Search

CN-122026618-A - Photovoltaic panel vibration monitoring method and system based on AI and heterogeneous fusion network communication

CN122026618ACN 122026618 ACN122026618 ACN 122026618ACN-122026618-A

Abstract

The application discloses a photovoltaic panel vibration monitoring method and system based on AI and heterogeneous fusion network communication, wherein the method comprises the steps that after vibration data acquisition and processing are completed in the same acquisition period by each sensor terminal, a local relative timestamp generated by a free running timer is sent to a gateway base station along with an uplink data message; the gateway base station calculates the relative transmission delay difference of any node pair according to the message receiving time and the local relative time stamp, constructs a frequency domain phase correction factor at each discrete frequency point according to the relative transmission delay difference, corrects uncorrected cross spectrum density to obtain corrected cross spectrum density, further calculates a coherence function value by combining each node self spectrum density, and positions a vibration abnormal region according to the spatial distribution of the coherence function value on the photovoltaic array topology. The method can improve the accuracy of multi-node vibration joint analysis and the reliability of abnormal positioning, and is suitable for on-line monitoring of the photovoltaic power station.

Inventors

  • REN SHUNING

Assignees

  • 广东华成电力能源股份有限公司

Dates

Publication Date
20260512
Application Date
20260416

Claims (10)

  1. 1. The photovoltaic panel vibration monitoring method based on AI and heterogeneous integration network communication is characterized by comprising the following steps: After each sensor terminal completes vibration data acquisition and processing in the same acquisition period, reading a counting difference value of a free running timer in a micro control unit between the starting moment and the current moment of the acquisition period as a local relative timestamp, and sending the local relative timestamp to a gateway base station along with an uplink data message; The gateway base station receives the uplink data messages of each sensor terminal in the same acquisition period and the same logic group, records the receiving time of each uplink data message and extracts the local relative time stamp thereof; Constructing a frequency domain phase correction factor at each discrete frequency point based on the relative transmission delay difference, wherein the phase angle of the frequency domain phase correction factor at each discrete frequency point is the product of the double circumference rate, the frequency value of the frequency point and the relative transmission delay difference; Multiplying the frequency domain phase correction factor by the uncorrected cross spectral density obtained by the vibration characteristic quantity sequence calculation of two nodes in the node pair, and obtaining corrected cross spectral density; and calculating a coherence function value of the node pair based on the corrected cross spectral density and the self spectral density of each node in the node pair, and positioning a vibration abnormal region according to the spatial distribution of the coherence function value on the photovoltaic array topology.
  2. 2. The method of claim 1, wherein the uplink data packets are transmitted between each sensor terminal and the gateway base station using a LoRa communication protocol, and the corrected cross spectral density or the coherence function value is transmitted between the gateway base station and a cloud server in a WIFI manner.
  3. 3. The method of claim 1 wherein the gateway base station records the relative transmission delay difference of the same node pair in a plurality of continuous acquisition periods, performs linear regression fitting with the corresponding accumulated running time of each acquisition period as an independent variable and the corresponding relative transmission delay difference as a dependent variable to obtain a slope, wherein the slope characterizes the crystal oscillator frequency offset between two nodes in the node pair, subtracts the product of the crystal oscillator frequency offset and the current accumulated running time from the relative transmission delay difference of the current acquisition period in the subsequent acquisition period to obtain the relative transmission delay difference after frequency offset compensation, and constructs the frequency domain phase correction factor with the relative transmission delay difference after frequency offset compensation.
  4. 4. The method of claim 3, wherein the acquisition cycle data used in the linear regression fit is taken from a sliding window of recent acquisition cycles, and when the residual error of the linear regression fit is outside a predetermined range, historical data in the sliding window is cleared and new acquisition cycle data is restarted.
  5. 5. The method of claim 1, wherein the corrected cross-spectral densities obtained over a plurality of consecutive acquisition periods are frequency point-wise computationally averaged for the same node to obtain an average cross-spectral density, and wherein the coherence function value is calculated by replacing the corrected cross-spectral densities for a single acquisition period with the average cross-spectral density.
  6. 6. The method of claim 5, wherein the corrected cross spectral density for each of the plurality of consecutive acquisition periods is derived from the frequency domain phase correction factor multiplied by the uncorrected cross spectral density from frequency point to frequency point, the frequency domain phase correction factor for each acquisition period being independently constructed based on the estimated relative transmission delay differences in each acquisition period.
  7. 7. The method of claim 1, wherein calculating the coherence function value of the node pair comprises calculating the ratio of the square of the modulus of the corrected cross spectral density to the product of the self spectral densities of the two nodes in the node pair at each discrete frequency point corresponding to the frequency band of interest to obtain the coherence function value at each discrete frequency point, and arithmetically averaging the coherence function values at each discrete frequency point in the frequency band of interest to obtain an in-band average coherence index, and taking the in-band average coherence index as the coherence function value to participate in the subsequent vibration anomaly region location.
  8. 8. The method of claim 7, wherein the upper frequency of the frequency band of interest corresponds to a cut-off frequency of a digital low pass filter in each sensor terminal and the lower frequency of the frequency band of interest corresponds to a start frequency of the photovoltaic panel environmental vibration energy distribution.
  9. 9. The method of claim 1, wherein locating the vibration anomaly region based on the spatial distribution of the coherence function values over the photovoltaic array topology comprises comparing the coherence function value of each node pair with a statistical baseline value for the node pair under historical normal operating conditions, marking the node pair as an anomaly node pair when the magnitude of the current coherence function value deviates from the statistical baseline value by more than a fluctuation range determined based on historical statistics, and determining the photovoltaic panel position corresponding to a node as the vibration anomaly region when node pairs each of which the node pair is composed of a plurality of adjacent nodes are marked as anomaly node pairs in the photovoltaic array topology.
  10. 10. Photovoltaic board vibration monitoring system based on AI and heterogeneous integration network communication, characterized in that includes: The method comprises the steps of including a plurality of sensor terminals and a gateway base station, wherein: Each sensor terminal is used for reading a counting difference value of a free running timer in a micro control unit between the starting moment and the current moment of the acquisition period as a local relative timestamp after the acquisition and the processing of vibration data are completed in the same acquisition period, and sending the local relative timestamp to the gateway base station along with an uplink data message; The gateway base station is used for receiving the uplink data messages of each sensor terminal in the same acquisition period and the same logic grouping, recording the receiving time of each uplink data message and extracting the local relative time stamp thereof; The gateway base station is further configured to construct a frequency domain phase correction factor at each discrete frequency point based on the relative transmission delay difference, where a phase angle of the frequency domain phase correction factor at each discrete frequency point is a product of a double circumference rate, a frequency value of the frequency point, and the relative transmission delay difference; The gateway base station is further used for multiplying the frequency domain phase correction factor by frequency points to obtain uncorrected cross spectral density through calculation of vibration characteristic quantity sequences of two nodes in the node pair, and obtaining corrected cross spectral density; The gateway base station is further configured to calculate a coherence function value of the node pair based on the corrected cross spectral density and the self spectral density of each node in the node pair, and locate a vibration abnormal region according to the spatial distribution of the coherence function value on the photovoltaic array topology.

Description

Photovoltaic panel vibration monitoring method and system based on AI and heterogeneous fusion network communication Technical Field The application relates to the technical field of cloud edge coordination, in particular to a photovoltaic panel vibration monitoring method and system based on AI and heterogeneous fusion network communication. Background Along with the continuous expansion of the installation scale of the photovoltaic power station, the photovoltaic array is in the complicated working conditions of wind load, temperature difference, equipment operation disturbance and the like for a long time, and the problems of loosening, fatigue, deformation and the like of the photovoltaic panel, the support, the connecting piece and other structural components of the photovoltaic panel are easy to occur. If the problems cannot be found in time, the abnormal stress of the assembly, the reduction of the power generation efficiency and even the damage of a local structure can be further caused. Therefore, developing vibration monitoring for the operation state of the photovoltaic panel and identifying abnormal areas according to the vibration monitoring has become an important technical direction in the structural health monitoring of the photovoltaic power station. In the prior art, a plurality of vibration sensors are generally distributed on the surface of a photovoltaic panel or at key positions of a support, vibration signals are collected by a terminal and then uploaded to a gateway or a background system, and multi-node data are subjected to joint analysis to judge whether an abnormal vibration position exists in an array. However, in a large-scale photovoltaic power station scene, the number of sensor nodes is large, the distribution range is wide, the data is often transmitted between the terminal and the gateway by adopting a low-power-consumption long-distance wireless link, and the data is transmitted back between the gateway and the background by adopting a broadband link. Under the heterogeneous communication network condition, different node data are easily affected by factors such as channel access competition, transmission delay difference, local clock frequency deviation and the like in the uploading process, so that the accurate time alignment relation of the multi-node vibration data is difficult to maintain. For applications requiring joint frequency domain analysis such as cross spectral density, coherence and the like, time dyssynchrony can directly influence the accuracy of analysis results, thereby reducing the reliability of abnormal positioning. The existing method generally adopts a time domain interpolation resampling mode to align different node data and then carries out frequency domain analysis, but the mode has large calculated amount, high requirement on gateway side embedded processing resources, and the problems of spectrum leakage, truncation error and the like can be introduced in the interpolation and resampling processes, so that the real-time performance, the precision and the engineering deployment cost are difficult to consider. Therefore, a technical scheme suitable for a photovoltaic panel multi-node vibration monitoring scene and capable of realizing efficient and reliable joint analysis under heterogeneous fusion network communication conditions is needed. Disclosure of Invention The embodiment of the application provides a photovoltaic panel vibration monitoring method and system based on AI and heterogeneous fusion network communication, which at least solve part of technical problems in the related art. According to a first aspect of the embodiment of the application, there is provided a photovoltaic panel vibration monitoring method based on communication between AI and heterogeneous fusion network, comprising: After each sensor terminal completes vibration data acquisition and processing in the same acquisition period, reading a counting difference value of a free running timer in a micro control unit between the starting moment and the current moment of the acquisition period as a local relative timestamp, and sending the local relative timestamp to a gateway base station along with an uplink data message; The gateway base station receives the uplink data messages of each sensor terminal in the same acquisition period and the same logic group, records the receiving time of each uplink data message and extracts the local relative time stamp thereof; Constructing a frequency domain phase correction factor at each discrete frequency point based on the relative transmission delay difference, wherein the phase angle of the frequency domain phase correction factor at each discrete frequency point is the product of the double circumference rate, the frequency value of the frequency point and the relative transmission delay difference; Multiplying the frequency domain phase correction factor by the uncorrected cross spectral density obtained by the vibration cha