Quality in-situ monitoring closed-loop control system and method of intelligent steamed stuffed bun forming machine

Abstract

The application discloses a quality in-situ monitoring closed-loop control system and method of an intelligent steamed stuffed bun forming machine, the system comprising a multi-view industrial camera erected above a conveying belt, a ring-shaped shadowless light source, a rotary encoder, an edge artificial intelligence processing unit, a high-speed pneumatic nozzle array and a bias adjusting servo motor connected with a forming disc. The method triggers the camera to synchronously collect the top closing and side profile images of the steamed stuffed bun through the rotary encoder, uses an edge end deep learning model to real-time extract quantitative features such as wrinkle quantity, closing eccentricity, height-diameter ratio and leakage filling color block area, and performs multi-dimensional quality evaluation according to preset threshold values; calculates the time stamp of the steamed stuffed bun reaching the nozzle array and implements pneumatic rejection when the steamed stuffed bun is determined as a defective product; and automatically adjusts the closing center of the forming disc when a consistent direction eccentricity is continuously detected. The application realizes in-situ millisecond-level detection of the forming quality of the steamed stuffed bun, automatic separation of defective products and online self-correction of the process.

Description

Technical Field

[0001] This invention relates to the field of intelligent food production technology, and in particular to a closed-loop control system and method for in-situ quality monitoring of an intelligent steamed bun forming machine. Background Technology

[0002] In the field of automated pastry production, automatic steamed bun forming machines typically use a forming disc (a pleating forming mechanism composed of multiple sealing blades) to seal and pleat the steamed buns. Due to the variable viscoelasticity of the dough, the adhesion of flour dust, and the long-term physical wear of the mechanical blades, defective products often occur during operation, such as incomplete sealing, filling leakage, eccentric (asymmetrical) pleats, or insufficient number of pleats.

[0003] Currently, existing steamed bun production lines mainly suffer from the following technical defects: 1. High reliance on manual labor and unhygienic conditions: The screening of defective products mainly relies on manual identification and sorting at the end of the production line, which is not only inefficient and labor-intensive, but also poses hygiene and safety risks due to human contact with food.

[0004] 2. Severe detection lag: Traditional industrial vision inspection often adopts the solution of uploading images to a centralized PC or cloud for offline processing. Due to the limitations of high-frequency production transmission bandwidth and computing latency, it is impossible to achieve "in-situ millisecond-level" evaluation of the forming quality of a single bun, let alone intervene in the front-end mechanical structure in real time based on the detection results.

[0005] 3. Lack of process closed-loop self-correction: When continuous molding eccentricity or filling leakage occurs, the existing equipment cannot automatically and quantitatively feed back the defect characteristics to the driving source of the molding machinery (such as servo motor), requiring the machine to be stopped and the engineers to manually adjust the cutters, resulting in production interruption and waste of raw materials. Summary of the Invention

[0006] This invention aims to solve at least one of the technical problems existing in the prior art. Therefore, this invention is proposed.

[0007] The technical solution adopted by one embodiment of the present invention to solve its technical problem is as follows: A closed-loop control system for in-situ quality monitoring of an intelligent steamed bun forming machine includes: a steamed bun forming production line architecture, and an intelligent monitoring and feedback component mounted on the production line architecture.

[0008] The steamed bun forming production line architecture includes: a forming disc mechanism for forming the steamed buns, a conveyor belt for transporting raw steamed buns forward, and a servo motor for adjusting the alignment that is physically connected to the forming disc mechanism.

[0009] The intelligent monitoring feedback component includes: In-situ image acquisition hardware module: mounted above the conveyor belt, including at least one vertically downward top high-speed industrial camera and two symmetrically tilted side-view industrial cameras facing the side of the bun, with a ring-shaped shadowless LED light source around the industrial cameras.

[0010] Conveyor belt synchronous speed measurement module: includes a rotary encoder coaxially connected to the conveyor belt drive shaft, used to collect dynamic displacement and speed signals of the conveyor belt in real time.

[0011] Edge AI processing unit: Deployed inside the steamed bun machine control cabinet, it contains a pre-trained lightweight instance segmentation and object detection neural network model with directional bounding box feature extraction capabilities.

[0012] Adaptive execution and feedback control module: includes a high-speed pneumatic nozzle array located at the end of the conveyor belt, and the control terminal of the adaptive execution and feedback control module and the deviation adjustment servo motor are electrically connected to the output pin of the edge artificial intelligence processing unit.

[0013] A closed-loop control method for in-situ quality monitoring of an intelligent steamed bun forming machine based on the above system includes the following steps: Step S1: Multi-view in-situ image trigger acquisition When the rotary encoder detects that the raw bun has moved directly below the in-situ image acquisition hardware module, the edge artificial intelligence processing unit triggers the top high-speed camera and the side view camera to simultaneously capture the top closure image and the side geometric contour image of the raw bun.

[0014] Step S2: Quantitative Extraction of AI Features and Instance Segmentation at the Edge The captured images are input into the deep learning model of the edge AI processing unit, which outputs and extracts the core geometric feature parameters of the raw steamed bun in real time, including: the three-dimensional coordinates of the ridge line of the bun's pleats, the total number of pleats N, the coordinates of the center point of the closure (X1, Y1), the coordinates of the geometric center point of the outer contour of the bun (X0, Y0), and the height-to-diameter ratio H / D of the side contour. At the same time, the area of ​​abnormal gray-scale feature blocks representing overflowing filling in the closure area is extracted. .

[0015] Step S3: Comprehensive in-situ evaluation of multi-dimensional molding quality The edge AI processing unit executes the following quantization discrimination rules based on the extracted geometric feature parameters: Calculate the eccentricity of the closing edge If Δ is greater than the preset eccentricity threshold δ1, a molding eccentricity warning will be triggered. If the total number of folds N is less than If the height-to-diameter ratio (H / D) exceeds the standard morphological range, a morphological deformity warning will be triggered. If the area of ​​abnormal grayscale feature color blocks If the value is greater than the zero threshold δ2, an alarm for defective products with filling leakage will be triggered.

[0016] Step S4: Spatiotemporal trajectory prediction and precise aerodynamic rejection of defective products If any step S3 determines that the bun is defective, the edge AI processing unit, in conjunction with the real-time speed of the rotary encoder, calculates the precise timestamp of when the defective bun reaches the high-speed pneumatic nozzle array at the end. When the time reaches At that time, the edge end outputs a control signal to drive the pneumatic solenoid valve to open, releasing a directional high-pressure airflow to blow the defective bun into the recycling tank.

[0017] Step S5: Adaptive process reverse closed-loop correction control When the edge AI processing unit detects molding eccentricity warnings M times consecutively within a preset sliding time window, and the eccentricity directions are consistent, it calculates the eccentricity vector. The processing unit converts the vector into a motor control fine-tuning pulse signal and outputs it in reverse to the alignment servo motor. The alignment servo motor automatically adjusts the physical position of the closing center of the forming disc blade to complete the online process self-correction.

[0018] The beneficial effects of this invention are: 1. In-situ real-time detection: Image processing and inference calculations are completed directly on the device through the edge artificial intelligence processing unit, without the need to upload images to the cloud or PC, avoiding transmission bandwidth and computing latency issues, and realizing millisecond-level in-situ evaluation of the forming quality of a single bun.

[0019] 2. Multi-dimensional quality comprehensive evaluation: Through multi-view image acquisition and deep learning instance segmentation, multiple geometric and grayscale feature parameters such as the number of folds, the eccentricity of the closing, the height-to-diameter ratio, and the area of ​​filling leakage can be extracted simultaneously to achieve a comprehensive and quantitative evaluation of the molding quality.

[0020] 3. Automatic rejection of defective products: By combining the real-time speed signal of the rotary encoder, the time when defective buns arrive at the rejection station can be accurately predicted, realizing precise pneumatic rejection and avoiding the hygiene hazards and inefficiency problems of manual sorting.

[0021] 4. Closed-loop self-correction of process: Through statistical analysis of continuous and consistent eccentricity, the defect characteristics are automatically and quantitatively fed back to the eccentricity servo motor, realizing online adaptive fine-tuning of the molding process, reducing production interruptions and raw material waste caused by manual adjustment due to machine downtime.

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0023] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a schematic diagram of the overall hardware architecture and in-situ monitoring installation and deployment of the intelligent steamed bun forming machine system of the present invention; Figure 2 This is a flowchart of the closed-loop monitoring and reverse adaptive correction control based on edge AI in the system of the present invention.

[0024] Explanation of key component symbols: 101. Top high-speed industrial camera; 102. Side-view industrial camera; 103. Ring-shaped shadowless LED light source; 201. Rotary encoder; 301. Edge artificial intelligence processing unit; 401. High-speed pneumatic nozzle array; 402. Tilt adjustment servo motor; 501. Forming disc mechanism; 502. Raw bun. Detailed Implementation

[0025] This section will describe in detail specific embodiments of the present invention. Preferred embodiments of the present invention are shown in the accompanying drawings. The purpose of the drawings is to supplement the textual description with graphics, so that people can intuitively and vividly understand each technical feature and overall technical solution of the present invention, but they should not be construed as limiting the scope of protection of the present invention.

[0026] In the description of this invention, "multiple" means two or more; "greater than," "less than," and "exceeding" are understood to exclude the stated number; "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0027] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0028] In this invention, unless otherwise explicitly defined, the terms "setting," "installing," and "connecting" should be interpreted broadly. For example, they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to a fixed connection, a detachable connection, or an integrally formed connection; they can refer to a mechanical connection; they can refer to the internal connection of two components or the interaction between two components. Those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0029] Example Reference Figure 1 and Figure 2 This system is directly integrated into the intelligent steamed bun forming machine. A food-grade conveyor belt is arranged below the output end of the forming disc mechanism 501 of the steamed bun machine. At the in-situ detection station approximately 10 cm behind the discharge port of the forming disc mechanism 501, a gantry mounting bracket is erected. A top high-speed industrial camera 101 is vertically mounted downwards in the center of the bracket, and two side-view industrial cameras 102 are mounted at a 45-degree angle on both sides of the bracket, aligned with the center of the conveyor belt. A ring-shaped shadowless LED light source 103 is mounted below the cameras. An incremental rotary encoder 201 is rigidly connected to the driven shaft of the conveyor belt.

[0030] When the forming disc mechanism 501 finishes cutting, a raw bun 502 falls onto the conveyor belt and is conveyed forward. The rotary encoder 201 continuously sends pulse counts to the edge artificial intelligence processing unit 301 inside the control cabinet. When the pulse count reaches a set value, the edge artificial intelligence processing unit 301 triggers the top camera 101 and the side camera 102 to simultaneously capture a set of images of the raw bun 502.

[0031] The NPU chip built into the edge AI processing unit 301 completes forward inference of the deep learning model within 12 milliseconds after receiving an image. For example... Figure 2 As shown in the process, the model first extracts the edge ridges of the bun folds in the top image and counts the total number of folds N. Then, the model segments out the small circle formed by the bun's closure and glued together using a mask, calculates its centroid coordinates as the closure center point (X1, Y1), and simultaneously segments out the outer circle outline of the entire bun and calculates its geometric physical center point (X0, Y0).

[0032] If the raw steamed bun 502 experiences uneven force due to wear on the blades of the forming disc during shaping, its sealing center point (X1, Y1) will deviate from the physical center point (X0, Y0). The processing unit 301 calculates the Euclidean distance Δ between the two. When this distance is greater than 3 mm, the system marks the steamed bun as a "defective product with eccentric forming" in its internal register. Additionally, if meat filling or vegetable juice spills out at the sealing point, the segmentation network will extract the area S of the abnormal color block. pie If the area is greater than 5 square millimeters, it will be marked as "severe filling leakage defect".

[0033] Once a bun is marked as having any of the aforementioned defects, the processing unit 301 accurately calculates the physical time required for the defective bun to move to the high-speed pneumatic nozzle array 401 located at the end of the conveyor belt, based on the current linear speed of the conveyor belt transmitted in real time by the rotary encoder 201. At the instant this time is reached, the GPIO pin of the processing unit 301 sends a control level to the pneumatic nozzle array 401, the solenoid valve opens, and a directional strong airflow is released to blow the defective bun away from the main conveyor belt and into the recycling trough.

[0034] To achieve fully automated adaptive closed-loop correction, when the edge AI processing unit 301 detects through timing sliding window analysis that five consecutive raw buns have triggered a "forming eccentricity warning" with consistent eccentricity directions, it indicates that the physical center of the forming disc mechanism 501 has physically drifted. At this point, the edge AI processing unit 301 converts the average distance of the continuous eccentricity into a physical adjustment amount for the forming disc and sends a position control command to the built-in alignment servo motor 402 of the bun machine via the industrial bus. This drives the alignment servo motor 402 to rotate forward by a specific angle, causing the mechanical centering mechanism of the forming disc to fine-tune, ensuring that the eccentricity Δ of subsequently produced buns falls back within the safe threshold. The system returns to normal production, achieving in-situ AI intelligent closed-loop control on the intelligent bun forming machine production line.

[0035] Of course, the present invention is not limited to the above-described embodiments. Those skilled in the art can make equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications and substitutions are included within the scope defined by the claims of this application.

Claims

1. A closed-loop control system for in-situ quality monitoring of an intelligent steamed bun forming machine, characterized in that, include: A steamed bun forming production line architecture, and an intelligent monitoring and feedback component mounted on the production line architecture; The steamed bun forming production line architecture includes: a forming disc mechanism (501) for forming the steamed buns, a conveyor belt for transporting raw steamed buns (502) forward, and a servo motor (402) for adjusting the alignment, which is physically connected to the forming disc mechanism (501). The intelligent monitoring feedback component includes: The in-situ image acquisition hardware module is mounted above the conveyor belt and includes at least one vertically downward-facing top high-speed industrial camera (101) and two symmetrically tilted side-view industrial cameras (102) facing the side of the bun. The industrial cameras are surrounded by a ring-shaped shadowless LED light source (103). The conveyor belt synchronous speed measurement module includes a rotary encoder (201) coaxially connected to the conveyor belt drive shaft, used to collect the dynamic displacement and speed signals of the conveyor belt in real time. The edge artificial intelligence processing unit (301) is deployed in the control cabinet of the steamed bun machine. It contains a trained lightweight instance segmentation and target detection neural network model with directional bounding box feature extraction capability. The adaptive execution and feedback control module includes a high-speed pneumatic nozzle array (401) disposed at the end of the conveyor belt. The control terminal of the adaptive execution and feedback control module and the deviation servo motor (402) are electrically connected to the output pin of the edge artificial intelligence processing unit (301).

2. As described in claim 1, characterized in that: The edge artificial intelligence processing unit (301) is an edge processing device with a built-in NPU chip.

3. As described in claim 1, characterized in that: The in-situ image acquisition hardware module is mounted on the in-situ detection station of the conveyor belt via a gantry mounting bracket. The top high-speed industrial camera (101) is vertically downward and aligned with the center of the conveyor belt. The two side-view industrial cameras (102) are symmetrically arranged on both sides of the conveyor belt with their lenses aligned with the center of the conveyor belt.

4. A closed-loop control method for in-situ quality monitoring of an intelligent steamed bun forming machine, characterized in that, The in-situ quality monitoring closed-loop control system for the intelligent steamed bun forming machine according to any one of claims 1 to 3 includes the following steps: Step S1: When the rotary encoder (201) detects that the raw bun (502) has moved to the underside of the in-situ image acquisition hardware module, the edge artificial intelligence processing unit (301) triggers the top high-speed camera and the side view camera to simultaneously capture the top closing image and the side geometric contour image of the raw bun (502). Step S2: Input the captured image into the deep learning model of the edge artificial intelligence processing unit (301), and output and extract the core geometric feature parameters of the raw bun (502) in real time, including: the three-dimensional coordinates of the ridge line of the bun pleats, the total number of pleats N, the coordinates of the center point of the closure (X1, Y1), the coordinates of the geometric center point of the outer contour of the bun (X0, Y0), and the height-to-diameter ratio H / D of the side contour. At the same time, extract the area of ​​abnormal gray-scale feature blocks representing overflowing filling in the closure area. ; Step S3: The edge AI processing unit (301) performs the following quantization discrimination rule based on the extracted geometric feature parameters: Calculate the eccentricity of the closing edge If Δ is greater than the preset eccentricity threshold δ1, a molding eccentricity warning will be triggered. If the total number of folds N is less than If the height-to-diameter ratio (H / D) exceeds the standard morphological range, a morphological deformity warning will be triggered. If the area of ​​abnormal grayscale feature color blocks If the value is greater than the zero threshold δ2, an alarm for defective products with leakage will be triggered. Step S4: If any item in step S3 determines that the bun is unqualified, the edge artificial intelligence processing unit (301) calculates the precise timestamp of the defective bun reaching the end high-speed pneumatic nozzle array (401) by combining the real-time speed of the rotary encoder (201). When the time reaches At that time, the edge end outputs a control signal to drive the pneumatic solenoid valve to open, releasing a directional high-pressure airflow to blow the defective bun into the recycling tank; Step S5: When the edge AI processing unit (301) detects molding eccentricity warning M times consecutively within a preset sliding time window and the eccentricity direction is consistent, the eccentricity vector is calculated. The processing unit converts the vector into a motor control fine-tuning pulse signal and outputs it in reverse to the adjustment servo motor (402), which automatically adjusts the physical position of the closing center of the forming disc closing blade.

5. The in-situ quality monitoring closed-loop control method for the intelligent steamed bun forming machine according to claim 4, characterized in that: In step S2, the deep learning model extracts the edge ridges of the bun folds in the top image to determine the total number of folds N. It then segments the centroid coordinates of the closing area using a mask as the closing center point (X1, Y1) and simultaneously segments the outer circle outline of the bun and calculates its geometric physical center point (X0, Y0).

6. The in-situ quality monitoring closed-loop control method for the intelligent steamed bun forming machine according to claim 4, characterized in that: In step S3, the eccentricity threshold δ1 ranges from 2 mm to 5 mm.

7. The in-situ quality monitoring closed-loop control method for the intelligent steamed bun forming machine according to claim 4, characterized in that: In step S3, the zero threshold δ2 ranges from 3 square millimeters to 8 square millimeters.

8. The in-situ quality monitoring closed-loop control method for the intelligent steamed bun forming machine according to claim 4, characterized in that: In step S5, the number of consecutive detections M within the sliding time window ranges from 3 to 10.

9. The in-situ quality monitoring closed-loop control method for the intelligent steamed bun forming machine according to claim 4, characterized in that: In step S5, the edge artificial intelligence processing unit (301) converts the average distance of continuous eccentricity into the physical adjustment amount of the forming disk, and sends a position control command to the eccentricity servo motor (402) through the industrial bus to drive the eccentricity servo motor (402) to rotate a specific angle, thereby driving the mechanical centering mechanism of the forming disk to make fine adjustments.

Images