KR102380374B1 - System for stacking container and method thereof - Google Patents

Abstract

In accordance with an embodiment, disclosed are a system and method for loading containers. The container loading system includes: a first module generating a prediction model for predicting a weight grade for a container; a second module predicating a plurality of weight grades for the container through the generated prediction model and allocating a stack to each of the plurality of weight grades based on the plurality of weight grades and a yard stack composition of a predetermined storage area, and determining one of all stacks of the storage area as a storage position for the container based on the allocated stacks and the loading status of the storage area; and a database (DB) storing the prediction model and the yard stack composition. Therefore, the present invention is capable of improving the productivity of container terminals.

Description

Container loading system and method therefor

Embodiments relate to container loading systems and methods.

With the development of container transport, efficient management of scarce storage space resources in container terminals has become an important role of maritime transport hubs. Containers arrive at storage locations randomly and are stacked in the form of yard blocks. Yard cranes are required to handle containers located at the top.

However, this inefficient processing causes excessive operational delays, creating a bottleneck in the container processing flow. Therefore, in order to improve the productivity of the container terminal, the most efficient method of determining the storage location of the arrival container is required. In order to facilitate the container processing flow in the future container shipping operation, the storage location of the container should be determined so that the weight of the container loaded for each yard stack is uniform.

Embodiments provide a container loading system and method capable of determining an optimal container storage location.

The problem to be solved in the embodiment is not limited thereto, and it will be said that the purpose or effect that can be grasped from the method of solving the problem described below or the embodiment is also included.

A container loading system according to an embodiment includes: a first module for generating a predictive model for predicting a weight class for a container; predicting a plurality of weight classes for the container by using the generated prediction model and allocating a stack to each of the plurality of weight classes based on the predicted plurality of weight classes and a yard stack configuration of a predetermined storage area; a second module for determining one of all stacks of the storage area as a storage location of the container based on the allocated stack and the loading status of the storage area; And it may include a DB (Database) for storing the prediction model and the yard stack configuration.

The first module may generate a unique Gaussian mixture model (GMM)-based prediction model for each container group by aggregating the container weight for each container group.

The second module allocates a stack for each of the plurality of weight classes based on the plurality of weight classes and a yard stack configuration of a predetermined storage area, calculates a contribution of each of the plurality of weight classes, and each calculated weight A weighted average value for each stack may be calculated based on the degree of contribution, and a stack having a maximum weighted average value among all stacks in the storage area may be determined as the storage location of the container based on the calculated weighted average value.

The second module may allocate all empty slots in the stack to be equal to the configuration of the predicted plurality of weight classes based on the plurality of weight classes and the yard stack configuration.

The second module may assign a value of the mixed weight for each weight class calculated by the prediction model to each of the K weight classes, and allocate all the slots empty in the stack to the K weight classes by the value of the mixed weight.

The second module may update the yard stack configuration with the determined storage location of the container.

Container loading method according to an embodiment comprises the steps of: storing a pre-generated prediction model and a yard stack configuration for each storage area in a DB (Database); predicting a plurality of weight classes for a container using the prediction model; and allocating a stack to each of the plurality of weight classes based on the predicted plurality of weight classes and a yard stack configuration of a predetermined storage area, and based on the loading status of the allocated stacks and the storage area, all stacks in the storage area. and determining one of the stacks as the storage location of the container.

The container loading method may further include generating a unique Gaussian mixture model (GMM)-based prediction model for each container group by aggregating the container weight for each container group.

The determining may include allocating a stack for each of the plurality of weight classes based on the plurality of weight classes and a yard stack configuration of a predetermined storage area, calculating a contribution of each of the plurality of weight classes, and calculating each of the calculated weights A weighted average value for each stack may be calculated based on the degree of contribution, and a stack having a maximum weighted average value among all stacks in the storage area may be determined as the storage location of the container based on the calculated weighted average value.

The determining may allocate all empty slots in the stack based on the plurality of weight classes and the yard stack configuration to be the same as the configuration of the predicted plurality of weight classes.

The determining may include assigning a value of a mixed weight for each weight class calculated by the prediction model to each of the K weight classes, and allocating all empty slots in the stack to the K weight classes by the value of the blending weight.

The container loading method may further include updating the yard stack configuration with the determined storage location of the container.

According to the embodiment, a plurality of weight classes of the container are predicted using a pre-generated prediction model, and a stack is allocated for each of the plurality of weight classes based on the predicted plurality of weight classes and the yard stack configuration of a predetermined storage area. By determining one of the allocated stacks as the storage location of the container, the optimal container storage location can be determined.

According to the embodiment, since it is possible to suppress the occurrence of excessive operation delay by determining the optimal container storage location, it is possible to improve the productivity of the container terminal.

Various and advantageous advantages and effects of the present invention are not limited to the above, and will be more easily understood in the course of describing specific embodiments of the present invention.

1 is a view showing the configuration of a container loading system according to an embodiment of the present invention.
2 is a view showing a yard block of a container.
3 is a diagram illustrating a configuration of a yard bay of a storage area.
4 is a view showing a method of loading a container according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating a process of determining a storage location of the container shown in FIG. 4 .
6A to 6D are diagrams for explaining a stack allocation process according to weight classes.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

However, the technical spirit of the present invention is not limited to some embodiments described, but may be implemented in various different forms, and within the scope of the technical spirit of the present invention, one or more of the components between the embodiments It can be optionally combined and substituted.

In addition, terms (including technical and scientific terms) used in the embodiments of the present invention may be generally understood by those of ordinary skill in the art to which the present invention belongs, unless specifically defined and described explicitly. It may be interpreted as a meaning, and generally used terms such as terms defined in advance may be interpreted in consideration of the contextual meaning of the related art.

In addition, the terminology used in the embodiments of the present invention is for describing the embodiments and is not intended to limit the present invention.

In this specification, the singular form may also include the plural form unless otherwise specified in the phrase, and when it is described as “at least one (or more than one) of A and (and) B, C”, it is combined with A, B, and C It may include one or more of all possible combinations.

In addition, in describing the components of the embodiment of the present invention, terms such as first, second, A, B, (a), (b), etc. may be used.

These terms are only for distinguishing the component from other components, and are not limited to the essence, order, or order of the component by the term.

And, when it is described that a component is 'connected', 'coupled' or 'connected' to another component, the component is not only directly connected, coupled or connected to the other component, but also with the component It may also include a case of 'connected', 'coupled' or 'connected' due to another element between the other elements.

In addition, when it is described as being formed or disposed on “above (above) or under (below)” of each component, the top (above) or bottom (below) is one as well as when two components are in direct contact with each other. Also includes a case in which another component as described above is formed or disposed between two components. In addition, when expressed as “upper (upper) or lower (lower)”, the meaning of not only the upward direction but also the downward direction based on one component may be included.

In the embodiment, a plurality of weight classes of the container are predicted using a pre-generated prediction model, and a stack is allocated to each of the plurality of weight classes based on the predicted plurality of weight classes and the yard stack configuration of a predetermined storage area, and the We want to determine one of all stacks in the allocated storage area as the storage location of the container.

1 is a view showing the configuration of a container loading system according to an embodiment of the present invention.

Referring to FIG. 1 , the container loading system according to an embodiment of the present invention may include a DB (Database, 100 ), a first module 200 , and a second module 300 .

The DB 100 may store specification information about the container for each yard block and the yard stack configuration of the storage area. Here, the specification information may include the weight, size, type, destination, and the like of the container.

The first module 200 may generate a prediction model for predicting a plurality of weight classes for a container. The first module 200 may generate a unique Gaussian mixture model (hereinafter referred to as GMM)-based prediction model for each container group by aggregating the container weight for the container group. The first module 200 may predict a plurality of weight classes for a container using a pre-generated prediction model. Here, GMM refers to a model that represents a population as a linear superposition of subpopulations, each subpopulation following a Gaussian distribution.

The second module 300 predicts a plurality of weight classes for the container using the prediction model generated from the first module 200, and based on the predicted weight class and the yard stack configuration of the predetermined storage area, The storage location can be dynamically determined.

FIG. 2 is a diagram illustrating a yard block of a container, and FIG. 3 is a diagram illustrating a configuration of a yard bay that is a storage area.

Referring to FIG. 2 , a yard block of a container is a bay, a row, and a tier, and describes the address system of a slot in which the container is located. Here, the bay means the distance in the direction of the long edge of the rectangular parallelepiped container, and the width means the distance in the direction of the short edge of the container.

Referring to FIG. 3 , a storage area of a container may be configured by several stacks of bays. Determining the storage location means selecting one stack from the storage area. Each stack may include a plurality of slots in which containers are loaded.

Each storage area specifies an area for containers belonging to the same container group.

4 is a diagram illustrating a method of loading a container according to an embodiment of the present invention, and FIG. 5 is a diagram illustrating a process of determining a storage location of the container illustrated in FIG. 4 .

4 to 5 , the container loading system according to an embodiment of the present invention may receive specification information of a container to be loaded in a predetermined yard block (S410).

Next, the container loading system may predict a plurality of weight classes for the container by using the prediction model generated in advance based on the input specification information of the container ( S420 ).

Specifically, the container loading system of FIG. 5 predicts a plurality of weight classes for a container using a pre-generated GMM-based prediction model, and calculates the responsibility for each of the predicted plurality of weight classes. can be saved

, 공분산(covariance)

, 혼합 가중치(mixture weight)

에 의해 변수화될 수 있다. 이러한 파라미터들은 다음의 [표 1]과 같이 정의될 수 있다.GMM is the mean

, covariance

, the mixture weight

can be parameterized by These parameters can be defined as follows [Table 1].

[Table 1]

은 다음의 [수학식 1]과 같이 나타낼 수 있다.First, the standard form of the prediction result of GMM, that is, the marginal probability distribution of the container weight x for the container group g

can be expressed as the following [Equation 1].

[Equation 1]

의 주변 우도(marginal likelihood)를 최대화하는 GMM의 3개의 파라미터를 추정해야 한다. 가장 널리 사용하는 방법은 EM(expectation maximization) 알고리즘인데, EM 알고리즘은 E-step과 M-step을 반복하여 모델 파라미터를 추정한다.

We need to estimate three parameters of the GMM that maximize the marginal likelihood of . The most widely used method is the EM (expectation maximization) algorithm, which estimates model parameters by repeating E-step and M-step.

을 구할 수 있다.Given the initialized model parameters and log-likelihood estimation values, E-step uses the model parameters to generate a responsibility as shown in [Equation 2].

can be obtained

[Equation 2]

M-step can re-estimate model parameters using responsibility values.

The container loading system according to the embodiment may use [Equation 2] to obtain respective responsibility values for a plurality of weight classes, for example, weight class 1, weight class 2, weight class 3, and weight class 4 for the container. .

Next, the container loading system allocates a stack for each weight class based on a plurality of predicted weight classes and a yard stack configuration of a predetermined storage area (S430), and a storage area based on the assigned stack and the loading status of the storage area By selecting one of all stacks of , the selected stack may be determined as a storage location of the container ( S440 ).

At this time, the container loading system updates the yard stack based on the determined storage location. This updated yard stack is used when the next container comes in.

Specifically, the container loading system of FIG. 5 may determine the weight class versus stack allocation based on the yard stack configuration of the predetermined storage area and respective responsibility values for the plurality of weight classes predicted.

Parameters used in the stack allocation process can be defined as shown in [Table 2] below.

[Table 2]

의 누적 범위

는 다음의 [수학식 3]과 같이 나타낼 수 있다.First, given the indices of the GMMs sorted in ascending order by weight, mixed weights for each weight class

cumulative range of

can be expressed as the following [Equation 3].

[Equation 3]

의 누적 범위

는 다음의 [수학식 4]와 같이 나타낼 수 있다.Similarly, given the indices of the stored stacks, the mixed weights for each stack

cumulative range of

can be expressed as the following [Equation 4].

[Equation 4]

는 {(0, 0.25), (0.25, 0.5), (0.5, 0.75), (0.75, 1)}로 설정될 수 있다.[Equation 4] calculates an accumulation range for each stack based on the number of remaining slots. For example, if the storage area consists of 4 empty stacks, the cumulative range

may be set to {(0, 0.25), (0.25, 0.5), (0.5, 0.75), (0.75, 1)}.

의 인자가 지정된 누적 범위

에 있는지에 의해 결정되는 중량 등급 대 스택 할당

은 다음의 [수학식 5]와 같이 나타낼 수 있다.The composition of the weight classes in the stack is the cumulative range

Cumulative range given the argument of

Weight class versus stack allocation determined by whether

can be expressed as the following [Equation 5].

[Equation 5]

에 영향을 미치는

를 변경하기 때문에, 변수

는 연속 조정될 수 있다.dynamic container placement

affecting

Since changing the variable

can be continuously adjusted.

Next, the container loading system may load the container at the determined storage location (S450).

6A to 6D are diagrams for explaining a stack allocation process according to weight classes.

이 0.25로 설정된 경우, 저장 영역 A, B 사이에 차이가 없는 초기화된

를 보여주고 있다.6A , it is assumed that containers belonging to the same container group are allocated to two storage areas, and mixed weights for all weight classes k=1, 2, 3, 4

is set to 0.25, there is no difference between storage areas A and B

is showing

That is, weight classes k=1, 2, 3, 4 may be assigned to stacks stack1, stack2, stack3, and stack4 of each storage area.

는 {(0, 0.2), (0.2, 0.4), (0.4, 0.8), (0.8, 1)}로 갱신될 수 있다.Referring to FIG. 6B , if 3, 3, 1, and 3 containers are sequentially loaded in each stack of the storage area A, the cumulative range

can be updated as {(0, 0.2), (0.2, 0.4), (0.4, 0.8), (0.8, 1)}.

이 0.25이고, 누적 범위

는 {(0, 0.2), (0.2, 0.4), (0.4, 0.8), (0.8, 1)}이기 때문에, 모든 중량 등급 k=1, 2, 3, 4은 스택 stack1, stack2, stack3, stack4에 할당될 수 있다.Referring to Figure 6c, mixed weights for all weight classes k=1, 2, 3, 4

is 0.25, the cumulative range

is {(0, 0.2), (0.2, 0.4), (0.4, 0.8), (0.8, 1)}, so all weight classes k=1, 2, 3, 4 are stacks stack1, stack2, stack3, stack4 can be assigned to

That is, all empty slots in the stack are allocated to be the same as the configuration of the plurality of weight classes, for example, a value of the mixed weight for each weight class calculated in the GMM is given to each K (K is a positive integer) weight class, and the stack It is possible to allocate all the empty slots to K weight classes as much as the value of the mixed weight.

Next, the second module may determine a storage location of the container based on the determined weight class versus stack allocation.

에 대한 중량 등급 k의 기여도

는 다음의 [수학식 6]과 같이 나타낼 수 있다.Specifically, since multiple weight classes are assigned to each stack, the contribution of multiple weight classes must be considered.

Contribution of weight class k to

can be expressed as the following [Equation 6].

[Equation 6]

에 따라

의 가중 평균값이 최대인 스택을 선택해야 한다. 주어진 저장 영역에서 컨테이너를 위해 선택될 스택

은 다음의 [수학식 7]과 같다.leveled

Depending on the

The stack with the largest weighted average of The stack to be selected for the container in the given storage area.

is the following [Equation 7].

[Equation 7]

는

와 같이 유도된다. 마찬가지로 중량 등급 1, 2를 위한 스택의

는

와

로 유도된다.Referring to FIG. 6d , of stack stack1 for weight class weight class 1

Is

is induced as Similarly, the stacks for weight classes 1 and 2

Is

Wow

is led to

에 따른

의 가중 평균값은

로 유도되고, 스택 stack2에 대한 가중 평균값은

로 유도된다.flattened to stack1

In accordance

The weighted average of

, and the weighted average for stack stack2 is

is led to

의 가중 평균값이 최대인 스택을 컨테이너의 저장 위치로 선택할 수 있다.Based on the weighted average calculated through this process, among all stacks in a given storage area,

The stack with the maximum weighted average value of can be selected as the storage location of the container.

The term '~ unit' used in this embodiment means software or hardware components such as field-programmable gate array (FPGA) or ASIC, and '~ unit' performs certain roles. However, '-part' is not limited to software or hardware. '~' may be configured to reside on an addressable storage medium or may be configured to refresh one or more processors. Accordingly, as an example, '~' indicates components such as software components, object-oriented software components, class components, and task components, and processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. The functions provided in the components and '~ units' may be combined into a smaller number of components and '~ units' or further separated into additional components and '~ units'. In addition, components and '~ units' may be implemented to play one or more CPUs in a device or secure multimedia card.

Although the above has been described with reference to preferred embodiments of the present invention, those skilled in the art can variously modify and change the present invention within the scope without departing from the spirit and scope of the present invention as set forth in the claims below. You will understand that it can be done.

100: DB
200: first module
300: second module

Claims (12)

a first module for generating a predictive model for predicting a weight class for a container;
predicting a plurality of weight classes for the container by using the generated prediction model and allocating a stack to each of the plurality of weight classes based on the predicted plurality of weight classes and a yard stack configuration of a predetermined storage area; a second module for determining one of all stacks of the storage area as a storage location of the container based on the allocated stack and the loading status of the storage area; and
Including a DB (Database) for storing the prediction model and the yard stack configuration,
The second module is
allocating a stack to each of the plurality of weight classes based on a yard stack configuration of the plurality of weight classes and a predetermined storage area;
calculating the contribution of each of the plurality of weight classes,
Calculate a weighted average value for each stack based on the calculated contribution of each weight class,
and determining, as the storage location of the container, a stack having a maximum weighted average value among all stacks in the storage area based on the calculated weighted average value. According to claim 1,
The first module is
A container loading system that aggregates container weights for each container group to create a unique Gaussian mixture model (GMM) based predictive model for each container group. delete According to claim 1,
The second module is
allocating all empty slots in a stack to be equal to the configuration of the predicted plurality of weight classes based on the plurality of weight classes and the yard stack configuration. 5. The method of claim 4,
The second module is
For each K (K is a positive integer) weight class, a value of the mixed weight for each weight class calculated from the prediction model is given,
allocating all empty slots in the stack to the K weight classes by the value of the mixed weight. According to claim 1,
The second module is
and updating the yard stack configuration with the determined storage location of the container. Storing, by the first module, the pre-generated prediction model and the yard stack configuration for each storage area in a DB (Database);
predicting, by a second module, a plurality of weight classes for the container using the prediction model; and
the second module allocates a stack for each of the plurality of weight classes based on the predicted plurality of weight classes and a yard stack configuration of a predetermined storage area, and stores the storage area based on the loading status of the allocated stacks and the storage area determining one of all stacks in a region as the storage location of the container;
The determining step is
allocating a stack to each of the plurality of weight classes based on a yard stack configuration of the plurality of weight classes and a predetermined storage area;
calculating the contribution of each of the plurality of weight classes,
Calculate a weighted average value for each stack based on the calculated contribution of each weight class,
and determining, as a storage location of the container, a stack having a maximum weighted average value among all stacks in the storage area based on the calculated weighted average value. 8. The method of claim 7,
The method of claim 1, further comprising: the first module aggregating container weights for each container group to generate a unique Gaussian mixture model (GMM) based predictive model for each container group. delete 8. The method of claim 7,
The determining step is
allocating all empty slots in a stack to be equal to the configuration of the predicted plurality of weight classes based on the plurality of weight classes and the yard stack configuration. 11. The method of claim 10,
The determining step is
A value of the mixed weight for each weight class calculated from the prediction model is given to each of the K weight classes,
Allocating all empty slots in the stack to K weight classes by a value of a mixed weight. 8. The method of claim 7,
The method further comprising the step of the second module updating the yard stack configuration with the determined storage location of the container.

Images