiled fin line claw structure design, which has good flexibility and adaptability, can conform to the shape of the oil-fried and conditioned food during grabbing, realize flexible grabbing, effectively avoid problems such as food breakage and skin peeling caused by rigid grabbing, and protect the appearance and quality of the food. The system takes "detection-determination-execution-collection" as the core logic, and the components corresponding to the numerals cooperatively realize the automatic intelligent classification of the oil-fried and conditioned food, and the specific process is as follows: 1. System initialization and installation positioning The whole system is deployed below the oil-frying and air-cooling terminal discharge port, and the support frame 1 is stably installed by aluminum profiles; after the power supply of the electric control box 20 is turned on, the 24V DC power supply inside is converted into the working voltage suitable for the Arduino development board (integrated in the electric control box 20) through the DC voltage reduction module (integrated in the electric control box 20), while the edge computing device (integrated in the electric control box 20), the 57-step motor closed-loop driving module (integrated in the electric control box 20) and other control components complete initialization, and the Delta robot 11 is reset to the standby state, preparing for the subsequent grading process.

[0018] 2. Food conveying and speed collection The motor 8 starts and drives the stainless steel mesh conveyor belt 7 to run at a constant speed, conveying the conditioned food after oil-frying and air-cooling from the oil-frying and air-cooling terminal discharge port to the detection area; in this process, the encoder 9 collects the running speed of the stainless steel mesh conveyor belt 7 in real time and transmits the speed signal to the edge computing device (integrated in the electric control box 20), which is used for subsequent food motion trajectory prediction and grabbing timing calibration to ensure accurate matching of grabbing action and food position.

[0019] 3. Image detection and substandard product determination The industrial camera 10 continuously collects images of the detection section of the stainless steel mesh conveyor belt 7 and transmits the real-time food image signals to the edge computing device (integrated in the electric control box 20); the edge computing device analyzes the food in each frame of image by its embedded improved CBAM_YOLOv8 deep learning model (distinguishes between qualified products and substandard products), and assigns a unique ID to each food entering the detection area to track its motion trajectory on the stainless steel mesh conveyor belt 7.

[0020] When the food with the same ID is determined as a "substandard product" category by the improved CBAM_YOLOv8 model more than a preset threshold number of times, the system locks the food as a "substandard product" category; at the same time, the edge computing device combines the conveyor belt speed collected by the encoder 9 and the food motion time to calculate and output the planar coordinates (in the image coordinate system) of the "substandard product" food when it reaches the "collision line" position (i.e. the grabbing trigger position) of the stainless steel mesh conveyor belt 7.

[0021] 4. Coordinate conversion and task scheduling The edge computing device (integrated in the electric control box 20) converts the “defective” food plane coordinates (image coordinate system) output in step 3 into a spatial coordinate system coordinate recognizable by the Delta robot 11, and then sends the converted coordinate data and the “grab defective” instruction to the task queue of the Arduino development board (integrated in the electric control box 20) through the serial port; the Arduino development board determines the optimal grabbing time window according to the real-time feedback of the conveyor belt speed and the current position of the “defective” food by the encoder 9, to avoid execution conflicts when multiple foods enter the detection area at the same time.

[0022] 5. Grading execution and classified collection 1. Grasping action control: The Arduino development board (integrated in the electric control box 20) sends a control signal to the 57-step motor closed-loop driving module (integrated in the electric control box 20), which drives the three parallel stepper motors 13 of the Delta robot 11; the three stepper motors 13 adjust the step angle according to the inverse kinematics formula of the robot, drive the main arm 15 and the driven arm 16 to link through the spherical hinge joint 17, and finally make the pneumatic flexible mechanical gripper 18 accurately move to the grabbing position of the “defective” food.

[0023] 2. Flexible grasping: At the same time, the Arduino development board controls the gas solenoid valve (integrated in the electric control box 20) to open, and the compressed air generated by the air compressor 12 is stabilized by the pressure regulating valve 19, then delivered to the pneumatic flexible mechanical gripper 18 through the gas circuit, to drive the gripper body to flexibly close, realizing non-destructive grabbing of the “defective” food (avoiding food breakage caused by rigid grabbing).

[0024] 3. Classified collection is completed: The Delta robot 11 drives the pneumatic flexible mechanical gripper 18 that grabs the “defective” food to move to the “defective” grading area at the end of the stainless steel mesh conveyor belt 7, then the gas solenoid valve is closed, the pneumatic flexible mechanical gripper 18 is released, and the “defective” food falls into the placement area until it reaches the end of the conveyor belt and falls into the defective product collection frame 21; the qualified products that are not determined as “defective” continue to be conveyed by the stainless steel mesh conveyor belt 7, and finally naturally fall into the qualified product collection frame 22 (the defective product collection frame 21 and the qualified product collection frame 22 are arranged side by side along the width direction of the stainless steel mesh conveyor belt 7, and correspond to the robot grabbing placement area and the conveyor belt natural conveying area respectively), completing the single grading process.

[0025] The above only describes the embodiments of the present application, and does not limit the patent scope of the present application, any equivalent structure or equivalent process transformation using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A smart grading system for fried prepared foods, characterized in that, It includes a support frame (1), a conveying mechanism (2), a detection mechanism (3), a hierarchical execution mechanism (4), a control mechanism (5), and a collection mechanism (6); The supporting frame (1) is constructed from aluminum profiles to form an installation foundation; The conveying mechanism (2) includes a stainless steel mesh conveyor belt (7), a motor (8) and an encoder (9). The motor (8) is connected to the stainless steel mesh conveyor belt (7) to drive it to convey fried and prepared food. The encoder (9) is used to collect the running speed of the stainless steel mesh conveyor belt (7). The graded actuator (4) includes a Delta robot (11), an air compressor (12), and a pressure regulating valve (19). The Delta robot (11) includes a stepper motor (13), a fixed base plate (14), an active arm (15), a driven arm (16), a ball joint (17), a stationary platform (23), and a pneumatic flexible mechanical claw (18). The air compressor (12) is connected to the pneumatic flexible mechanical claw (18) via the pressure regulating valve (19). The stepper motor (13) is mounted on the fixed base plate (14). One end of the active arm (15) is connected to the output end of the stepper motor (13), and the other end is connected to one end of the driven arm (16) via the ball joint (17). The other end of the driven arm (16) is connected to the stationary platform (23) via the ball joint (17). The stationary platform (23) is connected to the pneumatic flexible mechanical claw (18). The control mechanism (5) includes an electrical control box (20); The detection mechanism (3) includes an industrial camera (10) installed on the support frame (1) and an edge computing device installed in the electrical control box (20). The deep learning model runs on the edge computing device. The industrial camera (10) is used to collect image information of fried and prepared food on the stainless steel mesh conveyor belt (7) in real time. The electrical control box (20) is equipped with a 24V DC power supply, a gas solenoid valve, an Arduino development board, a DC step-down module, and a 57 stepper motor closed-loop drive module. The edge computing device is embedded with an improved CBAM_YOLO v8 deep learning model. The industrial camera (10) and encoder (9) are connected to the edge computing device via signal. The edge computing device is connected to the Arduino development board via a serial port. The Arduino development board is electrically connected to the gas solenoid valve and the 57 stepper motor closed-loop drive module, respectively. The 57 stepper motor closed-loop drive module is electrically connected to the stepper motor (13) of the Delta robot (11). The gas solenoid valve is connected in series with the pressure regulating valve (19).

2. The intelligent grading system for fried prepared foods according to claim 1, characterized in that, The dimensions of the active arm (15) and driven arm (16) of the Delta robot (11) are determined by the robot kinematic equation and the size of the graded planar region; the three stepper motors (13) of the Delta robot (11) are driven in parallel, and the pneumatic flexible mechanical claw (18) is moved to any position in the robot's workspace by changing the step angle. This position is calculated by the robot kinematic inverse kinematic formula.

3. The intelligent grading system for fried prepared foods according to claim 1, characterized in that, The improved CBAM_YOLO v8 deep learning model of the edge computing device performs frame-by-frame detection and classification of the images captured by the industrial camera (10). When the number of times the same ID product is judged as "defective" exceeds the set threshold, the product is locked as "defective" and the current plane coordinates and time value of the product are output.

4. The intelligent grading system for fried prepared foods according to claim 3, characterized in that, The edge computing device outputs the planar coordinates and then sends them to the task queue of the Arduino development board. The Arduino development board controls the Delta robot (11) to move to the grasping position through task scheduling based on the time value of the locked "defective" category received, and controls the gas solenoid valve to open so that the pneumatic flexible mechanical claw (18) performs the grasping action.

5. The intelligent grading system for fried prepared foods according to claim 1, characterized in that, The DC step-down module is used to convert the voltage output from the 24V DC power supply into the operating voltage adapted to the Arduino development board.

6. The intelligent grading system for fried prepared foods according to claim 1, characterized in that, The stainless steel mesh conveyor belt (7) is located below the end outlet of the frying and air cooling system, and the shooting area of ​​the industrial camera (10) covers the section of the stainless steel mesh conveyor belt (7) to be inspected.

7. The intelligent grading system for fried prepared foods according to claim 1, characterized in that, The collection mechanism (6) includes a defective product collection frame (21) and a qualified product collection frame (22), which are respectively set in different graded areas at the end of the stainless steel mesh conveyor belt (7).

8. The intelligent grading system for fried prepared foods according to claim 7, characterized in that, The defective product collection box (21) and the qualified product collection box (22) are arranged side by side along the width direction of the stainless steel mesh conveyor belt (7) and respectively correspond to the gripping and placement area of ​​the Delta robot (11).

9. The intelligent grading system for fried prepared foods according to claim 1, characterized in that, The pneumatic flexible mechanical claw (18) has a fin-like claw structure and is used to flexibly grasp fried and prepared foods.

Images