Buoy meteorological data calibration compensation method and system
Through multi-sensor collaborative calibration technology, the problem of sensor data deviation of buoys in complex ocean environments was solved, real-time adaptive calibration of buoy meteorological data was achieved, and the accuracy and stability of the data were improved.
Patent Information
- Application Number
- CN202510919413.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-04
- Publication Date
- 2025-10-03
AI Technical Summary
In existing technologies, buoy sensors are susceptible to mechanical vibration, electromagnetic interference, and biological attachment in complex marine environments, resulting in data drift or distortion. In addition, there is a lack of real-time adaptive calibration methods, which affects data reliability.
Multi-sensor collaborative calibration technology is used to collect environmental interference data through acceleration, electromagnetic field and ultrasonic sensors, perform time alignment and expansion processing, train dynamic interference models, and perform adaptive compensation to generate a calibration data set.
The accuracy and stability of buoy meteorological data have been significantly improved, making it suitable for long-term ocean observations.
Smart Images
Figure CN120742455A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of ocean monitoring technology, and in particular to a buoy meteorological data calibration and compensation method and system. Background Art
[0002] As the core equipment for marine meteorological observation, buoys are usually equipped with modules such as barometers, temperature and humidity sensors, and wind speed and direction meters for real-time monitoring of sea surface meteorological parameters. However, buoys are exposed to high salt fog, strong winds and waves, and drastic changes in temperature and humidity in the ocean environment for a long time. The sensors are easily affected by mechanical vibration, electromagnetic interference, biological attachment, etc., resulting in data drift or distortion. In the existing technology, calibration methods mostly rely on manual regular maintenance or static calibration, which cannot compensate for dynamic environmental interference in real time, resulting in reduced data reliability. In addition, traditional methods do not fully consider the correlation of multi-sensor collaborative calibration, making it difficult to achieve high-precision and adaptive compensation effects. Summary of the Invention
[0003] The object of the present invention is to provide a buoy meteorological data calibration and compensation method and system to solve the technical problem of low buoy data calibration accuracy in the prior art.
[0004] This application proposes a buoy meteorological data calibration and compensation method, which includes: S1: at least one first environmental interference sensor is set on a first buoy, at least one first environmental interference data set is collected by the at least one first environmental interference sensor, and first multi-source environmental interference data is obtained based on the at least one first environmental interference data set; S2: performing a first time alignment process and a first expansion process on the first multi-source environment interference data to obtain target multi-source environment interference data; S3: Based on the target multi-source environment interference data, train and obtain a first dynamic interference model; S4: Using the first dynamic interference model, adaptively compensate at least one first environmental data set to obtain at least one target compensated calibration data set.
[0005] Preferably, the S1 includes the following sub-steps: S11: installing four first acceleration sensors at four corners of the first buoy and performing initial settings on the first acceleration sensors; S12: Installing a first electromagnetic field sensor on top of the first buoy and performing initial settings on the first electromagnetic field sensor; S13: installing a first ultrasonic sensor on the surface of the first buoy and performing initial settings on the first ultrasonic sensor; S14: Acquire at least one first environmental interference data set through at least one of the first acceleration sensor, the first electromagnetic field sensor, and the first ultrasonic sensor, and perform a first preprocessing operation on the at least one first environmental interference data set to obtain first multi-source environmental interference data.
[0006] Preferably, the first pre-processing operation includes: Sampling frequency setting, data storage format adjustment and outlier filtering.
[0007] Preferably, the S1 includes the following sub-steps: S21: performing a first time alignment process on the first multi-source environmental interference data to obtain second multi-source environmental interference data; S22: performing a first feature extraction process on the second multi-source environmental interference data, and constructing a first multi-source environmental interference feature vector; S23: Perform a first expansion process on the first multi-source environment interference feature vector to obtain target multi-source environment interference data.
[0008] Preferably, the first expansion processing refers to performing further data expansion processing using historical data based on the first multi-source environmental interference feature vector obtained in real time.
[0009] Preferably, the specific steps of the first time alignment process include: timestamp adjustment and data interpolation.
[0010] Preferably, S3 includes the following sub-steps: S31: performing a second preprocessing operation on the target multi-source environment interference data to obtain first training sample data; S32: Based on the first training sample data, train and obtain the first dynamic interference model; S33: Performing model verification processing on the first dynamic interference model, and outputting the first dynamic interference model if the verification passes.
[0011] Preferably, the S4 includes the following sub-steps: S41: Obtaining a first coarse compensation coefficient based on the first dynamic interference model, and performing coarse compensation processing on at least one first environmental data set using the first coarse compensation coefficient to obtain at least one first compensated calibration data set; S42: Performing nonlinear fitting processing and residual correction processing on at least one of the first compensated calibration data sets to obtain a target compensated calibration data set.
[0012] The present application also proposes a buoy meteorological data calibration and compensation system for implementing the above-mentioned buoy meteorological data calibration and compensation method.
[0013] This application proposes a method and system for buoy meteorological data calibration and compensation, relating to the technical field of ocean monitoring technology. By using dynamic environmental interference modeling and multi-sensor collaborative calibration techniques, the system addresses the issue of sensor data deviation in complex ocean environments. The method includes environmental interference parameter acquisition, multi-source data fusion, dynamic interference model generation, adaptive compensation calculation, and compensation effect verification. This invention significantly improves the accuracy and stability of buoy meteorological data and is suitable for long-term ocean observation scenarios. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other implementation drawings based on the provided drawings without inventive effort.
[0015] Figure 1 The present invention is a flowchart of a method for calibrating and compensating buoy meteorological data.
[0016] Figure 2 This is a schematic diagram of the installation of the first acceleration sensor on the first buoy. DETAILED DESCRIPTION
[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0018] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The exemplary embodiments and descriptions are only used to explain the present invention but are not intended to limit the present invention.
[0019] The following is a detailed description of a buoy meteorological data calibration and compensation method and system of the present invention.
[0020] This embodiment proposes a buoy meteorological data calibration and compensation method. The specific process is as follows: Figure 1 shown.
[0021] S1: at least one first environmental interference sensor is set on a first buoy, at least one first environmental interference data set is collected according to the at least one first environmental interference sensor, and first multi-source environmental interference data is obtained based on the at least one first environmental interference data set.
[0022] In order to accurately detect data on the first buoy affected by mechanical vibration, electromagnetic interference, biological attachment, etc., in this step, it is necessary to install various sensors on the first buoy to be tested, and align the data obtained by various sensors in time and space dimensions to facilitate subsequent data analysis operations.
[0023] The first environmental interference sensor includes at least one of a first acceleration sensor, a first electromagnetic field sensor, and a first ultrasonic sensor.
[0024] The S1 includes the following sub-steps: S11: Install four first acceleration sensors at the four corners of the first buoy, and perform initial settings on the first acceleration sensors.
[0025] The vibration frequency of the buoy will directly affect the accuracy of its detection data. Therefore, in this step, the vibration frequency of the buoy is detected by an acceleration sensor.
[0026] Preferably, the first acceleration sensors are installed at the four corners of the first buoy to ensure that each sensor can independently detect vibrations in different directions. Figure 2 As shown, on the buoy A, four first acceleration sensors B are respectively arranged on the four corners of the buoy A.
[0027] After the first acceleration sensor is installed, it is necessary to perform an initial calibration on the first acceleration sensor, record a reference value in a static state, and set a vibration frequency threshold (eg, 5 Hz).
[0028] S12: Install a first electromagnetic field sensor on the top of the first buoy and perform initial settings on the first electromagnetic field sensor.
[0029] Typically, electromagnetic interference will also significantly affect the accuracy of the detection data of the first buoy. Therefore, in this step, it is necessary to install an electromagnetic field sensor on the first buoy and complete the detection of the electromagnetic sensing data.
[0030] Preferably, the first electromagnetic field sensor is installed on the top of the first buoy to ensure that it can cover the electromagnetic field changes within a 360-degree range around the buoy.
[0031] After the first electromagnetic field sensor is installed, it is necessary to perform an initial calibration on the first electromagnetic field sensor, record a reference value in a static state, and set an electromagnetic noise intensity threshold (eg, 50 μT).
[0032] S13: Install a first ultrasonic sensor on the surface of the first buoy and perform initial settings on the first ultrasonic sensor.
[0033] The thickness of biological attachments is also one of the key factors causing the low accuracy of buoy detection data. Therefore, in this step, an ultrasonic sensor needs to be installed on the first buoy to complete the detection of the attachment thickness.
[0034] Preferably, the first ultrasonic sensor is installed on the surface of the first buoy, and the specific coordinate point position can be selected and set according to specific needs.
[0035] After the first ultrasonic sensor is installed, it is necessary to perform initial calibration on the first ultrasonic sensor, record the baseline value when there is no attachment, and set an attachment thickness threshold (such as 1 mm).
[0036] In the above steps, the vibration frequency threshold, the electromagnetic noise intensity threshold and the object thickness threshold are benchmark values for comparing sensor data. When the measured data deviates from the above thresholds by more than the preset values, it means that the data detected by the buoy may have a high probability of being abnormal.
[0037] S14: Acquire at least one first environmental interference data set through at least one of the first acceleration sensor, the first electromagnetic field sensor, and the first ultrasonic sensor, and perform a first preprocessing operation on the at least one first environmental interference data set to obtain first multi-source environmental interference data.
[0038] The first pre-processing operation includes: 1. Sampling frequency setting: Accelerometer: 100Hz; Electromagnetic field sensor: 50Hz; Ultrasonic sensor: 10Hz.
[0039] 2. Data storage format: The sensor data is uniformly stored in JSON format, which contains the following fields: Timestamp (UTC time); Sensor type (accelerometer, electromagnetic field, ultrasonic); Sensor position (coordinates relative to the center of the buoy); Measurement values (vibration frequency, electromagnetic noise intensity, biofouling thickness).
[0040] 3. Outlier filtering: Threshold detection: If the sensor output value exceeds the preset threshold (such as vibration frequency > 5Hz, electromagnetic noise > 50μT, thickness > 1mm), it is marked as abnormal data.
[0041] Sliding window averaging: Perform sliding window averaging (window size = 5) on three consecutive abnormal data points to smooth out short-term interference.
[0042] S2: Perform a first time alignment process and a first expansion process on the first multi-source environment interference data to obtain target multi-source environment interference data.
[0043] In S1, different categories of the first multi-source environmental interference data are obtained through multiple categories of the first environmental interference sensors. In order to perform a comprehensive analysis of the multiple categories of data obtained, time alignment processing needs to be performed first in this step to ensure that multi-source data processing can be performed according to the specified time slice.
[0044] The first expansion processing refers to performing further data expansion processing using historical data on the basis of the first multi-source environmental interference data acquired in real time, so as to make the data required for analysis more sufficient.
[0045] The S2 may specifically include the following sub-steps: S21: Perform a first time alignment process on the first multi-source environment interference data to obtain second multi-source environment interference data.
[0046] The specific steps of the first time alignment process include: Timestamp adjustment: Align sensor data with different sampling frequencies to their lowest common multiple (e.g., 1 minute) using the timestamp alignment method. For example, align one minute of data (6,000 points) from an accelerometer (100 Hz) with one minute of data (1 point) from a meteorological sensor (1 Hz).
[0047] Data interpolation: Linearly interpolate low-frequency data (such as 10 Hz for ultrasonic sensors) to generate a data sequence with the same time resolution as high-frequency data (such as 100 Hz for acceleration sensors).
[0048] Through the above time alignment operation, the data of all sensors are aligned according to a unified timestamp to ensure that the time point of each set of data is consistent.
[0049] S22: Perform a first feature extraction process on the second multi-source environmental interference data, and construct a first multi-source environmental interference feature vector.
[0050] The second multi-source environmental interference data includes original data acquired by multiple types of sensors. In this step, feature extraction is required and intermediate data is obtained through calculation to construct the first multi-source environmental interference feature vector.
[0051] The first feature extraction process may specifically include: 1. Intermediate data calculation: The intermediate data that can be calculated by various environmental sensors may include: Accelerometer: Calculates the root mean square (RMS) value and peak frequency of vibration; Electromagnetic field sensor: calculates the peak intensity and frequency domain energy distribution of electromagnetic noise; Ultrasonic sensor: Calculates the mean and rate of change of biofouling thickness.
[0052] 2. Multi-dimensional feature splicing: The above features are concatenated into a multi-dimensional feature vector, as shown below: [Vibration frequency_RMS, electromagnetic noise_peak, biological attachment_mean, air pressure value, wind speed value].
[0053] S23: Perform a first expansion process on the first multi-source environment interference feature vector to obtain target multi-source environment interference data.
[0054] For each dimension of the first multi-source environmental interference feature vector, it only includes the sensor data obtained in real time by the sensor, which has certain limitations. In order to further reflect the corresponding data change rules in the first multi-source environmental interference feature vector, in this step, the first expansion processing is performed one by one on each data dimension in the first multi-source environmental interference feature vector to satisfactorily obtain the target multi-source environmental interference data.
[0055] The first extension process may specifically include: Time sliding window: This system incorporates the past 24 hours of historical environmental data (e.g., hourly average vibration frequency and electromagnetic noise intensity) and concatenates it with the current feature vector to form a context-aware input vector. The time range of this historical environmental data can also be adaptively adjusted and selected based on specific needs.
[0056] Data normalization: Min-Max normalization is performed on the concatenated feature vectors to eliminate the influence of different dimensions.
[0057] S3: Based on the target multi-source environment interference data, train and obtain a first dynamic interference model.
[0058] The target multi-source environmental interference data output by S2 includes time series change data of multi-dimensional data. The target multi-source environmental interference data can be used to train and generate the first dynamic interference model, and the time series change law of environmental interference data can be reflected in the first dynamic interference model.
[0059] The S3 may specifically include the following sub-steps: S31: Perform a second preprocessing operation on the target multi-source environment interference data to obtain first training sample data.
[0060] The second preprocessing operation is mainly used to standardize the fused target multi-source environmental interference data to eliminate the influence of different dimensions. Specifically, it may include: Missing value filling: The KNN algorithm is used to fill in missing data, and the eigenvalues of the nearest neighbor samples are used for interpolation.
[0061] Standardization: Perform Z-score standardization on the normalized data (mean is 0, standard deviation is 1) to improve model training efficiency.
[0062] S32: Based on the first training sample data, obtain the first dynamic interference model through training.
[0063] The training process may specifically include: Model architecture: The long short-term memory network (LSTM) is used as the basic model. The input layer is a multidimensional feature vector, and the output layer is the compensation coefficient of the target sensor.
[0064] Training process: Data partitioning: The dataset is divided into a training set (70%), a validation set (15%), and a test set (15%).
[0065] Parameter optimization: Use Adam optimizer (learning rate = 0.001) and cross entropy loss function to train the model until the validation set loss converges.
[0066] Overfitting control: Dropout layer (dropout rate = 0.2) and early stopping strategy (patience = 10) were added during training.
[0067] Preferably, after the first dynamic interference model is obtained through training in S32, a model verification step may be further included. Specifically, the step may include: S33: Performing model verification processing on the first dynamic interference model, and outputting the first dynamic interference model if the verification passes.
[0068] The model verification process may specifically include: Cross-validation: K-fold cross-validation (K=5) is used to evaluate the generalization ability of the model on different data subsets.
[0069] Performance metrics: Calculate the mean squared error (MSE) and coefficient of determination (R²) of the model on the test set. If MSE < 0.01 and R² > 0.95, the model passes validation.
[0070] S4: Using the first dynamic interference model, adaptively compensate at least one first environmental data set to obtain at least one target compensated calibration data set.
[0071] In this step, adaptive compensation is performed on at least one first environmental data set obtained by actual measurement using the first dynamic interference model obtained in S3. The adaptive compensation includes two steps: coarse compensation and fine compensation. The first environmental data set may exemplarily include: a barometer, an anemometer, etc.
[0072] Specifically, they may include: S41: Obtain a first coarse compensation coefficient based on the first dynamic interference model, and perform coarse compensation processing on at least one first environmental data set using the first coarse compensation coefficient to obtain at least one first compensated calibration data set.
[0073] The S41 may specifically include: Compensation formula: Based on the compensation coefficient K (such as K=0.98) output by the dynamic interference model, the raw data of the target sensor is linearly corrected.
[0074] Taking the barometer and anemometer as an example, the correction formula may include: Barometer correction: P c =P0×K (P0 is the original pressure value, P c is the coarse compensation value); Anemometer correction: V c =V0×K (V0 is the original wind speed value, V c is the coarse compensation value).
[0075] S42: Performing nonlinear fitting processing and residual correction processing on at least one of the first compensated calibration data sets to obtain a target compensated calibration data set.
[0076] The S42 may specifically include: 1. Nonlinear fitting processing Neural network model: A multi-layer perceptron (MLP) is used to perform nonlinear fitting on the coarsely compensated data.
[0077] Input layer: First compensation calibration dataset (such as P c 、V c ); Hidden layer: two fully connected layers (64 neurons per layer, ReLU activation function); Output layer: target compensation calibration dataset (P f 、V f ).
[0078] Residual correction Calculate the residual between the coarse compensation value and the fine compensation value (ΔP=P f -P c ), and feed it back to the dynamic disturbance model to optimize the compensation coefficient.
[0079] Preferably, specific optimization rules can be set according to specific needs. Specifically, they may include: 1. Relative error calculation: Barometer: |Pf-Pr| / Pr×100% (Pr is the measurement value of the redundant sensor); Anemometer: |Vf-Vr| / Vr×100% (Vr is the measurement value of the redundant sensor).
[0080] Threshold judgment: If the relative error exceeds the preset threshold (such as ±0.1hPa, ±0.5m / s), the model retraining process is triggered.
[0081] 2. Model update Online learning: Periodically receives redundant verification results (e.g., every 24 hours) and initiates the incremental learning process.
[0082] Incremental training: Merge new calibration data with historical data to retrain the dynamic interference model; Parameter optimization: Use transfer learning technology to freeze the parameters of the first half of the model and only fine-tune the compensation coefficient generation layer.
[0083] The present application also proposes a buoy meteorological data calibration and compensation system for executing the above-mentioned buoy meteorological data calibration and compensation method.
[0084] This application proposes a method and system for buoy meteorological data calibration and compensation, relating to the technical field of ocean monitoring technology. By using dynamic environmental interference modeling and multi-sensor collaborative calibration techniques, the system addresses the issue of sensor data deviation in complex ocean environments. The method includes environmental interference parameter acquisition, multi-source data fusion, dynamic interference model generation, adaptive compensation calculation, and compensation effect verification. This invention significantly improves the accuracy and stability of buoy meteorological data and is suitable for long-term ocean observation scenarios.
[0085] The above description is only a preferred embodiment of the present invention. Therefore, any equivalent changes or modifications made according to the structure, characteristics and principles described in the scope of the patent application of the present invention are included in the scope of the patent application of the present invention.
Claims
1. A buoy meteorological data calibration and compensation method, characterized in that: The method includes: S1: at least one first environmental interference sensor is set on a first buoy, at least one first environmental interference data set is collected by the at least one first environmental interference sensor, and first multi-source environmental interference data is obtained based on the at least one first environmental interference data set; S2: performing a first time alignment process and a first expansion process on the first multi-source environment interference data to obtain target multi-source environment interference data; S3: Based on the target multi-source environment interference data, train and obtain a first dynamic interference model; S4: Using the first dynamic interference model, adaptively compensate at least one first environmental data set to obtain at least one target compensated calibration data set.
2. A buoy meteorological data calibration and compensation method according to claim 1, characterized in that: The S1 includes the following sub-steps: S11: installing four first acceleration sensors at four corners of the first buoy and performing initial settings on the first acceleration sensors; S12: Installing a first electromagnetic field sensor on top of the first buoy and performing initial settings on the first electromagnetic field sensor; S13: installing a first ultrasonic sensor on the surface of the first buoy and performing initial settings on the first ultrasonic sensor; S14: Acquire at least one first environmental interference data set through at least one of the first acceleration sensor, the first electromagnetic field sensor, and the first ultrasonic sensor, and perform a first preprocessing operation on the at least one first environmental interference data set to obtain first multi-source environmental interference data.
3. A buoy meteorological data calibration and compensation method according to claim 2, characterized in that: The first pre-processing operation includes: Sampling frequency setting, data storage format adjustment and outlier filtering.
4. A buoy meteorological data calibration and compensation method according to claim 1, characterized in that: The S2 includes the following sub-steps: S21: performing a first time alignment process on the first multi-source environmental interference data to obtain second multi-source environmental interference data; S22: performing a first feature extraction process on the second multi-source environmental interference data, and constructing a first multi-source environmental interference feature vector; S23: Perform a first expansion process on the first multi-source environment interference feature vector to obtain target multi-source environment interference data.
5. A buoy meteorological data calibration and compensation method according to claim 4, characterized in that: The first expansion processing refers to performing further data expansion processing using historical data based on the first multi-source environmental interference feature vector obtained in real time.
6. A buoy meteorological data calibration and compensation method according to claim 4, characterized in that: The specific steps of the first time alignment process include: timestamp adjustment and data interpolation.
7. A buoy meteorological data calibration and compensation method according to claim 1, characterized in that: The S3 includes the following sub-steps: S31: performing a second preprocessing operation on the target multi-source environment interference data to obtain first training sample data; S32: Based on the first training sample data, train and obtain the first dynamic interference model; S33: Performing model verification processing on the first dynamic interference model, and outputting the first dynamic interference model if the verification passes.
8. A buoy meteorological data calibration and compensation method according to claim 6, characterized in that: The S4 includes the following sub-steps: S41: Obtaining a first coarse compensation coefficient based on the first dynamic interference model, and performing coarse compensation processing on at least one first environmental data set using the first coarse compensation coefficient to obtain at least one first compensated calibration data set; S42: Performing nonlinear fitting processing and residual correction processing on at least one of the first compensated calibration data sets to obtain a target compensated calibration data set.
9. A buoy meteorological data calibration and compensation system, used to implement the buoy meteorological data calibration and compensation method according to any one of claims 1 to 8.