Dynamic optimization method and system for cemented riprap construction parameters in wave flow environment

Abstract

The invention provides a dynamic optimization method and system for cemented riprap construction parameters in a wave flow environment, and relates to the technical field of construction parameter determination. The method comprises the following steps: firstly, extracting time sequence information of various environmental parameters in a specified construction area from multiple dimensions, and carrying out space-time alignment processing on the various environmental parameters, secondly, recording the construction parameter information in real time, and establishing an association relationship with the environmental parameter information and construction effect information by taking a construction round as a unit, and finally, carrying out time-space alignment processing on the environmental parameters. And training based on the data to obtain a time sequence prediction model for representing the construction condition, so that the optimal construction parameter information can be predicted through the environment parameters and the expected construction effect information in the future preset time interval. By means of the technical scheme, the cemented riprap construction parameters in the wave flow environment can be accurately predicted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of construction parameter determination, in particular to a dynamic optimization method and system for construction parameters of cemented riprap under wave-current environment. BACKGROUND

[0002] In a strong current and wave-current coupled marine environment, the traditional cemented riprap construction has problems such as difficulty in real-time matching of riprap particle size, riprap throwing height, cementing material pouring speed and other parameters with flow field changes, resulting in unstable cementing effect, material waste and decline in performance of the protection structure. The prior art lacks dynamic adjustment capability of construction parameters, and cannot realize intelligent construction. SUMMARY

[0003] The purpose of the present application is to provide a dynamic optimization method and system for construction parameters of cemented riprap under wave-current environment, which can solve the technical problem of inaccurate prediction of construction parameters in the cemented riprap construction process.

[0004] The present application provides a dynamic optimization method for construction parameters of cemented riprap under wave-current environment, which comprises the following steps: S1: collecting at least one multi-source data sequence, and performing data optimization extraction processing on the at least one multi-source data sequence to obtain at least one first flow field feature vector; S2: obtaining at least one construction parameter feature vector and at least one first construction effect information in the construction process, and determining a space-time correlation database according to the correlation between the at least one construction parameter feature vector, the at least one first construction effect information and the at least one first flow field feature vector; S3: predicting target construction parameter information according to a target flow field feature vector and target construction effect information; S4: after confirming the target construction parameter information, performing issuing and recording processing.

[0005] Preferably, the S1 comprises: S11: arranging at least one first multi-source sensor in a first ship and a first construction range to obtain a flow velocity data sequence, a wave data sequence and a ship motion data sequence; S12: performing space-time alignment fusion processing on the flow velocity data sequence, the wave data sequence and the ship motion data sequence to obtain a target flow velocity data sequence, a target wave data sequence and a target ship motion data sequence; S13: performing data quality optimization processing and feature extraction processing on the target flow velocity data sequence, the target wave data sequence and the target ship motion data sequence to obtain at least one first flow field feature vector.

[0006] Preferably, in each of the first flow field feature vectors, a combined vector of flow velocity data, wave data and ship movement data in a specified time interval is included.

[0007] Preferably, the S2 comprises: S21: Real-time data acquisition of construction parameters, thereby obtaining at least one construction parameter feature vector; S22: Determining at least one first construction effect information according to the first digital elevation model obtained after each construction; S23: Combining at least one of the first flow field feature vectors, at least one of the construction parameter feature vectors and at least one of the first construction effect information into a space-time correlation database.

[0008] Preferably, the S22 comprises: S221: After each construction is completed, full-coverage topographic survey of the construction area is performed, thereby obtaining a first digital elevation model; S222: Obtaining first cemented body information according to the first digital elevation model and the original digital elevation model; S223: Obtaining first construction effect information according to the first cemented body information and first design blueprint information.

[0009] Preferably, at least one data record is included in the space-time correlation database.

[0010] Preferably, the S3 comprises: S31: Training a construction parameter prediction model according to the space-time correlation database; S32: Inputting a target flow field feature vector and target construction effect information into the construction parameter prediction model to obtain target construction parameter information; S33: When the amount of construction records accumulated exceeds a first preset value, incrementally learning the construction parameter prediction model.

[0011] Preferably, in the training process of the construction parameter prediction model, the flow field feature vector and the construction effect information in the sample data are used as input data, and the construction parameter feature vector is used as output data.

[0012] Preferably, the S4 comprises: S41: Displaying the target construction parameter information on a digitalized board, and performing a delivery operation after a confirmation operation; S42: Recording the target construction parameter information and parameter adjustment information into an operation log.

[0013] The application also proposes a dynamic optimization system for cemented riprap construction parameters in a wave flow environment, which is used to implement the dynamic optimization method for cemented riprap construction parameters in a wave flow environment.

[0014] The wave flow environment under the cemented riprap construction parameter dynamic optimization method and system provided in the application relates to the technical field of construction parameter determination. First, the time sequence information of various environmental parameters in the specified construction area is extracted from multiple dimensions, and the various environmental parameters are subjected to spatiotemporal alignment processing. Second, the construction parameter information is recorded in real time, and the environmental parameter information and construction effect information are associated according to the construction round. Finally, a time sequence prediction model for representing the construction condition is obtained based on the above data, so that the best construction parameter information can be predicted based on the environmental parameters and the expected construction effect information in the future preset time interval. The technical solution of the application can accurately predict the cemented riprap construction parameters under the wave flow environment. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and other drawings can be obtained by the provided drawings without creative labor for those skilled in the art.

[0016] Figure 1 is the execution flowchart of the wave flow environment under the cemented riprap construction parameter dynamic optimization method in the embodiment of the present application.

[0017] Figure 2 is the layout schematic diagram of the flow velocity data acquisition sensor in the embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0020] The wave flow environment under the cemented riprap construction parameter dynamic optimization method and system of the present application will be described in detail below.

[0021] The wave flow environment under the cemented riprap construction parameter dynamic optimization method of the present embodiment is proposed, which aims to accurately predict the riprap construction parameters by combining big data and machine learning models. The specific process is as follows Figure 1As shown.

[0022] S1: Collect at least one multi-source data sequence, and perform data optimization extraction processing on at least one of the multi-source data sequences to obtain at least one first flow field feature vector.

[0023] This step aims to build a high-precision, high-frequency on-site environmental perception system. By deploying multiple sensors and coordinating their calibration and preprocessing, the system comprehensively and reliably captures dynamic flow field information in the construction area, providing a data foundation for subsequent intelligent decision-making.

[0024] The data optimization and extraction process includes steps such as spatiotemporal alignment, data optimization, and feature extraction.

[0025] S1 specifically includes the following sub-steps: S11: Deploy at least one first multi-source sensor within the first vessel and the first construction area to obtain flow velocity data sequence, wave data sequence and vessel motion data sequence.

[0026] To ensure that the subsequent rock-dropping parameters are accurately aligned with environmental and rock-dropping data, this step requires the deployment of appropriate sensors at designated points within the construction area and at designated locations on the vessel. These sensors will collect data on flow velocity, wave patterns, and vessel motion, thus preparing the data for predicting the subsequent rock-dropping parameters.

[0027] The specific acquisition process for the above-mentioned data types is as follows: 1. Flow velocity data acquisition like Figure 2 As shown, at least four Nortek Vectrino series point-type acoustic Doppler current meters are installed at node A 10m upstream, node B 10m downstream, and nodes C and D 5m to the sides of the foundation structure (such as a monopile). The sampling frequency can preferably be set to 16Hz to measure the three-dimensional instantaneous flow velocity at key points around the pile.

[0028] Meanwhile, a Nortek Signature 1000 ADCP is mounted on the bow of the construction vessel to collect complete velocity profile data from the seabed to the water surface below the waterline at a frequency of 1 Hz. The complete velocity profile data can be obtained by selecting multiple data points at specified distance intervals and combining multiple data points to form the complete velocity profile data.

[0029] A velocity data vector is constructed using three-dimensional instantaneous flow velocity and complete velocity profile data at key points around the pile. Preferably, the two velocity data sets can be simply concatenated to form the velocity data vector. Each velocity data vector corresponds to a specific time point.

[0030] 2. Wave data acquisition Three Valeport wave direction meters are arranged in a triangle outside the construction area, with a sampling frequency of 2 Hz, to synchronously measure wave height, wave period and wave direction, and the influence of ship motion on wave measurement is eliminated by a triangulation method. The specific operation process of the triangulation method can be known from the prior art.

[0031] After wave data acquisition, a wave data vector for representing wave data characteristics at a specified time point is obtained from wave height, wave period and wave direction data at the specified time point.

[0032] 3. Ship motion data acquisition An Applanix POS MV WaveMaster series high-precision integrated navigation system is installed at the center of gravity of the construction ship to output six-degree-of-freedom data (longitude, latitude, altitude, roll, pitch and yaw) of the ship at a frequency of 100 Hz. Exemplary accuracy requirements: plane positioning ≤2 cm, altitude ≤5 cm, attitude angle ≤0.05°.

[0033] The ship motion data is also combined from the six-degree-of-freedom data to obtain a ship motion vector at a specified time point.

[0034] Through the above steps, the flow velocity data vector, the wave data vector and the ship motion vector at different time points can be obtained.

[0035] After combining the data vectors at different time points, a flow velocity data sequence, a wave data sequence and a ship motion data sequence can be obtained.

[0036] S12: performing spatio-temporal alignment fusion processing on the flow velocity data sequence, the wave data sequence and the ship motion data sequence to obtain a target flow velocity data sequence, a target wave data sequence and a target ship motion data sequence.

[0037] In this step, spatio-temporal alignment fusion processing needs to be performed on the multiple data sequences obtained in S11, so that the processed data is more consistent with the actual situation. The spatio-temporal alignment fusion processing includes two steps of time alignment and space alignment.

[0038] For the time alignment processing process, a time series data synchronization module needs to be deployed on an industrial edge computing industrial computer. The module takes the PPS signal of GPS as an absolute time reference, synchronizes the internal clocks of all sensors through the PTP precise clock protocol, and ensures that the timestamp deviation of all data is less than a preset time interval, preferably 10 ms.

[0039] For the space alignment processing process, a coordinate unification conversion module needs to be developed to convert all sensor data to a local coordinate system of the construction area, preferably with a single pile center as the origin.

[0040] S13: performing data quality optimization processing and feature extraction processing on the target flow rate data sequence, the target wave data sequence, and the target ship motion data sequence to obtain at least one first flow field feature vector.

[0041] In the S12, the spatio-temporal alignment processing of various data sequences has been completed. In this step, further data quality optimization processing and feature extraction are needed to be performed on all data, so as to serve as input data for subsequent steps.

[0042] The data quality optimization processing step includes: for each type of original data, applying a phase space threshold method to remove spikes, and then using a 4th order low-pass Butterworth filter to set a specified cutoff frequency to eliminate high-frequency noise.

[0043] The feature extraction step includes: for each type of processed time series data, calculating the flow field feature values in each window in real time with a time window of 5 minutes. Taking the flow rate data as an example, the flow field feature values can include: average flow rate U_mean, flow rate standard deviation U_std, maximum flow rate U_max, average wave height Hs, and main wave period Tp. These feature values will serve as input vectors for the machine learning model.

[0044] At least one first flow field feature vector is formed by the extracted features. In each of the first flow field feature vectors, a combined vector of data values in a specified time interval is included. The specified time interval is preferably 5 minutes.

[0045] S2: obtaining at least one construction parameter feature vector and at least one first construction effect information during the construction process, and determining a spatio-temporal correlation database according to the correlation between at least one construction parameter feature vector, at least one first construction effect information, and at least one first flow field feature vector.

[0046] This step aims to create a structured, spatio-temporal correlation historical database, accurately correlate construction operations, environmental conditions, and final effects, and form a high-quality "problem-solution" sample library to prepare for subsequent model training.

[0047] The S2 includes the following sub-steps: S21: performing real-time data acquisition on construction parameters to obtain at least one construction parameter feature vector.

[0048] In this step, real-time acquisition of process parameters of construction machinery during the construction process is needed.

[0049] The collected construction parameters include: riprapping particle size, throwing height (continuous value), pouring rate (continuous value). Among them, the riprapping particle size is classified as 0, 1, 2 for classification, and the throwing height and pouring rate are continuous values.

[0050] In the specific collection process, the above key parameters of the riprapping machine are read and recorded in real time through the PLC controller of the construction machinery through the Modbus TCP protocol.

[0051] After the collection of the construction parameters is completed, at least one construction parameter feature vector is formed in a specified time interval. Preferably, the construction parameter feature vector is the same as the specified time interval of the first flow field feature vector, so as to facilitate subsequent data alignment.

[0052] Among them, in each of the construction parameter feature vectors, a feature vector composed of riprapping particle size, throwing height and pouring rate is included, and the specific form of the three types of data is as described above.

[0053] S22: According to the first digital elevation model obtained after each construction, at least one first construction effect information is determined.

[0054] In the S21, the real-time construction parameters have been collected, and in this step, full-coverage topographic survey is needed to be carried out in the construction area after the construction is completed according to the real-time construction parameters. After comparison with the design blueprint, the construction effect is quantitatively evaluated, so as to prepare for the subsequent model training.

[0055] The S22 includes the following sub-steps: S221: After each construction is completed, full-coverage topographic survey is carried out on the construction area, so as to obtain a first digital elevation model.

[0056] After each construction section is completed, Norbit iWBMS series multi-beam sounding system is used to carry out full-coverage topographic survey on the construction area with 200% overlap rate, and a digital elevation model with a resolution of 5cm×5cm is generated.

[0057] S222: According to the first digital elevation model and the original digital elevation model, first cemented body information is obtained.

[0058] In this step, the DEM after construction and the DEM before construction are subjected to difference operation (DoD). The elevation change threshold (such as +0.1m) is set, the cemented body area is automatically identified, the total volume, average thickness and coverage area are calculated, and the above information is taken as the first cemented body information.

[0059] S223: According to the first cemented body information and the first design blueprint information, first construction effect information is obtained.

[0060] In the first design blueprint information, the construction effect parameters expected to be achieved after the current construction are recorded, which can specifically include total volume, average thickness, coverage area, etc.

[0061] In this step, the first cemented body information is compared with the first design blueprint information in terms of similarity, so as to obtain the calculation coverage rate and volume compliance rate, thereby serving as the first construction effect information of the construction effect.

[0062] The first construction effect information can be obtained by performing a specified operation on the various compliance rates.

[0063] Preferably, the average operation or weighted summation operation mode can be selected.

[0064] Preferably, the various compliance rates can not be weighted and averaged, but the compliance rates of various parameters can be combined into a feature vector to serve as the first construction effect information.

[0065] Each of the first construction effect information corresponds to at least one of the specified time intervals.

[0066] S23: Combining at least one of the first flow field feature vectors, at least one of the construction parameter feature vectors, and at least one of the first construction effect information into a space-time correlation database.

[0067] In order to obtain complete sample data for subsequent model training, at least one of the first flow field feature vectors, at least one of the construction parameter feature vectors, and the first construction effect information are combined into a space-time correlation database in this step.

[0068] In the space-time correlation database, at least one data record is included, and each of the data records includes a one-to-one or one-to-many relationship between each of the first construction effect information and the first flow field feature vector and the construction parameter feature vector.

[0069] MySQL database is used to design two core data tables: Specifically, by using high-precision time stamp and spatial interpolation algorithm, the environmental data, construction operation and construction effect data in the same space-time range are associated to form a complete sample record. For example, the spatial position corresponding to the construction operation can be calculated according to the GPS coordinates of the construction ship and the single pile coordinates, so as to establish an association relationship between the environmental data, construction data and construction effect data in the specified time period in the corresponding spatial position.

[0070] S3: Predicting target construction parameter information according to target flow field feature vector and target construction effect information.

[0071] Since the environmental parameters in the same geographical area are relatively stable, in this step, the machine learning model for predicting the optimal construction parameters can be trained using the information stored in the spatio-temporal correlation database.

[0072] The S3 specifically includes the following sub-steps: S31: training a construction parameter prediction model according to the spatio-temporal correlation database.

[0073] In this step, each group of records in the spatio-temporal correlation database is taken as a training sample, so as to train a construction parameter prediction model.

[0074] In the training process, the environmental parameters and construction effect information are taken as inputs, and the construction parameter information is taken as output, so as to train the construction parameter prediction model. The input features of the model are at least one first flow field feature vector corresponding to the one-time construction time and the corresponding first construction effect information extracted in S13. The output of the model is three construction parameters to be optimized corresponding to the one-time construction time, i.e., the riprap particle size, the throwing height, and the pouring rate.

[0075] Preferably, the XGBoost regressor can be used as the construction parameter prediction model, so as to construct a multi-output regression model.

[0076] In the specific training process of the model, the grid search (GridSearchCV) can be used to cross-validate and optimize the model hyperparameters (such as max_depth, learning_rate, n_estimators), and the mean square error (MSE) and the mean absolute error (MAE) are used as the loss function and evaluation index. The specific optimization process can refer to the optimization method in the prior art, which will not be described here.

[0077] Since the time sequence information of the environmental parameters, construction parameters, and construction effect information is used in the training process of the construction parameter prediction model, the optimal construction parameter information in the future preset time interval can be predicted by using the trained construction parameter prediction model.

[0078] S32: inputting the target flow field feature vector and the target construction effect information into the construction parameter prediction model to obtain the target construction parameter information.

[0079] In this step, the target flow field feature vector in the preset time interval and the target construction effect information to be achieved are input into the construction parameter prediction model to obtain the target construction parameter information.

[0080] Since the target process feature vector for characterizing the environmental parameters usually does not change dramatically in a short time, the target construction parameter information in the future preset time interval can be predicted through this step.

[0081] S33: When the construction record accumulation amount exceeds the first preset value, incrementally learn the construction parameter prediction model.

[0082] Since the construction process is ongoing, the system automatically triggers the increment learning process every time a preset number of valid construction records are accumulated.

[0083] Preferably, the first preset value is 50.

[0084] In the specific implementation process, the partial_fit method of scikit-learn or the increment training interface of XGBoost can be used to update the existing model online with new data, while important historical samples are retained to prevent catastrophic forgetting. After the updated model is verified by the test set, the old model is automatically replaced to realize the continuous evolution of decision-making ability.

[0085] S4: After confirming the target construction parameter information, perform the issuing and recording process.

[0086] After obtaining the target construction parameter information through S3, the intelligent decision of the model can be converted into actual construction instructions in this step, and through the man-machine interaction interface or the industrial control system, the real-time and accurate execution of the construction parameters can be realized.

[0087] The S4 can specifically include the following sub-steps: S41: Display the target construction parameter information on the digital dashboard, and perform the issuing operation after the confirmation operation.

[0088] Develop a real-time data dashboard based on Web, preferably using Grafana or Vue.js + ECharts framework. The dashboard dynamically displays the optimal parameter setting value recommended by the model in the form of numbers and dashboards, and displays the current actual value side by side, and the difference part is highlighted in color.

[0089] Integrate a one-click issuing button on the dashboard. After the operator confirms the recommended parameters, clicking the button can issue the setting value to the PLC.

[0090] S42: Record the target construction parameter information and parameter adjustment information in the operation log.

[0091] In this step, any automatic or manual parameter adjustment will be recorded in the operation log table.

[0092] Exemplary fields can include timestamp, parameter_name, recommended_value, actual_set_value, operator_id.

[0093] The log is not only used for audit tracking, but its actual_set_value will be used as the real executed parameter, fed back to the database of S2, for subsequent model training and effect evaluation, to ensure the closure of the data chain.

[0094] The application also proposes a dynamic optimization system for cemented riprap construction parameters in a wave flow environment, which is used to execute the dynamic optimization method for cemented riprap construction parameters in a wave flow environment.

[0095] The dynamic optimization method and system for cemented riprap construction parameters in a wave flow environment proposed by the application relate to the technical field of construction parameter determination. First, time series information of various environmental parameters in a specified construction area is extracted from multiple dimensions, and various environmental parameters are subjected to spatiotemporal alignment processing. Second, construction parameter information is recorded in real time, and is associated with environmental parameter information and construction effect information according to construction rounds. Finally, a time series prediction model for representing construction conditions is obtained based on the above data, so that the best construction parameter information can be predicted based on environmental parameters and expected construction effect information in a future preset time interval. The technical solution of the application can accurately predict cemented riprap construction parameters in a wave flow environment.

[0096] The above description is only the preferred embodiment of the application, and any equivalent changes or modifications made according to the structure, features and principles described in the patent application scope of the application are included in the patent application scope of the application.

Claims

1. A method for dynamically optimizing construction parameters of cemented riprap under wave-current environment, characterized in that, The method comprises: S1: collecting at least one multi-source data sequence, performing data optimization extraction processing on at least one multi-source data sequence to obtain at least one first flow field feature vector; S2: obtaining at least one construction parameter feature vector and at least one first construction effect information in the construction process, and determining a space-time correlation database according to the correlation between at least one construction parameter feature vector, at least one first construction effect information and at least one first flow field feature vector; S3: obtaining target construction parameter information according to target flow field feature vector and target construction effect information; S4: after confirming the target construction parameter information, performing issuing and recording processing.

2. The method according to claim 1, wherein, The S1 comprises: S11: at least one first multi-source sensor is arranged in the first ship and the first construction range, so as to obtain flow velocity data sequence, wave data sequence and ship motion data sequence; S12: performing space-time alignment fusion processing on the flow velocity data sequence, the wave data sequence and the ship motion data sequence to obtain target flow velocity data sequence, target wave data sequence and target ship motion data sequence; S13: performing data quality optimization processing and feature extraction processing on the target flow velocity data sequence, the target wave data sequence and the target ship motion data sequence to obtain at least one first flow field feature vector.

3. The method according to claim 2, wherein, In each first flow field feature vector, a combined vector of flow velocity data, wave data and ship motion data in a specified time interval is included.

4. The method according to claim 2, wherein, The S2 comprises: S21: real-time data collection of construction parameters is performed to obtain at least one construction parameter feature vector; S22: at least one first construction effect information is determined according to the first digital elevation model obtained after each construction; S23: a space-time correlation database is combined according to at least one first flow field feature vector, at least one construction parameter feature vector and at least one first construction effect information.

5. The method according to claim 4, wherein, The S22 comprises: S221: after each construction is completed, full-coverage topographic measurement of the construction area is performed to obtain a first digital elevation model; S222: first cemented body information is obtained according to the first digital elevation model and an original digital elevation model; S223: first construction effect information is obtained according to the first cemented body information and first design blueprint information.

6. The method according to claim 5, wherein, At least one data record is included in the space-time correlation database.

7. The method according to claim 1, wherein, The S3 comprises: S31: a construction parameter prediction model is trained according to the space-time correlation database; S32: target flow field feature vector and target construction effect information are input into the construction parameter prediction model to obtain target construction parameter information; S33: when the construction record accumulation amount exceeds a first preset value, the construction parameter prediction model is incrementally learned.

8. The method according to claim 7, wherein, In the training process of the construction parameter prediction model, the flow field feature vector and the construction effect information in the sample data are used as input data, and the construction parameter feature vector is used as output data.

9. The method according to claim 1, wherein, The S4 comprises: S41: Display the target construction parameter information on the digital billboard, and perform the issuing operation after the confirmation operation; S42: Record the target construction parameter information and the parameter adjustment information into the operation log.

10. A system for dynamically optimizing the construction parameters of cemented riprap under wave-current environment, which is used to implement the method for dynamically optimizing the construction parameters of cemented riprap under wave-current environment according to any one of claims 1-9.

Images