Coefficient identification program for maneuvering motion model for vessel, utilization method for coefficient identified by maneuvering motion model, and coefficient identification system for maneuvering motion model

Abstract

To identify a coefficient for a maneuvering motion model by using maneuverability index often acquired for such as a sea trial for actual vessel in general.SOLUTION: A coefficient identification program for a maneuvering motion model of a vessel has a computer execute: a step S2 for acquiring maneuverability index data; a step S3-2 for acquiring an initial value for a coefficient of the maneuvering motion model; a step S4 for carrying out calculation of the maneuvering motion by applying the initial value for the coefficient to the maneuvering motion model; a step S5 for calculating maneuvering index based on calculation result of the maneuvering motion; a step S6 for calculating an evaluation value based on the maneuverability index data and the maneuverability index; a step S7 for determining termination based on the calculation result of evaluation value and termination determination condition; a step S8 for updating the maneuvering motion model coefficient based on optimization algorithm when termination is denied and repeating from the step S4 to the step S7; and a step S9 for outputting the latest used coefficient when termination is affirmed as the identified coefficient.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a coefficient identification program for a ship maneuvering motion model, a method for utilizing identified coefficients of the maneuvering motion model, and a coefficient identification system for the maneuvering motion model. [Background technology]

[0002] When developing an automated system for ship docking and undocking, it is necessary to perform simulation calculations of the target ship's maneuvering motion in order to tune the parameters before installing the system on an actual ship. Patent Document 1 discloses a method for keeping a ship on its course using a nonlinear autoregressive model, in which a statistical model is created having radial basis functions that represent the ship's movement from time series data of the ship at sea as regression coefficients of the nonlinear autoregressive model, and an optimal command rudder angle that minimizes the ship's course deviation from its desired course is obtained using optimal control theory, and the ship's position is controlled using this command rudder angle to minimize the course deviation. Furthermore, Patent Document 2 discloses an automatic steering device which, as a first means, calculates an optimal steering amount by taking into account a turn rate obtained from an azimuth angle detection means in addition to a course deviation correction, and controls the rudder angle based on the optimal steering amount, and, as a second means, sets a virtual set course that is shifted by a predetermined angle from a set course, measures the course deviation, which is the difference between the actual course and the virtual set course, calculates a course deviation offset value from the course deviation, calculates an optimal steering amount based on the course deviation offset value, and controls the rudder angle based on the optimal steering amount. Patent Document 3 discloses a self-replicating neuro-fuzzy knowledge acquisition device in which a model error evaluation section evaluates the error between output data from an input / output data storage section and a model output from a model parameter learning section, and based on the evaluation result, a rule / hidden layer unit self-replicating section automatically generates rules and / or hidden layer units and sets them in a neuro-fuzzy knowledge base, and if a knowledge acquisition completion determination section determines that the model error is equal to or less than an initially specified value, the entire knowledge acquisition process is terminated, and if not, the above-mentioned process is repeated, and the final learning result of the model parameters when the number of rules / hidden layer units is n is reflected in the initial value of the parameter learning process when the number of rules / hidden layer units is (n+1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2014-4911 A [Patent Document 2] JP 2008-230484 A [Patent Document 3] Japanese Patent Application Publication No. 6-35707 Summary of the Invention [Problem to be solved by the invention]

[0004] In order to perform a computer simulation of the maneuvering motion of a target ship, it is necessary to set the coefficients of the maneuvering motion model for each target ship. Until now, identification of steering control models for individual ships targeting the time series of free running motion, and identification of the coefficients of maneuvering motion models using restrained model tests and CFD calculations have been performed, but performing free running tests using actual ships or model ships and CFD calculations for the purpose of identifying the coefficients of maneuvering motion models requires a large amount of human, financial, and time costs. In addition, when identifying the coefficients of the maneuvering motion model from the time series data of free running tests, data processing and primary analysis are required for each implementer because the data format is not unified. Here, Patent Documents 1 to 3 do not attempt to identify the coefficients of a maneuvering motion model more simply and inexpensively than conventional methods that use model tests and CFD calculations. Patent Documents 1 and 2 both use data detected on a ship to directly optimize the control parameters of the ship, while Patent Document 3 uses a self-replicating fuzzy knowledge acquisition device to automatically set the minimum model structure that achieves a specified model accuracy and appropriate initial values ​​of model parameters, and is not related to ships. Therefore, the present invention aims to identify the coefficients of a maneuvering motion model using maneuverability indices that are generally obtained from sea trials of an actual ship, without requiring costly model tests or CFD calculations. [Means for solving the problem]

[0005] A coefficient identification program for a ship maneuvering motion model corresponding to claim 1 is a program for identifying coefficients of a ship maneuvering motion model, characterized in that the program causes a computer to execute the following steps: a maneuvering index data acquisition step for acquiring maneuvering index data of the ship; a coefficient initial value acquisition step for acquiring initial values ​​of coefficients of the set maneuvering motion model; a maneuvering motion calculation step for applying the acquired initial values ​​of coefficients to the maneuvering motion model to calculate a maneuvering motion; a maneuvering index calculation step for calculating a maneuvering index based on a result of the maneuvering motion calculation; an evaluation value calculation step for calculating an evaluation value based on the acquired maneuvering index data and the calculated maneuvering index; a termination determination step for determining termination based on the calculation result of the evaluation value and a predetermined termination determination condition; a repetition step for updating the coefficients of the maneuvering motion model based on an optimization algorithm and repeating the maneuvering motion calculation step, the maneuvering index calculation step, the evaluation value calculation step, and the termination determination step if the termination is determined to be positive; and an identification coefficient output step for outputting the most recently used coefficient as an identified coefficient if the termination is determined to be positive. According to the present invention as set forth in claim 1, the coefficients of a ship maneuvering motion model required for performing a maneuvering motion simulation can be identified more easily and at lower cost than with conventional methods by using maneuverability index data.

[0006] The present invention as recited in claim 2 is characterized in that a maneuvering motion model setting step of setting a maneuvering motion model selected from a plurality of maneuvering motion models is further executed. According to the present invention as set forth in claim 2, a user can select and set a maneuvering motion model that expresses motion that conforms to the application range of the automatic ship-steering system to be installed on an actual ship.

[0007] The present invention as described in claim 3 is characterized in that the maneuvering motion model is any one of the KT model, which is a first-order lag response model using gain and time constant, the MMG model, which is a model in which the fluid forces acting on the ship are expressed as a linear sum of the components of the hull, propeller, and rudder and their interaction terms, and this is input as an external force into the ship's equation of motion, and the low-speed maneuvering motion model. According to the present invention described in claim 3, by setting the KT model, the MMG model, or the low-speed maneuvering motion model as the maneuvering motion model, it is possible to accurately express motion that conforms to the application range of the automatic ship steering system to be installed on an actual ship.

[0008] The present invention as recited in claim 4 is characterized in that when a simplified estimation formula using the hull main specifications as initial values ​​is used in the coefficient initial value acquisition step, a hull main specifications acquisition step of acquiring the hull main specifications is further executed. According to the present invention as set forth in claim 4, the main hull specifications of the ship to be analyzed are acquired, and the initial coefficient values ​​can be obtained by utilizing a simplified estimation formula.

[0009] The present invention as defined in claim 5 is characterized in that the maneuverability index data is sea trial data of a ship. According to the present invention as set forth in claim 5, the coefficients of the maneuvering motion model can be identified by using the maneuverability index obtained in the sea trials which are always carried out when a ship is newly built or remodeled.

[0010] The present invention as recited in claim 6 is characterized in that a scalar quantity of sea trial data is used as the maneuverability index data. According to the present invention as set forth in claim 6, by using a scalar quantity of data that is widely recognized internationally as an index for determining the maneuverability of a ship, it is possible to identify the coefficients of the steering motion model, for example, without the need for primary processing of the track data, even when no track data remains.

[0011] The present invention as recited in claim 7 is characterized in that data to be used is selected from among the sea trial data in accordance with a set maneuvering motion model. According to the present invention as set forth in claim 7, the necessary sea trial data corresponding to the set maneuvering motion model can be selected, and the coefficients of the maneuvering motion model can be appropriately identified.

[0012] The present invention as recited in claim 8 is characterized in that the calculation of the evaluation value in the evaluation value calculation step is performed using a weighted sum of squares of the difference between the maneuverability index data and the maneuverability index. According to the present invention as defined in claim 8, it is possible to equalize the contribution of the types of maneuverability indexes to the improvement of the evaluation value.

[0013] The present invention as recited in claim 9 is characterized in that the termination determination condition for determining termination in the termination determination step is any one of the number of repeated calculations, convergence of the calculation result of the evaluation value, and a threshold value for the calculation result of the evaluation value. According to the present invention as defined in claim 9, the quality of identification of the coefficients of the maneuvering motion model can be ensured by a predetermined number of iterative calculations, convergence of the calculation results of the evaluation value, or a predetermined threshold value.

[0014] The present invention as defined in claim 10 is characterized in that a genetic algorithm is used as an optimization method for updating coefficients in the repetitive steps. According to the present invention as set forth in claim 10, the optimization problem can be solved using a genetic algorithm, and the coefficients of the control motion model can be identified.

[0015] A method of utilizing the identified coefficients of a ship's maneuvering motion model corresponding to claim 11 is characterized in that the coefficients of the ship's maneuvering motion model identified by a ship's maneuvering motion model coefficient identification program are used in simulation calculations of the ship's automatic docking / undocking maneuvering or automatic operation maneuvering motion. According to the present invention as set forth in claim 11, by using the coefficients of a maneuvering motion model identified using a coefficient identification program for a ship maneuvering motion model, it is possible to simplify and reduce the cost of maneuvering motion simulation calculations for parameter tuning, etc.

[0016] A method of utilizing the identified coefficients of a ship's maneuvering motion model corresponding to claim 12 is characterized in that the coefficients of the ship's maneuvering motion model identified by a ship's maneuvering motion model coefficient identification program are used as control parameters for automatic docking / undocking maneuvering or automatic operation of the ship. According to the present invention described in claim 12, the accuracy of control of automatic berthing / unberthing maneuvering or automatic operation can be improved by utilizing the coefficients of a ship maneuvering motion model identified by a coefficient identification program for the ship maneuvering motion model.

[0017] A coefficient identification system for a ship maneuvering motion model corresponding to claim 13 comprises a computer, an input means for inputting at least one of maneuverability index data and an initial value of the coefficient to the computer, and an output means for outputting at least the identified coefficient from the computer, and is characterized in that the computer executes a coefficient identification program for the ship maneuvering motion model. According to the present invention as set forth in claim 13, a system can be provided that identifies the coefficients of a ship maneuvering motion model required for performing a maneuvering motion simulation more simply and at lower cost than conventional methods by using maneuverability index data.

[0018] The present invention as set forth in claim 14 is characterized in that the input means and the output means are provided in remote locations and connected to the computer via an information and communication network. According to the present invention as set forth in claim 14, a user can identify the coefficients of the control motion model from a remote location via an information and communication network.

[0019] The present invention as recited in claim 15 is characterized in that the computer is linked to another computer that performs simulation calculations of automatic docking / undocking maneuvering or automatic navigation maneuvering movements of a ship. According to the present invention as set forth in claim 15, when performing simulation calculations of the maneuvering movements of automatic berthing / unberthing maneuvers or automatic navigation in another computer, the simulation calculations can be performed easily and at low cost by utilizing the coefficients of the maneuvering movement model identified by the coefficient identification program of the maneuvering movement model.

[0020] The present invention as recited in claim 16 is characterized in that a computer is linked to a control means for performing automatic docking / undocking maneuvering or automatic navigation of a ship. According to the present invention described in claim 16, the accuracy of control of automatic berthing and unberthing maneuvering or automatic operation can be improved by using the coefficients of a ship maneuvering motion model identified using a coefficient identification program for the ship maneuvering motion model. Effect of the Invention

[0021] According to the coefficient identification program for a ship maneuvering motion model of the present invention, the coefficients of the ship maneuvering motion model required for performing a maneuvering motion simulation can be identified more easily and at lower cost than conventional methods by using maneuverability index data.

[0022] In addition, when the steering motion model setting step of setting a steering motion model selected from a plurality of steering motion models is further executed, the user can select and set a steering motion model that expresses motion that conforms to the application range of the automatic steering system to be installed on an actual ship.

[0023] In addition, when the maneuvering motion model is any of the KT model, which is a first-order lag response model using gain and time constant, the MMG model, which expresses the fluid forces acting on the ship as a linear sum of the components of the hull, propeller, and rudder and their interference terms, and inputs this as an external force into the ship's equation of motion, and the low-speed maneuvering motion model, by setting the KT model, MMG model, or low-speed maneuvering motion model as the maneuvering motion model, it is possible to accurately express motion that is in line with the application range of the automatic ship steering system to be installed on an actual ship.

[0024] In addition, when a simplified estimation formula using hull main specifications as initial values ​​is used in the coefficient initial value acquisition step, if a hull main specifications acquisition step for acquiring hull main specifications is further executed, the hull main specifications of the ship to be analyzed can be acquired and the coefficient initial values ​​can be obtained using the simplified estimation formula.

[0025] In addition, when the maneuverability index data is data from sea trials of a ship, the coefficients of the maneuvering motion model can be identified using the maneuverability index obtained from sea trials, which are generally always conducted when a ship is newly built or modified.

[0026] In addition, when using scalar quantities from sea trial data as maneuverability index data, by using scalar quantities from data that are widely recognized internationally as indicators for determining the maneuverability of a ship, it is possible to identify the coefficients of the steering motion model, for example, without the need for primary processing of the track data, even if no track data remains.

[0027] In addition, when selecting data to be used from the sea trial data in accordance with the set maneuvering motion model, the necessary sea trial data corresponding to the set maneuvering motion model can be selected, and the coefficients of the maneuvering motion model can be appropriately identified.

[0028] Furthermore, if the calculation of the evaluation value in the evaluation value calculation step is performed using a weighted sum of squares of the difference between the maneuverability index data and the maneuverability index, the contribution of the type of maneuverability index to the improvement of the evaluation value can be equalized.

[0029] In addition, when the termination determination condition for determining termination in the termination determination step is any one of the number of iterative calculations, the convergence of the calculation result of the evaluation value, and a threshold value for the calculation result of the evaluation value, the quality of the identification of the coefficients of the steering motion model can be guaranteed by a predetermined number of iterative calculations, the convergence of the calculation result of the evaluation value, or a predetermined threshold value.

[0030] Furthermore, when a genetic algorithm is used as an optimization method for updating the coefficients in the repetitive steps, the optimization problem can be solved using the genetic algorithm, and the coefficients of the control motion model can be identified.

[0031] In addition, according to the method of utilizing the identified coefficients of a ship maneuvering motion model of the present invention, by using the coefficients of the maneuvering motion model identified using the coefficient identification program for the ship maneuvering motion model, it is possible to simplify and reduce the cost of maneuvering motion simulation calculations for parameter tuning, etc.

[0032] In addition, according to the method of utilizing the identified coefficients of a ship steering motion model of the present invention, the accuracy of control of automatic docking / undocking maneuvering or automatic operation can be improved by utilizing the coefficients of the ship steering motion model identified by the coefficient identification program for the ship steering motion model.

[0033] In addition, according to the coefficient identification system for a ship maneuvering motion model of the present invention, a system can be provided that identifies the coefficients of a ship maneuvering motion model required for performing a maneuvering motion simulation by using maneuverability index data more simply and at lower cost than conventional methods.

[0034] In addition, when the input means and output means are provided at a remote location and connected to a computer via an information and communication network, a user can identify the coefficients of the control motion model from a remote location via the information and communication network.

[0035] In addition, when the computer is linked to another computer that performs simulation calculations of the maneuvering movements of a ship for automatic berthing and unberthing or automatic navigation, when the other computer performs simulation calculations of the maneuvering movements of the ship for automatic berthing and unberthing or automatic navigation, the coefficients of the maneuvering movements model identified by the coefficient identification program for the maneuvering movements model can be used to perform simulation calculations easily and at low cost.

[0036] In addition, when the computer is linked to a control means that performs automatic docking and undocking maneuvering or automatic navigation of the ship, the accuracy of control of the automatic docking and undocking maneuvering or automatic navigation can be improved by using the coefficients of a maneuvering motion model identified using a coefficient identification program for the ship's maneuvering motion model. [Brief description of the drawings]

[0037] [Figure 1] FIG. 13 is a diagram showing a processing flow of a coefficient identification program for a ship steering motion model according to an embodiment of the present invention. [Diagram 2] A diagram showing the coefficient identification system for the maneuvering motion model of the ship. [Diagram 3] A diagram showing examples of tests and maneuverability indicators conducted during the sea trial. [Figure 4] A diagram showing how the coefficient identification system for the ship maneuvering motion model is used when it is connected to an information and communication network such as the Internet. [Diagram 5] FIG. 13 is a diagram showing the results of a ship maneuvering test of a z-test according to an embodiment of the present invention. [Figure 6] A diagram showing the genetic operation (coefficient update) method using the genetic algorithm. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0038] A coefficient identification program for a ship maneuvering motion model, a method for utilizing the identified coefficients of the maneuvering motion model, and a coefficient identification system for the maneuvering motion model according to an embodiment of the present invention will be described. FIG. 1 is a diagram showing the processing flow of a coefficient identification program for a ship's maneuvering motion model, FIG. 2 is a diagram showing a coefficient identification system for a ship's maneuvering motion model, and FIG. 3 is a diagram showing examples of tests conducted in sea trials and maneuverability indexes. The coefficient identification system for a ship maneuvering motion model shown in Fig. 2 (hereinafter sometimes abbreviated as "coefficient identification system") comprises a computer (calculator) 11 provided in a system main body 10, input means 20 such as a mouse or keyboard for inputting at least one of maneuverability index data and initial values ​​of coefficients to the computer 11, and output means 30 such as a printer or monitor for outputting at least the identified coefficients from the computer 11. The system main body 10 is provided with a first memory unit 12 and a second memory unit 13 connected to the computer 11, and a control unit 14 for controlling the computer 11 and the like.

[0039] The ship steering motion model coefficient identification program (hereinafter sometimes abbreviated as "coefficient identification program") causes the computer 11 of the coefficient identification system to execute predetermined processing shown in FIG. After the start of processing by the coefficient identification program, first, a maneuvering motion model selected from a plurality of maneuvering motion models is set (S1: maneuvering motion model setting step). In the maneuvering motion model setting step S1, the maneuvering motion model setting unit 11A of the computer 11 displays options or an input field for the maneuvering motion model on a monitor that is one of the output means 30, and when a user of the coefficient identification program selects or inputs one maneuvering motion model using the input means 20 in accordance with the display, the maneuvering motion model to be identified is set. In this way, the user selects and sets a maneuvering motion model that expresses motion that conforms to the scope of application of an automatic ship-steering system to be installed on an actual ship, such as an automated system for berthing and unberthing, immediately after starting the coefficient identification program. When a single maneuvering motion model to be identified is preset and incorporated in the computer 11 or the coefficient identification program, the maneuvering motion model setting step S1 is omitted and the process proceeds to the next step.

[0040] The maneuvering motion model is set to one of the following three models, for example. This makes it possible to accurately express motion that corresponds to the application range of the automatic ship-piloting system to be installed on an actual ship. 1.KT model. The KT model is a first-order lag response model that uses gain and time constant. 2. MMG model: The MMG model expresses the fluid forces acting on a ship as a linear sum of the components of the hull, propeller, and rudder, as well as their interaction terms, and inputs this as an external force into the ship's equation of motion. 3. Low-speed maneuvering motion model: The low-speed maneuvering motion model takes into account the fluid force characteristics specific to low speeds such as when maneuvering to and from a berth and external forces such as fluid forces due to side thrusters.

[0041] Next, maneuverability index data of the ship is acquired (S2: maneuverability index data acquisition step). In the maneuverability index data acquisition step S2, the maneuverability index data acquisition unit 11B of the computer 11 acquires maneuverability index data obtained in sea trials or the like, which is used as teacher data for identifying the maneuvering motion model coefficients. Types of maneuverability index data include model ship test data, CFD calculation data, full-scale ship test data, sea trial data, etc. Maneuverability index data is easy to handle and practical as an evaluation index for identifying the maneuvering motion model coefficients. The maneuverability index data acquired by the computer 11 in the maneuverability index data acquisition step S2 can be conveniently set so that the user reads numerical values ​​from the sea trial report or the like before starting the coefficient identification program, stores them in the first storage unit 12 or the second storage unit 13 in the form of a text file or the like, and the maneuverability index data acquisition unit 11B automatically reads them, but the user can also input the data by the input means 20 in response to a request from the maneuverability index data acquisition unit 11B. It is also possible to accumulate the maneuverability index data in a database in advance, and acquire the data from the database each time an acquisition instruction is issued.

[0042] When the maneuverability index data is sea trial data of a ship, the coefficients of the maneuvering motion model can be identified using the maneuverability index obtained in the sea trial, which is generally always conducted when a ship is newly built or remodeled. When the scalar quantity of the sea trial data is used as the maneuverability index data, the coefficients of the maneuvering motion model can be identified, for example, even when no wake data remains, without the need for primary processing of the wake data, by using the scalar quantity of data that is internationally recognized as an index for determining the quality of the maneuverability of a ship. Examples of maneuverability indices include the advance, transfer, and tactical diameter obtained in the turning test shown in Fig. 3(a), the first overshoot angle, the second overshoot angle, and the first and second overshoot times indicating the turning angles obtained in the zig-zag test shown in Fig. 3(b), or the track reach from the issuance of a reverse command to stopping in the stop test. In addition, the time to reach the advance or transfer, the lateral drift distance during emergency stopping maneuvers, and the speed and angular velocity in the steady motion state during the turning test can also be used as maneuverability indices. The type of maneuverability index data to be acquired corresponds to the maneuvering motion model selected in the maneuvering motion model setting step S1. For example, when the KT model is set, the overshoot angle and its time in the z test are acquired as the maneuverability index data, and when the MMG model is selected, the longitudinal distance and lateral distance in the turning test are acquired as the maneuvering index data. In this way, by selecting data to be used from the sea trial data in accordance with the set maneuvering motion model, the necessary sea trial data corresponding to the set maneuvering motion model can be selected, and the coefficients of the maneuvering motion model can be appropriately identified. In addition to sea trial data, maneuverability index data can also be obtained from sea trials that have similar data to sea trial data from an actual ship, or data from an actual ship sailing.

[0043] Next, the initial values ​​of the coefficients of the set maneuver motion model are acquired (S3-2: coefficient initial value acquisition step). The coefficient initial value acquisition step S3-2 is a step for setting the initial values ​​of the coefficients of the maneuver motion model, and the coefficient initial value acquisition unit 11D of the computer 11 accepts an input from a user or automatically reads the initial values ​​of the coefficients of the maneuver motion model from the first storage unit 12 or the second storage unit 13. All coefficients can be given as uniform random numbers, and for coefficients for which a simplified estimation formula using the hull's principal dimensions has been published, the simplified estimation formula can be used as the initial value. If it is expected that the coefficients will be identified as values ​​close to those obtained by the simplified estimation formula, it is preferable to use the simplified estimation formula as the initial value. Examples of simplified estimation formulas using the hull's principal dimensions include those that express the coefficients of the maneuvering motion model as regression formulas using principal dimensions such as the draft-to-ship length ratio d / L and the ship's breadth-to-ship length ratio B / L, as in the following formulas (1) and (2).

number

number

[0044] When using a simplified estimation formula using the hull's principal specifications as the initial values ​​of the coefficients of the maneuvering motion model, the hull's principal specifications required for estimation are obtained (S3-1: hull's principal specifications obtaining step) prior to the coefficient initial value obtaining step S3-2. In the hull principal particulars acquisition step S3-1, the hull principal particulars acquisition unit 11C of the computer 11 accepts input of the hull principal particulars from the user, or automatically reads them from the first storage unit 12 or the second storage unit 13. This makes it possible to acquire the hull principal particulars of the ship to be analyzed, and to obtain the initial coefficient values ​​using the simplified estimation formula.

[0045] Next, the initial values ​​of the coefficients acquired in the initial coefficient value acquisition step S3-2 are applied to the maneuver motion model, and the maneuver motion is calculated (S4: maneuver motion calculation step). In the maneuvering motion calculation step S4, the maneuvering motion calculation unit 11E of the computer 11 inputs the coefficients set in the coefficient initial value acquisition step S3-2 or in the coefficient update in the repetition step S8 described later, to the maneuvering motion model selected in the maneuvering motion model setting step S1, and calculates a motion time series of the turning angle, etc. by numerical simulation.

[0046] Next, a controllability index is calculated based on the calculation results of the control movement (S5: controllability index calculation step). In the controllability index calculation step S5, the controllability index calculation unit 11F of the computer 11 analyzes the time-series data of the controllability movements calculated in the controllability movement calculation step S4, and calculates the controllability index based on the analytical solution.

[0047] Next, an evaluation value is calculated based on the acquired maneuverability index data and the calculated maneuverability index (S6: evaluation value calculation step). In evaluation value calculation step S6, evaluation value calculation unit 11G of computer 11 calculates an evaluation value representing the quality of the coefficients used in the calculation in maneuvering motion calculation step S4. In evaluation value calculation step S6, for example, the sum of squares of the difference between the maneuverability index data acquired in maneuverability index data acquisition step S2 and the maneuverability index calculated in maneuverability index calculation step S5 is calculated as the evaluation value. Here, in the case where the absolute values ​​of the target values ​​differ greatly depending on the type of maneuverability index used to calculate the evaluation value, for example, when the target value of the controllability index a is 0.1 and the target value of the controllability index b is 10, if a simple sum of squares of differences (the sum of "the square of the difference between the target value and the calculated value of a" and "the square of the difference between the target value and the calculated value of b") is used as the evaluation value, an improvement in the calculated value of the controllability index b will contribute more to the improvement of the evaluation value than an improvement in the calculated value of the controllability index a. Therefore, by calculating the evaluation value in evaluation value calculation step S6 using "a weighted sum of squares of the difference between the controllability index data and the controllability index", it is possible to equalize the contribution of the type of controllability index to the improvement of the evaluation value. Note that when the weight is set to 1, it coincides with "the sum of squares of the difference between the controllability index data and the controllability index".

[0048] Next, the end of the process is determined based on the evaluation value calculation result and a predetermined end determination condition (S7: end determination step). In the end judgment step S7, the end judgment unit 11H of the computer 11 judges the end of the identification calculation of the maneuvering motion model coefficients. The end judgment unit 11H judges the end of the identification calculation of the maneuvering motion model coefficients based on one or more of the number of iterative calculations, the convergence of the calculation result of the evaluation value calculated in the evaluation value calculation step S6, and the magnitude of the evaluation value calculated in the evaluation value calculation step S6 relative to a threshold value. When the end judgment unit 11H judges the end of the identification calculation of the maneuvering motion model coefficients based on the number of iterative calculations, it judges that the end judgment is satisfied if a predetermined number of iterative calculations is reached, when the end judgment unit 11H judges the end of the identification calculation of the maneuvering motion model coefficients based on the convergence of the evaluation value, it judges that the end judgment is satisfied if the evaluation value converges, and when the end judgment unit 11H judges the end of the identification calculation of the maneuvering motion model coefficients based on a threshold value, it judges that the end judgment is satisfied if the evaluation value is smaller than the threshold value. This makes it possible to ensure the quality of the identification of the maneuvering motion model coefficients based on the predetermined number of iterative calculations, the convergence of the calculation result of the evaluation value, or the predetermined threshold value.

[0049] If it is determined in the termination determination step S7 that the termination determination is not satisfied ("No"), i.e., if the termination is denied, the repetition unit 11J of the computer 11 updates the coefficients of the maneuvering motion model based on the optimization algorithm and repeats the maneuvering motion calculation step S4, the maneuverability index calculation step S5, the evaluation value calculation step S6, and the termination determination step S7 (S8: repetition step) until the termination is determined to be affirmative. The repetition unit 11J updates the coefficients using an optimization algorithm (nonlinear programming (Nelder-Mead method), evolutionary computation (genetic algorithm, covariance matrix adaptive evolution strategy), simulated annealing, etc.). When a genetic algorithm is used as the optimization method in the repetition step S8, the optimization problem can be solved by the genetic algorithm, and the coefficients of the maneuvering motion model can be identified.

[0050] If it is determined in the termination determination step S7 that the termination determination is satisfied ("Yes"), i.e., if termination is confirmed, the identification coefficient output unit 11K outputs the most recently used coefficient as the identified coefficient to the output means 30 (S9: identification coefficient output step). In addition to the identified coefficients, the output to the output means 30 can output related information such as input information and information for each step. In this way, by identifying the coefficients of the maneuvering motion model using the maneuverability index (scalar quantity) known as a result of sea trials, etc., as a target value, it becomes possible to identify the coefficients of the maneuvering motion model of an individual ship required for performing a maneuvering motion simulation at a lower cost than the conventional method that requires a free running test, a restrained model test, and CFD calculations for coefficient identification. In recent years, along with technological developments, discussions on the formulation of international standards at the IMO (International Maritime Organization) have been progressing, and with the aim of practical application and dissemination of automatic navigation systems (automated systems for berthing and unberthing maneuvers, automatic navigation, etc.), which are elemental technologies for autonomous ships, the demand for maneuvering motion simulation for parameter tuning, etc. is expected to increase, and this embodiment can greatly contribute to the simplification and cost reduction of the simulation.

[0051] In the method of using the identified coefficients of the maneuvering motion model of the ship, the coefficients of the maneuvering motion model identified by the coefficient identification program can be used for simulation calculation of the maneuvering motion of the ship during automatic berthing or automatic navigation. In this case, it is also possible to link the computer 11 of the coefficient identification system with another computer that performs simulation calculation of the maneuvering motion of the ship during automatic berthing or automatic navigation. By using the coefficients of the maneuvering motion model identified using the coefficient identification program, it is possible to simplify and reduce the cost of maneuvering motion simulation calculations for parameter tuning and the like.

[0052] In addition, in the method of using the identified coefficients of the maneuvering motion model of the ship, the coefficients of the maneuvering motion model identified by the coefficient identification program can be used as control parameters for automatic berthing / unberthing maneuvering or automatic navigation of the ship. In this case, it is also possible to link the computer 11 of the coefficient identification system with a control means for performing automatic berthing / unberthing maneuvering or automatic navigation of the ship. This makes it possible to improve the accuracy of control of automatic berthing / unberthing maneuvering or automatic operation by using the coefficients of the maneuvering motion model identified using the coefficient identification program.

[0053] Here, the results of a comparison between this embodiment and the conventional method with respect to the cost required for coefficient identification of the maneuvering motion model will be shown. First, in the case of the conventional method, when the coefficients of the control motion model are identified using model tests and CFD calculations, the following costs are required: 1. Model test (1) Creating a model ship based on offset data (finished drawings at the time of construction) of the target ship - Cost of creating a model ship: Approximately 1 million yen / m (approximately 3 million yen for a 3m long model ship) - Time required to create the model ship: Approximately 3 months (if there are any adjustments to the offset data required to fit the test fixture, it may take longer) (2) Conducting confined model tests in a test tank - Test period: Approximately one week (assuming a drift angle β in the range of -20 to 20 degrees in 5 degree increments (9 conditions) × a dimensionless value r' of the turning angular velocity in the range of -0.8 to 0.8 in 0.2 increments (9 conditions)) - Test tank usage fee: Approximately 1 million yen / day - Required personnel: Approximately 3 people (3) Analysis of test results - Analysis period: about 3 days 2.CFD calculation (1) Creation of 3D mesh data based on the offset data of the target ship (assembly documents at the time of construction) - 3D mesh data creation time: about 1 day - 3D mesh data creation software usage fee: Approximately 3.5 million yen (2) Conducting CFD calculations simulating restraint model tests - Calculation time: Approximately 1 hour per condition (If calculations were performed using one computer under the same conditions as the model test described above, it would take approximately 80 hours. On the other hand, if a computer is prepared for each condition and calculations are performed in parallel, the calculation results can be obtained quickly, but the cost of procuring the computers will increase.) - CFD solver usage fee: Approximately 10 million yen - Computer price: Approximately 1 million yen per unit - Required personnel: 1 person (3) Analysis of calculation results - Analysis period: Approximately 3 days (same work as the restrained model test mentioned above)

[0054] In contrast, in this embodiment, it is possible to identify the coefficients of the control motion model in the following procedure and at the required cost. 1. Reference to maneuverability indexes obtained in sea trials, etc. - Time period: Immediately (read the values ​​on the sea trial report, etc.) - Cost: Basically free (no large-scale hardware or software required) 2. Coefficient identification of maneuvering motion model using the coefficient identification program according to this embodiment - Calculation time: several minutes to several hours - Computer price: Approximately 200,000 yen (this can be calculated using one computer with specifications suitable for general office work) - Required personnel: 1 person

[0055] In this way, by applying this embodiment to identify the coefficients of a maneuvering motion model, it is possible to significantly reduce human, financial, and time costs compared to identifying the coefficients of a maneuvering motion model using conventional methods.

[0056] In addition, the following comparison is made between the identification of coefficients using free sailing track data and time series data as teacher data and the handling of data in this embodiment. First, the data acquisition items and storage method for identifying coefficients using free sailing track data and time series data as training data are as follows. 1. Data Acquisition Items It is not specified what kind of steering time series should be used to acquire track or movement time series. 2. How we store your data When considering using the turning wake during the turning test of a sea trial as teacher data for free sailing wake data, for example, the sea trial report only contains a diagram showing the wake such as "Figure An1.4.2 Turning test" in the Rules for Steel Ships ("2023 Rules for Steel Ships Part B Appendix 2.3.1-1. Maneuvering performance test", Nippon Kaiji Kyokai, a general incorporated foundation, p201), so in order to use it as teacher data, it is necessary to transcribe it into numerical data such as two-dimensional position information. In other words, at present, when considering identifying coefficients using free navigation track data or time series data as training data, there is no established method for storing data in an easily usable format.

[0057] In contrast, the data acquisition items and storage method in this embodiment are as follows. 1. Data Acquisition Items It is a maneuverability index (scalar quantity) generally obtained during sea trials, etc., and is widely recognized internationally, including being adopted in the IMO maneuverability standard (MSC.137(76)) as an index for judging the maneuverability of a ship. 2. How we store your data The format is set out in the IMO explanatory document on maneuverability regulations (MSC / Circ.1053). Since the data is a scalar quantity, the numerical values ​​recorded in the sea trial report or the like can basically be used as they are as target values ​​for coefficient identification according to this embodiment.

[0058] In this way, the identification of the coefficients of the maneuvering motion model according to this embodiment allows for much clearer and easier data handling than the identification of the coefficients of the maneuvering motion model according to the conventional method.

[0059] FIG. 4 is a diagram showing a form of use in which the coefficient identification system for a ship maneuvering motion model is connected to an information and communication network such as the Internet. The coefficient identification system can be used independently without being connected to an information and communication network as shown in FIG. 2, and can also be used in conjunction with the outside via an information and communication network 50 as shown in FIG. 4, the database 40, company A, and engine B are located at a distance from the installation location of the system main body 10 of the coefficient identification system, and the computer 11 of the system main body 10 is connected to the database 40, the information processing device of company A, and the information processing device of engine B via an information and communication network 50. The information processing devices of company A and engine B each have an input means 20 and an output means 30. Company A and engine B are, for example, a shipping company or shipyard considering installing an automatic navigation system or the like on their own ships, a development company of an automatic navigation system, etc. In this usage form, the system main body 10 is connected to a main body input means 21 and a main body output means 31. The main body input means 21 and the main body output means 31 are basically used to start and stop the system and to check the operating status of the system. However, like the input means 20 and output means 30 provided in Company A and Engine B, the main body input means 21 can also be used to input maneuverability index data, hull main specifications, and coefficient initial values, and the main body output means 31 can be used to output identification coefficients, related information, and the like.

[0060] In this mode of use, the input means 20 and the output means 30 are provided at a remote location away from the computer 11 and connected to the computer 11 via an information and communication network 50, so that users at Company A and Engine B can identify the coefficients of the maneuvering motion model via the information and communication network 50 without having to go to the location where the system main body 10 is installed. It is also possible to provide a coefficient identification system on the cloud by connecting the system main body 10, the database 40, the information processing device of Company A, and the information processing device of Engine B to a virtual server. In this case, users of Company A and Engine B can identify maneuvering motion model coefficients on the cloud when performing maneuvering motion simulations on a computer for research and development of an automatic navigation system or for parameter tuning before installation on an actual ship. EXAMPLES

[0061] As a trial application example of the present invention, a trial calculation of the coefficients of a ship maneuvering motion model was carried out. In this trial calculation, the maneuvering motion model used in the numerical simulation, the coefficients to be identified, the optimization algorithm and other methods, and the maneuverability index to be set as the evaluation index were selected as follows. 1. Maneuvering motion model used in numerical simulation The maneuvering motion model selected was the KT model expressed by the following equation (3).

number

number

[0062] The method of setting the coefficients K and T to be identified in equation (4) is to estimate Y and N using the hull's principal dimensions shown in equation (5) below (K. Kijima, T. Katsuno, Y. Nakiri and Y. Furukawa: On the manoeuvring performance of a ship with the parameter of loading condition, Journal of the society of naval architects of Japan, Vol.168 (1990), pp.141-148.), and add a correction term to Y as shown in equation (6) below. * ,N * The coefficients K and T were calculated using the above.

number

number

[0063] FIG. 5 shows the results of the z-test maneuvering test. In the maneuvering motion calculation step S4, the time series of the heading angle in zig-zag maneuvering was calculated according to the KT model. In addition, in the maneuvering motion calculation step S4, the extreme values ​​of the time series of the turning angle calculated are analyzed, and the first over-pass angle, the second over-pass angle, the first over-pass time, and the second over-pass time, which are set as the maneuvering indexes, are obtained. In this way, in the maneuvering index calculation step S5, the maneuvering index is calculated using the following formula (7) based on the analytical solution of the zig-zag motion in the KT model (Kensaku Nomoto, Masashi Taguchi: On the Maneuverability of a Ship (2), Collection of Papers by the Society of Shipbuilders, No. 101 (1957), pp. 57-66.).

number

[0064] In the evaluation value calculation step S6, the error between the maneuverability index obtained in the maneuverability index calculation step S5 and the maneuverability index (target value) obtained in the sea trial was calculated. The evaluation value was set to the sum of squares of the difference from the target value (formula (8) below). In the genetic algorithm used in this trial calculation, the smaller the evaluation value, the better the individual is determined to be, with higher fitness.

number

[0065] In the termination judgment step S7, judgment was made based on the number of repeated calculations, and the calculation was deemed to be terminated when the number of repeated calculations reached 10,000 generations.

[0066] FIG. 6 is a diagram showing a genetic operation (coefficient update) method using a genetic algorithm. The generational change model used was “MGG (Minimal Generation Gap)” (Sato Hiroshi, Ono Isao, Kobayashi Shigenobu: Proposal and Evaluation of a Generational Change Model in Genetic Algorithms, Journal of the Japanese Society for Artificial Intelligence, Vol. 12, No. 5 (1997), pp. 734-744.), and was carried out using the following steps as shown in Figure 6(a). Step 1: Randomly select two individuals (parent individuals) from the current generation population. Step 2: Generate offspring individuals by crossover operation Step 3: Evaluate the fitness of offspring Step 4: Generate the next generation population by returning the best fitness individual and one randomly selected individual from the child and parent individuals to the current generation population. For the crossover operation in Step 2 (method of generating offspring individuals), we used "BLX-α" (LJ Eshleman and JD Schaffer: Real-Coded Genetic Algorithms and Interval-Schemata, Foundations of Genetic Algorithms, Vol.2 (1993), pp.187-202.) as shown in Figure 6(b). Figure 6(b) shows an image of the case where genes are represented by two variables.

[0067] The calculation conditions for this trial calculation are as follows: 1. Target ship: SR108 container ship (full scale) 2. Ship speed 20 knots (10.29 m / s), 10 / 10z test 3. Group size: 20 4. Number of children to be generated: 40 5. BLX-α coefficient: α=0.5 6. Number of repeated calculations (generations): 10,000 The calculation results are shown in Table 1 below. The target values ​​in Table 1 were taken from Hiroki Yasukawa and Yasuo Yoshimura: Naval and Ocean Engineering Series (3) Ship Motion Maneuvering Performance, Seizando Shoten (2012). [Table 1] As can be seen from Table 1, in this trial calculation, not only the maneuverability index that was explicitly given as an evaluation index of fitness, but also the coefficients K and T of the maneuvering model to be identified generally agreed with the target values. [Industrial Applicability]

[0068] By using the present invention to identify the coefficients of a maneuvering motion model, shipping companies and the like considering installing an automatic navigation system on their own ships can easily and at low cost perform maneuvering motion simulations on a computer for parameter tuning and the like before installing the system on an actual ship. In addition, the identified coefficients of the maneuvering motion model can be used for control of automatic berthing / unberthing maneuvering or automatic navigation of a ship. [Explanation of symbols]

[0069] 11. Computers 20 Input Method 30 Output Means 50 Information and Communication Network S1 Maneuvering motion model setting step S2 Maneuverability index data acquisition step S3-1 Hull Principal Specifications Acquisition Step S3-2 Initial coefficient value acquisition step S4 Maneuvering motion calculation step S5 Maneuverability index calculation step S6 Evaluation value calculation step S7 End judgment step S8 Repeat Step S9 Identification coefficient output step

Claims

1. A program for identifying coefficients of a ship steering motion model, comprising: On the computer, a maneuverability index data acquisition step of acquiring maneuverability index data of the vessel; a coefficient initial value acquisition step of acquiring initial values ​​of the coefficients of the set maneuvering motion model; a maneuver motion calculation step of calculating a maneuver motion by applying the obtained initial values ​​of the coefficients to the maneuver motion model; a controllability index calculation step of calculating a controllability index based on a result of the control motion calculation; an evaluation value calculation step of calculating an evaluation value based on the acquired maneuverability index data and the calculated maneuverability index; an end determination step of determining end based on a calculation result of the evaluation value and a predetermined end determination condition; a repeating step of updating the coefficients of the control motion model based on an optimization algorithm when the termination is denied, and repeating the control motion calculation step, the controllability index calculation step, the evaluation value calculation step, and the termination determination step; and an identification coefficient output step of outputting the most recently used coefficient as an identified coefficient if the termination is affirmative.

2. 2. The ship maneuvering motion model coefficient identification program according to claim 1, further comprising a maneuvering motion model setting step of setting the maneuvering motion model selected from a plurality of maneuvering motion models.

3. 3. The coefficient identification program for a ship maneuvering motion model as described in claim 2, characterized in that the maneuvering motion model is any one of a KT model, which is a first-order delay response model using gain and time constant, an MMG model, which is a model that expresses the fluid forces acting on the ship as a linear sum of components of the hull, propeller and rudder and their interference terms, and inputs this as an external force into the motion equation of the ship, and a low-speed maneuvering motion model.

4. 2. A coefficient identification program for a ship steering motion model as described in claim 1, characterized in that when a simplified estimation formula using hull main specifications as the initial values ​​is used in the coefficient initial value acquisition step, a hull main specifications acquisition step for acquiring the hull main specifications is further executed.

5. 2. The ship steering motion model coefficient identification program according to claim 1, wherein the maneuverability index data is sea trial data of the ship.

6. 6. The ship steering motion model coefficient identification program according to claim 5, wherein a scalar quantity of the sea trial data is used as the maneuverability index data.

7. 6. The ship maneuvering motion model coefficient identification program according to claim 5, further comprising:selecting data to be used from said sea trial data in accordance with said set maneuvering motion model.

8. 2. The ship steering motion model coefficient identification program according to claim 1, wherein the evaluation value in the evaluation value calculation step is calculated using a weighted sum of squares of a difference between the maneuverability index data and the maneuverability index.

9. 2. The ship steering motion model coefficient identification program according to claim 1, wherein the termination determination condition for determining the termination in the termination determination step is any one of the number of iterative calculations, convergence of the calculation result of the evaluation value, and a threshold value for the calculation result of the evaluation value.

10. 2. The ship steering motion model coefficient identification program according to claim 1, wherein a genetic algorithm is used as an optimization method for updating the coefficients in the repetitive step.

11. A method for utilizing identified coefficients of a ship's maneuvering motion model, characterized in that the coefficients of the ship's maneuvering motion model identified by the coefficient identification program for the ship's maneuvering motion model described in any one of claims 1 to 10 are used in simulation calculations of the ship's maneuvering motion for automatic docking / undocking maneuvering or automatic operation.

12. A method for utilizing identified coefficients of a ship's maneuvering motion model, characterized in that the coefficients of the ship's maneuvering motion model identified by the coefficient identification program for the ship's maneuvering motion model described in any one of claims 1 to 10 are used as control parameters for automatic docking / undocking maneuvering or automatic operation of the ship.

13. A system for identifying coefficients of a ship steering motion model comprising: a computer; an input means for inputting at least one of maneuverability index data and initial values ​​of coefficients to the computer; and an output means for outputting at least the identified coefficients from the computer, wherein the computer executes a coefficient identification program for a ship steering motion model according to any one of claims 1 to 10.

14. 14. The system for identifying coefficients of a ship steering motion model according to claim 13, wherein the input means and the output means are provided at remote locations and connected to the computer via an information and communication network.

15. The coefficient identification system for a ship maneuvering motion model as described in claim 13, characterized in that the computer is linked to another computer that performs simulation calculations of the maneuvering motion of the ship for automatic docking / undocking maneuvering or automatic navigation.

16. 14. The system for identifying coefficients of a ship steering motion model according to claim 13, characterized in that the computer is linked to a control means for performing automatic docking / undocking maneuvering or automatic navigation of the ship.

Images