Precision optimization method for electromagnetic sensor

Abstract

The invention relates to the technical field of sensor signal processing, in particular to a precision optimization method for an electromagnetic sensor. The invention aims to solve the problem that the precision and reliability of a measurement result are reduced due to the fact that an electromagnetic sensor is easily influenced by various interference factors in an actual working environment. The method comprises the steps of firstly obtaining an original measurement signal of an electromagnetic sensor; an adaptive noise suppression algorithm is used to suppress environmental noise; the abnormal data identification model monitors and rejects the abnormal data in real time; analyzing the sensor data through a parameter adjustment model, and adjusting the working parameters of the sensor; a built-in reference signal source is used for dynamically calibrating the electromagnetic sensor regularly; and finally, acquiring monitoring data of the plurality of electromagnetic sensors through the cloud platform so as to carry out remote monitoring and dynamic calibration on the electromagnetic sensors. According to the invention, the dynamic calibration function of the electromagnetic sensor and the remote monitoring and dynamic calibration of the cloud platform are combined, so that the measurement precision of the electromagnetic sensor is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sensor signal processing, and in particular to a precision optimization method for an electromagnetic sensor. Background Art

[0002] Electromagnetic sensor is a sensor that uses electromagnetic principle for measurement and detection. Its working principle is mainly based on electromagnetic induction and electrical processing signal. It senses the magnetic field change of the target object and converts it into an electrical signal, and then obtains the final output through signal processing. Electromagnetic sensors are widely used in industrial control, aerospace, national defense and security, etc., to measure physical quantities such as electromagnetic field strength, magnetic field direction and magnetic flux density. These technical parameters are crucial for many key applications. However, in the actual working environment, electromagnetic sensors are easily affected by various interference factors, resulting in reduced accuracy and reliability of the measurement results.

[0003] Therefore, there is an urgent need to develop an accuracy optimization method for electromagnetic sensors to overcome the shortcomings of the prior art. Summary of the invention

[0004] In order to solve the problem in the prior art that electromagnetic sensors are affected by various interference factors, resulting in reduced accuracy and reliability of measurement results, the present invention provides a method for optimizing the accuracy of electromagnetic sensors, comprising the following steps: S100: acquiring an original measurement signal of an electromagnetic sensor, and preprocessing the original measurement signal; S200: Uses adaptive noise suppression algorithm to dynamically adjust adaptive filter parameters to suppress environmental noise; S300: Establishing an abnormal data recognition model by collecting historical measurement data, and performing real-time monitoring and elimination of abnormal data through the abnormal data recognition model; S400: Analyze the electromagnetic sensor data and adjust the working parameters of the electromagnetic sensor; S500: using the built-in reference signal source, regularly and dynamically calibrating the electromagnetic sensor and updating the calibration coefficient.

[0005] Furthermore, the specific steps of using the adaptive noise suppression algorithm to dynamically adjust the adaptive filter parameters to suppress environmental noise are: S201: real-time collection of environmental noise data in the working environment of the electromagnetic sensor, and obtaining frequency characteristics of the noise by spectrum analysis; S202: adjusting the parameters of the adaptive filter, including the cutoff frequency and the order, using the least mean square LMS algorithm according to the frequency characteristics of the noise; S203: Compare the output signal of the electromagnetic sensor processed by the adaptive filter with the original signal to calculate the residual noise level; S204: When the residual noise level exceeds a preset threshold, S201 is repeatedly performed until the residual noise is effectively suppressed.

[0006] Furthermore, the abnormal data recognition model is constructed using random forest, and the specific construction steps are: S301: Collect electromagnetic sensor data under normal working conditions as a training set ; S302: training set Bootstrap sampling is performed on the data in to obtain Sample Set ; S303: For each sample set Both use the random forest algorithm to train a decision tree classifier ; S304: The decision tree classifier The integration is a random forest model, i.e., an abnormal data identification model; S305: For newly input electromagnetic sensor data , use the abnormal data recognition model to perform abnormal monitoring: calculate In each decision tree classifier The output in ; Calculate the average output value: ; in, is the predicted value of the abnormal data identification model, For the training set The number of samples, For the The predicted value output by a decision tree classifier; if If it is less than the preset threshold, is abnormal data, otherwise it is normal data.

[0007] Furthermore, the data measured by the electromagnetic sensor is analyzed to adjust the working parameters of the electromagnetic sensor, and the statistical characteristics of the measured data, including the mean and variance, are calculated. The statistical characteristics are used as optimization targets, and a parameter adjustment model is used to adjust the working parameters of the electromagnetic sensor.

[0008] Furthermore, the parameter adjustment model is constructed based on the gradient descent algorithm, and the specific calculation formula is: ; is the mean square error loss function, which is used to represent the difference between the measured value and the true value of the electromagnetic sensor. Specifically: ; is the adaptive weight coefficient used to balance the loss function and the regularization term The weights are: ; is the L2 regularization term, used to constrain the parameters The range of change is to avoid overfitting due to large parameter changes, specifically: ; in, To optimize the objective function, we solve ,get The smallest hour ; is the value of the electromagnetic sensor parameter that needs to be updated; The number of samples for the electromagnetic sensor measurement data; For the The standard value of samples; For parameter adjustment model The predicted value of samples; For the The predicted value of samples; the number of iterations to adjust the model for the parameters, Used to adjust the speed of weight changes; are the initial electromagnetic sensor working parameters.

[0009] Furthermore, a reference signal source and a program-controlled switching circuit are built into the electromagnetic sensor; The reference signal source is used to generate a stable reference signal; The switching circuit is used to automatically switch the working mode of the electromagnetic sensor, switching the normal working mode to the reference signal source or switching the reference signal source to the normal working mode.

[0010] Furthermore, the electromagnetic sensor is dynamically calibrated regularly by the built-in reference signal source, specifically comprising the following steps: S501: regularly switching the working mode of the electromagnetic sensor, connecting the electromagnetic sensor to the built-in reference signal source, and converting the electromagnetic sensor from the normal working mode to the calibration mode; S502: Under the premise that the output signal of the electromagnetic sensor and the output signal of the reference signal source are synchronously collected, compare the output signal of the electromagnetic sensor and the output signal of the reference signal source at the same time, and calculate the error value between the two. The error value can be calculated using absolute error, relative error and root mean square error; S503: Calculating a calibration coefficient according to the error value, and updating a calibration parameter of the electromagnetic sensor; S504: Switch the electromagnetic sensor from the calibration mode back to the normal operating mode, and continue to perform measurement using the updated calibration parameters.

[0011] Furthermore, the accuracy optimization method also includes deploying a plurality of the electromagnetic sensors when using the electromagnetic sensor to monitor data, and the cloud platform uses wireless communication to obtain the monitoring data of the plurality of the electromagnetic sensors, the cloud platform remotely monitors and dynamically calibrates the electromagnetic sensors, and uses a data fusion algorithm to fuse the monitoring data, the cloud platform calculates high-precision measurement data, and dynamically calibrates the plurality of the electromagnetic sensors.

[0012] Furthermore, the cloud platform remotely monitors and dynamically calibrates the electromagnetic sensor, specifically including the following steps: S601: The electromagnetic sensor uploads the measurement data and performance parameters to the cloud platform via wireless communication; S602: The cloud platform receives and stores monitoring data from a plurality of electromagnetic sensors, and establishes a historical record of electromagnetic sensor performance; S603: the cloud platform analyzes the monitoring data of the electromagnetic sensors, and monitors the working status and measurement accuracy of each of the electromagnetic sensors in real time, wherein the monitoring data of the electromagnetic sensors is analyzed by statistically analyzing the average value, variance and fluctuation index of the monitoring data; S604: When it is found that the performance indicator of the electromagnetic sensor is abnormal, the cloud platform determines whether the electromagnetic sensor needs to be dynamically calibrated; S605: For a situation where the electromagnetic sensor cannot be solved by its own dynamic calibration, the cloud platform generates optimized calibration parameters; S606: The cloud platform sends a calibration instruction to the corresponding electromagnetic sensor, triggering the electromagnetic sensor to execute a dynamic calibration process; S607: After the electromagnetic sensor completes the dynamic calibration, the calibration result is fed back to the cloud platform, and the cloud platform records the calibration history.

[0013] Furthermore, the cloud platform calculates high-precision measurement data specifically including the following steps: S611: The cloud platform receives measurement data from the plurality of electromagnetic sensors, and performs time synchronization and data alignment; S612: using a data fusion algorithm to fuse the measurement data of the plurality of electromagnetic sensors; S613: Outputting the fused high-precision measurement result as the final measurement output; S614: The cloud platform feeds back the final measurement output to each of the electromagnetic sensors as a basis for optimizing the performance of the electromagnetic sensors.

[0014] The beneficial effects of the present invention are: 1. The accuracy optimization method of a single electromagnetic sensor of the present invention systematically solves the problems of electromagnetic sensor measurement accuracy and stability from multiple aspects such as signal processing, noise suppression, abnormality elimination, parameter optimization and dynamic calibration; 2. The present invention periodically switches the working mode of the electromagnetic sensor and compares the reference signal and the measurement signal, automatically calculates the calibration coefficient and updates the calibration parameters, so that the electromagnetic sensor can realize real-time dynamic calibration of the electromagnetic sensor without manual intervention, effectively improving the long-term stability of the measurement accuracy; 3. The present invention ensures the continuous optimization of the performance of a single sensor by taking the dynamic calibration function of the electromagnetic sensor itself as the basis, while the cloud platform centrally manages multiple sensors through remote monitoring and dynamic calibration functions, performs overall optimization and special case processing, and the two complement each other to further improve the measurement accuracy of the electromagnetic sensor. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flow chart of the method of the present invention; Figure 2 A flow chart of dynamically adjusting filter parameters of the present invention; Figure 3 A flow chart of the dynamic calibration of the electromagnetic sensor according to the present invention; Figure 4 A flow chart of remote monitoring and dynamic calibration of electromagnetic sensors according to the present invention; Figure 5 The flowchart of the cloud platform of the present invention for calculating high-precision measurement data. DETAILED DESCRIPTION

[0016] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. Embodiment 1

[0017] A precision optimization method for electromagnetic sensors, as shown in the attached Figure 1 As shown, the following steps are included: S100: Acquire the original measurement signal of the electromagnetic sensor, and pre-process the original measurement signal.

[0018] S200: Use an adaptive noise suppression algorithm to dynamically adjust adaptive filter parameters to suppress environmental noise.

[0019] S300: By collecting historical data, an abnormal data identification model is established, and the abnormal data is monitored and eliminated in real time through the abnormal data identification model.

[0020] S400: Analyze the electromagnetic sensor data and adjust the sensor working parameters.

[0021] S500: Use the built-in reference signal source to dynamically calibrate the electromagnetic sensor regularly and update the calibration coefficients.

[0022] In this embodiment, the original measurement signal of the electromagnetic sensor is obtained, and the original measurement signal is preprocessed, and at the same time, the external electromagnetic environment parameters of the electromagnetic sensor are obtained, including electromagnetic field strength and frequency.

[0023] In this embodiment, an adaptive noise suppression algorithm is used to dynamically adjust adaptive filter parameters to suppress environmental noise.

[0024] Specifically, as attached Figure 2 As shown, the specific steps of using the adaptive noise suppression algorithm to dynamically adjust the filter parameters to suppress environmental noise are: S201: real-time acquisition of environmental noise data in the working environment of the electromagnetic sensor, and spectrum analysis of the noise signal using fast Fourier transform (FFT) to obtain frequency characteristics of the noise; S202: According to the noise frequency characteristics, the least mean square LMS algorithm is used to adjust the parameters of the adaptive filter, including the cutoff frequency and the order, etc.; S203: Compare the output signal of the electromagnetic sensor processed by the adaptive filter with the original signal to calculate the residual noise level; S204: When the residual noise level exceeds a preset threshold, S201 is repeatedly performed until the residual noise is effectively suppressed.

[0025] In this embodiment, historical data is collected to establish an abnormal data recognition model, and the abnormal data is monitored and eliminated in real time through the abnormal data recognition model.

[0026] Specifically, the abnormal data recognition model is constructed using random forest, and the specific construction steps are: S301: Collecting electromagnetic sensor monitoring data under normal working conditions as a training set ; Among normal samples, some monitoring data under abnormal conditions are artificially marked or simulated as abnormal samples; S302: training set Bootstrap sampling is performed on the data to obtain Sample Set ; S303: For each sample set Both use the random forest algorithm to train a decision tree classifier ; When building a decision tree, for each node, randomly select features; for the selected features, calculate the Gini index, and select the best features for classification; S304: Decision tree classifier The integration is a random forest model, i.e., an abnormal data identification model; S305: For newly input electromagnetic sensor data , use the abnormal data recognition model for anomaly monitoring: calculate In each decision tree classifier The output in ; Calculate the average output value: ; in, The predicted value of the model for identifying abnormal data, is the number of training set samples, For the The predicted value output by a decision tree classifier; if If it is less than the preset threshold, is abnormal data, otherwise it is normal data.

[0027] When constructing an abnormal data recognition model, multiple decision trees are constructed through Bootstrap sampling and random feature selection, which can prevent a single decision tree from overfitting. Multiple decision trees are then combined to form a random forest, which improves the generalization ability and robustness of the abnormal data recognition model.

[0028] In this embodiment, the monitoring data of the electromagnetic sensor is analyzed and the working parameters of the electromagnetic sensor are adjusted.

[0029] The electromagnetic sensor data is analyzed to adjust the sensor working parameters. The statistical characteristics of the electromagnetic sensor, including the mean and variance, are calculated, and the statistical characteristics are used as the optimization target. The parameter adjustment model is used to adjust the sensor working parameters.

[0030] Specifically, the parameter adjustment model is constructed based on the gradient descent algorithm, and the specific calculation formula is: ; Used to indicate the measured value of electromagnetic sensor and predicted values The mean square error between the two reflects the current parameters The following parameters adjust the prediction error of the model, specifically: ; is the adaptive weight coefficient, as the number of iterations increases Dynamically changes with the increase of, used to balance the loss function and the regularization term The weights are: ; is the L2 regularization term, used to constrain the parameters The range of change is to avoid overfitting due to large parameter changes, specifically: ; in, To optimize the objective function, we solve ,get The smallest hour ; is the value of the electromagnetic sensor parameter that needs to be updated; The number of samples for the electromagnetic sensor measurement data; For the The standard value of samples; For parameter adjustment model The predicted value of samples; For the The predicted value of samples; is the number of iterations for parameter tuning of the model, Used to adjust the speed of weight changes; are the initial working parameters of the electromagnetic sensor.

[0031] Electromagnetic sensor adjusts the parameters of the model according to the parameters optimized To adjust its own working parameters, so as to improve the measurement accuracy, in the parameters The electromagnetic sensor includes key parameters such as gain coefficient and bias voltage. These key parameters are configured by setting the corresponding register values ​​in the electromagnetic sensor chip. The key parameters in the paper are applied to the electromagnetic sensor to improve the output data quality of the electromagnetic sensor.

[0032] In this embodiment, a built-in reference signal source is used to periodically perform dynamic calibration on the electromagnetic sensor and update the calibration coefficient.

[0033] Specifically, a reference signal source and a program-controlled switching circuit are built into the electromagnetic sensor; The reference signal source dynamically adjusts the generated calibration signal according to the working state of the electromagnetic sensor and environmental changes to generate a stable reference signal; by comparing the reference signal with the measurement signal of the sensor itself, the calibration coefficient is accurately calculated to improve the accuracy of the calibration.

[0034] The switching circuit is used to automatically switch the working mode of the electromagnetic sensor, switching the normal working mode to the reference signal source or switching the reference signal source to the normal working mode; by periodically starting the switching circuit, the working mode of the electromagnetic sensor is switched so that it is connected to the built-in reference signal source for calibration, thereby realizing dynamic real-time calibration of the electromagnetic sensor, making the calibration of the electromagnetic sensor no longer dependent on human intervention, and improving the degree of automation.

[0035] Specifically, as attached Figure 3 As shown in the figure, the dynamic calibration of the electromagnetic sensor periodically through the built-in signal source specifically includes the following steps: S501: regularly switching the working mode of the electromagnetic sensor, connecting the electromagnetic sensor to the built-in reference signal source, and converting the electromagnetic sensor from the normal working mode to the calibration mode; S502: Under the premise that the output signal of the electromagnetic sensor and the output signal of the reference signal source are synchronously collected, that is, the time synchronization of the signals is ensured, the difference between the output signal of the electromagnetic sensor and the output signal of the reference signal source is compared, and the error value between the two is calculated. The error value can be calculated by absolute error, relative error and root mean square error. By calculating the error value, it can be judged whether the measurement accuracy of the electromagnetic sensor meets the requirements; if the output signal of the electromagnetic sensor and the output signal of the reference signal source are of different types, the signal is converted into a digital signal so that they can be compared in the same domain; S503: Calculating a calibration coefficient according to the error value to adjust the measurement result of the electromagnetic sensor so that the measurement result is closer to the standard value of the reference signal source, applying the calculated calibration coefficient to the electromagnetic sensor, updating its working parameters, so as to improve the measurement accuracy of the electromagnetic sensor; S504: Switch the working mode of the electromagnetic sensor from the calibration mode back to the normal working mode, and continue to measure using the updated calibration parameters.

[0036] By regularly switching the working mode of the electromagnetic sensor and comparing the reference signal and the measurement signal, the calibration coefficient is automatically calculated and the calibration parameters are updated, so that the electromagnetic sensor can be calibrated in real time and dynamically without human intervention, effectively improving the long-term stability of the measurement accuracy. Embodiment 2

[0037] In this embodiment, the accuracy optimization method also includes, when using electromagnetic sensors to monitor data, deploying multiple electromagnetic sensors, using wireless communication to obtain the monitoring data of the multiple electromagnetic sensors by the cloud platform, remotely monitoring and dynamically calibrating the electromagnetic sensors, and using a data fusion algorithm to fuse the monitoring data, calculate high-precision measurement data, and dynamically calibrate the multiple electromagnetic sensors. The data fusion algorithm can use methods such as weighted averaging and Kalman filtering.

[0038] When deploying multiple electromagnetic sensors, they are optimized according to the geometric shape and size of the target, including circular and rectangular array layouts. At the same time, the relative position and spacing between the electromagnetic sensors in the array are optimized according to the characteristics of the target, including uniform distribution around the target or dense arrangement along a specific direction. In addition to electromagnetic sensors, other types of auxiliary sensors can also be deployed, such as optical sensors and acoustic sensors, etc. Through data fusion algorithms, the measurement data of different types of sensors can be fused to further improve the overall measurement accuracy.

[0039] Specifically, as attached Figure 4 As shown in the figure, the cloud platform performs remote monitoring and dynamic calibration of electromagnetic sensors, including the following steps: S601: The electromagnetic sensor regularly uploads measurement data and performance parameters to the cloud platform via wireless communication.

[0040] S602: The cloud platform receives and stores data from multiple electromagnetic sensors and establishes a historical record of the electromagnetic sensor performance.

[0041] S603: The cloud platform analyzes the sensor data and monitors the working status and measurement accuracy of each sensor in real time. The sensor data is analyzed by using indicators such as the mean, variance and fluctuation index of the statistical sensor data.

[0042] S604: When the cloud platform finds that the performance indicator of a certain electromagnetic sensor is abnormal, the cloud platform triggers an alarm and records the abnormal situation. At the same time, the cloud platform determines whether dynamic calibration is required based on the analysis results of the electromagnetic sensor.

[0043] S605: For situations where the electromagnetic sensor cannot be resolved through its own dynamic calibration, the cloud platform calculates more optimized calibration parameters or strategies based on the analysis results of the sensor data and generates customized calibration instructions.

[0044] S606: The cloud platform sends the calibration instruction to the corresponding electromagnetic sensor through wireless communication, triggering the electromagnetic sensor to execute the dynamic calibration process.

[0045] S607: The cloud platform monitors the calibration process of the electromagnetic sensor in real time. When the electromagnetic sensor completes the dynamic calibration, the electromagnetic sensor feeds back the calibration result to the cloud platform, and the cloud platform records the calibration history of the electromagnetic sensor and the changes in the calibration parameters. The cloud platform also provides a user interface to display the calibration status and historical data of the electromagnetic sensor for users to view.

[0046] The cloud platform centrally monitors the performance of multiple sensors and detects overall abnormalities. Based on the analysis results of a large amount of electromagnetic sensor data, the cloud platform provides more optimized calibration strategies or parameters. For special situations where some electromagnetic sensors cannot be solved through dynamic calibration, the cloud platform can issue customized calibration instructions and record the global calibration history, providing a basis for performance management and optimization of electromagnetic sensors.

[0047] Specifically, as attached Figure 5 As shown, the cloud platform calculates high-precision measurement data specifically including the following steps: S611: The cloud platform receives measurement data of the same target from multiple electromagnetic sensors, and performs pre-processing such as time synchronization and data alignment on the data; S612: using a data fusion algorithm to perform fusion processing on multiple sensor data; S613: Outputting the fused high-precision measurement result as the final measurement output; S614: The cloud platform feeds back the fusion result, i.e., the final measurement output, to all electromagnetic sensors as a basis for performance optimization.

[0048] By taking the dynamic calibration function of the electromagnetic sensor itself as the basis, the continuous optimization of the performance of a single sensor is ensured, while the cloud platform centrally manages multiple sensors through remote monitoring and dynamic calibration functions, performs overall optimization and handles special situations. The two complement each other and further improve the measurement accuracy of the electromagnetic sensor.

[0049] The above-described embodiments only express the preferred implementation modes of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for those of ordinary skill in the art, several modifications, improvements and substitutions can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A method for optimizing the accuracy of an electromagnetic sensor, characterized in that: The following steps are included: S100: acquiring an original measurement signal of an electromagnetic sensor, and preprocessing the original measurement signal; S200: Uses adaptive noise suppression algorithm to dynamically adjust adaptive filter parameters to suppress environmental noise; S300: Establishing an abnormal data recognition model by collecting historical measurement data, and performing real-time monitoring and elimination of abnormal data through the abnormal data recognition model; S400: Analyze the electromagnetic sensor data and adjust the working parameters of the electromagnetic sensor; S500: using the built-in reference signal source, regularly and dynamically calibrating the electromagnetic sensor and updating the calibration coefficient.

2. The method for optimizing the accuracy of an electromagnetic sensor according to claim 1, characterized in that: The specific steps of using the adaptive noise suppression algorithm to dynamically adjust the adaptive filter parameters to suppress environmental noise are: S201: real-time collection of environmental noise data in the working environment of the electromagnetic sensor, and obtaining frequency characteristics of the noise by spectrum analysis; S202: adjusting the parameters of the adaptive filter, including the cutoff frequency and the order, using the least mean square LMS algorithm according to the frequency characteristics of the noise; S203: Compare the output signal of the electromagnetic sensor processed by the adaptive filter with the original signal to calculate the residual noise level; S204: When the residual noise level exceeds a preset threshold, S201 is repeatedly performed until the residual noise is effectively suppressed.

3. The method for optimizing the accuracy of an electromagnetic sensor according to claim 2, characterized in that: The abnormal data recognition model is constructed using random forest, and the specific construction steps are as follows: S301: Collect electromagnetic sensor data under normal working conditions as a training set ; S302: training set Bootstrap sampling is performed on the data in to obtain Sample Set ; S303: For each sample set Both use the random forest algorithm to train a decision tree classifier ; S304: The decision tree classifier The integration is a random forest model, i.e., an abnormal data identification model; S305: For newly input electromagnetic sensor data , use the abnormal data recognition model to perform abnormal monitoring: calculate In each decision tree classifier The output in ; Calculate the average output value: ； in, is the predicted value of the abnormal data identification model, For the training set The number of samples, For the The predicted value output by a decision tree classifier; if If it is less than the preset threshold, is abnormal data, otherwise it is normal data.

4. The method for optimizing the accuracy of an electromagnetic sensor according to claim 3, characterized in that: The data measured by the electromagnetic sensor are analyzed to adjust the working parameters of the electromagnetic sensor, and the statistical characteristics of the measured data, including the mean and variance, are calculated. The statistical characteristics are used as optimization targets, and the working parameters of the electromagnetic sensor are adjusted using a parameter adjustment model.

5. The method for optimizing the accuracy of an electromagnetic sensor according to claim 4, characterized in that: The parameter adjustment model is constructed based on the gradient descent algorithm, and the specific calculation formula is: ； is the mean square error loss function, which is used to represent the difference between the measured value and the true value of the electromagnetic sensor. Specifically: ； is the adaptive weight coefficient used to balance the loss function and the regularization term The weights are: ； is the L2 regularization term, used to constrain the parameters The range of change is to avoid overfitting due to large parameter changes, specifically: ； in, To optimize the objective function, we solve ,get The smallest hour ; is the value of the electromagnetic sensor parameter that needs to be updated; The number of samples for electromagnetic sensor measurement data; For the The standard value of samples; For parameter adjustment model The predicted value of samples; For the The predicted value of samples; the number of iterations to adjust the model for the parameters, Used to adjust the speed of weight changes; are the initial electromagnetic sensor working parameters.

6. The method for optimizing the accuracy of an electromagnetic sensor according to claim 5, characterized in that: A reference signal source and a program-controlled switching circuit are built into the electromagnetic sensor; The reference signal source is used to generate a stable reference signal; The switching circuit is used to automatically switch the working mode of the electromagnetic sensor, switching the normal working mode to the reference signal source or switching the reference signal source to the normal working mode.

7. The method for optimizing the accuracy of an electromagnetic sensor according to claim 6, characterized in that: The electromagnetic sensor is regularly dynamically calibrated by the built-in reference signal source, specifically comprising the following steps: S501: regularly switching the working mode of the electromagnetic sensor, connecting the electromagnetic sensor to the built-in reference signal source, and converting the electromagnetic sensor from the normal working mode to the calibration mode; S502: Under the premise that the output signal of the electromagnetic sensor and the output signal of the reference signal source are synchronously collected, compare the output signal of the electromagnetic sensor and the output signal of the reference signal source at the same time, and calculate the error value between the two. The error value can be calculated using absolute error, relative error and root mean square error; S503: Calculating a calibration coefficient according to the error value, and updating a calibration parameter of the electromagnetic sensor; S504: Switch the electromagnetic sensor from the calibration mode back to the normal operating mode, and continue to perform measurement using the updated calibration parameters.

8. The method for optimizing the accuracy of an electromagnetic sensor according to claim 7, characterized in that: The accuracy optimization method also includes, when using the electromagnetic sensor to monitor data, deploying multiple electromagnetic sensors, and having a cloud platform use wireless communication to obtain the monitoring data of the multiple electromagnetic sensors, the cloud platform remotely monitors and dynamically calibrates the electromagnetic sensors, and uses a data fusion algorithm to fuse the monitoring data, the cloud platform calculates high-precision measurement data, and dynamically calibrates the multiple electromagnetic sensors.

9. The method for optimizing the accuracy of an electromagnetic sensor according to claim 8, characterized in that: The cloud platform performs remote monitoring and dynamic calibration of electromagnetic sensors, specifically including the following steps: S601: The electromagnetic sensor uploads the measurement data and performance parameters to the cloud platform via wireless communication; S602: The cloud platform receives and stores monitoring data from a plurality of electromagnetic sensors, and establishes a historical record of electromagnetic sensor performance; S603: the cloud platform analyzes the monitoring data of the electromagnetic sensors, and monitors the working status and measurement accuracy of each of the electromagnetic sensors in real time, wherein the monitoring data of the electromagnetic sensors is analyzed by statistically analyzing the average value, variance and fluctuation index of the monitoring data; S604: When it is found that the performance indicator of the electromagnetic sensor is abnormal, the cloud platform determines whether the electromagnetic sensor needs to be dynamically calibrated; S605: For a situation where the electromagnetic sensor cannot be solved by its own dynamic calibration, the cloud platform generates optimized calibration parameters; S606: The cloud platform sends a calibration instruction to the corresponding electromagnetic sensor, triggering the electromagnetic sensor to execute a dynamic calibration process; S607: After the electromagnetic sensor completes the dynamic calibration, the calibration result is fed back to the cloud platform, and the cloud platform records the calibration history.

10. The method for optimizing the accuracy of an electromagnetic sensor according to claim 9, characterized in that: The cloud platform calculates high-precision measurement data specifically including the following steps: S611: The cloud platform receives measurement data from the plurality of electromagnetic sensors, and performs time synchronization and data alignment; S612: using a data fusion algorithm to fuse the measurement data of the plurality of electromagnetic sensors; S613: Outputting the fused high-precision measurement result as the final measurement output; S614: The cloud platform feeds back the final measurement output to each of the electromagnetic sensors as a basis for optimizing the performance of the electromagnetic sensors.

Images