CN113657021A - Marine measurement period evaluation method based on BP neural network - Google Patents
Marine measurement period evaluation method based on BP neural network Download PDFInfo
- Publication number
- CN113657021A CN113657021A CN202110799971.7A CN202110799971A CN113657021A CN 113657021 A CN113657021 A CN 113657021A CN 202110799971 A CN202110799971 A CN 202110799971A CN 113657021 A CN113657021 A CN 113657021A
- Authority
- CN
- China
- Prior art keywords
- output
- neural network
- hidden layer
- marine
- water depth
- 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.)
- Pending
Links
- 238000005259 measurement Methods 0.000 title claims abstract description 57
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 48
- 238000011156 evaluation Methods 0.000 title abstract description 15
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 51
- 238000000034 method Methods 0.000 claims abstract description 48
- 230000008859 change Effects 0.000 claims abstract description 36
- 238000013210 evaluation model Methods 0.000 claims abstract description 13
- 238000004458 analytical method Methods 0.000 claims abstract description 12
- 230000005284 excitation Effects 0.000 claims abstract description 10
- 238000012876 topography Methods 0.000 claims abstract description 3
- 238000012549 training Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 14
- 210000002569 neuron Anatomy 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 abstract description 14
- 238000004364 calculation method Methods 0.000 abstract description 11
- 238000003062 neural network model Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 abstract description 3
- 230000006872 improvement Effects 0.000 description 9
- 238000001514 detection method Methods 0.000 description 8
- 238000013499 data model Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000011002 quantification Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000013178 mathematical model Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 229910000906 Bronze Inorganic materials 0.000 description 1
- 101100313164 Caenorhabditis elegans sea-1 gene Proteins 0.000 description 1
- 238000013142 basic testing Methods 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 239000010974 bronze Substances 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- KUNSUQLRTQLHQQ-UHFFFAOYSA-N copper tin Chemical compound [Cu].[Sn] KUNSUQLRTQLHQQ-UHFFFAOYSA-N 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- 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
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C13/00—Surveying specially adapted to open water, e.g. sea, lake, river or canal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Hydrology & Water Resources (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a marine measurement period evaluation method based on a BP neural network. The method comprises the following steps: collecting the water depth change, port grade, scale, water depth change influence degree, AIS track analysis and shoreline topography of a sea area to be detected; inputting the data into a pre-established and trained marine measurement period evaluation model to obtain a sea area measurement period to be measured; the ocean measurement period evaluation model adopts an optimized and improved BP neural network structure. The method can effectively combine the past marine measurement period evaluation experience with the BP neural network model, is consistent with the actual situation of the water area of the marine coastal port channel in China, and realizes a scientific and quantitative marine measurement period evaluation method; and simultaneously, the characteristics of marine measurement data and the number of samples are considered, the BP neural network is optimized and improved, the selected hidden layer excitation function has a good calculation effect after testing, and the obtained final model has uniqueness.
Description
Technical Field
The invention relates to the technical field of marine surveying and mapping, in particular to a marine measurement period evaluation method based on a BP neural network.
Background
Ocean surveying is a direct source of chart data compiling production, and the accuracy and timeliness of chart data are directly related to the safety of personnel and ships. However, marine surveying is a capital, time, and labor intensive task, and surveying in the whole sea area is a task that is impossible to accomplish in a short period of time. According to the international sea-way survey organization (IHO), the current global ocean has accurately measured water area less than 15%, and even the measured area in 200 meters along the ocean is less than 50%. Under the background, how to solve the contradiction between insufficient conditions of capital, time and manpower and the requirement of navigation users is a problem that marine surveying needs to be solved urgently.
The nature of the problem is a multi-conditional decision analysis and optimization problem. The key point of the solution is to use the existing resources in important, urgent, large-demand, heavy-traffic and obviously-changed natural conditions water areas, namely when and how long to measure the same water area, so as to achieve the maximum utilization of the resources. However, the current marine survey cycle is mainly evaluated by expert scoring and working experience according to the change of natural conditions such as water depth and shoreline and the demand of a sea chart, and an objective, scientific and quantitative method is not formed.
An artificial neural network is a mathematical model that applies a structure similar to brain neurosynaptic connections for information processing. The artificial neural network is applied to solve the decision analysis and optimization problem under multiple conditions, and existing experience and knowledge are quantized to form a relatively fixed data model. In the artificial neural network, the BP neural network has arbitrary complex pattern classification capability and excellent multidimensional function mapping capability, and solves the problems of XOR and other problems which cannot be solved by a simple perceptron. Structurally, the BP network has an input layer, a hidden layer, and an output layer; basically, the BP algorithm calculates the minimum value of an objective function by using a network error square as the objective function and adopting a gradient descent method. The invention simultaneously considers the characteristics of marine measurement data and the maximum sample number, optimizes and improves the BP neural network, and the selected hidden layer excitation function has good calculation effect through testing, so that the obtained final model has uniqueness.
Disclosure of Invention
Aiming at the problem of work decision analysis under the condition of limited marine measurement resources, the invention aims to overcome the defects of the prior art and provides a marine measurement period evaluation method based on a BP neural network. Under the existing conditions of capital, manpower and time, the ocean surveying work is optimized, and resources are reasonably distributed.
In order to achieve the above object, the present invention provides a marine measurement cycle assessment method based on a BP neural network, the method including:
collecting the water depth change, port grade, scale, water depth change influence degree, AIS track analysis and shoreline topography of a sea area to be detected;
inputting the data into a pre-established and trained marine measurement period evaluation model to obtain a sea area measurement period to be measured;
the ocean measurement period evaluation model adopts an optimized and improved BP neural network structure.
As an improvement of the method, the input of the ocean surveying period evaluation model is water depth change, port grade, scale, influence degree of water depth change, AIS trajectory analysis and shoreline terrain, and the output is a surveying period; the marine measurement cycle assessment model comprises an input layer, a hidden layer and an output layer.
As an improvement of the above method, the method further comprises a training step of the marine survey cycle assessment model; the method specifically comprises the following steps:
establishing a training set;
setting network initialization parameters;
inputting training set data into a BP neural network;
calculating hidden layer output;
calculating a prediction output from the hidden layer output;
calculating a network prediction error based on the expectation and the output prediction output;
updating the network connection weight according to the network prediction error;
updating a network node threshold value according to the network prediction error;
and giving a learning rate and a neuron excitation function, and adjusting the number of input layer nodes, the number of hidden layer nodes and the number of output layer nodes of the network to obtain a BP neural network meeting constraint conditions, so as to obtain a trained marine measurement period evaluation model.
As an improvement of the above method, the establishing a training set; the method specifically comprises the following steps:
respectively selecting a plurality of groups of sample data of a north sea area, an east sea area and a south sea area of a Chinese coastal port channel, wherein each group of sample data comprises a scale, port ship flow, port throughput, water depth change conditions, water depth change range, influence degree of water depth change on a measurement period, shoreline topographic data and whether important navigation paths are included;
the water depth control method comprises the following steps of classifying the water depth into a large class, a middle class and a small class according to port properties, and classifying the water depth control method into the large class, the middle class and the small class according to the value range of water depth change; quantifying and scoring the ruler data;
scoring the port ship flow according to the range value of the ship flow, the passenger ship flow or the oil ship flow;
grading the port throughput according to the range value;
dividing into three types of large, medium and small according to the value range of the passenger ship circumferential flow; quantifying the water depth change condition;
quantifying whether the important air routes are contained;
respectively carrying out normalization processing on the size of the water depth change range, the influence degree of the water depth change on the measurement period, and the change conditions of the shoreline and the terrain;
and forming a group of records by the quantitative grading data and the normalized data, and taking the corresponding marine measurement period as a label to obtain a plurality of groups of training data so as to establish a training set.
As an improvement of the foregoing method, the network initialization parameter specifically includes: the number of input layer nodes, the number of hidden layer nodes, the number of output layer nodes, and the connection weights between the input layer, hidden layer and output layer neurons, the hidden layer threshold and the output layer threshold.
As an improvement of the foregoing method, the calculating the hidden layer output specifically includes:
calculating the output H of the jth node of the hidden layer according to the formulajComprises the following steps:
where l is the number of hidden layer nodes, i represents the ith input layer node, xiRepresenting the ith input vector, ωijRepresenting the connection weight between the ith input level node and the jth hidden level node, ajThe threshold value representing the j-th hidden layer node is f (-) which is a hidden layer excitation function and satisfies the following formula:
wherein y is an intermediate variable.
As an improvement of the above method, the computing of the prediction output from the hidden layer output; the method specifically comprises the following steps:
calculating the predicted output O of the kth node of the output layer according to the formulakComprises the following steps:
in the formula, wjkAs the connection weights between the j hidden layer nodes and the kth output layer node neurons, bkAnd m is the number of output layer nodes.
As an improvement to the above method, the calculating of the net prediction error is based on the expectation and the output prediction output; the method specifically comprises the following steps:
calculating the network prediction error e according tokComprises the following steps:
ek=Yk-Ok k=1,2,…m
in the formula, YkRepresenting the desired output of the kth output level node, and m is the number of output level nodes.
As an improvement of the above method, the network connection weight is updated according to the network prediction error; the method specifically comprises the following steps:
wjk-new=wjk+ηHjek j=1,2,…,l k=1,2,…m
in the formula, ωij-newThe updated connection weight between the ith input layer node and the jth hidden layer node is calculated, eta is the learning rate, wjk-newAnd (4) updating the connection weight between the j-th hidden layer node and the k-th output layer node neuron, wherein l is the number of hidden layer nodes, and m is the number of output layer nodes.
As an improvement of the above method, the updating of the network node threshold value is based on a network prediction error; the method specifically comprises the following steps:
bk-new=bk+ek k=1,2,…m
in the formula, aj-newFor updated threshold of the jth hidden layer node, bkThreshold for the kth output level node, bk-newFor the updated threshold of the kth output layer node, l is the number of hidden layer nodes, and m is the number of output layer nodes.
Compared with the prior art, the invention has the advantages that:
1. the method can effectively combine the past marine measurement period evaluation experience with the BP neural network model, is consistent with the actual situation of the water area of the marine coastal port channel in China, and realizes a scientific and quantitative marine measurement period evaluation method;
2. the invention simultaneously considers the characteristics of marine measurement data and the maximum sample quantity possible by the embodiment, optimizes and improves the BP neural network, and the selected hidden layer excitation function has good calculation effect through testing, so that the obtained final model has uniqueness.
Drawings
FIG. 1 is a flow chart of the marine survey cycle assessment method based on BP neural network of the present invention;
FIG. 2 is a diagram showing the prediction of the results of the basemeasurements in the sea area of the North sea;
FIG. 3 is a diagram showing the prediction of the results of the basic test in the sea area of the east China sea;
FIG. 4 is a diagram showing the prediction of the results of the basemeasurements in the sea area of the south China sea;
FIG. 5 is a diagram showing the prediction of the results of the test in the sea area of the North sea;
FIG. 6 is a diagram showing the prediction of the test results in the sea area of the east China sea;
FIG. 7 is a diagram showing the prediction of the test results in the sea area of the south China sea.
Detailed Description
The technical scheme of the invention is that factors influencing the marine cycle, such as water depth change, port grade, a scale, water depth change influence degree, AIS track analysis, shoreline terrain and the like, are learned through a BP neural network, an improved BP neural network structure and a hidden layer excitation function are established, a data model for evaluating the marine measurement cycle is formed, the model has uniqueness, and the method specifically comprises the following steps:
selecting factors of the ocean period, including water depth change, port grade, a scale, the influence degree of the water depth change, AIS trajectory analysis and shoreline terrain, and designing a BP neural network structure.
And (2) initializing the network. Determining the number n of nodes of a network input layer, the number l of nodes of a hidden layer, the number m of nodes of an output layer, initializing connection weights Wij and Wjk between neurons of the input layer, the hidden layer and the output layer, initializing a threshold value a of the hidden layer and a threshold value b of the output layer, and setting a learning rate and a neuron excitation function according to a system input and output sequence (X, Y).
And (3) outputting and calculating by a hidden layer. And calculating the hidden layer output H according to the input vector X, the connection weight Wij between the input layer and the hidden layer and the threshold a of the hidden layer.
Where l is the number of hidden layer nodes, i represents the ith input layer node, xiRepresenting the ith input vector, ωijRepresenting the connection weight between the ith input level node and the jth hidden level node, ajThe threshold value representing the j-th hidden layer node is f (-) which is a hidden layer excitation function and satisfies the following formula:
wherein y is an intermediate variable.
And (4) outputting the layer output calculation. And calculating the prediction output O of the BP neural network according to the hidden layer output H, the connection weight Wjk and the threshold b.
And (5) error calculation. And calculating the network prediction error e according to the network prediction output O and the expected output Y.
ek=Yk-Ok k=1,2,…m
And (6) updating the weight. Updating the network connection weight omega according to the network prediction error eij,wjk。
wjk-new=wjk+ηHjek j=1,2,…,l k=1,2,…m
Where η is the learning rate.
And (7) updating the threshold value. And updating the network node threshold values a and b according to the network prediction error e.
bk-new=bk+ek k=1,2,…m
And (8) judging whether algorithm iteration is constrained or not, if not, returning to the step 2.
And (9) outputting the marine survey cycle data model with the constraint in the step (8).
The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and examples.
Example 1
The technical solution of the present invention can be implemented by a person skilled in the art using computer software technology. The following provides a detailed description of embodiments of the invention, taken in conjunction with the accompanying drawings.
As shown in fig. 1, the embodiment provides a method for evaluating a marine survey cycle, which decomposes natural and social factors affecting the marine cycle into an input layer, a hidden layer and an output layer for learning simulation by using complex pattern classification capability and excellent multidimensional function mapping capability of a BP neural network, and obtains a corresponding data model for evaluating the marine survey cycle. Embodiments include processes for generating data models, and for testing mathematical models using sample data. The invention is not limited to six factors of port grade, water depth, scale and the like which participate in calculation in the example, and an application can simulate a more complex neural network based on the test method of the invention.
The embodiment mainly comprises the following steps:
1) and determining the network structure of the BP neural network.
In the embodiment, a BP neural network structure is constructed by taking sample data of a north sea area, an east sea map and a south sea area of a Chinese coastal port channel as a basis.
Table 1 example element model structure
In the table, 8 basic measurement input elements of each sea area are respectively a scale, port ship flow, port Throughput (TEU), water depth change condition, water depth change range, influence degree of water depth change on a measurement period, shoreline and terrain change condition, and whether important routes (four passengers, one danger, four areas, one line) are included. The number of detection input elements is 9 (except the above 8, the base detection period is one of the detection period input elements).
The three sea areas basic detection and detection cycle types (output) are respectively as follows: 4 types (8, 9, 10 and 12) are detected in the sea area of the North sea, and 6 types (1, 2, 3, 4, 5 and 0) are detected; measuring 4 types (2, 4, 6 and 8) and 5 types (0.25, 1, 2, 4 and 0) in the east sea area; and 2 types (4, 8) are detected at the sea area of the south China sea, and 4 types (1, 2, 4, 0) are detected.
The structures of BP neural networks in each sea area are respectively as follows:
TABLE 2 model network architecture
2) Sample selection and data preliminary processing.
■ sample selection:
an electronic chart of Chinese coastal port and channel chart planning catalogue is selected as sample data of the embodiment. Training samples required by model establishment are different according to the number of the three-sea-area maps, wherein 93 effective samples are available in the northern sea area, and 83 modeling training samples are taken; 104 effective samples in the east sea area and 94 modeling training samples; the number of effective samples in the sea area of south China sea is 100, the number of modeling training samples is 90, and the number of test samples is 10 in a unified mode. As shown in table 3.
TABLE 3 number of training samples and test samples
Training sample | Test specimen | |
North sea area | 83 | 10 |
Sea area of east China sea | 94 | 10 |
Sea area of south China sea | 90 | 10 |
■ data processing
Limiting the value ranges of port properties, water depth changes and the like into three types: large, medium and small; the sample data was quantified and scored to meet the modeling requirements, with the quantification results shown in tables 4-10.
TABLE 4 Scale quantification results
Serial number | Scale | Quantizing data | |
1 | 1:5000 | 0.5 | |
2 | 1:6000 | 0.6 | |
3 | 1:8000 | 0.8 | |
4 | 1:10000 | 1 | |
5 | 1:12000 | 1.2 | |
6 | 1:15000 | 1.5 | |
7 | 1:20000 | 2 | |
8 | 1:25000 | 2.5 | |
9 | 1:30000 | 3 | |
10 | 1:35000 | 3.5 | |
11 | 1:40000 | 4 | |
12 | 1:50000 | 5 | |
13 | 1:60000 | 6 | |
14 | 1:70000 | 7 | |
15 | 1:75000 | 7.5 | |
16 | 1:80000 | 8 | |
17 | 1:100000 | 10 |
TABLE 5 Port throughput data (Container traffic TEU) Scoring criteria
Serial number | Throughput (TEU) | Score of |
1 | 2 hundred million tons or more than 1000 million TEU | 20 |
2 | 1 hundred million tons or more than 500 million |
18 |
3 | 5000 ten thousand tons or more than 100 ten thousand |
16 |
4 | 2000 million tons or more than 50 million |
14 |
5 | 1000 ten thousand tons or more than 10 ten thousand |
12 |
6 | 500 ten thousand tons or more than 1 ten thousand |
10 |
7 | 200 ten thousand tons or more than |
8 |
8 | 100 ten thousand tons or more than |
6 |
9 | 100 ten thousand tons or less than |
5 |
TABLE 6 Port Ship traffic Scoring Standard
Serial number | Ship circumferential flow (pipe) | Score of |
1 | 10000 or more | 16 |
2 | Over 5000 | 14 |
3 | Above 2000 | 12 |
4 | Over 1000 | 10 |
5 | Over |
8 |
6 | Over 300 | 6 |
7 | 300 or less | 5 |
TABLE 7 Port Ship traffic-passenger ship bonus standard
Serial number | Passenger ship volume (pipe) | Score of |
1 | Over |
10 |
2 | Over 400 |
9 |
3 | Over 300 | 8 |
4 | Over 200 | 7 |
5 | Over 100 | 6 |
6 | More than 50 | 5 |
7 | Less than 50 | 4 |
TABLE 8 Port Ship flow-oil transport ship bonus standard
Serial number | Passenger ship volume (pipe) | Score of |
1 | Over 1000 | 10 |
2 | Over |
9 |
3 | Over 300 | 8 |
4 | Over 200 | 7 |
5 | Over 100 | 6 |
6 | More than 50 | 5 |
7 | Less than 50 | 4 |
TABLE 9 water depth variation data quantification results
Serial number | Passenger ship volume (pipe) | |
1 | Big (a) | 3 |
2 | In | 2 |
3 | |
1 |
Whether Table 10 contains important route data quantification results
Serial number | Passenger ship volume (pipe) | |
1 | Is that | 1 |
2 | Whether or not | 2 |
3) And (3) carrying out data normalization processing and initialization, training and classification calculation of a BP neural network marine measurement period model.
In the embodiment, Matlab software is used for processing initialization, training and classification calculation of a BP neural network, and data normalization processing and model processing codes are as follows:
the core steps are as follows: initializing the weight and the threshold of the BP neural network by using a random function of Matlab; and then training the network by establishing a circulation condition so as to enable the network to have the capability of distinguishing element combinations.
4) And analyzing the test result of the marine measurement periodic model.
The research uniformly selects 10 electronic chart areas as test samples of the neural network model in the north sea, the east sea and the south sea. Three seas 10 samples were each brought into the established BP neural network, and each seas test sample is shown in the following table (table 11).
Table 11 model validation test samples
Serial number | North sea area | Sea area of east China sea | Sea area of |
|
1 | Penglai port | In and around | Jiuzhougang | |
2 | Great linking harbor and its vicinity | Main channel of Linyunkong | |
|
3 | Liaohukou water channel (one) | Deepwater channel of |
|
|
4 | Small three mountain waterway | |
|
|
5 | Jingtang harbor basin and channel | Wurime mouth channel | Jing Bay and nearby | |
6 | From the ship lock of the new port to the second lock | Near the wharf of the customs | Three harbors and the |
|
7 | The yellow Ye is near the harbor basin | Sijing from Laoshu pond to Tang | Guangdong |
|
8 | Huang Ye gang channel (two) | From Daishan to Zhoushan | East island to |
|
9 | At and near the mountain area | Hair reef mountain to cylinder slit mountain | Around |
|
10 | Tianjin gang | Deepwater channel of Yangtze river mouth | Around the islands of Chuanshan |
As shown in fig. 1, there are 6 input elements, which are respectively a scale, a port level, a water depth variation condition, a water depth variation range size, a shoreline and terrain variation, and an AIS trajectory analysis condition. The number of detection input elements is 7 (except the above 6, the base detection period is one of the detection period input elements).
The sample data is input into the data model to obtain a cycle accuracy rate chart as shown in fig. 2-7, and the results show that the accuracy rate of the result obtained by the BP neural network model through classification calculation is high, the accuracy rate is maintained at about 90%, and the cycle evaluation requirements under different condition combinations can be basically reflected. But the latter is more concise, and meanwhile, the scoring can be performed again according to the annual change of the port, so that the requirement of a coding implementation program is more facilitated.
The invention discloses a marine measurement cycle evaluation method based on a BP neural network, which comprises positive and negative correlation evaluation of sample data; sample data cleaning; initializing a BP neural network; hidden layer output calculation; output layer output calculation; calculating an error; updating the weight value; updating a threshold value; judging iteration constraint of the algorithm; and (6) generating a model. Experiments show that: the method can effectively combine the past marine measurement period evaluation experience with the BP neural network model, is consistent with the actual situation of the water area of the marine coastal port channel in China, and realizes a scientific and quantitative marine measurement period evaluation method.
Example 2
The embodiment 2 of the invention provides a marine measurement cycle evaluation system based on a BP neural network, which is realized based on the method and comprises the following steps: the marine measurement cycle evaluation model comprises a marine measurement cycle evaluation model, an acquisition module and an output module; wherein the content of the first and second substances,
the acquisition module is used for acquiring the water depth change, the port grade, the scale, the influence degree of the water depth change, the AIS track analysis and the shoreline terrain of the sea area to be detected;
the output module is used for inputting the data into a pre-established and trained marine measurement period evaluation model to obtain a sea area measurement period to be measured;
the ocean measurement period evaluation model adopts an optimized and improved BP neural network structure.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention and are not limited. Although the present invention has been described in detail with reference to the embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (10)
1. A method for marine measurement cycle assessment based on a BP neural network, the method comprising:
collecting the water depth change, port grade, scale, water depth change influence degree, AIS track analysis and shoreline topography of a sea area to be detected;
inputting the data into a pre-established and trained marine measurement period evaluation model to obtain a sea area measurement period to be measured;
the ocean measurement period evaluation model adopts an optimized and improved BP neural network structure.
2. The BP neural network-based marine survey cycle assessment method according to claim 1, wherein the inputs of the marine survey cycle assessment model are water depth variation, port level, scale, water depth variation influence degree, AIS trajectory analysis and shoreline terrain, and the output is a survey cycle; the marine measurement cycle assessment model comprises an input layer, a hidden layer and an output layer.
3. The BP neural network-based marine measurement cycle assessment method according to claim 2, further comprising a training step of a marine measurement cycle assessment model; the method specifically comprises the following steps:
establishing a training set;
setting network initialization parameters;
inputting training set data into a BP neural network;
calculating hidden layer output;
calculating a prediction output from the hidden layer output;
calculating a network prediction error based on the expectation and the output prediction output;
updating the network connection weight according to the network prediction error;
updating a network node threshold value according to the network prediction error;
and giving a learning rate and a neuron excitation function, and adjusting the number of input layer nodes, the number of hidden layer nodes and the number of output layer nodes of the network to obtain a BP neural network meeting constraint conditions, so as to obtain a trained marine measurement period evaluation model.
4. The BP neural network-based marine measurement cycle assessment method according to claim 3, wherein the establishing a training set; the method specifically comprises the following steps:
respectively selecting a plurality of groups of sample data of a north sea area, an east sea area and a south sea area of a Chinese coastal port channel, wherein each group of sample data comprises a scale, port ship flow, port throughput, water depth change conditions, water depth change range, influence degree of water depth change on a measurement period, shoreline topographic data and whether important navigation paths are included;
the water depth control method comprises the following steps of classifying the water depth into a large class, a middle class and a small class according to port properties, and classifying the water depth control method into the large class, the middle class and the small class according to the value range of water depth change; quantifying and scoring the ruler data;
scoring the port ship flow according to the range value of the ship flow, the passenger ship flow or the oil ship flow;
grading the port throughput according to the range value;
dividing into three types of large, medium and small according to the value range of the passenger ship circumferential flow; quantifying the water depth change condition;
quantifying whether the important air routes are contained;
respectively carrying out normalization processing on the size of the water depth change range, the influence degree of the water depth change on the measurement period, and the change conditions of the shoreline and the terrain;
and forming a group of records by the quantitative grading data and the normalized data, and taking the corresponding marine measurement period as a label to obtain a plurality of groups of training data so as to establish a training set.
5. The method for marine measurement cycle assessment based on the BP neural network according to claim 3, wherein the network initialization parameters specifically include: the number of input layer nodes, the number of hidden layer nodes, the number of output layer nodes, and the connection weights between the input layer, hidden layer and output layer neurons, the hidden layer threshold and the output layer threshold.
6. The method for marine measurement cycle assessment based on the BP neural network according to claim 5, wherein the calculating the hidden layer output specifically comprises:
calculating the output H of the jth node of the hidden layer according to the formulajComprises the following steps:
where l is the number of hidden layer nodes, i represents the ith input layer node, xiRepresenting the ith input vector, ωijRepresenting the connection weight between the ith input level node and the jth hidden level node, ajA threshold value representing the jth hidden layer node, f (-) being a hidden layer excitation function, fullThe formula:
wherein y is an intermediate variable.
7. The BP neural network-based marine measurement cycle assessment method according to claim 6, wherein the calculating a prediction output from a hidden layer output; the method specifically comprises the following steps:
calculating the predicted output O of the kth node of the output layer according to the formulakComprises the following steps:
in the formula, wjkAs the connection weights between the j hidden layer nodes and the kth output layer node neurons, bkAnd m is the number of output layer nodes.
8. The BP neural network-based marine measurement cycle assessment method according to claim 7, wherein the calculating a network prediction error from an expectation and an output prediction output; the method specifically comprises the following steps:
calculating the network prediction error e according tokComprises the following steps:
ek=Yk-Ok k=1,2,…m
in the formula, YkRepresenting the desired output of the kth output level node, and m is the number of output level nodes.
9. The BP neural network-based marine measurement cycle assessment method according to claim 8, wherein the updating of the network connection weights is performed according to a network prediction error; the method specifically comprises the following steps:
wjk-new=wjk+ηHjek j=1,2,…,l k=1,2,…m
in the formula, ωij-newThe updated connection weight between the ith input layer node and the jth hidden layer node is calculated, eta is the learning rate, wjk-newAnd (4) updating the connection weight between the j-th hidden layer node and the k-th output layer node neuron, wherein l is the number of hidden layer nodes, and m is the number of output layer nodes.
10. The BP neural network-based marine measurement cycle assessment method according to claim 9, wherein the updating of the network node threshold according to network prediction error; the method specifically comprises the following steps:
bk-new=bk+ek k=1,2,…m
in the formula, aj-newFor updated threshold of the jth hidden layer node, bkThreshold for the kth output level node, bk-newFor the updated threshold of the kth output layer node, l is the number of hidden layer nodes, and m is the number of output layer nodes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110799971.7A CN113657021A (en) | 2021-07-15 | 2021-07-15 | Marine measurement period evaluation method based on BP neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110799971.7A CN113657021A (en) | 2021-07-15 | 2021-07-15 | Marine measurement period evaluation method based on BP neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113657021A true CN113657021A (en) | 2021-11-16 |
Family
ID=78489391
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110799971.7A Pending CN113657021A (en) | 2021-07-15 | 2021-07-15 | Marine measurement period evaluation method based on BP neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113657021A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115238569A (en) * | 2022-06-21 | 2022-10-25 | 中国人民解放军61540部队 | Temperature and salt reconstruction method and system based on information embedding depth FFNN model |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102681033A (en) * | 2012-04-27 | 2012-09-19 | 哈尔滨工程大学 | Sea surface wind measurement method based on X-band marine radar |
CN102982351A (en) * | 2012-11-15 | 2013-03-20 | 河北省电力公司电力科学研究院 | Porcelain insulator vibrational acoustics test data sorting technique based on back propagation (BP) neural network |
CN103077267A (en) * | 2012-12-28 | 2013-05-01 | 电子科技大学 | Parameter sound source modeling method based on improved BP (Back Propagation) neural network |
CN104268625A (en) * | 2014-10-09 | 2015-01-07 | 哈尔滨工程大学 | Autonomous underwater vehicle track predicating method based on marine environment information |
CN104613944A (en) * | 2015-01-27 | 2015-05-13 | 电子科技大学 | Distributed water depth prediction method based on GWR (geographically weighted regression) and BP (back propagation) neural network |
CN109344502A (en) * | 2018-09-30 | 2019-02-15 | 武汉理工大学 | Marine shafting intermediate bearing installation displacement method of adjustment based on GA-BP algorithm |
CN110887506A (en) * | 2019-10-14 | 2020-03-17 | 交通运输部水运科学研究所 | Motion amplitude detection method and system of inertial sensor influenced by sea waves |
-
2021
- 2021-07-15 CN CN202110799971.7A patent/CN113657021A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102681033A (en) * | 2012-04-27 | 2012-09-19 | 哈尔滨工程大学 | Sea surface wind measurement method based on X-band marine radar |
CN102982351A (en) * | 2012-11-15 | 2013-03-20 | 河北省电力公司电力科学研究院 | Porcelain insulator vibrational acoustics test data sorting technique based on back propagation (BP) neural network |
CN103077267A (en) * | 2012-12-28 | 2013-05-01 | 电子科技大学 | Parameter sound source modeling method based on improved BP (Back Propagation) neural network |
CN104268625A (en) * | 2014-10-09 | 2015-01-07 | 哈尔滨工程大学 | Autonomous underwater vehicle track predicating method based on marine environment information |
CN104613944A (en) * | 2015-01-27 | 2015-05-13 | 电子科技大学 | Distributed water depth prediction method based on GWR (geographically weighted regression) and BP (back propagation) neural network |
CN109344502A (en) * | 2018-09-30 | 2019-02-15 | 武汉理工大学 | Marine shafting intermediate bearing installation displacement method of adjustment based on GA-BP algorithm |
CN110887506A (en) * | 2019-10-14 | 2020-03-17 | 交通运输部水运科学研究所 | Motion amplitude detection method and system of inertial sensor influenced by sea waves |
Non-Patent Citations (1)
Title |
---|
JING DUAN等: "Research on the hydrographic survey cycle for updating navigational charts", 《THE JOURNAL OF NAVIGATION》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115238569A (en) * | 2022-06-21 | 2022-10-25 | 中国人民解放军61540部队 | Temperature and salt reconstruction method and system based on information embedding depth FFNN model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109800863B (en) | Logging phase identification method based on fuzzy theory and neural network | |
Kaya et al. | Groundwater level prediction using artificial neural network and M5 tree models | |
CN108469507A (en) | A kind of water outlet BOD flexible measurement methods based on Self organizing RBF Neural Network | |
WO2022021726A1 (en) | Pmu-based power system state estimation performance evaluation method | |
CN105608004A (en) | CS-ANN-based software failure prediction method | |
CN108346144A (en) | Bridge Crack based on computer vision monitoring and recognition methods automatically | |
CN109902329A (en) | A kind of reservoir modeling auxiliary history-matching method, system, storage medium and equipment | |
CN110675036A (en) | Plateau river ecological health evaluation method based on random forest optimization microbial index | |
CN108008099A (en) | A kind of pollution sources localization method | |
CN108960421A (en) | The unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network of improvement | |
CN112233734B (en) | Water quality data deduction acquisition method and system based on machine learning | |
CN110309609A (en) | A kind of architecture indoor air quality evaluation method based on rough set and wavelet neural network | |
CN110619418A (en) | Multi-feature water quality prediction method based on mixed model combination algorithm | |
CN113657021A (en) | Marine measurement period evaluation method based on BP neural network | |
CN108615090A (en) | River and lake water quality prediction based on Monte Carlo simulation and outrophication risk appraisal procedure | |
CN117330040B (en) | Tidal flat topography mapping method and system based on unmanned water mapping ship | |
CN117078114B (en) | Water quality evaluation method and system for water-bearing lakes under influence of diversion engineering | |
CN117076893B (en) | Sound velocity distribution forecasting method based on long-term and short-term memory neural network | |
CN112989526B (en) | Aviation network key node identification method based on kernel extreme learning machine | |
CN111914488B (en) | Data area hydrologic parameter calibration method based on antagonistic neural network | |
CN113743662A (en) | Ship key technical parameter prediction method and system based on machine learning | |
CN113962426A (en) | Offshore water navigation safety intelligent forecasting method and device | |
CN113343601A (en) | Dynamic simulation method for water level and pollutant migration of complex water system lake | |
CN112446168A (en) | Effluent BOD concentration soft measurement method based on MIC and RBFNN | |
CN109101759A (en) | A kind of parameter identification method based on forward and reverse response phase method |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20211116 |