CN116186907B - Method, system and medium for analyzing navigable state based on state of marine subsystem - Google Patents

Method, system and medium for analyzing navigable state based on state of marine subsystem Download PDF

Info

Publication number
CN116186907B
CN116186907B CN202310484257.8A CN202310484257A CN116186907B CN 116186907 B CN116186907 B CN 116186907B CN 202310484257 A CN202310484257 A CN 202310484257A CN 116186907 B CN116186907 B CN 116186907B
Authority
CN
China
Prior art keywords
subsystem
ship
sample
data
neural network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202310484257.8A
Other languages
Chinese (zh)
Other versions
CN116186907A (en
Inventor
肖剑波
俞翔
胡世峰
黎恒智
张振海
谢海燕
楼京俊
刘杰峰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Naval University of Engineering PLA
Original Assignee
Naval University of Engineering PLA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Naval University of Engineering PLA filed Critical Naval University of Engineering PLA
Priority to CN202310484257.8A priority Critical patent/CN116186907B/en
Publication of CN116186907A publication Critical patent/CN116186907A/en
Application granted granted Critical
Publication of CN116186907B publication Critical patent/CN116186907B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Computer Hardware Design (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Traffic Control Systems (AREA)
  • Navigation (AREA)

Abstract

The invention discloses a method, a system and a medium for analyzing a navigable status based on a status of a marine subsystem, wherein the method comprises the following steps: s1: acquiring ship navigation history data, wherein the ship navigation history data comprises subsystem history running state data and corresponding ship navigability states; s2: selecting a group of data from the ship navigation history data as a reference sample, acquiring current running state data of a ship subsystem as a sample to be tested, and inputting a pair of the subsystem history running state data in the reference sample and the sample to be tested into a trained twin neural network model to obtain two feature vectors output by the model; s3: calculating the similarity of the two feature vectors, judging whether the similarity meets the threshold requirement, if so, enabling the ship navigability state corresponding to the sample to be tested to be the same as the ship navigability state of the reference sample, and if not, entering step S4; s4: repeating the steps S2-S3 until the similarity meets the threshold requirement.

Description

Method, system and medium for analyzing navigable state based on state of marine subsystem
Technical Field
The invention relates to the field of computers, in particular to a method, a system and a medium for analyzing a navigable status based on a status of a marine subsystem.
Background
Ship seaworthiness is the risk performance and state that a ship hull, a ship plane, and the like can resist or reasonably foresee commonly occurring in contracted voyages in design, structure, and the like. Good navigability of the ship is of great importance for safe sailing of the ship and high quality completion of the transportation tasks during transportation.
How to evaluate the navigability of a ship in combination with the parameters of equipment or the running state of the equipment on the ship is a key problem. At present, a method for determining the weight of the overall navigability evaluation of the ship occupied by the state of each machine equipment parameter or subsystem by a human expert evaluation method is generally adopted, and then a fuzzy evaluation model is combined to realize the method for obtaining the navigability of the ship by the comprehensive evaluation result.
However, the method for evaluating the knock-out weights based on the expert puts a high requirement on the professional level of the expert, and in particular implementation, the number of the expert is also high, otherwise, the evaluation result lacks objectivity. Meanwhile, for different ships and different sailing tasks, the standards are different, and the reliability is lost by relying on expert evaluation methods.
Disclosure of Invention
To overcome the above-mentioned shortcomings of the prior art, the present invention provides a method, system and medium for analyzing a navigable status based on a status of a subsystem for a ship, which is used to solve at least one of the above-mentioned technical problems.
According to an aspect of the present invention, there is provided a method of analyzing a navigable status based on a status of a subsystem for a ship, comprising the steps of:
s1: acquiring ship navigation history data, wherein the ship navigation history data comprises subsystem history running state data and corresponding ship navigability states;
s2: selecting a group of data from ship navigation history data as a reference sample, wherein the reference sample comprises subsystem history running state data and corresponding ship navigability states; acquiring current running state data of a ship subsystem as a sample to be tested, and inputting a pair of the subsystem historical running state data in a reference sample and the sample to be tested into a trained twin neural network model to obtain two feature vectors output by the model;
s3: calculating the similarity of the two feature vectors, judging whether the similarity meets the threshold requirement, if so, enabling the ship navigability state corresponding to the sample to be tested to be the same as that of the reference sample, and if not, entering step S4;
s4: and selecting another group of data from the ship navigation history data excluding the selected reference samples as a new reference sample, and repeating the steps S2-S3 until the similarity meets the threshold requirement.
In the above technical solution, the subsystem comprises at least a propulsion system, a rudder and control system, a fuel system, a communication and navigation system and a cooling system. The status of each subsystem has a large influence on the navigability of the ship. Therefore, the historical data of ship navigation is firstly obtained, each ship navigability state corresponds to the operation state data of a plurality of groups of subsystems, and the operation state data of the subsystems refer to the data obtained by digitizing the operation states (including excellent, good, medium and poor) of the subsystems.
And obtaining two feature vectors of the reference sample and the sample to be tested by adopting the trained model, calculating the similarity between the sample to be tested and the reference sample based on the feature vectors, and searching out the reference sample with higher similarity with the sample to be tested, wherein the ship navigability state of the reference sample is the ship navigability state of the sample to be tested.
Further, the training process of the twin neural network model comprises the following steps:
constructing a positive sample pair and a negative sample pair based on the ship navigation history data, wherein the positive sample pair comprises two groups of subsystem history running state data and positive sample labels with the same ship navigability state; the negative sample pair comprises two groups of subsystem historical operation state data and negative sample labels with different ship navigability states;
creating a twin neural network model, and inputting the positive sample pair and the negative sample pair into the twin neural network model for training to obtain a trained twin neural network model.
Classifying the historical operating state data of the subsystems according to the ship navigability state, forming a pair of samples by two groups of the historical operating state data of the subsystems belonging to the same ship navigability state, and then giving a positive sample label to the pair of samples to obtain a positive sample pair. The historical running state data of two groups of subsystems belonging to different ship navigability states can also form a pair of samples, and then a negative sample label is given to the pair of samples, so that a negative sample pair can be obtained.
Further, the twin neural network model includes a first convolution layer, a second convolution layer, and a third convolution layer; the first convolution layer normalizes input data and filters the normalized data by 256 checks with the size of 5; the second convolution layer is provided with 384 kernels with the size of 3, takes the data vector filtered by the first convolution layer as input and outputs a group of feature mapping; the third convolution layer has 256 kernels of size 3, takes as input the feature map output by the second convolution layer and outputs another set of new feature maps.
The second convolution layer and the third convolution layer do not need any layer merging or normalization intervention, and the output of the first convolution layer is the output obtained by normalizing, merging and discarding the model input.
Further, an Adam optimizer is adopted to iteratively train the twin neural network model by continuously reducing the numerical value of the loss function through a gradient descent method.
Further, the iterative training specifically includes the following steps:
step 1: randomly initializing weights and biases;
step 2: inputting an input sample pair into a model, carrying out forward propagation through a twin neural network, and calculating to obtain a predicted output; the input sample pair refers to a positive sample pair and a negative sample pair;
step 3: calculating errors of the predicted output and the actual output to obtain a loss function;
step 4: calculating gradients of the loss function with respect to the weights and biases;
step 5: updating weights and biases using the negative direction of the gradient;
step 6: and repeating the steps 2-5 until the loss function reaches the minimum value.
Further, the loss function is:
；
wherein:in order to activate the function,y i for sample labels, when the sample pairs in the input model are positive sample pairsy i 1, when the sample pair in the input model is a negative sample pairy i Is 0;na number of all pairs of samples used to train the model;
activation functionThe method comprises the following steps: />；
Wherein: centering the sampleiSample numberkThe value of the operating state of the subsystem,centering the samplejSample numberkSubsystem operating state value +_>For an initial fixed vector of the network,ban initial fixed value for the network;Kis the number of subsystems;
。
further, the method for calculating the similarity comprises the following steps:
；
wherein:and->The two input vectors are respectively embedded with the representation of the high-dimensional space, the mathematical form is a vector, the data characteristic of the input vector is represented,Gwis an output of the model; />And->Respectively areAnd->Is used for the control of the degree of freedom of the composition,X1andX2respectively a sample to be detected and a reference sample;Kis the number of subsystems.
As another aspect of the present invention, there is provided a system for analyzing a navigable status based on a status of a subsystem for a ship, comprising:
and a data acquisition module: the method comprises the steps of acquiring ship navigation history data and current running state data of a ship subsystem, and constructing a positive sample pair and a negative sample pair based on the ship navigation history data;
model training module: the method is used for creating a twin neural network model and training the twin neural network model;
the navigability assessment module: the method is used for analyzing the current ship navigability state based on the current running condition data of the ship subsystem by adopting the trained twin neural network model.
In the technical scheme, the data acquisition module acquires ship navigation historical data, the acquired ship navigation historical data is used for constructing a positive sample pair and a negative sample pair required by model training, meanwhile, current running state data of a ship subsystem required by seaworthiness assessment is acquired, the data are respectively transmitted to the model training module and the seaworthiness assessment module, the model training module creates a twin neural network model and trains the model according to the received positive sample pair and the negative sample pair, a trained model is obtained, and the seaworthiness assessment module adopts the trained model to assess to obtain the current seaworthiness state of the ship.
Further, the model training module is further configured to complete the following steps: randomly initializing weights and biases; inputting an input sample pair into a model, carrying out forward propagation through a twin neural network, and calculating to obtain a predicted output; calculating errors of the predicted output and the actual output to obtain a loss function; calculating gradients of the loss function with respect to the weights and biases; updating weights and biases using the negative direction of the gradient to minimize the loss function; repeating the steps until the loss function reaches the minimum value.
As a further aspect of the present invention, there is provided a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a method of analyzing a navigable status based on a status of a marine subsystem.
Compared with the prior art, the invention has the beneficial effects that:
(1) The method for analyzing the navigable state based on the state of the marine subsystem, provided by the invention, can avoid the process of evaluating the weight of each parameter by an expert, and has the advantages of low cost and reliability compared with an evaluation system adopting a traditional expert evaluation method. Meanwhile, as the number of the ship subsystems is far less than that of all the machine equipment of the whole ship, the method can be used for more conveniently estimating the navigable state of the ship;
(2) According to the system for analyzing the navigability state based on the state of the marine subsystem, the data acquisition module acquires the ship navigation history data, the acquired ship navigation history data is used for constructing a positive sample pair and a negative sample pair required by model training, meanwhile, current running state data of the ship subsystem required by navigability evaluation are acquired, the data are respectively transmitted to the model training module and the navigability evaluation module, the model training module creates a twin neural network model and trains the model according to the received positive sample pair and the negative sample pair, a trained model is obtained, and the navigability evaluation module adopts the trained model to evaluate the current navigability state of the ship. When the system is used for analyzing the ship navigability state, only the ship navigation history data and the current running state of the subsystem are required to be collected, and then model analysis is adopted, so that artificial subjective evaluation is not required, the subjectivity of the artificial evaluation is avoided, the evaluation result is more objective and accurate, the evaluation efficiency is improved, and the labor cost is reduced.
Drawings
FIG. 1 is a flow chart of a method for analyzing a navigable status based on a status of a marine subsystem according to an embodiment of the invention;
FIG. 2 is a flowchart of an iterative training method according to an embodiment of the present invention;
fig. 3 is a system configuration diagram for analyzing a navigable status based on a status of a subsystem for a ship according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made in detail and with reference to the accompanying drawings, wherein it is apparent that the embodiments described are only some, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
Example 1
As shown in fig. 1, the present embodiment provides a method for analyzing a navigable status based on a status of a marine subsystem, including the following steps:
s1: acquiring ship navigation history data, wherein the ship navigation history data comprises subsystem history running state data and corresponding ship navigability states;
constructing a positive sample pair and a negative sample pair based on the ship navigation history data;
the positive sample pair comprises two groups of subsystem historical operation state data with the same ship navigability state and a positive sample label; the negative sample pair comprises two groups of subsystem historical running state data with different ship navigability states and a negative sample label.
In this embodiment, the ship navigability status is divided into four levels of good, medium and bad, the ship navigation history data includes multiple sets of subsystem history operation status data, each set of subsystem history operation status data refers to the data after the numerical analysis of the subsystem operation status rating, and the recording form of the subsystem operation status data is a data setC m =（c m1 ,c m2 ,..., c mk ) WhereinmFor the numbering of the subsystem operational state data sets,kis the number of the subsystem.
Constructing two sets of data in subsystem operation state data corresponding to the same ship navigability state into positive sample pairs, e.g. subsystem operation state data set C 1 =(c 11 ,c 12 ,...,c 1k ) And subsystem operational state data setC 2 =(c 21 ,c 22 ,...,c 2k ) The corresponding ship navigability states are all good, a positive sample pair can be constructed based on the two data sets:(C 1 、C 2 、1)wherein "1" is a positive sample tag; when subsystem operation state data setC 3 If the corresponding ship navigability state is good, the data set can be based onC 1 And a data setC 3 Constructing a negative sample pair:（C 1 ,C 3 ,0）wherein "0" is a negative sample label.
Creating a twin neural network model, and inputting a positive sample pair and a negative sample pair into the twin neural network model for training to obtain a trained twin neural network model;
the twin neural network model comprises a first convolution layer, a second convolution layer and a third convolution layer; the first convolution layer normalizes input data and filters the normalized data by 256 checks with the size of 5; the second convolution layer is provided with 384 kernels with the size of 3, takes the data vector filtered by the first convolution layer as input and outputs a group of feature maps, and the vector form of the feature maps is the same as that of the input data vector; the third convolution layer has 256 kernels of size 3, takes as input the feature map output by the second convolution layer and outputs another set of new feature maps, the vector form of which is the same as the vector form of the feature map input to the third convolution layer.
And adopting an Adam optimizer to iteratively train the twin neural network model by continuously reducing the numerical value of the loss function through a gradient descent method.
As shown in fig. 2, the iterative training specifically includes the following steps:
step 1: randomly initializing weights and biases;
step 2: inputting an input sample pair into a model, carrying out forward propagation through a twin neural network, and calculating to obtain a predicted output;
step 3: calculating errors of the predicted output and the actual output to obtain a loss function;
step 4: calculating gradients of the loss function with respect to the weights and biases;
step 5: updating the weight and the bias in the negative direction of the practical gradient;
step 6: and repeating the steps 2-5 until the loss function reaches the minimum value.
The loss function is:
；
wherein:in order to activate the function,y i for sample labels, when the sample pairs in the input model are positive sample pairsy i 1, when the sample pair in the input model is a negative sample pairy i Is 0;na number of all pairs of samples used to train the model;
activation functionThe method comprises the following steps: />；
Wherein:centering the sampleiSample numberkSubsystem operating state value +_>Centering the samplejSample numberkSubsystem operating state value +_>For an initial fixed vector of the network,ban initial fixed value for the network;Kis the number of subsystems;
。
s2: selecting a group of data from ship navigation history data as a reference sample, wherein the reference sample comprises subsystem history running state data and corresponding ship navigability states; and acquiring current running state data of the ship subsystem as a sample to be tested, and inputting a pair of the subsystem historical running state data in the reference sample and the sample to be tested into a trained twin neural network model to obtain two feature vectors output by the model.
For example: first, a subsystem operation state data set is obtained as a reference sampleC h =(c h1 ,c h2 ,...,c hk )And obtain the data setC h The corresponding ship navigability state is excellent, and then the current running state data set of the ship subsystem is obtained as a sample to be testedC 0 =(c 01 ,c 02 ,...,c 0k )Data setC h And a data setC 0 In the trained model as a pair of sample pairs, the model outputs a dataset C 0 Corresponding feature vectorG w (0)=（x 01 ,x 02 ,...,x 0k ）And data set C h Corresponding feature vectorG w (h)=(x h1 ,x h2 ,...,x hk )。
S3: and calculating the similarity of the two feature vectors, judging whether the similarity meets the threshold requirement, if so, enabling the ship navigability state corresponding to the sample to be tested to be the same as that of the reference sample, and if not, entering step S4.
The method for calculating the similarity comprises the following steps:
；
wherein:and->The two input vectors are respectively embedded with the representation of the high-dimensional space, the mathematical form is a vector, the data characteristic of the input vector is represented,Gwis an output of the model; />And->Are respectively->And->Is used for the control of the degree of freedom of the composition,X1andX2respectively a sample to be detected and a reference sample;Kis the number of subsystems.
In the present embodiment, whenE w When the value of (2) is 0.2 or less, the reference sample is describedC h Corresponding feature vector and sample to be testedC 0 The similarity of the corresponding feature vectors meets the threshold requirement, the similarity of the ship navigability states corresponding to the two samples is higher, and the ship navigability states corresponding to the samples to be tested can be better, ifE w Greater than 0.2, description reference sampleCCorresponding feature vector and sample to be testedC 0 And (4) if the similarity of the corresponding feature vectors does not meet the threshold requirement, namely, the similarity of the ship navigability states corresponding to the two samples is lower, and the current ship navigability state cannot be judged at the moment, the step (S4) is carried out.
S4: selecting another set of data from the ship voyage history data excluding the selected reference sample as a new reference sampleC q =（C q1 ,C q1 ,...,C qk ) Repeating the steps S2-S3 until the similarity meets the threshold requirement.
Re-selecting a set of subsystem operating state data sets from the voyage history data as another reference sampleC q Repeating the steps S2-S3, and judgingC q Corresponding feature vector and sample to be testedC 0 Whether the similarity of the corresponding feature vectors meets the threshold requirement or not, if so, terminating the test sampleC 0 Corresponding ship navigability stateC q If not, continuing to repeat the process until the similarity meets the threshold requirement, thereby obtaining the current ship navigability state.
Example 2
As shown in fig. 3, the present embodiment provides a system for estimating a navigable status based on a status of a subsystem for a ship, including:
and a data acquisition module: the method comprises the steps of acquiring ship navigation history data and current running state data of a ship subsystem, and constructing a positive sample pair and a negative sample pair based on the ship navigation history data;
model training module: the method is used for creating a twin neural network model and training the twin neural network model;
the navigability assessment module: the method is used for estimating the current ship navigability state based on the current running condition data of the ship subsystem by adopting the trained twin neural network model.
Further, the model training module adopts an Adam optimizer to iteratively train the twin neural network model by continuously reducing the value of the loss function through a gradient descent method.
Specifically, the model training module is further configured to complete the following steps: randomly initializing weights and biases; inputting an input sample pair into a model, carrying out forward propagation through a twin neural network, and calculating to obtain a predicted output; calculating errors of the predicted output and the actual output to obtain a loss function; calculating gradients of the loss function with respect to the weights and biases; the negative direction of the utility gradient updates the weights and biases to minimize the loss function; repeating the steps until the loss function reaches the minimum value.
Example 3
The present embodiment provides a computer storage medium having a computer program stored thereon, which when executed by a processor implements the method of analyzing a navigable status based on a status of a marine subsystem.
Those of ordinary skill in the art will appreciate that all or a portion of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, includes one or a combination of the steps of the method embodiments.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a medium if implemented in the form of software functional modules and sold or used as a stand-alone product.
The above-mentioned medium may be a read-only memory, a magnetic disk or an optical disk, etc.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced with equivalents; these modifications or substitutions do not depart from the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims (7)

1. The method for analyzing the navigable status based on the status of the marine subsystem is characterized by comprising the following steps:
s1: acquiring ship navigation history data, wherein the ship navigation history data comprises subsystem history running state data and corresponding ship navigability states;
s2: selecting a group of data from ship navigation history data as a reference sample, wherein the reference sample comprises subsystem history running state data and corresponding ship navigability states; acquiring current running state data of a ship subsystem as a sample to be tested, and inputting a pair of the subsystem historical running state data in a reference sample and the sample to be tested into a trained twin neural network model to obtain two feature vectors output by the model;
s3: calculating the similarity of the two feature vectors, judging whether the similarity meets the threshold requirement, if so, enabling the ship navigability state corresponding to the sample to be tested to be the same as that of the reference sample, and if not, entering step S4;
s4: selecting another group of data from the ship navigation history data excluding the selected reference sample as a new reference sample, and repeating the steps S2-S3 until the similarity meets the threshold requirement;
adopting an Adam optimizer to continuously reduce the numerical value of a loss function through a gradient descent method to carry out iterative training on the twin neural network model;
the iterative training specifically comprises the following steps:
step 1: randomly initializing weights and biases;
step 2: inputting an input sample pair into a model, carrying out forward propagation through a twin neural network, and calculating to obtain a predicted output;
step 3: calculating errors of the predicted output and the actual output to obtain a loss function;
step 4: calculating gradients of the loss function with respect to the weights and biases;
step 5: updating weights and biases using the negative direction of the gradient;
step 6: repeating the steps 2-5 until the loss function reaches the minimum value;
the loss function is:
wherein: ŷ i To activate the function, y i For sample labels, when the sample pairs in the input model are positive sample pairs y i 1, when the sample pair in the input model is a negative sample pair y i Is 0; n is the number of all pairs of samples used to train the model;
the activation function ŷ is:
wherein:the kth subsystem operating state value for the ith sample in the sample pair,/th sample pair>The kth subsystem operating state value for the jth sample of the sample pair,/for the sample pair>B is an initial fixed value of the network; k is the number of subsystems; />。
2. The method for analyzing the navigable status based on the status of the subsystem for a ship as recited in claim 1, wherein the training process of the twin neural network model comprises the steps of:
constructing a positive sample pair and a negative sample pair based on the ship navigation history data, wherein the positive sample pair comprises two groups of subsystem history running state data and positive sample labels with the same ship navigability state; the negative sample pair comprises two groups of subsystem historical operation state data and negative sample labels with different ship navigability states;
creating a twin neural network model, and inputting the positive sample pair and the negative sample pair into the twin neural network model for training to obtain a trained twin neural network model.
3. The method of analyzing a navigable status based on a status of a marine subsystem of claim 2, wherein the twin neural network model comprises a first convolution layer, a second convolution layer, and a third convolution layer; the first convolution layer normalizes input data and filters the normalized data by 256 checks with the size of 5; the second convolution layer is provided with 384 kernels with the size of 3, takes the data vector filtered by the first convolution layer as input and outputs a group of feature mapping; the third convolution layer has 256 kernels of size 3, takes as input the feature map output by the second convolution layer and outputs another set of new feature maps.
4. The method for analyzing the navigable status based on the status of the subsystem for a ship according to claim 1, wherein the similarity calculating method comprises the following steps:
，
wherein:and->Embedding representations of two input vectors into a high-dimensional space, respectively, in the mathematical form of vectors +.>Data features representing input vectors, gw being an output of the model; />And->Are respectively->And->X1 and X2 are the sample to be measured and the reference sample, respectively; k is the number of subsystems.
5. A system for analyzing a navigable status based on a status of a marine subsystem, for implementing the steps in the method for analyzing a navigable status based on a status of a marine subsystem as defined in any one of claims 1 to 4, comprising:
and a data acquisition module: the method comprises the steps of acquiring ship navigation history data and current running state data of a ship subsystem, and constructing a positive sample pair and a negative sample pair based on the ship navigation history data;
model training module: the method is used for creating a twin neural network model and training the twin neural network model;
the navigability assessment module: the method is used for analyzing the current ship navigability state based on the current running condition data of the ship subsystem by adopting the trained twin neural network model.
6. The marine subsystem state analysis seaworthiness state based system of claim 5, wherein the model training module is further configured to perform the steps of: randomly initializing weights and biases; inputting an input sample pair into a model, carrying out forward propagation through a twin neural network, and calculating to obtain a predicted output; calculating errors of the predicted output and the actual output to obtain a loss function; calculating gradients of the loss function with respect to the weights and biases; updating weights and biases using the negative direction of the gradient to minimize the loss function; repeating the steps until the loss function reaches the minimum value.
7. A computer storage medium, wherein the medium has stored thereon a computer program, wherein the computer program, when executed by a processor, implements the steps of the method of analyzing a navigable status based on a status of a marine subsystem as defined in any one of claims 1 to 4.
CN202310484257.8A 2023-05-04 2023-05-04 Method, system and medium for analyzing navigable state based on state of marine subsystem Active CN116186907B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310484257.8A CN116186907B (en) 2023-05-04 2023-05-04 Method, system and medium for analyzing navigable state based on state of marine subsystem

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310484257.8A CN116186907B (en) 2023-05-04 2023-05-04 Method, system and medium for analyzing navigable state based on state of marine subsystem

Publications (2)

Publication Number Publication Date
CN116186907A CN116186907A (en) 2023-05-30
CN116186907B true CN116186907B (en) 2023-09-15

Family

ID=86446660

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310484257.8A Active CN116186907B (en) 2023-05-04 2023-05-04 Method, system and medium for analyzing navigable state based on state of marine subsystem

Country Status (1)

Country Link
CN (1) CN116186907B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111275736A (en) * 2020-01-07 2020-06-12 中国科学院大学 Unmanned aerial vehicle video multi-target tracking method based on target scene consistency
CN111723632A (en) * 2019-11-08 2020-09-29 珠海达伽马科技有限公司 Ship tracking method and system based on twin network
CN112232269A (en) * 2020-10-29 2021-01-15 南京莱斯网信技术研究院有限公司 Twin network-based intelligent ship identity identification method and system
CN113177264A (en) * 2021-05-11 2021-07-27 北京邮电大学 Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network
CN113886973A (en) * 2021-10-25 2022-01-04 江苏远望仪器集团有限公司 Ship navigational speed processing method, device and processing equipment based on virtual-real mapping
CN113987251A (en) * 2021-10-26 2022-01-28 武汉理工大学 Method, system, equipment and storage medium for establishing ship face characteristic database
EP3974640A1 (en) * 2019-05-22 2022-03-30 National Institute of Maritime, Port and Aviation Technology Ship main engine monitoring method, main engine monitoring system, main engine state prediction system, and operation status prediction system

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3974640A1 (en) * 2019-05-22 2022-03-30 National Institute of Maritime, Port and Aviation Technology Ship main engine monitoring method, main engine monitoring system, main engine state prediction system, and operation status prediction system
CN111723632A (en) * 2019-11-08 2020-09-29 珠海达伽马科技有限公司 Ship tracking method and system based on twin network
CN111275736A (en) * 2020-01-07 2020-06-12 中国科学院大学 Unmanned aerial vehicle video multi-target tracking method based on target scene consistency
CN112232269A (en) * 2020-10-29 2021-01-15 南京莱斯网信技术研究院有限公司 Twin network-based intelligent ship identity identification method and system
CN113177264A (en) * 2021-05-11 2021-07-27 北京邮电大学 Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network
CN113886973A (en) * 2021-10-25 2022-01-04 江苏远望仪器集团有限公司 Ship navigational speed processing method, device and processing equipment based on virtual-real mapping
CN113987251A (en) * 2021-10-26 2022-01-28 武汉理工大学 Method, system, equipment and storage medium for establishing ship face characteristic database

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Modeling and simulation on movement of air cushion vehicle based on neural network;L. Xu-Ming, H. Da-Bin, X. Jian-Bo and H. Jin-Hui;《2017 2nd International Conference on Robotics and Automation Engineering (ICRAE)》;513-516 *
SiamFPN: A Deep Learning Method for Accurate and Real-Time Maritime Ship Tracking;Y. Shan, X. Zhou, S. Liu, Y. Zhang and K. Huang,;《IEEE Transactions on Circuits and Systems for Video Technology》;第31卷(第1期);315-325 *
基于层次注意力孪生网络的船舶身份甄别;苏俊杰,兰培真;《大连海事大学学报》;第48卷(第2期);31-39 *
舰船航行状态信息集中监测系统的研究与设计方法;李文辉,王璨;《舰船科学技术》;第43卷(第10期);31-33 *

Also Published As

Publication number Publication date
CN116186907A (en) 2023-05-30

Similar Documents

Publication Publication Date Title
CN111753101B (en) Knowledge graph representation learning method integrating entity description and type
CN110941734B (en) Depth unsupervised image retrieval method based on sparse graph structure
CN109902714B (en) Multi-modal medical image retrieval method based on multi-graph regularization depth hashing
WO2021240707A1 (en) Data classification system, data classification method, and recording medium
CN111353373A (en) Correlation alignment domain adaptive fault diagnosis method
CN112000689A (en) Multi-knowledge graph fusion method based on text analysis
CN116186907B (en) Method, system and medium for analyzing navigable state based on state of marine subsystem
CN113792753A (en) Dynamic hypergraph neural network classification method and system
CN112802011A (en) Fan blade defect detection method based on VGG-BLS
CN117112992A (en) Fault diagnosis method for polyester esterification stage
CN116127846A (en) Intelligent evaluation method and system for anti-rolling technology
CN108898157B (en) Classification method for radar chart representation of numerical data based on convolutional neural network
CN115329139A (en) Knowledge graph embedding method based on group representation theory
CN114742199A (en) Attention mechanism-based neural network macro architecture searching method and system
CN114528906A (en) Fault diagnosis method, device, equipment and medium for rotary machine
CN113269678A (en) Fault point positioning method for contact network transmission line
CN112580797A (en) Incremental learning method of multi-mode multi-label prediction model
CN110728292A (en) Self-adaptive feature selection algorithm under multi-task joint optimization
CN116703001B (en) Oil consumption prediction method and system of intelligent ship, intelligent ship and medium
CN117233520B (en) AUV propulsion system fault detection and evaluation method based on improved Sim-GAN
CN113627494B (en) Gesture data preprocessing method of ship gesture six-degree-of-freedom system
KR102665023B1 (en) Method and Apparatus for Determining a Monitoring Variable for a Target System
CN114383846B (en) Bearing composite fault diagnosis method based on fault label information vector
CN115964661B (en) Rotary machine fault diagnosis method and system based on domain-by-domain network
CN109255722B (en) Complex network hierarchical analysis system and method based on neighbor topology

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant