Intelligent early warning method based on automated stereoscopic warehouse

Abstract

An intelligent early warning method based on an automated stereoscopic warehouse first collects all data in the management system of the automated stereoscopic warehouse, then designs an action library according to the devices configured for the normal operation of the automated stereoscopic warehouse and the collected data, and pre-processes the collected pictures of the normally operating devices of the automated stereoscopic warehouse to construct a picture environment library; a learning model is trained to select appropriate operating actions from the action library, based on a PID algorithm, according to the characteristics of the automated stereoscopic warehouse, based on the operating control algorithm and the constructed picture environment library, combined with the early warning decision-making strategy of the automated stereoscopic warehouse, the early warning learning model is constructed and trained; finally, the early warning learning model is connected to the management system of the automated stereoscopic warehouse, the real-time received pictures of the devices operating in the automated stereoscopic warehouse are processed, and safety early warning is issued in time, so as to ensure the safe production of the automated stereoscopic warehouse; at the same time, the possible risks are investigated, and the early warning effect is further improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent decision-making technology, and in particular to an intelligent early warning method based on an automated three-dimensional warehouse. Background Technology

[0002] With the development of artificial intelligence and integrated sensor technology, automated storage and retrieval systems (AS / RS) for single-roll material storage have emerged. These systems are controlled by a management system that manages the conveying system, packaging system, and barcode scanners. The conveying system includes a roll conveyor, a tray conveyor, a handling vehicle, an inbound conveyor, and a stacker crane located on the same conveyor line. It also includes an outbound conveyor on a separate conveyor line connected to the stacker crane. The packaging system includes a packaging line located at the front end of the roll conveyor and a barcode labeling system used in conjunction with the packaging line. The barcode scanners are used in conjunction with the conveying system. The AS / RS itself is the carrier for transporting materials. The normal operation of an AS / RS involves all related aspects and operator steps. Therefore, studying the operational status of AS / RS and its various systems, and monitoring, diagnosing, and isolating faults, is crucial for ensuring the safe operation of AS / RS. Summary of the Invention

[0003] The technical problem solved by this invention is to provide an intelligent early warning method based on an automated three-dimensional warehouse, so as to solve the problems in the background art mentioned above.

[0004] The technical problem solved by this invention is achieved by the following technical solution:

[0005] An intelligent early warning method based on automated storage and retrieval systems (AS / RS) comprises the following steps:

[0006] 1) Data Collection

[0007] The data acquisition module uses bus technology to collect all data from the automated storage and retrieval system management system. To ensure the security and reliability of the data acquisition process, the data acquisition module adopts a unidirectional flow method, which will not interfere with the automated storage and retrieval system management system.

[0008] 2) Real-time storage

[0009] The data acquisition module stores the acquired data in real time.

[0010] 3) Design an action library

[0011] The motion library is designed based on the equipment and data collected required for the normal operation of the automated storage and retrieval system (AS / RS). The equipment that enables the normal operation of the AS / RS includes a management system, a conveying system, a packaging system, and a barcode scanner. The conveying system includes a roll conveyor, a tray conveyor, a handling vehicle, an inbound conveyor, a stacker crane, and an outbound conveyor on another conveyor line located on the same conveyor line as the stacker crane. The packaging system includes a packaging line located at the front end of the roll conveyor and a labeling system used in conjunction with the packaging line.

[0012] The equipment that constitutes the normal operation of the automated storage and retrieval system is broken down according to the operating status of individual equipment and numbered sequentially to generate an action library. The action library includes basic operating actions: ① uniform linear motion; ② intermittent linear motion; ③ minimum deceleration linear motion; ④ maximum acceleration linear motion; ⑤ maximum climbing motion; ⑥ maximum diving motion.

[0013] 4) Image Acquisition

[0014] Images are captured from the normally operating automated storage and retrieval system (AS / RS) equipment, and then the captured images are preprocessed to build an image environment library. The preprocessing includes normalizing the pixels and size of the images.

[0015] 5) Construct an early warning training learning model

[0016] The training learning model selects appropriate running actions from the action library in step 3), converts the conveying manipulation quantities obtained from the automated storage and retrieval system into conveying positions based on the PID algorithm, and then, based on the characteristics of the automated storage and retrieval system, and combined with the automated storage and retrieval system early warning decision strategy, constructs an early warning training learning model by learning and training the automated storage and retrieval system early warning decision.

[0017] The early warning training and learning model includes neural network structure design and selection of training and learning algorithms;

[0018] The training and learning model adopts the PPO deep reinforcement learning method, and the neural network structure design includes the design of the automated storage and retrieval system early warning decision network structure and the evaluation of the automated storage and retrieval system early warning decision network structure.

[0019] 6) Training the early warning training model

[0020] Various images of different devices in operation are input into the early warning training and learning model to train the model, and the training results are visualized until the early warning training and learning model meets the requirements for the formulation of early warning decisions for automated three-dimensional warehouses.

[0021] 7) Networked early warning

[0022] The early warning training model is integrated into the management system of the automated storage and retrieval system. The early warning training model receives real-time images of equipment operation taken in the automated storage and retrieval system, processes the images, and issues a safety warning in a timely manner through the management system when it detects dangerous or abnormal information in the images.

[0023] In this invention, in step 5), the automated storage and retrieval system (AS / RS) early warning decision is set with an upper limit and a lower limit for the operation of the AS / RS equipment.

[0024] In this invention, in step 5), the automated storage and retrieval system (AS / RS) early warning decision establishes a one-to-one correspondence between the AS / RS operation actions and the image environment inventory, which is used to determine the AS / RS operating status.

[0025] In this invention, in step 7), the equipment operation images of the automated storage and retrieval system are captured using streaming media or a camera.

[0026] In this invention, in step 7), the early warning training learning model pre-stores historical data of automated storage and retrieval systems (AS / RS) equipment. The historical data of AS / RS equipment is periodically compared with the existing equipment data to investigate potential risks and further improve the early warning effect.

[0027] Beneficial effects: This invention integrates the constructed early warning training learning model into the management system of the automated storage and retrieval system, enabling the monitoring, fault diagnosis, and fault isolation of the automated storage and retrieval system and its various systems. When dangerous or abnormal information is detected in the operation images, the management system issues a timely safety warning, thereby ensuring the safe production of the automated storage and retrieval system. At the same time, it investigates potential risks, further improving the early warning effect. Detailed Implementation

[0028] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below with reference to specific embodiments.

[0029] An intelligent early warning method based on automated storage and retrieval systems (AS / RS) comprises the following steps:

[0030] 1) Data Collection

[0031] The data acquisition module uses bus technology to collect all data from the automated storage and retrieval system management system. To ensure the security and reliability of the data acquisition process, the data acquisition module adopts a unidirectional flow method, which will not interfere with the automated storage and retrieval system management system.

[0032] 2) Real-time storage

[0033] The data acquisition module stores the acquired data in real time.

[0034] 3) Design an action library

[0035] The motion library is designed based on the equipment and data collected required for the normal operation of the automated storage and retrieval system (AS / RS). The equipment that enables the normal operation of the AS / RS includes a management system, a conveying system, a packaging system, and a barcode scanner. The conveying system includes a roll conveyor, a tray conveyor, a handling vehicle, an inbound conveyor, a stacker crane, and an outbound conveyor on another conveyor line located on the same conveyor line as the stacker crane. The packaging system includes a packaging line located at the front end of the roll conveyor and a labeling system used in conjunction with the packaging line.

[0036] The equipment that constitutes the normal operation of the automated storage and retrieval system is broken down according to the operating status of individual equipment and numbered sequentially to generate an action library. The action library includes basic operating actions: ① uniform linear motion; ② intermittent linear motion; ③ minimum deceleration linear motion; ④ maximum acceleration linear motion; ⑤ maximum climbing motion; ⑥ maximum diving motion.

[0037] 4) Image Acquisition

[0038] Images are captured from the normally operating automated storage and retrieval system (AS / RS) equipment, and then the captured images are preprocessed to build an image environment library. The preprocessing includes normalizing the pixels and size of the images.

[0039] 5) Construct an early warning training learning model

[0040] The training learning model selects appropriate running actions from the action library in step 3). Based on the PID algorithm, linear speed control is based on the difference between the tangential overload command and the actual tangential overload of the conveying system, and pitch control is based on the difference between the normal overload command and the actual normal overload of the conveying system. The conveying manipulation quantities obtained from the automated storage and retrieval system are converted into conveying positions. Then, based on the characteristics of the automated storage and retrieval system, and based on the running control algorithm and the constructed image environment library, combined with the automated storage and retrieval system early warning decision strategy, an early warning training learning model is constructed by learning and training the automated storage and retrieval system early warning decision.

[0041] The early warning training and learning model includes neural network structure design and selection of training and learning algorithms;

[0042] The training and learning model adopts the PPO deep reinforcement learning method, and the neural network structure design includes the design of the automated storage and retrieval system early warning decision network structure and the evaluation of the automated storage and retrieval system early warning decision network structure.

[0043] 6) Training the early warning training model

[0044] Various images of different devices in operation are input into the early warning training and learning model to train the model, and the training results are visualized until the early warning training and learning model meets the requirements for the formulation of early warning decisions for automated three-dimensional warehouses.

[0045] 7) Networked early warning

[0046] The early warning training model is integrated into the management system of the automated storage and retrieval system. The early warning training model receives real-time images of equipment operation taken in the automated storage and retrieval system, processes the images, and issues a safety warning in a timely manner through the management system when it detects dangerous or abnormal information in the images.

[0047] In this embodiment, in step 5), the automated storage and retrieval system (AS / RS) early warning decision is set with an upper limit and a lower limit for the operation of the AS / RS equipment.

[0048] In this embodiment, in step 5), the automated storage and retrieval system (AS / RS) early warning decision establishes a one-to-one correspondence between the AS / RS operation actions and the image environment inventory, which is used to determine the AS / RS operating status.

[0049] In this embodiment, in step 7), the equipment operation images of the automated storage and retrieval system are captured using streaming media or a camera.

[0050] In this embodiment, in step 7), the early warning training learning model pre-stores historical data of automated storage and retrieval systems (AS / RS) equipment. The historical data of AS / RS equipment is periodically compared with the existing equipment data to investigate potential risks and further improve the early warning effect.

Claims

1. An intelligent early warning method based on an automated storage and retrieval system (AS / RS), characterized in that, The specific steps are as follows: 1) Data Collection The data acquisition module uses a unidirectional flow method to collect all data from the automated warehouse management system; 2) Real-time storage The data acquisition module stores the acquired data in real time. 3) Design an action library Based on the equipment and data collected required for the normal operation of an automated storage and retrieval system (AS / RS), the equipment is broken down into individual operating states and numbered sequentially to generate an action library. The action library includes basic operating actions: ① uniform linear motion; ② intermittent linear motion; ③ minimum deceleration linear motion; ④ maximum acceleration linear motion; ⑤ maximum climbing motion; ⑥ maximum diving motion. 4) Image Acquisition Images are captured from the normally operating automated storage and retrieval system (AS / RS) equipment, and then the captured images are preprocessed to build an image environment library. The preprocessing includes normalizing the pixels and size of the images. 5) Construct an early warning training learning model The training learning model selects appropriate running actions from the action library in step 3), converts the conveying manipulation quantities obtained from the automated storage and retrieval system into conveying positions based on the PID algorithm, and then, based on the characteristics of the automated storage and retrieval system, and combined with the automated storage and retrieval system early warning decision strategy, constructs an early warning training learning model by learning and training the automated storage and retrieval system early warning decision. The early warning training and learning model includes neural network structure design and selection of training and learning algorithms. Neural network structure design includes the design of the automated three-dimensional warehouse early warning decision network structure and the design of the evaluation automated three-dimensional warehouse early warning decision network structure. 6) Training the early warning training model Images of various devices in operation are input into the early warning training and learning model to train the model, and the training results are visualized until the early warning training and learning model meets the requirements for the formulation of early warning decisions in the automated three-dimensional warehouse. 7) Networked early warning The early warning training model is integrated into the management system of the automated storage and retrieval system. The early warning training model receives real-time images of equipment operation taken in the automated storage and retrieval system, processes the images, and issues a safety warning in a timely manner through the management system when it detects dangerous or abnormal information in the images.

2. The intelligent early warning method based on an automated storage and retrieval system (AS / RS) according to claim 1, characterized in that, In step 5), the automated storage and retrieval system (AS / RS) early warning decision-making process sets the upper and lower limits of the AS / RS equipment operation actions, and there is a one-to-one correspondence between the AS / RS operation actions and the image environment inventory.

3. The intelligent early warning method based on an automated storage and retrieval system (AS / RS) according to claim 1, characterized in that, In step 7), the images of the automated storage and retrieval system (AS / RS) equipment in operation are captured using streaming media or cameras.

4. The intelligent early warning method based on an automated storage and retrieval system (AS / RS) according to claim 1, characterized in that, In step 7), the early warning training learning model contains historical data of automated storage and retrieval system equipment, and the historical data of automated storage and retrieval system equipment is compared with the existing equipment data periodically.