CN-121981632-A - Container dispatching method, electronic device, storage medium and program product

CN121981632ACN 121981632 ACN121981632 ACN 121981632ACN-121981632-A

Abstract

The application provides a container dispatching method, electronic equipment, a storage medium and a program product. The method comprises the steps of determining a target container and a storage yard where the target container is located based on a box fetching instruction. Inputting configuration information corresponding to a storage yard into a container handling model, so that the container handling model simulates a process of taking out all containers in the storage yard according to a priority order, selecting a handling path with the minimum actual repositioning times from a plurality of handling paths corresponding to all containers obtained through simulation, and outputting the handling path as a target handling path, wherein in the simulation process, the container handling model simulates the state of the storage yard after each container handling action, predicts the repositioning times required for taking out all the containers from the plurality of states obtained after updating, and keeps one or more states with the minimum repositioning times to enter the next simulation. The technical scheme shortens the calculation time of the model.

Inventors

LV HAN
DONG LIN
SUN YAJING
JIAO BO
ZHANG GUANTING
Tian Xidan

Assignees

上海振华重工集团科技有限公司
上海振华重工(集团)股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260120

Claims (10)

1. A method for shipping a container, comprising: Receiving a box receiving instruction; Determining a target container and a storage yard in which the target container is positioned based on the container fetching instruction, and acquiring configuration information corresponding to the storage yard, wherein the configuration information comprises the specification of the storage yard and the position and priority of each container in the storage yard; Inputting the configuration information corresponding to the storage yard into a container handling model, so that the container handling model simulates a process of taking out all containers in the storage yard according to a priority order, and selecting a handling path with the minimum actual repositioning times as a target handling path to output according to a plurality of handling paths corresponding to all containers taken out obtained through simulation, wherein in the simulation process, the container handling model simulates the state of the storage yard after each container handling action, predicts the repositioning times required for taking out all the containers from the various states obtained after updating, and keeps one or more states with the minimum repositioning times to enter the next simulation, and the repositioning times represent the total times of taking out all the containers in the storage yard according to the priority order and needing to carry the containers; and carrying the corresponding container in the storage yard based on the target carrying path so that the target container can be taken out.
2. The method of claim 1, wherein the container handling model renumbers the priorities corresponding to the remaining containers in the yard if one container in the yard is taken out during the simulation.
3. The method for dispatching containers according to claim 1, wherein the container handling model employs a plurality of long-short-term memory network units, and the container handling model adjusts the number of the long-short-term memory network units according to the specification of the yard, wherein one long-short-term memory network unit is used for capturing the vertical dependency relationship between containers located in the same column in the yard.
4. The method for shipping containers according to claim 1, wherein the training process of the container handling model comprises: Constructing an initial container handling model; Acquiring a storage yard set for model training, wherein the storage yard set comprises a plurality of storage yards and configuration information corresponding to each storage yard; Determining a lower limit value of repositioning times corresponding to each storage yard; Pre-training network parameters in the initial container handling model based on each storage yard and the corresponding lower limit value of the repositioning times, wherein the network parameters are used for estimating the corresponding repositioning times according to the state of the storage yard; Selecting a storage yard for reinforcement training from the storage yard set aiming at the initial container handling model after pre-training, and performing reinforcement training on the initial container handling model in an actual simulation mode based on the selected storage yard so as to optimize the network parameters and obtain the container handling model.
5. The method for dispatching containers as set forth in claim 4, wherein said determining the lower limit value of the number of relocations corresponding to each of said yards comprises: counting the sum of the number of blocking containers corresponding to each container in the storage yard, wherein the number of blocking containers represents the number of containers positioned on the upper layer of the containers; and taking the sum of the number of the blocking containers as the lower limit value of the repositioning times.
6. The method of shipping containers as set forth in claim 4, wherein said training the initial container handling model based on the selected yard in a manner of actual simulation to optimize the network parameters includes: Simulating a process of taking out all containers in the storage yard according to a priority order, and estimating the repositioning times of the storage yard by adopting the network parameters in the simulation process to obtain the estimated repositioning times corresponding to the storage yard in different states; According to the simulation process, obtaining the corresponding actual repositioning times of the storage yard in different states in a backtracking mode; And performing reinforcement training on the initial container handling model based on the actual repositioning times and the estimated repositioning times corresponding to the storage yard in different states so as to optimize the network parameters.
7. The method for shipping containers as set forth in claim 4, wherein said pre-training network parameters in said initial container handling model based on each of said yards and corresponding lower threshold for said number of relocations comprises: The initial container handling model predicts the repositioning times of the storage yard based on the network parameters to obtain the predicted repositioning times corresponding to the storage yard; Based on the lower limit value of the repositioning times and the estimated repositioning times corresponding to each storage yard, taking the mean square error between the lower limit value of the repositioning times and the estimated repositioning times as a target, and pre-training network parameters in the initial container handling model.
8. An electronic device comprising a processor and a memory, wherein the memory stores at least one instruction or at least one program, the at least one instruction or the at least one program loaded and executed by the processor to implement the container dispatch method of any one of claims 1-7.
9. A computer readable storage medium having stored therein at least one instruction or at least one program loaded and executed by a processor to implement the container dispatching method of any one of claims 1-7.
10. A computer program product comprising a computer program or instructions which, when executed by a processor, implements the container dispatching method of any one of claims 1 to 7.

Description

Container dispatching method, electronic device, storage medium and program product Technical Field The application relates to the field of automated wharfs, in particular to a container dispatching method, electronic equipment, storage medium and program product. Background With the continued growth of the global shipping industry, container ports occupy a critical location in global trade networks and sustainable development. In recent years, the number of containers handled in ports has also been continuously rising due to the increasing global sea traffic. Most containers for port transportation require several days to be stored in the area of the block. In practice, since an early arriving container may be placed at the bottom of the area, there may be other containers above it that obstruct removal when it is desired to remove the container. In addition, since the external trucks are operated by different logistics companies, the order in which they reach the port is unknown to port management personnel, and thus it is difficult to plan a placement mode that can avoid the above blocking phenomenon when placing high density containers. This means that when a container subsequently needs to be removed, it is likely that it will be blocked by other containers in the same yard, and that these blocked containers need to be relocated to other empty spaces in the yard, but that after placing these blocked containers in a new location, it is also possible to cause blocking of the containers located below it. Therefore, when the number of containers in a yard is large, it is necessary to estimate in an acceptable time which empty space the blocked container is to be specifically relocated to, so as to avoid affecting the removal of other containers as much as possible, and it is not easy to solve the container relocation problem (container relocation problem, CRP). Aiming at the problem of repositioning the large-scale container of the wharf, the traditional solution mainly depends on an accurate algorithm and a heuristic algorithm. For the exact algorithm, since the container relocation problem is a problem of non-deterministic polynomial time-difficulties (nondeterministic polynomial time, NP), that is to say its computational complexity grows exponentially with the problem size. For the accurate algorithm, although the optimal solution can be found, the calculation time is long, and the optimal solution fluctuates with the increase of the storage yard scale. While it can find solutions in a short time for heuristic algorithms, it often fails to get a high enough quality solution to support real-time decisions for large-scale container application scenarios. Therefore, some recent researches begin to try to solve the problem by using a machine learning method, but the model obtained by training on the basis of the machine learning algorithm has the problem of long calculation time in actual application, especially in the case of large storage yard scale and large number of stored containers, the calculation time of the model is longer, and the model is difficult to adapt to the requirements of real-time operation of a wharf. Disclosure of Invention In view of the above, the present application provides a container positioning method, an electronic device, a storage medium, and a program product, which can solve the problem that the existing model trained based on the machine learning method has long calculation time in practical application in the repositioning scene of a large-scale container at a wharf. Some embodiments of the application provide a container dispatching method, electronic equipment, storage medium and program product. The application is described in terms of several aspects, embodiments and advantages of which can be referenced to one another. In a first aspect, the present application provides a container dispatching method, including: receiving a box receiving instruction. And determining a target container and a storage yard in which the target container is positioned based on the container fetching instruction, and acquiring configuration information corresponding to the storage yard, wherein the configuration information comprises the specification of the storage yard and the position and priority of each container in the storage yard. Inputting configuration information corresponding to a storage yard into a container handling model, so that the container handling model simulates a process of taking out all containers in the storage yard according to a priority order, and selecting a handling path with the minimum actual repositioning times as a target handling path to output according to various handling paths obtained in the simulation process, wherein in the simulation process, the container handling model simulates the state of the storage yard after each container handling action, predicts the repositioning times required for taking out all the containers from various states obt