Capacitance measurement system based on big data multi-source calculation and measurement method thereof

Abstract

The invention provides a capacitance measurement system based on big data multi-source calculation and a measurement method thereof. Relates to the technical field of electrical measurement and data processing, and comprises a multi-source data acquisition module for performing real-time acquisition on a capacitor through a multi-source sensor to generate multi-source heterogeneous data. And the synchronous sampling module is used for performing unified time synchronization on the multi-source heterogeneous data through a hybrid synchronization mechanism to generate a hybrid synchronization sequence. According to the capacitance measurement system based on big data multi-source calculation and the measurement method thereof, by adopting a self-adaptive correction mechanism and a multi-model self-adaptive weighted fusion technology, a measurement algorithm can be adjusted in real time to cope with dynamic changes under complex working conditions. By combining the powerful processing capability of a big data computing platform, the online monitoring precision of the capacitor is improved, a more reliable basis is provided for life prediction and performance evaluation of the capacitor, and the reliability and high efficiency in industrial production and engineering application are ensured.

Description

Technical Field

[0001] This invention relates to the field of electrical measurement and data processing technology, specifically to a capacitance measurement system and method based on big data multi-source computing. Background Technology

[0002] Currently, most capacitance measurement technologies are based on single-point sensors or single signal channels, utilizing methods such as the bridge method, impedance analysis, or frequency response method to detect capacitor parameters. Common system structures are simple, and data processing relies on fixed algorithms, meeting basic measurement needs in typical laboratory environments. In industrial production and engineering applications, with the increasing performance requirements of power electronic devices, large energy storage equipment, and new energy vehicles, the introduction of multi-source data is gaining attention. Multi-channel acquisition and digital analysis improve the measurement accuracy of capacitance parameters. Big data processing and cloud computing platforms are increasingly being applied to capacitance monitoring, processing complex operational data and providing remote diagnostic support. Existing technologies have initially demonstrated multi-source fusion and big data computing capabilities in terms of measurement systems and data processing models.

[0003] However, existing capacitance measurement methods still have core shortcomings, lacking a unified calculation and dynamic correlation mechanism for multi-source heterogeneous data. Inconsistencies in data formats, sampling accuracy, and time synchronization standards among different sensors make effective data fusion difficult in large-scale operating environments. Accumulated noise and biases among the data often lead to unstable or distorted measurement results, affecting the accurate assessment of capacitance values, loss factors, and lifespan parameters. Traditional algorithms often employ single correction methods, which are ill-suited to adapt to dynamic changes under complex operating conditions and cannot provide a reliable basis for online monitoring and lifespan prediction of capacitors. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a capacitance measurement system and method based on big data multi-source computing. The technical problem this invention aims to solve is: how to achieve high-precision capacitance measurement and reliable performance prediction under complex working conditions through the synchronous fusion of multi-source data acquisition and big data multi-source computing.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a capacitance measurement system based on big data multi-source computing, comprising: a multi-source data acquisition module, which acquires multi-source heterogeneous data of the capacitance in real time through multi-source sensors.

[0006] The synchronization sampling module generates a hybrid synchronization sequence by uniformly synchronizing the multi-source heterogeneous data through a hybrid synchronization mechanism. The synchronization sampling module then performs integral-differential modulation on the hybrid synchronization sequence to obtain high-resolution sampling data.

[0007] The feature processing module includes a feature extraction unit and a parameter aggregation unit. The feature extraction unit performs time-frequency analysis on the high-resolution sampled data to extract transient response characteristic parameters of the capacitor. The parameter aggregation unit integrates the transient response characteristic parameters of the capacitor to generate an initial feature parameter set.

[0008] The adaptive correction module processes the initial feature parameter set through a residual feedback mechanism to generate residual feedback values. Based on the residual feedback values, the adaptive correction module performs multi-model adaptive weighted fusion to generate a corrected feature parameter set.

[0009] The output module performs unified calculations on the modified feature parameter set through a big data computing platform and outputs the capacitance measurement results.

[0010] Preferably, the multi-source sensor includes a voltage sensor, a current sensor, and an environmental parameter sensor, and the multi-source heterogeneous data includes voltage data, current data, temperature data, and humidity data.

[0011] Preferably, the hybrid synchronization mechanism includes a PTP mechanism and a 1PPS mechanism. The synchronization sampling module uses the PTP mechanism to synchronize the clocks of network nodes in the multi-source heterogeneous data to generate a synchronization sequence. The synchronization sampling module uses the 1PPS mechanism to time-stamp the multi-source heterogeneous data to generate a time-stamped sequence. The synchronization sequence and the time-stamped sequence are fused to generate the hybrid synchronization sequence.

[0012] Preferably, the integral-differential modulation includes the following steps: S21. The hybrid synchronization sequence is sampled to form a sampled data stream.

[0013] S22. Perform integral-differential modulation on the sampled data stream to generate a quantized bit stream.

[0014] S23. The quantized bit stream is extracted to generate the high-resolution sampled data.

[0015] Preferably, the transient response characteristic parameters of the capacitor include voltage and current phase difference parameters, equivalent series resistance parameters, and dielectric loss parameters.

[0016] Preferably, the residual feedback mechanism generates the residual feedback value by calculating the difference between the initial feature parameter set and the standard feature data. The standard feature data consists of historical measurement feature parameters stored in the capacitance measurement database. The model formula for the multi-model adaptive weighted fusion is: in, To correct the feature parameter set, it is normalized to become dimensionless values. These are the characteristic parameters of the Kalman filter output. The feature parameters for deep learning predictions are normalized to become dimensionless values. For residual feedback value The dynamic weighting function, whose value ranges from 0 to 1, is a dimensionless quantity. As the residual evaluation index, the normalization process adopts the residual normalization weighted method, and the dynamic weight function is determined by the normalized residual. The weights are adjusted in real time according to the residual size and the sum of the weights is equal to 1.

[0017] Preferably, the big data computing platform includes a data computing unit and a data storage unit. The data computing unit performs parallel calculations on the modified feature parameter set to generate the capacitance measurement results. The data storage unit stores the capacitance measurement results in a structured manner, supporting dynamic association and subsequent querying of historical data.

[0018] A capacitance measurement method based on big data multi-source computing includes: S1. Real-time acquisition of multi-source heterogeneous data of the capacitor is performed using multi-source sensors, including voltage sensors, current sensors, and environmental parameter sensors.

[0019] S2. A hybrid synchronization sequence is generated by uniformly synchronizing the multi-source heterogeneous data through a hybrid synchronization mechanism, and high-resolution sampling data is obtained by integral-differential modulation of the hybrid synchronization sequence. The hybrid synchronization mechanism includes PTP and 1PPS.

[0020] S3. Perform time-frequency analysis on the high-resolution sampled data to extract the transient response characteristic parameters of the capacitor, and integrate the transient response characteristic parameters of the capacitor to generate an initial feature parameter set. The transient response characteristic parameters of the capacitor include voltage and current phase difference parameters, equivalent series resistance parameters, and dielectric loss parameters.

[0021] S4. The initial feature parameter set is compared to generate residual feedback values. The comparison process adopts a residual feedback mechanism, and the residual feedback values ​​are subjected to multi-model adaptive weighted fusion to generate a corrected feature parameter set.

[0022] S5. The modified feature parameter set is uniformly calculated and the capacitance measurement results are output through a big data computing platform.

[0023] This invention provides a capacitance measurement system and method based on big data multi-source computing. It has the following beneficial effects: This invention, by introducing multi-source data acquisition and big data multi-source computing technologies, can effectively fuse heterogeneous data from different sensors, achieving unified data calculation and dynamic correlation. Through high-resolution sampling and time-frequency analysis, the transient response characteristic parameters of capacitance are accurately extracted, thereby significantly improving the accuracy and stability of capacitance measurement.

[0024] This capacitance measurement system and method, based on big data multi-source computing, employs an adaptive correction mechanism and multi-model adaptive weighted fusion technology to adjust the measurement algorithm in real time to cope with dynamic changes under complex operating conditions. Combined with the powerful processing capabilities of the big data computing platform, it improves the accuracy of online capacitor monitoring, providing a more reliable basis for life prediction and performance evaluation, and ensuring reliability and efficiency in industrial production and engineering applications. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a capacitance measurement system based on big data multi-source computing. Figure 2 This is a schematic diagram of a multi-source data acquisition and synchronization mechanism; Figure 3 This is a schematic diagram of the adaptive correction and weighted fusion process. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1

[0028] like Figure 1-3 As shown, this embodiment of the invention provides a capacitance measurement system based on big data multi-source computing, including a multi-source data acquisition module that uses multi-source sensors to collect capacitance data in real time and generate multi-source heterogeneous data. The multi-source sensors include voltage sensors, current sensors, and environmental parameter sensors, and the multi-source heterogeneous data includes voltage data, current data, temperature data, and humidity data.

[0029] A voltage sensor measures the voltage across a capacitor. By contacting the capacitor, it converts the voltage signal into a digital signal, which is then transmitted to the data acquisition system.

[0030] A current sensor is used to measure changes in the current of a capacitor. By connecting to a circuit, the current sensor monitors current fluctuations in real time during the capacitor's charging and discharging process. For example, during system operation, a current sensor might collect a current reading of 0.12A, which is crucial for analyzing the capacitor's performance.

[0031] Environmental parameter sensors, including temperature and humidity sensors, are used to monitor changes in temperature and humidity in the capacitor's operating environment. This effectively complements capacitor performance analysis, as temperature and humidity changes significantly impact the capacitor's operating state.

[0032] The synchronization sampling module uses a hybrid synchronization mechanism to uniformly synchronize multi-source heterogeneous data and generate a hybrid synchronization sequence. The module then performs integral-differential modulation on the hybrid synchronization sequence to obtain high-resolution sampled data. The hybrid synchronization mechanism includes PTP and 1PPS mechanisms. The PTP mechanism synchronizes the clocks of network nodes in the multi-source heterogeneous data to generate a synchronization sequence, while the 1PPS mechanism timestamps the multi-source heterogeneous data to generate a time-stamped sequence. The synchronization sequence and the time-stamped sequence are then fused to generate the hybrid synchronization sequence. The integral-differential modulation includes the following steps: S21. Sample the mixed synchronization sequence to form a sampled data stream.

[0033] S22. Perform integral-differential modulation on the sampled data stream to generate a quantized bit stream.

[0034] S23. Extract the quantized bit stream to generate high-resolution sampled data.

[0035] The feature processing module includes a feature extraction unit and a parameter aggregation unit. The feature extraction unit performs time-frequency analysis on the high-resolution sampled data to extract transient response characteristic parameters of the capacitor. The parameter aggregation unit integrates the transient response characteristic parameters of the capacitor to generate an initial feature parameter set. The transient response characteristic parameters of the capacitor include voltage and current phase difference parameters, equivalent series resistance parameters, and dielectric loss parameters.

[0036] The adaptive correction module processes the initial feature parameter set through a residual feedback mechanism to generate residual feedback values. Based on these residual feedback values, the module performs multi-model adaptive weighted fusion to generate a corrected feature parameter set. The residual feedback mechanism generates residual feedback values ​​by calculating the difference between the initial feature parameter set and standard feature data. The standard feature data consists of historical measurement feature parameters stored in a capacitance measurement database. The formula for the multi-model adaptive weighted fusion is as follows: in, To correct the feature parameter set, it is normalized to become dimensionless values. These are the characteristic parameters of the Kalman filter output. The feature parameters for deep learning predictions are normalized to become dimensionless values. For residual feedback value The dynamic weight function has a value range between 0 and 1 and is a dimensionless quantity. As the residual evaluation index, the normalization process adopts the residual normalization weighted method. The dynamic weight function is determined by the normalized residual. The weights are adjusted in real time according to the size of the residual and the sum of the weights is equal to 1.

[0037] The output module uses a big data computing platform to perform unified calculations on the corrected feature parameter set and outputs the capacitance measurement results. The big data computing platform includes a data computing unit and a data storage unit. The data computing unit performs parallel calculations on the corrected feature parameter set to generate the capacitance measurement results, while the data storage unit stores the capacitance measurement results in a structured manner, supporting dynamic association and subsequent querying of historical data.

[0038] A capacitance measurement method based on big data multi-source computing includes: S1. Real-time acquisition of capacitance data is performed using multi-source sensors to generate multi-source heterogeneous data. The multi-source sensors include voltage sensors, current sensors, and environmental parameter sensors.

[0039] S2. A hybrid synchronization sequence is generated by uniformly synchronizing multi-source heterogeneous data through a hybrid synchronization mechanism. The hybrid synchronization sequence is then subjected to integral-differential modulation to obtain high-resolution sampled data. The hybrid synchronization mechanism includes PTP and 1PPS.

[0040] S3. Perform time-frequency analysis on the high-resolution sampled data to extract the transient response characteristic parameters of the capacitor. Integrate the transient response characteristic parameters of the capacitor to generate an initial characteristic parameter set. The transient response characteristic parameters of the capacitor include voltage and current phase difference parameters, equivalent series resistance parameters, and dielectric loss parameters.

[0041] S4. The initial feature parameter set is compared to generate residual feedback values. The comparison process adopts a residual feedback mechanism, and the residual feedback values ​​are subjected to multi-model adaptive weighted fusion to generate a corrected feature parameter set.

[0042] S5. The capacitance measurement results are output by uniformly calculating the corrected feature parameter set through a big data computing platform.

[0043] Example 2

[0044] This embodiment demonstrates how to use PTP and 1PPS mechanisms to synchronize data based on the synchronous sampling module, and how to obtain high-resolution sampled data through integral-differential modulation and decimation processing.

[0045] 1. Implementation of Hybrid Synchronization Mechanism The synchronous sampling module synchronizes the clocks of multi-source heterogeneous data using the PTP mechanism. In one embodiment, the network has multiple nodes, each with asynchronous sampling times. Through the PTP mechanism, the system can synchronize the clocks of all nodes to a unified standard clock, ensuring that data collected by different sensors can be compared and analyzed simultaneously.

[0046] For example, in one embodiment, the sampling timestamp of the voltage sensor is T1=10:00:01.000, the sampling timestamp of the current sensor is T2=10:00:01.050, and the timestamp of the ambient temperature sensor is T3=10:00:00.999, with slight differences between the timestamps. Under the action of the PTP mechanism, the timestamps of all sensors will be synchronized to the same moment T0=10:00:01.000, thus forming a unified synchronization sequence.

[0047] The 1PPS mechanism is used to generate time-stamped sequences. The 1PPS mechanism marks a standard of time by generating a pulse signal every second, ensuring that each data sampling point has a clear time marker. For example, the 1PPS signal emits a pulse signal at the beginning of each second, and this pulse signal will calibrate the specific sampling time across all sensors.

[0048] After processing by the PTP and 1PPS mechanisms, the system generates a hybrid synchronization sequence, which combines information from the synchronization sequence and the time-stamped sequence. In one embodiment, the fusion result of the synchronization sequence and the time-stamped sequence at a certain moment is as follows: Synchronization sequence: T0=10:00:01.000, T1=10:00:02.000, T2=10:00:03.000.

[0049] Time-stamped sequence: PPS0=10:00:01.000, PPS1=10:00:02.000, PPS2=10:00:03.000.

[0050] All sensor data were precisely aligned, laying the foundation for subsequent data processing.

[0051] 2. Integral-Differential Modulation Process The synchronization sampling module performs integral-differential modulation on the mixed synchronization sequence, and the specific steps are as follows: S21. Sample the hybrid synchronization sequence to form a sampled data stream. The system samples the synchronized hybrid synchronization sequence. In one embodiment, the sampling frequency is 1 kHz, and each data point represents a sampled value of parameters such as capacitance, temperature, and humidity. Sampled data streams from voltage, current, and environmental parameter sensors are recorded. For example, the system may acquire voltage data streams of [5.6V, 5.7V, 5.8V, 5.9V, 6.0V], and also acquire current data streams, temperature data streams, etc., forming a multi-channel sampled data stream.

[0052] S22. Perform integral-differential modulation on the sampled data stream to generate a quantized bit stream. By performing an integration operation on the sampled data stream, the system generates a cumulative signal result by integrating each sample point, which helps improve signal smoothness and reduce high-frequency noise. After the integration operation, differential modulation is performed to calculate the difference between the current data point and the previous data point, reducing the impact of sampling deviations or noise.

[0053] In one embodiment, the voltage data in the sampled data stream are 5.6V, 5.7V, 5.8V, 5.9V, and 6.0V. After integration and differential calculations, a quantized bit stream of 110011, 110100, 110101, 110110, and 110111 is generated.

[0054] S23. Perform decimation processing on the quantized bitstream to generate high-resolution sampled data. The quantized bitstream undergoes decimation processing to reduce redundant data and improve data resolution. By retaining the most representative data points within each time period, decimation effectively reduces the amount of data while preserving the main features of the signal. In one embodiment, after decimation, 30 data points are selected from 100 data points in the original quantized bitstream, ultimately generating a streamlined and high-resolution sampled data stream.

[0055] The extracted high-resolution sampled data stream is 110011, 110101, 110111, 111000, 111001.

[0056] A high-resolution sampling dataset is generated, which will be used for subsequent feature extraction and analysis to improve the accuracy and reliability of capacitance measurements.

[0057] Example 3

[0058] In this embodiment, the adaptive correction module processes the initial feature parameter set through a residual feedback mechanism and generates residual feedback values. Then, it performs multi-model adaptive weighted fusion to generate the final corrected feature parameter set.

[0059] 1. Residual Feedback Mechanism The residual feedback mechanism generates residual feedback values ​​by calculating the difference between the initial set of characteristic parameters and the standard characteristic data. The standard characteristic data consists of historical measurement characteristic parameters stored in the capacitance measurement database, representing verified and calibrated capacitance parameters.

[0060] In one embodiment, the residual feedback value r is [0.1, 0.01, 0.2].

[0061] 2. Adaptive weighted fusion of multiple models The adaptive correction module performs multi-model adaptive weighted fusion based on this feedback value to generate a corrected feature parameter set. By weightedly combining the outputs of different models, a more accurate measurement result is obtained. The specific formula is as follows: in, To correct the feature parameter set, it is normalized to become dimensionless values. These are the characteristic parameters of the Kalman filter output. The feature parameters for deep learning predictions are normalized to become dimensionless values. For residual feedback value The dynamic weight function has a value range between 0 and 1 and is a dimensionless quantity. As the residual evaluation index, the normalization process adopts the residual normalization weighted method. The dynamic weight function is determined by the normalized residual. The weights are adjusted in real time according to the size of the residual and the sum of the weights is equal to 1.

[0062] In one embodiment, at a certain moment, the Kalman filter model output is: =[5.55,0.11,27.95].

[0063] =[5.60,0.12,28.05].

[0064] The residual feedback value r = [0.1, 0.01, 0.2].

[0065] Weighting function Adaptive adjustment based on residual magnitude. =0.7.

[0066] Corrected feature parameter set Will be: The calculation result is: 3. Generate the corrected feature parameter set Through the adaptive weighted fusion steps described above, the final corrected feature parameter set is obtained. This provides high-precision capacitance measurement results. The corrected characteristic parameters will be used in subsequent analysis and decision-making processes, improving the system's accuracy in predicting capacitance performance.

[0067] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A capacitance measurement system based on big data multi-source computing, characterized in that, include: The multi-source data acquisition module generates multi-source heterogeneous data by acquiring data from the capacitor in real time through multi-source sensors. The synchronous sampling module generates a hybrid synchronization sequence by uniformly synchronizing the multi-source heterogeneous data through a hybrid synchronization mechanism, and the synchronous sampling module performs integral-differential modulation on the hybrid synchronization sequence to obtain high-resolution sampling data. The feature processing module includes a feature extraction unit and a parameter aggregation unit. The feature extraction unit performs time-frequency analysis on the high-resolution sampled data to extract transient response characteristic parameters of the capacitor. The parameter aggregation unit integrates the transient response characteristic parameters of the capacitor to generate an initial feature parameter set. The adaptive correction module processes the initial feature parameter set through a residual feedback mechanism to generate residual feedback values. Based on the residual feedback values, the adaptive correction module performs multi-model adaptive weighted fusion to generate a corrected feature parameter set. The output module performs unified calculations on the modified feature parameter set through a big data computing platform and outputs the capacitance measurement results.

2. The capacitance measurement system based on big data multi-source computing according to claim 1, characterized in that: The multi-source sensor includes a voltage sensor, a current sensor, and an environmental parameter sensor, and the multi-source heterogeneous data includes voltage data, current data, temperature data, and humidity data.

3. The capacitance measurement system based on big data multi-source computing according to claim 1, characterized in that: The hybrid synchronization mechanism includes a PTP mechanism and a 1PPS mechanism. The synchronization sampling module uses the PTP mechanism to synchronize the clocks of network nodes in the multi-source heterogeneous data to generate a synchronization sequence. The synchronization sampling module uses the 1PPS mechanism to time-stamp the multi-source heterogeneous data to generate a time-stamped sequence. The synchronization sequence and the time-stamped sequence are fused to generate the hybrid synchronization sequence.

4. The capacitance measurement system based on big data multi-source computing according to claim 1, characterized in that: The integral-differential modulation includes the following steps: S21. Sample the hybrid synchronization sequence to form a sampled data stream; S22. Perform integral-differential modulation on the sampled data stream to generate a quantized bit stream; S23. The quantized bit stream is extracted to generate the high-resolution sampled data.

5. The capacitance measurement system based on big data multi-source computing according to claim 1, characterized in that: The transient response characteristic parameters of the capacitor include voltage and current phase difference parameters, equivalent series resistance parameters, and dielectric loss parameters.

6. The capacitance measurement system based on big data multi-source computing according to claim 1, characterized in that: The residual feedback mechanism generates the residual feedback value by calculating the difference between the initial feature parameter set and the standard feature data. The standard feature data consists of historical measurement feature parameters stored in the capacitance measurement database. The model formula for the multi-model adaptive weighted fusion is as follows: ， in, For the set of corrected feature parameters, These are the characteristic parameters of the Kalman filter output. To predict and output feature parameters for deep learning, For residual feedback value The dynamic weighting function.

7. The capacitance measurement system based on big data multi-source computing according to claim 1, characterized in that: The big data computing platform includes a data computing unit and a data storage unit. The data computing unit performs parallel calculations on the modified feature parameter set to generate the capacitance measurement results, and the data storage unit stores the capacitance measurement results in a structured manner.

8. A capacitance measurement method based on big data multi-source computing, characterized in that, include: S1. Real-time acquisition of the capacitor data using multi-source sensors to generate multi-source heterogeneous data, wherein the multi-source sensors include voltage sensors, current sensors, and environmental parameter sensors. S2. A hybrid synchronization sequence is generated by uniformly synchronizing the multi-source heterogeneous data through a hybrid synchronization mechanism, and high-resolution sampling data is obtained by integral-differential modulation of the hybrid synchronization sequence. The hybrid synchronization mechanism includes PTP and 1PPS. S3. Perform time-frequency analysis on the high-resolution sampled data to extract capacitor transient response characteristic parameters, and integrate the capacitor transient response characteristic parameters to generate an initial feature parameter set. The capacitor transient response characteristic parameters include voltage and current phase difference parameters, equivalent series resistance parameters, and dielectric loss parameters. S4. The initial feature parameter set is compared to generate residual feedback values. The comparison is performed using a residual feedback mechanism. The residual feedback values ​​are then subjected to multi-model adaptive weighted fusion to generate a corrected feature parameter set. S5. The modified feature parameter set is uniformly calculated and the capacitance measurement results are output through a big data computing platform.

Images