Neural network-based aeromagnetic compensation method and device, and storage medium

By using one-dimensional convolutional neural network and Attention mechanism in the neural network avionic compensation method, geomagnetic, interference and diurnal networks are established, the problem of poor avionic interference compensation effect in complex magnetic measurement environments is solved, and aviation geophysical exploration data is improved quality and computational efficiency.

CN118938331AActive Publication Date: 2024-11-12CHINA UNIV OF GEOSCIENCES (WUHAN)

Patent Information

Application Number
CN202411107845.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-13
Publication Date
2024-11-12
Estimated Expiration
2044-08-13

AI Technical Summary

Technical Problem

The existing neural network avionic compensation method has poor compensation effect in complex magnetic measurement environments, low computational efficiency, and it is difficult to effectively deal with magnetic interference problems with complex flight conditions.

Method used

Adopt avionic compensation method based on neural networks, including calibrating flight data acquisition, building input groups, establishing geomagnetic networks, interference networks and diurnal networks, and improving generalization capabilities through one-dimensional convolutional neural networks and Attention mechanism (ECA module). After training, the interference network is retained as an avionic compensation model for magnetic interference compensation of flight data.

Benefits of technology

It improves the compensation effect and data quality of aeronautical magnetic interference, improves the accuracy and calculation efficiency of aeronautical geophysical exploration data, and performs well in complex magnetic measurement environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118938331A_ABST
    Figure CN118938331A_ABST
Patent Text Reader

Abstract

The invention provides an aeromagnetic compensation method and device based on a neural network and a storage medium, and relates to the field of aeromagnetic detection, and the method comprises the steps: calibrating flight, obtaining calibration flight data, selecting a time point, processing the calibration flight data, and constructing an input group; constructing and dividing a data set by taking the input group as a unit, establishing an aeromagnetic compensation architecture based on a neural network, specifically comprising a geomagnetic network, an interference network and a daily variation network, training the aeromagnetic compensation architecture by using the data set, reserving the interference network as an aeromagnetic compensation model after training, normalizing flight data to be processed, and inputting the normalized flight data into the aeromagnetic compensation model, therefore, aeromagnetic compensation is realized; the device and the storage medium are used for implementing the method. The method has the beneficial effects that the compensation effect of the verification flight experiment is improved, a good aeromagnetic interference compensation effect can also be achieved in a complex magnetic survey environment, and the quality of airborne geophysical prospecting data is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of aeromagnetic detection, and in particular to an aeromagnetic compensation method, device and storage medium based on a neural network. Background Art

[0002] The spatial distribution of the Earth's magnetic field provides an important basis for the exploration of solid and oil and gas resources, the detection of unexploded ordnance, anti-submarine warfare and underground ore exploration. With the rapid development of the economy and the continuous consumption of mineral resources, the country's demand for underground resources is constantly increasing. As the near-surface mineral resources that are easy to develop are close to exhaustion, the focus of exploration work has gradually shifted to deep minerals and mining in remote areas, which has put high demands on previous exploration technologies. As an important method in the field of geophysical exploration, aeromagnetic exploration is widely used in earth science research, resource exploration, military exploration and other fields. During the aeromagnetic survey process, since the flight platform carrying the magnetic detection instrument cannot be completely magnetized, magnetic interference will be formed during the movement and mixed in the magnetic survey data. This will cause the magnetic survey data to fail to correctly reflect the magnetic characteristics of the area to be measured, so it needs to be compensated.

[0003] There are two types of aeromagnetic compensation methods: The physical model-based method achieves compensation by physically modeling the interfering magnetic field (such as the TL model) and solving the magnetic interference coefficient in the model. Its essence is to simplify the nonlinear problem into a linear problem. The calculation is simple, but there are many problems such as multicollinearity; The non-physical model-based method: first establish a neural network magnetic compensation model, then obtain the interfering magnetic field network through iterative training, and finally measure the corresponding data during flight to obtain the output interfering magnetic field, thereby completing the compensation.

[0004] At present, in the research of neural network magnetic compensation methods, in order to improve its compensation accuracy and generalization performance, related work is mainly carried out from two aspects: model structure and training process, such as introducing random noise to regularize the model, using recurrent neural network to improve the long-term generalization performance of the model, introducing genetic algorithm to optimize initialization parameters, etc. However, there is less attention paid to the compensation research for magnetic survey scenes with very complex and harsh flight conditions. In complex magnetic survey environments, the existing neural network methods have poor compensation effects and low computational efficiency. Summary of the invention

[0005] The purpose of the present invention is to solve the problem that the prior art has poor aeromagnetic interference compensation effect in a complex magnetic survey environment. The present invention provides an aeromagnetic compensation method, device and storage medium based on a neural network. The aeromagnetic compensation method based on a neural network mainly includes the following steps: S1. Calibration flight, obtain calibration flight data: S2, selecting a time point, processing the calibration flight data and then constructing an input group, and constructing and dividing the data set based on the input group; The input group includes the total field magnetic value and the normalized processing results of time, direction cosines and spatial variables obtained by processing the calibration flight data; S3. Establish an aeromagnetic compensation framework based on neural networks, including geomagnetic network, interference network and diurnal variation network; S4, using the data set to train the aeromagnetic compensation architecture, and retaining the interference network as the aeromagnetic compensation model after training; S5. The flight data to be processed is normalized and then input into the aeromagnetic compensation model to achieve aeromagnetic compensation.

[0006] Furthermore, the calibration flight is the flight of the aircraft at a preset flight altitude and a flight altitude variation range, during which small maneuvers such as pitch, roll and yaw are repeated; The calibration flight data includes the aircraft's attitude, position and time information as well as the three-axis fluxgate magnetometer and optically pumped magnetometer measurements.

[0007] Furthermore, step S2 is specifically as follows: S21, selecting a time point in the calibration flight process according to a preset time interval; S22, process the calibration flight data corresponding to the time point, and construct the original variable group corresponding to the time point, specifically: S221, combining the selected time point and the square of the calculated time point as a time variable; S222, calculating the direction cosines by combining the attitude information of the aircraft corresponding to the selected time point with the measurement value of the three-axis fluxgate magnetometer, and calculating and obtaining the quadratic term and the differential term of the direction cosines, and combining them as the direction cosine variable; S223, selecting and obtaining the longitude, latitude and altitude from the position information of the aircraft corresponding to the selected time point, and combining the calculated longitude square value, latitude square value and product of longitude and latitude as the position variable; S224, combining the time variable, direction cosine variable and position variable of the selected time point as an original variable group; S23, obtaining the optical pump magnetometer measurement value at the time point as the total field magnetic value; S24. Normalize the data in all original variable groups according to the type of variables, and then combine the total field magnetic value corresponding to the time point as a training group, construct a data set with the training group, and divide it into a training set and a validation set.

[0008] Furthermore, the geomagnetic network is a one-dimensional convolutional neural network, which consists of a convolutional layer, an activation layer, a pooling layer, a Flatten layer and a fully connected layer. The input is the normalized position variable, and the output is the fitted geomagnetic field value.

[0009] Furthermore, the interference network consists of a one-dimensional convolutional neural network and an ECA module, with the input being the normalized direction cosine variables and the output being the fitted interference magnetic field value.

[0010] Furthermore, the diurnal variation network consists of 4 fully connected layers. The first three fully connected layers have 20 neurons and use tanh activation function. The last fully connected layer has only one neuron and uses linear activation function. It inputs the normalized time variable and outputs the fitted geomagnetic diurnal variation value.

[0011] Furthermore, the training process of the aeromagnetic compensation architecture is as follows: Furthermore, step S5 is specifically as follows: The flight data to be tested is obtained during the test flight, and after normalization, it is input into the flight compensation network to obtain the predicted magnetic interference value, which is then eliminated from the total field magnetic value measured by the optically pumped magnetometer to complete the aeromagnetic compensation operation.

[0012] A storage medium stores instructions and data for implementing an aeromagnetic compensation method based on a neural network.

[0013] A computer device comprises: a processor and the storage medium; the processor loads and executes instructions and data in the storage medium to implement an aeromagnetic compensation method based on a neural network.

[0014] The beneficial effects of the technical solution provided by the present invention are as follows: the present invention refines the aeromagnetic compensation architecture based on a neural network, uses a 1D CNN-Attention model in the interference network, and improves the generalization ability of the convolutional network by adding an ECA attention mechanism between convolutional layers, thereby improving the compensation effect of the verification flight experiment, and having a good aeromagnetic interference compensation effect in a complex magnetic survey environment, thereby improving the quality of aerial geophysical exploration data. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which: Figure 1 is a flow chart of an aeromagnetic compensation method based on a neural network in an embodiment of the present invention; Figure 2 is a schematic diagram of an aeromagnetic compensation architecture structure in an embodiment of the present invention; Figure 3 is a schematic diagram of the ECA module structure in an embodiment of the present invention; Figure 4 is a schematic diagram of global pooling in an embodiment of the present invention; Figure 5 It is a schematic diagram of the operation of the hardware device in the embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.

[0017] Embodiments of the present invention provide a neural network-based aeromagnetic compensation method, device, and storage medium.

[0018] Please refer to Figure 1 , Figure 1 : is a flowchart of an aeromagnetic compensation method based on a neural network in an embodiment of the present invention, which specifically includes the following steps: The first step is to calibrate the flight and obtain the calibration flight data.

[0019] Calibration flight is the flight of the aircraft at a preset flight altitude and a flight altitude variation range, during which small maneuvers such as pitch, roll, and yaw are repeated; The calibration flight data includes the aircraft's attitude, position and time information as well as the three-axis fluxgate magnetometer and optically pumped magnetometer measurements.

[0020] The second step is to select time points, process the calibration flight data and then construct input groups, and construct and divide the data set based on the input groups. The input groups contain the total field magnetic value and the normalized processing results of time, direction cosines, and spatial variables obtained by processing the calibration flight data, specifically: Step 1: Select a time point during the calibration flight according to a preset time interval.

[0021] Step 2: Process the calibration flight data corresponding to the time point and construct the original variable group corresponding to the time point, specifically: (1) Combine the selected time point and the square of the calculated time point as the time variable; (2) Calculate the direction cosines by combining the attitude information of the aircraft corresponding to the selected time point with the measurement value of the three-axis fluxgate magnetometer, and calculate the quadratic term and the differential term of the direction cosines, and combine them as the direction cosine variable; (3) The longitude, latitude and altitude are obtained from the position information of the aircraft corresponding to the selected time point, and the calculated longitude square value, latitude square value and longitude and latitude product are used as the position variable; (4) Combine the time variable, direction cosine variable, and position variable at the selected time point as an original variable group.

[0022] Step three, obtaining the optical pump magnetometer measurement value at a time point as the total field magnetic value.

[0023] Step 4: Normalize the data in all original variable groups according to the type of variable, and then combine the total field magnetic value corresponding to the time point as the training group, construct a data set with the training group, and divide it into a training set and a validation set.

[0024] The third step is to establish an aeromagnetic compensation framework based on neural networks, which includes geomagnetic networks, interference networks and diurnal variation networks, such as Figure 2 shown.

[0025] The design of these three networks is based on the measurement characteristics. The data collected by the total field magnetometer in the aeromagnetic survey contains various magnetic survey sources:

[0026] In the formula, H Total-Mag is the measured value of the optical pump magnetometer, H G is the geomagnetic field value, H Sun is the diurnal variation of the geomagnetic field, H I is the interference magnetic field value.

[0027] The geomagnetic network is a one-dimensional convolutional neural network, which consists of a convolutional layer, an activation layer, a pooling layer, a Flatten layer, and a fully connected layer. The input is the normalized position variable, and the output is the fitted geomagnetic field value.

[0028] The formula for forward propagation in a one-dimensional convolutional layer is as follows:

[0029] in, is the input of the jth neuron in the lth layer, is the bias of the jth neuron in the lth layer, is the output of the i-th neuron in the l-1th layer, is the kernel from the i-th neuron in the l-1th layer to the j-th neuron in the lth layer, Used to perform one-dimensional convolution. The size of the convolution kernel in the geomagnetic network is set to 3, and the mode of the convolution layer is zero padding. At this time, the convolution operation starts when the center of the filter coincides with the corner of the input tensor. The input array With the output array The dimensions are the same. Activation function in the activation layer The intermediate output for:

[0030] In the one-dimensional convolutional neural network, this paper uses Relu as the function of the activation layer , the formula of Relu activation function is as follows:

[0031] Therefore, when input In the negative half zone, the intermediate output of the Relu activation function is zero, when the input In the positive half zone, the intermediate output With input The value of is the same. The derivative of the Relu function in the negative half is 0, and the output of the activation function is 0, and the neuron is not activated. The derivative of the Relu function in the positive half is 1. Therefore, the Relu function will suppress the activation of some neurons, making the neurons sparsely activated. In neural network compensation, the calculation method of the Relu activation function is simple, but it can achieve excellent compensation effect with the shortest training time. If the tanh activation function is used, the training time will increase and the compensation effect will be reduced. The compensation effect of using the Leaky Relu activation function is close to that of the Relu activation function, but it will significantly increase the time cost of the neural network compensation method.

[0032] The interference network consists of a one-dimensional convolutional neural network and an ECA module. The input is the normalized direction cosine variable, and the output is the fitted interference magnetic field value.

[0033] Since the interference network has 18 input items and the causal relationship between input and output is relatively complex under complex magnetic measurement conditions, using a deeper convolutional network in the interference network can achieve better nonlinear fitting ability by superimposing nonlinear functions, thereby better learning the nonlinear characteristics of the interference network.

[0034] The ECA module is used between the convolutional layers to calculate the attention weights corresponding to the interference magnetic field features of each channel in the convolutional layer. The structure of the ECA module is as follows: Figure 3 As shown in the figure, 3*1 convolution and relu function are used to extract the characteristic information of magnetic interference contained in the input item, and the ECA module is used between the convolution layers to calculate the attention weights corresponding to the interference magnetic field characteristics of each channel in the convolution layer, which makes the interference magnetic field characteristics of different channels have different effects on the output results of the interference network.

[0035] The ECA module receives the interference magnetic field features input from the convolution layer and performs global average pooling on them. The global average pooling is as follows: Figure 4 shown.

[0036] Global average pooling compresses the interference magnetic field features on each channel in the convolution layer into a scalar and converts it into A matrix of size is the number of channels. The matrix of size is input into the one-dimensional convolution layer. The number of channels of the one-dimensional convolution is 1 and the convolution kernel size is With the number of channels Related:

[0037] To be closest When the SoftMax function is used in the ECA module to calculate the weights of each channel, a weight of 0 to 1 is assigned to each interfering magnetic field feature input into the SoftMax function. The SoftMax function is:

[0038] in, is the interference magnetic field feature of the ith channel after convolution and global average pooling, and n is the number of channels. The SoftMax function converts the exponential interference magnetic field feature of the ith channel into Divided by the sum of the exponentially interfering magnetic field characteristics of all channels, Mapped to a value of (0,1), which is the weight of the i-th channel. Finally, the attention weight is multiplied by the interference magnetic field characteristics of different channels to obtain the output result of the attention layer.

[0039] The diurnal variation network consists of 4 fully connected layers. The first three fully connected layers have 20 neurons and use the tanh activation function. The last fully connected layer has only one neuron and uses a linear activation function. It inputs the normalized time variable to fit the diurnal magnetic field changes and outputs the fitted geomagnetic diurnal variation value.

[0040] In the fourth step, the dataset is used to train the aeromagnetic compensation architecture, and after training, the interference network is retained as the aeromagnetic compensation model.

[0041] During training, a training group is extracted from the training set, and the original variable groups in the group are respectively input into the aeromagnetic compensation framework, specifically: the normalized position variables are input into the geomagnetic network to obtain the fitted geomagnetic magnetic field value, the normalized direction cosine variables are input into the interference network to obtain the fitted interference magnetic field value, and the normalized time variables are input into the diurnal variation network to obtain the fitted geomagnetic diurnal variation value. The three are added together to obtain the predicted total field value obtained by the entire model, and the predicted total field value is compared with the total field magnetic value in the training group to calculate the loss. The model is updated based on the calculated loss until the entire aeromagnetic compensation framework converges.

[0042] The validation set data is input into the converged aeromagnetic compensation framework for verification to ensure the feasibility of the aeromagnetic compensation framework. Finally, an aeromagnetic compensation framework that meets the accuracy requirements is output, and the interference network is extracted as a separate aeromagnetic compensation model.

[0043] The fifth step is to normalize the flight data to be processed and input it into the aeromagnetic compensation model to achieve aeromagnetic compensation. Specifically: The flight data to be tested is obtained during the test flight, and after normalization, it is input into the flight compensation network to obtain the predicted magnetic interference value, which is then eliminated from the total field magnetic value measured by the optically pumped magnetometer to complete the aeromagnetic compensation operation.

[0044] During the test flight, the measured data is directly input into the interference network to obtain the predicted magnetic interference value, and the magnetic interference is eliminated in the subsequent total field data to complete the compensation operation.

[0045] After the compensation is completed, the compensation result is quantitatively evaluated by the improvement ratio IR. IR is the ratio of the standard deviation of the uncompensated total magnetic field after bandpass filtering to the standard deviation of the compensated total magnetic field after bandpass filtering.

[0046] In order to evaluate the effectiveness of the proposed method, this experiment conducted a calibration flight and a test flight. Both flights were rectangular paths with orthogonal headings, the flight altitude was about 2500m, the altitude change amplitude was about 150m, and 4-6 small-amplitude maneuvers of pitch (±5°), roll (±10°), and yaw (±5°) were repeated in each direction during the flight.

[0047] In order to compare the effectiveness of the proposed method, the compensation results of this method are compared with those of the existing BPNN, GRNN, ResNet and market products (RMS). The input of the magnetic interference network in the BP method is consistent with that in this paper, and the compensation effect is compared as shown in Table 1.

[0048] Table 1 Comparison of compensation effects

[0049] In the calibration flight experiment, the proposed method was improved by 113%, 81%, and 164% compared with BPNN, GRNN, and Resnet, respectively, and also improved by 11% compared with the RMS aeromagnetic compensator. In the verification flight experiment, the proposed method was improved by 107%, 813%, and 419% compared with BPNN, GRNN, and Resnet, respectively. This shows that the proposed method can effectively improve the accuracy of the neural network-based aeromagnetic compensation method. Since the network trained in the calibration flight experiment is used for the verification flight, the compensation effect of the verification flight experiment will decrease. The decline rates of BPNN, GRNN, Resnet and the proposed method are 15%, 84%, 58%, and 18%, respectively. Although the generalization ability of the proposed method is not the best, its comprehensive ability in terms of accuracy and generalization ability is the best.

[0050] See also Figure 5, Figure 5 40 is a schematic diagram of the working of the hardware device of an embodiment of the present invention, wherein the hardware device specifically comprises: a computer device 401, a processor 402 and a storage medium 403.

[0051] A computer device 401: The computer device 401 implements the neural network-based aeromagnetic compensation method.

[0052] Processor 402: The processor 402 loads and executes the instructions and data in the storage medium 403 to implement the aeromagnetic compensation method based on a neural network.

[0053] Storage medium 403: The storage medium 403 stores instructions and data; the storage medium 403 is used to implement the aeromagnetic compensation method based on a neural network.

[0054] The beneficial effects of the present invention are as follows: the present invention refines the aeromagnetic compensation architecture based on a neural network, uses a 1D CNN-Attention model in the interference network, and improves the generalization ability of the convolutional network by adding an ECA attention mechanism between convolutional layers, thereby improving the compensation effect of the verification flight experiment, and having a good aeromagnetic interference compensation effect in a complex magnetic survey environment, thereby improving the quality of aerial geophysical exploration data.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An aeromagnetic compensation method based on neural network, It is characterized in that The specific steps include: S1. Calibration flight, obtaining calibration flight data; S2, selecting a time point, processing the calibration flight data and then constructing an input group, and constructing and dividing the data set based on the input group; The input group includes the total field magnetic value and the normalized processing results of time, direction cosines and spatial variables obtained by processing the calibration flight data; S3. Establish an aeromagnetic compensation framework based on neural networks, including geomagnetic network, interference network and diurnal variation network; S4, using the data set to train the aeromagnetic compensation architecture, and retaining the interference network as the aeromagnetic compensation model after training; S5. The flight data to be processed is normalized and then input into the aeromagnetic compensation model to achieve aeromagnetic compensation.

2. The aeromagnetic compensation method based on a neural network as claimed in claim 1, It is characterized in that The calibration flight is a flight of the aircraft at a preset flight altitude and a flight altitude variation range, during which small maneuvers such as pitch, roll and yaw are repeated; The calibration flight data includes the attitude, position and time information of the aircraft and the measurement values ​​of the three-axis fluxgate magnetometer and the optical pump magnetometer.

3. The aeromagnetic compensation method based on a neural network as claimed in claim 2, It is characterized in that Step S2 is specifically as follows: S21, selecting a time point in the calibration flight process according to a preset time interval; S22, processing the calibration flight data corresponding to the time point, and constructing the original variable group corresponding to the time point; S23, obtaining the optical pump magnetometer measurement value at the time point as the total field magnetic value; S24. Normalize the data in all original variable groups according to the type of variables, and then combine the total field magnetic value corresponding to the time point as a training group, construct a data set with the training group, and divide it into a training set and a validation set.

4. The aeromagnetic compensation method based on a neural network as claimed in claim 3, It is characterized in that Step S22 is specifically as follows: S221, combining the selected time point and the square of the calculated time point as a time variable; S222, calculating the direction cosines by combining the attitude information of the aircraft corresponding to the selected time point with the measurement value of the three-axis fluxgate magnetometer, and calculating and obtaining the quadratic term and the differential term of the direction cosines, and combining them as the direction cosine variable; S223, selecting and obtaining the longitude, latitude and altitude from the position information of the aircraft corresponding to the selected time point, and combining the calculated longitude square value, latitude square value and product of longitude and latitude as the position variable; S224. Combine the time variable, direction cosine variable, and position variable of the selected time point as an original variable group.

5. The aeromagnetic compensation method based on neural network as claimed in claim 1, It is characterized in that The geomagnetic network is a one-dimensional convolutional neural network, which consists of a convolutional layer, an activation layer, a pooling layer, a Flatten layer and a fully connected layer. The input is a normalized position variable, and the output is a fitted geomagnetic field value.

6. The aeromagnetic compensation method based on neural network as claimed in claim 1, It is characterized in that The interference network consists of a one-dimensional convolutional neural network and an ECA module, the input is a normalized direction cosine variable, and the output is a fitted interference magnetic field value.

7. The aeromagnetic compensation method based on a neural network as claimed in claim 1, It is characterized in that The diurnal variation network consists of 4 fully connected layers. The first three fully connected layers have 20 neurons and use the tanh activation function. The last fully connected layer has only one neuron and uses a linear activation function. It inputs the normalized time variable and outputs the fitted geomagnetic diurnal variation value.

8. The aeromagnetic compensation method based on neural network as claimed in claim 1, It is characterized in that Step S5 is specifically as follows: The flight data to be tested is obtained during the test flight, and after normalization, it is input into the flight compensation network to obtain the predicted magnetic interference value, which is then eliminated from the total field magnetic value measured by the optically pumped magnetometer to complete the aeromagnetic compensation operation.

9. A storage medium, Features: The storage medium stores instructions and data for implementing the neural network-based aeromagnetic compensation method described in any one of claims 1 to 8.

10. A computer device, Features: include: Processor and storage medium; the processor loads and executes instructions and data in the storage medium to implement the neural network-based aeromagnetic compensation method described in any one of claims 1 to 8.

Citation Information

Patent Citations

  • Aeromagnetic compensation method based on generalized regression neural network

    CN110187393A

  • Method and device for compensating magnetic interference caused by airborne electronic equipment

    CN112858959A

  • Magnetic measurement data processing method and system based on multi-machine cooperation anti-diurnal variation magnetic interference

    CN118426062A

  • Magnetic compensation method based on aeromagnetic compensation error model

    US20200301040A1

Cited By

  • Aeromagnetic data compensation method and system for multi-rotor unmanned aerial vehicle

    CN120428344A

  • A method and system for compensating aeromagnetic data of a multi-rotor UAV

    CN120428344B

  • Lightweight magnetic compensation neural network method for aerial computing platform of multi-rotor unmanned aerial vehicle

    CN120706476A

  • Global magnetic disturbance compensation method in aviation magnetic survey

    CN120993504A