Photovoltaic ammeter data communication and dynamic processing method and system based on Internet of Things

Abstract

The invention provides a photovoltaic ammeter data communication and dynamic processing method and system based on the Internet of Things, and relates to the fields of communication, the Internet of Things and electric power. According to the method, data such as current, voltage, environment temperature, illumination intensity and power grid load state are acquired in real time by deploying sensors, and dynamic encryption is performed by adopting a generative adversarial network (GAN), so that the security of data transmission is ensured. The encrypted data is transmitted to a cloud server through an Internet of Things communication module, and after decryption, the system decouples time components and space components in the data by adopting a derivative analysis method to generate a feature matrix to provide a basis for subsequent risk modeling; and based on the generated feature matrix, systematically constructing a dynamic risk model, updating a risk heat map in real time through an alternating direction implicit algorithm, and dynamically adjusting a risk diffusion coefficient and a risk bearing capacity threshold. An optimized power dispatching strategy is generated through a deep reinforcement learning algorithm in combination with a real-time electricity price signal and a risk heat map of a power grid, and system benefits are maximized.

Description

Technical Field

[0001] The present invention relates to the fields of communication, Internet of Things and electric power, and in particular to a photovoltaic meter data communication and dynamic processing method and system based on Internet of Things. Background Art

[0002] With the large-scale deployment of distributed photovoltaic power generation systems, photovoltaic meters, as core monitoring units, need to collect multi-dimensional data such as current, voltage, environmental parameters and grid status in real time, and realize cross-regional collaborative management through the Internet of Things. However, complex outdoor environments (such as extreme temperatures, shadows, and component aging) cause data to fluctuate violently, and traditional centralized communication architectures face problems such as high data security risks and delayed dynamic responses. For example, static encrypted transmission is easy to crack, feature-coupled data is difficult to accurately decouple environmental interference and equipment failures, risk propagation models lack temporal and spatial correlation, and anomaly detection relies on manual experience, resulting in low operation and maintenance efficiency and safety risks.

[0003] At present, the industry generally adopts SSL / TLS protocol to ensure data transmission security, and combines threshold alarm mechanism to realize abnormal monitoring. Existing technology encrypts meter data through fixed encryption algorithm (such as AES), extracts power generation characteristics through time series analysis, and triggers power adjustment based on rule engine. Some solutions introduce edge computing nodes to realize local data processing and improve reliability through redundant storage. In terms of risk prediction, linear regression or support vector machine is often used to establish static models, and operation and maintenance strategies are generated in combination with historical data.

[0004] However, existing technologies have significant defects: static encryption is difficult to deal with dynamic attacks, and key management vulnerabilities can easily lead to data leakage; feature decoupling methods ignore the coupling effect of environmental parameters and device status, and cannot accurately identify hidden risks such as shadow occlusion and hot spot effects; fixed threshold alarm mechanisms lack foresight for complex risk propagation (such as chain failures caused by high temperatures), and response delays can reach minutes. In addition, centralized storage faces the risk of single point failures, and the visual interface has poor interactivity, making it difficult to support real-time decision-making. These defects seriously restrict the efficient coordination capabilities of photovoltaic systems in smart grids. Summary of the invention

[0005] In response to the needs raised in the above-mentioned background technology, the embodiments of the present invention provide a photovoltaic meter data communication and dynamic processing method based on the Internet of Things and a system thereof, which aims to collect multi-dimensional data such as current, voltage, ambient temperature, etc. of the photovoltaic meter in real time through the photovoltaic meter data communication and dynamic processing method based on the Internet of Things, and use technologies such as generative adversarial networks (GAN) to encrypt data to ensure security during data transmission; at the same time, through risk propagation modeling and real-time optimization scheduling, the operating efficiency and safety of the photovoltaic system are improved to avoid the risks caused by equipment failure or performance degradation.

[0006] A photovoltaic meter data communication and dynamic processing method based on the Internet of Things, the specific steps include:

[0007] Step 1: Multimodal data encryption collection

[0008] This step collects raw data such as current, voltage, ambient temperature, light intensity, and grid load status in real time through sensors deployed at photovoltaic arrays and grid nodes.

[0009] To prevent data from being stolen or tampered with during transmission, dynamic encryption technology is used to protect the original data. Specifically, the system constructs a generative adversarial network (GAN). The generator performs pixel-level XOR operations on the original meter data and the random encryption matrix to generate an encrypted data stream that appears random but retains data characteristics; the discriminator continuously learns the difference between the real data and the encrypted data, and ultimately makes the encrypted data indistinguishable from the real data in terms of statistical characteristics. For example, when the light sensor detects an irradiance of 800W / m 2 When the data is received, the generator will convert it into a specific encrypted format (such as "1011→0100") to ensure that even if the data is intercepted, the real physical value cannot be directly parsed out.

[0010] The encrypted data is transmitted to the cloud server via an IoT communication module (such as LoRa or NB-IoT).

[0011] Step 2: Data feature decoupling and reconstruction

[0012] After the encrypted data reaches the cloud, it is necessary to separate the core factors that affect power generation efficiency (such as light changes, component aging, shadows, etc.). The system uses mathematical derivative analysis methods to calculate the sensitivity of different environmental parameters to power generation. For example, by analyzing historical data, it is found that when the component temperature increases by 1°C, the power generation efficiency decreases by 0.4%, and when the irradiance increases by 100W / m 2 , the current output is increased by 2.1A. Based on these sensitivity coefficients, the system decomposes the mixed data stream into independent time-space feature components: the time component reflects the impact of the day and night cycle and weather mutations, and the space component identifies the performance differences of different photovoltaic panel groups. In the final reconstructed feature matrix, each column corresponds to an independent influencing factor (such as "temperature influence factor column" and "irradiance response column"), providing structured input for subsequent risk analysis.

[0013] Step 3: Dynamic risk modeling

[0014] Based on the decoupled characteristic data, the system simulates the risk propagation process of photovoltaic systems in complex environments. For example, high temperature weather may cause components to overheat, and the risk will spread from local hot spots to surrounding areas like "heat diffusion". The mathematical model associates the risk diffusion rate with parameters such as temperature gradient and wind speed: when the temperature in a certain area reaches 60°C, the risk diffusion coefficient D increases to 0.08, causing the risk value of adjacent components to increase by 15% within 10 minutes. At the same time, the system introduces the concept of "risk carrying capacity". For example, under strong light conditions (irradiance>1000W / m 2 ), the maximum risk tolerance of the component K c Increased to 0.9 (full value is 1), and in rainy weather K c The Alternating Direction Implicit (ADI) algorithm updates the risk heat map of the entire field every 5 minutes, with red areas indicating high-risk locations that require immediate intervention (such as abnormal temperature at a certain inverter connection point).

[0015] Step 4: Real-time optimization scheduling

[0016] Combined with the real-time power price signal P(t) of the power grid (such as 1.2 yuan / kWh during peak hours and 0.3 yuan / kWh during valley hours) and the risk heat map H(x,t), the system dynamically adjusts the power dispatch strategy. For example, when the risk value of a certain area exceeds 0.7, the output power of the area is automatically reduced by 10% to avoid equipment overload; at the same time, the power of low-risk areas is preferentially called during peak electricity price periods to maximize revenue. The optimization process is achieved through deep reinforcement learning: the neural network simulates the long-term benefits under different dispatch strategies (such as total revenue in the next 24 hours = electricity sales revenue - equipment maintenance cost), and back-propagates to correct the strategy parameters. For example, when a thunderstorm is predicted in the afternoon, the system reserves electricity in low-risk periods in advance and releases it during peak electricity price periods, increasing the daily revenue by 8% to 12%.

[0017] Step 5: Anomaly Detection and Self-Healing

[0018] The system continuously compares the deviation between the actual power generation and the predicted value to locate potential faults. For example, if the actual current of a group of photovoltaic panels is 30% lower than the predicted value, and the residual index exceeds the threshold for three consecutive times, the fault diagnosis process is triggered. The algorithm first eliminates environmental factors (such as cloud cover), and then determines the fault location through topological analysis: if multiple adjacent components are abnormal at the same time, it is judged to be a junction box failure; if only a single component is abnormal, it may be that the panel is damaged or the wiring is detached. After determining the fault, the system automatically switches to the backup circuit and maintains power supply through redundant lines. For example, when the No. 5 string is detected to be abnormal, the control relay disconnects it from the main circuit and enables the pre-stored No. 12 backup string to be connected. The whole process is completed within 200ms.

[0019] Step 6: Distributed storage verification

[0020] To ensure the credibility of data storage, the system stores encrypted data in blocks on multiple edge nodes (such as regional gateways and local servers). Each data block is attached with a "commitment label" based on a mathematical problem, such as using a lattice to generate an unforgeable hash value. When data integrity needs to be verified, the verifier does not need to obtain the original data, but only needs to confirm through a zero-knowledge proof protocol that the storage node actually holds the correct data block.

[0021] Step 7: Visualization and User Interaction

[0022] The system integrates the processing results of each link into a visual interface to help operation and maintenance personnel intuitively grasp the overall status. For example:

[0023] Risk heat map: Use color gradient to display the risk distribution of the entire field. Click the red area to view specific parameters (such as "Coordinate (3,5) risk value 0.85, temperature 62℃");

[0024] Power curve: The predicted value and the actual value are displayed in a superimposed manner, and the abnormal points are automatically marked with possible reasons (such as "power drop at 14:25: cloud cover caused a sudden drop in irradiance");

[0025] Equipment topology: The connection relationship of the PV array is presented in three dimensions, and faulty equipment flashes to indicate the fault and provides maintenance suggestions (such as "Replace the MC4 connector of string 5").

[0026] Users can use gestures to zoom in and out to view details, or set threshold alerts (such as "send SMS notifications when risk > 0.8"). The interface data is refreshed every 30 seconds and supports multi-terminal access (PC, tablet, AR glasses).

[0027] Further: A photovoltaic meter data communication and dynamic processing system based on the Internet of Things, comprising:

[0028] Data acquisition and encryption module, which is used to collect raw data in real time through sensors deployed in photovoltaic arrays and grid nodes, including current, voltage, ambient temperature, light intensity and grid load status, and encrypt the collected raw data.

[0029] The data decryption and feature decoupling module is used to receive the encrypted data stream transmitted from the data acquisition and encryption module, decrypt the data, calculate the sensitivity of environmental parameters to power generation through the derivative analysis method, decouple the time component and space component in the data stream, and generate a feature matrix.

[0030] Dynamic risk modeling and updating module: This module builds a dynamic risk model for the photovoltaic system based on the feature matrix, and updates it in real time according to factors such as temperature, wind speed, and irradiance to generate a full-field risk heat map;

[0031] The power dispatch and optimization module uses a deep reinforcement learning algorithm to dynamically adjust the power output strategy and optimize the power call of the photovoltaic system based on the real-time power price signal and risk heat map of the power grid;

[0032] Anomaly detection and fault location module, which is used to monitor the deviation between the power generation of the photovoltaic system and the predicted value. When the residual exceeds the preset threshold, fault diagnosis is performed and the fault location is located;

[0033] The data storage and verification module is used to store encrypted data in multiple edge nodes and attach commitment tags to ensure the integrity of the data;

[0034] Visualization and user interaction module, which is used to generate and display the system's visual interface, help operation and maintenance personnel intuitively understand the system's operating status, and provide threshold alarm function.

[0035] Beneficial effects of the present invention: The present invention adopts generative adversarial networks (GAN) for dynamic encryption, which can effectively deal with dynamic attacks and improve the security of data transmission compared to traditional static encryption (such as AES). Even if the data is intercepted during the transmission process, the real data cannot be directly parsed, ensuring the privacy and integrity of photovoltaic meter data. Through the feature decoupling method, the factors affecting the power generation efficiency (such as temperature, irradiance, etc.) are processed independently, and hidden risks such as shadow occlusion and hot spot effect are identified more accurately, thereby improving the accuracy of risk prediction.

[0036] By constructing a dynamic risk propagation model, the present invention can simulate the process of risk spreading from a local area to the entire photovoltaic system and generate a risk heat map in real time. This model has a strong temporal and spatial correlation and can timely detect and warn of potential risks during system operation, avoiding the spread of faults and losses caused by response delays in traditional methods.

[0037] The system of the present invention adopts distributed storage, stores encrypted data in multiple edge nodes, and uses zero-knowledge proof protocol to verify the integrity of the data. This avoids the single point failure problem of centralized storage and enhances the data security and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0039] Figure 1A flow chart of the method of the present invention is shown.

[0040] Figure 2 A schematic diagram of the composition of the system of the present invention is shown. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described in conjunction with the drawings in the embodiment of the present invention. It should be understood that the drawings in the present invention only serve the purpose of illustration and description and are not used to limit the scope of protection of the present invention. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowcharts used in the present invention illustrate the operations implemented according to some embodiments of the present invention. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, under the guidance of the content of the present invention, those skilled in the art can add one or more other operations to the flowchart, or remove one or more operations from the flowchart.

[0042] In addition, the embodiments described in the present invention are only some embodiments of the present invention, rather than all embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0043] It should be noted that the term "comprising" will be used in the embodiments of the present invention to indicate the existence of the features declared thereafter, but does not exclude the addition of other features. It should also be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In the description of the present invention, it should also be noted that the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0044] The case is described in detail below with reference to the accompanying drawings in the specification.

[0045] See also Figure 1 The present invention provides a photovoltaic meter data communication and dynamic processing method based on the Internet of Things, which specifically includes the following steps:

[0046] Step 1: Multimodal data encryption collection

[0047] This step collects raw data such as current, voltage, ambient temperature, light intensity, and grid load status in real time through sensors deployed at photovoltaic arrays and grid nodes.

[0048] To prevent data from being stolen or tampered with during transmission, dynamic encryption technology is used to protect the original data. Specifically, the system constructs a generative adversarial network (GAN). The generator performs pixel-level XOR operations on the original meter data and the random encryption matrix to generate an encrypted data stream that appears random but retains data characteristics; the discriminator continuously learns the difference between the real data and the encrypted data, and ultimately makes the encrypted data indistinguishable from the real data in terms of statistical characteristics. For example, when the light sensor detects an irradiance of 800W / m 2 When the data is received, the generator will convert it into a specific encrypted format (such as "1011→0100") to ensure that even if the data is intercepted, the real physical value cannot be directly parsed out.

[0049] The encrypted data is transmitted to the cloud server via an IoT communication module (such as LoRa or NB-IoT).

[0050] The encryption process is defined as the XOR operation of the generator G and the random key matrix K, as shown in formula (1):

[0051]

[0052] Where X is the original meter data, Indicates pixel-level XOR operation, K is the dynamically generated encryption key, E is the encrypted data stream, and the encrypted data stream E is generated by the generator G. After the cloud receives E, it needs to use the key K to decrypt and restore the original data Then perform feature decoupling on X′.

[0053] Discriminator D φ Minimize the distinction between encrypted data and real data through adversarial training (Formula 2):

[0054]

[0055] By iteratively optimizing G and D, the distribution of encrypted data E is finally G Approximate the real data distribution P data ;

[0056] Generator G performs XOR operation on the original data and the random key matrix For example, convert the current value "10A" into binary "1010", and then XOR it with the key "1100" to get the encrypted value "0110" (corresponding to 6A). φThrough adversarial training, it is ensured that the encrypted data (such as "0110") is indistinguishable from the real data (such as "1010") in terms of statistical distribution, and cannot be restored even if it is intercepted.

[0057] Step 2: Data feature decoupling and reconstruction

[0058] After the encrypted data reaches the cloud, it is necessary to separate the core factors that affect power generation efficiency (such as light changes, component aging, shadows, etc.). The system uses mathematical derivative analysis methods to calculate the sensitivity of different environmental parameters to power generation. For example, by analyzing historical data, it is found that when the component temperature increases by 1°C, the power generation efficiency decreases by 0.4%, and when the irradiance increases by 100W / m 2 , the current output is increased by 2.1A. Based on these sensitivity coefficients, the system decomposes the mixed data stream into independent time-space feature components: the time component reflects the impact of the day and night cycle and weather mutations, and the space component identifies the performance differences of different photovoltaic panel groups. In the final reconstructed feature matrix, each column corresponds to an independent influencing factor (such as "temperature influence factor column" and "irradiance response column"), providing structured input for subsequent risk analysis.

[0059] To calculate the sensitivity of different environmental parameters to power generation, eigenvalue decomposition is first used to separate the spatiotemporal features (Equation 3):

[0060] E=T·S+ε(3)

[0061] Among them, T is the temporal modal matrix (reflecting the diurnal cycle and weather mutations), S is the spatial modal matrix (identifying component performance differences), and ε is the noise term;

[0062] By solving the coupled characteristic equation (Equation 4):

[0063]

[0064] Get the eigenvalue λ t , s , respectively represent the sensitivity of environmental parameters to power generation (such as temperature sensitivity λ t =-0.4% / ℃, irradiance sensitivity λ s =2.1A / 100W / m 2 ).

[0065] Step 3: Dynamic risk modeling

[0066] Based on the decoupled characteristic data, the system simulates the risk propagation process of photovoltaic systems in complex environments. For example, high temperature weather may cause components to overheat, and the risk will spread from local hot spots to surrounding areas like "heat diffusion". The mathematical model associates the risk diffusion rate with parameters such as temperature gradient and wind speed: when the temperature in a certain area reaches 60°C, the risk diffusion coefficient D increases to 0.08, causing the risk value of adjacent components to increase by 15% within 10 minutes. At the same time, the system introduces the concept of "risk carrying capacity". For example, under strong light conditions (irradiance>1000W / m 2 ), the maximum risk tolerance of the component K c Increased to 0.9 (full value is 1), and in rainy weather K c The Alternating Direction Implicit (ADI) algorithm updates the risk heat map of the entire field every 5 minutes, with red areas indicating high-risk locations that require immediate intervention (such as abnormal temperature at a certain inverter connection point).

[0067] Specifically, the wildfire risk assessment method is used to establish a photovoltaic system risk model. First, the risk propagation equation (Equation 5) is defined:

[0068]

[0069] The diffusion coefficient D = 0.05exp(0.2T / °C) represents the speed of risk propagation, and T is the component temperature. For example, when the temperature T = 50°C, D = 0.05×22026≈1101, indicating that the risk in the high temperature area will spread rapidly.

[0070] H represents the risk value at the spatial position (x, y) at time t;

[0071] Carrying capacity K c =1-exp(-I / 1000), I is the irradiance; when irradiance I = 1000W / m 2 When K c =1-e -1 ≈0.63, indicating the maximum risk tolerance of the system under the current illumination;

[0072] β1 represents the growth coefficient, which is set to 0.1, indicating the natural accumulation rate of risk (e.g., the risk value increases by 0.1× the current value every 5 minutes).

[0073] Then the ADI method is used to discretize the risk propagation equation (Equation 5), and the process is shown in Equations (6) and (7):

[0074]

[0075] Where Δt is the time step (uniformly in seconds); represents the spatial second-order difference operator; the explicit Euler method is used to discretize the equation, the time step Δt = 300 seconds (i.e. 5 minutes), and the risk heat map H(x, t) is updated every 5 minutes, which is consistent with the data update cycle.

[0076] Step 4: Real-time optimization scheduling

[0077] Combined with the real-time power price signal P(t) of the power grid (such as 1.2 yuan / kWh during peak hours and 0.3 yuan / kWh during valley hours) and the risk heat map H(x,t), the system dynamically adjusts the power dispatch strategy. For example, when the risk value of a certain area exceeds 0.7, the output power of the area is automatically reduced by 10% to avoid equipment overload; at the same time, the power of low-risk areas is preferentially called during peak electricity price periods to maximize revenue. The optimization process is achieved through deep reinforcement learning: the neural network simulates the long-term benefits under different dispatch strategies (such as total revenue in the next 24 hours = electricity sales revenue - equipment maintenance cost), and back-propagates to correct the strategy parameters. For example, when a thunderstorm is predicted in the afternoon, the system reserves electricity in low-risk periods in advance and releases it during peak electricity price periods, increasing the daily revenue by 8% to 12%.

[0078] Combining the electricity price signal P(t) and the risk heat map H(x, t), the optimization objective function is defined (Equation 8):

[0079]

[0080] Where Q(t) is the power generation and β2 is the risk cost coefficient. The Hamilton-Jacobi-Bellman equation (Equation 9) is solved by deep reinforcement learning:

[0081]

[0082] Among them, V is the value function, u is the scheduling strategy (such as reducing the power of high-risk areas by 10%), and r(t,u) is the immediate benefit.

[0083] Step 5: Anomaly Detection and Self-Healing

[0084] The system continuously compares the deviation between the actual power generation and the predicted value to locate potential faults. For example, if the actual current of a group of photovoltaic panels is 30% lower than the predicted value, and the residual index exceeds the threshold for three consecutive times, the fault diagnosis process is triggered. The algorithm first eliminates environmental factors (such as cloud cover), and then determines the fault location through topological analysis: if multiple adjacent components are abnormal at the same time, it is judged to be a junction box failure; if only a single component is abnormal, it may be that the panel is damaged or the wiring is detached. After the fault is determined, the system automatically switches to the backup circuit and maintains power supply through redundant lines. For example, when the No. 5 string is detected to be abnormal, the control relay disconnects it from the main circuit and enables the pre-stored No. 12 backup string to be connected. The whole process is completed within 200ms.

[0085] Define the abnormal index A (Formula 10):

[0086]

[0087] Where TV(r) is the total variation regularization term, which is used to penalize severe power fluctuations; r t is the deviation between the actual power and the predicted value; ω τ is the time-decay weight (e.g., weight 0.5 when τ = 1, 0.3 when τ = 2), highlighting recent anomalies;

[0088] Step 6: Distributed storage verification

[0089] To ensure the credibility of data storage, the system divides the encrypted data into blocks and stores them in multiple edge nodes (such as regional gateways and local servers). Each data block is attached with a "commitment tag" based on a mathematical problem, such as using a lattice to generate an unforgeable hash value. When the data integrity needs to be verified, the verifier does not need to obtain the original data, but only needs to confirm through a zero-knowledge proof protocol that the storage node actually holds the correct data block. For example, the commitment value of a data block stored by a node is Com(E i )=H(K i ·E i +e i ), where K i is a common matrix, e i is the tolerance noise. The verification is verified by zero-knowledge proof (Equation 11 and Equation 12):

[0090]

[0091] And H(K i ·E i +e i )=Com t (12)

[0092] Step 7: Visualization and User Interaction

[0093] The system integrates the processing results of each link into a visual interface to help operation and maintenance personnel intuitively grasp the overall status. For example:

[0094] Risk heat map: Use color gradient to display the risk distribution of the entire field. Click the red area to view specific parameters (such as "Coordinate (3,5) risk value 0.85, temperature 62℃");

[0095] Power curve: The predicted value and the actual value are displayed in a superimposed manner, and the abnormal points are automatically marked with possible reasons (such as "power drop at 14:25: cloud cover caused a sudden drop in irradiance");

[0096] Equipment topology: The connection relationship of the PV array is presented in three dimensions, and faulty equipment flashes to indicate the fault and provides maintenance suggestions (such as "Replace the MC4 connector of string 5").

[0097] Users can use gestures to zoom in and out to view details, or set threshold alerts (such as "send SMS notifications when risk > 0.8"). The interface data is refreshed every 30 seconds and supports multi-terminal access (PC, tablet, AR glasses).

[0098] See also Figure 2 The present invention provides a photovoltaic meter data communication and dynamic processing system based on the Internet of Things, characterized in that the system includes the following modules:

[0099] Data acquisition and encryption module, which is used to collect raw data in real time through sensors deployed in photovoltaic arrays and grid nodes, including current, voltage, ambient temperature, light intensity and grid load status, and encrypt the collected raw data.

[0100] Specifically, the generator receives the original meter data and generates a random encryption matrix; performs an XOR operation on the encryption matrix and the original data element by element to generate an encrypted data stream; and transmits the encrypted data stream to the cloud server through an IoT communication module (such as LoRa or NB-IoT).

[0101] Data decryption and feature decoupling module, which is used to receive the encrypted data stream transmitted from the data acquisition and encryption module, decrypt the data, and calculate the sensitivity of environmental parameters to power generation through derivative analysis method, decouple the time component and space component in the data stream, and generate a feature matrix. Specifically, it includes: decrypting the encrypted data stream to restore the original data; calculating the sensitivity coefficient of different environmental parameters (such as temperature, irradiance) to power generation; decomposing the data into time component and space component, and reconstructing the feature matrix.

[0102] Dynamic risk modeling and updating module: This module builds a dynamic risk model for the photovoltaic system based on the feature matrix generated in step 2, and updates it in real time according to factors such as temperature, wind speed, and irradiance to generate a full-field risk heat map. Specifically, it includes: calculating the risk diffusion speed and updating the risk heat map every 5 minutes through the ADI algorithm; dynamically adjusting the risk diffusion coefficient and risk carrying capacity threshold; marking and processing high-risk areas.

[0103] The power dispatch and optimization module uses a deep reinforcement learning algorithm to dynamically adjust the power output strategy and optimize the power call of the photovoltaic system based on the real-time power price signal and risk heat map of the power grid. Specifically, it includes: combining the power price signal of the power grid and the risk heat map to adjust the power output in high-risk areas to avoid overload; optimizing the power call in low-risk areas to maximize the system benefits; and continuously adjusting the power dispatch strategy through the reinforcement learning algorithm to optimize the long-term benefits of the system.

[0104] The module for abnormal detection and fault location is used to monitor the deviation between the power generation of the photovoltaic system and the predicted value. When the residual exceeds the preset threshold, the module performs fault diagnosis and locates the fault location. Specifically, it includes: calculating the deviation of the power generation and determining the source of the fault through topological analysis; eliminating the influence of environmental factors and accurately locating the fault location; and automatically switching to the backup circuit to ensure normal power supply of the system.

[0105] The data storage and verification module is used to store the encrypted data in multiple edge nodes and attach commitment tags to ensure the integrity of the data. Specifically, it includes: storing the encrypted data in blocks in multiple edge nodes to avoid single point failure; attaching a commitment tag generated based on a lattice cipher to each data block; and using a zero-knowledge proof protocol to verify the integrity of the data to ensure the credibility of the data.

[0106] Visualization and user interaction module, which is used to generate and display the visual interface of the system, help operation and maintenance personnel to intuitively grasp the operating status of the system, and provide threshold alarm function. Specifically, it includes: displaying risk heat map, intuitively presenting the risk status of each area; displaying power curve, and automatically marking possible abnormal reasons; displaying device topology, and providing real-time maintenance suggestions for faulty equipment; this module supports multi-terminal access, provides access interfaces for mobile phones, tablets, PCs and other devices, and sets real-time alarm function.

[0107] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A photovoltaic meter data communication and dynamic processing method based on the Internet of Things, characterized in that: The specific steps include: Step 1: Use sensors deployed at photovoltaic arrays and grid nodes to collect raw data of current, voltage, ambient temperature, light intensity and grid load status in real time, encrypt the raw data using dynamic encryption technology, generate encrypted data streams, and transmit them to the cloud server through the Internet of Things communication module; Step 2: decrypt the encrypted data stream in the cloud server, use the derivative analysis method to calculate the sensitivity coefficient of the environmental parameters to the power generation, decouple the mixed data stream into time components and space components, and reconstruct a feature matrix containing a temperature influence factor column and an irradiance response column; Step 3: Based on the characteristic matrix, a dynamic risk model of the photovoltaic system is established. The dynamic risk model updates the full-field risk heat map every 5 minutes through the alternating direction implicit (ADI) algorithm, and dynamically adjusts the risk diffusion coefficient and the risk carrying capacity threshold according to the temperature gradient, wind speed and irradiance; Step 4: Combine the real-time electricity price signal of the power grid and the risk heat map, generate a power dispatch strategy through a deep reinforcement learning algorithm, dynamically adjust the output power of high-risk areas and optimize the power call of low-risk areas; Step 5: Compare the deviation between the actual power generation and the predicted value in real time. When the residual index exceeds the preset threshold for three consecutive times, locate the fault location through topological analysis and automatically switch to the backup circuit to maintain power supply. Step 6: The encrypted data is divided into blocks and stored in multiple edge nodes. A commitment tag generated based on the lattice cryptography is attached to each data block, and the data integrity is verified through a zero-knowledge proof protocol. Step 7: Generate a visualization interface to dynamically display the risk heat map, power curve, and device topology relationship, and support multi-terminal access and threshold alarm functions.

2. The method according to claim 1, characterized in that The dynamic encryption technology in step 1 includes: The generator receives the original meter data and generates a random encryption matrix; Performing an XOR operation on the random encryption matrix and the original data element by element to generate an encrypted data stream that retains data characteristics; The discriminator simultaneously verifies the indistinguishability of the encrypted data stream and the real data through adversarial training until the statistical characteristics and feature correlation indicators in the time-frequency domain reach the preset threshold.

3. The method according to claim 1, characterized in that The calculation of the sensitivity coefficient in step 2 includes: Eigenvalue decomposition is used to separate spatiotemporal characteristics to obtain a time modal matrix reflecting the diurnal cycle and weather changes, as well as a spatial modal matrix identifying component performance differences. By solving the coupled characteristic equations, the eigenvalues ​​representing the sensitivity of environmental parameters to power generation and the irradiance sensitivity are obtained.

4. The method according to claim 1, characterized in that it comprises: In step five, if multiple adjacent components are abnormal at the same time, it is judged that the junction box is faulty; if only a single component is abnormal, it may be that the panel is damaged or the wiring is detached. After the fault is determined, the system automatically switches to the backup circuit and maintains power supply through redundant lines.

5. A photovoltaic meter data communication and dynamic processing system based on the Internet of Things, characterized in that: include: Data collection and encryption module, which is used to collect raw data in real time through sensors deployed at photovoltaic arrays and grid nodes, including current, voltage, ambient temperature, light intensity and grid load status, and encrypt the collected raw data; Data decryption and feature decoupling module, which is used to receive the encrypted data stream transmitted from the data acquisition and encryption module, decrypt the data, calculate the sensitivity of environmental parameters to power generation through derivative analysis method, decouple the time component and space component in the data stream, and generate a feature matrix; Dynamic risk modeling and updating module: This module builds a dynamic risk model for the photovoltaic system based on the feature matrix, and updates it in real time according to factors such as temperature, wind speed, and irradiance to generate a full-field risk heat map; The power dispatch and optimization module uses a deep reinforcement learning algorithm to dynamically adjust the power output strategy and optimize the power call of the photovoltaic system based on the real-time power price signal and risk heat map of the power grid; Anomaly detection and fault location module, which is used to monitor the deviation between the power generation of the photovoltaic system and the predicted value. When the residual exceeds the preset threshold, fault diagnosis is performed and the fault location is located; The data storage and verification module is used to store encrypted data in multiple edge nodes and attach commitment tags to ensure the integrity of the data; Visualization and user interaction module, which is used to generate and display the system's visual interface, help operation and maintenance personnel intuitively understand the system's operating status, and provide threshold alarm function.

Images