CN-122006895-A - Multi-table full-flow automatic control integrated system based on AI image recognition

CN122006895ACN 122006895 ACN122006895 ACN 122006895ACN-122006895-A

Abstract

The invention belongs to the technical field of automatic control of concentrating tables, and discloses a full-flow automatic control integrated system of a plurality of tables based on AI image recognition. The full-flow automatic control integrated system for the multiple shaking tables based on the AI image recognition is adopted, the screw rod moves among the multiple shaking tables, meanwhile, the screw rod carries the cameras and the control switchboard, so that the screw rod moves accurately along the multiple shaking table arrays, full-coverage acquisition of the multiple shaking tables by a single group of cameras is realized, the acquisition, the shooting and the image recognition are carried out together, the processing is not required to be carried out after the acquisition is completed, the illumination intensity is changed according to different light rays, the AI auxiliary recognition is arranged in the control switchboard, the AI deep learning model is utilized to replace a traditional algorithm, the boundary recognition accuracy of a mine belt in a complex environment is improved, and the cyclic detection of the multiple shaking tables is realized.

Inventors

JI CUICUI
Cheng Yuanji
TONG XIONG
XIE XIAN
Liu Jiachenkang
LIU ZEFANG
Tang Zhangyi

Assignees

昆明理工大学

Dates

Publication Date: 20260512
Application Date: 20251211

Claims (10)

1. The full-flow automatic control integrated system of a plurality of shaking tables based on AI image recognition is characterized by comprising an image recognition module and a control module; the image recognition module comprises a camera shooting mechanism, and the control module comprises a flow controller, a gradient regulator and a mineral receiving plate mechanism.
2. The AI image recognition-based full-flow automatic control integrated system for a plurality of shaking tables, as set forth in claim 1, wherein the camera mechanism comprises a camera, a light supplementing device and a control switchboard; The camera is arranged on a moving platform of the screw rod through a horizontal adjusting type supporting frame, the light supplementing devices are symmetrically arranged on two sides of the camera through adjustable brackets, the control switchboard is arranged on the inner side of the moving platform, and a CSI interface of the control switchboard is connected with the camera.
3. The system of claim 2, wherein the control switchboard is a processor of a programmable data processing device, and the control main board is provided with a camera interface, a power interface, a USB interface, a network interface, an expansion interface and a video output interface.
4. The full-flow automatic control integrated system of a plurality of shaking tables based on AI image recognition as set forth in claim 2, wherein one end of the screw rod is connected with a coupler, the coupler is connected with a screw rod motor, the other end of the screw rod is fixed by a bearing seat, and the screw rod is fixed by a bearing seat between the screw rod and the coupler.
5. The full-flow automatic control integrated system of a plurality of shaking tables based on AI image recognition as set forth in claim 4, wherein the screw rod penetrates through a ball nut inside the screw rod sliding table, and the screw rod sliding table is fixedly connected with the moving platform; The whole camera shooting mechanism is arranged on the cross beam of the screw rod supporting frame in a flip-chip mode that a camera lens is downwards aligned to the table surface of the shaking table and the control switchboard is upwards to avoid ore pulp dripping.
6. The AI image recognition-based full-flow automatic control integrated system for a plurality of shaking tables, as set forth in claim 5, is characterized in that the screw rod supporting frame is fixed on the floor of a workshop through expansion bolts and is positioned on two sides of the middle of the shaking tables close to the tail end, and the screw rod supporting frame spans the plurality of shaking tables to ensure that each shaking table surface can be completely shot when the camera moves.
7. The AI image recognition-based full-flow automatic control integrated system for a plurality of shaking tables, as set forth in claim 6, is characterized in that an electric cabinet is mounted on a support column of the screw rod support frame, and comprises a motor driver and a fixed power supply; The power end of the motor driver is connected with a fixed power supply, and the input of the driving end of the motor driver extends out of the electric cabinet and is connected to the output end of the motor; the output of the driving end of the motor driver extends from the electric cabinet and is connected to the control total expansion interface; The fixed power supply output end is connected with the screw motor and the ore receiving plate rotating motor, and the fixed power supply input end is connected with a 220v power supply.
8. The full-flow automatic control integrated system of a plurality of shaking tables based on AI image recognition as set forth in claim 7, wherein the electric control box is provided with an electric supply switch and an electric switch for supplying power to the cameras, the control switchboard, the control steering engine and the screw rod motor, and a touch screen is arranged on the outer side of the box body on the electric control box to control the running speed of the shaking table motor.
9. The AI image recognition-based full-flow automatic control integrated system for a plurality of shaking tables, as set forth in claim 1, wherein the ore receiving plate mechanism comprises an ore receiving plate, a control steering engine and an ore receiving plate supporting frame; The ore receiving plate is connected with the control steering engine, and the control steering engine is connected with the ore receiving plate supporting frame which is arranged below the tail end of the shaking table, so that the ore receiving plate supporting frame and the control steering engine are ensured to be hidden under the shaking table; The water feeding tank of the cradle is provided with a water feeding end flow regulator for controlling the water flow, the middle lower part of the cradle is provided with a cradle gradient regulating knob, and the cradle gradient regulating knob is provided with a gradient regulator for regulating the gradient according to the detected mineral property so as to improve the sorting efficiency.
10. The AI-image-recognition-based multiple-shaker full-flow automatic control integrated system of any one of claims 1-9, wherein the workflow is as follows: Step S1, presetting a maximum shooting angle point of each shaking table by a control switchboard, calibrating through debugging and storing in the control switchboard; S2, the screw rod drives the camera and the control switchboard to move to a calibration point, and then the control switchboard stops, and the horizontal calibrator detects the verticality of the camera and ensures that the axis of the lens is absolutely vertical to the bed surface; Step S3, controlling the switchboard to start illumination detection, stabilizing the illumination of the bed surface in a specific interval through the light supplementing device, and starting to shoot an image by the camera after the illumination is supplemented and stabilized; S4, transmitting the image acquired by the camera into a control switchboard in real time, performing AI image processing by using a deep learning model, and identifying the boundary of the mine belt; S5, issuing a judging result to a mineral receiving plate control steering engine, and controlling the mineral receiving plate to rotate to a specified angle by the mineral receiving plate control steering engine to receive minerals; and S6, controlling the switchboard to comprehensively judge whether the vibration frequency, the water supply flow rate and the gradient of the table surface of the shaking table are in an optimal state or not by using the AI through the processed image data, and issuing commands to the gradient regulator, the water supply end flow regulator and the shaking table motor to regulate.

Description

Multi-table full-flow automatic control integrated system based on AI image recognition Technical Field The invention relates to the technical field of automatic control of concentrating tables, in particular to a full-flow automatic control integrated system of a plurality of tables based on AI image recognition. Background In the automatic development process of the mineral processing industry, the technology of identifying and controlling the ore receiving of the shaking table ore belt is iterated continuously, but the prior art still has a plurality of limitations, and the production requirements of high efficiency, accuracy and synergy are difficult to meet. The existing method and device for automatically identifying and collecting the ore belt of the cradle with the publication number of CN113304869B takes a camera fixedly arranged above the cradle as core acquisition equipment, processes continuous images of the ore belt through a gray threshold segmentation algorithm, identifies characteristics of the ore belt and determines an ore collecting area, and then drives an ore collecting plate to move for collecting ores by a cylinder to form a single cradle image acquisition-identification-ore collecting process. However, the technology has the obvious defects that a camera can only cover a single shaking table, the equipment utilization rate is low, jump motion is easy to occur when an air cylinder runs at a low speed, the positioning precision of a mineral receiving plate is only +/-2 mm, the adaptability of the traditional threshold segmentation algorithm to gray level fluctuation caused by illumination and ore pulp concentration change is poor, the boundary misjudgment rate of a mineral strip is high, and the recognition accuracy is seriously influenced. The utility model provides an automatic device and method of patrolling and examining of shaking table of concentrating under publication number CN108490960A adopts the AGV dolly as the removal carrier of patrolling and examining, and along magnetic stripe navigation driving, through on-vehicle arm adjustment camera angle collection ore deposit area image and transmission to the backstage, possess electric leakage detection and automatic function of charging simultaneously. The technical problems are also remarkable, that the AGV trolley path is limited by the magnetic strips, the turning and positioning time of a narrow workshop is long, the single-wheel inspection period exceeds 2.5min, the high-frequency detection requirement cannot be met, the mobile transmission structure is difficult to realize accurate displacement along the length direction of the cradle, the risk of missing the mine belt is caused, the system only realizes 'image acquisition-transmission', the identification and the ore receiving control are not integrated, a closed loop cannot be formed, in addition, the trolley is multiple in parts and complex in structure, the faults are easily caused by dust and ore pulp in the workshop, and the maintenance difficulty and the cost are far higher than those of the fixed transmission structure. An intelligent mineral processing method based on an image processing system with the publication number of CN116017160A is characterized in that an inspection robot is used for collecting images of a shaking table ore belt and the like, the images are transmitted to a cloud end through 5G, the images are processed through traditional algorithms such as gray scale integration and Gaussian smoothing, a mineral processing report is generated, and then parameters are manually adjusted. The technology has the key defects that the technology is only stopped in the acquisition-analysis-report stage, no ore receiving control executing mechanism is needed, manual intervention adjustment is needed, response is delayed for 3-5min, the best ore receiving time is easy to miss, the traditional algorithm has fuzzy boundary and weak color gradual change processing capability caused by the change of the ore pulp flow velocity, the ore bandwidth identification error reaches +/-8 mm, the high-purity concentrate receiving requirement is difficult to meet, the inspection robot acquires a single shaking table image according to a fixed route, a plurality of shaking table unified scheduling and linkage control logics are not needed, and the integral ore dressing index fluctuation is large. Therefore, in view of the above-mentioned drawbacks of the prior art, there is a need in the art for a full-flow automatic control integrated system for multiple shaking tables based on AI image recognition technology, which meets the efficient, accurate and collaborative production requirements of the mineral separation industry. Disclosure of Invention The invention aims to provide an AI image recognition-based full-flow automatic control integrated system for a plurality of shaking tables, which solves the problems of insufficient acquisition and transmission precision, weak image