Safety early warning method and system for floating type wind turbine

Abstract

The invention relates to the technical field of safety monitoring of floating structures, and provides a safety early warning method and system for a floating wind turbine, and the method comprises the steps: obtaining multi-modal data at historical moments, and predicting the response data of the wind turbine at the current moment through a deep neural network model; obtaining deterministic environment parameters, and calculating wind turbine response data at the current moment through the wind turbine numerical model; calculating the difference between the prediction result and the monitoring result and the difference between the calculation result and the monitoring result, and taking the weighted sum of the two differences as a weighted error; respectively carrying out statistical analysis on the prediction result, the calculation result and the monitoring result; recording the difference value between the statistical value of the calculation result and the statistical value of the monitoring result as a first difference value; recording the difference value between the statistical value of the prediction result and the statistical value of the monitoring result as a second difference value; and carrying out grading early warning on the wind turbine. The abnormal information of the wind turbine system can be comprehensively identified, and the state of the wind turbine can be timely mastered and intelligently evaluated.

Description

Technical Field

[0001] The present invention belongs to the technical field of floating structure safety monitoring, and in particular relates to a floating wind turbine safety early warning method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] As the global demand for renewable resources increases, deep-sea floating wind turbines, as an emerging wind energy technology, are gaining more and more attention. Deep-sea wind energy technology expands the development boundaries of wind energy resources, helps reduce dependence on land and nearshore resources, and further promotes the optimization of the global energy structure. Floating wind turbines are deployed in deep water areas far from the coast, where the environment is complex and changeable. It is particularly important to develop and improve the safety monitoring and early warning system for floating wind turbines.

[0004] Current floating wind turbine safety monitoring technology mainly focuses on the measurement of basic parameters such as environmental conditions and structural vibrations. Although it can provide some key data, it is often unable to capture all potential risks faced by floating wind turbines in real time and comprehensively, and it is even more unable to realize real-time early warning functions based on monitoring data. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a floating wind turbine safety early warning method and system, which integrates environmental data such as wind speed, ocean current, and waves, and relies on numerical analysis models for analysis, integrates multimodal data such as slight changes in the marine environment, swaying and motion offset of the floating platform, and relies on deep learning algorithms for analysis to predict the performance and risks of the equipment under various conditions, and performs weighted analysis on the differences between simulation data, machine learning prediction data and actual monitoring data to comprehensively identify abnormal information of the wind turbine system, realize real-time early warning function, and realize timely grasp and intelligent evaluation of the status of the wind turbine.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A first aspect of the present invention provides a floating wind turbine safety early warning method, comprising:

[0008] Obtain multimodal data at historical moments, and use a deep neural network model to predict the wind turbine response data at the current moment as the prediction result;

[0009] Obtaining deterministic environmental parameters, and calculating the wind turbine response data at the current moment through the wind turbine numerical model as a calculation result;

[0010] The wind turbine response data at the current moment monitored by the sensor is obtained as the monitoring result; the difference between the predicted result and the monitored result and the difference between the calculated result and the monitored result are calculated, and the weighted sum of the two differences is used as the weighted error; the predicted result, the calculated result and the monitored result are statistically analyzed respectively; the difference between the statistical value of the calculated result and the statistical value of the monitored result is recorded as the first difference; the difference between the statistical value of the predicted result and the statistical value of the monitored result is recorded as the second difference; based on the weighted error, the first difference and the second difference, a graded warning of the wind turbine is performed.

[0011] Furthermore, the multimodal data is collected by wind speed and direction sensors, wave monitoring sensors, current monitoring sensors, temperature sensors, strain gauges, tension gauges, inclination sensors, vibration sensors and / or motion sensors.

[0012] Furthermore, the wind turbine response data includes motion data on six degrees of freedom of the wind turbine, nacelle acceleration data and mooring tension data.

[0013] Furthermore, the motion data of the wind turbine in six degrees of freedom include sway, surge, heave, roll, pitch and bow pitch.

[0014] Furthermore, the statistical analysis includes calculation of maximum value, minimum value, numerical range, mean value and standard deviation.

[0015] Furthermore, the deterministic environmental parameters include a wind time series and a wave time series.

[0016] Furthermore, the multimodal data at historical moments are input into a deep neural network model after data anomaly elimination and feature extraction.

[0017] A second aspect of the present invention provides a floating wind turbine safety warning system, comprising:

[0018] A prediction module is configured to: obtain multimodal data at historical moments, and predict the wind turbine response data at the current moment as a prediction result through a deep neural network model;

[0019] The numerical calculation module is configured to: obtain deterministic environmental parameters, and calculate the wind turbine response data at the current moment as a calculation result through the wind turbine numerical model;

[0020] The early warning module is configured to: obtain the wind turbine response data monitored by the sensor at the current moment as the monitoring result; calculate the difference between the prediction result and the monitoring result and the difference between the calculation result and the monitoring result, and use the weighted sum of the two differences as the weighted error; perform statistical analysis on the prediction result, the calculation result and the monitoring result respectively; record the difference between the statistical value of the calculation result and the statistical value of the monitoring result as the first difference; record the difference between the statistical value of the prediction result and the statistical value of the monitoring result as the second difference; and perform graded early warning of the wind turbine based on the weighted error, the first difference and the second difference.

[0021] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-mentioned floating wind turbine safety warning method.

[0022] A fourth aspect of the present invention provides a computer device, comprising a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, wherein when the processor executes the program, the steps in the floating wind turbine safety warning method as described above are implemented.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] The present invention integrates environmental data such as wind speed, ocean current, and waves, and relies on numerical analysis models for analysis. It integrates multimodal data such as slight changes in the marine environment, swaying and motion deviation of floating platforms, and relies on deep learning algorithms for analysis to predict the performance and risks of equipment under various conditions. It also performs weighted analysis on the differences between simulation data, machine learning prediction data and actual monitoring data to comprehensively identify abnormal information of the wind turbine system, realize real-time early warning function, and achieve timely grasp and intelligent evaluation of the status of wind turbines.

[0025] The multimodal data used in the present invention is collected through wind speed and direction sensors, wave monitoring sensors, current monitoring sensors, temperature sensors, strain gauges, tension gauges, inclination sensors, vibration sensors and / or motion sensors. The acquired monitoring data such as slight changes in the marine environment, swaying and motion deviation of the floating platform contain deeper information, which can reveal the actual performance and potential risks of floating wind turbines in complex marine environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0027] Figure 1 is a flow chart of a floating wind turbine safety early warning method according to a first embodiment of the present invention;

[0028] Figure 2 This is a flow chart of numerical calculation of wind turbine response under determined parameters according to the first embodiment of the present invention;

[0029] Figure 3 This is a data processing flow chart of a wind turbine response neural network prediction model based on deep learning according to Embodiment 1 of the present invention;

[0030] Figure 4 This is a schematic diagram of dividing a data set of a wind turbine response neural network prediction model based on deep learning according to the first embodiment of the present invention;

[0031] Figure 5 This is a structural diagram of the LSTM neural network prediction model of the first embodiment of the present invention;

[0032] Figure 6 This is a schematic diagram of an early warning method based on numerical calculation results, deep learning prediction results and monitoring data differences according to Embodiment 1 of the present invention;

[0033] Figure 7 It is a structural diagram of a computer device according to a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0034] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0035] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0036] Embodiment 1

[0037] This embodiment provides a floating wind turbine safety early warning method.

[0038] Monitoring data such as subtle changes in the marine environment, the swaying and movement deviation of the floating platform contain deeper information that can reveal the actual performance and potential risks of floating wind turbines in complex marine environments.

[0039] The floating wind turbine safety warning method provided in this embodiment should not only have the ability to monitor environmental factors in real time, but also be able to accurately evaluate and warn the structural health status of the equipment, and realize the real-time warning function based on online monitoring data; at the same time, it can integrate environmental data such as wind speed, current, and waves, and rely on deep learning algorithms and numerical analysis models for analysis to predict the performance and risks of the equipment under various environmental conditions; in addition, the real-time transmission and processing of data should also be taken into account to ensure the real-time transmission and accurate analysis of data, and by introducing advanced sensing technology, data analysis algorithms and intelligent warning mechanisms, the safety and operation efficiency of floating wind turbines can be ensured.

[0040] A floating wind turbine safety warning method provided in this embodiment monitors the collected monitoring data in real time based on sensors installed on the wind turbine, and compares it with the numerical analysis results of the model to preliminarily identify abnormal information of the wind turbine system. Then, the collected monitoring data is immediately analyzed and predicted through a machine learning algorithm. Based on the difference between the numerical analysis results and the deep learning prediction results and the actual monitoring information, a warning weight assessment is performed to realize a real-time warning function.

[0041] This embodiment provides a floating wind turbine safety warning method, such as Figure 1 As shown, including:

[0042] Step 1: Multimodal data collection and real-time monitoring.

[0043] The purpose of the multimodal data acquisition center is to collect the operating data of offshore floating wind turbines from different dimensions (i.e. different sensor types) and monitor their status in real time.

[0044] In this embodiment, the data source includes:

[0045] (1) Environmental sensors: including wind speed and direction sensors, wave monitoring sensors, ocean current monitoring sensors, and temperature sensors, which are used to measure wind, wave, current, and temperature data. These data help evaluate the operation of wind turbines under current meteorological conditions.

[0046] (2) Structural sensors: including strain gauges, tension gauges, inclination sensors, vibration sensors and motion sensors, which monitor the structural health of wind turbines. The vibration, strain and other data detected by these sensors can reveal potential mechanical failures or fatigue damage.

[0047] In this embodiment, the transmission of multimodal monitoring data applies many communication technologies, including wired communication and wireless communication technologies, to ensure real-time transmission and update of data. Specifically, the monitoring data transmission includes:

[0048] (1) Transmit sensor data to the central processing unit in real time for real-time data processing.

[0049] (2) After the sensor data signal is transmitted to the monitoring data center, the monitored data is visualized. The visualization can help the operator to instantly understand the operating status of the wind turbine, report abnormal signals in a timely manner, respond before a fault occurs, and reduce operation and maintenance costs.

[0050] (3) Transmit the data signal to the cloud for storage and backup to a backup database.

[0051] Step 2: determine the numerical calculation model of wind turbine response under the parameters.

[0052] A floating wind turbine is a complex rigid-flexible coupling system, which is subject to the combined effects of multiple environmental loads during actual operation, including aerodynamic loads on blades, elastic effects of blades and towers, wave and flow loads on floating foundations, mooring loads, etc. Numerical simulation of floating wind turbines is a common method for predicting wind turbine loads and motions. Through aerodynamic-hydrodynamic-servo-elastic coupling analysis theory or computational fluid dynamics methods, the response of a given wind turbine under given environmental parameters, such as motion, stress, and deformation, can be predicted more accurately, thereby identifying potential design defects or risks.

[0053] By building a mathematical model through numerical simulation to simulate the structural response and performance of wind turbines under given environmental conditions, and statistically analyzing these data with multimodal monitoring data, the abnormal operating behavior of wind turbines can be preliminarily identified as part of the weight of early warning assessment.

[0054] The numerical calculation model of wind turbine response under certain parameters is used to predict the behavior of the wind turbine under certain conditions. The given parameters and coupled analysis are explained as follows:

[0055] (1) The established parameters refer to the statistical information and spectral characteristics of the sea conditions based on the actual wind, wave and current conditions;

[0056] (2) The coupled analysis considers the time domain response of the floating wind turbine under the combined action of wind and waves;

[0057] (3) When wind and waves act together, the input parameters required for numerical analysis are given based on the sea state level and the corresponding relationship between wind and waves based on the spectral characteristics.

[0058] Table 1. Sea state classification table

[0059]

[0060]

[0061] Wind spectrum: In the design and analysis of offshore floating structures, the wind spectrum is usually used to describe the impact of wind on offshore floating structures from an energy perspective. Currently, the two most widely used wind spectra in the engineering community are the API wind spectrum and the NPD wind spectrum. In the latest API RP 2A specification, the NPD wind spectrum has replaced the API wind spectrum as the recommended wind spectrum.

[0062] The NPD wind spectrum describes the fluctuation of the longitudinal wind speed energy density at a certain point on the sea level, and the expression is:

[0063]

[0064] Where U0 represents the wind speed at a certain reference height, usually 10 meters, that is, U0 is usually U 10 ; S NPD (f) is the energy spectral density of frequency f, in m 2 / s; f is frequency, the unit is Hz.

[0065]

[0066] Wave spectrum: including PM spectrum and JONSWAP. Taking JONSWAP as an example, it is composed of H S , T P , γ are determined by three parameters, H S is the significant wave height, T P is the spectrum peak period, the average value of γ is 3.3, and γ is the spectrum peak factor, which describes the concentration of wave energy. The spectral density function is:

[0067]

[0068] Among them, α and g are constant factors, which are related to the amplitude and intensity of the fluctuation and are often obtained through experiments or theoretical derivation; ω is the angular frequency, ω p is the characteristic frequency; σ is the standard deviation.

[0069] Based on the sea condition classification table in Table 1, considering the correlation between sea conditions and wind speed, using the irregular wave JONSWAP spectrum and the IEC wind spectrum used in the wind turbine industry, the working conditions required for numerical analysis can be divided into 8 types, as shown in Table 2.

[0070] Table 2. Numerical analysis working condition classification table

[0071] Working conditions Average wind speed (m / s) <![CDATA[Significant wave height H S (m)]]> <![CDATA[Spectral peak period T P (s)]]> A 0.9 0.05 5.0 B 2.5 0.3 7.5 C 4.4 0.88 7.5 D 6.7 1.84 8.3 E 9.4 3.61 8.3 F 12.3 5.0 8.3 G 15.5 7.5 15.0 H 19 11.5 16.4

[0072] The basic process of numerical calculation is as follows Figure 2 As shown, specifically including:

[0073] Step 201: Build a numerical model of a wind turbine, verify it through a water tank experiment, and perform model correction.

[0074] In this embodiment, building a wind turbine numerical model refers to inputting the target wind turbine structural parameters (size, mass, center of gravity position, moment of inertia, etc.) into the software.

[0075] Among them, the water tank experiment refers to making a scaled physical model based on the target wind turbine model and conducting the model experiment in a laboratory water tank.

[0076] Among them, model correction refers to evaluating numerical analysis results based on experimental results. If the experimental results are significantly different from the numerical analysis results, the problem is found and the numerical analysis results or model experimental results are corrected until the results of the two are consistent within the allowable error range.

[0077] Step 202: Input different operating condition parameters (deterministic environmental parameters) into the wind turbine numerical model to obtain numerical calculation results and generate a wind turbine response data set (denoted as W).

[0078] In this embodiment, the deterministic environmental parameters include a wind time series and a wave time series.

[0079] The generation process of wind time series is as follows: select a certain working condition from "Table 2, numerical analysis working condition classification table", such as working condition A, and determine the average wind speed U0 = 0.9 under this working condition, and substitute it into formula (1) and formula (2) to generate wind speed frequency domain data S NDF (f) Then the inverse Fourier transform is used to obtain the time series of wind speed.

[0080] The generation process of the wave time series is as follows: select a certain working condition from "Table 2, numerical analysis working condition classification table", such as working condition A, then determine the significant wave height Hs = 0.05m and the spectrum peak period Tp = 5s under this working condition, substitute them into formula (3) to generate the wave frequency domain data S(w), and then use the inverse Fourier transform to obtain the wave time series.

[0081] In this embodiment, the generated wind turbine response data set includes: motion data of the wind turbine in six degrees of freedom (called sway, surge, heave, roll, pitch, and bow) at several time points within the current moment, cabin acceleration data, and mooring tension data.

[0082] Step 203: Perform statistical analysis on the wind turbine response data set to determine the baseline value of the wind turbine response under the design condition. Here, the statistical analysis is explained as follows:

[0083] Statistical analysis refers to determining the maximum value W in a data set. max , minimum value W min , value range W range , average value W mean , standard deviation W std ,have:

[0084] W range =W max -W min (4)

[0085]

[0086]

[0087] Among them, N represents the number of data in the data set, x i Represents each data point in the dataset.

[0088] Step 3: Wind turbine response neural network prediction model based on deep learning.

[0089] The deep learning-based wind turbine response prediction neural network model uses machine learning algorithms to analyze and predict wind turbine responses based on monitoring data.

[0090] The implementation steps of a deep learning-based wind turbine response neural network prediction model usually include:

[0091] (1) Data preprocessing: The original multimodal data collected from environmental sensors and structural sensors at historical moments are preprocessed into a format suitable for training neural networks. This includes steps such as data anomaly removal, data set segmentation, and feature extraction. The implementation process is as follows: Figure 3 As shown, the data set is split as Figure 4 shown.

[0092] (2) Selection of deep neural network models: Deep neural network models such as convolutional neural network (CNN) and recurrent neural network (RNN) are used to process time series data. Taking the long short-term memory neural network (LSTM) in the recurrent neural network as an example, the specific process of the LSTM model predicting monitoring data is as follows: Figure 5 shown.

[0093] Several important parameter settings in the training process of the wind turbine response neural network prediction model based on deep learning are explained as follows:

[0094] Input data type: refers to the type of data to be predicted by the neural network, the scale and dimension of the input feature data;

[0095] Output data: wind turbine response data at the current moment, including: motion data of the wind turbine in six degrees of freedom (called sway, surge, heave, roll, pitch, and bow), cabin acceleration data, and mooring tension data;

[0096] Hidden layer dimension: that is, the number of hidden layer nodes. This parameter affects the number of variables to establish weights. The dimension is increased or decreased for different data types or prediction effects.

[0097] The number of neural network layers: This is equivalent to the number of neural network cycles. Each additional layer will recalculate the result of the previous neural network processing. This parameter setting needs to be tested during the training process.

[0098] Normalization processing: It is used to solve the errors caused by the difference in data magnitude, and it can also speed up the convergence speed and improve the operation efficiency. For example, the expression of the maximum and minimum normalization processing method is:

[0099]

[0100] Among them, X* is the normalized data, X is the original data, and X min is the minimum value in the data set, X max is the maximum value in the data set;

[0101] Denormalization: After performing neural network operations, all data is denormalized to restore the data.

[0102] The trained neural network model is evaluated using the following four indicators:

[0103] Root mean square error (RMSE),

[0104] Mean absolute error (MAE),

[0105] Root mean square error (MSE),

[0106] Coefficient of determination (R 2 ),

[0107] in, is the predicted value of each monitoring data, y i is the actual value of each monitoring data, is the average value of the actual value of each monitoring data, n is the size of the test set, s i is the error between the actual value and the predicted value, and Nr is the degrees of freedom of the prediction.

[0108] (3) Training and verification: Use historical data to train and verify the deep learning-based neural network prediction model for wind turbine response. Iteratively optimize model parameters to improve prediction accuracy. When the prediction error of the training model is less than the operating error, the model training is considered complete and can be applied to the monitoring and early warning system. After the monitoring and early warning system is put into operation, the real-time monitoring data is input into the trained model to predict the short-term response of the wind turbine as part of the early warning indicator evaluation.

[0109] Step 4: Early warning method based on the difference between numerical calculation results, deep learning prediction results and monitoring data. The main steps include:

[0110] Step 401, data comparison: compare the real-time monitoring data (actual response data of the wind turbine at the current moment) with the expected results of the numerical calculation model and the predicted results of the deep learning model.

[0111] The difference between the numerical calculation result and the actual response data of the wind turbine at the current moment is recorded as R1, the difference between the deep learning prediction result and the actual response data of the wind turbine at the current moment is recorded as R2, the weight of the numerical analysis error is recorded as K1, the weight of the deep learning prediction error is recorded as K2, and the weighted error R is recorded as:

[0112] R=K1×R1+K2×R2

[0113] Step 402, error evaluation: evaluate the weighted error and the set benchmark value, analyze the deviation between the predicted result and the actual result, identify anomalies, and make graded warnings. The numerical simulation statistical value uses each working condition as a subscript to indicate the current working condition type. For example, the statistical value of a parameter in working condition A is recorded as S A , the machine learning prediction value is recorded as S neural , the current monitoring value is recorded as S real , the difference between the numerical analysis statistical value and the monitoring value is recorded as the first difference δ1, and the difference between the machine learning prediction value and the monitoring value is recorded as the second difference δ2, and we have:

[0114] δ1=|S A -S real |

[0115] δ2=|S neural -S real |

[0116] For example, for motion response data, the average sway data of the current moment (within 1 hour) of numerical analysis may be 10m, and the maximum value is 20m, the average sway data of the neural network prediction value within 1 hour may be 11m, and the maximum value is 21m, and the average sway data of the monitoring value within 1 hour may be 12m, and the maximum value is 22m. The difference can be calculated.

[0117] It should be noted that the predicted value is obtained using numerical methods and deep learning methods when there is no abnormality in the wind turbine system. However, during the actual monitoring process, the wind turbine may have abnormalities, such as damage to mechanical equipment, sensor failure, etc. This is why an early warning is required when there is a large difference between the predicted value and the actual value. The early warning when there is a large difference between the predicted value and the actual value is to allow managers to detect potential abnormalities or failures in a timely manner. A large difference indicates that the wind turbine system may have equipment damage, sensor failure, significant changes in environmental conditions, etc.

[0118] Step 403: Graded warning: The yellow warning threshold is denoted as δ yellow , the orange warning threshold is denoted as δ orange , the red warning threshold is denoted as δ red The three thresholds increase from small to large, with the yellow threshold being the smallest, the orange threshold being in the middle, and the red threshold being the largest. For example, the yellow threshold is set to 1.1 times the maximum value, the orange threshold is set to 1.2 times the maximum value, and the red threshold is set to 1.3 times the maximum value. This setting is more in line with the human recognition of color and danger, with red being more dangerous, orange being in the middle, and yellow being lower.

[0119] Yellow warning is recorded as: δ1>δ yellow ;

[0120] Orange warning is recorded as: δ2>δ orange ;

[0121] Red warning is recorded as: δ3=R>δ red .

[0122] Figure 6 The warnings in "Preliminary Warning Signal" and "Warning Signal Issued" are three-level warnings. For example, if a preliminary warning signal is issued, yellow, is it determined whether the sea conditions have changed? If the sea conditions change, the warning weight is re-evaluated or the alarm information is manually lifted, and the cycle returns to normal. If the sea conditions do not change, a yellow warning signal is issued.

[0123] Table 3. Level 3 warning evaluation table

[0124] Warning level Early warning indicators Early warning indicators 1 Yellow Warning <![CDATA[δ1>d yellow ]]> 2 Orange warning <![CDATA[δ2>d orange ]]> 3 Red Alert <![CDATA[δ3>d red ]]>

[0125] The principle diagram of the early warning method based on the difference between numerical calculation results, deep learning prediction results and monitoring data is as follows Figure 6 shown.

[0126] The wind turbine online monitoring and early warning system of the present invention monitors the vibration, tilt, stress and other signal data of the unit in real time and analyzes them through specific information collection and alarm strategies; if an abnormality is found, an alarm will be automatically triggered, and the wind turbine monitoring and management personnel will be immediately notified via email, text message or App, etc., so that they can conduct a detailed assessment of the equipment status.

[0127] The core of this embodiment is to perform weighted analysis on the differences between simulation data, machine learning prediction data and actual monitoring data, based on which accurate equipment data can be obtained in real time. The collected data is processed by a high-precision intelligent monitoring module and combined with artificial diagnosis technology to achieve timely understanding and intelligent evaluation of the status of the wind turbine.

[0128] Embodiment 2

[0129] This embodiment provides a floating wind turbine safety warning system, which specifically includes:

[0130] A prediction module is configured to: obtain multimodal data at historical moments, and predict the wind turbine response data at the current moment as a prediction result through a deep neural network model;

[0131] The numerical calculation module is configured to: obtain deterministic environmental parameters, and calculate the wind turbine response data at the current moment as a calculation result through the wind turbine numerical model;

[0132] The early warning module is configured to: obtain the wind turbine response data monitored by the sensor at the current moment as the monitoring result; calculate the difference between the prediction result and the monitoring result and the difference between the calculation result and the monitoring result, and use the weighted sum of the two differences as the weighted error; perform statistical analysis on the prediction result, the calculation result and the monitoring result respectively; record the difference between the statistical value of the calculation result and the statistical value of the monitoring result as the first difference; record the difference between the statistical value of the prediction result and the statistical value of the monitoring result as the second difference; and perform graded early warning of the wind turbine based on the weighted error, the first difference and the second difference.

[0133] It should be noted here that each module in this embodiment corresponds to each step in Example 1 one by one, and the specific implementation process is the same, which will not be repeated here.

[0134] Embodiment 3

[0135] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the floating wind turbine safety warning method described in the first embodiment are implemented.

[0136] Embodiment 4

[0137] This embodiment provides a computer device, such as Figure 7 As shown, it includes a display device, an input device, a computer-readable storage medium (volatile memory and non-volatile storage medium), a processor, a communication interface (i.e., a network interface), and a computer program stored on the computer-readable storage medium and executable on the processor, wherein the processor, the communication interface, and the computer-readable storage medium can be connected via a bus or other means. The communication interface is used to receive and send data, and when the processor executes the program, the steps in the safety warning method for a floating wind turbine described in the first embodiment are implemented.

[0138] Among them, any reference to memory, storage, database or other media provided by the present application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0139] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0140] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0141] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A floating wind turbine safety early warning method, characterized in that: include: Obtain multimodal data at historical moments, and use a deep neural network model to predict the wind turbine response data at the current moment as the prediction result; Obtaining deterministic environmental parameters, and calculating the wind turbine response data at the current moment through the wind turbine numerical model as a calculation result; Obtain the wind turbine response data at the current moment monitored by the sensor as the monitoring result; Calculate the difference between the prediction result and the monitoring result and the difference between the calculation result and the monitoring result, and take the weighted sum of the two differences as the weighted error; perform statistical analysis on the prediction result, the calculation result and the monitoring result respectively; record the difference between the statistical value of the calculation result and the statistical value of the monitoring result as the first difference; The difference between the statistical value of the prediction result and the statistical value of the monitoring result is recorded as the second difference; Based on the weighted error, the first difference and the second difference, a graded warning of the wind turbine is performed.

2. A floating wind turbine safety early warning method according to claim 1, characterized in that: The multimodal data is collected by wind speed and direction sensors, wave monitoring sensors, ocean current monitoring sensors, temperature sensors, strain gauges, tension gauges, inclination sensors, vibration sensors and / or motion sensors.

3. A floating wind turbine safety early warning method according to claim 1, characterized in that: The wind turbine response data includes motion data on six degrees of freedom of the wind turbine, nacelle acceleration data and mooring tension data.

4. A floating wind turbine safety early warning method as claimed in claim 3, characterized in that: The motion data of the wind turbine on six degrees of freedom include sway, surge, heave, roll, pitch and bow.

5. A floating wind turbine safety early warning method as claimed in claim 1, characterized in that: The statistical analysis includes the calculation of maximum value, minimum value, range, mean value and standard deviation.

6. A floating wind turbine safety early warning method according to claim 1, characterized in that: The deterministic environmental parameters include a wind time series and a wave time series.

7. A floating wind turbine safety early warning method according to claim 1, characterized in that: The multimodal data at the historical moments are input into the deep neural network model after data anomaly elimination and feature extraction.

8. A floating wind turbine safety warning system, characterized in that: include: A prediction module is configured to: obtain multimodal data at historical moments, and predict the wind turbine response data at the current moment as a prediction result through a deep neural network model; The numerical calculation module is configured to: obtain deterministic environmental parameters, and calculate the wind turbine response data at the current moment as a calculation result through the wind turbine numerical model; The early warning module is configured to: obtain the wind turbine response data at the current moment monitored by the sensor as a monitoring result; Calculate the difference between the prediction result and the monitoring result and the difference between the calculation result and the monitoring result, and take the weighted sum of the two differences as the weighted error; perform statistical analysis on the prediction result, the calculation result and the monitoring result respectively; record the difference between the statistical value of the calculation result and the statistical value of the monitoring result as the first difference; The difference between the statistical value of the prediction result and the statistical value of the monitoring result is recorded as the second difference; Based on the weighted error, the first difference and the second difference, a graded warning of the wind turbine is performed.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in a floating wind turbine safety early warning method as described in any one of claims 1 to 7 are implemented.

10. A computer device comprising a computer-readable storage medium, a processor, and a computer program stored in the computer-readable storage medium and executable on the processor, characterized in that: When the processor executes the program, the steps in the floating wind turbine safety early warning method according to any one of claims 1-7 are implemented.

Images