DE102004031977A1 - Method for creating a stowage plan for container ships - Google Patents

Abstract

A method for speeding up the congestion planning process for container ships is hereby proposed. The process is based on the process of remembering how old similar problems have been solved. Old stowage plans are systematically stored in the case base. Every new stowage task relies on such cases. The planner has the opportunity to select a similar case, applying its solution concept to the current stowage planning problem and often modifying it. The method is a learning system, with each stowage plan newly stored in the case base increasing the system's ability to solve.

Description

One Method for accelerating the process of congestion planning for container ship is hereby proposed.

A container ship, a ship that transports containers, usually has a fixed route, see 2 , A stowage plan is a diagram showing the use of the staging areas aboard the ship. The task is to develop a method that is able to (semi-) automatically make the process of stowage planning in a port of loading (POL).

In front When the ship arrives, the planner begins sketching the ship Storage plan. The initial state of the stowage plan is the charge on Board after unloading the container destined for this port (POL) are. This condition can be calculated starting from the departure state in the previous port (Previous Port of Loading or PPOL) and from the container unloading list of the POL (Container Discharging List or CDL). Another input is the number of containers to load (Container Loading List or CLL). This CLL is usually not exactly, often even until shortly before the departure of the ship. It is factual a planning task with uncertainties.

To the Generating a stowage plan are skills in the field of marine engineering, e.g. ship stability and strength required. Depending on the route, the ports of call, weather conditions, type of ship and size must the ship has different criteria, e.g. Ship stability or draft, meet. Often There is a situation in which a container has one or more other containers that need to be unloaded are blocked. You have to do that Damming the blocking container, i. unloaded and loaded. There Congestion is expensive, belongs the congestion to the goal of minimization, especially if the ship no charger on board.

The attempts to automate the stowage planning process for container ships have a long way to go. Most experiments used mathematical or optimization techniques. The difference lies in the way in which the objective functions and constraints have been manipulated, see Schott, R., _" Containment Planning for Container Vessels" in Traffic Sciences Studies from the Institute of Transportation Science of the University of Hamburg, Vandenhoeck & Ruprecht, Göttingen, 1989, and Wilson, ID et al, "Container stowage pre-planning: using search to generate solutions, a case study", Knowledge-Based Systems 14 (2001) 137-145 An interesting attempt "method for determining a stowage plan" by Davidor, Y. and Avihail, M. is under US 5,809,489 patented. The difficulty with the methods is the limited ability to use the method for a different vessel size or route. The loading strategy for a ship or route can not be applied to another ship or route. The subjective aspects of the stowage planning process are hardly considered, see also [4].

Despite these attempts and the progress of information technology, stowage planning is still carried out manually. This means that the planner must himself develop the stowage plan concept. The support of the computer program is limited to the visualization of the stowage planning process and to calculations of ship conditions, eg ship stability and strength, by modules ( 18 ), see in 3 ,

The The practice of stowage planning shows that an experienced planner has a stowage plan faster and even better than a new or inexperienced planner. From experience, an experienced planner can anticipate the uncertainties. He knows specific requirements of every single port or each single route and even specific requirements of important customers. This knowledge leaves not (simply) into a form of mathematical formulas or Formulate rules.

Experience is basically a collection of the known congestion planning situations. The more stowage plans one has done, the more experienced one has become. If a stowage planner gets a job to create a stowage plan, he tries to mentally tie in with the old stowage planning situations he knew. The Solutions more similar old stowage planning situations could be the starting point to get to your destination faster. The basic one The idea is to create a stowage plan by remembering like that Jam planning problem solved had been.

Next books and experienced planners, the stowage plans are in essence a source of Knowing about it, how to solve the stowage planning problems Despite the complexity of the dam planning task and Despite the difficulties in acquiring knowledge, an old one can Stauplan a useful means to create a new stowage plan.

The The aim of this invention is to develop a method that the Planner will be able to design a stowage plan faster.

Jam planning is a search process in a space of many possibilities. The stowage plan must fulfill boundary conditions, eg the draft. The stowage planning can also be considered as a learning process. The more stooping plans you have, the more experienced you are, and the quicker you do it a stowage planning task. The following factors play a role in the stowage task: specific requirements and subjectivity. Each route has specific cargo patterns, and each port is often unique. Important customers often require specific parking spaces for their containers. A stowage planer also has certain preferences in solving a stowage planning problem. For the same stowage problem, two planners can create different stowage plans, which are both equally good. The solution is not unique.

One case-based closing (Case-Based Reasoning) is a new process Problem solves, that reminds you, like a similar problem solved had been. The thesis is that most problems are not unique are, but certain similarities with the already known problems.

The procedure consists of four steps: a. To retrieve). Reuse, c. Revise, d. Retain, see Aamodt, A. and Plaza, E. _" Case-Based Reasoning: Foundational Issues, Methodological Variations and System Approaches"; AI Communications 7 (i), 39-59, 1994, and Watson, I. "Applying Case-Based Reasoning: Techniques for Enterprise Systems"; ISBN 1-55860-462-6, Morgan-Kaufmann Publishers, 1997.

1 shows the overview of the case-based jam planning system. A query means the task of creating a stowage plan for the current loading port (POL). The input of a query contains two components, the initial stoppage before loading and the number of containers to be loaded (Container Loading List = CLL). Since the stowage plans are often not well documented, the initial stowage plans are to be derived from the departure schedules from the previous port (PPOL) and from the list of containers to be unloaded (Container Discharging List = CDL). A query now includes three components: PPOL's departure schedule, POL's CDL and POL's CLL ( 10 . 11 . 12 ) in 5 ,

A case base ( 14 ) is a place, eg a database, in which the stowage planning knowledge is kept. The case base is a collection of congestion planning cases. A jam case, or case, is a jam planning problem solving session. This is a package that is a problem ( 91 ), the corresponding solution ( 92 ) and a comment ( 93 ), see 4 , The problem consists of three components, see [8]. The solution is the generated stowage plan. A comment is an assessment and / or comment on the stowage plan regarding the quality of the stowage plan and / or additional information about the stowage plan. The comment can be presented in the simplest form as a number or a code. Alternatively, a comment can be detailed from several components ( 102 - 105 ) to be discribed. Both good and less good stowage plans can be stored in the case base. A good stowage plan is a positive indication of a promising solution. In contrast, a bad stowage plan means a warning to the planner, that is, the solution concept must be avoided or should be used with caution.

Similar stowage plans are taken from the case base ( 14 ) recovered, 3 , The user becomes similar to old stowage planning concepts ( 15 ), from which the user selects a solution concept that best suits his ideas. The user can then apply the selected solution concept to the current stowage planning problem and modify the stowage plan ( 20 ). With regard to aspects of shipping, such as vessel strength and stability, the stowage plan is based on the available modules ( 18 ) ( 19 ).

Today's congestion planning software has these modules ( 18 ). The completed, good or bad, stowage plan can with a comment ( 23 ) before storing it in the case base. The acceptable stowage plan is basically ready for use ( 24 ).

There are two types of similarity calculations, namely numerical and layout similarity calculations. 5 shows the process of the entire recovery process. CDL and CLL- To recover similar cases, POL 'CDL, POL's CLL and PPOL's stowage plan are from input and case base ( 10 . 11 . 12 . 91 . 92 . 93 ), see 5 , Corresponding similarity calculations are carried out ( 31 . 32 . 33 ), and then the overall similarity value is calculated ( 35 ). The cases are sorted according to their similarity values ( 36 ). Similar cases are proposed to the user as solution concepts. In 6 , the flow of numerical similarity calculations, for CDL or CLL, is illustrated.

Similarity value ≤ 1.0. Attributes similarity calculations can use one of the following methods:
Method 1: similarity value = (number of similar objects) / (total number of objects), see illustration in [18].
Method 2: Fuzzy membership function defined as follows, see 7 : Similarity = 0 if x <a
(x - a) / (b - a) if a ≤ x ≤ b
1.0, if b ≤ x ≤ c
(d - x) / (d - c) if c ≤ x ≤ d
0 if x> d,
where b ≤ (attribute value of the query) ≤ c.

Total similarity value ≤ 1.0. One of the three methods can be used to calculate the Gosness similarity value:
Method 1: total similarity value = Σ (weight _i × similarity value _i ) / Σ (weight _i )
Method 2: overall similarity value = max (similarity value _i )
Method 3: Combination of method 1 and method 2, eg total similarity value = ((weight ₅ × max (similarity value ₁ , similarity value ₂ )) + (weight ₆ × similarity value ₃ )) / (weight ₅ + weight ₆ ).

following Example shows the attribute similarity calculations. An attribute is the smallest element of an object. An object, e.g. CLL, consists of the number of containers to be loaded by a POL after all POD's.

The CLLs of the query and the current case are written in the following format [POD; POLE; Number of containers to be loaded] represents:
Query [11; 12; 250]
Case (1) [11; 12; 300]

The similarity value is determined as follows, see [12]:
Method 1: Similarity value = 250/300 = 0.83.
Method 2: Similarity value = 0.67 if a = 150, b = 225, c = 250 and d = 350.

Stowage plan similarity calculations consist of two components, a. Bay similarity calculations b. Longitudinal similarity value calculations, see 8th , The total similarity value is calculated using a method shown in [13]. The Stauplan similarity calculations are started with the lowest level, the bay similarity and then the longitudinal direction similarity. The Bay Plan similarity occurs, as by 9 will be shown.

9 shows the process of Bayikenlichkeitsberechnung. For bay similarity calculations, the following criteria are due, see 10 :
Criterion 1: Two containers or two groups Containers stowed on the deck in the same partbay are considered identical.
Criterion 2: Two containers or two groups Containers stowed in the same partbay in the hold are considered identical.
Criterion 3: Two containers or two groups of containers considered identical.
Criterion 4: Two containers or two groups of containers that have been jammed in Partbays, with the partbays being symmetric
along the center line are considered identical.
Criterion 5: Two containers or two groups of containers jammed in partbays with the partbays next to each other are considered to be non-identical.
Criterion 6: Two containers or two groups of containers jammed in Partbays with the partbays rotated 180 degrees are considered non-identical. Partbay encoding, as in 18 , helps with the implementation in the computer program.

In addition to the above-mentioned criteria [16], the following longitudinal directional criteria should be taken into account for the overall estimation of similarity, see 11 ,
Criterion 1: Two bays in the same hold or in the same part of the hold are considered identical.
Criterion 2: Two bays in the adjacent cargo holds are considered identical.

12 to 21 show an example for the calculation of the similarity value between two bays, see 12 (a) , 12 (a) to 12 (d) show possible positions of the Bay, which are considered identical according to [16]. The similarity calculation is carried out as follows, see also 9 , Table in 13 - 17 show the number of containers in partbays of the query and the current case, see the list of references. The table in 18 to 21 are the minimum values of the query-case combination.

The Number of containers having identical parking spaces is: max (75; 67; 90; 82) = 90.

The similarity value can be calculated as follows, see [14]:
Method 1: Bay Similarity Value = 90/99 = 0.91.
Method 2: Bay similarity value = 0.88 if a = 75, b = 92, c = 99, d = 99.

23 illustrates the similarity calculation in the longitudinal direction, see also [15] and 9 , The bay positions that are considered identical are identified by the 23 (a) , (b), and (c) illustrates. First, the positions of the bays of the query are varied in order to create a unique query-case combination, see also 11 , Sim _i is the similarity value of the query-case combination whose similarity calculation takes place according to [12]. The total similarity value is the maximum value of the above-mentioned query-case combinations. As an alternative to the overall similarity calculation, one of the methods mentioned in [13] can be used.

24 - 26 provide an example of the implementation of the proposed congestion planning especially as part of the recovery of similar cases.

27 illustrates an example of reuse and auditing procedures. The POL's stowage plan ( 111 ), the POL 'CDL ( 113 ) and CLL ( 117 ) form the query (stowage planning task). The stowage plan after unloading the containers transported for the POL ( 115 ) becomes ( 111 ) and ( 113 ) derived. The selected case, ( 16 ) in 3 , contains the components ( 112 . 114 . 116 . 118 and 120 ). The stowage plan ( 120 ) is considered a solution for the current stowage planning problem. By means of the criteria mentioned in [16] the stowage planning problem can be dealt with faster, whereby several solutions ( 119 . 121 ) possible are.

28 shows a relationship between the number of schedules created and the ability of the system to assist the planner during the stowage planning process.

In the earlier phase of stowage planning, eg before a ship starts a new route, there are no stowage plans in the case base that apply to a real journey. Nevertheless, studies have usually been done for certain loading situations. These are hypothetical stowage plans, noted with trip number 0, in 28a , When the planner makes a stowage plan, it looks for similar cases that are still limited in scope at the moment. Each time the planner has created a stowage plan, this stowage plan is stored in the case base. This stowage plan is immediately available and can be used directly in the following stowage planning task. The process is a learning system, that is, each newly created stowage plan increases the ability of the system to assist the planner in performing the stowage planning task. The more stowage plans have been stored in the case base, the greater the chance of finding one or more and / or better case likenesses in the case base for a stowage problem, see 28b ,

LIST OF REFERENCE NUMBERS

• 1 Overview of the method
• 2 route
• 3 Case-based congestion planning architecture
• 4 case basis
• 5 recovery
• 6 CDL or CLL similarity calculations
• 7 Fuzzy membership function
• 8th Overview of stowage plan similarity calculations
• 9 Bayähnlichkeitsberechnungen
• 10 Bayähnlichkeitskriterien
• 11 Longitudinally similarity criteria
• 12 Example: Bay similarity calculations
• 13 Table 1: Number of containers of the query
• 14 Table 2: Number of containers of mirrored Partbay on deck
• 15 Table 3: Number of containers of the mirrored partbay in the hold
• 16 Table 4: Number of containers of the mirrored bay
• 17 Table 5: Number of containers of the current case
• 18 Table 6: (min (Table 1, Table 5))
• 19 Table 7: (min (Table 2, Table 5))
• 20 Table 8: (min (Table 3, Table 5))
• 21 Table 9: (min (Table 4, Table 5))
• 22 Partbaykodierung
• 23 Example of the validity calculation
• 24 Example Implementation - Input
• 25 Example implementation - CLL input
• 26 Example Implementation - Recovery Output
• 27 Example of a stowage planning process
• 28 Growing case basis: a learning system

Claims (6)

A method for speeding up the congestion planning process for container ships, characterized in that a new stowage plan is created by reminding us how a similar stowage planning problem has been solved. Method according to claim 1, characterized in that that the following processes take place: • Recovery of similar cases • Selection a proper stowage plan concept • Use of the concept, through the application of the solution concept to the current stowage problem and by modifying the stowage plan, where necessary. • Storage of the newly created Stowage plans in the case base. Method according to claim 1 and / or 2, characterized that recovery is more similar Cases out consists of the following components: • CDL similarity calculations of Query and the case base • CLL similarity calculations the query and the case base • PPOL stowage similarity calculations the query and the case base Method according to claim 1 and / or 2, characterized that • of the Users have the opportunity to select a proposed stowage plan he for the best • the user the selected one Stowage plan as the basis for the solution the current stowage problem uses. Method according to claim 1 and / or 2 and / or 3, characterized in that • the solution concept of the selected stowage plan the current stowage problem is being applied • the user modify the stowage plan until it's done • the user by means of the available Stowage planning modules ship stability and / or stability and / or control other aspects. Method according to claim 1 and / or 2, characterized that • one Jamming case, or a case, consists of three components, namely problem, solution and comment. The problem consists of the PPOL 'Stauplan, POL' CDL and POL 'CLL. The solution contains the created stowage plan. And the comment contains the Evaluation of and comments on the created stowage plan. • the cases systematically stored in the case base. • Old and newly created stowage plans systematically stored in the case base • become, thus the recovery similar stowage plans is made possible, • the more Cases in the case base are stored, the greater the chance for a stowage planning problem one or more and / or better case similarities in the case base to find.