Distributed photovoltaic data acquisition method and device based on time synchronization

Abstract

The invention relates to the technical field of data acquisition, and discloses a distributed photovoltaic data acquisition method and device based on time synchronization. The method comprises the following steps: performing distributed transmission processing on nanosecond synchronous acquisition reference signals through a time delay compensation network to obtain trigger synchronization signals of a plurality of acquisition nodes, and performing data acquisition processing through a multi-channel parallel sampling circuit to obtain original voltage and current data; and carrying out data verification processing on the original voltage and current data through a digital filtering algorithm to obtain effective acquisition data, carrying out string matching analysis processing to obtain string feature distribution, and carrying out data screening processing through a state evaluation algorithm to obtain real-time performance data of the photovoltaic module. Through high-precision clock synchronization and parallel sampling technologies, nanosecond sampling synchronization of a plurality of acquisition nodes is realized, and the acquisition data is ensured to have a strict time sequence corresponding relation, so that an accurate data basis is provided for performance evaluation and fault diagnosis of the photovoltaic module.

Description

Technical Field

[0001] This application relates to the field of data acquisition, and in particular to a distributed photovoltaic data acquisition method and device based on time synchronization. Background Art

[0002] The operation monitoring of a distributed photovoltaic power generation system requires real-time acquisition and analysis of the operation parameters of each photovoltaic module. The existing photovoltaic data acquisition methods mainly include two technical solutions: centralized acquisition and time-sharing sampling. Centralized acquisition sets up data acquisition devices at the busbar boxes and polls and samples parameters such as the voltage and current of each photovoltaic module through communication methods such as RS485 or CAN bus. Some systems use GPS timing technology as the time reference to attempt to achieve time synchronization of multiple acquisition devices. At the same time, some systems use the time-sharing sampling method to sequentially acquire data from different modules according to a preset time sequence, and analyze and evaluate the acquisition results through data processing algorithms. These technical solutions have been widely used in the actual operation monitoring of photovoltaic power stations.

[0003] However, the existing technologies have the following deficiencies: Due to the lack of an accurate time synchronization mechanism, there are time differences at the microsecond to millisecond level in the sampling moments of each acquisition point, resulting in a lack of a strict time sequence correspondence relationship for the acquired data. This time sequence inconsistency makes it impossible to accurately analyze the working state differences of different photovoltaic modules at the same moment. Especially in dynamic working conditions such as rapid changes in light intensity or local shading, the time deviation of the sampling data will cause large errors in the component performance evaluation results. In addition, the time-sharing sampling method is also difficult to meet the requirements of large-scale photovoltaic power stations for high spatio-temporal resolution data acquisition. The time differences in the sampling data lead to a lack of reliability in the comparison of component performances. At the same time, the existing data processing methods often ignore the influence of the sampling time sequence and are difficult to accurately reflect the true operating state of the components. Summary of the Invention

[0004] This application provides a distributed photovoltaic data acquisition method and device based on time synchronization, which is used to achieve nanosecond-level sampling synchronization of multiple acquisition nodes through high-precision clock synchronization and parallel sampling technologies, ensuring that the acquired data has a strict time sequence correspondence relationship, thereby providing an accurate data basis for the performance evaluation and fault diagnosis of photovoltaic modules.

[0005] In a first aspect, the present application provides a method for collecting distributed photovoltaic data based on time synchronization. The method for collecting distributed photovoltaic data based on time synchronization includes: performing clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronous acquisition reference signal; performing distributed transmission processing on the synchronous acquisition reference signal through a time delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes; performing data acquisition processing on the photovoltaic module parameters of the multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain raw voltage and current data with timestamps; performing data verification processing on the raw voltage and current data through a digital filtering algorithm to obtain effectively acquired data with normalized time series; performing string matching analysis processing on the effectively acquired data through a data clustering algorithm to obtain the string feature distribution of the acquired data; and performing data screening processing on the string feature distribution through a state evaluation algorithm to obtain the real-time performance data of the photovoltaic modules.

[0006] In a second aspect, the present application provides a device for collecting distributed photovoltaic data based on time synchronization. The device for collecting distributed photovoltaic data based on time synchronization includes:

[0007] a synchronization module, configured to perform clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronous acquisition reference signal;

[0008] a transmission module, configured to perform distributed transmission processing on the synchronous acquisition reference signal through a time delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes;

[0009] an acquisition module, configured to perform data acquisition processing on the photovoltaic module parameters of the multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain raw voltage and current data with timestamps;

[0010] a verification module, configured to perform data verification processing on the raw voltage and current data through a digital filtering algorithm to obtain effectively acquired data with normalized time series;

[0011] an analysis module, configured to perform string matching analysis processing on the effectively acquired data through a data clustering algorithm to obtain the string feature distribution of the acquired data;

[0012] a screening module, configured to perform data screening processing on the string feature distribution through a state evaluation algorithm to obtain the real-time performance data of the photovoltaic modules.

[0013] In the technical solution provided by this application, the GPS reference time signal is processed through a dynamic clock calibration algorithm, achieving a clock synchronization accuracy at the nanosecond level and avoiding the data distortion problem caused by clock deviation in traditional acquisition systems. Through a delay compensation network for distributed transmission processing, the transmission delay differences between different acquisition nodes are effectively eliminated, ensuring the phase consistency of the trigger synchronization signals of each node. A multi-channel parallel sampling circuit is used for data acquisition, combined with high-precision voltage and current sensors, to achieve synchronous sampling of component parameters and ensure the timing correlation of the sampled data. The digital filtering algorithm verifies and processes the original data, effectively filtering out the sampling noise, improving the signal-to-noise ratio of the data. At the same time, through timing normalization processing, the validity and comparability of the data are ensured. The data clustering algorithm performs string matching analysis on the acquired data, accurately identifying the strings with abnormal performance, providing a reliable basis for fault diagnosis. The state evaluation algorithm filters the data on the string feature distribution, realizing the real-time evaluation of the performance of photovoltaic modules and providing data support for operation and maintenance decisions. The entire solution realizes the full-process processing from data acquisition to performance evaluation, overcoming problems such as timing chaos, data distortion, and analysis lag in traditional acquisition systems, and significantly improving the operation and maintenance efficiency and fault diagnosis ability of photovoltaic power stations. Through high-precision time synchronization and data processing technologies, this solution not only ensures the accuracy of the acquired data but also realizes the precise evaluation of the operating state of photovoltaic modules through systematic data analysis methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0015] Figure 1 FIG. is a schematic diagram of an embodiment of the distributed photovoltaic data acquisition method based on time synchronization in an embodiment of this application;

[0016] Figure 2 FIG. is a schematic diagram of an embodiment of the distributed photovoltaic data acquisition device based on time synchronization in an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The embodiments of the present application provide a method and device for distributed photovoltaic data acquisition based on time synchronization. Terms such as "first", "second", "third", "fourth", etc. (if any) in the specification, claims, and the above-mentioned drawings of the present application are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that such data used can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0018] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the method for distributed photovoltaic data acquisition based on time synchronization in the embodiments of the present application includes:

[0019] Step S101, perform clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronous acquisition reference signal;

[0020] Step S102, perform distributed transmission processing on the synchronous acquisition reference signal through a delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes;

[0021] Step S103, perform data acquisition processing on the photovoltaic module parameters of multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain raw voltage and current data with timestamps;

[0022] Step S104, perform data verification processing on the raw voltage and current data through a digital filtering algorithm to obtain effectively acquired data with normalized time series;

[0023] Step S105, perform string matching analysis processing on the effectively acquired data through a data clustering algorithm to obtain the string feature distribution of the acquired data;

[0024] Step S106, perform data screening processing on the string feature distribution through a state evaluation algorithm to obtain real-time performance data of the photovoltaic modules.

[0025] It can be understood that the execution entity of the present application can be a distributed photovoltaic data acquisition device based on time synchronization, or a terminal or a server. Specifically, it is not limited here. The embodiments of the present application will be described by taking the server as the execution entity as an example.

[0026] Specifically, in clock synchronization processing, the GPS receiving module receives GPS satellite signals to obtain accurate UTC time information. The dynamic clock calibration algorithm calculates the frequency deviation by comparing the phase difference between the GPS second pulse and the local crystal oscillator signal. Specifically, let the arrival time of the GPS second pulse be t1, and the corresponding second pulse time of the local crystal oscillator be t2, then the phase difference δt = t1 - t2. According to the magnitude of the phase difference, different calibration strategies are adopted: when |δt| > 1 μs, a large step size is used for rapid adjustment; when |δt| ≤ 1 μs, a small step size is used for fine adjustment, and finally the phase difference is controlled within 10 ns. In the time delay compensation network processing, first, the delays of each transmission path are measured. For a transmission path with a distance of L, its theoretical delay τ = L / v, where v is the signal propagation speed. The actual delay is obtained through on-line measurement and compensated by a programmable delay unit. For example, if a certain acquisition node is 100 meters away from the master station, the theoretical delay is about 500 ns, and the measured delay is 520 ns, then a compensation value of -20 ns needs to be set.

[0027] Data acquisition processing uses a high-precision 24-bit Σ-Δ ADC for parallel sampling. For voltage signals, a resistor network with a voltage division ratio of R2 / (R1 + R2) is used for conditioning to make the sampling voltage within the ADC range. Current signals are sampled after being converted into voltage signals by Hall sensors. Each sampled data is appended with a microsecond-level timestamp to record the absolute time of the sampling moment. Data verification processing first checks the timing relationship of the sampled data. The data is smoothed by a linear-phase FIR filter, the filter order N = 64, and the cut-off frequency fc is set to 1 / 4 of the sampling frequency. At the same time, range normalization is performed to unify the data ranges of all channels to the [0, 1] interval.

[0028] In the string matching and analysis processing, first, the voltage and current data of each string are subjected to feature extraction, including statistical quantities such as the average value and standard deviation. Then, the strings are divided into different categories through the K-means clustering algorithm, and the number of clustering centers k is adaptively determined according to the total number of strings, generally taking k = √n / 2, where n is the total number of strings. The final status evaluation processing performs anomaly detection by setting thresholds. Normal operating ranges are set for the voltage and current parameters: the voltage deviation does not exceed ±5%, and the current deviation does not exceed ±10%. Data outside the range is marked as an anomaly point, and strings with consecutive anomalies will be determined to have abnormal performance.

[0029] For example, the system synchronously samples a photovoltaic array with 160 acquisition points at a sampling frequency of 1 kHz. Through clock synchronization and delay compensation, the timing deviation of each acquisition point is controlled within ±5 ns. The 24-bit ADC data collected is classified for the strings through FIR filtering and normalization processing, and then K-means clustering (k = 6) is used. Finally, 3 voltage anomaly strings (deviation > 5%) and 2 current anomaly strings (deviation > 10%) are identified, providing accurate positioning information for subsequent maintenance.

[0030] In the embodiment of the present application, the GPS reference time signal is processed through a dynamic clock calibration algorithm, achieving a clock synchronization accuracy at the nanosecond level and avoiding the data distortion problem caused by clock deviation in traditional acquisition systems. Through a delay compensation network for distributed transmission processing, the transmission delay difference between different acquisition nodes is effectively eliminated, ensuring the phase consistency of the trigger synchronization signals of each node. A multi-channel parallel sampling circuit is used for data acquisition, combined with high-precision voltage and current sensors, to achieve synchronous sampling of component parameters and ensure the timing correlation of the sampled data. The digital filtering algorithm verifies and processes the original data, effectively filtering out sampling noise, improving the signal-to-noise ratio of the data. At the same time, through timing normalization processing, the validity and comparability of the data are ensured. The data clustering algorithm performs string matching analysis on the collected data, accurately identifying strings with abnormal performance and providing a reliable basis for fault diagnosis. The state evaluation algorithm screens the data on the string feature distribution, realizing the real-time evaluation of the performance of photovoltaic modules and providing data support for operation and maintenance decisions. The entire solution realizes the full-process processing from data acquisition to performance evaluation, overcomes problems such as chaotic timing, data distortion, and analysis lag in traditional acquisition systems, and significantly improves the operation and maintenance efficiency and fault diagnosis ability of photovoltaic power stations. Through high-precision time synchronization and data processing technologies, this solution not only ensures the accuracy of the collected data but also realizes the precise evaluation of the operating state of photovoltaic modules through systematic data analysis methods.

[0031] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0032] (1) Demodulate the received GPS satellite signal to obtain a second pulse signal and serial port time code information;

[0033] (2) Compare the phases of the second pulse signal and the output signal of the local crystal oscillator to obtain a clock frequency deviation value;

[0034] (3) Rapidly capture the clock frequency deviation value through a segmented adjustment coefficient to obtain an initial frequency calibration value;

[0035] (4) Fine-tune and track the initial frequency calibration value through small-step iteration to obtain stable calibration parameters;

[0036] (5) Perform frequency correction processing on the local crystal oscillator through stable calibration parameters to obtain a 10 MHz reference clock signal;

[0037] (6) Perform synchronous pulse generation processing on the 10 MHz reference clock signal through counting and frequency division to obtain a nanosecond-level synchronous acquisition reference signal.

[0038] Specifically, the GPS receiving module receives satellite signals at 1575.42 MHz in the L1 band, and obtains two types of key information through frequency conversion and demodulation processing: second pulse signals and serial port time code information. The second pulse signal generates an accurate pulse per second, which is used as the time synchronization reference; the serial port time code contains complete UTC time information, which is used for absolute time calibration. During the clock synchronization process, a temperature-compensated crystal oscillator is used as the local clock source, and its frequency stability reaches one in a hundred thousand. By comparing the arrival times of the GPS second pulse signal and the output signal of the local crystal oscillator, the phase difference between the two is calculated. This phase difference reflects the offset degree of the local clock relative to the GPS time reference, and is an important basis for subsequent calibration.

[0039] The clock calibration adopts a segmented adjustment strategy, and different adjustment coefficients are selected according to the size of the phase difference in the fast capture stage. When the phase difference is greater than 1 microsecond, a larger adjustment step is adopted to quickly narrow the clock gap; when the phase difference is between 100 nanoseconds and 1 microsecond, a medium adjustment step is adopted to avoid oscillations caused by over-adjustment. Through this segmented adjustment method, the local clock frequency quickly approaches the target value. After obtaining the preliminary frequency calibration value, it enters the fine tracking stage. In this stage, a smaller adjustment step is adopted, and the clock parameters are gradually optimized through multiple iterations. The current phase difference is measured in each iteration, and the frequency value is finely adjusted accordingly until the phase difference is stabilized within 10 nanoseconds. The stable calibration parameters generated in this process will be used for continuous calibration of the local crystal oscillator.

[0040] The calibrated local crystal oscillator outputs a stable 10 MHz reference clock signal, which has extremely high frequency accuracy and stability. Finally, the reference clock signal is divided by a counting and frequency division circuit to generate a synchronous sampling trigger signal with the required frequency. For example, to achieve a sampling rate of 1 kHz, set a 10,000 frequency division, and a sampling trigger signal with a period of 1 millisecond can be obtained.

[0041] For example: In a data acquisition system of a certain photovoltaic power station, after the GPS receiving module demodulates to obtain an accurate second pulse signal, the local clock phase difference is first measured to be 2.5 microseconds. The fast capture program is started and large-step adjustments are made. After about 10 adjustments, the phase difference is reduced to 200 nanoseconds. Subsequently, it enters the fine tracking stage, and small-step adjustments are used. After about 20 more iterations, the phase difference is finally stabilized within 8 nanoseconds. Such time synchronization accuracy ensures that the acquisition nodes distributed at different positions can perform data sampling at almost the same moment, providing a reliable timing basis for subsequent data analysis. The 1 kHz sampling trigger signal obtained through counting and frequency division is distributed to each acquisition node, achieving strict synchronization of the sampling moments, and the sampling moment difference between adjacent acquisition points is controlled within 10 nanoseconds.

[0042] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0043] (1) Measure the transmission delay of the synchronous acquisition reference signal on each transmission path to obtain the path delay value;

[0044] (2) Configure the compensation parameters for the path delay value through a programmable delay unit to obtain the compensation coefficients for each path;

[0045] (3) Delay-calibrate the allocated signals of each acquisition node through the compensation coefficients to obtain the phase adjustment signal;

[0046] (4) Re-form the phase adjustment signal through a low-jitter buffer amplification circuit to obtain a multi-channel buffer output;

[0047] (5) Lock the multi-channel buffer output through a phase-locked loop circuit to obtain the trigger synchronization signals for multiple acquisition nodes.

[0048] Specifically, accurately measure the delay values of each transmission path. The measurement method uses the two-way time transfer technology: send a synchronous signal to the target acquisition node and record the sending moment. After the acquisition node receives the signal, it immediately returns a response signal. After the master station receives the response signal, it records the receiving moment. The signal transmission delay is calculated through the round-trip time. For example, for a 100-meter transmission path, if the measured round-trip time is 1000 nanoseconds, the one-way transmission delay is 500 nanoseconds. After obtaining the path delay value, perform delay compensation through a programmable delay unit. The delay unit uses a digitally adjustable delay line with an adjustment accuracy of up to 100 picoseconds. The configuration of the compensation parameters is based on the measured delay value. For the 500-nanosecond delay of the aforementioned 100-meter transmission path, a compensation value of -500 nanoseconds is set at the signal sending end to align the signal arrival moment at the target node with the reference moment.

[0049] During the delay calibration process, corresponding compensation coefficients are applied to the allocated signals of each acquisition node. The compensation circuit stores the compensation values of each path in a look-up table manner and selects the corresponding compensation coefficient according to the number of the target node for delay adjustment. In this way, the synchronous signals received by all acquisition nodes are aligned in time. To ensure signal quality, a low-jitter buffer amplifier circuit is used to shape the adjusted signal. The jitter index of the buffer amplifier circuit is better than 10 picoseconds, and the differential signal transmission method is used to suppress common-mode interference. After signal shaping, the output signal has stable amplitude and phase characteristics.

[0050] Finally, the phase-locked loop circuit is used to lock the shaped signal. The loop bandwidth of the phase-locked loop is set to 1 kHz, which can not only track the slow-changing phase drift but also effectively suppress high-frequency jitter. The trigger synchronization signal output by the phase-locked loop has extremely high phase stability, and the trigger time difference between adjacent acquisition nodes is controlled at the nanosecond level.

[0051] Taking the on-site application of a certain photovoltaic power station as an example: There are 160 acquisition nodes in this power station, and the farthest transmission distance is 200 meters. The delays of each path are measured through two-way time transmission. The shortest path delay is 100 nanoseconds, and the longest path delay is 1000 nanoseconds. The delay unit configures compensation parameters for each path respectively, and the maximum compensation value is -1000 nanoseconds. After compensation and signal shaping, an oscilloscope is used to measure the phase difference of the trigger signals of adjacent acquisition nodes. The test results show that the maximum phase difference does not exceed 5 nanoseconds, which proves that this compensation scheme can effectively solve the synchronization deviation problem caused by transmission delay. The above delay compensation process not only overcomes the time difference caused by signal transmission but also ensures the signal quality through means such as low-jitter buffer amplification and phase-locked loop locking, laying a solid foundation for subsequent synchronous data acquisition. The entire scheme adopts a fully digital processing method, with high reliability and stability.

[0052] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0053] (1) The voltage signal of the photovoltaic module is subjected to voltage division processing through a high-impedance voltage sampling circuit to obtain a normalized voltage signal;

[0054] (2) The current signal of the photovoltaic module is subjected to current sampling processing through a Hall sensor to obtain a normalized current signal;

[0055] (3) The normalized voltage signal and the normalized current signal are subjected to signal conditioning processing through an anti-aliasing filter to obtain a filtered sampling signal;

[0056] (4) The filtered sampling signal is subjected to high-precision quantization processing through a 24-bit Σ-Δ analog-to-digital converter to obtain digital quantization data;

[0057] (5) Time - stamp the digital quantization data through a clock counter to obtain the original voltage - current data with time stamps.

[0058] Specifically, for voltage sampling, a high - impedance voltage sampling circuit is adopted, which is composed of a precision resistor voltage - dividing network. For a photovoltaic module with a rated voltage of 40V, a voltage - dividing network with a voltage - division ratio of 1:10 is formed by using high - precision resistors of 100kΩ and 10kΩ, converting the 40V input voltage into a 4V sampling voltage. The voltage - dividing resistors are precision resistors with a temperature coefficient less than 5ppm / °C to ensure that the measurement accuracy is not affected by temperature. For current sampling, Hall - sensor technology is adopted. The range of the sensor is selected as 0 - 20A, and the output sensitivity is 100mV / A. The Hall sensor has good linearity and temperature stability, with a zero - point drift less than 0.1% and a full - scale accuracy better than 0.5%. When the output current of the photovoltaic module is 10A, the Hall sensor outputs a corresponding voltage signal of 1V.

[0059] Signal conditioning is pre - processed by an anti - aliasing filter. The filter adopts a third - order Butterworth low - pass structure. Considering the variation characteristics of photovoltaic module parameters, the cut - off frequency of the filter is set to 1 / 4 of the sampling frequency. Taking a sampling rate of 1kHz as an example, the cut - off frequency of the filter is set to 250Hz, which has a flat amplitude - frequency characteristic in the pass - band and a good attenuation effect on high - frequency interference. The signal ripple after being processed by the filter is effectively suppressed. High - precision quantization is achieved by a 24 - bit Σ - Δ analog - to - digital converter. Its full - scale input range is ±5V, and the effective number of bits reaches 20 bits. The integration time of the converter is set to 20ms, which can effectively suppress power - frequency interference. For a 4V voltage signal and a 1V current signal, changes in the order of one - millionth can be distinguished respectively. The converter adopts a differential input mode, and the common - mode rejection ratio reaches 120dB, effectively reducing the influence of external interference.

[0060] The time - stamping process uses a high - precision clock counter. The reference clock frequency of the counter is 10MHz, providing a time resolution of 100ns. Each sampled data is appended with a 32 - bit time stamp, recording the exact interval between the sampling moment and the reference time point.

[0061] Taking the actual acquisition process as an example: A certain photovoltaic module outputs a voltage of 35.6V under standard conditions, which is converted into a sampling voltage of 3.56V through a voltage division network; the output current is 8.5A, which is converted into a sampling voltage of 0.85V through a Hall sensor. These signals are processed by an anti-aliasing filter to eliminate high-frequency interference components. A 24-bit ADC quantifies the processed signals, and the digital quantity output by the voltage channel is 7,864,321, and the digital quantity output by the current channel is 1,876,543. Each data is attached with a timestamp relative to UTC time, and the time resolution is 100ns. During the sampling process, the time interval between two adjacent sampling points is stable within the range of 1.000ms ± 100ns, demonstrating the timing accuracy of sampling. The entire sampling and processing link realizes high-precision synchronous acquisition of the parameters of the photovoltaic module through means such as high-impedance sampling, precise sensing, anti-aliasing filtering, high-resolution quantization, and accurate time stamping. The sampled data has high accuracy and reliable timing characteristics.

[0062] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0063] (1) Perform timing comparison processing on the timestamps of the original voltage and current data to obtain a data synchronization deviation value;

[0064] (2) Perform data screening processing on the data synchronization deviation value and a preset threshold to obtain timing valid data;

[0065] (3) Perform signal smoothing processing on the timing valid data through a linear-phase FIR filter to obtain filtered output data;

[0066] (4) Perform amplitude normalization processing on the filtered output data to obtain range-normalized data;

[0067] (5) Perform timing regularization processing on the range-normalized data through data reorganization to obtain effectively acquired data with regularized timing.

[0068] Specifically, in the time series comparison process, the timestamps of the data of each acquisition node are compared, and the node with the highest sampling rate is selected as the reference, and its sampling period is 1 millisecond. By calculating the time deviation of other nodes relative to the reference node, the data synchronization deviation value is obtained. For example, if the sampling timestamp of the reference node at a certain moment is 10,000,000 nanoseconds, and the corresponding sampling timestamps of other nodes are 10,000,005 nanoseconds and 9,999,995 nanoseconds respectively, the corresponding time deviation values are +5 nanoseconds and -5 nanoseconds. In the data screening stage, a strict time series validity judgment standard is set, and the preset synchronization deviation threshold is ±100 nanoseconds. The time deviation of all acquired data is judged, and the data with deviation exceeding the threshold range is marked as invalid data and excluded. Taking 160 acquisition nodes as an example, in a certain acquisition, it is found that the time deviation of 3 nodes exceeds 100 nanoseconds, and the data of these nodes are excluded, and the valid data of 157 nodes are retained.

[0069] The signal smoothing process uses a 128th-order linear-phase FIR filter, and its cut-off frequency is 0.25 times the sampling frequency. The filter performs a moving average process on the input data, and each output point is obtained by weighted averaging of 128 adjacent input data. Taking voltage data as an example, the original data fluctuates in the range of ±0.2V near 35.6V, and after filtering, the fluctuation is reduced to ±0.05V, while maintaining the phase characteristics of the signal without distortion. The amplitude normalization process unifies the voltage and current data of different ranges into a standard interval. The full scale of the voltage signal is 60V, and the full scale of the current signal is 20A. The normalization process divides the actual sampling value by its respective full scale value to obtain the normalized data between 0 and 1. For example, the normalized value of the voltage value of 35.6V is 0.593, and the normalized value of the current value of 8.5A is 0.425.

[0070] The time series regularization process rearranges the normalized data according to a unified time reference. First, a reference time series is established with a time interval of 1 millisecond. Then, the normalized data of each node is aligned to the closest reference time point. For the data between the reference points, linear interpolation is used to fill it to ensure that all node data has a unified sampling interval. In practical applications, 160 acquisition nodes of a certain photovoltaic power station respectively acquired 1000 data points within 1 second. After time series verification, it is found that more than 95% of the data points have a time deviation within 50 nanoseconds, and the maximum deviation does not exceed 80 nanoseconds. The filtering process reduces the measurement noise by 75%, while maintaining the dynamic response characteristics of the data. The normalized data is convenient for comparative analysis between different parameters, and the time series regularization ensures the equal time interval characteristics of the data. The entire processing process ensures the time series accuracy and numerical reliability of the acquired data, providing high-quality basic data for subsequent data analysis.

[0071] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0072] (1) Process the effectively collected data with time series normalization by grouping the data according to the string number to obtain grouped string data;

[0073] (2) Extract features from the voltage and current parameters in the grouped string data to obtain string feature parameters;

[0074] (3) Perform statistical distribution calculation on the string feature parameters to obtain the parameter distribution interval;

[0075] (4) Classify the strings by K-means clustering on the parameter distribution interval to obtain string class identifiers;

[0076] (5) Perform string association processing on the string class identifiers through correlation analysis to obtain the string feature distribution.

[0077] Specifically, first perform data grouping processing. Taking a single photovoltaic string as a unit, divide the data groups. For a photovoltaic array with 160 acquisition points, group it according to 20 strings, and each string contains data of 8 photovoltaic modules. Each group of data contains voltage and current values of 1000 sampling points, forming a complete string operation data set. Feature extraction processing calculates multiple feature quantities for the voltage and current parameters of each string. Including statistical features such as mean value, standard deviation, maximum value, minimum value, and coefficient of variation. Taking a certain string as an example, the average voltage of 8 modules is 35.6V, the standard deviation is 0.4V, the maximum value is 36.2V, the minimum value is 35.1V, and the coefficient of variation is 1.12%. The average value of the current parameter is 8.5A, the standard deviation is 0.15A, the maximum value is 8.8A, the minimum value is 8.2A, and the coefficient of variation is 1.76%.

[0078] Statistical distribution calculation is based on the extracted feature parameters to establish a parameter distribution histogram. Divide the voltage parameters into 20 intervals and count the data distribution in each interval. The voltage distribution interval is 34V - 37V, and the data in the interval of 35.5V - 35.8V has the highest proportion, reaching 35%. The current distribution interval is 8.0A - 9.0A, and the data in the interval of 8.4A - 8.6A has the highest proportion, reaching 40%. String classification uses the K-means clustering algorithm, and the number of clustering centers is selected as 4 according to the number of strings. The clustering features include key parameters such as voltage average value, current average value, and power output. After iterative calculation, the 20 strings are divided into 4 categories: 8 optimal operation strings, 6 strings with slightly decreased performance, 4 strings with significantly decreased performance, and 2 abnormal operation strings. Each string is marked with the corresponding category number 1 - 4.

[0079] String correlation analysis studies the mutual relationships between different categories of strings. By calculating the parameter correlation coefficients between strings, a correlation matrix is established. Strings with a correlation coefficient greater than 0.9 are determined to be strongly correlated, indicating similar operating states. Taking three strings numbered 5, 6, and 7 as an example, their voltage correlation coefficients are all greater than 0.95, and the current correlation coefficients are greater than 0.92, indicating that these three strings are in similar operating states.

[0080] In practical applications, the operation data of 20 strings in a certain photovoltaic power station after the above processing shows that: the voltage parameter distributions of 12 strings are concentrated in the range of 35.5V - 36.0V, which belongs to the normal operation range; the voltage distributions of 5 strings are in the range of 35.0V - 35.5V, showing slight attenuation; the voltages of 2 strings are lower than 35.0V, and their correlations with adjacent strings are relatively low, which need to be focused on; the voltage fluctuation range of 1 string exceeds 1V, which is determined to be in an abnormal operating state. These analysis results intuitively reflect the operating conditions of each string, providing data support for operation and maintenance decisions. Through this multi-level data analysis method, abnormal strings in the photovoltaic array are accurately identified, improving the operation and maintenance efficiency.

[0081] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0082] (1) Perform parameter threshold detection processing on the string feature distribution to obtain abnormal feature marks;

[0083] (2) Perform fault type identification processing on the abnormal feature marks to obtain fault category data;

[0084] (3) Perform statistical frequency analysis processing on the fault category data to obtain the fault probability distribution;

[0085] (4) Perform association rule extraction processing on the fault probability distribution to obtain state feature indicators;

[0086] (5) Perform performance parameter calculation processing on the state feature indicators to obtain the real-time performance data of the photovoltaic modules.

[0087] Specifically, parameter threshold detection first sets multi-level thresholds for core parameters such as voltage, current, and power. The voltage parameter is set with three levels of thresholds: the normal operation range is 35.5V - 36.5V, the slightly abnormal range is 34.5V - 35.5V, and the severely abnormal range is less than 34.5V. The current parameter is set with corresponding thresholds: the normal range is 8.0A - 9.0A, the slightly abnormal range is 7.0A - 8.0A, and the severely abnormal range is less than 7.0A. Threshold judgment is performed on the measured data to generate abnormal feature marks. Fault type recognition is based on the combined features of the abnormal feature marks. A low voltage and normal current are marked as a component attenuation fault; a normal voltage but low current is marked as a series fault; both low voltage and low current are marked as an occlusion fault; excessive parameter fluctuation is marked as a contact fault. Taking a certain string as an example, the measured voltage is 34.2V and the current is 8.5A, and it is determined as a component attenuation fault according to the feature combination.

[0088] Fault frequency analysis statistically analyzes the occurrence frequencies of various faults. Among the long-term monitoring data of 20 strings, the component attenuation fault accounts for 25%, the series fault accounts for 15%, the occlusion fault accounts for 10%, the contact fault accounts for 5%, and the normal operation accounts for 45%. By statistically analyzing the time distribution of each type of fault, it is found that the occlusion fault mostly occurs in the morning and evening, and the contact fault is highly frequent during periods of drastic temperature change. Association rule extraction analyzes the correlation between fault types and environmental parameters. By analyzing environmental factors such as temperature, humidity, and light at the time of fault occurrence, association rules are extracted. When the environmental temperature exceeds 45 degrees and lasts for more than 4 hours, the probability of component attenuation fault increases by 30%; when the humidity exceeds 85%, the probability of contact fault increases by 25%.

[0089] Performance parameter calculation evaluates the component performance based on state characteristic indicators. The actual power generation efficiency of each string is calculated and compared with the theoretical power generation efficiency to obtain the performance attenuation rate. The performance attenuation rate of normally operating strings is within 5%, the attenuation rate of slightly abnormal strings is between 5% - 10%, and the attenuation rate of severely abnormal strings exceeds 10%. In actual operation, a certain photovoltaic power station continuously monitors 20 strings for 30 days. Through threshold detection, it is found that the voltages of 3 strings continuously remain lower than 34.5V, and after fault identification, they are determined as component attenuation faults. The fault occurrence frequency of these strings increases significantly during high-temperature periods, and the correlation coefficient with the environmental temperature reaches 0.85. According to performance calculation, the actual power generation efficiency of these strings is 12% lower than the theoretical value, which has exceeded the severely abnormal threshold. At the same time, it is found that 2 strings have periodic occlusion faults in the early morning. After on-site investigation, it is confirmed that they are shaded by the shadows of nearby buildings, and then the bracket angle is adjusted to eliminate the occlusion effect. This systematic fault diagnosis method effectively improves the accuracy and efficiency of power station operation and maintenance.

[0090] The above describes the method for distributed photovoltaic data collection based on time synchronization in the embodiments of the present application. Next, the device for distributed photovoltaic data collection based on time synchronization in the embodiments of the present application will be described. Please refer to Figure 2 In one embodiment, the device for distributed photovoltaic data collection based on time synchronization in the embodiments of the present application includes:

[0091] A synchronization module 201, configured to perform clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronous acquisition reference signal;

[0092] A transmission module 202, configured to perform distributed transmission processing on the synchronous acquisition reference signal through a delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes;

[0093] An acquisition module 203, configured to perform data acquisition processing on the photovoltaic module parameters of the multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain raw voltage and current data with timestamps;

[0094] A verification module 204, configured to perform data verification processing on the raw voltage and current data through a digital filtering algorithm to obtain effectively acquired data with normalized timings;

[0095] An analysis module 205, configured to perform string matching analysis processing on the effectively acquired data through a data clustering algorithm to obtain the string feature distribution of the acquired data;

[0096] A screening module 206, configured to perform data screening processing on the string feature distribution through a state evaluation algorithm to obtain real-time performance data of the photovoltaic modules.

[0097] Through the collaborative cooperation of the above-mentioned various components, the GPS reference time signal is processed by the dynamic clock calibration algorithm, achieving a clock synchronization accuracy at the nanosecond level and avoiding the data distortion problem caused by clock deviation in traditional acquisition systems. Through the delay compensation network for distributed transmission processing, the transmission delay difference between different acquisition nodes is effectively eliminated, ensuring the phase consistency of the trigger synchronization signals of each node. The multi-channel parallel sampling circuit is used for data acquisition, and together with the high-precision voltage and current sensors, synchronous sampling of component parameters is achieved, ensuring the timing correlation of the sampled data. The digital filtering algorithm verifies and processes the original data, effectively filtering out the sampling noise, improving the signal-to-noise ratio of the data. At the same time, through the timing normalization process, the validity and comparability of the data are ensured. The data clustering algorithm performs string matching analysis on the collected data, accurately identifying the strings with abnormal performance, providing a reliable basis for fault diagnosis. The state evaluation algorithm screens the data on the string feature distribution, realizing the real-time evaluation of the performance of photovoltaic modules, and providing data support for operation and maintenance decisions. The entire solution realizes the full-process processing from data acquisition to performance evaluation, overcoming problems such as chaotic timing, data distortion, and analysis lag in traditional acquisition systems, and significantly improving the operation and maintenance efficiency and fault diagnosis ability of photovoltaic power plants. Through the high-precision time synchronization and data processing technology, this solution not only ensures the accuracy of the collected data, but also realizes the precise evaluation of the operating state of photovoltaic modules through systematic data analysis methods.

[0098] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distributed photovoltaic data acquisition method based on time synchronization, characterized in that, The distributed photovoltaic data acquisition method based on time synchronization includes: Performing clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronous acquisition reference signal; Performing distributed transmission processing on the synchronous acquisition reference signal through a delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes; Performing data acquisition processing on the photovoltaic module parameters of the multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain raw voltage and current data with timestamps; Performing data verification processing on the raw voltage and current data through a digital filtering algorithm to obtain effectively acquired data with normalized time series; Performing string matching analysis processing on the effectively acquired data through a data clustering algorithm to obtain the string feature distribution of the acquired data; Performing data screening processing on the string feature distribution through a state evaluation algorithm to obtain the real-time performance data of the photovoltaic module.

2. The distributed photovoltaic data acquisition method based on time synchronization according to claim 1, characterized in that The performing clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronous acquisition reference signal includes: Performing signal demodulation processing on the received GPS satellite signal to obtain a second pulse signal and serial port time code information; Performing phase comparison processing on the second pulse signal and the output signal of the local crystal oscillator to obtain a clock frequency deviation value; Performing fast capture processing on the clock frequency deviation value through a segmented adjustment coefficient to obtain an initial frequency calibration value; Performing fine tracking processing on the initial frequency calibration value through small-step iteration to obtain stable calibration parameters; Performing frequency correction processing on the local crystal oscillator through the stable calibration parameters to obtain a 10 MHz reference clock signal; Performing synchronous pulse generation processing on the 10 MHz reference clock signal through counting frequency division to obtain the nanosecond-level synchronous acquisition reference signal.

3. The distributed photovoltaic data acquisition method based on time synchronization according to claim 1, wherein The performing distributed transmission processing on the synchronous acquisition reference signal through a delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes includes: Performing transmission delay measurement on the synchronous acquisition reference signal on each transmission path to obtain path delay values; Performing compensation parameter configuration on the path delay values through a programmable delay unit to obtain compensation coefficients for each path; Performing delay calibration on the assigned signals of each acquisition node through the compensation coefficients to obtain phase adjustment signals; Performing signal reshaping on the phase adjustment signals through a low-jitter buffer amplifier circuit to obtain multiple buffered outputs; Performing signal locking on the multiple buffered outputs through a phase-locked loop circuit to obtain the trigger synchronization signals for the multiple acquisition nodes.

4. The distributed photovoltaic data acquisition method based on time synchronization according to claim 1, wherein The performing data acquisition processing on the photovoltaic module parameters of the multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain raw voltage and current data with timestamps includes: Performing voltage division processing on the voltage signal of the photovoltaic module through a high-impedance voltage sampling circuit to obtain a normalized voltage signal; Performing current sampling processing on the current signal of the photovoltaic module through a Hall sensor to obtain a normalized current signal; Performing signal conditioning processing on the normalized voltage signal and the normalized current signal through an anti-aliasing filter to obtain filtered sampling signals; The filtered sampled signal is subjected to high-precision quantization processing by a 24-bit Σ-Δ analog-to-digital converter to obtain digital quantization data; The digital quantization data is subjected to time stamping processing by a clock counter to obtain the original voltage and current data with time stamps; 5. The distributed photovoltaic data acquisition method based on time synchronization according to claim 1, wherein The original voltage and current data is subjected to data verification processing by a digital filtering algorithm to obtain effectively acquired data with normalized time series, including: Performing time series comparison processing on the time stamps of the original voltage and current data to obtain a data synchronization deviation value; Performing data screening processing on the data synchronization deviation value and a preset threshold to obtain time series valid data; Performing signal smoothing processing on the time series valid data by a linear-phase FIR filter to obtain filtered output data; Performing amplitude normalization processing on the filtered output data to obtain range-normalized data; Performing time series regularization processing on the range-normalized data by data recombination to obtain the effectively acquired data with normalized time series; 6. The distributed photovoltaic data acquisition method based on time synchronization according to claim 1, wherein The effectively acquired data is subjected to string matching analysis processing by a data clustering algorithm to obtain the string feature distribution of the acquired data, including: Performing data grouping processing on the effectively acquired data with normalized time series according to string numbers to obtain string grouped data; Performing feature extraction processing on the voltage and current parameters in the string grouped data to obtain string feature parameters; Performing statistical distribution calculation processing on the string feature parameters to obtain a parameter distribution interval; Performing string classification processing on the parameter distribution interval by K-means clustering to obtain string category identifiers; Performing string association processing on the string category identifiers through correlation analysis to obtain the string feature distribution; 7. The distributed photovoltaic data acquisition method based on time synchronization according to claim 1, wherein The string feature distribution is subjected to data screening processing by a state evaluation algorithm to obtain the real-time performance data of the photovoltaic module, including: Performing parameter threshold detection processing on the string feature distribution to obtain abnormal feature marks; Performing fault type identification processing on the abnormal feature marks to obtain fault category data; Performing statistical frequency analysis processing on the fault category data to obtain a fault probability distribution; Performing association rule extraction processing on the fault probability distribution to obtain state feature indicators; Performing performance parameter calculation processing on the state feature indicators to obtain the real-time performance data of the photovoltaic module; 8. A distributed photovoltaic data acquisition device based on time synchronization is used to implement the distributed photovoltaic data acquisition method based on time synchronization as described in any one of claims 1-7, and is characterized in that, The distributed photovoltaic data acquisition device based on time synchronization includes: A synchronization module, configured to perform clock synchronization processing on the GPS reference time signal through a dynamic clock calibration algorithm to obtain a nanosecond-level synchronized acquisition reference signal; A transmission module, configured to perform distributed transmission processing on the synchronized acquisition reference signal through a delay compensation network to obtain trigger synchronization signals for multiple acquisition nodes; An acquisition module, configured to perform data acquisition processing on the photovoltaic module parameters of the multiple acquisition nodes through a multi-channel parallel sampling circuit to obtain the original voltage and current data with time stamps; A verification module, configured to perform data verification processing on the original voltage and current data through a digital filtering algorithm to obtain effectively acquired data with normalized time series; An analysis module for performing string matching analysis and processing on the effective collected data through a data clustering algorithm to obtain the string feature distribution of the collected data; A screening module for performing data screening processing on the string feature distribution through a state evaluation algorithm to obtain the real-time performance data of the photovoltaic modules.

Images