Abstract: The decision-making system for the prohibition of berthing/deberthing operations depending on the vessel and environmental conditions has been developed based on an expert system. Parameters such as the size of the ship, loading conditions, arrival/departure status, berthing direction(port or starboard), wind speed and directions, tidal current conditions, available tug capacity were chosen as the major factors which might influence the maneuvering difficulty when berthing or deberthing. FTS(Fast Time Simulation) technique was used to obtain the vast amount of learning data to train the expert system. The expert system has been applied to a specific product-tanker terminal and its estimated results were compared with those by human berth master, which showed general agreements. In this paper, the procedures for the development of the decision-making system is described and its application results is shown

　

 The maneuvering difficulties felt by shiphandlers when berthing or deberthing may vary depending on the various parameters prevailing at that time. The variety of the size of the vessels as well as their various loading conditions, together with the complicated combinations of environmental conditions may make it difficult for the berth master to determine whether berthing or deberthing is safe or not at specific situation.

　

 Most tanker terminal operating companies have their own guidelines to allow or to prohibit berthing/ deberthing operations under some environmental conditions. They are, however, not definite enough so that they can hardly be used as practical and objective guidelines. Furthermore, there exists a room for the subjective judgment on the safety of a ship during berthing/deberthing operations in ambiguous situations, which may result in not only the allowance of berthing/deberthing operations in dangerous situations but also its prohibition in safe situations.

　

 To establish more objective and reasonable guidelines on the allowance or prohibition of berthing/deberthing operations at various situations, it has been tried to develop discrimination software based on an expert system.

　

 Quantitative assessment on the maneuvering difficulties during berthing/deberthing operations in various situations was made through fast time simulations for a number of combinations of various parameters . B y correlating thus estimated maneuvering difficulties with those assessed from the human berth master, a set of learning data was created which could be used as input to train the expert system.

　

 In order to develop an expert system to determine whether to allow or to prohibit berthing/deberthing operations depending on the various parameters, these procedures were applied to a specific product-tanker terminal located at southern part of Korea. The estimated results by the expert system were compared with those by human berth master, which showed general agreement.

　

 The correlation analysis between the estimated results by the expert system and those by human berth master is being carried out regularly, by which it is expected to be possible to improve the accuracy of the expert system.

　

2.1 Synopsis

　

 In case of product-tanker terminal of E-company (hereafter described as "E-Berth"), a number of product carriers with various sizes and loading conditions are berthing and deberthing everyday. The company has its own guidelines and manuals that roughly describe when to allow or to prohibit the berthing/deberthing operations depending on the environmental conditions. They are, however, not so definite enough that they can hardly be used as practical and objective guidelines. Furthermore, there exists a room for the subjective judgment on the safety of a ship during berthing/deberthing operations in ambiguous situations, which may result in not only the allowance of berthing/deberthing operations in dangerous situations but also its prohibition in safe situations .

　

It became necessary for the company to establish more objective and reasonable guidelines on the allowance or prohibition of berthing/deberthing operations by taking various parameters such as vessel conditions and environmental conditions into consideration.

　

 In order to meet these requirements, an intelligent expert system was introduced, which could assess maneuvering difficulty at specific situations by considering various parameters. To construct the knowledge databases to train the expert system, simulations were carried out for almost all situations that the ships calling at the E-Berth might be encountered with. The simulation results were analyzed and correlated with the assessment data by human berth master to give final learning database set for the expert system. The schematic diagram for the construction of this expert system is shown in Fig. 1.

　

　

2.2 Selection of Target Berth and Target Ship

　

 Owing to the variety of ship sizes and berth locations, it was difficult to consider all of the possible combinations of these parameters. Considering the characteristics (maximum and most frequent vessel size) of the vessels calling at each berth, representative ships and berths were selected as Table l through discussions with the personnel working in the E-Berth.

　

 By selecting these berths and ships as representatives of entire E-Berth, it was expected that the results could be applied to nearby berths and vessel sizes in between.

　

　

2.3 Selection of Parameters affecting Maneuvering Difficulty

　

 There may exist a number of parameters that may affect the maneuvering difficulty of a berthing or deberthing ship. Among these, following factors were determined as critical and selected as major parameters.

　

　

2.4 Assessment of Maneuvering Difficulty by Fast Time Simulation

　

 In order to assess the maneuvering difficulties of a berthing and deberthing vessel, various combinations of environmental conditions were constructed. For each combination, fast time simulations were carried out and the results were analyzed as a function of various environmental conditions.

　

 The simulations were carried out by using the FMBS (Full Mission Bridge Simulator) system of KRISO and by adopting offline auto-tracking algorithms which modeled human pilots' decision-making procedures. The brief algorithms of these fast time simulation and various indices of maneuvering difficulty can be found in reference [1, 2]. Fig. 2 shows an example of off-line fast time simulation.

　

　

2.4.1 Selection of Fast Time Simulation Scenario

　

 Among the huge number of possible combinations of various parameters listed in Section 2 .3, following combinations were selected as final scenarios for each berth. For each berth, 14,112 scenarios were constructed as Table 2.

　

　

 Fast time simulations were carried out for Berth A and B, and for berthing and deberthing operations, respectively

　

2.4.2 Analysis of Fast Time Simulation

　

 As a maneuvering index representing quantitative difficulty during berthing and deberthing operations, tug usage parameter was adopted. This parameter was a measure on how much efforts were put to maintain the course and heading of a ship as desired during berthing or deberthing operations against environmental disturbances such as wind and tidal currents.

　

 Fig. 3 shows one part of the analysis results of maneuvering difficulty for arrival (berthing) conditions for Berth A. It is seen in this Figure that as ship size is increasing, the maneuvering difficulty generally increases as well.

　

　

 Since it was difficult to directly relate these tug indices with maneuvering difficulty, correlation analysis was carried out. After obtaining maneuvering difficulties ( I =Very Easy, 2=Easy, 3=Moderate, 4=Difficult, 5=Very Difficult) felt by human berth master for some typical situations, they were compared with the calculated tug usage parameters at the same conditions . Even though, the number of the data that can be used for the correlation was very limited, it could give a clue to find a relationship between the calculated tug usage parameters and the real maneuvering difficulties felt by human berth master.