Time domain electromagnetic transmitter performance evaluation method based on machine learning

Abstract

The invention discloses a machine learning-based time domain electromagnetic transmitter performance evaluation method, which comprises the following steps of: extracting key parameter data from an operation log of an electromagnetic transmitter, and preprocessing the key parameter data to obtain a standardized data set; constructing a multi-dimensional feature set according to the standardized data set, and dividing the multi-dimensional feature set into an input feature vector and a target output variable; training a regression prediction model according to the multi-dimensional feature set to obtain a trained performance prediction model; and predicting the to-be-evaluated data according to the performance prediction model to obtain a prediction result, and performing feature importance analysis on the prediction result to obtain an analysis result. According to the method, intelligent evaluation of transmitter performance under the conditions of multiple devices and multiple working conditions can be achieved, the limitation of a traditional experience or single-index method is overcome, the precision and interpretability of performance analysis are improved, and a new data driving scheme is provided for instrument operation optimization and abnormality diagnosis.

Description

Technical Field

[0001] This invention belongs to the field of geophysical instrument performance testing and intelligent analysis technology, and in particular relates to a time-domain electromagnetic transmitter performance evaluation method based on machine learning. Background Technology

[0002] Ground-based electromagnetic transmitters are core equipment in time-domain electromagnetic exploration systems, used to generate a primary excitation field in underground media. The stability of their output current and the precise control of their turn-off time directly determine the quality of the received signal, the signal-to-noise ratio, and the inversion accuracy. In actual operation, transmitters are affected by the coupling of various physical parameters such as coil inductance, loop resistance, output voltage, AC component, transmission frequency, and internal temperature, exhibiting complex nonlinear characteristics and operational condition dependencies. Existing technologies mainly employ performance evaluation methods based on statistical analysis. For example, by collecting the transmitter's output voltage, current, and frequency signals, calculating their standard deviation and trends, and using the variance analysis results as performance criteria, a macroscopic assessment of transmission frequency stability can be achieved. In recent years, with the accumulation of operational data from geophysical instruments, machine learning technology has shown great application potential in geophysical modeling and power system analysis, providing new technical approaches for the intelligent evaluation of complex systems.

[0003] However, the aforementioned statistical analysis methods can only reflect the overall fluctuations of the transmitter under steady-state or single operating conditions. The evaluation indicators are limited and it is difficult to reveal the coupling effects between multiple parameters, making it impossible to achieve predictive analysis in complex and variable environments. At the same time, traditional methods lack the ability to explain the physical laws of electromagnetic transmitters, cannot quantitatively assess the specific contributions of each parameter to turn-off performance and current stability, and are difficult to achieve intelligent evaluation and performance prediction under multiple device, multiple load, and multiple environmental conditions.

[0004] Therefore, there is currently a lack of an electromagnetic transmitter performance analysis technology that can integrate multi-dimensional physical parameters, possess high-precision prediction capabilities, and offer physical interpretability. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a time-domain electromagnetic transmitter performance evaluation method based on machine learning, comprising:

[0006] Key parameter data is extracted from the operation log of the electromagnetic transmitter, and the key parameter data is preprocessed to obtain a standardized dataset;

[0007] A multidimensional feature set is constructed based on the standardized dataset, and the multidimensional feature set is divided into input feature vectors and target output variables;

[0008] The regression prediction model is trained based on the multidimensional feature set to obtain the trained performance prediction model.

[0009] The performance prediction model is used to predict the data to be evaluated, and the prediction results are obtained. Feature importance analysis is then performed on the prediction results to obtain the analysis results.

[0010] Preferably, the process of obtaining a standardized dataset includes:

[0011] Extract parameters such as turn-off time, output current, operating voltage, coil inductance, loop resistance, AC component, operating frequency, and temperature to form the original dataset;

[0012] The original dataset is cleaned by removing outliers according to a preset anomaly detection algorithm.

[0013] The cleaned dataset is aligned according to a time reference to obtain a standardized dataset.

[0014] Preferably, the process of dividing the multidimensional feature set into input feature vectors and target output variables includes:

[0015] The inductance, resistance, voltage, AC component, frequency, and temperature parameters are combined to construct the input feature vector;

[0016] Define the turn-off time and output current as the target output variables;

[0017] The multidimensional feature set is grouped and labeled according to the device identification information to obtain a training sample set with group labels.

[0018] Preferably, the process of obtaining the trained performance prediction model includes:

[0019] The gradient boosting tree algorithm is used to construct the initial regression model;

[0020] The multidimensional feature set is divided into a training set and a validation set according to the group identifier;

[0021] The initial regression model is iteratively trained based on the training set, and the model hyperparameters are adjusted based on the validation set to obtain an optimized performance prediction model.

[0022] Preferably, the process of constructing the initial regression model using the gradient boosting tree algorithm includes:

[0023] Multiple base learners based on decision trees are constructed in parallel;

[0024] A comprehensive prediction model is obtained by weighted ensemble of the prediction results of the base learners.

[0025] Based on the performance evaluation results of the comprehensive prediction model, the weights of the base learners are dynamically adjusted.

[0026] Preferably, the process of dividing the multidimensional feature set into a training set and a validation set according to the grouping identifier includes:

[0027] A grouped cross-validation strategy is adopted, and data is grouped according to file identifier or device identifier;

[0028] Based on the grouping results, data from different sources are allocated to different training and validation subsets to obtain non-overlapping training and validation sets.

[0029] Preferably, the process of predicting the data to be evaluated based on the performance prediction model and obtaining the prediction result includes:

[0030] The operating data of the electromagnetic transmitter to be evaluated is input into the performance prediction model, and the predicted values ​​of the turn-off time and output current are calculated based on the performance prediction model.

[0031] The predicted turn-off time and predicted output current are compared with their corresponding measured values ​​to obtain prediction deviation data.

[0032] Preferably, the process of performing feature importance analysis on the prediction results includes:

[0033] The contribution of each input feature to the prediction result is calculated using an interpretability algorithm based on Shapley values;

[0034] Based on the contribution values, a feature importance ranking is generated to obtain the set of dominant influence parameters;

[0035] The influence direction of each physical parameter on the performance of the electromagnetic transmitter is determined based on the set of dominant influencing parameters.

[0036] Preferably, the process of calculating the contribution of each input feature to the prediction result using a Shapley value-based interpretability algorithm includes:

[0037] Calculate the marginal gain of each input feature in different samples to obtain the SHAP value of individual samples;

[0038] A feature influence distribution map is constructed based on the SHAP values ​​of all samples to obtain the non-linear relationship between features and prediction results;

[0039] Based on the characteristic influence distribution map, the coupling effect between parameters is analyzed to obtain the parameter interaction relationship matrix.

[0040] Preferably, the method is used for performance evaluation of electromagnetic transmitters in a time-domain electromagnetic exploration system;

[0041] The electromagnetic transmitter is used to generate a primary excitation field in the underground medium.

[0042] The analysis results are used for transmitter operation optimization and anomaly diagnosis.

[0043] Compared with the prior art, the present invention has the following advantages and technical effects:

[0044] This invention enables intelligent evaluation and high-precision prediction of electromagnetic transmitter performance under multiple devices and operating conditions. It effectively reveals the nonlinear coupling law between multiple physical parameters such as electricity, heat, and frequency, and provides quantitative analysis results of feature contribution with physical interpretability. It significantly improves the accuracy and reliability of transmitter performance analysis and provides data-driven technical support for instrument health monitoring, operation optimization, and anomaly diagnosis. Attached Figure Description

[0045] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0046] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0047] Figure 2 This is a comparison chart of the current prediction results and measured values ​​in an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of the SHAP feature importance analysis for current prediction in an embodiment of the present invention.

[0049] Figure 4 This is a comparison chart of the predicted shutdown time and the measured value in an embodiment of the present invention;

[0050] Figure 5 This is a schematic diagram illustrating the importance analysis of SHAP features of shutdown time in an embodiment of the present invention. Detailed Implementation

[0051] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0052] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0053] like Figure 1 As shown, this embodiment provides a time-domain electromagnetic transmitter performance evaluation method based on machine learning, including:

[0054] Key parameter data were extracted from the operation logs of the electromagnetic transmitter, and the key parameter data was preprocessed to obtain a standardized dataset.

[0055] A multidimensional feature set is constructed based on a standardized dataset, and the multidimensional feature set is divided into input feature vectors and target output variables.

[0056] A regression prediction model is trained based on a multidimensional feature set to obtain a performance prediction model after training.

[0057] The performance prediction model is used to predict the data to be evaluated, and the prediction results are obtained. The feature importance analysis of the prediction results is then performed to obtain the analysis results.

[0058] This embodiment extracts key parameters such as turn-off time, output current, operating voltage, coil inductance, loop resistance, AC component, operating frequency, and temperature from the electromagnetic transmitter's operation log. It then performs anomaly detection, time alignment, and standardization preprocessing on the data to obtain a high-quality, uniformly formatted standardized dataset. This effectively eliminates noise interference and dimensional differences in the original data, providing a reliable data foundation for subsequent modeling.

[0059] Furthermore, the process of obtaining a standardized dataset includes:

[0060] Extract parameters such as turn-off time, output current, operating voltage, coil inductance, loop resistance, AC component, operating frequency, and temperature to form the original dataset;

[0061] The original dataset is cleaned by removing outliers based on a pre-defined anomaly detection algorithm.

[0062] The cleaned dataset is aligned based on a time reference to obtain a standardized dataset.

[0063] Furthermore, the data acquisition and preprocessing involved in this embodiment include:

[0064] Key parameter data were extracted from the operation log files of multiple electromagnetic transmitters. The logs included turn-off time (Toff), output current (I), operating voltage (U), transmitter coil inductance (L), resistance (R), AC component (AC_V), operating frequency (Freq), and information from temperature sensors (T1, T2, T3). The raw data was then subjected to outlier detection, time-base alignment, and standardization to remove invalid records, thus constructing a training sample set.

[0065] This embodiment constructs an input feature vector by combining inductance, resistance, voltage, AC component, frequency, and temperature parameters, and sets the turn-off time and output current as target output variables. At the same time, it groups and marks data according to device identifiers, thus clarifying the input-output relationship of the model and realizing the standardized organization and efficient management of multi-device and multi-operating-condition data.

[0066] Furthermore, the process of dividing the multidimensional feature set into input feature vectors and target output variables includes:

[0067] The inductance, resistance, voltage, AC component, frequency, and temperature parameters are combined to construct the input feature vector;

[0068] Define the turn-off time and output current as the target output variables;

[0069] The multidimensional feature set is grouped and labeled according to the device identification information to obtain a training sample set with group labels.

[0070] Furthermore, the feature construction and sample preparation involved in this embodiment include:

[0071] In each log record, [L, R, U, AC_V; T1, T2, T3, Freq] is used as the input feature vector; and the off-time (Toff) and output current (I) are used as the target output variables. For data from multiple days or multiple devices, file-level grouping identifiers (file_id) are used to prevent sample confusion, and auxiliary labels such as load type and power level are retained.

[0072] Furthermore, the process of obtaining the trained performance prediction model includes:

[0073] The gradient boosting tree algorithm is used to construct the initial regression model;

[0074] The multidimensional feature set is divided into a training set and a validation set based on the grouping identifier;

[0075] The initial regression model is iteratively trained using the training set, and the model hyperparameters are adjusted using the validation set to obtain an optimized performance prediction model.

[0076] This embodiment constructs an initial regression model using the gradient boosting tree algorithm and combines it with a grouped cross-validation strategy to divide the multidimensional feature set into non-overlapping training and validation sets. This effectively prevents data leakage caused by the simultaneous occurrence of data from the same device in both the training and validation sets, and significantly improves the model's generalization ability and robustness.

[0077] Furthermore, the process of constructing the initial regression model using the gradient boosting tree algorithm includes:

[0078] Multiple base learners based on decision trees are constructed in parallel;

[0079] A comprehensive prediction model is obtained by weighted ensemble of the prediction results from the base learners.

[0080] Based on the performance evaluation results of the comprehensive prediction model, the weights of the base learners are dynamically adjusted.

[0081] This embodiment fully leverages the advantages of ensemble learning by constructing multiple decision tree-based base learners in parallel and performing weighted ensemble and dynamic weight adjustment based on the prediction results of each base learner, thereby further improving the overall performance and stability of the comprehensive prediction model.

[0082] Furthermore, the process of dividing the multidimensional feature set into a training set and a validation set based on the grouping identifier includes:

[0083] A grouped cross-validation strategy is adopted, and data is grouped according to file identifier or device identifier;

[0084] Based on the grouping results, data from different sources are allocated to different training and validation subsets to obtain non-overlapping training and validation sets.

[0085] Furthermore, this embodiment uses LightGBM and XGBoost models for regression prediction modeling, wherein:

[0086] Model input: Multidimensional physical parameters X;

[0087] Model outputs: Toff prediction and current prediction;

[0088] GroupKFold cross-validation is used, with file_id grouping to prevent data leakage;

[0089] The main parameters of the model (such as num_leaves, learning_rate, max_depth, etc.) are determined through experiments, and then manually optimized based on the performance on the validation set.

[0090] After training, the model weights are saved for real-time prediction of new data.

[0091] This embodiment obtains a fully tuned performance prediction model by iteratively training the model based on the training set and optimizing hyperparameters such as num_leaves, learning_rate, and max_depth based on the performance of the validation set, thus achieving high accuracy in modeling complex nonlinear relationships.

[0092] Furthermore, the process of obtaining prediction results by making predictions based on the performance prediction model for the data to be evaluated includes:

[0093] The operating data of the electromagnetic transmitter to be evaluated is input into the performance prediction model, and the predicted values ​​of the turn-off time and output current are calculated based on the performance prediction model.

[0094] The predicted turn-off time and predicted output current are compared with their corresponding measured values ​​to obtain prediction deviation data.

[0095] This embodiment inputs the operating data of the electromagnetic transmitter to be evaluated into the trained performance prediction model, calculates the predicted values ​​of the turn-off time and output current, and compares and analyzes them with the measured values ​​to generate prediction deviation data, thereby realizing quantitative prediction and error assessment of the transmitter performance indicators.

[0096] Furthermore, the process of performing feature importance analysis on the prediction results includes:

[0097] The contribution of each input feature to the prediction result is calculated using an interpretability algorithm based on Shapley values;

[0098] Based on the contribution values, feature importance is ranked to obtain the set of dominant influence parameters;

[0099] The direction of influence of each physical parameter on the performance of the electromagnetic transmitter is determined based on the set of dominant influencing parameters.

[0100] This embodiment uses an interpretability algorithm based on Shapley values ​​to calculate the contribution of each input feature to the prediction result, and generates a feature importance ranking based on the contribution value, thereby identifying the set of dominant parameters affecting the performance of the electromagnetic transmitter and providing clear guidance on the direction of physical quantity influence for performance analysis.

[0101] Furthermore, the process of calculating the contribution of each input feature to the prediction result using an interpretability algorithm based on Shapley values ​​includes:

[0102] Calculate the marginal gain of each input feature in different samples to obtain the SHAP value of individual samples;

[0103] A feature influence distribution map is constructed based on the SHAP values ​​of all samples to obtain the non-linear relationship between features and prediction results;

[0104] The coupling effect between parameters is analyzed based on the feature influence distribution map, and the parameter interaction relationship matrix is ​​obtained.

[0105] Furthermore, to enhance the interpretability of the results, this embodiment employs the SHAP algorithm to calculate the contribution of each input parameter to Toff and I. SHAP values ​​reflect the positive or negative impact of parameter changes on the model output, intuitively demonstrating the dominance of different physical quantities on transmitter performance under different operating conditions. If the SHAP results are consistent with physical laws (e.g., an increase in R leads to a decrease in I, and an increase in L leads to an increase in Toff), the physical rationality of the model can be further verified. In practical implementation, the feature contribution calculation module based on Shapley values ​​is used to evaluate the magnitude of the impact of input parameters on the prediction results according to their marginal gains, thereby obtaining the importance ranking and directionality analysis of the parameters.

[0106] This embodiment obtains the SHAP value of individual samples by calculating the marginal gain of each input feature in different samples, and constructs a feature influence distribution map and parameter interaction matrix based on the SHAP values ​​of all samples, revealing the nonlinear coupling effect and complex interaction law between multiple physical parameters of electricity, heat and frequency.

[0107] Furthermore, the method is used for performance evaluation of electromagnetic transmitters in time-domain electromagnetic exploration systems;

[0108] Among them, the electromagnetic transmitter is used to generate a primary excitation field in the underground medium;

[0109] The analysis results are used for transmitter operation optimization and anomaly diagnosis.

[0110] This embodiment, through the aforementioned data-driven modeling and interpretability analysis techniques, achieves intelligent evaluation and high-precision prediction of electromagnetic transmitter performance under multiple devices and operating conditions. It overcomes the limitations of traditional statistical methods, which are only applicable to a single operating condition and lack physical interpretation capabilities, and provides reliable technical support for instrument health monitoring, operation optimization, and anomaly diagnosis.

[0111] This embodiment applies the method specifically to the performance evaluation of electromagnetic transmitters in time-domain electromagnetic exploration systems, enabling it to effectively handle data characteristic fluctuations caused by complex loads and environmental changes in underground exploration scenarios. This significantly improves the performance analysis accuracy and reliability of electromagnetic transmitters in actual field operations, providing an intelligent solution for ensuring the quality of electromagnetic exploration data.

[0112] This embodiment discloses a time-domain electromagnetic transmitter performance evaluation method based on machine learning. It extracts key parameters such as turn-off time (Toff), current (I), inductance (L), resistance (R), voltage (U), AC component (AC_V), frequency (Freq), and temperature (T1, T2, T3) from the operation logs of multiple transmitters. The data undergoes anomaly detection and standardization to construct a multi-dimensional feature set. Predictive models for Toff and I are established using gradient boosting tree algorithms (LightGBM, XGBoost). A SHAP value interpretation mechanism is introduced to quantitatively analyze the impact of each physical parameter on transmitter turn-off performance and current stability. This embodiment enables intelligent evaluation of transmitter performance under multiple devices and operating conditions, overcoming the limitations of traditional empirical or single-index methods, improving the accuracy and interpretability of performance analysis, and providing a new data-driven solution for instrument operation optimization and anomaly diagnosis.

[0113] In an optional implementation, the output current (I) of the electromagnetic transmitter is predicted using the LightGBM and XGBoost models according to the method of this embodiment. The input features of this method include parameters such as the transmitter's operating voltage (U), inductance (L), resistance (R), temperature (T1, T2, T3), AC component (AC), and operating frequency (Freq). The model is trained using GroupKFold cross-validation to ensure generalization capability under multiple device and load conditions.

[0114] The results of the current prediction are shown in the figure below. Figure 2 As shown in the figure, the comparison between the current predicted by the model and the measured data demonstrates that the model can effectively predict the current fluctuations of the transmitter. Figure 2 The small error indicates that the model has high prediction accuracy and can accurately track current changes in most time periods.

[0115] Furthermore, such as Figure 3 As shown in the figure, this plot illustrates the contribution (measured by gain) of each input feature to the current prediction residual in the LightGBM model. The results indicate that the resistance... With inductance It is the main factor affecting current prediction error; temperature sensor With voltage Importance is secondary; the weight of communication ,temperature With frequency The impact on the model output is relatively small. This feature ordering conforms to the physical laws of electromagnetic transmitters, namely, the current is controlled by the combined effects of resistance, voltage, and inductance.

[0116] In an optional implementation, this embodiment uses the same model structure and training method to predict the transmitter's shutdown time (Toff). By optimizing the model hyperparameters (such as num_leaves, learning_rate, and max_depth) and using GroupKFold cross-validation for verification, a relatively ideal prediction effect is achieved. A comparison of the predicted and actual shutdown times is shown in the figure below. Figure 4 As shown.

[0117] Furthermore, such as Figure 5 As shown, Figure 5 This demonstrates the importance of the LightGBM model's gain on various features in turn-off time prediction. Results show: voltage It is the dominant feature, contributing the most to the model output; resistance Secondly; the amount of exchange The influence of voltage is also quite significant; the influence of temperature-related characteristics (T1, T2, T3) is relatively small. This ranking of characteristics is consistent with the physical laws of electromagnetic transmitters: the off-time is directly related to the voltage, and an increase in inductance will lead to a slower energy release, thereby prolonging Toff.

[0118] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A time-domain electromagnetic transmitter performance evaluation method based on machine learning, characterized in that, include: Key parameter data is extracted from the operation log of the electromagnetic transmitter, and the key parameter data is preprocessed to obtain a standardized dataset; A multidimensional feature set is constructed based on the standardized dataset, and the multidimensional feature set is divided into input feature vectors and target output variables; The regression prediction model is trained based on the multidimensional feature set to obtain the trained performance prediction model. The performance prediction model is used to predict the data to be evaluated, and the prediction results are obtained. Feature importance analysis is then performed on the prediction results to obtain the analysis results.

2. The method according to claim 1, characterized in that, The process of obtaining a standardized dataset includes: Extract parameters such as turn-off time, output current, operating voltage, coil inductance, loop resistance, AC component, operating frequency, and temperature to form the original dataset; The original dataset is cleaned by removing outliers according to a preset anomaly detection algorithm. The cleaned dataset is aligned according to a time reference to obtain a standardized dataset.

3. The method according to claim 1, characterized in that, The process of dividing the multidimensional feature set into input feature vectors and target output variables includes: The inductance, resistance, voltage, AC component, frequency, and temperature parameters are combined to construct the input feature vector; Define the turn-off time and output current as the target output variables; The multidimensional feature set is grouped and labeled according to the device identification information to obtain a training sample set with group labels.

4. The method according to claim 1, characterized in that, The process of obtaining a fully trained performance prediction model includes: The gradient boosting tree algorithm is used to construct the initial regression model; The multidimensional feature set is divided into a training set and a validation set according to the group identifier; The initial regression model is iteratively trained based on the training set, and the model hyperparameters are adjusted based on the validation set to obtain an optimized performance prediction model.

5. The method according to claim 4, characterized in that, The process of constructing an initial regression model using the gradient boosting tree algorithm includes: Multiple base learners based on decision trees are constructed in parallel; A comprehensive prediction model is obtained by weighted ensemble of the prediction results of the base learners. Based on the performance evaluation results of the comprehensive prediction model, the weights of the base learners are dynamically adjusted.

6. The method according to claim 4, characterized in that, The process of dividing the multidimensional feature set into a training set and a validation set according to the group identifier includes: A grouped cross-validation strategy is adopted, and data is grouped according to file identifier or device identifier; Based on the grouping results, data from different sources are allocated to different training and validation subsets to obtain non-overlapping training and validation sets.

7. The method according to claim 1, characterized in that, The process of predicting the data to be evaluated based on the performance prediction model and obtaining the prediction results includes: The operating data of the electromagnetic transmitter to be evaluated is input into the performance prediction model, and the predicted values ​​of the turn-off time and output current are calculated based on the performance prediction model. The predicted turn-off time and predicted output current are compared with their corresponding measured values ​​to obtain prediction deviation data.

8. The method according to claim 1, characterized in that, The process of performing feature importance analysis on the prediction results includes: The contribution of each input feature to the prediction result is calculated using an interpretability algorithm based on Shapley values; Based on the contribution values, a feature importance ranking is generated to obtain the set of dominant influence parameters; The influence direction of each physical parameter on the performance of the electromagnetic transmitter is determined based on the set of dominant influencing parameters.

9. The method according to claim 8, characterized in that, The process of calculating the contribution of each input feature to the prediction result using an interpretable algorithm based on Shapley values ​​includes: Calculate the marginal gain of each input feature in different samples to obtain the SHAP value of individual samples; A feature influence distribution map is constructed based on the SHAP values ​​of all samples to obtain the non-linear relationship between features and prediction results; Based on the characteristic influence distribution map, the coupling effect between parameters is analyzed to obtain the parameter interaction relationship matrix.

10. The method according to claim 1, characterized in that, The method is used for performance evaluation of electromagnetic transmitters in time-domain electromagnetic exploration systems. The electromagnetic transmitter is used to generate a primary excitation field in the underground medium. The analysis results are used for transmitter operation optimization and anomaly diagnosis.

Images