5G-based electric power Internet of Things equipment fault judgment system and 5G-based electric power Internet of Things equipment fault judgment method

Abstract

The invention discloses a 5G-based power Internet of Things equipment fault judgment system and method, and relates to the technical field of solar power stations, and the system comprises a multi-dimensional sensing network deployment module, an aging feature classification and recognition module, an aging risk grade evaluation module and an aging type aggregation analysis module. The multi-dimensional sensing module deploys a terahertz sensing array and a microenvironment monitoring network, and obtains the multi-dimensional data characteristics of the perovskite thin film. The aging feature classification and identification module analyzes the evaluation value and the aging risk level accordingly. And the aging risk grade evaluation module obtains a 9-dimensional feature vector and analyzes an aging type. And the aging type aggregation analysis module divides areas, calculates a spatial aggregation index and analyzes an aggregation level, thereby realizing comprehensive monitoring and analysis of aging of the perovskite battery of the solar power station.

Description

Technical Field

[0001] The present invention relates to the technical field of solar power stations, and in particular to a 5G-based power Internet of Things equipment fault diagnosis system and method. Background Art

[0002] As a new generation of high-efficiency photovoltaic technology, perovskite solar cells have become a research hotspot in the energy field due to their advantages such as high photoelectric conversion efficiency and low-cost preparation. However, their thin film materials are easily affected by environmental factors such as humidity, temperature, and light, and are prone to aging phenomena such as microcracks, ion migration, and interface degradation, which leads to rapid degradation of device performance and seriously restricts their large-scale application. Therefore, a 5G-based power Internet of Things equipment fault diagnosis system and method is needed.

[0003] Existing technology, such as the invention application patent with publication number CN118096122A, discloses an intelligent power station equipment fault management method and system, which relates to the field of data processing technology. The method includes: obtaining basic information of the equipment of the target power station, combining the Internet of Things technology to build an equipment network topology diagram of the target power station, and deploying equipment monitoring sensor modules; through the equipment monitoring sensor modules, collecting multiple sets of real-time status data of the target power station equipment in real time, and combining the fault prediction model to perform fault analysis to obtain multiple equipment fault prediction results; conducting a power station operation safety risk assessment, and selecting the optimal fault management plan based on the safety risk assessment results to perform equipment fault management of the target power station. The present invention solves the technical problem that the traditional fault management method in the prior art has limitations and cannot meet the trend of increasing complexity of the power system, and achieves the technical effect of performing equipment fault early warning through multi-source data analysis to improve the reliability and operation efficiency of power station equipment.

[0004] Regarding the above-mentioned solution, the inventors of this application have discovered that the above-mentioned technology has at least the following technical problems: 1. The existing technology has significant limitations in the perception dimension of perovskite solar cell aging monitoring. Traditional solutions often focus on single-indicator monitoring, such as judging performance degradation solely by Voc decay rate, or relying on offline means to observe surface microcracks. These solutions fail to establish a full-chain monitoring system of "nano-defects-interface environment-macro-performance". As a result, it is impossible to analyze the coupling effects of aging inducements, such as the complex relationship between accelerated ion migration after microcrack formation and the synergistic exacerbation of interface degradation by humidity penetration and temperature fluctuations. This makes the diagnosis of the root causes of aging one-sided and difficult to support precise intervention.

[0005] 2. In terms of data processing and intelligent analysis, existing technologies face dual bottlenecks: transmission efficiency and recognition accuracy. On the one hand, traditional wireless networks cannot meet the real-time transmission requirements of high-bandwidth image data from terahertz sensor arrays and high-frequency sampling from microenvironmental probes. Furthermore, they lack an edge-cloud collaborative computing architecture, resulting in data analysis delays often reaching minutes. On the other hand, aging type identification often relies on a single algorithm that fails to integrate spatial and temporal information, making it difficult to distinguish between similar aging patterns such as moisture penetration and ion migration. Consequently, recognition accuracy is generally low, which can easily lead to misjudgments and failure of operational and maintenance strategies.

[0006] 3. In addition, existing technologies lack the ability to spatially distribute and close-loop control aging faults. Traditional solutions only evaluate individual panels in isolation, fail to quantify the regional clustering patterns of aging types within the power station, and fail to analyze whether there are specific commonalities, resulting in a "one-size-fits-all" operation and maintenance strategy. At the same time, monitoring data is disconnected from control strategies, with most alarms being issued after the fact rather than proactive warnings. Furthermore, there is a lack of differentiated intervention measures for different aging types, making it difficult to prevent the spread of aging, ultimately leading to high maintenance costs and delayed fault response. Summary of the Invention

[0007] In response to the above-mentioned technical deficiencies, the purpose of the present invention is to provide a 5G-based power Internet of Things equipment fault diagnosis system and method.

[0008] In order to solve the above technical problems, the present invention adopts the following technical solutions: In the first aspect, the present invention provides a 5G-based power Internet of Things equipment fault judgment system, including: a multi-dimensional perception network deployment module: used to deploy terahertz sensor arrays and micro-environment monitoring networks in each solar panel of the target solar power station, and then obtain the multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel.

[0009] Aging feature classification and identification module: It is used to analyze the multi-dimensional data features of the perovskite film corresponding to each monitoring node in each solar panel to obtain the perovskite film evaluation value corresponding to each solar panel, and then evaluate the aging risk level of the perovskite film corresponding to each solar panel.

[0010] Aging risk level assessment module: used to obtain the 9-dimensional feature vector corresponding to each solar panel, and then analyze the aging type of the perovskite film corresponding to each solar panel.

[0011] Aging type aggregation analysis module: used to divide the target solar power station into several identical areas, so as to calculate the spatial aggregation index corresponding to the aging type of the perovskite film in each area, and then analyze the aggregation level corresponding to the aging type of the perovskite film in each area.

[0012] In a second aspect, the present invention provides a 5G-based method for fault diagnosis of electric power Internet of Things equipment, including: step one, deployment of a multidimensional sensing network: deploying a terahertz sensor array and a microenvironment monitoring network in each solar panel of a target solar power station, and then obtaining multidimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel.

[0013] Step 2: Identification of aging feature classification: Based on the multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel, the perovskite film evaluation value corresponding to each solar panel is analyzed to evaluate the aging risk level of the perovskite film corresponding to each solar panel.

[0014] Step 3: Assessment of aging risk level: Obtain the 9-dimensional feature vector corresponding to each solar panel, and then analyze the aging type of the perovskite film corresponding to each solar panel.

[0015] Step 4: Analysis of aging type clustering: Divide the target solar power station into several identical areas to calculate the spatial clustering index corresponding to the perovskite film aging type in each area, and then analyze the clustering level corresponding to the perovskite film aging type in each area.

[0016] The beneficial effects of the present invention are: 1. The embodiments of the present invention realize multi-dimensional precise perception and efficient data transmission: through the coordinated deployment of the terahertz sensor array and the microenvironment monitoring network, it covers the full-dimensional characteristics from nanoscale defects to macroscopic performance, solves the one-sided perception of traditional technology, and provides a comprehensive and timely data source for subsequent analysis.

[0017] 2. The embodiments of the present invention improve the accuracy of aging type identification and risk assessment: by fusing spatiotemporal information through a 9-dimensional feature vector and dynamically weighting spatial and temporal features with the help of the attention mechanism of the perovskite aging pattern recognition model, the accuracy of aging type identification is greatly improved. In addition, through the normalized fusion of multi-dimensional indices such as the crack risk index and the interface degradation index, a quantitative assessment of the aging risk level is achieved, overcoming the problems of high misjudgment rate and fuzzy assessment of traditional single algorithms, and providing an accurate basis for differentiated control strategies.

[0018] 3. The embodiments of the present invention realize spatial aggregation analysis and targeted operation and maintenance: through regional division and spatial aggregation index calculation, the regional distribution pattern of aging types is accurately located, and a graded early warning plan is formulated in conjunction with the aggregation level, so that the operation and maintenance strategy is transformed from a "one-size-fits-all" to a "regional customization". For example, common inducements are prioritized in high-aggregation areas, reducing operation and maintenance costs by more than 30%. At the same time, through the closed-loop linkage of control strategies and early warnings, the spread of aging is effectively prevented, and the loss of power generation efficiency is reduced to within 5%.

[0019] 4. The embodiments of the present invention enhance full-scenario adaptability and intelligent closed-loop capabilities: the system deeply couples monitoring data, analysis results, and control strategies through an edge-cloud collaborative architecture, achieving an end-to-end response from perception to intervention, and realizing an intelligent upgrade from "passive alarm" to "active warning" - precise disposal. This significantly improves the reliability and service life of perovskite solar cells in complex environments, providing key technical support for their large-scale application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 This is a schematic diagram of the system module connection of the present invention.

[0022] Figure 2 The present invention is a flowchart of the steps for implementing the method. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0024] The present invention is implemented as follows Figure 1 As shown, a 5G-based power Internet of Things equipment fault judgment system includes: a multi-dimensional perception network deployment module, an aging feature classification and identification module, an aging risk level assessment module and an aging type aggregation analysis module.

[0025] The aging feature classification and identification module is connected to the multi-dimensional perception network deployment module and the aging risk level assessment module respectively, and the aging risk level assessment module is connected to the aging type aggregation analysis module.

[0026] Multi-dimensional perception network deployment module: used to deploy terahertz sensor arrays and micro-environment monitoring networks in each solar panel of the target solar power station, and then obtain the multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel.

[0027] In a specific embodiment, the terahertz sensor array and microenvironment monitoring network are deployed in each solar panel of the target solar power station. The specific deployment process is as follows: a terahertz sensor array and a microenvironment monitoring network are constructed on the surface and inside of the perovskite solar cell module. The terahertz sensor array is distributed with a grid density of 2cm×2cm. Each node integrates a 0.1-10THz frequency band transmitter and receiver. Three groups of microenvironment probes are synchronously deployed in the edge area of ​​the battery. Each group contains humidity, temperature and light intensity sensors, and the perception data is uploaded to the edge computing node in real time through the 5G module.

[0028] Aging feature classification and identification module: It is used to analyze the multi-dimensional data features of the perovskite film corresponding to each monitoring node in each solar panel to obtain the perovskite film evaluation value corresponding to each solar panel, and then evaluate the aging risk level of the perovskite film corresponding to each solar panel.

[0029] In a specific embodiment, the analysis obtains the perovskite film evaluation value corresponding to each solar panel. The specific analysis process is as follows: the crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are analyzed and normalized. At the same time, they are input into the perovskite film evaluation value analysis model to output the perovskite film evaluation value corresponding to each solar panel.

[0030] It should be noted that the analysis process of the perovskite film evaluation value corresponding to each solar panel is as follows: the crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are recorded as Q ab 、W ab and E ab , where a represents the number of each solar panel, a=1,2...m, m is a positive integer, and m is also the sum of all solar panels; b represents the number of each monitoring node, b=1,2...n, n is a positive integer, and n is also the sum of all monitoring nodes. Substitute into the analysis formula: The perovskite film evaluation value Ω corresponding to each solar panel is obtained. a .

[0031] It should be noted that the weightings are based on clear physical meaning and experimental evidence. The weightings for the crack risk index (0.4), interface degradation index (0.3), and ion migration index (0.3) are based on research on the failure mechanisms of perovskite films. The crack length weight of 0.4 is derived from the Griffith fracture mechanics criterion, and experiments show that cell efficiency decays by 35% when the crack length exceeds 2 mm. The interface degradation weight of 0.3 is based on the IEC62788-7-2 standard, and the interface degradation rate increases by three times when the humidity exceeds 60% RH. The ion migration weight of 0.3 is most sensitive when the Voc decay rate exceeds 5% / month. All input parameters are normalized to eliminate dimensionality.

[0032] In a specific embodiment, the crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are analyzed, and the specific analysis process is as follows: multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel are obtained, and the multi-dimensional data characteristics of the perovskite film include the length, density, and expansion rate corresponding to microcracks; humidity, temperature and light intensity corresponding to the interface; Voc attenuation rate and power fluctuation coefficient, and are normalized. At the same time, they are input as input items into the crack risk index analysis module, the interface degradation index analysis module and the ion migration index analysis module respectively, and the crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are output.

[0033] It should be noted that by deploying terahertz sensor arrays in a 2cm×2cm grid on the surface and inside of each solar panel, and utilizing the penetration characteristics of terahertz waves in the 0.1-10THz frequency band, the length, density and expansion rate of microcracks are captured in real time, and the length difference between adjacent time points is calculated through continuous scanning; simultaneously, microenvironment probes with integrated humidity, temperature, and light intensity sensors are installed in the edge area of ​​the panel to collect interface environmental parameters at high frequency; at the same time, the real-time Voc data of each panel is obtained through the communication interface of the photovoltaic inverter, and the Voc decay rate ((initial Voc-current Voc) / initial Voc×100%) and power fluctuation coefficient (difference between the maximum and minimum power values ​​within 1 hour / average value) are calculated. All data are uploaded to the edge computing node in real time via the 5G module, realizing full-time and high-resolution acquisition of multi-dimensional data characteristics of perovskite films.

[0034] It should also be noted that the analysis process of the crack risk index corresponding to each monitoring node in each solar panel is as follows: the length, density, and expansion rate of the microcracks corresponding to each monitoring node in each solar panel are respectively denoted as ω ab ,ξ ab and ψ ab , substitute into the calculation formula: In the process of analyzing the crack risk index, the crack risk index corresponding to each monitoring node in each solar panel is obtained, and the interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are obtained according to the analysis process of the crack risk index.

[0035] It should also be noted that the multidimensional data features of the perovskite film corresponding to each monitoring node in each solar panel are normalized. These include the length, density, and growth rate of microcracks; the humidity, temperature, and light intensity at the interface; and the Voc decay rate and power fluctuation coefficient. After normalization, all three are dimensionless values ​​between 0 and 1, satisfying the mathematical premise of weighted addition.

[0036] In a specific embodiment, the perovskite film aging risk level corresponding to each solar panel is evaluated, and the specific evaluation process is as follows: the perovskite film evaluation value corresponding to each solar panel is compared with the set perovskite film evaluation value interval corresponding to each perovskite film aging risk level; if the perovskite film evaluation value corresponding to a certain solar panel is within the set perovskite film evaluation value interval corresponding to a certain perovskite film aging risk level, the set perovskite film aging risk level is used as the perovskite film aging risk level corresponding to the solar panel;

[0037] The control strategies corresponding to each solar panel are analyzed based on the aging risk level of the perovskite film corresponding to each solar panel. The aging risk levels of the perovskite film include high risk, medium risk and low risk.

[0038] In a specific embodiment, the control strategy corresponding to each solar panel is analyzed, and the specific analysis process is as follows: B1. If the aging risk level of the perovskite film corresponding to each solar panel is high risk, an emergency stop-loss operation and maintenance linkage strategy is executed.

[0039] It should be noted that an emergency stop-loss operation and maintenance linkage strategy is implemented: the sampling frequency of the micro-environment probes in the area is maintained at a normal level through 5GmMTC slicing (temperature / humidity once every 15 minutes, light once every 30 minutes), and the terahertz sensor array only performs a special scan once a day on nodes that have shown slight aging characteristics in historical data; the edge node generates a simple early warning report every 24 hours, focusing on the difference in aging types between the area and adjacent areas (such as the difference in the number of microcracks), and simultaneously pushes the data to the cloud for incremental model learning. Active intervention will not be triggered for the time being. The warning level will be automatically upgraded only when the number of a certain type of aging is monitored to increase by more than 10% month-on-month for three consecutive times.

[0040] B2. If the aging risk level of the perovskite film corresponding to each solar panel is medium risk, the edge autonomous adaptation strategy is executed.

[0041] It should be noted that the edge autonomous adaptation strategy is as follows: for microcrack-dominated aging, the pulsed cooling fan is used to reduce the surface temperature difference of the solar panel to within 3°C, and at the same time, the high-frequency scanning mode of the terahertz sensor array is triggered to continuously track the crack expansion; for humidity penetration-dominated aging, the inert gas circulation system of the packaging layer is started to control the interface humidity permeability coefficient below 0.01% RH / min; for ion migration-dominated aging, the intelligent shading system is activated for local shading to reduce the light intensity to 600W / m 2 Next, the lightweight phase-field model at the edge node updates the ion migration prediction every 5 minutes to ensure that no cloud intervention is required before the risk level is further escalated.

[0042] B3. If the aging risk level of the perovskite film corresponding to each solar panel is low risk, the trend warning forward strategy is executed.

[0043] It should be noted that parameter adjustment commands are issued in batches through 5GmMTC slices to dynamically fine-tune the cooling fan speed according to the real-time temperature, maintaining the operating temperature of the solar panel in the range of 25-35°C; the micro-environmental probe maintains a regular sampling frequency and continuously accumulates aging characteristic data for incremental model learning; the cloud platform generates a trend analysis report once a week and adjusts the monitoring focus for the next week in advance based on regional weather forecasts. For example, coastal areas prioritize strengthening the monitoring weight of the humidity permeability coefficient to ensure that potential risks are identified in the budding stage.

[0044] Aging risk level assessment module: used to obtain the 9-dimensional feature vector corresponding to each solar panel, and then analyze the aging type of the perovskite film corresponding to each solar panel.

[0045] In a specific embodiment, the analysis of the perovskite film aging type corresponding to each solar panel is performed as follows: a 9-dimensional feature vector corresponding to each solar panel is obtained, the 9-dimensional feature vector including the perovskite film evaluation value, 30-minute average, 1-hour peak and change rate, and input into the perovskite aging pattern recognition model. The perovskite aging pattern recognition model extracts spatial distribution characteristics, and the LSTM captures the temporal change law. The outputs of the two are dynamically weighted through the attention mechanism, and the probability distribution of the three types of aging patterns is obtained through the softmax function. The item with the highest probability is the perovskite film aging type of each solar panel.

[0046] It should be noted that the crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are analyzed, and a 9-dimensional feature vector including the perovskite film evaluation value, 30-minute average, 1-hour peak and change rate is constructed according to the time series. That is, each type of index corresponds to 3 dimensions of the evaluation value, 30-minute average, 1-hour peak and change rate, with a total of 3 types × 3 dimensions = 9 dimensions. After the feature vector is input into the CNN-LSTM fusion model, the model first performs a 3-layer convolution operation on the spatial distribution characteristics in the feature vector through the CNN module, such as the difference in crack risk index of different monitoring nodes and the regional gradient of the interface degradation index, to extract local spatial patterns such as microcrack aggregation areas and humidity penetration hotspots; then the LSTM module performs a 2-layer bidirectional recursive processing on the time series features, such as the fluctuation curve of the ion migration index with the illumination duration and the 1-hour peak change trend of the evaluation value, to capture long-term and short-term dependencies, focusing on identifying The hysteresis correlation between Voc decay and ion migration, and the temperature response law of crack growth rate are identified; then, weights are dynamically assigned to the spatial features output by CNN and the temporal features output by LSTM through the attention mechanism, with the weight range being 0-1. When the product of the microcrack feature weight and the crack risk index change rate is greater than 0.4, the output probability of "microcrack-dominated type" is enhanced; when the correlation weight between the 30-minute average value of the interface degradation index and the humidity change rate is greater than 0.35, the confidence of "interface degradation-dominated type" is improved; when the temporal weight between the 1-hour peak value of the ion migration index and the Voc decay rate is greater than 0.3, the prediction score of "ion migration-dominated type" is increased. Finally, the probability distribution of the three categories of labels is output through the softmax function. The one with the highest probability is the perovskite film aging type corresponding to the solar panel. After the model is trained with 1,000 sets of samples, the output accuracy is stable at above 92%, realizing accurate mapping from multidimensional features to aging types.

[0047] It should also be noted that the CNN-LSTM model:

[0048] 1. Specific interpretation of input data content:

[0049] Each item in the 9-dimensional eigenvector corresponds to a quantifiable physical indicator, with the following specific meanings:

[0050] Spatial Dimension Data: Three types of indices are collected in real time by a 20×20 grid node, all with normalized values ​​ranging from 0 to 1. The Crack Risk Index reflects the density and length of microcracks in the perovskite film; the Interface Degradation Index characterizes the degree of delamination between the film and the substrate; and the Ion Migration Index reflects the diffusion activity of anions and cations.

[0051] Time dimension data: The three time series features of each index type are calculated using a sliding window. The 30-minute mean is the arithmetic average of 60 consecutive 10-second sampling values; the 1-hour peak is the maximum value within 60 minutes; and the rate of change is (current value - value 30 minutes ago) / 30 minutes.

[0052] Aging type aggregation analysis module: used to divide the target solar power station into several identical areas, so as to calculate the spatial aggregation index corresponding to the aging type of the perovskite film in each area, and then analyze the aggregation level corresponding to the aging type of the perovskite film in each area.

[0053] In a specific embodiment, the aggregation level corresponding to the aging type of the perovskite film in each region is analyzed, and the specific analysis process is as follows: C1. Count the number of aging types of the perovskite film corresponding to each region, and then calculate the spatial aggregation index corresponding to the aging type of the perovskite film in each region, and the spatial aggregation index = the number of aging types of the perovskite film corresponding to each region / the total number of solar panels in each region.

[0054] C2. If the spatial aggregation index corresponding to the aging type of the perovskite film in each region is less than 0.2, it means that the aggregation level corresponding to the aging type of the perovskite film in this region is low aggregation.

[0055] C3. If the spatial aggregation index corresponding to the aging type of the perovskite film in each region is between 0.2 and 0.5, it means that the aggregation level corresponding to the aging type of the perovskite film in this region is medium aggregation.

[0056] C4. If the spatial aggregation index corresponding to the aging type of the perovskite film in each region is greater than 0.5, it means that the aggregation level corresponding to the aging type of the perovskite film in this region is high aggregation.

[0057] C5. The aggregation level corresponding to the aging type of the perovskite film in each region is analyzed to analyze the aging fault aggregation early warning plan for each region.

[0058] In a specific embodiment, the analysis of the aging fault aggregation warning scheme for each area is as follows: D1. If the aggregation level corresponding to the aging type of the perovskite film in a certain area is low aggregation, the distributed lightweight monitoring and warning scheme is activated.

[0059] It should be noted that the "distributed lightweight early warning solution" was launched: the sampling frequency of the micro-environment probes in the area was maintained at a normal level through 5GmMTC slicing, with temperature / humidity measured every 15 minutes and light exposure measured every 30 minutes. The terahertz sensor array only conducts a special scan once a day on nodes that have shown slight aging characteristics in historical data; the edge node generates a simple early warning report every 24 hours, focusing on the differences in aging types between the area and adjacent areas, and simultaneously pushes the data to the cloud for incremental model learning. Active intervention will not be triggered for the time being. The warning level will be automatically upgraded only when the number of a certain type of aging is monitored to increase by more than 10% month-on-month for three consecutive times.

[0060] D2. If the aggregation level corresponding to the aging type of the perovskite film in each region is medium aggregation, the regional targeted tracking and early warning plan will be activated.

[0061] It should be noted that the "regional targeted early warning plan" was launched: for the dominant aging type in the area, the scanning range of the terahertz sensor array was expanded to the boundary zone between two adjacent areas, the scanning density was increased to 1cm×1cm grid, and the humidity / temperature sampling frequency of the micro-environment probe was encrypted to once every 5 minutes to capture the aging diffusion trend in real time; the edge node generated a dynamic early warning map every 6 hours, marked high-risk sub-areas, and pushed preventive intervention suggestions to the operation and maintenance terminal through 5GURLLC slicing, while locking the aging characteristic parameters of the area, and triggering the medium-risk control strategy once touched.

[0062] D3. If the aggregation level corresponding to the aging type of the perovskite film in each region is high, the global emergency response warning plan will be activated.

[0063] It should be noted that the "global emergency warning plan" was launched: the continuous scanning mode of the terahertz sensor array and the millisecond-level data acquisition of the microenvironment probe were immediately activated, and high-resolution images and real-time parameters were synchronized to the cloud and edge nodes through 5GeMBB slices; the cloud platform generated a three-dimensional aggregation heat map, with the core aggregation area marked in red and the diffusion warning zone marked in orange, and combined with the LSTM model to predict the aggregation and diffusion path in the next 24 hours, and issued an emergency isolation command to the power station control system through URLLC slices, and simultaneously pushed an emergency work order containing fault location, aging type cause analysis and priority repair order to the operation and maintenance team to ensure that the first round of on-site disposal is completed within 1 hour.

[0064] The present invention is implemented as follows Figure 2 As shown, a 5G-based power Internet of Things equipment fault diagnosis method includes: Step 1, deployment of a multi-dimensional perception network: deploying a terahertz sensor array and a micro-environment monitoring network in each solar panel of the target solar power station, and then obtaining the multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel.

[0065] Step 2: Identification of aging feature classification: Based on the multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel, the perovskite film evaluation value corresponding to each solar panel is analyzed to evaluate the aging risk level of the perovskite film corresponding to each solar panel.

[0066] Step 3: Assessment of aging risk level: Obtain the 9-dimensional feature vector corresponding to each solar panel, and then analyze the aging type of the perovskite film corresponding to each solar panel.

[0067] Step 4: Analysis of aging type clustering: Divide the target solar power station into several identical areas to calculate the spatial clustering index corresponding to the perovskite film aging type in each area, and then analyze the clustering level corresponding to the perovskite film aging type in each area.

[0068] The above content is merely an example and explanation of the concept of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the scope of protection of the present invention.

Claims

1. A 5G-based power Internet of Things equipment fault diagnosis system, characterized in that: include: Multi-dimensional sensing network deployment module: used to deploy terahertz sensor arrays and micro-environment monitoring networks in each solar panel of the target solar power plant, thereby obtaining multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel; aging Feature classification and identification module: used to analyze the multi-dimensional data features of the perovskite film corresponding to each monitoring node in each solar panel to obtain the corresponding perovskite film evaluation value of each solar panel, and then evaluate the aging risk level of the perovskite film corresponding to each solar panel; Aging risk level assessment module: used to obtain the 9-dimensional feature vector corresponding to each solar panel, and then analyze the aging type of the perovskite film corresponding to each solar panel; Aging type aggregation analysis module: used to divide the target solar power station into several identical areas, so as to calculate the spatial aggregation index corresponding to the aging type of the perovskite film in each area, and then analyze the aggregation level corresponding to the aging type of the perovskite film in each area.

2. A 5G-based power Internet of Things equipment fault diagnosis system according to claim 1, characterized in that: The terahertz sensor array and microenvironment monitoring network are deployed in each solar panel of the target solar power station. The specific deployment process is as follows: A terahertz sensor array and microenvironment monitoring network are constructed on the surface and inside of the perovskite solar cell module. The terahertz sensor array is distributed in a 2cm×2cm grid density. Each node integrates a 0.1-10THz frequency band transmitter and receiver. Three groups of microenvironment probes are deployed simultaneously in the edge area of ​​the battery. Each group contains humidity, temperature and light intensity sensors, and the perception data is uploaded to the edge computing node in real time through the 5G module.

3. A 5G-based power Internet of Things equipment fault diagnosis system as claimed in claim 2, characterized in that: The analysis results in the evaluation value of the perovskite film corresponding to each solar panel. The specific analysis process is as follows: The crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel are analyzed and normalized. At the same time, they are input into the perovskite film evaluation value analysis model to output the perovskite film evaluation value corresponding to each solar panel.

4. A 5G-based power Internet of Things equipment fault diagnosis system as described in claim 3, characterized in that: The analysis of the crack risk index, interface degradation index and ion migration index corresponding to each monitoring node in each solar panel is as follows: The multidimensional data characteristics of the perovskite film corresponding to each monitoring node in each solar panel are obtained. The multidimensional data characteristics of the perovskite film include the length, density, and expansion rate corresponding to the microcracks; the humidity, temperature, and light intensity corresponding to the interface; the Voc decay rate and the power fluctuation coefficient, and are normalized. At the same time, they are input as input items into the crack risk index analysis module, the interface degradation index analysis module, and the ion migration index analysis module, and the crack risk index, interface degradation index, and ion migration index corresponding to each monitoring node in each solar panel are output.

5. A 5G-based power Internet of Things equipment fault diagnosis system as claimed in claim 4, characterized in that: The aging risk level of the perovskite film corresponding to each solar panel is evaluated. The specific evaluation process is as follows: Comparing the perovskite film evaluation value corresponding to each solar panel with the set perovskite film evaluation value interval corresponding to each perovskite film aging risk level; if the perovskite film evaluation value corresponding to a solar panel is within the set perovskite film evaluation value interval corresponding to a set perovskite film aging risk level, the set perovskite film aging risk level is used as the perovskite film aging risk level corresponding to the solar panel; The control strategies corresponding to each solar panel are analyzed based on the aging risk level of the perovskite film corresponding to each solar panel. The aging risk levels of the perovskite film include high risk, medium risk and low risk.

6. A 5G-based power Internet of Things equipment fault diagnosis system according to claim 5, characterized in that: The control strategy corresponding to each solar panel is analyzed, and the specific analysis process is as follows: B1. If the aging risk level of the perovskite film corresponding to each solar panel is high, an emergency stop-loss operation and maintenance linkage strategy will be implemented; B2. If the aging risk level of the perovskite film corresponding to each solar panel is medium risk, the edge autonomous adaptation strategy is implemented; B3. If the aging risk level of the perovskite film corresponding to each solar panel is low risk, the trend warning forward strategy is executed.

7. A 5G-based power Internet of Things equipment fault diagnosis system according to claim 6, characterized in that: The specific analysis process for analyzing the aging type of the perovskite film corresponding to each solar panel is as follows: The 9-dimensional feature vector corresponding to each solar panel is obtained. The 9-dimensional feature vector includes the perovskite film evaluation value, 30-minute average, 1-hour peak and change rate. After inputting it into the perovskite aging pattern recognition model, the perovskite aging pattern recognition model extracts spatial distribution characteristics, and the LSTM captures the temporal change law. The outputs of the two are dynamically weighted through the attention mechanism, and the probability distribution of the three types of aging patterns is obtained through the softmax function. The item with the highest probability is the perovskite film aging type of each solar panel.

8. A 5G-based power Internet of Things equipment fault diagnosis system according to claim 7, characterized in that: The aggregation level corresponding to the aging type of the perovskite film in each region is analyzed, and the specific analysis process is as follows: C1. Count the number of perovskite film aging types corresponding to each region, and then calculate the spatial aggregation index corresponding to the perovskite film aging type in each region. Spatial aggregation index = number of perovskite film aging types corresponding to each region / total number of solar panels in each region; C2. If the spatial aggregation index corresponding to the aging type of the perovskite film in each region is less than 0.2, it means that the aggregation level corresponding to the aging type of the perovskite film in this region is low aggregation; C3. If the spatial aggregation index corresponding to the aging type of the perovskite film in each region is between 0.2 and 0.5, it means that the aggregation level corresponding to the aging type of the perovskite film in this region is medium aggregation; C4. If the spatial aggregation index corresponding to the aging type of the perovskite film in each region is greater than 0.5, it means that the aggregation level corresponding to the aging type of the perovskite film in this region is high aggregation; C5. The aggregation level corresponding to the aging type of the perovskite film in each region is analyzed to analyze the aging fault aggregation early warning plan for each region.

9. A 5G-based power Internet of Things equipment fault diagnosis system as claimed in claim 8, characterized in that: The analysis of the aging fault aggregation early warning scheme for each area is as follows: D1. If the aggregation level corresponding to the perovskite film aging type in a certain area is low, the distributed lightweight monitoring and early warning solution is activated; D2. If the aggregation level corresponding to the aging type of the perovskite film in each region is medium aggregation, a regional targeted tracking and early warning plan will be initiated; D3. If the aggregation level corresponding to the aging type of the perovskite film in each region is high, the global emergency response warning plan will be activated.

10. A 5G-based electric power Internet of Things device fault judgment method for executing the 5G-based electric power Internet of Things device fault judgment system according to any one of claims 1 to 9, characterized in that: include: Step 1: Deployment of a multi-dimensional sensing network: Deploy a terahertz sensor array and a micro-environment monitoring network on each solar panel in the target solar power plant to obtain multi-dimensional data characteristics of the perovskite film corresponding to each monitoring node on each solar panel; Step 2: Identification of aging feature classification: Based on the multi-dimensional data features of the perovskite film corresponding to each monitoring node in each solar panel, the perovskite film evaluation value corresponding to each solar panel is analyzed to evaluate the aging risk level of the perovskite film corresponding to each solar panel; Step 3: Aging risk level assessment: Obtain the 9-dimensional feature vector corresponding to each solar panel, and then analyze the aging type of the perovskite film corresponding to each solar panel; Step 4: Analysis of aging type clustering: Divide the target solar power station into several identical areas to calculate the spatial clustering index corresponding to the perovskite film aging type in each area, and then analyze the clustering level corresponding to the perovskite film aging type in each area.

Images