Method for measuring electrical variable of electronic component

Abstract

The invention relates to the technical field of electronic measurement, and discloses an electronic component electrical variable measurement method, which comprises the following steps of: 1, arranging a high-precision differential voltage sensor, a shunt resistor, a constant current source and impedance method capacitance measurement equipment at each key node of an electronic component so as to realize synchronous acquisition of voltage, current, resistance and capacitance, the differential voltage sensor is used for measuring start-end voltage and end-end voltage of an electronic component, the shunt resistor is used for calculating current through the start-end voltage and end-end voltage, the constant current source method is used for calculating resistance through known current and measured voltage, and the impedance method is used for measuring capacitance through alternating current signal frequency response. By combining a hardware low-pass filter and a digital signal processing algorithm, real-time filtering and noise suppression are carried out on collected data, efficient purification of signals is achieved, and the effects of improving the overall quality of the signals and providing clean and accurate data input for subsequent data analysis are achieved.

Description

Technical Field

[0001] The invention relates to the technical field of electronic measurement, and in particular to a method for measuring an electric variable of an electronic component. Background Art

[0002] In the electronics industry, accurately measuring and analyzing the electrical variables of electronic components (such as voltage, current, resistance, and capacitance) is essential to ensure equipment performance, reliability, and safety. Electronic measurements provide key information for design verification, quality control, fault diagnosis, and performance optimization. However, with the increasing complexity of electronic equipment and the continuous pursuit of high performance, traditional electrical variable measurement methods have shown limitations and shortcomings in many aspects.

[0003] The analog-to-digital converters commonly used in existing electrical variable measurement systems are limited in accuracy and resolution, resulting in quantization errors under high-precision measurement requirements. Especially when measuring tiny changes in electrical signals, the errors significantly affect the accuracy and reliability of the data. In addition, with the advancement of technology, the resolution requirements for measurement systems are constantly increasing, and traditional technologies can no longer meet the needs of emerging applications.

[0004] In practical applications, the measurement signal is affected by various interferences and noises, such as electromagnetic interference, temperature fluctuations and equipment aging, which causes the collected data to contain high-frequency noise. If not handled properly, it will seriously affect the accuracy of subsequent analysis and the availability of signals. Although the existing technology uses a variety of filtering technologies, it is unable to effectively suppress noise while reducing signal loss, limiting the effectiveness of signal processing technology in high-demand applications.

[0005] Existing measurement methods perform poorly when dealing with fast-changing or transient electronic signals, mainly due to the lack of tools that can effectively perform time-frequency analysis. Especially in scenarios where high-frequency changes need to be monitored or extremely short events need to be captured, the compromise between time resolution and frequency resolution of traditional tools such as fast Fourier transforms makes it impossible to provide enough details to identify problems.

[0006] Timely detection and diagnosis of faults are key to maintaining the reliability of electronic equipment. Traditional fault diagnosis technologies rely on post-analysis and lack effective real-time monitoring and prediction mechanisms, which increases maintenance costs and leads to equipment damage or more serious consequences. In addition, traditional methods cannot effectively extract learning and useful information from complex data patterns when processing data, which affects the accuracy of fault prediction.

[0007] Therefore, those skilled in the art provide a method for measuring electric variables of an electronic component to solve the above-mentioned problems. Summary of the invention

[0008] In view of the deficiencies of the prior art, the present invention provides a method for measuring electrical variables of electronic components to solve the problems raised in the above background technology.

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for measuring electrical variables of electronic components, comprising:

[0010] Step 1: Arrange high-precision differential voltage sensors, shunt resistors, constant current sources and impedance capacitance measurement equipment at key nodes of electronic components to achieve synchronous acquisition of voltage, current, resistance and capacitance. The differential voltage sensor is used to measure the voltage at the beginning and end of the electronic component, the shunt resistor is used to calculate the current through the voltage at the beginning and end, the constant current source method is used to calculate the resistance through the known current and the measured voltage, and the impedance method is used to measure the capacitance through the frequency response of the AC signal;

[0011] Step 2: digitize the data collected by the sensor using a high-precision analog-to-digital converter with multi-channel synchronous sampling function for data acquisition;

[0012] Step 3: Filter and suppress noise on the data collected by the digitizer. The filtering process includes using a hardware low-pass filter to remove high-frequency noise and using a software algorithm to purify the signal.

[0013] Step 4: Perform short-time Fourier transform on the preprocessed data, analyze the time-frequency characteristics of the electrical variable data, extract the dynamic frequency characteristics in the signal, and provide signal information for subsequent fault feature extraction;

[0014] Step 5: Based on the dynamic signal analysis, the threshold method and change rate analysis are used to detect anomalies in the electrical variable data, identify fault characteristics, and pay attention to the sudden changes and long-term trend changes of the electrical variables, providing key indicators for fault prediction;

[0015] Step 6: Use the long short-term memory network to perform time series trend analysis and prediction on the identified fault features, and combine the single-class support vector machine to classify and diagnose the fault type;

[0016] Step 7: Upload the measured, processed and analyzed data to the cloud through the wireless communication module to achieve remote storage and access of the data. On the cloud platform, the data will be used to generate detailed fault diagnosis reports and displayed through a visual interface, allowing users to monitor the status and health of electronic components in real time.

[0017] Preferably, in step 1, the shunt resistor is used to calculate the current using the following formula:

[0018]

[0019] Where I is the current, V Ris the voltage at the end of the shunt resistor, R shunt is a known shunt resistance value;

[0020] The voltage measured by the differential voltage sensor is V pos and V neg , the difference is calculated as:

[0021] V diff =V pos -V neg ,

[0022] Among them, V diff is the differential voltage, V pos and V neg It is the voltage value at the beginning and end of the electronic component.

[0023] Preferably, the high-precision analog-to-digital converter used in step 2 has a 24-bit resolution, and the quantization error is calculated by the following formula:

[0024]

[0025] Where Q is the quantization error, V ref is the reference voltage of the digital converter, and n is the resolution of the digital converter.

[0026] Preferably, the hardware low-pass filter in step 3 adopts a second-order active filter, and the transfer function of the filter is:

[0027]

[0028] Where H(f) is the frequency response of the filter, f c is the cutoff frequency of the filter, f is the frequency of the input signal, and j is the imaginary unit in the complex domain.

[0029] Preferably, the signal purification in step 3 uses a digital Kalman filter algorithm to suppress noise. The Kalman filter algorithm includes state prediction and observation update. The state prediction equation is:

[0030] x k =Ax k-1 +w k ,

[0031] Among them, x k is the current state, A is the state transfer matrix, w k is the process noise, x k-1 is the state vector of the system at time step k-1,

[0032] The observation update equation is:

[0033] x k =x k|k-1+K k (z k -Hx k|k-1 ),

[0034] Among them, x k is the current state, z k is the observed value, K k is the Kalman gain, x k|k-1 is the prior state estimate based on the prediction at time step k-1, Hx k|k-1 is the predicted measurement value.

[0035] Preferably, the short-time Fourier transform in step 4 is used to analyze the dynamic frequency characteristics of the electric variable signal, which is specifically expressed as:

[0036]

[0037] Among them, X(t, f) is the time-frequency domain signal, w(#) is the time window function, t is time, f is frequency, x(#) is the original signal,

[0038] w(#-t) is the time window function, centered at t, used to localize the signal;

[0039] e -j2sf# is the complex exponential kernel of the Fourier transform, representing the frequency component f;

[0040] d# is the integration variable.

[0041] Preferably, the time window function w(#) in step 4 adopts a Hanning window function, and the expression is:

[0042]

[0043] Where N is the window length, w(#) is the time window function, # is the time index in the window function, is the periodic cosine term in the window function calculation.

[0044] Preferably, the long short-term memory network in step 6 is used to predict the time series trend of the fault characteristics, and the state update formula of the long short-term memory network hidden layer is:

[0045] h t =L(W xh x t +W hh h t-1 +b h ),

[0046] Among them, h t is the current hidden state, x t is the input eigenvalue, W xh and Whh is the weight matrix from input to hidden layer and the weight matrix from hidden layer to hidden layer, b h is the bias, L is the activation function, h t-1 is the hidden state at time step t-1.

[0047] Preferably, the fault type diagnosis in step 6 adopts a single-class support vector machine, and the classification function is: f(x)=sgn(<w,φ(x)> +b),

[0048] Among them, f(x) is the classification function, w is the normal vector of the classification hyperplane, φ(x) is the feature mapping function, and b is the bias term.

[0049] Preferably, the long short-term memory network in step 6 is further combined with the structure of a recursive neural network. When performing time series trend prediction, the unit state update formula of the long short-term memory network is:

[0050]

[0051] Among them, c t is the cell state at time step t, f t is the output of the forget gate, c t-1 is the cell state at time step t-1, i t Is the output of the input gate, controlling the current input Impact on status.

[0052] The present invention provides a method for measuring electric variables of electronic components. It has the following beneficial effects:

[0053] 1. The present invention uses a high-precision analog-to-digital converter with a multi-channel synchronous sampling function to collect data, thereby achieving high-precision digitization of data, greatly reducing quantization errors and improving the accuracy of measurement data.

[0054] 2. The present invention realizes efficient signal purification by combining hardware low-pass filters and digital signal processing algorithms to perform real-time filtering and noise suppression on the collected data, thereby improving the overall signal quality and providing clean and accurate data input for subsequent data analysis.

[0055] 3. The present invention applies short-time Fourier transform to perform time-frequency analysis on the preprocessed data, thereby extracting accurate key frequency features in dynamically changing electronic signals, and achieving the effect of being able to monitor and analyze signal time-frequency characteristics in real time and effectively identify changes in equipment performance.

[0056] 4. The present invention combines long short-term memory networks and single-class support vector machines to perform time series trend analysis and classification diagnosis on the identified fault features, thereby achieving high-precision prediction and instant diagnosis of potential faults of electronic components, thereby improving the fault prediction accuracy and diagnostic capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0058] In order to make the technical personnel in the technical field understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in combination with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a partial embodiment of the present invention, not a complete embodiment. Based on the embodiment of the present invention, other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0059] The present invention is described in detail below in conjunction with the accompanying drawings:

[0060] Example:

[0061] Please see attached Figure 1 The embodiment of the present invention provides a method for measuring an electric variable of an electronic component, comprising:

[0062] Step 1: Arrange high-precision differential voltage sensors, shunt resistors, constant current sources and impedance capacitance measurement equipment at key nodes of electronic components to achieve synchronous acquisition of voltage, current, resistance and capacitance. The differential voltage sensor is used to measure the voltage at the beginning and end of the electronic component, the shunt resistor is used to calculate the current through the voltage at the beginning and end, the constant current source method is used to calculate the resistance through the known current and the measured voltage, and the impedance method is used to measure the capacitance through the frequency response of the AC signal;

[0063] Step 2: digitize the data collected by the sensor using a high-precision analog-to-digital converter with multi-channel synchronous sampling function for data acquisition;

[0064] Step 3: Filter and suppress noise on the data collected by the digitizer. The filtering process includes using a hardware low-pass filter to remove high-frequency noise and using a software algorithm to purify the signal.

[0065] Step 4: Perform short-time Fourier transform on the preprocessed data, analyze the time-frequency characteristics of the electrical variable data, extract the dynamic frequency characteristics in the signal, and provide signal information for subsequent fault feature extraction;

[0066] Step 5: Based on the dynamic signal analysis, the threshold method and change rate analysis are used to detect anomalies in the electrical variable data, identify fault characteristics, and pay attention to the sudden changes and long-term trend changes of the electrical variables, providing key indicators for fault prediction;

[0067] Step 6: Use the long short-term memory network to perform time series trend analysis and prediction on the identified fault features, and combine the single-class support vector machine to classify and diagnose the fault type;

[0068] Step 7: Upload the measured, processed and analyzed data to the cloud through the wireless communication module to achieve remote storage and access of the data. On the cloud platform, the data will be used to generate detailed fault diagnosis reports and displayed through a visual interface, allowing users to monitor the status and health of electronic components in real time.

[0069] Step 1 Benefits: High-precision sensors are placed at key nodes of electronic components, including differential voltage sensors, shunt resistors, constant current sources, and impedance capacitance measurement devices. This enables synchronous acquisition of voltage, current, resistance, and capacitance, ensuring real-time and accuracy of data. This is critical for capturing electrical variables of electronic components under different working conditions, and helps detect and respond to possible performance deviations or failures.

[0070] Step 2 Benefits: The use of a high-precision digital converter with multi-channel synchronous sampling ensures seamless conversion of sensors to digital signals, improves data acquisition accuracy and reduces information loss, which is particularly important for subsequent data analysis and fault detection;

[0071] Step 3 Benefits: Perform hardware low-pass filtering and software algorithm processing on the collected data to effectively remove high-frequency noise and improve data quality;

[0072] Step 4 Benefits: Applying short-time Fourier transform to analyze the preprocessed data and extracting dynamic frequency features can reveal the time-varying characteristics and frequency distribution in the electrical variable data, which is extremely important for understanding the behavior patterns of electronic components in different states;

[0073] Step 5 Benefits: Through the threshold method and change rate analysis, abnormal detection of electrical variable data is carried out, fault characteristics are identified in a timely manner, and attention is paid to sudden changes and long-term trend changes of electrical variables, providing key indicators for fault prediction, thereby enhancing the system's preventive maintenance capabilities and reducing unexpected downtime;

[0074] Step 6 Benefits: Using long short-term memory network combined with single-class support vector machine to conduct in-depth time series trend analysis and accurate classification of fault characteristics, greatly improving the accuracy of fault prediction and the immediacy of diagnosis, and improving the reliability and efficiency of equipment operation;

[0075] Step 7 Benefits: Data is uploaded to the cloud for storage and processing through the wireless communication module. Users can access detailed fault diagnosis reports through the cloud platform and use the visual interface for real-time monitoring, providing a flexible and accessible solution to enhance the convenience of data management and the real-time nature of monitoring.

[0076] In step 1, the shunt resistor is used to calculate the current using the following formula:

[0077]

[0078] Where I is the current, V R is the voltage at the end of the shunt resistor, R shunt is a known shunt resistance value;

[0079] The voltage measured by the differential voltage sensor is V pos and V neg , the difference is calculated as:

[0080] V diff =V pos -V neg ,

[0081] Among them, V diff is the differential voltage, V pos and V neg It is the voltage value at the beginning and end of the electronic component.

[0082] Using shunt resistors to calculate current allows the user to obtain current data without directly intervening in the main circuit, reducing the system's sensitivity to measurement disturbances and improving measurement accuracy and reliability.

[0083] Differential voltage sensors effectively suppress noise caused by power supply fluctuations or other external interference factors by measuring the difference between voltages, especially in environments with strong electromagnetic interference.

[0084] Synchronous acquisition of voltage, current, resistance and capacitance ensures consistency and time synchronization of data acquisition, which is crucial for subsequent analysis and fault diagnosis, and can accurately reflect the real-time working status of electronic components.

[0085] In summary, step 1 realizes the synchronization and accurate measurement of key electrical variables of electronic components by arranging high-precision differential voltage sensors and shunt resistors. The measurement method improves the accuracy and anti-interference ability of the data, and provides solid data support for the reliability and safety of the system. By optimizing the data acquisition strategy, the present invention ensures the efficiency and accuracy of performance monitoring and fault prediction of electronic components under various working conditions, and is particularly suitable for high-tech fields with strict requirements on the performance of electronic equipment.

[0086] The high-precision analog-to-digital converter used in step 2 has a 24-bit resolution, and the quantization error is calculated by the following formula:

[0087]

[0088] Where Q is the quantization error, V ref is the reference voltage of the digital converter, and n is the resolution of the digital converter.

[0089] By using a 24-bit resolution digital converter, the device can provide very low quantization error and extremely high measurement accuracy, which means that very small voltage changes can be accurately captured and converted into digital signals, ensuring high data accuracy.

[0090] High-resolution digitizers can improve measurement accuracy and expand the dynamic range of the system, allowing the system to handle a wider range of input signals, from weak signals to strong signals, which can be accurately measured and recorded without distortion.

[0091] Reducing quantization errors helps reduce noise and errors in subsequent signal processing, improving the effectiveness and reliability of signal processing algorithms. High-precision input data can significantly improve the sensitivity and accuracy of fault detection algorithms.

[0092] In summary, by adopting a high-precision digital converter with a 24-bit resolution in step 2, the present invention significantly improves the measurement accuracy of electrical variable data and the overall performance of the system. The high-precision data acquisition method is particularly important for application scenarios that require high reliability and high accuracy measurements. It optimizes data quality, improves the efficiency and breadth of application of electronic measurement systems, and ensures that high-quality measurement results can be obtained in various operating environments.

[0093] The hardware low-pass filter in step 3 uses a second-order active filter, and the transfer function of the filter is:

[0094]

[0095] Where H(f) is the frequency response of the filter, f c is the cutoff frequency of the filter, f is the frequency of the input signal, and j is the imaginary unit in the complex domain.

[0096] The signal purification in step 3 uses a digital Kalman filter algorithm to suppress noise. The Kalman filter algorithm includes state prediction and observation update. The state prediction equation is:

[0097] x k =Ax k-1 +w k ,

[0098] Among them, x kis the current state, A is the state transfer matrix, w k is the process noise, x k-1 is the state vector of the system at time step k-1,

[0099] The observation update equation is:

[0100] x k =x k|k-1 +K k (z k -Hx k|k-1 ),

[0101] Among them, x k is the current state, z k is the observed value, K k is the Kalman gain, x k|k-1 is the prior state estimate based on the prediction at time step k-1, Hx k|k-1 is the predicted measurement value.

[0102] The use of hardware low-pass filters can effectively reduce high-frequency noise near the signal source, reduce interference during signal acquisition and transmission, and ensure signal purity and accuracy.

[0103] The digital Kalman filter algorithm provides an effective means of dealing with noise, and can optimize the estimation process by adapting to changes in process and observation noise, making it suitable for dynamically changing measurement environments.

[0104] Kalman filtering can adjust and optimize the filtering effect in real time by continuously updating state estimates, which is particularly important for applications that need to process time series data or dynamic systems.

[0105] Combining hardware filtering and advanced digital algorithms can improve the quality of data processing, enhance the system's adaptability to complex environments, and thus enhance the system's overall performance and reliability.

[0106] In summary, the strategy of combining hardware low-pass filter and digital Kalman filter algorithm in step 3 significantly improves the quality of electrical variable data and the efficiency of signal processing. The hardware low-pass filter directly reduces high-frequency noise at the signal source, while the digital Kalman filter algorithm adjusts and optimizes the signal processing process through precise mathematical models. The dual filtering mechanism ensures that each link from data acquisition to processing can maintain high-standard signal quality, providing a solid foundation for subsequent data analysis and fault diagnosis, and is particularly suitable for precision electronic measurement and control systems with extremely high data quality requirements.

[0107] The short-time Fourier transform in step 4 is used to analyze the dynamic frequency characteristics of the electrical variable signal, which is specifically expressed as:

[0108]

[0109] Among them, X(t, f) is the time-frequency domain signal, w(#) is the time window function, t is time, f is frequency, x(#) is the original signal,

[0110] w(#-t) is the time window function, centered at t, used to localize the signal;

[0111] e -j2sf# is the complex exponential kernel of the Fourier transform, representing the frequency component f;

[0112] d# is the integration variable.

[0113] The time window function w(#) in step 4 adopts the Hanning window function, and its expression is:

[0114]

[0115] Where N is the window length, w(#) is the time window function, # is the time index in the window function, is the periodic cosine term in the window function calculation.

[0116] By using a local window function at each time point, the short-time Fourier transform allows the analysis to consider both time and frequency information, allowing for a more detailed observation of the dynamic changes of the signal. It is particularly useful for rapidly changing electronic signals, such as signal analysis of electronic equipment during transient operation.

[0117] The use of the Hanning window function can effectively reduce spectrum leakage caused by window truncation and improve the detection accuracy of frequency components. The Hanning window smoothes the window edge and reduces the discontinuity of the signal edge, thereby improving the accuracy of spectrum analysis.

[0118] Short-time Fourier transform makes it possible to specifically analyze the frequency components of a signal within a specific time period, which is crucial for detecting transient phenomena such as failures or performance degradation of electronic components during operation. It can reveal the specific moment and frequency characteristics of the problem, providing key information for fault diagnosis and system maintenance.

[0119] In summary, the short-time Fourier transform and Hanning window function applied in step 4 provide an efficient and accurate method for electrical variable signal analysis, so that the time-frequency characteristics of the signal can be comprehensively and carefully evaluated. Through the application of appropriate window functions and short-time Fourier transform, the present invention can reduce spectrum leakage during the analysis process while maintaining good time and frequency resolution, thereby significantly improving the testing and monitoring capabilities of electronic components, and has a direct and positive impact on improving the reliability and performance analysis of electronic equipment, and is particularly suitable for application scenarios with high requirements for signal quality and analysis accuracy.

[0120] The long short-term memory network in step 6 is used to predict the time series trend of fault characteristics. The state update formula of the long short-term memory network hidden layer is:

[0121] h t =L(W xh x t +W hh h t-1 +b h ),

[0122] Among them, h t is the current hidden state, x t is the input eigenvalue, W xh and W hh is the weight matrix from input to hidden layer and the weight matrix from hidden layer to hidden layer, b h is the bias, L is the activation function, h t-1 is the hidden state at time step t-1.

[0123] The fault type diagnosis in step 6 uses a single-class support vector machine, and the classification function is:

[0124] f(x)=sgn(<w,φ(x)> +b),

[0125] Among them, f(x) is the classification function, w is the normal vector of the classification hyperplane, φ(x) is the feature mapping function, and b is the bias term.

[0126] The long short-term memory network in step 6 is further combined with the structure of the recursive neural network. When performing time series trend prediction, the unit state update formula of the long short-term memory network is:

[0127]

[0128] Among them, c t is the cell state at time step t, f t is the output of the forget gate, c t-1 is the cell state at time step t-1, i t Is the output of the input gate, controlling the current input Impact on status.

[0129] The LSTM network can effectively process and memorize long-term dependent information through its internal structure, which is critical for the processing of time series data. Through the adjustment of the forget gate and the input gate, the LSTM network can flexibly retain or forget information, making it able to adapt to the complex dynamic changes of electronic components in long time series.

[0130] Long short-term memory networks can learn the normal and abnormal behavior patterns of electronic components from historical data, and then predict possible failures in future operations. The prediction is based on past data patterns, taking into account the latest changes in the data, and providing high-accuracy failure prediction.

[0131] The single-class support vector machine identifies anomalies by learning the data patterns of normal operations, allowing the system to identify potential anomalies and failures without significant fault labels. It is particularly suitable for anomaly detection.

[0132] The combined LSTM and SVM approach provides a powerful adaptive mechanism to adjust its fault detection and prediction model over time, which is particularly important to cope with the performance changes of electronic components under different environmental and load conditions.

[0133] In summary, the LSTM network and single-class support vector machine technology applied in step 6 greatly enhance the fault prediction and diagnosis capabilities of the electronic component electrical variable measurement system. The long-term memory capability of the LSTM network combined with the anomaly detection function of the vector machine enables the system to achieve efficient and accurate time series trend analysis and fault classification, thereby improving the reliability and maintenance efficiency of electronic components. The integrated method optimizes the fault handling process and enhances the stability and safety of equipment operation.

[0134] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for measuring electrical variables of electronic components, characterized in that: include: Step 1: Arrange high-precision differential voltage sensors, shunt resistors, constant current sources and impedance capacitance measurement equipment at key nodes of electronic components to achieve synchronous acquisition of voltage, current, resistance and capacitance. The differential voltage sensor is used to measure the voltage at the beginning and end of the electronic component, the shunt resistor is used to calculate the current through the voltage at the beginning and end, the constant current source method is used to calculate the resistance through the known current and the measured voltage, and the impedance method is used to measure the capacitance through the frequency response of the AC signal; Step 2: digitize the data collected by the sensor using a high-precision analog-to-digital converter with multi-channel synchronous sampling function for data acquisition; Step 3: Filter and suppress noise on the data collected by the digitizer. The filtering process includes using a hardware low-pass filter to remove high-frequency noise and using a software algorithm to purify the signal. Step 4: Perform short-time Fourier transform on the preprocessed data, analyze the time-frequency characteristics of the electrical variable data, extract the dynamic frequency characteristics in the signal, and provide signal information for subsequent fault feature extraction; Step 5: Based on the dynamic signal analysis, the threshold method and change rate analysis are used to detect anomalies in the electrical variable data, identify fault characteristics, and pay attention to the sudden changes and long-term trend changes of the electrical variables, providing key indicators for fault prediction; Step 6: Use the long short-term memory network to perform time series trend analysis and prediction on the identified fault features, and combine the single-class support vector machine to classify and diagnose the fault type; Step 7: Upload the measured, processed and analyzed data to the cloud through the wireless communication module to achieve remote storage and access of the data. On the cloud platform, the data will be used to generate detailed fault diagnosis reports and displayed through a visual interface, allowing users to monitor the status and health of electronic components in real time.

2. A method for measuring electrical variables of electronic components according to claim 1, characterized in that: In step 1, the shunt resistor is used to calculate the current using the following formula: Where I is the current, V R is the voltage at the beginning and end of the shunt resistor, R shunt is a known shunt resistance value; The voltage measured by the differential voltage sensor is V pos and V neg , the difference is calculated as: V diff =V pos -V neg , Among them, V diff is the differential voltage, V pos and V neg It is the voltage value at the beginning and end of the electronic component.

3. The method for measuring electric variables of electronic components according to claim 1, characterized in that: The high-precision analog-to-digital converter used in step 2 has a 24-bit resolution, and the quantization error is calculated by the following formula: Where Q is the quantization error, V ref is the reference voltage of the digital converter, and n is the resolution of the digital converter.

4. The method for measuring electric variables of electronic components according to claim 1, characterized in that: The hardware low-pass filter in step 3 adopts a second-order active filter, and the transfer function of the filter is: Where H(f) is the frequency response of the filter, f c is the cutoff frequency of the filter, f is the frequency of the input signal, and j is the imaginary unit in the complex domain.

5. The method for measuring electric variables of electronic components according to claim 4, characterized in that: The signal purification in step 3 uses a digital Kalman filter algorithm to suppress noise. The Kalman filter algorithm includes state prediction and observation update. The state prediction equation is: x k =Ax k-1 +w k , Among them, x k is the current state, A is the state transfer matrix, w k is the process noise, x k-1 is the state vector of the system at time step k-1, The observation update equation is: x k =x k|k-1 +K k (z k -Hx k|k-1 ), Among them, x k is the current state, z k is the observed value, K k is the Kalman gain, x k|k-1 is the prior state estimate based on the prediction at time step k-1, Hx k|k-1 is the predicted measurement value.

6. The method for measuring electric variables of electronic components according to claim 1, characterized in that: The short-time Fourier transform in step 4 is used to analyze the dynamic frequency characteristics of the electric variable signal, which is specifically expressed as: Among them, X(t, f) is the time-frequency domain signal, w(#) is the time window function, t is time, f is frequency, x(#) is the original signal, w(#-t) is the time window function, centered at t, used to localize the signal; e -j2sf# is the complex exponential kernel of the Fourier transform, representing the frequency component f; d# is the integration variable.

7. A method for measuring electric variables of electronic components according to claim 6, characterized in that: The time window function w(#) in step 4 adopts the Hanning window function, and the expression is: Where N is the window length, w(#) is the time window function, # is the time index in the window function, is the periodic cosine term in the window function calculation.

8. The method for measuring electric variables of electronic components according to claim 1, characterized in that: The long short-term memory network in step 6 is used to predict the time series trend of fault features. The state update formula of the long short-term memory network hidden layer is: h t =L(W xh x t +W hh h t-1 +b h ), Among them, h t is the current hidden state, x t is the input eigenvalue, W xh and W hh is the weight matrix from input to hidden layer and the weight matrix from hidden layer to hidden layer, b h is the bias, L is the activation function, h t-1 is the hidden state at time step t-1.

9. A method for measuring electric variables of electronic components according to claim 8, characterized in that: The fault type diagnosis in step 6 adopts a single-class support vector machine, and the classification function is: f(x)=sgn(<w,φ(x)> +b), Among them, f(x) is the classification function, w is the normal vector of the classification hyperplane, φ(x) is the feature mapping function, and b is the bias term.

10. The method for measuring electric variables of electronic components according to claim 9, characterized in that: The long short-term memory network in step 6 is further combined with the structure of the recursive neural network. When performing time series trend prediction, the unit state update formula of the long short-term memory network is: Among them, c t is the cell state at time step t, f t is the output of the forget gate, c t-1 is the cell state at time step t-1, i t Is the output of the input gate, controlling the current input Impact on status.

Images