CN114091339A - Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU - Google Patents

Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU Download PDF

Info

Publication number
CN114091339A
CN114091339A CN202111409133.0A CN202111409133A CN114091339A CN 114091339 A CN114091339 A CN 114091339A CN 202111409133 A CN202111409133 A CN 202111409133A CN 114091339 A CN114091339 A CN 114091339A
Authority
CN
China
Prior art keywords
data
model
gru
instantaneous yield
bow
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
Application number
CN202111409133.0A
Other languages
Chinese (zh)
Inventor
章亮
袁伟
俞孟蕻
苏贞
周泊龙
齐亮
杨奕飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University of Science and Technology
Original Assignee
Jiangsu University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu University of Science and Technology filed Critical Jiangsu University of Science and Technology
Priority to CN202111409133.0A priority Critical patent/CN114091339A/en
Publication of CN114091339A publication Critical patent/CN114091339A/en
Pending legal-status Critical Current

Links

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
    • 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
    • G06F2119/22—Yield analysis or yield optimisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a bow blowing instantaneous yield prediction method of a trailing suction dredge based on GRU. The method comprises the steps of firstly, collecting historical construction data and instantaneous yield data of bow blowing of the drag suction dredger; analyzing according to the relevant mathematical model and the bow blowing process, selecting variables related to instantaneous yield prediction as the input of the model, and taking the instantaneous yield as the output; and then preprocessing the selected data, including denoising, abnormal value processing and normalization processing. Dividing the preprocessed data into training set data and verification set data; then establishing a GRU neural network model for predicting bow blowing instantaneous yield of the trailing suction dredge; and finally, training the model by using the training set data, and verifying the model by using the verification set data to obtain a prediction model of the instantaneous yield of the trailing suction hopper dredger. The prediction method is accurate, rapid and stable.

Description

Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU
Technical Field
The invention relates to a prediction method of engineering dredging yield, in particular to a bow blowing instantaneous yield prediction method of a trailing suction dredger based on GRU.
Background
Dredging engineering is the widening and deepening of an excavated area with special mechanical equipment of earthwork sediments, sand or rocks excavated under water. People have higher and higher requirements on the performance, efficiency, automation level and environmental protection level of the trailing suction dredger, and the improvement of the performance and efficiency of the dredger becomes the key point of dredging work. Bow blowing is one of the common hydraulic fill construction methods of a drag suction dredger, and is mainly used for sand taking in deep offshore areas and for bow blowing and land building in offshore areas, such as a girth project in the area of the Yangtze river mouth, a container wharf project in the south port of the Schland Karren slope and a port expansion project in the Abbe of Kotedawa. The bow blowing operation mode of the drag suction dredger is widely applied. But aiming at the complexity of two flows, the mechanism complexity of bow blowing is difficult to carry out mechanical analysis and accurate prediction on bow blowing. And as an important index for evaluating the bow blowing construction efficiency of the trailing suction hopper dredger, the prediction of the bow blowing productivity of the trailing suction hopper dredger is beneficial to improving the construction efficiency, and the optimal operating parameters of the dredger are selected to realize the optimal control of the dredger. Therefore, the prediction of the instantaneous dredging yield of the drag suction dredger has important significance for optimizing the dredging process and controlling the dredging cost.
Chinese patent CN201921046801.6 discloses a novel dredger yield meter, which uses a detection cylinder side wall with a driving mechanism to be rotatably connected with a retaining sleeve, so as to detect from different angles and improve the reliability of detection data; chinese patent CN202010875940.0 discloses a dredger yield meter, which further obtains the dredger yield by measuring the slurry concentration and the pipeline flow in the dredger pipeline. The passive sensor based on the electrical tomography method improves the accuracy of the transmission concentration and the flow measurement precision of the liquid-solid two-phase fluid pipeline. Although the instantaneous yield of the dredger can be directly and conveniently measured by using the yield meter, the instantaneous yield of the dredger cannot be estimated in advance, and the prediction effect cannot be achieved.
The dredging process of the trailing suction hopper dredger is a time continuous construction process, the generated construction data is data based on time series, the problem of time lag exists between the data, the GRU neural network has good processing capacity on the time series data, and a good prediction effect can be obtained on nonlinear and multistep prediction problems, so that a model for predicting the bow blowing instantaneous yield of the trailing suction hopper dredger is constructed by adopting the GRU neural network. The existing method for predicting the bow blowing instantaneous yield of the drag suction dredger in the prior art is less, has low precision, and does not consider the problems that the actual construction data of the dredger is based on the data of a time sequence.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a high-precision method for predicting bow blowing instantaneous yield of a trailing suction dredge based on GRUs.
The technical scheme is as follows: the invention provides a bow blow instantaneous yield prediction method of a trailing suction dredge based on GRU, which comprises the following steps:
step 1: and collecting historical construction data and instantaneous yield data of bow blowing of the drag suction dredger. The construction process of the trailing suction hopper dredger is complex, a plurality of key devices of the trailing suction hopper dredger are needed, and therefore collected historical data can also comprise a plurality of device operation data. And the descriptors for the instantaneous production of the trailing suction hopper dredger mainly include the pipe flow and the mud concentration.
Step 2: the pipe flow and mud concentration, which are more direct to the instantaneous production, are selected as model outputs. By analyzing a related mathematical model and a bow blowing process, starting from a direct calculation formula of yield and a calculation formula of slurry concentration in a pipeline, and analyzing according to knowledge of related industries, factors such as the rotating speed of a mud pump, the opening degree of a mud door and the like can influence or reflect the instantaneous yield of bow blowing of the drag suction dredger to a great extent, so that the rotating speed of the mud pump, the stroke of 14 pumping doors, the opening degree of a water guide valve, the rotating speed of a high-pressure flushing pump, the slurry concentration and the flow of the pipeline are selected as the factors influencing the yield as input variables of the model.
And step 3: and extracting actual data related to the steps from the collected historical data, and preprocessing the selected data. The preprocessing content comprises the filtering processing, the abnormal value processing and the normalization processing of the data. And the data after the preprocessing is subjected to data set division.
And 4, step 4: and establishing a GRU neural network prediction model for dredger yield prediction, and further, compared with a hidden layer of an original RNN network, the GRU neural network has only one state which is very sensitive to short-term input, and the GRU network adds a unit state to store a long-term state. The GRU unit mainly comprises an input gate, a forgetting gate and an output gate, and realizes and controls information through the three basic structures. The forgetting gate determines how much the unit state at the previous moment is reserved to the current moment; the input gate determines how much input of the network is saved to the unit state at the current moment; the reset gate can discard the history information which is irrelevant to prediction, and the update gate is used for controlling the degree of memorizing the information at the previous moment, thereby being beneficial to capturing long-term dependency relationship.
And 5: and training the model by using the training set data, selecting the model with the best Root Mean Square Error (RMSE) and the best average percent error (MAPE), namely the model with the lowest Root Mean Square Error (RMSE) and the lowest average percent error (MAPE), as the trained model by using the Root Mean Square Error (RMSE) and the average percent error (MAPE) as model evaluation indexes, and verifying the trained model by using the verification set data to obtain the bow blowing instantaneous yield prediction model of the drag suction dredger.
Compared with the prior art, the construction characteristics of the trailing suction hopper dredger are fully considered, the yield prediction model is established by adopting the gate control cycle unit (GRU), the prediction precision is improved, and the problems of long-term dependence and lag between time sequence data are solved.
Has the advantages that: compared with the prior art, the invention has the following advantages:
1. the gated cyclic unit (GRU) neural network is used as an important branch of the Recurrent Neural Network (RNN), and simultaneously forgets and memorizes information by using the same gated unit, namely an update gate, so that the calculation is more convenient, the training efficiency is improved, and a prediction result can be obtained more accurately, quickly and stably.
2. Aiming at the characteristics of mutual operation and mutual influence among key equipment of the trailing suction hopper dredger in the dredging engineering and the problems of lag and dependence among data, a gated circulation unit neural network (GRU) is selected, so that the problems can be well solved, and the long-time step length prediction can be realized.
3. Through the analysis of the mathematical model and the combination of related industry knowledge and operation experience, the input and output variables of the model are selected, and the advantages of mechanism-based modeling and data-driven modeling are fully utilized.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The invention relates to a method for predicting bow blow instantaneous yield of a trailing suction dredge based on GRU, which comprises the steps of firstly collecting bow blow data of the trailing suction dredge, wherein the bow blow data comprises historical construction data and corresponding instantaneous yield related data; then analyzing the existing parameters according to the bow blowing process, and determining the input and output parameters of the model; then, preprocessing selected parameter data, including data filtering, abnormal value processing and normalization processing, and dividing the preprocessed data, wherein the first 80% of the preprocessed data are used as training set data and verification set data; then, the GRU neural network model is used for forecasting the output of the dredger; and finally, training the model by using the training set data, and verifying the model by using the verification set data to obtain a bow blowing instantaneous yield prediction model of the drag suction dredger.
As shown in fig. 1, the method for predicting bow blow instantaneous yield of a GRU-based trailing suction dredge according to the present invention comprises the following steps:
step 1: historical construction data and instantaneous yield data of the trailing suction hopper dredger are collected. The construction process of the trailing suction hopper dredger is complex, a plurality of key devices of the trailing suction hopper dredger are needed, therefore, collected historical data can also comprise a plurality of device operation data, up to 306 dimensions, wherein the description quantity of the instantaneous yield of the trailing suction hopper dredger mainly comprises mud concentration and pipeline flow. It is impractical to use all the data for production prediction, so the following steps are required to select the variables.
Step 2: according to practical experience analysis, for the same drag suction dredger, the density and the flow of the silt mixture of the bow-blowing sludge discharge pipeline determine the actual yield of bow-blowing dredging. Dredger output Qs，m3The product of the volume flow of the slurry and the volume concentration of the solid is:
Qs＝Qm·Cvd·3600
wherein: qs(m3In/s) is yield, Qm(m3S) is the mud volume flow, Cvd(%) is the solids delivery volume concentration.
Step 2.1: from the above steps it can be seen that the mud concentration directly affects the production rate of the dredger, and therefore the mud concentration in the pipeline is analysed:
Figure BDA0003372107490000031
wherein d is the inner diameter of the pipeline, V is the mud flow rate, and V ismIs the volume of silt entering the pipe in a unit time
Step 2.2: it can be seen from the above steps that under the conditions that the inner diameter of the pipeline is unchanged and the flow rate of the slurry is not changed greatly, the volume of the silt entering the pipeline determines the concentration of the slurry in the pipeline, the volume of the silt entering the pipeline is mainly controlled by the pumping door, the opening degree of the pumping door directly determines the volume of the silt entering the mud pipe, and the high-pressure flushing water on the upper side of the pumping door also has great influence on the silt entering the pipeline.
Step 2.3: the change of the slurry concentration can be well reflected by analyzing the opening of the pumping chamber and the rotating speed of the high-pressure flushing pump according to the operation experience and the knowledge of related industries, and the rotating speed of the slurry pump influences the flow of the pipeline to a great extent. Therefore, in combination with the analysis process, mud flow, the rotating speed of the underwater pump, the suction vacuum of the underwater pump, the traversing speed and the current of the reamer motor are selected as the input variables of the model, wherein the factors influence the yield.
And step 3: extracting actual data of the steps related to six parameters such as yield per second, slurry flow rate, underwater pump rotating speed and the like from the collected historical data, and preprocessing the selected data, wherein the specific steps of data preprocessing are as follows:
step 3.1: filtering the data by adopting wavelet threshold denoising, and removing noise and interference signals in the data;
step 3.2: abnormal value processing is carried out on the data by adopting a Leitet criterion (3 times error method), and the construction data is measured for N times to obtain x1，x2，x3，L，xnWhen data xkCorresponding residual error RkThe following formula is satisfied, and if sigma is constant, the data is considered to be an abnormal value,
Figure BDA0003372107490000041
in the formula, x1，x2，x3，L，xnFor the construction data of the N measurements,
Figure BDA0003372107490000042
is the average of construction data of N measurements, RkIs the residual, σ is a constant.
Step 3.3: the data are normalized, and the specific method is as follows:
Figure BDA0003372107490000043
wherein X is normalized data,
Figure BDA0003372107490000044
for the raw data, XminIs the minimum value, X, in the raw datamaxIs the maximum value in the raw data.
Step 3.4: and dividing the preprocessed data into training set data and verification set data, wherein the first 80% of the data is used as a training sample set, and the rest of the data is used as a verification sample set.
And 4, step 4: establishing a GRU neural network prediction model for predicting the output of the dredger: the whole network structure mainly comprises an input layer, an output layer and a hidden layer. The GRU neural network is a variant of the recurrent neural network, and simultaneously forgets and memorizes information by using the same gate control unit, namely an updating gate, so that the calculation is more convenient and faster, and the training efficiency is improved.
The forward propagation formula of the GRU network comprises the following specific steps:
Figure BDA0003372107490000051
in the formula: xtInput, H, as a hidden layertHidden layer output, Ht-1Output of last moment, YtIs the output of the output layer, and RtTo reset the gate output, ZtFor updating the door input,
Figure BDA0003372107490000052
As a candidate memory cell output, RtTo reset the gate state, ZtTo update the door state, Whr、Wxr、Whz、Wxz、
Figure BDA0003372107490000053
Why、br、bz、
Figure BDA0003372107490000054
Are parameters.
In the above formula, Xt、HtInput and output, Y, of the hidden layer, respectivelytIs the output of the output layer, and Rt、ZtAnd
Figure BDA0003372107490000055
then respectively in the hidden layer structureA reset gate output, a refresh gate output, and a candidate memory cell output. In learning the structure of the GRU network, first, the output H at the previous time is passedt-1And input X of the current timetTo obtain two gating states-reset gate RtAnd a refresh door ZtBoth of which output a value range of [0, 1 ]]The value of (c). The reset gate determines how the hidden state at the last moment flows into the candidate hidden state at the current moment, so the reset gate can discard the historical information which is not related to prediction, and the capture of the short-term dependency relationship is facilitated. The update gate is used to control the degree of information being memorized at the previous moment, which is helpful to capture the long-term dependency relationship. Parameter Whr、Wxr、Whz、Wxz、
Figure BDA0003372107490000056
Why、br、bz、
Figure BDA0003372107490000057
The learning update process of (2) is an error back-propagation process, wherein the partial derivatives of the output of the hidden layer at time t-1 are composed of the partial derivatives of the inputs of the gates at time t:
Figure BDA0003372107490000058
in the formula, deltah，t、
Figure BDA0003372107490000059
δz，t、δr，tThe partial derivatives of the hidden layer, the candidate memory unit, the update gate and the reset gate of the network at the time t are respectively.
And 5: training a model by using the training set data, and verifying the trained model by using the verification set data to obtain a bow blowing instantaneous yield prediction model of the drag suction dredger, which comprises the following specific steps:
step 5.1: and training the constructed GRU network by using the data of the divided training sample set, and taking a Root Mean Square Error (RMSE) and a mean percentage error (MAPE) as indexes as model evaluation indexes.
Step 5.2: determining the coefficient R2The specific calculation process of (2) is as follows:
Figure BDA00033721074900000510
Figure BDA00033721074900000511
in the formula, yiRepresenting the ith real value in the dataset,
Figure BDA00033721074900000512
denotes the ith prediction value, and n denotes the number of values.
Step 5.3: and when the model is trained, carrying out fine adjustment updating on the model according to the root mean square error and the average percentage error to obtain a relatively accurate prediction model. And selecting the model with the minimum root mean square error and average percentage error as a trained model, and verifying the trained model by using the verification set data to obtain a bow blowing instantaneous yield prediction model of the drag suction dredger.
The invention solves the defects in the existing bow blowing instantaneous yield prediction algorithm of the drag suction dredger, provides the method for predicting the bow blowing instantaneous yield of the drag suction dredger based on the GRU algorithm, fully utilizes the memory of the GRU network to data, and improves the accuracy of the bow blowing instantaneous yield prediction of the drag suction dredger.

Claims (6)

1. A bow blowing instantaneous yield prediction method of a trailing suction dredge based on GRU is characterized by comprising the following steps:
step 1: collecting historical construction data and instantaneous yield data of the dredger;
step 2: selecting variables related to instantaneous yield prediction as input of the model and instantaneous yield as output by a relevant mathematical model and a bow blow process;
and step 3: preprocessing the selected data, and dividing the preprocessed data into training set data and verification set data;
and 4, step 4: establishing a GRU neural network prediction model for predicting bow blowing instantaneous yield of the trailing suction dredge;
and 5: and training the model by using the training set data, selecting the root mean square error and the average percentage error as model evaluation indexes, and verifying the trained model by using the verification set data to obtain a bow blowing instantaneous yield prediction model of the drag suction dredger.
2. The method for predicting bow blow instantaneous yield of a GRU-based trailing suction hopper dredger according to claim 1, wherein the historical construction data in step 1 comprises undisturbed soil weight, earthwork amount, current mud tank liquid level, current mud tank capacity, left high-pressure flushing pump speed, right high-pressure flushing pump speed, mud pump density, mud pump flow rate, left 1 pumping compartment door stroke, left 2 pumping compartment door stroke, and time during the bow blow construction of the trailing suction hopper dredger.
3. The method for predicting bow blow instantaneous yield of a GRU-based trailing suction dredge according to claim 2, wherein the step 2 is analyzed according to the existing mathematical model and related industry knowledge, and the mud pump rotating speed, 14 pumping door strokes, the opening degree of a water diversion valve, the high-pressure flushing pump rotating speed, the mud concentration and the pipeline flow are selected as the input variables of the model; the mud concentration and pipe flow rate directly representing the instantaneous production are selected as model outputs.
4. The method for predicting bow blow instantaneous yield of a GRU-based trailing suction dredge according to claim 1, characterized in that the collected data is preprocessed in step 3 to remove the interference signals and abnormal values in the data, and the data is converted to the interval range of 0-1 through the normalization process of the data, and the data set is divided.
5. The method for predicting bow blow instantaneous yield of a GRU-based trailing suction dredge according to claim 1, wherein the GRU neural network prediction model is established in step 4, the GRU neural network prediction model comprises a GRU input layer, a GRU neural network layer, a full connection layer and an output layer, data is input into the network through the input layer, and the final prediction value is output from the output layer through the processing of the GRU neural network layer and the full connection layer.
6. The method for predicting bow blow instantaneous yield of a GRU-based trailing suction dredge according to any one of claims 1-4, wherein the prediction method of step 5 is as follows:
step 5.1: training the constructed GRU network by using the data of the divided training sample set, and taking a root mean square error and an average percentage error as indexes to serve as model evaluation indexes;
step 5.2: determining the coefficient R2The specific calculation process of (2) is as follows:
Figure FDA0003372107480000021
Figure FDA0003372107480000022
in the formula, yiRepresenting the ith real value in the dataset,
Figure FDA0003372107480000023
representing the ith predicted value, and n represents the number of numerical values;
step 5.3: when the model is trained, the model is subjected to fine adjustment updating according to the root mean square error and the average percentage error to obtain a relatively accurate prediction model, the model with the smallest root mean square error and the smallest average percentage error is selected as the trained model, and the trained model is verified by using the verification set data to obtain the bow blowing instantaneous yield prediction model of the drag suction dredger.
CN202111409133.0A 2021-11-24 2021-11-24 Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU Pending CN114091339A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111409133.0A CN114091339A (en) 2021-11-24 2021-11-24 Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111409133.0A CN114091339A (en) 2021-11-24 2021-11-24 Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU

Publications (1)

Publication Number Publication Date
CN114091339A true CN114091339A (en) 2022-02-25

Family

ID=80304219

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111409133.0A Pending CN114091339A (en) 2021-11-24 2021-11-24 Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU

Country Status (1)

Country Link
CN (1) CN114091339A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114662782A (en) * 2022-04-08 2022-06-24 江苏科技大学 Method for predicting instantaneous yield of trailing suction hopper dredger based on LSTM neural network
CN115600746A (en) * 2022-10-24 2023-01-13 哈尔滨工程大学（Cn） Convolutional neural network-based drag suction ship energy efficiency prediction method

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114662782A (en) * 2022-04-08 2022-06-24 江苏科技大学 Method for predicting instantaneous yield of trailing suction hopper dredger based on LSTM neural network
CN115600746A (en) * 2022-10-24 2023-01-13 哈尔滨工程大学（Cn） Convolutional neural network-based drag suction ship energy efficiency prediction method

Similar Documents

Publication Publication Date Title
CN109978226B (en) Shield construction ground settlement prediction method based on cyclic neural network
Bai et al. Data mining approach to construction productivity prediction for cutter suction dredgers
CN114091339A (en) Method for predicting bow blowing instantaneous yield of drag suction dredger based on GRU
CN113779690B (en) Advanced geological prediction method and system based on while-drilling sensing
CN109543828B (en) Water absorption profile prediction method based on small sample condition
CN113553780B (en) Stratum pore pressure prediction method based on machine learning
CN110826790A (en) Intelligent prediction method for construction productivity of cutter suction dredger
CN102169077B (en) Hybrid intelligent soft measuring method of overflow granularity index in wet grinding process
CN111967079A (en) Foundation pit deformation prediction method based on improved artificial bee colony algorithm and BP neural network
CN113642801A (en) Cutter suction dredger yield prediction method based on LSTM
CN111460737B (en) Intelligent settlement prediction method and system for slurry air pressure balance shield
CN113792479A (en) Physical constraint-based coalbed methane reservoir fracturing effect evaluation method
CN114661006A (en) Optimized control system and method for cabin pumping and bank blowing process of trailing suction hopper dredger
CN111810115A (en) Underground real-time monitoring method and device for formation water pollution rate
CN115293316A (en) Prediction method for deep-buried thick coal seam water flowing fractured zone based on SSA-ELMAN
CN113486806B (en) Suspended sand monitoring method based on one-dimensional residual neutral network
CN114488311A (en) Transverse wave time difference prediction method based on SSA-ELM algorithm
CN117251802B (en) Heterogeneous reservoir parameter prediction method and system based on transfer learning
CN112765880B (en) Method for monitoring stratum saturated brine invasion amount based on Bi-LSTM
Bai et al. A new measuring method of dredging concentration based on hybrid ensemble deep learning technique
CN115952590A (en) Shield tunnel optimization design method and device based on BO-RF-MOMPA hybrid algorithm
CN115929285A (en) Geothermal gradient prediction method based on Lagrange support vector machine algorithm
CN114662782A (en) Method for predicting instantaneous yield of trailing suction hopper dredger based on LSTM neural network
Mokhtari et al. Feasibility study of EPB shield automation using deep learning
CN106919725A (en) Synthetic rubber conversion ratio online test method and device based on intelligent algorithm

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