KR102252785B1 - Shipping liner operational manageability estimation method for resilient port operations - Google Patents

Abstract

Container ports are composed of various subsystems such as marinas, blocks, and gates, and are composed of different types of equipment that handles containers. In the past decades, many studies have been conducted on improving the operational capacity of container ports by increasing processing efficiency in each subsystem, eliminating unproductive movement, and synchronizing handshake activities between them. As long as unexpected events do not interfere with the container's workflow in the operation process, high operational performance is achieved by efficiently scheduling container handling activities for each vessel based on a standardized operation process. Therefore, it is necessary to effectively control the operation process by looking at the manageability of the container workflow. The present invention analyzes the difference in operation of the container workflow in the operation process using an event-based learning approach, and evaluates operation manageability by considering the management factors (ie, logistics, equipment, and time) suggested from the analysis result. Comparative experiments were performed on actual datasets to evaluate the operational manageability of container ports for ships of various shipping companies.

Description

Shipping liner operational manageability estimation method for resilient port operations}

The present invention relates to a shipping carrier operation manageability evaluation method, and more particularly, to a shipping carrier management management evaluation method for flexible port operation.

As container traffic continues to increase under low-speed streaming and shortened on-time requirements, the implementation of technologies for operational capabilities is triggered at major ports to meet the demands of the competitive port industry. High operational capability eliminates unproductive movement along container workflows (CWs), synchronizes handshake activities between subsystems with different types of processing equipment, and improves the operating cycle of processing equipment. And, by implementing advanced scheduling algorithms in the system. Much research has been done on the (automated) container port operation schedule and strategy to improve operational capabilities.

According to recent research studies (see non-patent documents 6, 7, 8, 10 and 16), new operational processes are introduced, processing equipment is upgraded, and decision-making subsystems (i.e. marinas, yards, blocks, gates, etc.) ), it was reported that it could be improved by integrating the operational process. Operational capabilities are generally evaluated in terms of productivity, vessel call, vessel transition time, and other indicators representing the efficiency and utilization of the subsystem (see Non-Patent Document 14). The focus on performance measurement with a focus on operational efficiency is reasonable according to Lun et al., who reported a very positive association between operating profit and throughput, and between operating cost and operating profit (see Non-Patent Document 18). Many commercial decision support packages for container ports also apply various functions to improve operational capabilities (see Non-Patent Document 15).

Although port operators have paid great attention to improving operational capabilities, the recent port disruption event has a direct impact on the economic loss of the industry adjacent to the port, and the sequential effect affects other ports in the supply chain. It is required to understand the surrounding supply chain framework (see Non-Patent Document 27). The supply chain perspective allows port operators to manage operational resilience. Liu et al. (Refer to Non-Patent Document 17) investigated only ports along the actual shipping route in order to analyze the vulnerability of maritime networks. The proposed analysis framework can contribute to creating valuable management implications for stakeholders such as shipping companies, ports, and port states to ensure the soundness of the investigated maritime supply chain. Almutairi et al. (see Non-Patent Literature 1) have multiple stakeholder groups (e.g., terminal user groups, terminal services) to prioritize planning across all scenarios (e.g., traffic congestion, recession, high operating costs, environmental mitigation). Provides a robust analysis framework for port operation that solves the level of participation of provider groups and community groups). The authors reviewed the common impact of stakeholder groups and scenarios that hindered the priority ranking of the plan.

Port supply chain management needs a different perspective from that of manufacturers. In terms of organizational structure, ports are involved in single sales (ports) and multiple purchases (shipping) service systems, and manufacturers can have the characteristics of single purchases (companies) and multi-supply manufacturing systems. In terms of market and risk structure, it is estimated that ports face specific demands from the monopoly shipping market, while producers face framework demands from comparative markets (see Non-Patent Document 21). According to Cheon et al. (Refer to Non-Patent Document 11), shipping companies prefer ports with less ambiguity in handling capacity for regular shipments, but prefer ambiguous ports for urgent shipments. In this regard, accordingly, port operators should improve the management of operational processes as well as operational capabilities in order to provide services that less disruptive to ships, taking into account the characteristics of other ships.

One concern is whether only container ports with these technologies can produce the desired results. For example, while automation is generally known as a popular technology for improving operational capabilities, Mileski et al. believe that the level of operational stability at the desired throughput is the main motive for introducing automation technology, and production capacity throughout the operational process. It was empirically analyzed that management was essential. The container port system standardizes the operational process when handling containers on subsystems and other types of equipment. Based on the standardized operation process, operation strategies and algorithms were developed and executed. Therefore, effective management of CWs for ships and consignees in the navigation process is valuable when port operators need to achieve operational performance.

[Non-patent literature]

-Non-Patent Document 1: Almutairi A, Collier ZA, Hendrickson D, Palma-Oliveira JM, Polmateer TL, Lambert JH. Stakeholder mapping and disruption scenarios with application to resilience of a container port. Reliability Engineering and System Safety 2019; 182: 219-232.

-Non-Patent Document 2: Agresti A. Categorical Data Analysis. 2nd ed. New Jersey: Wiley; 2002.

-Non-Patent Document 3: de Medeiros AKA, Weijters AJMM, van der Aalst WMP. Genetic process mining: an experimental evaluation. Data Mining and Knowledge Discovery 2007; 14(2): 245-304.

-Non-Patent Document 4: Barabadi A, Ayele YZ. Post-disaster infrastructure recovery: Prediction of recovery rate using historical data. Reliability Engineering and System Safety 2018; 169: 209-223.

-Non-Patent Document 5: Bezerra F, Wainer J, van der Aalst WMP. Anomaly detection using process mining. Enterprise, In: Halpin T, Krogstie J, Nurcan S, Proper E, Schmidt R, Soffer P, Ukor R, editors. Business Process and Information Systems Modeling, Berlin: Springer; 2009, p.149-161.

-Non-Patent Document 6: Carlo HJ, Vis IFA, Roodbergen KJ. Seaside operations in container terminals: literature overview, trends, and research directions. Flexible Services and Manufacturing Journal 2015; 27: 224-262.

-Non-Patent Document 7: Carlo HJ, Vis IFA, Roodbergen KJ. Storage yard operations in container terminals: Literature overview, trends, and research directions. European Journal of Operational Research 2014a; 235: 412-430.

-Non-Patent Document 8: Carlo HJ, Vis IFA, Roodbergen KJ. Transport operations in container terminals: Literature overview, trends, research directions and classification scheme. European Journal of Operational Research 2014b; 236: 1-13.

-Non-Patent Document 9: Gehla P, Cavalierib F, Franchin P. Approximate Bayesian network formulation for the rapid loss assessment of real-world infrastructure systems. Reliability Engineering and System Safety 2018; 177: 80-93.

-Non-Patent Document 10: Gharehgozli AH, Roy D, de Koster R. Sea container terminals: New technologies and OR models. Maritime Economics and Logistics 2015; 18: 103-140.

-Non-Patent Document 11: Cheon SH, Lee CY, Wang Y. Processing time ambiguity and port competitiveness. Production and Operations Management 2017; 26(12): 2187-2206.

-Non-Patent Document 12: Goedertiera S, de Weerdta J, Martensa D, Vanthienena J, Baesensa B. Process discovery in event logs: An application in the telecom industry. Applied Soft Computing 2011; 11: 1697-1710.

-Non-Patent Document 13: Gunther CW, van der Aalst WMP. Fuzzy mining-Adaptive process simplification based on multi-perspective metrics. In: Alonso G, Dadam P, Rosemann M, editors. Business Process Management. BPM 2007. Lecture Notes in Computer Science, vol 4714, Berlin: Springer; 2007, p. 328-343.

-Non-Patent Document 14: Ha MH, Yang Z. Comparative analysis of port performance indicators: Independency and interdependency. Transportation Research Part A: Policy and Practice 2017; 103: 264-278.

-Non-Patent Document 15: Kim KH, Lee H. Container Terminal Operation: Current Trends and Future Challenges. In: Lee CY, Meng Q, editors. Handbook of Ocean Container Transport Logistics, Cham: Springer; 2015, p.43-73.

-Non-Patent Document 16: Lehnfeld J, Knust S. Loading, unloading and premarshalling of stacks in storage areas: Survey and classification. European Journal of Operational Research 2014; 239: 297-312.

-Non-Patent Document 17: Liu H, Tian Z, Huang A, Yang Z. Analysis of vulnerabilities in maritime supply chains. Reliability Engineering and System Safety 2018; 169: 475-484.

-Non-Patent Document 18: Lun YHV, Lai KH, Cheng TCE. Managing container terminals. Shipping and logistics management, London: Springer; 2010, p. 193-204.

-Non-Patent Document 19: Mileski JP, Wang P, Zeng Q. Alignments between strategic content and process structure: the case of container terminal service process automation. Maritime Economics and Logistics 2017.

-Non-Patent Document 20: van der Aalst WMP, de Medeiros AKA, Weijters AJMM. Genetic Process Mining. In: Ciardo G, Darondeau P, editors. Applications and Theory of Petri Nets 2005. ICATPN 2005. Lecture Notes in Computer Science, vol 3536. Berlin: Springer; 2005, p. 48-69.

-Non-Patent Document 21: Ng AKY, Liu JJ. Global supply chains and trade logistics: From firm-focal to port-focal. PortFocal Logistics and Global Supply Chains, New York: Palgrave Macmillan; 2014, p. 32-57.

-Non-Patent Document 22: Reder M, Yuru?en NY, Melero JJ. Data-driven learning framework for associating weather conditions and wind turbine failures. Reliability Engineering and System Safety 2018; 169: 554-569.

-Non-Patent Document 23: van der Aalst WMP. Process mining: Discovery, conformance and enhancement of business process. Berlin: Springer; 2011.

-Non-Patent Document 24: van der Aalst WMP, Weijters T, Maruster L. Workflow mining: discovering process models from event logs. IEEE Transactions on Knowledge and Data Engineering 2004; 16(9): 1128-1142.

-Non-Patent Document 25: Weijters AJMM, van der Aalst WMP, de Medeiros AKA. Process mining with the Heuristics miner algorithm. BETA working paper series WP 166, Eindhoven University of Technology, Netherlands. 2006.

-Non-Patent Document 26: Yuan Y, Zhou S, Sievenpiper C, Mannar K, Zheng Y. Event log modeling and analysis for system failure prediction. IIE Transactions 2011; 43(9): 647-660.

-Non-Patent Document 27: Zhang Y, Lam JSL. Estimating economic losses of industry clusters due to port disruptions. Transportation Research Part A: Policy and Practice 2016; 91: 17-33.

-Non-Patent Document 28: Zhou Y, Li C, Zhou C, Luo H. Using Bayesian network for safety risk analysis of diaphragm wall deflection based on field data. Reliability Engineering and System Safety 2018; 180: 152-167.

The present invention is a method for evaluating the management of shipbuilding operations, by investigating CWs according to the event (activity processing) analysis framework, assuming the manageability of the operation process, and suggesting the components of specific CWs management and managing them through a statistical estimation model. It is intended to provide a method for evaluating the ship's operation manageability for flexible port operation that provides implications for improving the manageability composed by the elements, a computer program for executing it, and a computer-readable recording medium recording the same.

As a first aspect for solving the above problems, the present invention provides a method for evaluating a ship's operation manageability through container workflow analysis, comprising: generating a container workflow by extracting an event log from a database of a container port information system; And determining an operational difference between the planned container workflow and the actual container workflow; wherein the event log includes an ID attribute indicating a container ID, an activity attribute indicating a task to be performed, a time attribute indicating a work time, and a performing equipment. Shipping carrier operation management for flexible port operation including resource attributes indicated, and including logistics elements related to the performing work, equipment elements related to the performing equipment, and time elements related to the working time as components of the container workflow Provides a method for assessing sex.

In addition, the logistics element relates to the type and order of the performed operation, the equipment element relates to the allocation of the performing equipment, and the time element relates to the work end time. Provides an evaluation method.

In addition, the difference between the logistics elements is the number of elements in the union of the (1 × t ) task vector and ( t × t ) matrix (in the above matrix, t is the planned container workflow task and the actual container workflow task). Provides a method for evaluating shipboard operation manageability for flexible port operation, characterized in that it is calculated by calculating a context vector represented by ).

In addition, the work vector is determined according to the following Algorithm 1, and the context vector is determined according to the following Algorithm 2;

[Algorithm 1]

는 j 번째 컨테이너 워크플로우에서 d 번째 작업의 성분이고,In Algorithm 1, j is the index of the container workflow, N _j is the union of the planned container workflow task and the actual container workflow task, and d is the task index of N _j (d =1, ..., t). , A _j is the task vector of the j th container workflow, nd _d is the d th task,

Is a component of the d- th task in the j- th container workflow,

[Algorithm 2]

는 j 번째 컨테이너 워크플로우에서 d 번째 및

번째 작업 사이의 성분이고,

는 j 번째 컨테이너 워크플로우에서 d 번째 및

번째 작업 사이의 작업 전환 수이고, T _j 는 j 번째 컨테이너 워크플로우의 컨텍스트 벡터이다.In Algorithm 2,

Is the d- th and

Is the component between the first tasks,

Is the d-th and

Is the number of task transitions between th task, and T _j is the context vector of the j th container workflow.

In addition, there is provided a method for evaluating a ship's operation manageability for flexible port operation, characterized in that the operation of the work vector and the context vector is performed according to Algorithm 3 below:

[Algorithm 3]

In Algorithm 3, ∧ is and (and).

In addition, there is provided a method for evaluating the ship's operation manageability for flexible port operation, characterized in that the difference between the equipment elements is determined according to the following Algorithm 4:

[Algorithm 4]

는 j 번째 컨테이너 워크플로우의 d 번째 작업의 수행 장비이고, m _d 는 계획된 컨테이너 워크플로우의 d 번째 작업의 장비이다.In Algorithm 4, t is the number of tasks in the j th container workflow, D is the set of tasks in the planned container workflow,

Is the equipment of the d- th task of the j- th container workflow , and m _d is the equipment of the d- th task of the planned container workflow.

In addition, there is provided a method for evaluating the management of shipbuilding operations for flexible port operation, characterized in that the difference in the time factor is determined according to the following Algorithm 5:

[Algorithm 5]

In Algorithm 5, ET is the estimated execution time of the planned container workflow, ET _j is the time difference of the j th container workflow, and [ Dj = { nd ₁ , nd ₂ , · ··, nd _t }] is the j th It is the working set of the container workflow, Sd _j and Ed _j are the start and end times of the d- th task in the j- th container workflow, respectively, and LP is a positive value that makes ET _{j positive.}

In addition, the step of estimating the operational manageability of each shipping company by representing the probability of each component of the container workflow having a difference from the identified operational difference, and comparing the relative degree of association of the shipping company; characterized in that it further comprises Provides a method of evaluating the management of shipbuilding operations for flexible port operation.

In addition, the probability of each component having a difference is provided by the following equation (1), characterized in that it provides a method for evaluating the ship's management manageability for flexible port operation:

[Equation 1]

는 선사(s)의 구성요소(e)가 차이를 가질 확률이고, n _se 는 s 선사의 e 구성요소의 차이이고, n _s 는 s 선사의 구성요소 차이의 합이다.In Equation 1,

It is a variety (s) component (e) the probability that the difference, n _se is the difference between the components of the e s shipping, n _s is the sum of the components of the difference s offering.

In addition, the relative degree of association of the ships is provided by a method for evaluating the ship's operation manageability for flexible port operation, characterized in that it is represented by an odds ratio of Equation 2 below:

[Equation 2]

는 m 선사에 대한 l 선사의 구성요소(e)의 승산비이고,

는 l 선사의 구성요소(e)가 차이를 가질 확률이고,

는 m 선사의 구성요소(e)가 차이를 가질 확률이다.In Equation 2,

Is the odds ratio of the component ( e ) of l line to m line,

And l is the component (e) is likely to have a difference in variety,

Is the probability that the component ( e ) of the line m has a difference.

In addition, the relative degree of association of the ships is provided by a method for evaluating the ship's operation manageability for flexible port operation, characterized in that it is represented by an odds ratio of Equation 3 below:

[Equation 3]

는 m 선사에 대한 l 선사의 p 구성요소 대신 q 구성요소가 차이를 가질 확률에 관한 승산비이고, n _lq 는 l 선사의 q 구성요소의 차이이고, n _lp 는 l 선사의 p 구성요소의 차이이고, n _mq 는 m 선사의 q 구성요소의 차이이고, n _mp 는 m 선사의 p 구성요소의 차이이다.In Equation 3,

Is the multiplication ratio of the l p component instead of the probability that q component have a difference brings about the m variety, n _lq is a difference between the q component of the l variety, n _lp is the difference between the p component of l gives And n _mq is the difference between the q component of the m line, and n _mp is the difference between the p component of the m line.

In addition, the step of determining the difference in operation includes applying an algorithm that allows the difference of components to be equal or decreased as the elasticity criterion is lower by assigning an elasticity criterion of 1 or less. Provides a method for evaluating the management of shipbuilding operations.

In addition, as a second aspect for solving the above problem, the present invention provides a computer program for executing the method.

In addition, as a third aspect for solving the above problem, the present invention provides a computer-readable recording medium in which a computer program for executing the method is recorded.

The method for evaluating the management of shipbuilding operations for flexible port operation according to the present invention assumes the need for management of the container port's operation process in terms of three management elements: logistics, equipment, and time, and provides a process analysis framework.

In addition, the present invention provides an algorithm for grasping the operating differences of three management elements for application to process mining, which is a recent event-based learning approach.

In addition, the statistical estimation model in the present invention analyzes learning outcomes and provides implications for improving manageability composed of three management elements.

In the present invention, the actual data set collected for the regular ships of 4 ships over 6 months was used in the analysis result, and the comparison result was a different level of operation management, which is configured by management elements to control the operating differences of different ships. Show that you need sex.

1 is a schematic diagram showing a container processing process in a container port;
2 is a schematic diagram showing a process analysis framework incorporating process mining and statistical analysis proposed in the present invention;
3 is a diagram illustrating an example of analysis of differences in logistics in the present invention;
4 is a diagram illustrating an example of detecting a difference in equipment in the present invention;
5 is a flowchart showing simplified managed CWs in an experimental example of the present invention;
6 is a flowchart showing CWs identified as heuristic mining in an experimental example of the present invention;
7 is a graph showing the observation ratio of the difference in logistics for the shipping line according to the elasticity criterion in the experimental example of the present invention;
8 is a graph showing an observation ratio of differences in equipment for ships according to the elasticity criterion in the experimental example of the present invention;
9 is a graph showing the observation ratio of the time difference with respect to the line according to the elasticity criterion in the experimental example of the present invention;
10 is a probability of occurrence of a difference in logistics instead of a difference in equipment in the ship of carrier 2 with respect to the ship of carrier 1 in the elasticity criterion in the experimental example of the present invention (logistics-equipment), probability of occurrence of a difference in logistics instead of the time difference (logistics-time ) And a graph showing the probability that the equipment difference will occur (equipment-time) instead of the time difference,
11 is a probability of occurrence of a difference in logistics instead of a difference in equipment in the ship of carrier 3 with respect to the ship of carrier 1 in terms of elasticity in the experimental example of the present invention (logistics-equipment), probability of occurrence of a difference in logistics instead of the time difference (logistics-time ) And a graph showing the probability that the equipment difference will occur (equipment-time) instead of the time difference,
12 is a probability of occurrence of a difference in logistics instead of the difference in equipment in the ship of carrier 4 with respect to the ship of carrier 1 in the elasticity criterion in the experimental example of the present invention (logistics-equipment), probability of occurrence of a difference in logistics instead of the time difference (logistics-time ) And a graph showing the probability that the equipment difference will occur (equipment-time) instead of the time difference.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, when it is determined that a detailed description of a related known technology may obscure the subject matter of the present invention, a detailed description thereof will be omitted. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and similar reference numerals are assigned to similar parts throughout the specification. In addition, throughout the specification, when a certain part "includes" a certain component, it means that other components may be further included rather than excluding other components unless otherwise stated.

The method for evaluating the management of shipboard operations for flexible port operation according to the present invention is conducted by investigating container workflows (hereinafter, referred to as'CWs' or'CW') according to the event (activity processing) analysis framework. It assumes the manageability of the operation process and proposes three components of container workflow management such as logistics, equipment and time.

Accordingly, the present invention provides a method for evaluating the management of shipbuilding operations through container workflow analysis, comprising: generating a container workflow by extracting an event log from a database of a container port information system; And determining an operational difference between the planned container workflow and the actual container workflow; wherein the event log includes an ID attribute indicating a container ID, an activity attribute indicating a task to be performed, a time attribute indicating a work time, and a performing equipment. Shipping carrier operation management for flexible port operation including resource attributes indicated, and including logistics elements related to the performing work, equipment elements related to the performing equipment, and time elements related to the working time as components of the container workflow Disclosing a method for evaluating sex.

In addition, the method for evaluating the operational manageability of ships for flexible port operation according to the present invention is represented by the probability that each component of the container workflow will have a difference from the identified operating difference, and the relative degree of association between the ships is compared to each shipping company. Estimating the operational manageability of the; may further include.

Looking at CWs from the perspective of the three management components, differences in CWs over time may occur due to weak operating schedules, information differences, inappropriate processing power, and other unexpected events (see Non-Patent Document 17). In the present invention, the operational difference between the planned CWs and the actual CWs for the three management elements is identified, a comparative analysis is performed between CWs generated from ships of different ships, and operational manageability is estimated based on the estimation of statistical analysis. .

Operational differences are identified with an event-based learning approach, a customized process mining technique that integrates management elements. While process mining is basically a kind of data mining, it aims at the automatic construction of a process model that describes the behavior observed in the event log, whereas data mining is the analysis of data on previously undiscovered relationships and patterns (non-patent literature 23). Container handling event data can be organized in a way that includes the history of events that occurred during process execution, so the use of event data enables practical analysis of hidden and unexpected tasks occurring in the workplace. The determined difference result is a sample of statistical analysis when estimating the ship's operational manageability. The sample's (regional) odds ratio estimates provide implications for operational differences.

When analyzing differences, studies on system reliability apply data analysis and learning approaches to understand system failure prediction and detection. Yuan et al. (Refer to Non-Patent Document 26) analyzed the event log and event sequence of the system and proposed a statistical model suitable for predicting failure. (See Non-Patent Document 22) Reder et unsupervised k - to extract the rules in the data set by the mean clustering (unsupervised k -means clustering) developed the operation and maintenance strategy. Gehl et al. (see Non-Patent Document 9) and Zhou et al. (See Non-Patent Document 28) adopted Bayesian network approaches based on past datasets and conducted risk analysis of infrastructure systems. Barabadi and Ayele (refer to Non-Patent Document 4) developed a framework for selecting the most appropriate regression model for recovery rate analysis based on past datasets with the Bayesian network when there is little data. Unlike studies on failure detection, this study detects operational variances for CWs through a process mining approach, since variance detection is concerned with the suitability of operational processes to scheduling activities for CWs.

The present invention proposes a process for measuring the operational manageability of a port used by ships by examining CWs in the operation process. In order to achieve this object, the present invention contributes to developing a framework approach that identifies operational differences through event-based learning techniques and analyzes the implications of differences results through statistical data analysis. This is considered to be the first study to estimate the operational manageability of container ports using a learning approach.

In the following, the contents of the preliminary research on CWs and the process search algorithm in the process mining technique are summarized, the difference analysis framework is introduced in the three management elements, and the discovery of the present invention is explained in detail through actual experimental results. Draw conclusions.

Prerequisites for the operational process

This section describes CWs for the standardized operation process of container ports, and introduces a process mining approach.

(1) CWs for the container port subsystem

Container ports are facilities for transporting containers between different modes of transport, providing an activity/service package for handling and controlling the flow of containers from ships to rail or roads, and vice versa. The main object of the container port is a container, and various processing equipment such as cranes, trucks, and carriers performs container processing. The container handling process differs depending on the port characteristics and container type, but the general container handling process is as shown in Fig. 1 for discharging, loading, receiving (gate-in), and delivery (gate-out). It can be classified into 4 types. Each process consists of a number of tasks handing over unit containers called CWs.

The CWs of the handling process are as follows. When the ship arrives, Quay Cranes (QCs) unload containers from the ship according to the QC work schedule. Containers are relayed to transport vehicles, transported to storage yards, and temporarily stored in Yard Slots (YSs) by Yard Cranes (YCs). Depending on the destination, the container can be transported by other ships or delivered through a gate for transport by truck or train after inspection. The former relates to the handling process for shipment, and the latter to the process for delivery. The handling process for shipment is the reverse of the unloading process. YC lifts the container and loads it into a transport vehicle, which transports the container to the dock where QCs loads the container into the vessel. Regarding the handling process for the takeover, road trucks come into the port with containers, usually transported by YCs to the yard, which is temporarily stored in YSs.

The container port has an advanced information system (see Non-Patent Document 15) that controls container processing activities for subsystems and processing equipment, and manages data processing between them. Each activity processing generates a large number of event logs throughout the container processing process. The operational difference between planned CWs and actual CWs can be identified by event log analysis. The present invention adopts a process mining approach to learn operational differences from data.

(2) Process mining and its discovery algorithm

Process mining has been widely used to discover process models for process analysis. Among the three process mining techniques of process discovery, conformance checking and enhancement, the present invention focuses on process discovery. The basic concept of process discovery is simple: extract the event log from the system and construct an appropriate process model that describes the operation behavior in the event log (see Non-Patent Document 12). This process also consists of many structures such as join/split, join/split and loop. Considering an event log containing events for a finite number of process cases, the process miner accurately summarizes the actions in the event log and strikes the right balance between generality (allowing enough actions) and specificity (not allowing too many actions). Find a model that fits. By using a model that accurately describes the behavior of the event log, advanced analysis is possible, such as identifying performance bottlenecks or localizing paths in process models that do not conform to existing regulations. α-algorithm (refer to non-patent document 24), genetic process mining (refer to non-patent document 20), heuristic mining (refer to non-patent document 25), fuzzy mining There are various process discovery techniques for mining algorithms, such as mining) (see Non-Patent Document 13).

The α-algorithm is one of the first process discovery algorithms that can properly handle concurrency. It assumes that the event log is complete for all allowable binary sequences, and that the event log does not contain noise. Therefore, the α-algorithm is sensitive to noise and imperfections of the event log. Moreover, the original α-algorithm cannot find short loops or non-local, non-free selection structures. Nevertheless, the α-algorithm is simple and many ideas are embedded in more complex and powerful technologies. The initial approach of the process discovery algorithm was local search, and it was not possible to find non-trivial configurations such as non-free choices, invisible tasks, and redundant tasks.

Genetic process mining to support global search was developed by de Medeiros et al. (see Non-Patent Document 3). Genetic process mining has four main steps similar to normal genetic algorithms: (a) initiation, (b) selection, (c) regeneration, and (d) termination, the concept of which is to define appropriate internal representations, measure appropriate fit, and Three main concerns must be addressed: genetic operators that ensure full search space coverage. Genetic process mining defines the search space in terms of a causal matrix representing job dependence, and it is necessary to provide an initial population for a significant improvement in computation time. When the internal representation is specified and the initial population is determined, an appropriate fitness function is required. The goodness-of-fit function evaluates how well an entity (i.e., a process model) can reproduce the behavior of the event log, and then runs the algorithm, constructing a new entity for each generation using genetic operators. Oilfield process mining is flexible and robust because it can handle noise and imperfections and can be easily adopted and extended. You can give preference to a specific configuration by changing the fitness function. Despite the advantages of these genetic processes, they are not efficient for large models and logs. It can take a very long time to find a model with an acceptable fit.

Weijters et al. (see Non-Patent Document 25) developed a heuristic-based process discovery method called heuristic mining. Heuristic mining takes the frequency of events related to the relationship between the event log and the three types of activity: direct dependence, concurrency and non-direct connectivity. The basic concept of heuristic mining is that infrequent paths should not be incorporated into the model. Heuristic mining constructs a dependency graph using the frequency of dependencies between events in the event log. In this step, three threshold parameters are used: the dependent threshold, the positive observation threshold, and the relative value to the optimal threshold to differentiate between noise and low frequency behavior. The main idea of heuristic mining is to derive a causal matrix of preconditions from a dependency graph that considers only the highest dependency measure. The causal matrix distinguishes between different dependencies between activities, and consequently AND/XOR-splits/joins can be successfully identified and represented in the process model.

Fuzzy mining provides a dynamic view of the process using thematic cartography method. The main concept in thematic map cartography is generalization, the abstraction from specific details that can change over time. In fuzzy mining, two indicators of importance and correlation are confirmed. The importance index calculates the frequency and sequence of events. First of all, the more often the relationship is observed, the more important it is. Correlation indicators measure how closely events are related, which can be measured by the similarity of the data or activity names they share. In creating a process model, fuzzy mining simplifies the process based on the degree of important and correlated events. That is, an action with high importance is maintained, and when the action is highly correlated, but when the importance is low, the events are aggregated into a cluster, and when the correlation of the action is low, it is excluded. Based on these measurements, a view of the event log can be dynamically created. This allows analysts to zoom in and out on certain aspects of the process model.

Difference Analysis Framework

This section introduces a process analysis framework developed for analyzing operational differences through process discovery as shown in FIG. 2. The framework generates CWs by extracting event logs from the database of the container port information system (item (1)), and analyzes the operational differences between planned CWs and actual CWs from three perspectives: logistics, equipment, and time ((2) ) Item), and the analysis results are prepared to estimate operational manageability ((3) item)). The framework starts by extracting the event log for a set of CWs. CWs are generated in the order of the container processing operations performed, as shown in Tables 1 and 2 below. The generated CWs include the planned work history and the actual work history, which is useful for both process discovery and difference detection in sub-items.

(1) Event log and CWs generation

The proposed framework generates CWs using process mining techniques to extract event logs, and evaluates their operational differences. In general, event logs are extracted from a system database composed of cases, where cases are composed of events. In a case, an event is expressed in the form of a trace, a series of special events. Cases can have attributes, and each case has a special mandatory attribute trace, which is a definite sequence of events such that each event appears only once. The event log is a set of cases, and each event appears at most once in the entire log. When a timestamp is included in the event log, the order of tracing should take this timestamp into account (see Non-Patent Document 23).

Process mining extracts a sequence of chronological events often referred to as "history", "audit trail" or "transaction log". Each event refers to an action or activity (ie, a distinct step in the process), and the extracted event log is the number of "identification", "activity", "time" and "resource". It consists of four attributes. The "ID" and "Activity" attributes refer to cases and events, respectively. In the present invention, the four attributes of ID, activity, time, and resource refer to container ID, task to be performed, task end time, and equipment to be performed, respectively, and these are database entities of the terminal operating system.

CWs are generated by arranging the performed tasks of each container ID in ascending order of the task end time. Since the next task can only be started after completing the previous task, the task end time is the only factor in the task sequence. A method of generating CWs will be described using the exemplary event data in Table 1, which is sample event data extracted based on attributes. For container TCKU02, it contains 3 jobs: YardJobGateIn, QuayJobLoad and YardJobLoad. When sorting the jobs in ascending order by job end time, the container CW of TCKU02 becomes YardJobGateIn → YardJobLoad → QuayJobLoad as shown in Table 2 below.

(2) Analysis of differences

Individual containers are processed according to a sequence of planned or predefined workflows, and pre-allocated equipment performs activity processing through the container port's subsystem. Bezerra et al. (see Non-Patent Literature 5) include (i) rare or infrequent events, (ii) conditions outside the range of normal fluctuations, (iii) unexpected results, or (iv) deviations from normal forms or rules. The general definition of abnormal situations in the same business process is described. (i) and (ii) are related to the noise of the event data when extracting event data from the database, and (iii) and (iv) are related to the exceptional business execution of the business process. Based on the general definition of anomalous situations (especially definitions (iii) and (iv)) provided by Beczerra et al. (see Non-Patent Document 5), the present invention does not make this difference in a planned workflow with pre-allocated equipment. It is regarded as an individual container's workflow. In response to the three management factors, the framework detects three types of operational differences: logistics, equipment and time differences, the most important CW differences recognized by many field operators in this survey. Logistics differences measure the similarity between actual CWs and planned CWs according to the processing process. Equipment differences detect CWs when handling containers with equipment that has not been pre-allocated. The time difference measures the time difference in processing an activity according to CWs.

Logistics difference analysis

Logistics differences allow understanding of the possibility of operational control over CWs to facilitate the processing of activities in the planned sequence of CWs. When analyzing logistical differences, it is necessary to construct a work vector and a context vector. The task vector is used to detect the presence or absence of the same task of the actual j- th CW compared to the planned CW, while the context vector is used to detect the task sequence of the j-th CW compared to the planned CW. The task vector and context vector represent the result in the form of values from 0 to 1. If the value is close to 1, it means that the j- th CW is likely to be similar to the planned CW. The task and context vector of the j- th CW may be composed of two processes shown in Tables 3 and 4, respectively.

는 계획된 CW의 d 번째 및 d 번째 작업 사이의 요소가 되도록 한다. ∧ 및 ∨는 각각 'and' 및 'or'를 나타낸다. j 번째 CW에 대한 물류 차이는 하기 알고리즘 3으로 계산될 수 있다. 계획된 CW의 작업 및 컨텍스트 벡터도 각각 위의 두 가지 과정에 따라 계산될 수 있다.The task and context vectors are composed of matrices (1× t ) and ( t × t ), respectively. Let A and T be the work and context vectors of the planned CW, respectively,

Is the element between the d- th and d- th tasks of the planned CW. ∧ and ∨ represent ' and ' and ' or ', respectively. The logistical difference for the j- th CW can be calculated with Algorithm 3 below. The task and context vector of the planned CW can also be calculated according to the above two processes, respectively.

[Algorithm 3]

및

는 각각 작업 및 컨텍스트를 이용하여 유사도를 측정하기 위한 것이다. 유사도는 차이의 역개념이라는 점에 유의한다. 도 3은 CWs 1 및 2의 물류 차이의 예를 보여주고 있다. 계획된 CW는 순차적으로 작업 1, 2, 3, 4로 구성되며, CW1 및 CW2는 순차적으로 작업 1, 2, 3 및 작업 1, 3, 2, 4, 5로 구성된다. 계획된 CW과 비교하여 CW1 및 CW2의 워크플로우 차이는 CW1에는 작업 4가 없고, CW2에는 작업 시퀀스가 다른 추가 작업 5가 있다는 것이다.

And

Is to measure the similarity using task and context, respectively. Note that similarity is the inverse concept of difference. 3 shows an example of the difference in logistics between CWs 1 and 2. The planned CW consists of tasks 1, 2, 3, and 4 sequentially, and CW1 and CW2 sequentially consists of tasks 1, 2, and 3 and tasks 1, 3, 2, 4, and 5. The difference in the workflow of CW1 and CW2 compared to the planned CW is that CW1 has no task 4, and CW2 has an additional task 5 with a different task sequence.

The planned working set of CW and the union of CW1 and CW2 can be N ₁ ={1, 2, 3, 4} and N ₂ ={1, 2, 3, 4, 5}, respectively. Because all of the proposed CW operation include, as an element of N _1, operation of the proposed CW vector (A) compared to the N ₁ may be {1, 1,1,1}. Similarly, since there is 4 in N ₁ but not in CW1 , the working vector of CW1 (A1 ) compared to N ₁ can be {1, 1, 1, 0}. Since there is no task 5 in the planned CW, the task vector of the planned CW( A ) compared to N ₂ can be {1, 1, 1, 1, 0}. Similarly, the working vector of CW2 (A2) as compared to N _2, because all the operation is included as an element of N ₂ in the CW2 may be {1, 1, 1, 1, 1}.

은 T 및 T1의 모든 요소의 합으로 계산되기 때문에 6.833(=1 + 1/2 + 1/3 + 1 + 1/2 + 1 + 1 + 1/2 +1)이다.

은 T 및 T1에서 모두 0이 아닌 요소의 합으로 계산되기 때문에 5(=1 + 1 + 1/2 + 1/2 + 1 + 1)이다. 같은 방법으로,

및

는 각각 10.75(=1 + 1/2 + 1/2 + 1 + 1/2 + 1 + 1/2 + 1 + 1/3 + 1/4 + 1 + 1/2 + 1 + 1/2 + 1/3 + 1) 및 6.667(=1 + 1/2 + 1/2 + 1 + 1/3 + 1/3 + 1/2 + 1 + 1 + 1/2)로 계산된다. CW1 및 CW2의 물류 차이는 각각 0.74 및 0.71로 구해진다. 따라서, CW1은 작업 구조 면에서 CW2와 비교하여 계획된 CW와 더 유사할 가능성이 있다.When comparing the planned CW and CW1, the number of intersection and union elements of A and A _{1 is obtained, respectively.} That is, A ∩ A ₁ = 3 and A ∪ A ₁ = 4. Comparing CW planned in the same way with CW2, the number of intersection and union elements of A and A ₂ is obtained as A ∩ A ₂ = 4 and A ∪ A ₂ = 5. In the planned CW, 1, 2, and 3 transitions between tasks occurred at {1→2, 2→3, 3→4}, {1→3, 2→4} and {1→4}, respectively. In CW1, one and two transitions between tasks occurred at {1→2, 2→3} and {1→3}, respectively. In CW2, 1, 2, 3, and 4 transitions between tasks are respectively {1→3, 3→2, 2→4, 4→5}, {1→2, 3→4, 2→5}, {1→ It occurred in 4, 3→5} and {1→5}. By recognizing the transition between tasks in the planned CW, CW1 and CW2, context vectors ( T , T ₁ , T ₂ ) are obtained as shown in FIG. 3.

Is 6.833 (=1 + 1/2 + 1/3 + 1 + 1/2 + 1 + 1 + 1/2 +1) because it is calculated as the sum of all elements of T and T1.

Is 5(=1 + 1 + 1/2 + 1/2 + 1 + 1) because it is calculated as the sum of non-zero elements in both T and T1. In the same way,

And

Is 10.75 each (=1 + 1/2 + 1/2 + 1 + 1/2 + 1 + 1/2 + 1 + 1/3 + 1/4 + 1 + 1/2 + 1 + 1/2 + 1 /3 + 1) and 6.667 (=1 + 1/2 + 1/2 + 1 + 1/3 + 1/3 + 1/2 + 1 + 1 + 1/2). The logistical differences between CW1 and CW2 are obtained as 0.74 and 0.71, respectively. Therefore, it is possible that CW1 is more similar to the planned CW compared to CW2 in terms of the work structure.

Equipment difference analysis

는 j 번째 CW의 d 번째 작업의 수행 장비이고, m _d 는 계획된 CW의 d 번째 작업의 장비이다.When performing activity processing with equipment differences, it makes it possible to check the planned allocation between tasks (containers according to CWs) and equipment. Equipment differences are detected by Algorithm 4 below. [ t ] is the number of jobs in the j- th CW, D is the set of jobs in the planned CW,

Is the equipment for the d- th task of the j- th CW , and m _d is the equipment for the d- th task of the planned CW.

[Algorithm 4]

4 shows an example of the equipment difference between CWs 1 and 2. The implemented equipment of CW1 is the same as that of the planned CW, but CW2 has two different equipment compared to the planned CW (Vehicle 07 and RMG 04). Thus, MD1 and MD2 are calculated as 1 and 0.5, respectively. In short, CW1 is considered more similar to the planned CW than CW2 in terms of the equipment performed.

Time difference analysis

로 계산되며, ET _j 는 하기 알고리즘 5로 계산될 수 있다.

는 실제 작업 시간이 계획된 시간 도과 여부에 따라 음 또는 양의 값을 가지며, 큰 양의 상수를 더하면 항상 양의 값을 갖는다. LP는 큰 양의 값이다.

가 ET와 같으면 LP는 그 값을 유지하게 되고, 그렇지 않으면

와 ET의 차이만큼 줄어든다. 물류 및 장비 차이와 같이, 시간 차이도 0과 1 사이에 평준화한다. 값이 1에 가까울 경우, 실제 컨테이너 처리 시간은 계획된 실행 시간과 보다 유사한 것으로 간주된다는 것을 의미한다.The time difference makes it possible to check the processing efficiency when the allocated equipment performs the activity processing for the container according to the CWs. The time difference measures the difference in actual container processing time compared to the planned execution time. Let ET be the expected execution time of the planned CW, ET _j is the time difference of the j- th CW, D _j = { nd ₁ , nd ₂ , ··· , nd _t } is the working set of the j- th CW, S _dj and Let E _{dj be} the start time and end time of the d- th job in the j- th CW, respectively. The actual working time of the j-th CW is

And ET _j can be calculated with Algorithm 5 below.

Has a negative or positive value depending on whether the actual working time exceeds the planned time, and always has a positive value when a large positive constant is added. LP is a large positive value.

If is equal to ET , LP maintains that value, otherwise

It decreases by the difference between and ET. Like logistics and equipment differences, time differences are equalized between 0 and 1. If the value is close to 1, it means that the actual container processing time is considered more similar to the planned execution time.

[Algorithm 5]

(3) Evaluation of manageability through statistical analysis

The analysis result of the operating difference may be in the form of a contingency table as shown in Table 5 below. When constructing a contingency table, it is possible to estimate the statistical interdependence between variables of a category of interest (see Non-Patent Document 2). Based on the difference detection results, the (local) odds ratio will adequately estimate the ship's operational manageability across operational differences.

은

로 나타낸다. 선사 1의 장비 및 시간 차이에 대해서는 각각

및

로 나타낸다. n ₁₁ , n ₁₂ 및 n ₁₃ 은 각각 선사 1의 물류, 장비 및 시간 차이에 대해 관찰로 감지된 수를 나타내고, n ₁ 은 n ₁₁ , n ₁₂ 및 n ₁₃ 의 합을 나타낸다.

> 1.0 일 경우, 선사 1의 운영 프로세스는 다른 차이보다 물류 차이를 가질 확률이 더 높다. 선사에 대한 시간 및 장비의 차이도 마찬가지이다.Probability of shipping company 1 to have a logistical difference

silver

Represented by For shipper 1's equipment and time difference,

And

Represented by n ₁₁ , n ₁₂ and n ₁₃ represent the number of observations for the difference in logistics, equipment and time of shipping line 1, respectively, and n ₁ represents the sum of n ₁₁ , n ₁₂ and n _13.

If> 1.0, shipping company 1's operating process is more likely to have a logistical difference than any other difference. The same is true of the differences in time and equipment for ships.

로 나타낸다. 즉, 선사 1의 물류 차이가 발생할 확률은 선사 2보다

배 더 높다.

의 값이 1.0에서 멀리 있을 때, 선사 1에서 물류 차이에 대한 연관 정도가 더 큰 것을 알 수 있다.With respect to the relative associations between ships, odds ratios will provide useful implications. The relative degree of correlation between shipping companies 1 and 2 to the difference in logistics is the odds ratio.

Represented by In other words, the probability of a difference in logistics between shipping company 1 is more

Times higher.

When the value of is far from 1.0, it can be seen that the degree of correlation to the logistical difference in shipping company 1 is greater.

는 선사 1에 대한 비차이(non-discrepancy) 대신 물류 차이가 발생할 확률이 선사 2에 대하여

배로 추정된다는 것을 의미한다. 이와 유사하게,

를 통해 선사 1에 대한 장비 차이 대신 물류 차이가 발생할 확률이 선사 2에 대하여

배로 추정될 수 있다.The local odds ratio is suitable for estimating the implications between the line and the difference.

Is the probability of a logistical difference occurring instead of a non-discrepancy for shipping company 1.

It means that it is estimated to be a doubling. Similarly,

The probability of a logistical difference occurring instead of the equipment difference to the shipping company 1 is increased with respect to the shipping company 2.

It can be estimated as a ship.

Experimental example

Actual data sets of regular ships of 4 shipping companies over 6 months were collected from Korean container ports. When extracting event logs related to unloading, loading, gate-in and gate-out processes, approximately 18,620 events are obtained from the event log, and 1,233 CWs are generated from an average of 4 container ships upon arrival at the port. In the demonstration experiment, it is assumed that storage space allocation and QC work schedule are prepared, and the workflow of individual containers should be referenced. Each QC has a work schedule consisting of the order of loading (unloading) individual containers and the designated location of the ship. Note that in general, container ports in Korea allocate 3 to 4 vehicles to QC for loading and unloading operations. Remarshalling and housekeeping tasks are excluded to simplify the work process. YCs do not specify YC in container handling, because YCs mainly handle the various kinds of work required by road tracks and vehicles in yard blocks.

5 shows a simplified CW managed in a container port called a managed CW. Events include Gate In (GF), Unloading (GD), Unloading for Shift (SD), Lift Off for Unloading (DF), Bay Shift Off (YY), Shifting to Another Block or Zone (YO), and Another Block or Zone. Yard operations for shift-off (YF), lift-on for gate-out (GO), lift-on for shipment (LO), shipment (GL) and shipment for shift (SL) are included. In general, individual containers are processed according to a planned workflow with pre-allocated equipment (same as managed CWs). However, as described above, even if the workflow is accurately defined in advance, the (actually) performed workflow may be more diverse than the managed CWs. This can be easily verified by discovering CWs from event logs through process mining and comparing them with managed CWs.

6 shows the necessity of analyzing the differences between the planned and executed tasks and the causes of these differences. 6 shows CWs identified using heuristic mining. Heuristic mining has a dependency graph structure where each node represents an event. The node number is the total number of events occurring in the node (e.g., 14,202 occurrences of GD), the number of calls is the flow that occurs between two events (e.g., 4,157 occurrences between GD and DF) and reliability dependence between calls ( For example, the reliability dependence between GD and DF is 0.998). The reliability dependence between events is calculated by dividing the difference in the positive and negative incidence between two events by the positive and negative incidence plus 1 (Weijters et al. 2006). In FIG. 6, three abnormal CWs that cannot be explained in the managed CW could be detected (this can be seen as a difference). Red dotted lines indicate abnormal CWs. For example, the link between DF and SL does not appear in managed CWs, whereas it appears in identified CWs (marked by ②). The connection of DF and SL indicates that some containers have been loaded for shift by QC without pickup from the yard. These containers were loaded under QC instead of being loaded on the regular yard.

Logistics, equipment, and time differences are calculated by the above algorithms 3 to 5, and descriptive statistics are shown in Table 6 below. Note that a large positive value (LP) of 100 minutes is used to detect the time difference. The level of detection for the three types of differences is determined by the desired elasticity criterion. The higher the elasticity criterion, the higher the level of detection. For example, when applying 1.0 as the elasticity criterion, the difference between the planned and actual processes for CW is calculated as discrepancy. As can be seen in Figures 7-9, the observation of the perceived difference clearly decreases when lowering the elasticity criterion.

Table 7 below shows the probability of occurrence of differences in logistics, equipment, and time in ships of carriers 1, 2, and 3 when the elasticity criterion is set to 1.0. According to the odds ratio estimation in Table 7 below, in particular, carrier 1 has a 1.032 times higher probability of occurrence of a logistical difference than carrier 4, while the probability of occurrence of a time difference between carrier 4 is 1.053 times higher than that of carrier 1. On the other hand, shipping company 2 has a 1.105 times and 1.444 times higher probability of logistics and equipment differences than shipping company 3, respectively.

Table 8 below shows the ranks of ships based on the difference when comparing the ranks of ships with respect to the elasticity criterion, that is, relative degrees of association based on the estimated odds ratio. With the elasticity criterion of 1.0, the ranking from the highest difference to the lowest difference was found to be 2, 3, 1, and 4. The equipment and time differences are similarly explained. The differences in equipment and time were ranked by ships 2, 1, 4, 3 and ships 4, 3, 1, and 2, respectively.

When the port operator controls the CWs' operational process for ships with a high elasticity criterion (i.e., a elasticity criterion of 0.98 to 1.00), the container workflow and work-equipment allocation for ships in Shipping 2 For ships, it is recommended to pay more attention to the efficiency of processing unit equipment at the port. If the elasticity criterion is lowered (i.e., the elasticity criterion of 0.97), the operator should pay more attention to the CWs' operational process control and processing efficiency of unit equipment for ship 1's and work equipment allocation for ship 2's. Recommended.

Table 9 below shows the differences between the three types through estimation of the relative association of the entire shipping company and the local odds ratio.When the elasticity criterion is set to 1.0, the differences between ships 1, 2, and 3 differ from each other with respect to different ships. Instead, it represents the probability of having a difference in logistics, equipment and time. The probability of logistic differences instead of equipment and time differences in Shipping Line 1 is 0.791 times and 0.945 times higher (0.209 times and 0.055 times lower) than in Shipping Line 3, respectively, while the probability of equipment differences occurring instead of time differences in Shipping Line 1 is 1.195 times higher. It is estimated to be. On the other hand, the probability of logistics differences instead of equipment and time differences in shipping company 3 is 1.261 times and 1.113 times higher than that in shipping carrier 4, respectively, whereas the probability of equipment differences occurring in shipping company 3 instead of time differences is 0.883 times higher than in shipping company 4 (0.117). Times lower). This requires the operator's attention to control the operation of CWs rather than allocating work devices or unit device efficiency in ships of ships 1 compared to ships of ships 1, but the focus of the management process is unit devices in ships of ships 1 compared to ships 3. It means that it is expected to move from efficiency to work equipment allocation. Nevertheless, ships of shipper 3 are likely to make ships of shipper 1 more focus on managing the allocation of work equipment instead of efficiency of unit equipment than ships of shipper 4. Shipping company 3's ship is likely to cause ship 1's ship to focus less on the operational control management of CWs instead of unit equipment efficiency, while shipping company 4's ship is more likely to force ship 1's ship to focus more on the same management. .

In the previous analysis, the limit of the elasticity criterion was established. Lowering the elasticity criterion is expected to reduce the likelihood of operating gaps. However, the criterion of low resilience can be interpreted as a high readiness of the port for operational differences. The level of elasticity criterion can also be important in estimating operational manageability for a shipping company. 10 to 12 show changes in estimated local odds ratios for the three types of differences with respect to the elasticity criterion from the viewpoint of line 1. The estimated odds ratio is highly dependent on the observation of the differences derived in Figs. 7-9. In particular, the probability of not coincident with the work equipment allocation (equipment difference) instead of unit processing efficiency (time difference) in the ship of Shipping Line 2 is always higher than that of the ship of Shipping Line 1, and generally higher when the elasticity criterion is lowered (Fig. 10). ). This trend of increase is due to the fact that less efficient unit processing (time difference) is observed in ships of ship 1 than ships of ship 2. In addition, in the observation of less efficient unit processing (time difference), the reduction rate of ships in Shipping Line 2 is, on average, 25% higher than when the working equipment allocation (equipment difference) does not match. In FIG. 12, the probability of change is derived from the fact that the observation of the ship of ship 1 steadily decreases, but the observation of the equipment and time difference gradually decreases.

Because operational performance can be achieved by efficiently managing the container flow in the operational process, it is important to control the operational differences between CWs, which measure how closely actual CWs follow the planned CWs according to the operational process. The operational difference was divided into three management factors: logistics, equipment, and time difference, and the operational manageability could be measured by estimating these three management factors. The proposed process mining approach discovers the operational differences of CWs from the event log, and the (local) odds ratio estimation model estimates operational manageability in terms of management factors.

The experimental example of the present invention was carried out using the actual data set of container ports for 4 ships. The comparison results show that different shipping companies require different levels of operational manageability, organized by management elements, to control operational variances. The elasticity criterion is effective for detecting operational differences, and due to the relative ratio of differences observation, the estimated operational manageability among shipping companies was varied.

The above-described method for evaluating ship management management for flexible port operation according to an embodiment of the present invention may be implemented in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. . The computer-readable recording medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded in the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in the computer software field. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floptical disks. optical media), and hardware devices specially configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those produced by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the processing according to the present invention, and vice versa.

In the above, preferred embodiments of the present invention have been described in detail with reference to the drawings. The description of the present invention is for illustrative purposes only, and those of ordinary skill in the art to which the present invention pertains will be able to understand that other specific forms can be easily modified without changing the technical spirit or essential features of the present invention.

Therefore, the scope of the present invention is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning, scope and equivalent concepts of the claims are included in the scope of the present invention. It must be interpreted.

Claims (14)

A computer program stored in a medium in order to execute a method for evaluating the shipbuilding operation manageability through container workflow analysis on a computer, the evaluation method comprising:
Generating a container workflow by extracting an event log from a database of a container port information system; And
Including; identifying operational differences between the planned container workflow and the actual container workflow
The event log includes an ID attribute indicating a container ID, an activity attribute indicating a task to be performed, a time attribute indicating a work time, and a resource attribute indicating a performing equipment,
As a component of the container workflow, including a logistic element related to the performing work, an equipment element related to the performing equipment, and a time factor related to the working time,
Estimating the operational manageability of each shipping company by representing the probability of each component of the container workflow having a difference from the identified operational difference, and comparing the relative degree of association of the shipping company;
A computer program stored in the medium in order to execute the method for evaluating the ship's operation manageability for flexible port operation further comprising a. The method of claim 1,
The logistics factor relates to the type and order of the performing work, the equipment factor relates to the allocation of the performing equipment, and the time factor relates to the work end time. A computer program stored on a medium to run it. The method of claim 2,
The difference between the logistical factors is the operation vector represented by the (1 × t ) matrix and the (t × t ) matrix (in the above matrix, t is the number of elements in the union of the planned container workflow task and the actual container workflow task). A computer program stored in a medium to execute a method for evaluating a ship's operation manageability for flexible port operation, characterized in that it is obtained by calculating a displayed context vector. The method of claim 3,
The work vector is determined according to Algorithm 1 below, and the context vector is determined according to Algorithm 2 below:
[Algorithm 1]

In Algorithm 1, j is the index of the container workflow, N _j is the union of the planned container workflow task and the actual container workflow task, and d is the task index of N _j (d =1, ..., t). , A _j is the task vector of the j th container workflow, nd _d is the d th task,

Is a component of the d- th task in the j- th container workflow,
[Algorithm 2]

In Algorithm 2,

Is the d- th and

Is the component between the first tasks,

Is the d-th and

Is the number of task transitions between th task, and T _j is the context vector of the j th container workflow. The method of claim 4,
A computer program stored in a medium to execute a method for evaluating ship management management for flexible port operation, characterized in that the calculation of the work vector and the context vector is performed according to Algorithm 3 below:
[Algorithm 3]

In Algorithm 3, ∧ is and (and). The method of claim 2,
A computer program stored in the medium to execute the method for evaluating the shipboard operation manageability for flexible port operation, characterized in that the difference between the equipment elements is determined according to the following Algorithm 4:
[Algorithm 4]

In Algorithm 4, [ t ] is the number of jobs in the j th container workflow, D is the set of jobs in the planned container workflow,

Is the equipment of the d- th task of the j- th container workflow , and m _d is the equipment of the d- th task of the planned container workflow. The method of claim 2,
A computer program stored in a medium in order to execute the method for evaluating the shipboard operation manageability for flexible port operation, characterized in that the difference in the time factor is determined according to the following Algorithm 5:
[Algorithm 5]

In Algorithm 5, ET is the estimated execution time of the planned container workflow, ET _j is the time difference of the j th container workflow, and [ Dj = { nd ₁ , nd ₂ , · ··, nd _t }] is the j th It is the working set of the container workflow, Sd _j and Ed _j are the start and end times of the d- th task in the j- th container workflow, respectively, and LP is a positive value that makes ET _{j positive.} delete The method of claim 1,
A computer program stored in a medium to execute a method for evaluating ship management management for flexible port operation, characterized in that the probability of each component having a difference is represented by Equation 1 below:
[Equation 1]

In Equation 1,

It is a variety (s) component (e) the probability that the difference, n _se is the difference between the components of the e s shipping, n _s is the sum of the components of the difference s offering. The method of claim 9,
A computer program stored in a medium in order to execute a method for evaluating the management of a ship for flexible port operation, characterized in that the relative degree of association of the ships is represented by an odds ratio of Equation 2 below:
[Equation 2]

In Equation 2,

Is the odds ratio of the component ( e ) of l line to m line,

And l is the component (e) is likely to have a difference in variety,

Is the probability that the component ( e ) of the line m has a difference. The method of claim 9,
A computer program stored in a medium in order to execute a method for evaluating the management of a ship for flexible port operation, characterized in that the relative degree of association of the ships is represented by an odds ratio of Equation 3 below:
[Equation 3]

In Equation 3,

Is the multiplication ratio of the l p component instead of the probability that q component have a difference brings about the m variety, n _lq is a difference between the q component of the l variety, n _lp is the difference between the p component of l gives And n _mq is the difference between the q component of the m line, and n _mp is the difference between the p component of the m line. The method of claim 1,
The step of determining the operational difference comprises applying an algorithm that allows the difference of components to be equal or decreased as the elasticity criterion is lower by assigning an elasticity criterion of 1 or less. A computer program stored on a medium to execute the operational manageability evaluation method. delete A computer-readable recording medium on which the computer program of claim 1 is recorded.

Images