CN116451591A - Ocean floating structure mooring force assessment method based on deep learning - Google Patents
Ocean floating structure mooring force assessment method based on deep learning Download PDFInfo
- Publication number
- CN116451591A CN116451591A CN202310698549.1A CN202310698549A CN116451591A CN 116451591 A CN116451591 A CN 116451591A CN 202310698549 A CN202310698549 A CN 202310698549A CN 116451591 A CN116451591 A CN 116451591A
- Authority
- CN
- China
- Prior art keywords
- neural network
- network model
- floating structure
- training
- sea
- 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
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000013135 deep learning Methods 0.000 title claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 111
- 238000003062 neural network model Methods 0.000 claims abstract description 99
- 238000004364 calculation method Methods 0.000 claims abstract description 47
- 230000008878 coupling Effects 0.000 claims abstract description 45
- 238000010168 coupling process Methods 0.000 claims abstract description 45
- 238000005859 coupling reaction Methods 0.000 claims abstract description 45
- 230000004044 response Effects 0.000 claims abstract description 24
- 238000011156 evaluation Methods 0.000 claims abstract description 17
- 238000012544 monitoring process Methods 0.000 claims abstract description 10
- 238000005259 measurement Methods 0.000 claims abstract description 8
- 230000007613 environmental effect Effects 0.000 claims description 41
- 238000012360 testing method Methods 0.000 claims description 33
- 238000013528 artificial neural network Methods 0.000 claims description 24
- 238000005457 optimization Methods 0.000 claims description 24
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000013461 design Methods 0.000 claims description 6
- 238000013016 damping Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 230000005855 radiation Effects 0.000 claims description 4
- 238000002474 experimental method Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims 1
- 238000012423 maintenance Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
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
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
- G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
- G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
- G01C21/165—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation combined with non-inertial navigation instruments
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01L—MEASURING FORCE, STRESS, TORQUE, WORK, MECHANICAL POWER, MECHANICAL EFFICIENCY, OR FLUID PRESSURE
- G01L5/00—Apparatus for, or methods of, measuring force, work, mechanical power, or torque, specially adapted for specific purposes
- G01L5/0028—Force sensors associated with force applying means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01L—MEASURING FORCE, STRESS, TORQUE, WORK, MECHANICAL POWER, MECHANICAL EFFICIENCY, OR FLUID PRESSURE
- G01L5/00—Apparatus for, or methods of, measuring force, work, mechanical power, or torque, specially adapted for specific purposes
- G01L5/04—Apparatus for, or methods of, measuring force, work, mechanical power, or torque, specially adapted for specific purposes for measuring tension in flexible members, e.g. ropes, cables, wires, threads, belts or bands
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
- G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/42—Determining position
- G01S19/45—Determining position by combining measurements of signals from the satellite radio beacon positioning system with a supplementary measurement
- G01S19/47—Determining position by combining measurements of signals from the satellite radio beacon positioning system with a supplementary measurement the supplementary measurement being an inertial measurement, e.g. tightly coupled inertial
-
- 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
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Abstract
The invention relates to the technical field of mooring force monitoring, in particular to a deep learning-based ocean floating structure mooring force evaluation method, which comprises the following steps: establishing and checking a time domain coupling calculation model; building and optimizing a neural network model; dividing the sea state of a training set of the optimal neural network model, calculating mooring force and six-degree-of-freedom response data of the floating structure under each sea state, constructing the training set of the optimal neural network model, and training to obtain a training neural network model; the global positioning system and the inertial measurement unit monitor six-degree-of-freedom motion of the floating structure, transmit monitoring data to the training neural network model in real time, and estimate six-degree-of-freedom motion response and mooring force of the floating structure through the training neural network model. The method provided by the invention can effectively guide the operation and maintenance of the ocean floating structure in real time, and has lower cost.
Description
Technical Field
The invention relates to the technical field of mooring force monitoring, in particular to a deep learning-based ocean floating structure mooring force evaluation method.
Background
The offshore floating structures are important equipment for the development and utilization of ocean resources, and comprise a semi-submersible drilling platform for the development and utilization of oil and gas resources, a floating production, storage and unloading device and the like, a floating wind driven generator for the development and utilization of offshore wind energy, and a floating net cage for fishery cultivation, wherein the offshore floating structures are positioned in a working sea area by means of a mooring system, so that whether the mooring system is safe or not is directly related to whether the offshore floating structures can safely and stably work. Mooring line stress is the most direct manifestation of whether a mooring system is safe or not, so it is critical to monitor mooring line stress to prevent potential damage.
At present, a common mooring stress monitoring method comprises the steps of installing a tension sensor on a mooring rope to directly monitor the mooring stress, installing an inclinometer on the mooring rope, and deducing the mooring stress through a static balance formula. However, the deployment of sensors or inclinometers on mooring lines is expensive, while the marine environment is harsh, the monitoring equipment is prone to failure, resulting in discontinuous monitoring data, and further, the cost of long-term maintenance and repair is also very expensive.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a deep learning-based ocean floating structure mooring force assessment method, wherein time domain coupling calculation data is used as training data of a neural network model and is optimized, then training is carried out to obtain the training neural network model, data monitored by a global positioning system and an inertial measurement unit are used as input data of the training neural network model and are input into the training neural network model, so that real-time mooring stress of the floating structure is obtained, operation and maintenance of the ocean floating structure can be effectively guided in real time, and the cost is lower.
The invention is realized by the following technical scheme:
a deep learning-based ocean floating structure mooring force assessment method comprises the following steps:
s1: establishing a time domain coupling calculation model according to the entity of the floating structure, carrying out a pool experiment on the floating structure, recording floating structure pool experimental data, and checking the time domain coupling calculation model through the floating structure pool experimental data to obtain a checking time domain coupling calculation model;
s2: calculating mooring force and six-degree-of-freedom response data of the floating structure according to the checking time domain coupling calculation model,
then building a neural network model, and optimizing the structure of the neural network model to obtain an optimal neural network model;
s3: dividing the sea state of a training set of an optimal neural network model according to the rated sea state designed by the floating structure, calculating mooring force and six-degree-of-freedom response data of the floating structure under each sea state, constructing the training set of the optimal neural network model, and training the optimal neural network model to obtain a training neural network model;
s4: the global positioning system and the inertial measurement unit monitor six-degree-of-freedom motion of the floating structure, transmit monitoring data to the training neural network model in real time, and estimate six-degree-of-freedom motion response and mooring force of the floating structure through the training neural network model.
When the time-coupled calculation model is checked through floating structure water pool experimental data in the step S1, the check content comprises an amplitude response operator of the hydrodynamic model, additional mass, radiation damping, pretension of a mooring model, a rigidity curve and the maximum stress of time domain dynamic calculation.
The parameters for optimizing the structure of the neural network model in step S2 include the length of the input data time window, the number of hidden layers of the neural network, the number of neurons of each hidden layer, and an optimization algorithm.
Further, in step S2, before optimizing the structure of the neural network model, the calculated mooring force and six-degree-of-freedom response data of the floating structure are preprocessed by using a maximum and minimum normalization processing method, so as to obtain a preprocessed time domain coupling calculation value, then the preprocessed time domain coupling calculation value is input into the neural network model for learning and estimating, so as to obtain a neural network estimation value, and then an evaluation index average absolute error of the neural network model estimation result is calculated according to formula (1)MAEAverage absolute percentage errorMAPERoot mean square errorRMSEFitting degreeR，Comparing each evaluation index with corresponding preset value, if the average absolute errorMAEAverage absolute percentage errorMAPE、Root mean square errorRMSE is smaller than or equal to the corresponding preset value, and the fitting degreeRIf the structural value is larger than or equal to the corresponding preset value, the optimization of the structure of the neural network model is completed;
（1），
wherein:total number of data sets>For data set sequence number>Is->Individual neural network estimates, < >>Is->And time domain coupling calculated values.
Further, in step S3, the step of dividing the training set sea state of the optimal neural network model according to the rated sea state of the floating structure design includes the following steps:
d1: determining an estimation interval of each environmental factor: firstly analyzing environmental factors under rated sea conditions, taking the sea conditions formed by the rated sea conditions and one of the changed environmental factors as a training set sea condition, taking any sea condition in the range of a training set sea condition interval as a test set sea condition, inputting the training set sea condition into a checking time domain coupling calculation model, respectively calculating the mooring force of a floating structure and training data responded by six degrees of freedom, normalizing the training data to form a training set, inputting the training set into an optimal neural network model, training the optimal neural network model to form a training neural network model, inputting the test set sea condition into the checking time domain coupling calculation model, respectively calculating the mooring force of the floating structure and testing data responded by six degrees of freedom, normalizing the testing data to form a test set, inputting the six degrees of freedom movements in the test set into the training neural network model, and testing the training neural network model;
d2, comparing the test result with the test set sea state time domain coupling calculation result, if the preset condition is met, determining the training set sea state interval as an estimation interval of the environmental factors, if the preset condition is not met, reducing the difference between the environmental factors and the rated sea state, and returning to the step D1 until the test result and the time domain coupling calculation result meet the preset condition;
d3: dividing the total training set sea state according to the estimated interval and rated sea state of the environmental factors: dividing critical sea states according to the determined environmental factor estimation interval length and the rated sea states of the floating structure design, setting two adjacent critical sea states as a group to obtain a group training set sea state, superposing the group training set sea states to form a total training set sea state, inputting the total training set sea state into a time domain coupling calculation model to perform time domain calculation to obtain total training data, normalizing the total training data to form a total training set, inputting the total training set into an optimal neural network model to perform training, and obtaining a training neural network model.
The invention has the beneficial effects that:
according to the method, motion response and mooring stress of the floating structure are calculated according to given environmental conditions, a nonlinear relation between input data and output data is fitted by a deep learning model in advance, then the motion response of the floating structure is accurately measured through a GPS positioning device and an Inertial Measurement Unit (IMU) of the floating structure, and the mooring force of the marine floating structure is calculated by time domain coupling according to the fitted nonlinear relation.
Drawings
FIG. 1 is a schematic flow chart of the evaluation method of the present invention.
Detailed Description
The ocean floating structure mooring force evaluation method based on deep learning is shown in the flow chart of figure 1, and specifically comprises the following steps:
s1: establishing a time domain coupling calculation model according to the entity of the floating structure, carrying out a pool experiment on the floating structure, recording floating structure pool experimental data, and checking the time domain coupling calculation model through the floating structure pool experimental data to obtain a checking time domain coupling calculation model;
the content of the specific check can comprise an amplitude response operator of the hydrodynamic model, additional mass, radiation damping, pretension of the mooring model, a stiffness curve, a maximum stress calculated dynamically in time, and the like. During checking, firstly, respectively calculating an amplitude response operator, additional mass, radiation damping, pretension of a mooring model, a rigidity curve and maximum stress of time dynamic calculation according to the established time domain coupling calculation model, then comparing with corresponding floating structure water tank experimental data, if the data fitness meets the preset requirement, checking the time domain coupling model, and if the data fitness does not meet the preset requirement, adjusting the time domain coupling calculation model, and then comparing until the data fitness meets the preset requirement;
s2: calculating mooring force and six-degree-of-freedom response data of the floating structure according to the checking time domain coupling calculation model,
then building a neural network model, and optimizing the structure of the neural network model to obtain an optimal neural network model;
parameters for optimizing the structure of the neural network model comprise the length of a time window of input data, the number of hidden layers of the neural network, the number of neurons of each hidden layer and an optimization algorithm;
optimization of input data time window length: the optimization basis is as follows: the loss function is MSE; the optimizer is Rmspprop; an activation function is tanh, and the number of neurons is 50; batch size was 100; training number is 200;
the specific input data time window length optimization variable can be shown in a first table, then training is carried out on the built neural network model, mooring stress is estimated, and an evaluation index is calculated, wherein the statistics of the evaluation index is shown in a second table:
list one
Watch II
It can be seen from Table II that the root mean square error of the estimated results of the neural network model to the mooring force is determined by other parametersRMSEAverage absolute errorMAEPercentage of mean absolute errorMAPEThe root mean square error, the average absolute error and the average absolute error percentage are minimum when the time window length of the input data is 40; fitting degree of neural network model to estimated result of mooring forceRThe fitting degree is optimal when the time window length of the input data is 40 as the time window length of the input data is increased first and then decreased. The estimated performance of the neural network on the mooring force is synthesized, so that the error of the estimated result of the mooring leg force by the neural network model when the time window length of the input data is 40 is minimum and the fitting degree is highest can be determined, and the time window length of the input data of the neural network model is 40, thereby completing the optimization of the time window length of the input data of the neural network model.
Optimization of the number of hidden layers of the neural network: the optimization basis is as follows: the loss function is MSE; the optimizer is Rmspprop; an activation function is tanh, the time window length of input data is 40, and the neuron number is 50; batch size was 100; training number is 200;
the optimized variable of the number of hidden layers of the specific neural network can be shown in a table III, then the built neural network model is trained, the mooring stress is estimated, and an evaluation index is calculated, wherein the statistics of the evaluation index is shown in a table IV:
watch III
Table four
As can be seen from table four, under the condition that other parameters are determined, the root mean square error, the average absolute error and the average absolute error percentage of the estimated result of the mooring force by the neural network model become larger along with the increase of the number of hidden layers of the neural network, and when the number of hidden layers of the neural network is 2, the root mean square error, the average absolute error and the average absolute error percentage are the smallest; the fitting degree of the neural network model to the estimated result of the mooring force is reduced along with the increase of the number of the hidden layers of the neural network, and the fitting degree is optimal when the number of the hidden layers of the neural network is 2, so that the number of the hidden layers of the neural network model is determined to be 2, and the optimization of the number of the hidden layers of the neural network model is completed;
the optimization of the number of neurons of each hidden layer is carried out by adopting the method on the basis that the length of a time window of input data is 40 and the number of hidden layers of a neural network is 2;
the optimization algorithm is used for managing and updating learning parameters of the neural network model, and assuming that the number of neurons of each hidden layer is determined to be 256, the optimization variables of the optimization algorithm are shown in a fifth table, then the built neural network model is trained, mooring stress is estimated, evaluation indexes are calculated, and statistics of the evaluation indexes are shown in a sixth table:
TABLE five
TABLE six
As can be seen from table six, under the condition of determining other parameters, the root mean square error, the average absolute error and the average absolute error percentage of the estimated result of the mooring force by the neural network model are minimum when the neural network model optimization algorithm is Adam, and are maximum when the neural network model optimization algorithm is Sgd; the fitting degree of the neural network model to the estimated result of the mooring force is optimal when the neural network model optimization algorithm is Adam, and worst when the neural network model optimization algorithm is Sgd. And (3) synthesizing the estimated performance of the neural network on the tension of the mooring leg, and determining the optimization algorithm of the neural network model as Adam, thereby completing the optimization of the optimization algorithm of the neural network model.
S3: dividing the sea state of a training set of an optimal neural network model according to the rated sea state designed by the floating structure, calculating mooring force and six-degree-of-freedom response data of the floating structure under each sea state, constructing the training set of the optimal neural network model, and training the optimal neural network model to obtain a training neural network model;
s4: the method comprises the steps of installing a global positioning system and an inertial measurement unit on a floating structure, monitoring six-degree-of-freedom motion of the floating structure in real time, transmitting monitoring data to a training neural network model in real time, and estimating six-degree-of-freedom motion response and mooring force of the floating structure through the training neural network model.
According to the method, the motion response and mooring force of the floating structure are calculated according to given environmental conditions, a nonlinear relation between input data and output data is fitted by a deep learning model in advance, then the motion response of the floating structure is accurately measured through a GPS positioning device and an Inertial Measurement Unit (IMU) of the floating structure, and the mooring force of the marine floating structure is calculated by time domain coupling according to the fitted nonlinear relation.
Further, in step S2, before optimizing the structure of the neural network model, the calculated mooring force and six-degree-of-freedom response data of the floating structure are preprocessed by using a maximum and minimum normalization processing method, so as to obtain a preprocessed time domain coupling calculation value, then the preprocessed time domain coupling calculation value is input into the neural network model for learning and estimating, so as to obtain a neural network estimation value, and then an evaluation index average absolute error of the neural network model estimation result is calculated according to formula (1)MAEAverage absolute percentage errorMAPERoot mean square errorRMSEFitting degreeR，Comparing each evaluation index with corresponding preset value, if the average absolute errorMAEAverage absolute percentage errorMAPE、Root mean square errorRMSE is smaller than or equal to the corresponding preset value, and the fitting degreeRIf the structural value is larger than or equal to the corresponding preset value, the optimization of the structure of the neural network model is completed;
（1），
wherein:total number of data sets>For data set sequence number>Is->Individual neural network estimates, < >>Is->And time domain coupling calculated values.
Further, in step S3, the step of dividing the training set sea state of the optimal neural network model according to the rated sea state of the floating structure design includes the following steps:
d1: determining an estimation interval of each environmental factor: firstly analyzing environmental factors under rated sea conditions, taking the sea conditions formed by the rated sea conditions and one of the changed environmental factors as a training set sea condition, taking any sea condition in the range of a training set sea condition interval as a test set sea condition, inputting the training set sea condition into a checking time domain coupling calculation model, respectively calculating the mooring force of a floating structure and training data responded by six degrees of freedom, normalizing the training data to form a training set, inputting the training set into an optimal neural network model, training the optimal neural network model to form a training neural network model, inputting the test set sea condition into the checking time domain coupling calculation model, respectively calculating the mooring force of the floating structure and testing data responded by six degrees of freedom, normalizing the testing data to form a test set, inputting the six degrees of freedom movements in the test set into the training neural network model, and testing the training neural network model;
environmental factors here include: sense wave height, spectrum peak period, wave direction, wind speed, wind direction, flow velocity and flow direction;
when a specific flow rate estimation interval is determined, sea conditions for training and testing are shown in a table seven, three-hour time domain calculation is carried out on working conditions of a training set and a testing set, working condition calculation results of the training set subjected to normalization and feature extraction are input into an optimal neural network model for training, then the testing working conditions are estimated, and an estimated performance index statistical table for obtaining a neural network estimation result is shown in a table eight:
watch seven
Table eight
In the eighth table, it is obvious that the fitting degree of the estimated results of the neural network is greater than 90%, the average absolute percentage error is less than 1%, and the predetermined requirement is satisfied, so that the estimated interval of the flow velocity of the neural network can be determined to be [0.83-1.35] m/s, the estimated interval length is determined to be [0-0.52] m/s, if the fitting degree of the estimated results of the neural network is equal, the average absolute percentage error does not satisfy the predetermined requirement, the value range of the flow velocity is narrowed, and then the estimated is performed according to the above method until the fitting degree of the estimated results of the neural network is equal, the average absolute percentage error satisfies the predetermined requirement, thereby determining the estimated interval of the flow velocity of the neural network, and the determination of the estimated intervals of other environmental factors is also determined one by one according to the method.
D2, comparing the test result with the test set sea state time domain coupling calculation result, if the preset condition is met, determining the training set sea state interval as an estimation interval of the environmental factors, if the preset condition is not met, reducing the difference between the environmental factors and the rated sea state, and returning to the step D1 until the test result and the time domain coupling calculation result meet the preset condition;
d3: dividing the total training set sea state according to the estimated interval and rated sea state of the environmental factors: dividing critical sea states according to the determined environmental factor estimation interval length and the rated sea states of the floating structure design, setting two adjacent critical sea states as a group to obtain a group training set sea state, superposing the group training set sea states to form a total training set sea state, inputting the total training set sea state into a time domain coupling calculation model to perform time domain calculation to obtain total training data, normalizing the total training data to form a total training set, inputting the total training set into an optimal neural network model to perform training, and obtaining a training neural network model.
Assuming that the century-specific sea conditions are shown in table nine, the estimated intervals of the determined environmental factors are shown in table ten:
table nine
Ten meters
Dividing the annual environmental load into four critical sea conditions (extreme sea conditions, severe sea conditions, common sea conditions, mild sea conditions) according to the estimated interval of the environmental factors, as shown in a table eleven; dividing four critical sea conditions into three situations, as shown in a table twelve; these three conditions include a range that covers substantially all sea conditions that may be encountered during the floating structure's duty cycle.
Table eleven
Twelve watches
The working condition of the floating structure can change along with the change of sea condition and the direction of environmental factors, and any one of the factors can change to generate different working conditions, so that different environmental factors need to be combined, the estimated interval of the environmental load direction determined in the prior stage is assumed to be 30 degrees, the included angle between waves and wind is known to be within 30 degrees according to DNV specification, the included angle between waves and wind is known to be within 45 degrees, and the combined form of the environmental load direction of stormy waves and currents is set as shown in the table thirteen: in the table, a represents the included angle between the wave and the heading of the floating structure, and the estimated interval of the wave direction is 0-30 degrees, so the value range of a is [0, 30, 60, 90, 120, 150, 180], and the combination form of the environmental load direction is 42 possibilities in total;
watch thirteen
The method comprises the steps of determining the direction combination of environmental loads, then combining the sizes of the environmental loads, combining the sizes of the environmental loads under three conditions according to the estimated range of the environmental factors determined in the prior art, taking a situation I as an example, combining the environmental loads as shown in a table fourteen, wherein the number A is an extreme working condition, the number I is a bad sea condition, and 16 combination forms are obtained in total, namely all combination forms of four environmental factors, so that a total training set sea state is obtained, the situation that the training set covers sea states with different sizes of any environmental loads under typhoon is ensured, and the coverage range of a training neural network model finally obtained is ensured.
Fourteen watch
In summary, the ocean floating structure mooring force evaluation method based on deep learning provided by the invention calculates the motion response and mooring stress of the floating structure according to given environmental conditions, fits the nonlinear relation between input data and output data by a deep learning model in advance, then accurately measures the motion response by a GPS positioning device and an Inertial Measurement Unit (IMU) of the floating structure, and calculates the ocean floating structure mooring force according to the fitted nonlinear relation by time domain coupling.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (5)
1. The ocean floating structure mooring force assessment method based on deep learning is characterized by comprising the following steps of: the method comprises the following steps:
s1: establishing a time domain coupling calculation model according to the entity of the floating structure, carrying out a pool experiment on the floating structure, recording floating structure pool experimental data, and checking the time domain coupling calculation model through the floating structure pool experimental data to obtain a checking time domain coupling calculation model;
s2: calculating mooring force and six-degree-of-freedom response data of the floating structure according to the checking time domain coupling calculation model,
then building a neural network model, and optimizing the structure of the neural network model to obtain an optimal neural network model;
s3: dividing the sea state of a training set of an optimal neural network model according to the rated sea state designed by the floating structure, calculating mooring force and six-degree-of-freedom response data of the floating structure under each sea state, constructing the training set of the optimal neural network model, and training the optimal neural network model to obtain a training neural network model;
s4: the global positioning system and the inertial measurement unit monitor six-degree-of-freedom motion of the floating structure, transmit monitoring data to the training neural network model in real time, and estimate six-degree-of-freedom motion response and mooring force of the floating structure through the training neural network model.
2. The deep learning-based ocean floating structure mooring force assessment method according to claim 1, wherein: in the step S1, when the time domain coupling calculation model is checked through the floating structure water pool experimental data, the check content comprises an amplitude response operator of the hydrodynamic model, additional mass, radiation damping, pretension of a mooring model, a rigidity curve and the maximum stress of time domain dynamic calculation.
3. The deep learning-based ocean floating structure mooring force assessment method according to claim 1, wherein: in step S2, parameters for optimizing the structure of the neural network model include the length of the input data time window, the number of hidden layers of the neural network, the number of neurons of each hidden layer, and an optimization algorithm.
4. A deep learning based ocean floating structure mooring force assessment method according to claim 3, wherein: in step S2, before optimizing the structure of the neural network model, preprocessing the calculated mooring force and six-degree-of-freedom response data of the floating structure by adopting a maximum and minimum normalization processing method to obtain a preprocessed time domain coupling calculation value, inputting the preprocessed time domain coupling calculation value into the neural network model for learning and estimating to obtain a neural network estimation value, and calculating the average absolute error of the evaluation index of the neural network model estimation result according to the formula (1)MAEAverage absolute percentage errorMAPERoot mean square errorRMSEFitting degreeR，Comparing each evaluation index with corresponding preset value, if the average absolute errorMAEAverage absolute percentage errorMAPE、Root mean square errorDifference of differenceRMSE is smaller than or equal to the corresponding preset value, and the fitting degreeRIf the structural value is larger than or equal to the corresponding preset value, the optimization of the structure of the neural network model is completed;
（1），
wherein:total number of data sets>For data set sequence number>Is->Individual neural network estimates, < >>Is->And time domain coupling calculated values.
5. The deep learning-based ocean floating structure mooring force assessment method according to claim 4, wherein: in step S3, the step of dividing the training set sea state of the optimal neural network model according to the rated sea state of the floating structure design includes the following steps:
d1: determining an estimation interval of each environmental factor: firstly analyzing environmental factors under rated sea conditions, taking the sea conditions formed by the rated sea conditions and one of the changed environmental factors as a training set sea condition, taking any sea condition in the range of a training set sea condition interval as a test set sea condition, inputting the training set sea condition into a checking time domain coupling calculation model, respectively calculating the mooring force of a floating structure and training data responded by six degrees of freedom, normalizing the training data to form a training set, inputting the training set into an optimal neural network model, training the optimal neural network model to form a training neural network model, inputting the test set sea condition into the checking time domain coupling calculation model, respectively calculating the mooring force of the floating structure and testing data responded by six degrees of freedom, normalizing the testing data to form a test set, inputting the six degrees of freedom movements in the test set into the training neural network model, and testing the training neural network model;
d2, comparing the test result with the test set sea state time domain coupling calculation result, if the preset condition is met, determining the training set sea state interval as an estimation interval of the environmental factors, if the preset condition is not met, reducing the difference between the environmental factors and the rated sea state, and returning to the step D1 until the test result and the time domain coupling calculation result meet the preset condition;
d3: dividing the total training set sea state according to the estimated interval and rated sea state of the environmental factors: dividing critical sea states according to the determined environmental factor estimation interval length and the rated sea states of the floating structure design, setting two adjacent critical sea states as a group to obtain a group training set sea state, superposing the group training set sea states to form a total training set sea state, inputting the total training set sea state into a time domain coupling calculation model to perform time domain calculation to obtain total training data, normalizing the total training data to form a total training set, inputting the total training set into an optimal neural network model to perform training, and obtaining a training neural network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310698549.1A CN116451591A (en) | 2023-06-14 | 2023-06-14 | Ocean floating structure mooring force assessment method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310698549.1A CN116451591A (en) | 2023-06-14 | 2023-06-14 | Ocean floating structure mooring force assessment method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116451591A true CN116451591A (en) | 2023-07-18 |
Family
ID=87134127
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310698549.1A Pending CN116451591A (en) | 2023-06-14 | 2023-06-14 | Ocean floating structure mooring force assessment method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116451591A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109871609A (en) * | 2019-02-18 | 2019-06-11 | 中国海洋大学 | The prediction technique that marine floating type platform mooring system is responded based on BP-FEM |
| CN110826290A (en) * | 2019-10-31 | 2020-02-21 | 中国海洋大学 | Safety early warning method for offshore floating system |
| CN111241740A (en) * | 2020-01-25 | 2020-06-05 | 哈尔滨工程大学 | Fast and accurate calculation method for FPSO soft rigid arm stress |
| CN114580152A (en) * | 2022-02-10 | 2022-06-03 | 中国电建集团华东勘测设计研究院有限公司 | Floating wind power structure foundation local stress time domain analysis method based on multi-body coupling analysis |
-
2023
- 2023-06-14 CN CN202310698549.1A patent/CN116451591A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109871609A (en) * | 2019-02-18 | 2019-06-11 | 中国海洋大学 | The prediction technique that marine floating type platform mooring system is responded based on BP-FEM |
| CN110826290A (en) * | 2019-10-31 | 2020-02-21 | 中国海洋大学 | Safety early warning method for offshore floating system |
| CN111241740A (en) * | 2020-01-25 | 2020-06-05 | 哈尔滨工程大学 | Fast and accurate calculation method for FPSO soft rigid arm stress |
| CN114580152A (en) * | 2022-02-10 | 2022-06-03 | 中国电建集团华东勘测设计研究院有限公司 | Floating wind power structure foundation local stress time domain analysis method based on multi-body coupling analysis |
Non-Patent Citations (1)
| Title |
|---|
| 杨洁: "《基于深度学习的单点系泊系统动态张力预测方法研究》", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》, no. 5, pages 27 - 44 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hegseth et al. | Integrated design optimization of spar floating wind turbines | |
| Dong et al. | Long-term fatigue analysis of multi-planar tubular joints for jacket-type offshore wind turbine in time domain | |
| Si et al. | Modelling and optimization of a passive structural control design for a spar-type floating wind turbine | |
| Jiang et al. | A parametric study on the final blade installation process for monopile wind turbines under rough environmental conditions | |
| Ciuriuc et al. | Digital tools for floating offshore wind turbines (FOWT): A state of the art | |
| Ren et al. | Active heave compensation of floating wind turbine installation using a catamaran construction vessel | |
| Wei et al. | Directional effects on the reliability of non-axisymmetric support structures for offshore wind turbines under extreme wind and wave loadings | |
| Li et al. | Splash zone lowering analysis of a large subsea spool piece | |
| Li et al. | Operability analysis of monopile lowering operation using different numerical approaches | |
| Hines et al. | Structural instrumentation and monitoring of the block island offshore wind farm | |
| Sheng et al. | Reliability and fragility assessment of offshore floating wind turbine subjected to tropical cyclone hazard | |
| Harnois | Analysis of highly dynamic mooring systems: peak mooring loads in realistic sea conditions | |
| Bachynski et al. | Linear and nonlinear analysis of tension leg platform wind turbines | |
| Yeter et al. | Fatigue reliability assessment of an offshore supporting structure | |
| CN116341358A (en) | Large buoy motion response and anchoring tension prediction method combined with deep learning | |
| Tang et al. | Real-time monitoring system for scour around monopile foundation of offshore wind turbine | |
| Hong et al. | Hydrodynamic and environmental modelling influence on numerical analysis of an innovative installation method for floating wind | |
| CN116451591A (en) | Ocean floating structure mooring force assessment method based on deep learning | |
| Lim et al. | Direct Time Domain Life Cycle Loading Analysis on a Floating Platform | |
| CN113283138B (en) | Deep-learning-based dynamic response analysis method for deep-sea culture platform | |
| Isnaini et al. | Real-time wave prediction for floating offshore wind turbine based on the kalman filter | |
| CN110826290B (en) | Safety early warning method for offshore floating system | |
| Isnaini et al. | Real-time prediction of incoming wave profile surrounding floating offshore wind turbine using kalman filter | |
| KR102121246B1 (en) | Damage detection method for mooring lines of submersible structures based on deep learning | |
| Beurskens | Converting Offshore Wind into Electricity |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20230718 |