High-precision electrometer based on intelligent algorithm and signal processing method thereof

Abstract

The invention relates to the technical field of electrometer measurement, and particularly discloses a high-precision electrometer based on an intelligent algorithm and a signal processing method thereof. The high-precision electrometer based on the intelligent algorithm comprises a capacitive sensor module, and the capacitive sensor module obtains an electrostatic signal and converts the electrostatic signal into a digital signal. The preprocessing module carries out noise suppression processing on the digital signals, external interference is removed, and the stability of the signals is improved. And the signal processing module performs feature extraction and analysis on the corrected signal by using a deep learning algorithm, so that the measurement precision is further improved. The correction output module obtains an error correction value through historical data fitting, calculates a final static value in combination with a static predicted value, and ensures the accuracy and reliability of a measurement result.

Description

Technical Field

[0001] This application relates to the technical field of electrometer measurement, and particularly to a high-precision electrometer based on intelligent algorithms and its signal processing method. Background Art

[0002] With the continuous progress of modern technology, the importance of precision measurement has become increasingly prominent in various fields. Especially in industries such as industry, scientific research, and medical care, the measurement and control of static electricity have become very important topics. As a key measurement instrument, electrometers are widely used in these fields to detect and monitor the generation, distribution, and dissipation of static electricity. However, existing electrometers face some challenges in practical applications and need to be further improved and optimized.

[0003] Existing electrometers usually face the following problems: insufficient measurement accuracy, large environmental noise interference, slow response speed, etc. These problems are particularly obvious under high-precision measurement requirements, and the performance of traditional electrometers often fails to meet application needs. Specifically, existing electrometers have limitations in measurement accuracy. Especially in complex environments, measurement results are easily affected by external noise, resulting in unstable and inaccurate data. In addition, traditional electrometers have a slow response speed and cannot capture the rapid changes of static electricity in a timely manner, which is a prominent defect in dynamic measurement environments.

[0004] Existing electrometers also have deficiencies in adaptability and long-term stability. Under different environmental conditions, such as places with large temperature and humidity changes, the measurement accuracy of traditional electrometers will significantly decrease. In addition, the long-term use of the device will cause drift in measurement results, requiring frequent calibration, which increases the maintenance cost and the difficulty of use. These problems limit the widespread use of electrometers in application scenarios with high-precision and high-reliability requirements, and there is an urgent need for a new technology that can effectively solve these problems. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a high-precision electrometer based on intelligent algorithms and its signal processing method to solve at least one of the above technical problems.

[0006] To achieve the above object, in a first aspect, the present invention provides a high-precision electrometer based on intelligent algorithms, including:

[0007] A capacitance sensor module for acquiring static electricity signals;

[0008] A signal conversion circuit for converting the static electricity signal into a digital signal;

[0009] A microprocessor for outputting the final static electricity value by using a deep learning algorithm for the digital signal;

[0010] In the microprocessor, it includes:

[0011] A preprocessing module for performing noise suppression processing on the digital signal using a filtering algorithm to obtain a corrected signal;

[0012] A signal processing module for extracting features from the corrected signal to obtain a number of signal features, and using the number of signal features as the input of an intelligent algorithm to obtain an electrostatic prediction value;

[0013] A correction output module for obtaining historical electrostatic data, fitting the historical electrostatic data to obtain an error correction value, and calculating a final electrostatic value based on the error correction value and the electrostatic prediction value.

[0014] In a second aspect, the present invention provides a signal processing method for a high-precision electrometer based on an intelligent algorithm, including the following steps:

[0015] Performing noise suppression processing on the digital signal using a filtering algorithm to obtain a corrected signal;

[0016] Extracting features from the corrected signal to obtain a number of signal features, and using the number of signal features as the input of an intelligent algorithm to obtain an electrostatic prediction value;

[0017] Obtaining historical electrostatic data, fitting the historical electrostatic data to obtain an error correction value, and calculating a final electrostatic value based on the error correction value and the electrostatic prediction value.

[0018] The above technical solutions have the following beneficial technical effects:

[0019] The capacitance sensor module can capture electrostatic signals and convert them into analog signals to ensure accurate signal capture. Through a highly sensitive capacitance sensor, it can effectively detect charge changes in the air and improve the measurement accuracy.

[0020] The signal conversion circuit converts the analog signal into a digital signal to ensure that the signal is not lost during transmission and provides high-quality data for subsequent digital signal processing.

[0021] The preprocessing module performs noise suppression processing on the digital signal to remove noise introduced by electromagnetic interference or other external factors, thereby reducing the processing burden of subsequent filters. Through technologies such as Kalman filtering, it can effectively suppress environmental noise interference and improve the stability of the signal.

[0022] The signal processing module uses a deep learning algorithm to process the corrected signal, extracts signal features, further enhances the signal recognition ability, and corrects errors caused by factors such as equipment deviation. The deep learning algorithm can automatically identify and optimize data, thereby further improving the measurement accuracy.

[0023] The correction output module obtains historical static electricity data, performs fitting to obtain an error correction value, calculates the final static electricity value based on the error correction value and the static electricity prediction value, and ensures the accuracy and stability of the measurement result. Through the fitting of historical data, the measurement error can be effectively reduced and the measurement reliability can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings are used to better understand the present invention and do not constitute

[0025] an improper limitation on the present invention. Among them:

[0026] Figure 1 is a structural block diagram of a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0027] Figure 2 is a schematic structural diagram of a signal conversion circuit in a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0028] Figure 3 is a schematic structural diagram of a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0029] Figure 4 is a structural block diagram of a preprocessing module in a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0030] Figure 5 is a structural block diagram of a signal processing module in a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0031] Figure 6 is a structural block diagram of a correction output module in a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0032] Figure 7 is a flowchart of a signal processing method of a high-precision electrometer based on an intelligent algorithm according to an embodiment of the present invention;

[0033] Figure 8 is a schematic structural diagram of a computer system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] The following describes exemplary embodiments of the present invention with reference to the drawings. Various details of the embodiments of the present invention are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description below omits descriptions of well-known functions and structures.

[0035] Embodiment 1

[0036] As Figure 1 shown, this embodiment provides a high-precision electrometer based on an intelligent algorithm, including:

[0037] A capacitance sensor module for acquiring electrostatic signals;

[0038] A signal conversion circuit for converting the electrostatic signal into a digital signal;

[0039] A microprocessor for outputting a final electrostatic value by using a deep learning algorithm for the digital signal;

[0040] In the microprocessor, it includes:

[0041] A preprocessing module for performing noise suppression processing on the digital signal by using a filtering algorithm to obtain a corrected signal;

[0042] A signal processing module for extracting features from the corrected signal to obtain a number of signal features, and using the number of signal features as the input of an intelligent algorithm to obtain an electrostatic prediction value;

[0043] A correction output module for acquiring historical electrostatic data, fitting the historical electrostatic data to obtain an error correction value, and calculating a final electrostatic value according to the error correction value and the electrostatic prediction value.

[0044] Specifically, the capacitance sensor module is a high-sensitivity capacitance sensor, and the sensing electrode of the capacitance sensor module interacts with the surface of the object to be measured or the electric field in the measurement space,

[0045] so as to capture external electrostatic signals. The capacitance sensor module is arranged at the top of the electrometer and directly contacts or approaches the object to be measured or the static power source.

[0046] As Figure 2As shown, the signal conversion circuit includes an analog front-end circuit and an analog-to-digital converter. The analog front-end circuit is used to amplify the electrostatic signal to obtain an amplified signal, and the analog-to-digital converter is used to convert the amplified signal into a digital signal. The signal input terminal of the analog front-end circuit is connected to the signal output terminal of the capacitance sensor module, the signal output terminal of the analog front-end circuit is connected to the signal input terminal of the analog-to-digital converter, and the signal output terminal of the analog-to-digital converter is connected to the signal input terminal of the microprocessor through a serial communication interface. The analog front-end circuit includes components such as an operational amplifier, a filter, and a gain regulator, so as to ensure that the electrostatic signal is not lost during transmission and perform noise filtering processing according to the actual situation. The analog-to-digital converter is used to convert the analog signal output from the analog front-end circuit into a digital signal, which is convenient for subsequent digital signal processing. The serial communication interface is preferably an Inter-Integrated Circuit (I2C) or a Serial Peripheral Interface (SPI).

[0047] Specifically, as Figure 3 shown, a high-precision electrometer based on an intelligent algorithm further includes a display module, a communication module, a power supply module, and a housing. The signal input terminal of the display module is connected to the signal output terminal of the microprocessor, the signal input terminal of the communication module is connected to the signal output terminal of the microprocessor, the power supply module is used to supply power to the capacitance sensor module, the signal conversion circuit, the microprocessor, the display module, and the communication module, and the communication module is used to upload the final static electricity value to a preset terminal.

[0048] The display module displays the final static electricity value result to the user in real time through an LCD screen or an OLED screen. The display module presents the processed static electricity data (such as electric field strength, charge quantity, etc.) through a graphical interface, and the user can interact with the system through a touch screen or buttons to obtain real-time data and adjust parameters.

[0049] The communication module selects wired communication or wireless communication (such as Wi-Fi or Bluetooth). The communication module is used to upload the final static electricity value to a preset terminal (such as the cloud or a local computer). Data upload through the communication module facilitates subsequent data analysis, data storage, and subsequent use. When wireless communication is selected, data interaction is performed through an adapted protocol.

[0050] The power supply module includes a battery, a charging circuit, and a voltage regulator, so as to ensure that each module can work stably and normally. The power supply module and other modules are connected through a circuit board to provide stable power input.

[0051] The outer shell is set at the outermost layer. The outer shell is made of anti-static and impact-resistant materials to ensure that the high-precision electrometer based on intelligent algorithms is not disturbed by the external environment during use. All components are installed through standard interfaces and fixtures to ensure the stability and safety of each module. The capacitive sensor module is installed at the front end of the shell, and other modules are installed in reasonable positions inside the shell to ensure the compactness and ease of use of the electrometer.

[0052] Specifically, Figure 4 As shown, in the preprocessing module, it includes:

[0053] A preliminary processing unit, configured to remove low-frequency noise and invalid data from the digital signal by using a filtering algorithm to obtain first preprocessed data;

[0054] A Kalman prediction unit, configured to process the first preprocessed data using a Kalman filter algorithm to obtain a Kalman prediction value;

[0055] A correction unit, configured to calculate correction data according to the Kalman prediction value and the first preprocessed data;

[0056] The normalization unit is used to perform normalization processing on the correction data to obtain a correction signal.

[0057] Specifically, in the preliminary processing unit, statistical methods are used to remove invalid data, such as the Chauville criterion, the Dixon criterion, the Grubbs criterion and the Laida criterion, etc., to remove outliers in the signal, that is, burrs that are too large or too small, and add interpolation points, and the interpolation points are calculated by the average value or the median method. The low-frequency noise can be removed by filtering algorithms such as digital filters, wavelet transform methods, moving average filters, and median filter singular value decomposition denoising methods. The low-frequency noise refers to a noise signal with a frequency range between 20Hz and 200Hz. The wavelet transform method decomposes the signal by wavelet transform, removes the low-frequency noise, and then reconstructs it. The moving average filter calculates the moving average value of the digital signal, smoothes the digital signal, and removes the low-frequency noise. The median filter takes the median in the filter window as the estimated value of the current point, smoothes the digital signal, and removes the low-frequency noise. The singular value decomposition denoising method performs singular value decomposition on the signal matrix of the digital signal to remove smaller singular values, thereby achieving the purpose of noise reduction.

[0058] Specifically, in the Kalman prediction unit, the Kalman filter is first initialized to initialize the state estimate value and the error covariance matrix. During the initialization process, the microprocessor sets the initial values based on the preset working state. Assuming the initial state estimate is x0 and the initial error covariance is P0, these initial values will affect the performance of the Kalman filter and can be obtained through experimental data or system preset. Then, the dynamic model is described by the state transition equation, and the Kalman prediction value is obtained according to the dynamic model. The expression of the dynamic model is as follows:

[0059] x k = Ax k-1 + Bu k + w k ;

[0060] In the formula, x k is the Kalman prediction value, A is the state transition matrix, B is the control matrix, u k is the control input, w k is the process noise. The Kalman filter predicts the state at the current moment by using the dynamic model based on the estimated state at the previous moment.

[0061] Specifically, in the correction unit, it includes:

[0062] The Kalman gain calculation unit is used to calculate the Kalman gain according to the Kalman filtering algorithm;

[0063] The observation residual calculation unit is used to calculate the observation residual according to the Kalman prediction value and the first preprocessed data;

[0064] The correction term calculation unit is used to multiply the observation residual by the Kalman gain to obtain the correction term;

[0065] The corrected data calculation unit is used to add the Kalman prediction value at the previous moment and the correction term to obtain the corrected data.

[0066] In the Kalman gain calculation unit, the calculation formula of the Kalman gain is as follows:

[0067]

[0068] In the formula, P k-1 is the estimated error covariance at the previous moment, H k is the observation matrix, R is the observation noise covariance matrix, H k T is the transpose matrix of the observation matrix, and K k is the Kalman gain. The corrected data is calculated according to the Kalman gain and the Kalman prediction value. The calculation formula of the corrected data is as follows:

[0069] X k = x k-1 + K k (y k - H k x k-1 );

[0070] In the formula, X k represents the corrected data, x k-1 represents the Kalman prediction value at the previous moment, y k is the measured value (i.e., the first preprocessed data), (y k - H k x k-1 ) represents the observation residual, [K k (y k - H k x k-1 )] represents the correction term. The state at the current moment is estimated through the actual measured value and the predicted value, making the corrected data closer to the true signal value. Finally, the error covariance matrix P k in the Kalman filter is updated to evaluate the uncertainty of the current estimate. Through the update step, it is ensured that after each state estimate, the system can adaptively adjust the confidence level of the signal. The expression for the update is:

[0071] P k = (I - K k H k )P k-1 ;

[0072] Specifically, in the normalization unit, the corrected data is first normalized. The min - max normalization method is selected for the normalization process. The formula of the min - max normalization method is as follows:

[0073]

[0074] In the formula, X is the corrected data, X' is the corrected signal, X min is the minimum value of the corrected data, X max is the maximum value of the corrected data. Preferably, the corrected signal is segmented, and the continuous time - series signal is converted into signal segments with a fixed window size for subsequent processing.

[0075] Specifically, as Figure 5 shown, in the signal processing module, it includes:

[0076] A feature extraction unit, which is used to input the corrected signal into a convolutional neural network to obtain several signal features;

[0077] An identification unit for inputting a plurality of the signal features into a classifier to obtain probabilities corresponding to each signal type, and selecting the signal type with the maximum probability as the classification result;

[0078] A prediction unit for inputting a plurality of the signal features into a preset regression model to obtain an electrostatic prediction value;

[0079] A signal processing and output unit for outputting the classification result and the electrostatic prediction value.

[0080] Specifically, in this embodiment, the intelligent algorithm selects a deep learning algorithm.

[0081] Specifically, in the feature extraction unit, the convolutional neural network includes a convolutional layer, a pooling layer, and a fully connected layer. First, input the corrected signal into the convolutional layer, and the convolutional layer scans the input corrected signal through a plurality of one-dimensional convolutional kernels (for example, 3×1 kernels)

[0082] to extract local temporal features. The formula of the convolutional layer is as follows:

[0083] y = f(W * x + b);

[0084] where W is the convolutional kernel, x is the corrected signal, b is the bias, f() is a non-linear activation function (for example, ReLU), and y is the local temporal feature. Secondly, input the local temporal features into the pooling layer for dimensionality reduction. The pooling layer uses max pooling or average pooling to reduce the dimension of the local temporal features to obtain pooled features. Finally, input the pooled features into the fully connected layer, and the fully connected layer maps the pooled features to a signal feature as the input for subsequent classification or regression. The convolutional neural network automatically extracts the signal features of the corrected signal, avoiding manually designing features in traditional methods, and the convolutional neural network can learn local patterns from temporal signals and is applicable to electrostatic signals in different environments.

[0085] Specifically, in the identification unit, the classifier selects a Softmax classifier, and the formula of the Softmax classifier is as follows:

[0086]

[0087] where P(y i ) represents the probability of the category being i, and z i represents the activation value output by the network. The classifier selects the category with the highest probability as the classification result, automatically classifies different types of electrostatic signals through deep learning, improves the intelligence level of the system, and combines real-time signal analysis to achieve early warning of abnormal signals in the electrostatic monitoring scenario.

[0088] In the signal processing module of the high-precision electrometer in this embodiment, the signal types specifically include the following categories: normal signals, abnormal signals, stable fluctuation signals, error signals, and trend abnormal signals.

[0089] Normal signals indicate that the measurement results of the electrometer are within the normal range and are not affected by external interference or other abnormal factors. The signals are stable and have high precision, and are classified as normal signals. Such signals indicate that the device is working properly and no further intervention is required.

[0090] Abnormal signals indicate that there are problems with the measurement results of the electrometer, which are caused by external interference, device failures, or other reasons. Classified as abnormal signals, this indicates that the measurement results are abnormal and affect the normal operation of the device. Abnormal signals can be further subdivided into the following categories: noise interference, whose signals contain excessive noise, affecting the accuracy of the measurement. Excessive fluctuations, whose signal volatility is relatively large and exceeds the preset threshold, which is caused by external interference or unstable factors. Drift, whose measured value gradually deviates from the normal value over time, which is a long-term deviation caused by device aging or environmental changes.

[0091] Stable fluctuation signals indicate that there are certain fluctuations in the signals, but these fluctuations are regular and do not fall within the abnormal range. The fluctuations of the signals are caused by minor environmental changes or the operating state of the device, and no emergency intervention is required, but monitoring is still needed.

[0092] Error signals indicate that due to device failures or other hardware problems, the measurement results deviate severely from the normal range. Such signals usually show extremely large deviations or unpredictable behaviors, seriously affecting the accuracy of the electrometer. Once error signals are identified, emergency treatment is required, such as equipment maintenance

[0093] or recalibration.

[0094] Trend abnormal signals indicate that the signals show a continuous upward or downward trend over time. This is caused by sensor aging, long-term environmental changes, or other factors. Such signals are predicted through time series analysis and regression models, and can be marked as "trend abnormal" or "need calibration" to remind operators to regularly check the device or perform calibration to avoid long-term measurement errors.

[0095] In summary, through the automatic learning and classification of signal features, the Softmax classifier can intelligently judge whether the electrostatic signals are normal, and realize the timely early warning of abnormal signals during the monitoring process, enhancing the intelligent level of the system.

[0096] Specifically, in the prediction unit, the calculation formula of the regression model is as follows:

[0097] V predicted =W reg *F + b;

[0098] In the formula, V predicted is the electrostatic prediction value, F is the signal feature, and W reg is the weight of the regression model, and b is the bias term.

[0099] Specifically, the prediction unit obtains the electrostatic prediction value by inputting several of the signal features into a preset regression model, and labels the electrostatic prediction value with the classification result. Through automatic fitting by the regression model, the measurement error is reduced and the measurement accuracy is improved. When the correction signal fluctuates greatly, the drift is corrected through time series prediction, improving the measurement stability.

[0100] Specifically, as Figure 6 shown, the correction output module includes:

[0101] A historical data acquisition unit for acquiring historical electrostatic data;

[0102] An error calculation unit for calculating an error correction value according to the historical electrostatic data by using a fitting error algorithm;

[0103] An output unit for calculating a final electrostatic value according to the error correction value and the electrostatic prediction value.

[0104] Specifically, the historical electrostatic data includes historical prediction data and historical real data.

[0105] Specifically, in the error calculation unit, the fitting error algorithm includes calculating the mean squared error (MSE), the root mean squared error (RMSE), and the mean absolute error (MAE). The formula for the mean squared error is as follows:

[0106]

[0107] In the formula, y i is the historical real data, is the historical prediction data.

[0108] The formula for the root mean squared error is as follows:

[0109]

[0110] The formula for the mean absolute error is as follows:

[0111]

[0112] Specifically, the error correction value can be directly obtained by calculating the mean square error, root mean square error, or mean absolute error, or can be obtained by weighted calculation based on the mean square error, root mean square error, and mean absolute error.

[0113] Specifically, in the output unit, the formula for calculating the final static electricity value according to the static electricity prediction value and the error correction value is as follows:

[0114] V corrected = V predict + △V error ;

[0115] In the formula, V corrected is the final voltage value, and △V error is the correction value.

[0116] Specifically, in the correction output module, there is also a dynamic correction unit. The dynamic correction unit is used to optimize the error correction value by using an adaptive learning algorithm to obtain an error correction value. The error correction value can replace the error correction value to calculate the final voltage value. The error correction value dynamically corrects the error correction value through a Kalman filter or an adaptive learning algorithm, thereby improving long-term stability.

[0117] Specifically, in the high-precision electrometer based on an intelligent algorithm, there is also an environment sensor and a composite calibration module. The environment sensor is used to obtain the environmental data around the capacitance sensor module. The signal output end of the environment sensor is connected to the signal input end of the composite calibration module. The signal input end of the composite calibration module is also connected to the signal output end of the preprocessing module.

[0118] Specifically, the environmental data includes temperature data, humidity data, and time data. The environment sensor includes a temperature sensor and a humidity sensor. The humidity sensor is preferably a capacitive humidity sensor. The time data is a time drift parameter, and the time data is obtained by recording with a built-in clock. The environmental data can be processed in the composite calibration module. The composite calibration module can be set independently or can be set in the microprocessor.

[0119] Specifically, the composite calibration module is used to establish a compensation function according to the environmental data, and compensate the correction signal according to the compensation function to obtain the final static electricity value. The expression of the compensation function is as follows:

[0120] f(T, H, t, A);

[0121] Wherein, T is the temperature data, H is the humidity data, t is the time data, and A is the aging factor. This aging factor can be estimated through historical data or usage status, and the sensor output is corrected by gradually changing the compensation function. As the device usage time increases, the performance of the sensor will gradually decline, and the aging factor A can be dynamically updated based on long-term data. The aging factor can adjust the compensation function through a gradually increasing error, so that even when the device ages, the compensation can maintain accuracy.

[0122] In a further embodiment, the specific formula of the compensation function is:

[0123] f(T, H, t, A) = aT 2 + bH 3 + c(t·H)+ d(T·t)+ e + δ(A);

[0124] Wherein, a, b, c, and d are coefficients in the compensation function, used to adjust the influence degree of temperature, humidity, and time on the measurement result. e is a constant term, used to compensate for other errors or reference deviations unrelated to environmental factors, and ensure the accuracy of the model. δ(A) is a correction term related to the aging factor A. A is a dynamically changing factor, used to consider the performance decline of the sensor over time. This correction term can be a function, used to adjust the compensation result according to the aging degree of the sensor, so as to improve the measurement accuracy in long-term use.

[0125] Furthermore, the parameters or independent variables that can be considered in the compensation function include: air pressure, power supply voltage, and the initial deviation of the sensor. The compensation function includes several compensation terms. The compensation terms include one or more combinations of temperature compensation, humidity compensation, time compensation, air pressure compensation, power supply voltage compensation, and initial deviation compensation of the capacitive sensor module, and also include aging compensation. In high-precision measurements, the air pressure factor can be considered as a supplementary parameter, especially in an environment where the atmospheric pressure changes greatly. The air pressure data can be obtained through an air pressure sensor and used for compensation. The stability of the electrometer is closely related to the power supply voltage, and voltage fluctuations will directly affect the working stability of the sensor. Therefore, by monitoring the power supply voltage and using it as a compensation variable, the measurement accuracy of the electrometer under different voltage conditions can be improved. The initial deviation of the sensor is an important compensation factor, especially when there is a certain deviation at the time of factory. Adding the initial deviation parameter can help correct the basic offset of the electrometer and ensure the accuracy of each measurement. In an alternative embodiment, the compensation function formula is as follows:

[0126] f(T, H, t, A, P, V power , B init ) = aT 2 + bH 3+c(t·H)+d(T·t)+e+δ(A)+f(P)+g(V power )+h(B init );

[0127] Wherein, P represents the air pressure, and V power represents the power supply voltage. Voltage fluctuations will directly affect the stability of the electrometer. B init is the initial deviation of the capacitance sensor module (the factory deviation of the electrometer). a, b, c, d, e, f, g, and h are the coefficients in the compensation function, which are used to adjust the influence of the corresponding parameters on the electrostatic value. δ(A) is the compensation term of the aging factor, which is a correction term that changes with the increase of the sensor usage time.

[0128] Specifically, in the step of compensating the correction signal according to the compensation function to obtain the electrostatic value of environmental restoration, the calculation formula of the electrostatic value of environmental restoration is as follows:

[0129] V corrected = V raw + f(T, H, t, A);

[0130] In the formula, V raw is the correction signal, that is, the measured value. When calculating the electrostatic value of environmental restoration through environmental data, there is no need for a signal processing module and an error correction module. The correction signal is optimized through environmental data, so as to obtain the electrostatic value of environmental restoration. The method of correction through environmental data has full environmental adaptability, can compensate multiple environmental factors simultaneously, rather than a single parameter, and improves the measurement stability. Secondly, the real-time dynamic adjustment mechanism ensures that the system can automatically adapt to environmental changes without manual intervention. In addition, the time compensation parameter can correct the drift error during the long-term use of the device and extend the accurate measurement life of the device. Finally, the machine learning optimization enables the system to adaptively adjust, reduce errors, and improve the accuracy and reliability of long-term measurement. It is applicable to multiple high-precision application scenarios, such as precision laboratories (requiring extremely low error electrostatic measurement), industrial environmental monitoring (electrostatic monitoring in complex environments), medical devices (electrostatic sensors that need to work stably for a long time), and outdoor applications with large environmental changes (such as aerospace, weather stations, etc.). By integrating a multi-dimensional compensation mechanism, it is ensured that the electrometer maintains high precision in different environments and realizes true full-environment adaptive calibration.

[0131] Specifically, in the microprocessor, there is also an analysis module. The analysis module is used to analyze the final electrostatic value to obtain an analysis result, and adjust the parameters of the preprocessing module and the signal processing module according to the analysis result. For the preprocessing module, the correspondingly adjusted parameters include: filter parameters, signal gain, sampling rate. For the signal processing module, the correspondingly adjusted parameters include: Kalman gain, threshold setting, noise filter strength, signal smoothness.

[0132] Specifically, when analyzing the final static electricity value, it is analyzed from three dimensions, namely signal-to-noise ratio, signal volatility, and measurement error. The signal-to-noise ratio is the ratio of the signal to the noise. The higher the signal-to-noise ratio, the higher the quality of the signal. Conversely, noise suppression needs to be strengthened. The signal volatility represents the stability of the final static electricity value. When the signal volatility is greater than the preset threshold, it indicates that there is external interference, and an alarm should be issued in time to call personnel for inspection. The measurement error is judged by comparing the final static electricity value with the reference signal, so as to determine the accuracy of the final static electricity value.

[0133] Specifically, when adjusting the parameters of the preprocessing module and the signal processing module according to the analysis results, it is implemented by using a Kalman filter. The Kalman filter determines the weight between the predicted value and the actual measured value through the Kalman gain. In actual operation, the adjustment of the Kalman gain is very important, which directly affects the noise suppression effect and signal accuracy. The system dynamically adjusts the Kalman gain according to the signal quality evaluation results. Specifically, the better the signal quality, the smaller the Kalman gain can be set to reduce overcorrection; when the signal quality is poor, the Kalman gain will increase to enhance noise suppression.

[0134] Specifically, in the microprocessor, there is also a deep learning module. When the signal processing method of the high-precision electrometer based on the intelligent algorithm works, the deep learning module will be automatically optimized as the data volume increases and the measurement environment changes. In the real-time feedback mechanism, the system continuously inputs newly collected signal data, and the deep learning module will perform adaptive updates based on these data. Specifically, the deep learning module will train each batch of new data and update the weights of the neural network to enable it to better fit the current static electricity signal. This adaptive learning based on real-time data not only improves the accuracy of the model but also enables the system to gradually adapt to different working environments.

[0135] The update process can be achieved through online learning or incremental learning, allowing the deep learning module to continuously optimize by introducing new data without the need for retraining. Through adaptive learning, the deep learning module can automatically optimize according to real-time data and maintain a high measurement accuracy in a dynamically changing environment. After the system adjusts the Kalman filter gain and the deep learning module in real time, it is necessary to continuously evaluate the adjusted performance. The microprocessor regularly monitors the signal quality, the output stability of the system, and the error rate to ensure that after the algorithm is adjusted, the measurement results of the system tend to be optimal. If the evaluation results show that the measurement error is large or the output fluctuation is too large, the system will readjust the algorithm parameters and perform re-feedback to ensure that the system can provide accurate and stable measurement results in any environment.

[0136] Finally, after real-time feedback and adaptive adjustment processing, the system outputs the adjusted measurement results to the display module or storage device for the user to view. The results of real-time optimization will be transmitted to the cloud or remote server for data storage and analysis. Through remote feedback, the system can not only ensure the accuracy of current data but also provide data support and model optimization for future measurements.

[0137] Embodiment 2

[0138] As Figure 7 shown, this embodiment provides a signal processing method for a high-precision electrometer based on an intelligent algorithm, including the following steps:

[0139] S10: Perform noise suppression processing on the digital signal using a filtering algorithm to obtain a corrected signal;

[0140] S20: Extract features from the corrected signal to obtain several signal features, and use the several signal features as the input of the intelligent algorithm to obtain an electrostatic prediction value;

[0141] S30: Obtain historical electrostatic data, fit the historical electrostatic data to obtain an error correction value, and calculate the final electrostatic value based on the error correction value and the electrostatic prediction value.

[0142] Specifically, in the step S10, it includes the following steps:

[0143] S11: Use a filtering algorithm to remove low-frequency noise and invalid data in the digital signal to obtain first preprocessed data;

[0144] S12: Process the first preprocessed data using a Kalman filtering algorithm to obtain a Kalman prediction value;

[0145] S13: Calculate corrected data based on the Kalman prediction value and the first preprocessed data;

[0146] S14: Perform normalization processing on the corrected data to obtain a corrected signal.

[0147] Specifically, in the step S11, statistical methods are used to remove invalid noise, such as Chauvenet's criterion, Dixon's criterion, Grubbs' criterion, and the 3-sigma criterion, etc., to eliminate the outliers in the signal, that is, the over-large or over-small burr points, and appropriate interpolation points are added. The interpolation points are obtained by methods such as the average value or the median. To remove the low-frequency noise, methods such as digital filters, wavelet transform methods, moving average filters, median filters, and singular value decomposition noise reduction methods can be used. In the wavelet transform method, the signal is decomposed by wavelet transform to remove the low-frequency noise and then reconstructed. In the moving average filter, the digital signal is smoothed by calculating the moving average value of the digital signal to remove the low-frequency noise. In the median filter, the median value within the filter window is taken as the estimated value of the current point to smooth the digital signal and remove the low-frequency noise. In the singular value decomposition noise reduction method, the signal matrix of the digital signal is subjected to singular value decomposition, and the smaller singular values are removed to achieve the purpose of noise reduction.

[0148] Specifically, in the step S12, the Kalman filter is first initialized to initialize the state estimate value and the error covariance matrix. During the initialization process, the microprocessor sets the initial values based on the preset working state. Assuming the initial state estimate is x0 and the initial error covariance is P0, these initial values will affect the performance of the Kalman filter and can be obtained through experimental data or system preset. Then, the dynamic model is described by the state transition equation, and the Kalman prediction value is obtained according to the dynamic model. The expression of the dynamic model is as follows:

[0149] x k = Ax k-1 + Bu k + w k ;

[0150] In the formula, x k is the Kalman prediction value, A is the state transition matrix, B is the control matrix, u k is the control input, and w k is the process noise. The Kalman filter predicts the state at the current moment by using the dynamic model based on the estimated state at the previous moment.

[0151] Specifically, in the step S13, the Kalman gain of the Kalman filter is first calculated. The calculation formula of the Kalman gain is as follows:

[0152]

[0153] In the formula, P k-1 is the estimated error covariance at the previous moment, H k is the observation matrix, R is the observation noise covariance matrix, and H k Tis the transpose matrix of the observation matrix, and K k is the Kalman gain. The correction data is calculated according to the Kalman gain and the Kalman predicted value. The calculation formula of the correction data is as follows:

[0154] X k = x k-1 + K k (y k - H k x k-1 );

[0155] In the formula, X k represents the correction data, x k-1 represents the Kalman predicted value at the previous moment, and y k is the measured value (i.e., the first preprocessed data). The state at the current moment is estimated through the actual measured value and the predicted value, making the correction data closer to the true signal value. Finally, the error covariance matrix P k in the Kalman filter is updated to evaluate the uncertainty of the current estimate. By the update step, it is ensured that after each state estimate, the system can adaptively adjust the confidence of the signal. The expression of the update is:

[0156] P k = (I - K k H k )P k-1 ;

[0157] Specifically, in the step S14, the correction data is first normalized. The minimum-maximum normalization method is selected for the normalization process. The formula of the minimum-maximum normalization method is as follows:

[0158]

[0159] In the formula, X is the correction data, X' is the correction signal, X min is the minimum value of the correction data, and X max is the maximum value of the correction data. Preferably, the correction signal is segmented, and the continuous time series signal is converted into signal segments with a fixed window size for subsequent processing.

[0160] Specifically, in the step S20, the following steps are included:

[0161] S21: Input the correction signal into the convolutional neural network to obtain several signal features;

[0162] S22: Input several of the signal features into the classifier

[0163] to obtain the classification result;

[0164] S23: Input several of the said signal features into a preset regression model to obtain an electrostatic prediction value;

[0165] S24: A signal processing output unit, configured to output the classification result and the electrostatic prediction value.

[0166] Specifically, in the step S21, the convolutional neural network includes a convolutional layer, a pooling layer, and a fully connected layer. First, input the corrected signal into the convolutional layer. The convolutional layer scans the input corrected signal through several one-dimensional convolutional kernels (for example, 3×1 kernels) to extract local temporal features. The formula of the convolutional layer is as follows:

[0167] y = f(W * x + b);

[0168] In the formula, W is the convolutional kernel, x is the corrected signal, b is the bias, f() is a non-linear activation function (for example, ReLU), and y is the local temporal feature. Secondly, input the local temporal features into the pooling layer for dimensionality reduction. The pooling layer uses max pooling or average pooling to reduce the dimension of the local temporal features to obtain pooled features. Finally, input the pooled features into the fully connected layer. The fully connected layer maps the pooled features to a signal feature as the input for subsequent classification or regression. The convolutional neural network automatically extracts the signal features of the corrected signal, avoiding manually designing features in traditional methods, and the convolutional neural network can learn local patterns from temporal signals and is applicable to electrostatic signals in different environments.

[0169] Specifically, in the step S22, the classifier selects a Softmax classifier. The formula of the Softmax classifier is as follows:

[0170]

[0171] In the formula, P(y i ) represents the probability of the category being i, and z i represents the activation value output by the network. The classifier selects the category with the highest probability as the classification result, automatically classifies different types of electrostatic signals through deep learning, improves the intelligence level of the system, and combines real-time signal analysis to achieve early warning of abnormal signals in the electrostatic monitoring scenario.

[0172] Specifically, in the step S23, the calculation formula of the regression model is as follows:

[0173] V predicted = W reg * F + b;

[0174] In the formula, V predicted is the electrostatic prediction value, F is the signal feature, and W regw is the weight of the regression model, and b is the bias term.

[0175] Specifically, in the step S30, it specifically includes:

[0176] S31: Obtain historical static electricity data;

[0177] S32: Calculate an error correction value according to the historical static electricity data by using a fitting error algorithm;

[0178] S33: Calculate a final static electricity value according to the error correction value and the static electricity prediction value.

[0179] Specifically, in the historical static electricity data, it includes historical prediction data and historical real data.

[0180] Specifically, in the step S32, the fitting error algorithm includes calculating the mean square error (MSE), the root mean square error (RMSE), and the mean absolute error (MAE).

[0181] The calculation formula of the mean square error is as follows:

[0182]

[0183] In the formula, y i is the historical real data, is the historical prediction data.

[0184] The calculation formula of the root mean square error is as follows:

[0185]

[0186] The calculation formula of the mean absolute error is as follows:

[0187]

[0188] Specifically, the error correction value can be directly obtained by solving the mean square error, the root mean square error, or the mean absolute error, or can be obtained by weighted calculation according to the mean square error, the root mean square error, and the mean absolute error.

[0189] Specifically, in the step S33, the formula for calculating the final static electricity value according to the static electricity prediction value and the error correction value is as follows:

[0190] V corrected = V predict + △V error ;

[0191] In the formula, V corrected is the final voltage value, and △V error is the correction value.

[0192] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0193] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements a signal processing method of a high-precision electrometer based on an intelligent algorithm as described above.

[0194] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above method embodiments of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc. Of course, there are other ways of readable storage media, such as quantum memory, graphene memory, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0195] The present invention also provides an electronic device. The electronic device according to an embodiment of the present invention includes: one or more processors; a storage device for storing one or more programs, which when executed by the one or more processors, cause the one or more processors to implement a signal processing method of a high-precision electrometer based on an intelligent algorithm provided by the present invention.

[0196] Reference is now made to Figure 8 , which shows a schematic structural diagram of a computer system 800 suitable for use in implementing the electronic device according to an embodiment of the present invention. Figure 8 The illustrated electronic device is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of the present invention.

[0197] As Figure 8 shown, the computer system 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes according to the programs stored in the read-only memory (ROM) 802 or the programs loaded from the storage section 808 into the random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the computer system 800 are also stored. The CPU 801, ROM 802, and RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0198] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as required. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as required so that a computer program read from it is installed into the storage section 808 as required.

[0199] Specifically, according to the embodiments disclosed in the present invention, the process described in the above main step diagram can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for performing the method shown in the main step diagram. In the above embodiment, the computer program can be downloaded and installed from the network through the communication section 809, and / or installed from the removable medium 811. When the computer program is executed by the central processing unit 801, the above functions defined in the system of the present invention are executed.

[0200] It should be noted that the computer-readable medium shown in the present invention can be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted by any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0201] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the above-mentioned module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0202] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub - combinations and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A high-precision electrometer based on an intelligent algorithm, characterized in that, Including: A capacitance sensor module for acquiring electrostatic signals; A signal conversion circuit for converting the electrostatic signals into digital signals; A microprocessor for outputting a final electrostatic value by using a deep learning algorithm for the digital signals; In the microprocessor, including: A preprocessing module for performing noise suppression processing on the digital signals by using a filtering algorithm to obtain a corrected signal; A signal processing module for extracting features from the corrected signal to obtain a number of signal features, and using the number of signal features as inputs of an intelligent algorithm to obtain an electrostatic prediction value; A correction output module for acquiring historical electrostatic data, fitting the historical electrostatic data to obtain an error correction value, and calculating a final electrostatic value according to the error correction value and the electrostatic prediction value.

2. The high-precision electrometer based on an intelligent algorithm according to claim 1, characterized in that, The preprocessing module includes: A preliminary processing unit for removing low-frequency noise and invalid data in the digital signals by using a filtering algorithm to obtain first preprocessed data; A Kalman prediction unit for processing the first preprocessed data by using a Kalman filtering algorithm to obtain a Kalman prediction value; A correction unit for calculating corrected data according to the Kalman prediction value and the first preprocessed data; A normalization unit for performing normalization processing on the corrected data to obtain a corrected signal.

3. The high-precision electrometer based on an intelligent algorithm according to claim 1, characterized in that, The signal processing module includes: A feature extraction unit for inputting the corrected signal into a convolutional neural network to obtain a number of signal features; An identification unit for inputting the number of signal features into a classifier to obtain probabilities corresponding to each signal type, and selecting the signal type with the maximum probability as a classification result; A prediction unit for inputting the number of signal features into a preset regression model to obtain an electrostatic prediction value; A signal processing output unit for outputting the classification result and the electrostatic prediction value.

4. The high-precision electrometer based on an intelligent algorithm according to claim 1, characterized in that, The correction output module includes: A historical data acquisition unit for acquiring historical electrostatic data; An error calculation unit for calculating an error correction value by using a fitting error algorithm according to the historical electrostatic data; An output unit for calculating a final electrostatic value according to the error correction value and the electrostatic prediction value; The historical electrostatic data includes historical prediction data and historical real data; Calculating the mean square error, root mean square error, mean absolute error or a weighted sum value of the three as the error correction value according to the historical prediction data and the historical real data; The final electrostatic value is the sum of the error correction value and the electrostatic prediction value.

5. The high-precision electrometer based on an intelligent algorithm according to claim 1, characterized in that It further includes an environmental sensor and a composite calibration module; The environmental sensor is used for acquiring environmental data around the capacitance sensor module, a signal output end of the environmental sensor is connected to a signal input end of the composite calibration module, and the signal input end of the composite calibration module is connected to a signal output end of the preprocessing module; The composite calibration module is used for establishing a compensation function according to the environmental data, and compensating the corrected signal according to the compensation function to obtain an environmentally repaired electrostatic value.

6. The high-precision electrometer based on an intelligent algorithm according to claim 3, characterized in that, The convolutional neural network includes a convolutional layer, a pooling layer and a fully connected layer; The convolutional layer is configured to receive the correction signal and perform feature extraction on the correction signal to obtain a number of local temporal features; The pooling layer is configured to reduce the dimensionality of a number of the local temporal features to obtain a number of pooled features; The fully connected layer is configured to map the pooled features to signal features.

7. An electrostatic meter with high precision based on an intelligent algorithm according to claim 1, characterized in that, The microprocessor further includes an analysis module configured to analyze the final electrostatic value from three aspects of signal-to-noise ratio, signal volatility, and measurement error to obtain an analysis result, and adjust the parameters of the preprocessing module and the signal processing module according to the analysis result.

8. An intelligent algorithm-based high-precision electrometer according to claim 3, wherein In the feature extraction unit, the signal conversion circuit includes an analog front-end circuit and an analog-to-digital converter; A signal input end of the analog front-end circuit is connected to a signal output end of the capacitance sensor module, a signal output end of the analog front-end circuit is connected to a signal input end of the analog-to-digital converter, and a signal output end of the analog-to-digital converter is connected to a signal input end of the microprocessor through a serial communication interface; The analog front-end circuit is configured to amplify the electrostatic signal to obtain an amplified signal; The analog-to-digital converter is configured to convert the amplified signal into a digital signal.

9. An electrostatic meter with high precision based on an intelligent algorithm according to claim 5, characterized in that The compensation function includes a number of compensation terms, and the compensation terms include temperature compensation, humidity compensation, time compensation, air pressure compensation, power supply voltage compensation, initial deviation compensation of the capacitance sensor module, and aging compensation.

10. A signal processing method for a high-precision electrometer based on an intelligent algorithm, characterized in that, The method is based on the intelligent algorithm-based high-precision electrometer according to any one of claims 1-9, and the method includes the following steps: S10: Perform noise suppression processing on the digital signal using a filtering algorithm to obtain a correction signal; S20: Perform feature extraction on the correction signal to obtain a number of signal features, and use the number of signal features as an input to an intelligent algorithm to obtain an electrostatic prediction value; S30: Obtain historical electrostatic data, perform fitting on the historical electrostatic data to obtain an error correction value, and calculate a final electrostatic value according to the error correction value and the electrostatic prediction value.

Images