CN114462444A - Navigation state identification method and system for polar region inspection ship - Google Patents
Navigation state identification method and system for polar region inspection ship Download PDFInfo
- Publication number
- CN114462444A CN114462444A CN202111576087.3A CN202111576087A CN114462444A CN 114462444 A CN114462444 A CN 114462444A CN 202111576087 A CN202111576087 A CN 202111576087A CN 114462444 A CN114462444 A CN 114462444A
- Authority
- CN
- China
- Prior art keywords
- signal
- time
- real
- power system
- state
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000007689 inspection Methods 0.000 title description 2
- 238000012545 processing Methods 0.000 claims abstract description 62
- 238000011835 investigation Methods 0.000 claims abstract description 44
- 238000012549 training Methods 0.000 claims abstract description 9
- 238000001914 filtration Methods 0.000 claims description 33
- 230000007613 environmental effect Effects 0.000 claims description 32
- 238000000354 decomposition reaction Methods 0.000 claims description 27
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 230000010354 integration Effects 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 5
- 238000005183 dynamical system Methods 0.000 claims 2
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000011176 pooling Methods 0.000 description 8
- 230000004913 activation Effects 0.000 description 6
- 230000001537 neural effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013488 ordinary least square regression Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/048—Activation functions
-
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Signal Processing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a navigation state identification method and a navigation state identification system for a polar region investigation ship, which relate to the technical field of computers and comprise the following steps: step S1, collecting real-time power system operation signals and real-time environment state signals generated in the process of sailing of a polar investigation ship; step S2, processing the real-time power system operation signal and the real-time environment state signal to obtain corresponding real-time comprehensive navigation data; and step S3, inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training, and processing to obtain a real-time navigation state of the polar investigation ship. The method has the beneficial effect of improving the navigation state identification efficiency of the polar region investigation ship.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a method and a system for identifying the navigation state of a polar region investigation ship.
Background
The polar region investigation ship is a professional marine investigation ship special for marine investigation and investigation in north and south polar sea areas. Due to the fact that navigation conditions of the south and north sea areas are severe, people need to pay attention to the navigation state of the polar investigation ship all the time, and corresponding measures can be taken timely when the navigation state of the polar investigation ship is abnormal. The relationship among the state of the power system, the state of the environment and the navigation state of the polar investigation ship in the navigation process is deeply explored through the operation data of the power system, the state data of the environment, the navigation log and the like.
Currently, in order to analyze the sailing state of a polar exploration ship, an engineer with a rich polar exploration experience needs to analyze the power system running state data and the environment state data generated when the polar exploration ship runs to judge the sailing state of the polar exploration ship.
Since the data volume related to the power system operation data and the environmental state data generated during the operation of the polar exploration ship is large, a data analyst is required to have strong data sensitivity and data engineering capability when analyzing the data characteristics of a large number of parameters. In addition, when the data of the power system penetrated by the polar region survey and the environmental filling data are collected, the time lines corresponding to various data can not be accurately mapped, and the navigation state of the polar region survey ship is not easy to recognize in real time.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a method for identifying the navigation state of a polar region investigation ship, which comprises the following steps:
step S1, collecting real-time power system operation signals and real-time environment state signals generated in the process of sailing of a polar investigation ship;
step S2, processing according to the real-time power system operation signal and the real-time environment state signal to obtain corresponding real-time comprehensive navigation data;
and step S3, inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training, and processing to obtain a real-time navigation state of the polar region investigation ship.
Preferably, the step S2 includes:
step S21, respectively carrying out noise reduction processing on the real-time power system operation signal and the real-time environment state signal to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal;
step S22, respectively performing feature dimension reduction processing on the power system noise reduction signal and the environment state noise reduction signal to obtain a corresponding power system dimension reduction signal and a corresponding environment state dimension reduction signal;
step S23, respectively carrying out standardization processing on the power system dimension reduction signal and the environment state dimension reduction signal to obtain a corresponding power system standard signal and an environment state standard signal, and processing according to the power system standard signal and the environment state standard signal to obtain the real-time comprehensive navigation data.
Preferably, in step S21, the real-time power system operating signal and the real-time environment state signal are respectively input to a median filter to obtain a corresponding power system filtering signal and a corresponding environment state filtering signal, and then the power system filtering signal and the environment state filtering signal are respectively input to a pre-configured third-order low-pass butterworth filter to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal.
Preferably, in step S22, the power system noise reduction signal and the environmental state noise reduction signal are respectively subjected to adaptive normalization processing to obtain the corresponding power system dimension reduction signal and the corresponding environmental state dimension reduction signal.
Preferably, the step S23 includes:
step S231, respectively carrying out normal standardization processing on the power system dimension reduction signal and the environmental state dimension reduction signal to obtain a power system standard signal and an environmental state standard signal;
and step S232, processing according to the power system standard signal and the environment state standard signal to obtain the real-time comprehensive navigation data.
Preferably, before executing the step S3, the method includes:
step A1, acquiring historical power system operation signals and historical environment state signals which are respectively generated by the polar region investigation ship in a plurality of power operation modes, wherein the power operation modes comprise an ice breaking mode, a pitch propeller mode and a combination mode;
step A2, respectively constructing corresponding time sequence data sets according to the historical power system operation signals and the historical environment state signals, and respectively extracting and obtaining frequency domain signals and time domain signals according to the time sequence data sets;
step A3, respectively constructing corresponding time domain weak classifiers according to the time domain signals, and respectively constructing corresponding frequency domain weak classifiers according to the frequency domain signals;
step A4, obtaining and storing the identification model according to the integration of each time domain weak classifier and each frequency domain weak classifier;
in step S3, the real-time integrated navigation data is input to the identification model integrated by each of the time domain weak classifiers and each of the frequency domain weak classifiers, and the identification model outputs the corresponding power operation mode and outputs the power operation mode as the real-time navigation state.
Preferably, the step a2 includes:
step A21, respectively sampling each time sequence data set according to a preset sampling frequency to obtain a plurality of sub data sets;
step A22, performing empirical mode decomposition on the historical power system operation signals and the historical environment state signals in each subdata set respectively to obtain corresponding historical power system decomposition signals and corresponding historical environment state decomposition signals;
step A23, filtering the historical power system decomposition signal and the historical environment state decomposition signal respectively to obtain a corresponding historical power system filtering signal and a corresponding historical environment state filtering signal, then superposing the historical power system filtering signal and the historical environment state filtering signal to obtain a corresponding one-dimensional time domain signal, decomposing the one-dimensional time domain signal to obtain the frequency domain signal and outputting the frequency domain signal, and reconstructing the one-dimensional time domain signal to obtain a two-dimensional time domain signal as the time domain signal to output.
Preferably, there is also provided a navigation state recognition system for polar exploration ships, applying the navigation state recognition method according to any one of the above items, the navigation state recognition system comprising:
the acquisition module is used for acquiring real-time power system operation signals and real-time environment state signals generated in the sailing process of a polar region investigation ship;
the preprocessing module is connected with the acquisition module and used for processing the real-time power system running signal and the real-time environment state signal to obtain a corresponding real-time environment state signal and outputting the real-time environment state signal;
and the recognition module is connected with the preprocessing module and used for inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training and processing the recognition model to obtain a real-time navigation state of the polar region investigation ship.
Preferably, the preprocessing module comprises:
the noise reduction unit is used for respectively carrying out noise reduction processing on the real-time power system operation signal and the real-time environment state signal to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal;
the dimensionality reduction unit is connected with the noise reduction unit and is used for respectively carrying out characteristic dimensionality reduction processing on the power system noise reduction signal and the environmental state noise reduction signal to obtain a corresponding power system dimensionality reduction signal and a corresponding environmental state dimensionality reduction signal;
and the standardization unit is connected with the dimension reduction unit and used for respectively carrying out standardization processing on the power system dimension reduction signal and the environment state dimension reduction signal to obtain a corresponding power system standard signal and an environment state standard signal and processing according to the power system standard signal and the environment state standard signal to obtain the real-time comprehensive navigation data.
Preferably, the system further comprises a model building module connected to the identification module, wherein the model building module comprises:
the system comprises a collecting unit, a data processing unit and a control unit, wherein the collecting unit is used for acquiring historical power system running signals and historical environment state signals generated by the polar region investigation ship in a plurality of power operation modes, and the power operation modes comprise an ice breaking mode, a pitch propeller mode and a combination mode;
the signal extraction unit is connected with the acquisition unit and used for respectively constructing corresponding time sequence data sets according to the historical power system operation signals and the historical environment state signals and respectively extracting and obtaining frequency domain signals and time domain signals according to the time sequence data sets;
the signal processing unit is connected with the signal extraction unit and used for respectively constructing corresponding time domain weak classifiers according to the time domain signals and respectively constructing corresponding frequency domain weak classifiers according to the frequency domain signals;
the model construction unit is connected with the signal processing unit and used for obtaining and storing the identification model according to the integration of each time domain weak classifier and each frequency domain weak classifier;
and in the identification model, inputting the real-time comprehensive navigation data into the identification model obtained by integrating each time domain weak classifier and each frequency domain weak classifier, and outputting the corresponding power operation mode as the real-time navigation state by the identification model.
The technical scheme has the following advantages or beneficial effects: the comprehensive navigation data are obtained by processing the real-time power system operation signals and the real-time environment state signals generated in the navigation process of the polar investigation ship, and then the comprehensive navigation data are input into the identification model to obtain the real-time navigation state, so that the time and labor consumed by manually identifying the real-time navigation state of the polar investigation ship are reduced, and the identification efficiency of the navigation state identification is improved because the polar investigation ship does not need to manually analyze the data characteristics of a large number of parameters.
Drawings
FIG. 1 is a flow chart of a navigation status recognition method according to a preferred embodiment of the present invention;
FIG. 2 is a flowchart illustrating a step S2 of the navigation state recognition method according to the preferred embodiment of the present invention;
FIG. 3 is a flowchart illustrating a step S23 of the navigation state recognition method according to the preferred embodiment of the present invention;
FIG. 4 is a flowchart illustrating the construction of a recognition model according to a preferred embodiment of the present invention;
FIG. 5 is a flowchart illustrating a step A2 of the navigation state recognition method according to the preferred embodiment of the present invention;
FIG. 6 is a network structure diagram of a first neural convolutional network model according to a preferred embodiment of the present invention;
FIG. 7 is a network structure diagram of a second convolutional neural network model in a preferred embodiment of the present invention;
FIG. 8 is a control schematic diagram of a navigation status recognition system in accordance with a preferred embodiment of the present invention;
FIG. 9 is a schematic diagram illustrating the detailed control of the signal extraction unit according to the preferred embodiment of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present invention is not limited to the embodiment, and other embodiments may be included in the scope of the present invention as long as the gist of the present invention is satisfied.
In a preferred embodiment of the present invention, based on the above problems in the prior art, there is provided a method for identifying a voyage state of a polar research ship, as shown in fig. 1, including:
step S1, collecting real-time power system operation signals and real-time environment state signals generated in the process of sailing of a polar investigation ship;
step S2, processing according to the real-time power system operation signal and the real-time environment state signal to obtain corresponding real-time comprehensive navigation data;
and step S3, inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training, and processing to obtain a real-time navigation state of the polar investigation ship.
Specifically, in the embodiment, in the process of identifying the real-time navigation state of the polar investigation ship, the real-time power system operation signal and the real-time environment state signal are combined, and the real-time environment state signal reflecting the environmental factors is used as one of identification bases of the real-time navigation state of the polar investigation ship, so that the real-time navigation state output by the identification model is more consistent with the actual navigation state corresponding to the polar investigation ship, and the applicability and reliability of the navigation state identification method are improved.
When the polar investigation ship is used for judging the real-time navigation state, the related data volume is large, the data is redundant, and workers for judging the real-time navigation state are required to have hard professional ability, so that when the polar investigation ship executes a polar investigation task, suitable workers need to be equipped, more consumed workers can incline to the polar investigation task, and convenience is brought to the polar investigation.
In a preferred embodiment of the present invention, as shown in fig. 2, step S2 includes:
step S21, respectively carrying out noise reduction processing on the real-time power system operation signal and the real-time environment state signal to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal;
step S22, respectively performing feature dimension reduction processing on the power system noise reduction signal and the environmental state noise reduction signal to obtain a corresponding power system dimension reduction signal and a corresponding environmental state dimension reduction signal;
and step S23, respectively carrying out standardization processing on the power system dimension reduction signal and the environmental state dimension reduction signal to obtain a corresponding power system standard signal and an environmental state standard signal, and processing according to the power system standard signal and the environmental state standard signal to obtain real-time comprehensive navigation data.
In a preferred embodiment of the present invention, in step S21, the real-time driving system operating signal and the real-time environment state signal are respectively input to a median filter to obtain a corresponding driving system filtering signal and a corresponding environment state filtering signal, and then the driving system filtering signal and the environment state filtering signal are respectively input to a pre-configured three-order low-pass butterworth filter to obtain a corresponding driving system noise reduction signal and a corresponding environment state noise reduction signal.
Specifically, in this embodiment, the gain of the nth order butterworth low-pass filter can be expressed by the following formula:
where n denotes the order of the Butterworth low-pass filter, ωcRepresenting a turning frequency, omegacApproximately equal to the frequency, omega, at which the amplitude drops to-3 dBpWhich represents the passband edge frequencies, is shown,denotes | H (ω) & gtnon2The value of the passband edge.
In a preferred embodiment of the present invention, in step S22, the adaptive normalization processing is performed on the power system noise reduction signal and the environmental state noise reduction signal respectively to obtain a corresponding power system dimension reduction signal and a corresponding environmental state dimension reduction signal.
Specifically, in this embodiment, the formula of the adaptive normalization is:
In another embodiment, the adaptive normalization formula is adjusted by selecting inconsistent comprehensive navigation data for the existing variables to obtain the following adjusted formula:
wherein the content of the first and second substances,represents a rightThe weight of the steel is heavy,are coefficients derived by the ordinary least squares method.
In a preferred embodiment of the present invention, as shown in fig. 3, step S23 includes:
step S231, respectively carrying out normal standardization processing on the power system dimension reduction signal and the environmental state dimension reduction signal to obtain a power system standard signal and an environmental state standard signal;
and step S232, processing according to the standard signal of the power system and the standard signal of the environmental state to obtain real-time comprehensive navigation data.
Specifically, in this embodiment, the normal normalization process is used to balance the weight fraction of the data with different dimensions, and the formula of the normal normalization is as follows:
wherein, x represents the comprehensive navigation data after dimensionality reduction, mu represents a mean value, and sigma represents a standard deviation. The formula for calculating σ is as follows:
where N represents a number of samples.
In a preferred embodiment of the present invention, before executing step S3, as shown in fig. 4, the method includes:
a1, acquiring historical power system running signals and historical environment state signals which are respectively generated by a polar region investigation ship in a plurality of power operation modes, wherein the power operation modes comprise an ice breaking mode, a pitch propeller mode and a combination mode;
step A2, respectively constructing corresponding time sequence data sets according to historical power system operation signals and historical environment state signals, and respectively extracting and obtaining frequency domain signals and time domain signals according to the time sequence data sets;
step A3, respectively constructing corresponding time domain weak classifiers according to the time domain signals, and respectively constructing corresponding frequency domain weak classifiers according to the frequency domain signals;
step A4, obtaining and storing an identification model according to the integration of each time domain weak classifier and each frequency domain weak classifier;
in step S3, the real-time integrated navigation data is input to the recognition models integrated by the time domain weak classifiers and the frequency domain weak classifiers, and the recognition models output the corresponding dynamic operation modes and output as the real-time navigation states.
Specifically, in this embodiment, before executing step a2, the method further includes preprocessing the historical operating signals and the historical environmental status signals to facilitate the subsequent construction of the corresponding time-series data set.
In step a1, an expert with experience of judging the navigation state for more than 2 years manually marks the power operation mode, wherein the marking content comprises an ice breaking mode, a fixed pitch propeller mode and a combination mode, and the historical power system operation signals and the historical environment state signals are classified into the corresponding power operation modes.
In the preferred embodiment of the present invention, as shown in fig. 5, step a2 includes:
step A21, sampling each time sequence data set according to a preset sampling frequency to obtain a plurality of sub data sets;
step A22, performing empirical mode decomposition on the historical power system operation signals and the historical environment state signals in each subdata set respectively to obtain corresponding historical power system decomposition signals and corresponding historical environment state decomposition signals;
step A23, respectively filtering the historical power system decomposition signal and the historical environment state decomposition signal to obtain a corresponding historical power system filtering signal and a corresponding historical environment state filtering signal, then superposing the historical power system filtering signal and the historical environment state filtering signal to obtain a corresponding one-dimensional time domain signal, decomposing the one-dimensional time domain signal to obtain a frequency domain signal and outputting the frequency domain signal, and reconstructing the one-dimensional time domain signal to obtain a two-dimensional time domain signal as a time domain signal to output.
Specifically, in the present embodiment, the time-series data set S is set to { S } according to the sampling frequency f(i)And (3) carrying out sample collection on | i ═ 1, 2.., N }, so as to obtain a plurality of sub data sets.
When each subdata set is subjected to empirical mode decomposition, the data set s is subjected to empirical mode decomposition(i)Decomposition into m decomposed signalsWherein, the j-th decomposition signal is:
wherein, beta is a penalty term coefficient, alpha is a constant term, and w(i)As a data set s(i)The center frequency of (a) is,is the center frequency of the jth decomposed signal,the calculation formula of (a) is as follows:
subsequently, the frequency domain signal is extracted: decomposing the m signalsAfter Gaussian filtering processing, superposition processing is carried out to form an intermediate process signal P(i)Then decomposing the decomposed signal based on wavelet transform to obtain frequency domain signal V(i);
Then, extracting a time domain signal: taking a minimum period as a breakpoint to transmit the intermediate process signal P(i)Reconstruction into a two-dimensional time-domain signal T (i.e., a one-dimensional time-domain signal)(i)Two-dimensional time domain signal T(i)And outputting the time domain signal obtained by extraction.
Step a3 includes a time domain weak classifier constructing process and a frequency domain weak classifier constructing process.
The process of constructing the time-domain weak classifier is as follows, as shown in fig. 6:
from a two-dimensional time-domain signal T(i)Training a first neural convolution network model to obtain a corresponding time domain weak classifier
Wherein the first neural convolutional network model comprises:
a first input layer 100, where an input time domain sample size is 64 × 16, and the number of channels is 1;
a first convolution layer 200 connected to the first input layer 100, wherein the convolution kernel size is 3 × 3, the step size is 1, the zero padding pad is 1, the feature map size is 64 × 16, the feature map depth is 6, and relu is an activation function;
a first pooling layer 300 connected to the first convolution layer 200, wherein the feature size is 32 × 8, the step size is 2, and the zero padding pad is 0;
a second convolution layer 201 connected to the first input layer 100, where the convolution kernel size is 3 × 3, the step size is 1, the zero padding pad is 1, the feature map size is 32 × 8, the feature map depth is 24, and relu is an activation function;
a second pooling layer 301 connected to the second convolution layer 201, wherein the feature size is 16 × 4, the step size is 2, and the zero padding pad is 0;
a first fully-connected layer 400, which is connected to the first pooling layer 300 and the second pooling layer 301, respectively, and has a neuron number of 64, a Dropout parameter of 0.5, and an activation function threshold of 0.5;
the first output layer 500 is connected to the first full connection layer 400, and has an output category of 4, which corresponds to the ice breaking mode, the fixed pitch propeller mode, and the combination mode, respectively.
The process of constructing the frequency domain weak classifier is as follows, as shown in fig. 7:
from the frequency-domain signal V(i)Training a second neural convolution network model to obtain a corresponding frequency domain weak classifier
Wherein the second neural convolutional network model comprises:
a second input layer 101 having an input frequency domain sample size of 28 × 28 and a number of channels of 1;
a third convolution layer 202 connected to the second input layer 101, where the convolution kernel size is 5 × 5, the step size is 1, the zero padding pad is 0, the feature map size is 24 × 24, the feature map depth is 6, and relu is an activation function;
a third pooling layer 302 connected to the third convolutional layer 202, having a feature size of 12 × 12, a step size of 2, and a zero padding pad of 0;
a fourth convolution layer 203 connected to the second input layer 101, where the convolution kernel size is 5 × 5, the step size is 1, the zero padding pad is 0, the feature map size is 8 × 8, the feature map depth is 24, and relu is an activation function;
a fourth pooling layer 303 connected to the fourth convolution layer 203, having a feature size of 4 × 4, a step size of 2, and a zero padding pad of 0;
a second fully connected layer 401, which is connected to the third pooling layer 302 and the fourth pooling layer 303, respectively, and sets the number of neurons to 36, the Dropout parameter to 0.5, and the threshold of the activation function to 0.5;
and the second output layer 501 is connected with the second full connection layer 401, has an output category of 3, and corresponds to the ice breaking mode, the fixed pitch propeller mode and the combined mode respectively.
And then integrating the 2f time domain weak classifiers and the frequency domain weak classifiers to form an identification model, wherein the formula is as follows:
wherein the content of the first and second substances,is a set of coefficients of the time-domain weak classifier,is the coefficient set of the frequency domain weak classifier.
In a preferred embodiment of the present invention, there is further provided a navigation state recognition system for polar exploration of a ship, which applies any one of the above navigation state recognition methods, as shown in fig. 8, the navigation state recognition system includes:
the acquisition module 1 is used for acquiring real-time power system operation signals and real-time environment state signals generated in the process of sailing of a polar region survey ship;
the preprocessing module 2 is connected with the acquisition module 1 and is used for processing the real-time power system operation signal and the real-time environment state signal to obtain a corresponding real-time environment state signal and outputting the real-time environment state signal;
and the recognition module 3 is connected with the preprocessing module 2 and used for inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training and processing the recognition model to obtain a real-time navigation state of the polar investigation ship.
In a preferred embodiment of the present invention, the preprocessing module 2 includes:
the noise reduction unit 21 is configured to perform noise reduction processing on the real-time power system operating signal and the real-time environment state signal respectively to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal;
the dimension reduction unit 22 is connected with the noise reduction unit 21 and is used for respectively performing feature dimension reduction processing on the power system noise reduction signal and the environmental state noise reduction signal to obtain a corresponding power system dimension reduction signal and a corresponding environmental state dimension reduction signal;
and the standardization unit 23 is connected with the dimension reduction unit 22 and is used for respectively carrying out standardization processing on the dimension reduction signal of the power system and the dimension reduction signal of the environmental state to obtain a corresponding standard signal of the power system and a corresponding standard signal of the environmental state, and processing the standard signals according to the standard signal of the power system and the standard signal of the environmental state to obtain real-time comprehensive navigation data.
In a preferred embodiment of the present invention, the system further includes a model building module 4, the connection identification module 3, and the model building module 4 includes:
the acquisition unit 41 is used for acquiring historical power system operation signals and historical environment state signals generated by the polar region investigation ship in a plurality of power operation modes, wherein the power operation modes comprise an ice breaking mode, a pitch propeller mode and a combination mode;
the signal extraction unit 42 is connected to the acquisition unit 41, and is configured to respectively construct corresponding time sequence data sets according to the historical operating signals of the power system and the historical environmental status signals, and respectively extract and obtain each frequency domain signal and each time domain signal according to each time sequence data set;
the signal processing unit 43 is connected to the signal extracting unit 42, and is configured to respectively construct corresponding time-domain weak classifiers according to the time-domain signals, and respectively construct corresponding frequency-domain weak classifiers according to the frequency-domain signals;
the model construction unit 44 is connected with the signal processing unit 43 and is used for obtaining and storing an identification model according to the integration of each time domain weak classifier and each frequency domain weak classifier;
and in the identification model, inputting the real-time comprehensive navigation data into the identification model obtained by integrating each time domain weak classifier and each frequency domain weak classifier, and outputting a corresponding dynamic operation mode as a real-time navigation state by the identification model.
Specifically, in the present embodiment, as shown in fig. 9, the signal extraction unit 42 includes:
the sampling sub 421 unit is configured to sample each time sequence data set according to a preset sampling frequency to obtain a plurality of sub data sets;
the decomposition subunit 422 is connected to the sampling subunit 421, and is configured to perform empirical mode decomposition on the historical power system operation signal and the historical environment state signal in each sub data set respectively to obtain a corresponding historical power system decomposition signal and a corresponding historical environment state decomposition signal;
and the processing subunit 423 is connected to the decomposition subunit 422, and is configured to perform filtering processing on the historical power system decomposition signal and the historical environment state decomposition signal respectively to obtain a corresponding historical power system filtering signal and a corresponding historical environment state filtering signal, then perform superposition processing on the historical power system filtering signal and the historical environment state filtering signal to obtain a corresponding one-dimensional time domain signal, perform decomposition on the one-dimensional time domain signal to obtain a frequency domain signal, output the frequency domain signal, and reconstruct the one-dimensional time domain signal to obtain a two-dimensional time domain signal, which is output as the time domain signal.
In conclusion, the corresponding comprehensive navigation data is obtained by processing the real-time power system operation signal and the real-time environment state signal generated in the navigation process of the polar region investigation ship, then the comprehensive navigation data is input into the identification model integrated by each time domain weak classifier and each frequency domain weak classifier, and the real-time navigation state of the polar region investigation ship is output by the identification model, so that the time and labor consumed for identifying the real-time navigation state of the polar region investigation ship are reduced, and the identification efficiency of the navigation state identification is improved.
While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.
Claims (10)
1. A method for identifying the navigation state of a polar region exploration ship is characterized by comprising the following steps:
step S1, collecting real-time power system operation signals and real-time environment state signals generated in the process of sailing of a polar investigation ship;
step S2, processing according to the real-time power system operation signal and the real-time environment state signal to obtain corresponding real-time comprehensive navigation data;
and step S3, inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training, and processing to obtain a real-time navigation state of the polar region investigation ship.
2. The navigation state recognition method according to claim 1, wherein the step S2 includes:
step S21, respectively carrying out noise reduction processing on the real-time power system operation signal and the real-time environment state signal to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal;
step S22, respectively performing feature dimension reduction processing on the power system noise reduction signal and the environment state noise reduction signal to obtain a corresponding power system dimension reduction signal and a corresponding environment state dimension reduction signal;
step S23, respectively carrying out standardization processing on the power system dimension reduction signal and the environment state dimension reduction signal to obtain a corresponding power system standard signal and an environment state standard signal, and processing according to the power system standard signal and the environment state standard signal to obtain the real-time comprehensive navigation data.
3. The navigation state recognition method of claim 2, wherein in step S21, the real-time driving system operating signal and the real-time environment state signal are respectively input to a median filter to obtain a corresponding driving system filtering signal and a corresponding environment state filtering signal, and then the driving system filtering signal and the environment state filtering signal are respectively input to a pre-configured third-order low-pass butterworth filter to obtain a corresponding driving system noise reduction signal and a corresponding environment state noise reduction signal.
4. The navigation state recognition method according to claim 2, wherein in step S22, the dynamical system noise reduction signal and the environmental state noise reduction signal are respectively subjected to adaptive normalization processing to obtain the dynamical system dimension reduction signal and the environmental state dimension reduction signal.
5. The navigation state recognition method according to claim 2, wherein the step S23 includes:
step S231, respectively carrying out normal standardization processing on the power system dimension reduction signal and the environmental state dimension reduction signal to obtain a power system standard signal and an environmental state standard signal;
and step S232, processing according to the power system standard signal and the environment state standard signal to obtain the real-time comprehensive navigation data.
6. The navigation state recognition method according to claim 1, wherein before performing the step S3, the method includes:
step A1, acquiring historical power system operation signals and historical environment state signals which are respectively generated by the polar region investigation ship in a plurality of power operation modes, wherein the power operation modes comprise an ice breaking mode, a pitch propeller mode and a combination mode;
step A2, respectively constructing corresponding time sequence data sets according to the historical power system operation signals and the historical environment state signals, and respectively extracting and obtaining frequency domain signals and time domain signals according to the time sequence data sets;
step A3, respectively constructing corresponding time domain weak classifiers according to the time domain signals, and respectively constructing corresponding frequency domain weak classifiers according to the frequency domain signals;
step A4, obtaining and storing the identification model according to the integration of each time domain weak classifier and each frequency domain weak classifier;
in step S3, the real-time integrated navigation data is input to the identification model integrated by each of the time domain weak classifiers and each of the frequency domain weak classifiers, and the identification model outputs the corresponding power operation mode and outputs the power operation mode as the real-time navigation state.
7. The navigation state recognition method of claim 6, wherein the step A2 includes:
step A21, respectively sampling each time sequence data set according to a preset sampling frequency to obtain a plurality of sub data sets;
step A22, performing empirical mode decomposition on the historical power system operation signals and the historical environment state signals in each subdata set respectively to obtain corresponding historical power system decomposition signals and corresponding historical environment state decomposition signals;
step A23, filtering the historical power system decomposition signal and the historical environment state decomposition signal respectively to obtain a corresponding historical power system filtering signal and a corresponding historical environment state filtering signal, then superposing the historical power system filtering signal and the historical environment state filtering signal to obtain a corresponding one-dimensional time domain signal, decomposing the one-dimensional time domain signal to obtain the frequency domain signal and outputting the frequency domain signal, and reconstructing the one-dimensional time domain signal to obtain a two-dimensional time domain signal as the time domain signal to output.
8. A navigation state recognition system of a polar exploration ship, characterized in that the navigation state recognition method according to any one of claims 1 to 7 is applied, the navigation state recognition system comprising:
the acquisition module is used for acquiring real-time power system operation signals and real-time environment state signals generated in the sailing process of a polar region investigation ship;
the preprocessing module is connected with the acquisition module and used for processing the real-time power system running signal and the real-time environment state signal to obtain a corresponding real-time environment state signal and outputting the real-time environment state signal;
and the recognition module is connected with the preprocessing module and used for inputting the real-time comprehensive navigation data into a recognition model obtained by pre-training and processing the recognition model to obtain a real-time navigation state of the polar region investigation ship.
9. The navigation state recognition system of claim 8, wherein the preprocessing module comprises:
the noise reduction unit is used for respectively carrying out noise reduction processing on the real-time power system operation signal and the real-time environment state signal to obtain a corresponding power system noise reduction signal and a corresponding environment state noise reduction signal;
the dimensionality reduction unit is connected with the noise reduction unit and is used for respectively carrying out characteristic dimensionality reduction processing on the power system noise reduction signal and the environmental state noise reduction signal to obtain a corresponding power system dimensionality reduction signal and a corresponding environmental state dimensionality reduction signal;
and the standardization unit is connected with the dimension reduction unit and used for respectively carrying out standardization processing on the power system dimension reduction signal and the environment state dimension reduction signal to obtain a corresponding power system standard signal and an environment state standard signal and processing according to the power system standard signal and the environment state standard signal to obtain the real-time comprehensive navigation data.
10. The navigation state recognition system of claim 8, further comprising a model building module coupled to the recognition module, the model building module comprising:
the system comprises a collecting unit, a data processing unit and a control unit, wherein the collecting unit is used for acquiring historical power system running signals and historical environment state signals generated by the polar region investigation ship in a plurality of power operation modes, and the power operation modes comprise an ice breaking mode, a pitch propeller mode and a combination mode;
the signal extraction unit is connected with the acquisition unit and used for respectively constructing corresponding time sequence data sets according to the historical power system operation signals and the historical environment state signals and respectively extracting and obtaining frequency domain signals and time domain signals according to the time sequence data sets;
the signal processing unit is connected with the signal extraction unit and used for respectively constructing corresponding time domain weak classifiers according to the time domain signals and respectively constructing corresponding frequency domain weak classifiers according to the frequency domain signals;
the model construction unit is connected with the signal processing unit and used for obtaining and storing the identification model according to the integration of each time domain weak classifier and each frequency domain weak classifier;
and in the identification model, inputting the real-time comprehensive navigation data into the identification model obtained by integrating each time domain weak classifier and each frequency domain weak classifier, and outputting the corresponding power operation mode as the real-time navigation state by the identification model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111576087.3A CN114462444B (en) | 2021-12-21 | Navigation state identification method and system of polar survey ship |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111576087.3A CN114462444B (en) | 2021-12-21 | Navigation state identification method and system of polar survey ship |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114462444A true CN114462444A (en) | 2022-05-10 |
CN114462444B CN114462444B (en) | 2024-07-09 |
Family
ID=
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106327610A (en) * | 2016-08-27 | 2017-01-11 | 南通中远川崎船舶工程有限公司 | Intelligent ship for arctic navigation |
CN108313236A (en) * | 2018-01-24 | 2018-07-24 | 深圳远航股份有限公司 | A kind of ship's navigation method for early warning and system |
CN110532613A (en) * | 2019-07-26 | 2019-12-03 | 中国船舶重工集团公司第七一九研究所 | Ship power system operation mode recognition method and device |
CN110737267A (en) * | 2019-09-30 | 2020-01-31 | 智慧航海(青岛)科技有限公司 | Multi-objective optimization method for unmanned ships and intelligent comprehensive management and control system for unmanned ships |
CN111259949A (en) * | 2020-01-13 | 2020-06-09 | 山东超越数控电子股份有限公司 | Fault identification model construction method, model and identification method for aircraft environmental control system |
US20200365035A1 (en) * | 2019-05-13 | 2020-11-19 | Hoseo University Academic Cooperation Foundation | Method and system for recognition of objects near ship by using deep neural network |
CN112001450A (en) * | 2020-08-27 | 2020-11-27 | 山东超越数控电子股份有限公司 | Ship inertial navigation system fault identification method |
CN112163461A (en) * | 2020-09-07 | 2021-01-01 | 中国海洋大学 | Underwater target identification method based on multi-mode fusion |
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106327610A (en) * | 2016-08-27 | 2017-01-11 | 南通中远川崎船舶工程有限公司 | Intelligent ship for arctic navigation |
CN108313236A (en) * | 2018-01-24 | 2018-07-24 | 深圳远航股份有限公司 | A kind of ship's navigation method for early warning and system |
US20200365035A1 (en) * | 2019-05-13 | 2020-11-19 | Hoseo University Academic Cooperation Foundation | Method and system for recognition of objects near ship by using deep neural network |
CN110532613A (en) * | 2019-07-26 | 2019-12-03 | 中国船舶重工集团公司第七一九研究所 | Ship power system operation mode recognition method and device |
CN110737267A (en) * | 2019-09-30 | 2020-01-31 | 智慧航海(青岛)科技有限公司 | Multi-objective optimization method for unmanned ships and intelligent comprehensive management and control system for unmanned ships |
CN111259949A (en) * | 2020-01-13 | 2020-06-09 | 山东超越数控电子股份有限公司 | Fault identification model construction method, model and identification method for aircraft environmental control system |
CN112001450A (en) * | 2020-08-27 | 2020-11-27 | 山东超越数控电子股份有限公司 | Ship inertial navigation system fault identification method |
CN112163461A (en) * | 2020-09-07 | 2021-01-01 | 中国海洋大学 | Underwater target identification method based on multi-mode fusion |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102721545B (en) | Rolling bearing failure diagnostic method based on multi-characteristic parameter | |
CN111523509B (en) | Equipment fault diagnosis and health monitoring method integrating physical and depth expression characteristics | |
CN102607845B (en) | Bearing fault characteristic extracting method for redundantly lifting wavelet transform based on self-adaptive fitting | |
Li et al. | Mechanical fault diagnosis based on redundant second generation wavelet packet transform, neighborhood rough set and support vector machine | |
CN111880157B (en) | Method and system for detecting target in radar image | |
CN111783558A (en) | Satellite navigation interference signal type intelligent identification method and system | |
CN113707176A (en) | Transformer fault detection method based on acoustic signal and deep learning technology | |
CN110333076B (en) | Bearing fault diagnosis method based on CNN-Stacking | |
Li et al. | Life grade recognition method based on supervised uncorrelated orthogonal locality preserving projection and K-nearest neighbor classifier | |
CN102539159A (en) | Fault diagnosis method for valve mechanism of diesel engine | |
Bhadane et al. | Bearing fault identification and classification with convolutional neural network | |
CN104504403A (en) | Rotating machinery failure forecasting method based on scattering conversion | |
Fong et al. | Mean shift clustering-based analysis of nonstationary vibration signals for machinery diagnostics | |
CN107478418A (en) | A kind of rotating machinery fault characteristic automatic extraction method | |
CN111855202A (en) | Gear box fault diagnosis method and system | |
CN107705256A (en) | A kind of forward direction Vehicular video image enchancing method semantic based on contact net | |
CN113707175B (en) | Acoustic event detection system based on feature decomposition classifier and adaptive post-processing | |
CN115452376A (en) | Bearing fault diagnosis method based on improved lightweight deep convolution neural network | |
CN111458145A (en) | Cable car rolling bearing fault diagnosis method based on road map characteristics | |
CN114330430A (en) | Elevator fault judgment method and system based on big data characteristic analysis | |
CN114462444A (en) | Navigation state identification method and system for polar region inspection ship | |
CN114462444B (en) | Navigation state identification method and system of polar survey ship | |
CN112151067A (en) | Passive detection method for digital audio tampering based on convolutional neural network | |
CN115061196B (en) | Micro-seismic signal identification method based on empirical mode decomposition (IMF) guidance | |
CN115563480A (en) | Gear fault identification method for screening octave geometric modal decomposition based on kurtosis ratio coefficient |
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 |