Intelligent photovoltaic power generation regulation and management system based on internet of things

Abstract

The application relates to the cross field of photovoltaic power generation intelligent regulation and Internet of Things, and discloses a photovoltaic power generation intelligent regulation and management system based on the Internet of Things. The system adopts a three-layer architecture of a sensing layer, an edge network layer and a cloud platform layer, and solves the problems of edge communication link congestion of a distributed photovoltaic cluster, transmission time delay imbalance of regulation and control instructions and power matching imbalance of a photovoltaic array. The sensing layer classifies and collects multi-dimensional data and uploads the data; the edge network layer realizes independent transmission of data and instructions and dynamic allocation of bandwidth through link differentiation scheduling, combines instruction priority allocation and data preprocessing, reduces transmission volume and guarantees the transmission efficiency of key instructions; and the cloud platform layer dynamically adjusts the transmission time sequence of instructions through a time delay monitoring and power matching cooperation module, combines regional clustering, time sequence regulation, feedback iteration and time delay prediction, and dynamically adjusts the transmission time sequence of instructions. Through the whole-process solution, the application guarantees the real-time performance and stability of the transmission of regulation and control instructions, and realizes dynamic and accurate matching of photovoltaic power.

Description

Technical Field

[0001] This invention relates to the intersection of intelligent regulation and control of photovoltaic power generation and the Internet of Things (IoT), and particularly to an intelligent regulation and control management system for photovoltaic power generation based on the Internet of Things. Background Technology

[0002] With the maturation of photovoltaic power generation technology and the promotion of new energy policies, photovoltaic power generation is gradually developing towards distributed and clustered development. The decentralized deployment of large-scale photovoltaic nodes has placed higher demands on intelligent control and management. The integration of Internet of Things (IoT) technology provides support for real-time photovoltaic data collection and remote control. However, the operation of photovoltaic clusters requires high-frequency collection of multi-dimensional data such as photovoltaic status, environmental parameters, and grid operation. At the same time, it is necessary to quickly issue control commands such as fault emergency response and power optimization. The coordination efficiency of data transmission and command execution directly affects power generation efficiency and grid connection stability.

[0003] Existing IoT-based photovoltaic (PV) control systems mostly employ a design where sensing data and control commands share a common communication link, without differentiated allocation of link resources. As the scale of PV nodes rapidly increases, link congestion easily occurs, leading to increased transmission latency of control commands and latency imbalances among nodes. This latency imbalance causes asynchronous control responses from PV nodes, resulting in power mismatch in the PV array, reduced overall power generation efficiency, and even impacting the safe and stable operation of the power grid. Current solutions primarily focus on optimizing cloud-based scheduling algorithms or improving the communication performance of individual devices, which is insufficient to meet the precise control requirements of distributed PV clusters. Summary of the Invention

[0004] This invention solves three core problems in distributed photovoltaic clusters: edge communication link congestion caused by the shared communication link between sensing data and control commands; imbalance in the transmission delay of control commands; and imbalance in photovoltaic array power matching caused by asynchronous control responses of photovoltaic nodes.

[0005] To address the aforementioned technical problems, this invention provides an intelligent control and management system for photovoltaic power generation based on the Internet of Things (IoT), comprising a sensing layer, an edge network layer, and a cloud platform layer. The sensing layer is communicatively connected to the edge network layer, and the edge network layer is communicatively connected to the cloud platform layer. Its key feature is: The edge network layer is equipped with a link differentiation scheduling module and an instruction priority allocation module. The link differentiation scheduling module logically separates the sensing data uploaded by the sensing layer from the control instructions issued by the cloud platform layer and allocates independent transmission links. It dynamically allocates link bandwidth resources according to the real-time load status of the links. The instruction priority allocation module assigns transmission priorities to control instructions according to control requirements, with higher priority control instructions taking priority in occupying transmission resources. The cloud platform layer is equipped with a latency monitoring and power matching coordination module, which collects the transmission latency data of the control commands of each edge node in real time, and adjusts the timing of the control command issuance of each photovoltaic node according to the latency data, so that the synchronization of the control response of each photovoltaic node matches the power output of the photovoltaic array.

[0006] Preferably, the sensing layer includes a photovoltaic status acquisition unit, an environmental parameter acquisition unit, and a power grid parameter acquisition unit. Each acquisition unit classifies the acquired data according to preset rules and uploads the classified sensing data to the edge network layer. The sensing data includes photovoltaic operating status data, environmental monitoring data, and power grid operating parameter data.

[0007] Preferably, the edge network layer is further provided with an edge data preprocessing module to filter, aggregate and reduce noise in the sensing data uploaded by the sensing layer, and remove redundant and invalid data.

[0008] Preferably, the link differentiation scheduling module is equipped with a bandwidth adaptive adjustment submodule, which dynamically and adaptively adjusts the bandwidth of the separated transmission links based on the real-time transmission volume of sensing data and control commands.

[0009] Preferably, the delay monitoring and power matching coordination module includes a delay data acquisition submodule, a timing control submodule, and a power matching submodule. The delay data acquisition submodule acquires the delay data of each edge node in real time and generates a delay distribution map. The timing control submodule adjusts the timing of the control command issuance for the corresponding photovoltaic node according to the delay distribution map. The power matching submodule verifies and optimizes the timing of the control command issuance according to the real-time power output status of the photovoltaic array, thereby achieving dynamic matching of power output.

[0010] Preferably, the cloud platform layer is further provided with an intelligent decision-making module and a data storage module. The intelligent decision-making module generates the optimal control instructions based on the sensing data, latency data and power grid dispatching requirements. The data storage module stores the sensing data, latency data, control instructions and photovoltaic array power data throughout the entire lifecycle.

[0011] Preferably, the cloud platform layer is further provided with an intelligent decision-making module and a data storage module. The intelligent decision-making module generates the optimal control instructions based on the sensing data, latency data and power grid dispatching requirements. The data storage module stores the sensing data, latency data, control instructions and photovoltaic array power data throughout the entire lifecycle.

[0012] Preferably, the cloud platform layer is also equipped with a latency prediction module, which predicts the transmission latency of control commands for each edge node based on historical latency data and the operating status of the photovoltaic cluster. The latency monitoring and power matching coordination module combines the predicted latency data to adjust the timing of the control commands issued by the photovoltaic nodes in advance, thereby achieving forward-looking power matching control.

[0013] Preferably, the edge data preprocessing module includes a machine learning-based intelligent data classification submodule and a sampling frequency dynamic adaptation submodule. The intelligent data classification submodule combines the dynamic characteristics of the photovoltaic cluster operation scenario and the contribution weight of the sensing data to the control decision. Through a preset multi-dimensional feature recognition model, the sensing data is divided into core decision data, key operation data and redundant auxiliary data. Among them, photovoltaic module fault characteristic data and grid frequency / voltage change data are classified as core decision data, photovoltaic array conventional operation parameter data are classified as key operation data, and repetitive redundant data of non-core environmental monitoring are classified as redundant auxiliary data. The sampling frequency dynamic adaptation submodule dynamically adjusts the sampling period of the corresponding sensing and acquisition unit based on the priority ranking results of each data level and the real-time load utilization of the link in the edge network layer. Core decision data maintains a preset maximum sampling frequency to ensure the timeliness of the adjustment response, key operating data is sampled using adaptive frequency conversion according to the link load utilization, and redundant auxiliary data uses a preset minimum sampling frequency to transmit data volume.

[0014] Preferably, the delay monitoring and power matching coordination module further includes a photovoltaic node region clustering submodule and a control parameter feedback iteration submodule; The photovoltaic node regional clustering submodule divides the photovoltaic cluster into several independent collaborative control zones based on the geographical deployment characteristics of photovoltaic nodes, the network topology of IoT communication links, and historical transmission delay fluctuation characteristics. Photovoltaic nodes within the same collaborative control zone share a unified delay calibration benchmark and control timing benchmark. The regulation parameter feedback iteration submodule collects the power output matching error value of each collaborative regulation zone in real time, and corrects the time delay regulation threshold and the timing offset of the regulation command issuance in reverse according to the error amplitude, thus constructing a closed-loop collaborative regulation mechanism, including real-time time delay acquisition, dynamic adjustment of regulation timing, power output matching verification, and iterative optimization of regulation parameters.

[0015] Compared with related technologies, the IoT-based intelligent control and management system for photovoltaic power generation provided by this invention has the following advantages: 1. This solution achieves independent transmission and dynamic bandwidth allocation of two types of data through the link differentiation scheduling module of the edge network layer. Combined with the instruction priority allocation module, it ensures the transmission of high-priority instructions. Furthermore, the intelligent hierarchical classification and frequency conversion sampling of the edge data preprocessing module reduce the amount of data, thereby reducing the link load from the source and significantly improving the real-time performance and stability of control instruction transmission. This avoids control delays caused by link congestion for critical instructions.

[0016] 2. This solution utilizes a cloud platform layer latency monitoring and power matching coordination module, combined with regional clustering, timing control, and feedback iteration mechanisms, to dynamically adjust the command issuance timing based on the latency data of each node. This verifies and optimizes the power output matching degree, ensuring that the response synchronization of each photovoltaic node is accurately matched with the array power output, fundamentally improving the power matching imbalance and ensuring the safe and stable operation of the power grid.

[0017] 3. This solution incorporates a latency prediction module, which predicts latency changes based on historical data and cluster operating status, enabling proactive timing adjustments. The data storage module stores multi-dimensional data throughout the entire lifecycle, providing data support for intelligent decision-making, latency prediction, and parameter iterative optimization. The system adopts a layered architecture and modular design, adaptable to various scenarios such as distributed clusters and centralized power plants. Through a closed-loop control mechanism, it dynamically adapts to the communication characteristics and power requirements of different scenarios, improving control accuracy and long-term operational performance.

[0018] In summary, this invention provides a comprehensive solution encompassing link optimization, latency control, power matching, and forward-looking regulation. This solution effectively addresses three core issues in distributed photovoltaic (PV) clusters: edge communication link congestion, latency imbalance of regulation commands, and power matching imbalance of PV arrays. It ensures the real-time performance and stability of regulation command transmission, achieves dynamic and precise matching of PV power output, and possesses strong adaptability to various scenarios and continuous optimization capabilities. This significantly improves the overall efficiency of PV power generation and the stability of grid connection. Attached Figure Description

[0019] Figure 1 A three-layer architecture connection diagram of the IoT-based intelligent control and management system for photovoltaic power generation provided by the present invention; Figure 2 The logical block diagram of the core module of the edge network layer provided by this invention; Figure 3 This is an interaction diagram of the core modules of the cloud platform layer provided by the present invention; Figure 4 The core data flow diagram of the system provided by this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “group,” “class,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0022] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0023] Please refer to the following: Figures 1-4 A photovoltaic power generation intelligent control and management system based on the Internet of Things (IoT) includes a sensing layer, an edge network layer, and a cloud platform layer. The sensing layer communicates with the edge network layer, and the edge network layer communicates with the cloud platform layer. The specific collaborative logic is as follows: the sensing layer collects and uploads multi-dimensional data according to classification rules; the edge network layer reduces transmission volume through data preprocessing, ensures transmission efficiency through differentiated link scheduling, and ensures the timeliness of key instructions through instruction priority allocation; the cloud platform layer collects latency data in real time, realizes photovoltaic node response synchronization through regional clustering and time-series control, iteratively optimizes control parameters by combining power matching error feedback, and finally generates the optimal control instruction through intelligent decision-making. After being proactively adjusted by the latency prediction module, the instruction is issued to achieve precise collaborative control of the photovoltaic cluster. The sensing layer includes a photovoltaic status acquisition unit, an environmental parameter acquisition unit, and a power grid parameter acquisition unit. Each acquisition unit classifies the acquired data according to preset rules and uploads the classified sensing data to the edge network layer. The sensing data includes photovoltaic operating status data, environmental monitoring data, and power grid operating parameter data. The specific implementation process of each unit is as follows: Each data collection unit collects data in a targeted manner according to a preset classification logic: The photovoltaic status acquisition unit focuses on the operating status parameters of photovoltaic modules and acquires data such as voltage, current, and temperature of photovoltaic modules in real time through module-level sensors; The environmental parameter acquisition unit collects environmental monitoring data from the photovoltaic deployment area, and obtains data such as irradiance, wind speed, and ambient temperature through environmental sensors deployed in the photovoltaic array area; The power grid parameter acquisition unit acquires the operating parameters of the power grid side and collects data such as power grid frequency, voltage and grid-connected power in real time through the power grid side metering device; During the data collection process, each unit uses a built-in identification module to classify and label the collected data, ensuring that the corresponding data source and data type can be quickly identified after the data is uploaded to the edge network layer. To achieve standardized classification and transmission of data, the classification identifier for perceived data is denoted as follows: Its value is 1, 2, or 3, corresponding to photovoltaic operation status data, environmental monitoring data, and power grid operation parameter data, respectively; the classified sensing data of the same type are integrated to form a sensing data set, denoted as... ; Each data acquisition unit determines the attribute and source of the currently acquired data according to preset rules, determines its corresponding Td value, and then... All collected data corresponding to the value are grouped into the same category. In the process, the data is classified and organized. After classification, each acquisition unit uses IoT communication protocols (such as RS485, LoRa, NB-IoT, etc.) to transmit the data carrying... The sign Upload to the nearest edge node in the edge network layer to avoid delays caused by excessively long data transmission paths, ensuring targeted and efficient data transmission; It should be noted that the module design of this perception layer is applicable to various photovoltaic application scenarios such as distributed photovoltaic clusters, centralized ground power stations, and park microgrid photovoltaic clusters. Through the accurate collection and classification of multi-dimensional data, it can provide complete and standardized basic data support for data preprocessing of the edge network layer and intelligent decision-making of the cloud platform layer, avoiding deviations in control strategies caused by data chaos or missing data.

[0024] The edge network layer comprises a link differentiation scheduling module, an instruction priority allocation module, and an edge data preprocessing module. The edge data preprocessing module filters, aggregates, and reduces noise in the sensing data uploaded from the sensing layer, eliminating redundant and invalid data to reduce the amount of sensing data transmitted and lower the link transmission pressure. The link differentiation scheduling module logically separates the sensing data uploaded from the sensing layer from the control instructions issued by the cloud platform layer and allocates independent transmission links, dynamically allocating link bandwidth resources based on the real-time load status of the links. The instruction priority allocation module assigns transmission priorities to control instructions according to control requirements, with higher-priority control instructions occupying transmission resources first. The edge data preprocessing module includes a machine learning-based intelligent data grading submodule and a sampling frequency dynamic adaptation submodule. The specific implementation of each module is as follows: The core function of the edge data preprocessing module is to process the data uploaded from the perception layer. The system performs filtering, aggregation, and noise reduction to eliminate redundant and invalid data, reduce the amount of transmitted sensing data, and lower the transmission pressure on the link. Its intelligent data classification submodule and sampling frequency dynamic adaptation submodule work together to achieve this function. The specific implementation process is as follows: The intelligent data classification submodule, combining the dynamic characteristics of photovoltaic cluster operation scenarios and the contribution weight of sensing data to control decisions, divides the sensing data into core decision-making data, key operational data, and redundant auxiliary data through a pre-set multi-dimensional feature recognition model. Specifically, photovoltaic module fault characteristic data and grid frequency / voltage mutation data are classified as core decision-making data, photovoltaic array routine operating parameter data as key operational data, and repetitive redundant data from non-core environmental monitoring as redundant auxiliary data. The specific implementation is as follows: The preset data contribution weight is denoted as Specifically, by analyzing the impact of various sensing data on the control decision results in historical control cases, and combining the dynamic characteristics of the real-time operation scenario of photovoltaic clusters (such as whether it is in a period of high failure incidence, whether there are fluctuations in the power grid, etc.), the results are obtained through machine learning model training, which is used to quantify the importance of different sensing data to control decisions. Let the multi-dimensional feature vector of the perceived data be denoted as... The core dimensions of the perceived data, such as timeliness, scope of impact, and decision-making relevance, are extracted. The timeliness feature is derived from the interval between the data collection time and the current time, the scope of impact feature is obtained by counting the number of photovoltaic nodes associated with the data, and the decision-making relevance feature is obtained by analyzing the matching degree between historical data and control instructions. The feature bias term, denoted as b, is used to correct system errors caused by factors such as sensor accuracy and environmental interference during feature extraction. A multi-dimensional feature recognition model is used for data classification, and a multi-class activation function is introduced. This is used to map the feature weighting result to the corresponding classification identifier, and the data classification result is denoted as: Its value is 1, 2, or 3, corresponding to core decision data, key operational data, and redundant auxiliary data, respectively; the calculation formula for data classification is as follows: The data classification processing logic is as follows: and After performing weighted calculations and adding b for error correction, the input is... The function determines the category of the perceived data by identifying the maximum value of the function's output. Values ​​are selected to complete data classification; among them, photovoltaic module fault characteristic data and grid frequency / voltage mutation data correspond to The corresponding normal operating parameter data of photovoltaic array The duplicate and redundant data corresponding to non-core environmental monitoring .

[0025] The sampling frequency dynamic adaptation submodule dynamically adjusts the sampling period of the corresponding sensing and acquisition unit based on the priority ranking results of each data level and the real-time load utilization of the edge network layer links. Core decision data maintains a preset maximum sampling frequency to ensure the timeliness of the control response, key operational data is sampled using adaptive frequency conversion according to the link load utilization, and redundant auxiliary data uses a preset minimum sampling frequency to reduce the amount of data transmitted. While further reducing the transmission pressure of the edge communication link, it ensures the integrity and real-time performance of the core data required for intelligent control. The specific implementation is as follows: The real-time load utilization rate of the edge network layer links is denoted as... The link monitoring unit at the edge network layer calculates the ratio of the current link's used bandwidth to the total bandwidth in real time, providing a direct reflection of the link's transmission pressure. The preset maximum sampling frequency is denoted as Based on the control and response requirements of core decision data, and combined with the preset performance of sensor hardware, ensure the real-time capture of emergency data; The preset minimum sampling frequency is denoted as Based on the actual application value of redundant auxiliary data and combined with the preset system energy consumption control requirements, the amount of transmission is reduced while ensuring basic information. Let the sampling frequency adjustment coefficient be denoted as ,pass The ratio to the preset link load baseline value is derived from this. The higher, The larger the value, the more it can significantly reduce the sampling frequency of non-core data when the link load is too high; The real-time sampling frequency of the acquisition unit is denoted as... According to data hierarchy and Dynamic adjustment; Based on the data classification results , combined and The following formula is used to... Make dynamic adjustments: when (Core decision data) when: The highest sampling frequency is always maintained to ensure that emergencies such as photovoltaic module failures and power grid sudden changes can be captured in a timely manner. when (Critical runtime data) It adaptively adjusts the sampling frequency according to the link load. When the link load is low, the sampling frequency is increased to obtain more detailed data, and when the load is high, the frequency is reduced to alleviate the transmission pressure. when When there is redundant auxiliary data: Use the lowest sampling frequency to reduce invalid data transmission; The calculated The data is sent to the corresponding sensing and acquisition unit, and the sampling period of the unit is dynamically adjusted to achieve transmission optimization that prioritizes core data and simplifies non-core data as needed. This ensures the core data requirements for intelligent control while minimizing the transmission pressure on the edge communication link.

[0026] The link differentiation scheduling module logically separates the sensing data uploaded from the sensing layer from the control commands issued by the cloud platform layer and allocates them to independent transmission links. It dynamically allocates link bandwidth resources based on the real-time load status of the links. The link differentiation scheduling module also includes a bandwidth adaptive adjustment submodule. This submodule dynamically and adaptively adjusts the bandwidth of the separated transmission links based on the real-time transmission volume of sensing data and control commands, ensuring sufficient bandwidth resources for the control command transmission links. The specific implementation is as follows: Let the total bandwidth resources of the edge network layer be denoted as... The total schedulable communication bandwidth of the current edge node is obtained in real time through the bandwidth monitoring unit of the edge network layer; The real-time transmission volume of the sensed data is denoted as... The real-time byte transmission rate of the preprocessed sensing data currently received by the edge network layer is statistically analyzed. The real-time transmission volume of the control command is denoted as The real-time byte transmission rate of control commands issued from the cloud platform layer to the edge network layer is statistically analyzed. The bandwidth allocation factor is set as follows: Based on the minimum bandwidth requirement for control command transmission and the link load status, it is dynamically derived by comparing the minimum bandwidth ratio of non-blocking transmission of control commands in historical transmission data with the current... The weighted calculation is used to balance the bandwidth allocation between the sensed data and the control commands; Let the sensing data transmission link be denoted as The control command transmission link is denoted as The two are logically separate independent transmission links; Link The allocated bandwidth is denoted as ,link The allocated bandwidth is denoted as ; The link differentiation scheduling module allocates independent transmission links for sensing data and control commands. and This achieves physical isolation between the two links, avoiding mutual contention caused by sharing the link. Subsequently, the bandwidth resources of the two links are dynamically allocated using the following formula: , The processing logic for dynamically allocating bandwidth resources between two links is as follows: first, based on... and The bandwidth is initially allocated based on the proportion of bandwidth used, and then... Correction Get the value, ensure Always meet the minimum requirements for the transmission of control commands, and avoid the bandwidth of control commands being squeezed due to excessive data transmission of sensing data. Real-time monitoring of the bandwidth adaptive adjustment submodule and Load status: when When the load rate exceeds the preset threshold, automatically increase The value of is reduced And supplement ;when When the load rate is too high, reduce it appropriately. This balances the transmission pressure on the two links, ensuring the stability and efficiency of link transmission.

[0027] The instruction priority allocation module assigns transmission priorities to control instructions according to control requirements. Higher-priority control instructions occupy transmission resources first, as implemented in the following way: The urgency coefficient of the control order is denoted as: Based on the control scenarios corresponding to the instructions, scenarios such as fault emergency control and grid connection emergency control are deduced. By quantifying the "severity level of faults / grid fluctuations," conventional optimized control... Set according to the "urgency of regulating demand"; The influence range coefficient of the control order is denoted as The number of photovoltaic nodes associated with the command and the size of the grid coverage area are counted, and the results are obtained by weighting the importance of the associated nodes / regions. The priority threshold is denoted as θ, and is determined based on the lowest priority command in historical control data. The value is dynamically adjusted based on the current power grid operating status (such as whether it is during peak electricity consumption or whether there are stability risks). The transmission priority of control commands is denoted as Its value is 1 or 2, which correspond to high priority and normal priority, respectively; The transmission priority of control commands is determined using the following formula: ; The processing logic for determining the transmission priority of control commands is as follows: and Multiply the results to obtain a comprehensive priority score for the instruction. Compare this score with θ. If the score meets the standard, the instruction is classified as high priority. This corresponds to emergency control for photovoltaic faults and emergency control for grid connection; failure to meet the standard results in a normal priority ( 2) Corresponding to instructions such as routine photovoltaic power optimization and status inspection; The transmission scheduling unit of the edge network layer is based on Priority allocation of transmission resources: priority scheduling The instructions are passed through Transmission will proceed only after high-priority instructions have been transmitted. The 2-level instruction ensures the unblocked transmission of key control instructions, avoiding control delays caused by routine instructions consuming resources. It is suitable for various photovoltaic control scenarios with high requirements for control response timeliness.

[0028] The cloud platform layer is equipped with a latency monitoring and power matching coordination module, an intelligent decision-making module, a data storage module, and a latency prediction module. The latency monitoring and power matching coordination module includes a latency data acquisition submodule, a timing control submodule, and a power matching submodule, as well as a photovoltaic node area clustering submodule and a control parameter feedback iteration submodule. The intelligent decision-making module generates optimal control commands based on sensing data, latency data, and grid dispatch requirements; the data storage module stores sensing data, latency data, control commands, and photovoltaic array power data throughout the entire lifecycle, providing data support for optimizing control strategies; the latency prediction module predicts the transmission latency of control commands for each edge node based on historical latency data and the operating status of the photovoltaic cluster; and the latency monitoring and power matching coordination module adjusts the timing of control command issuance for photovoltaic nodes in advance based on the predicted latency data, achieving proactive power matching control. The specific implementation of each module is as follows: The latency monitoring and power matching coordination module is used to collect the transmission latency data of the control commands from each edge node in real time. Based on the latency data, it adjusts the timing of the control commands issued by each photovoltaic node to ensure that the synchronization of the control response of each photovoltaic node matches the power output of the photovoltaic array, thus solving the problem of power matching imbalance in the photovoltaic array. The specific implementation of each sub-module is as follows: The latency data acquisition submodule acquires latency data from each edge node in real time and generates a latency distribution map. The specific implementation is as follows: Let the transmission delay of the control command of the i-th edge node be denoted as . This is achieved by setting a timestamp synchronization mechanism between the cloud platform layer and edge nodes: when the cloud platform layer issues a control command, it marks the sending timestamp. After receiving the instruction, the edge node immediately returns an acknowledgment message along with the receiving timestamp. The cloud platform layer performs calculations. Real-time collection of data from each edge node ; The time delay distribution map is denoted as It uses the edge node number as the horizontal axis, The value is taken as the vertical axis. This represents the values ​​of all edge nodes collected in real time. Sorted by edge node number, and generated through visualization processing. It intuitively reflects the time delay distribution characteristics and time delay differences of each edge node; Will After generation, the data is pushed to the timing control submodule and the photovoltaic node region clustering submodule in real time, enabling the two modules to quickly identify edge nodes with time delay anomalies and time delay distribution patterns. This ensures that the control strategy matches the actual transmission time delay status and avoids control deviations caused by time delay data lag.

[0029] The photovoltaic node regional clustering submodule, based on the geographical deployment characteristics of photovoltaic nodes, the network topology of IoT communication links, and historical transmission delay fluctuation characteristics, uses a density clustering algorithm to divide the photovoltaic cluster into several independent collaborative control zones. Photovoltaic nodes within the same collaborative control zone share a unified delay calibration benchmark and control timing benchmark. The specific implementation is as follows: Let the geographical deployment feature vector of the i-th photovoltaic node be denoted as The data is obtained by collecting information such as the latitude and longitude of photovoltaic nodes, installation spacing, and terrain features of the area, and then processing these information through feature standardization. Let the communication link topology feature vector of the i-th photovoltaic node be denoted as... The network topology detection tool obtains information such as the connection method between nodes and edge nodes, link length, and link type (wired / wireless), and extracts features to form the network topology detection method. Let the historical time delay fluctuation characteristic vector of the i-th photovoltaic node be denoted as... The history of this node can be retrieved through the data storage module. The data is generated by calculating statistical characteristics such as delay variance, delay mean, and maximum delay deviation. Let the comprehensive feature vector of the i-th photovoltaic node be denoted as It is used to integrate three core features: geography, topology, and historical latency. Let the characteristic distance between the i-th and j-th photovoltaic nodes be denoted as . This is used to quantify the feature similarity between nodes; Let the kth coordinated control zone be denoted as ; The cluster density threshold is denoted as ρ, which is adjusted based on the node distribution density of the photovoltaic cluster and in combination with historical clustering effect feedback, and is used to determine the core cluster nodes. The clustering distance threshold is denoted as δ. It is set according to the transmission characteristics of the communication link (such as the latency fluctuation range of the wireless link and the latency stability of the wired link) and combined with the coordination requirements of regional regulation. It is used to determine whether nodes belong to the same cluster. First, construct the comprehensive feature vector. It integrates three core features—geography, topology, and historical latency—to comprehensively reflect the regulation and correlation characteristics of photovoltaic nodes. Then, the Euclidean distance formula is used to calculate the characteristic distance between any two nodes i and j. Quantify the feature similarity between nodes; Finally, density clustering algorithm is used for partitioning: traversing all nodes, groups nodes whose density is ≥ ρ and whose density is similar to that of the node are... Nodes whose number of nodes meets the preset condition are designated as core nodes, and all nodes that meet the core node condition are... Nodes are grouped into the same This forms an independent, coordinated control zone; same The photovoltaic nodes within the region share a unified time delay calibration benchmark and control timing benchmark: the time delay calibration benchmark is taken from all nodes within the region. The average value ensures the consistency of the delay reference within the partition; the timing reference is set according to the communication characteristics and power output requirements of the partition, so that the timing control is more in line with the actual situation of the partition; through this design, the large-scale photovoltaic cluster is divided into several small-scale collaborative partitions, reducing the complexity of timing control and improving the accuracy of control.

[0030] The timing control submodule adjusts the timing of the control commands issued to the corresponding photovoltaic nodes based on the time delay distribution map. The specific implementation is as follows: Let the average time delay of the kth coordinated control zone be denoted as . By calculating all edge nodes within the partition The arithmetic mean is obtained; The system's preset reference delay is denoted as Based on the time requirements of power grid dispatch for control response and the synchronization requirements of photovoltaic array power matching, combined with the time delay data of historical optimal control, a unified benchmark is provided for timing adjustment; The timing offset for issuing instructions to the k-th coordinated control partition is denoted as... This is used to quantify the deviation between the partition latency and the ideal state; Calculated using the following formula : The processing logic is as follows: The timing offset that needs to be adjusted for a partition is obtained by using the difference between the partition's average latency and the system reference latency. ,when When this occurs, it indicates that the partition latency is greater than the reference latency; when When this occurs, it indicates that the partition latency is less than the reference latency; according to Adjust the timing of instruction issuance: when At that time, control commands are issued to the photovoltaic nodes of that partition in advance by Δtk; when At that time, the photovoltaic node delay in this partition The instruction was issued. This adjustment ensures that the actual control response time of each zone's photovoltaic nodes is consistent with... Consistency is achieved, enabling synchronized control and response across the entire cluster and resolving the issue of asynchronous responses caused by latency imbalance.

[0031] The power matching submodule verifies and optimizes the timing of control command issuance based on the real-time power output status of the photovoltaic array, thereby achieving dynamic matching of power output. The specific implementation is as follows: Let the actual power output of the kth coordinated control zone be denoted as . Data is collected in real time by the power output sensors of all photovoltaic modules within the partition, and then summarized to obtain the result. Let the target power output of the kth coordinated control zone be denoted as... It is calculated and generated by the intelligent decision-making module based on power grid dispatching requirements, environmental parameters, and the operating status of the photovoltaic array; Let the power output matching error of the k-th coordinated control zone be denoted as... This is used to quantify the degree of matching between actual power and target power; Calculated using the following formula : The processing logic is as follows: Quantify the degree of matching between the partitioned power output by using the relative error between the actual power and the target power. The larger the value, the lower the match. Compared with the preset error threshold: If If the error is less than or equal to the preset error threshold, it indicates that the current timing adjustment is effective, and the current distribution timing should be maintained; if... The preset error threshold indicates that the timing adjustment has not met the power matching requirements. Feedback is sent to the timing control submodule for recalculation. And adjust the timing until... The requirements are met; This mechanism enables the linkage between timing control and power matching, ensuring that the synchronous control response ultimately translates into power output matching, thereby improving the overall power generation efficiency and grid connection stability of the photovoltaic array.

[0032] The regulation parameter feedback iteration submodule collects the power output matching error value of each coordinated regulation zone in real time. Based on the error amplitude, it reversely corrects the time delay regulation threshold and the timing offset of the regulation command issuance for that zone, constructing a closed-loop coordinated regulation mechanism. This mechanism includes real-time time delay acquisition, dynamic adjustment of regulation timing, power output matching verification, and iterative optimization of regulation parameters. This enables the regulation strategy of each coordinated regulation zone to dynamically and accurately adapt to the communication transmission characteristics and power output coordination requirements of that zone. The specific implementation is as follows: Let the time delay control threshold of the k-th partition in the nth iteration be denoted as... The initial value is determined by the system reference delay. With the weighting of partition characteristics, subsequent iteration values ​​are derived from the previous iteration result and the power matching error; Let the time offset of the k-th partition in the nth iteration be denoted as The initial value is the value calculated for the first time by the timing control submodule. Subsequent iteration values ​​are obtained by correcting the previous offset and power matching error; The feedback adjustment coefficient is denoted as β. It is set according to the sensitivity of the power matching error to the control parameters. It is obtained by statistical analysis of the "ratio of error change to parameter adjustment" in historical iteration data to ensure that the parameter adjustment range is compatible with the error magnitude. Let the delay control threshold for the (n+1)th iteration of the k-th partition be denoted as... The time offset of the k-th partition in the (n+1)-th iteration is denoted as .

[0033] The control parameters are iteratively corrected using the following formula: , The processing logic is as follows: based on power matching error As a feedback signal, the amplitude is adjusted by controlling β for the current iteration. and Make corrections to obtain the parameter values ​​for the next iteration, thus achieving dynamic optimization of parameters as errors occur; The revised and Feedback is sent to the time-series regulation submodule and the region clustering submodule respectively: the time-series regulation submodule uses... As a new latency reference standard, This serves as the basis for new time-series adjustments; the regional clustering submodule is combined with... Optimize the clustering threshold to make the partitioning more aligned with current control needs; construct a closed loop of time-delay acquisition, time-series control, power matching, and parameter optimization through this mechanism to ensure that the control strategies of each partition are dynamically adapted to communication characteristics and power requirements, thereby improving the accuracy of power coordination matching for large-scale photovoltaic clusters.

[0034] The intelligent decision-making module generates optimal control commands based on sensing data, time delay data, and grid dispatch requirements; the data storage module stores sensing data, time delay data, control commands, and photovoltaic array power data throughout the entire lifecycle, providing data support for optimizing control strategies. The specific implementation of the intelligent decision-making module is as follows: The preprocessed sensing data is denoted as... It receives a set of perception data from the edge network layer after hierarchical, filtering and aggregation processing, which contains effective information such as core decision data and key operational data. Let the latency data set of each edge node be denoted as Obtain all edge node data from the latency data acquisition submodule. With each partition ; Let the power grid dispatch demand parameter be denoted as The information is obtained from the power grid dispatch center interface and includes core requirements such as grid-connected power limits, frequency stability requirements, and voltage fluctuation thresholds. The decision weights are denoted as follows: , , , respectively corresponding Based on the current control scenario settings (such as a scenario prioritizing power grid stability) Higher weighting, photovoltaic efficiency priority scenarios (Higher weight) Let the ideal reference values ​​for the sensed data, latency data, and power grid demand be denoted as follows: ,in Based on the optimal operating parameters of the photovoltaic array That is, system reference delay , Standard parameter settings based on the safe and stable operation of the power grid; The optimal control command is denoted as ; Generate using the following formula : The processing logic is as follows: calculate separately. and , and , and The Euclidean distance is used to find the control instruction that minimizes the weighted sum after weighted summation of decision weights. This instruction is the optimal control instruction that makes the system operating state closest to the ideal reference state. The generated The command is sent to the edge network layer, where it is allocated an independent link by the link differentiation scheduling module and the command priority allocation module determines the priority before being transmitted to the corresponding photovoltaic node for control. This module generates the optimal command by integrating multi-dimensional data to ensure that the control is both in line with the photovoltaic operating status and meets the grid dispatch requirements, while also adapting to communication latency characteristics to ensure the accuracy and effectiveness of the control.

[0035] The data storage module stores sensing data, latency data, control commands, and photovoltaic array power data throughout the entire lifecycle, providing data support for optimizing control strategies. The specific implementation is as follows: Let the full-cycle data set be denoted as It contains sensing data ( ), latency data ( ), control instructions ( ), power data ( All data is pushed to the corresponding module in real time, ensuring data integrity and timeliness. A distributed storage architecture is adopted, and data storage indexes are set. Where T is the data acquisition / generation timestamp, and Type is the data type identifier (and...). Correspondingly, 1 = sensing data, 2 = time delay data, 3 = control command, 4 = power data), ID is the identifier of the data source node (edge ​​node number / coordinated control zone number); Data is categorized and stored by Index: first, it is divided into major categories by Type, then by time dimension (such as hour / day / month) by T, and finally by ID to associate specific source nodes, forming a structured storage system to ensure that data can be quickly retrieved and accessed; It should be noted that this module enables the stored data to provide data support for various modules of the system. For example, the intelligent decision-making module retrieves historical data... The system optimizes decision weights and reference standard values; the latency prediction module trains a prediction model using historical latency data; the regulation parameter feedback iteration submodule optimizes the iteration strategy using historical power data and parameter data; and the full-cycle data can be used for system operation status tracking and fault diagnosis, providing complete data support for subsequent technology upgrades.

[0036] The latency prediction module predicts the transmission latency of control commands for each edge node based on historical latency data and the operating status of the photovoltaic cluster. The latency monitoring and power matching coordination module combines the predicted latency data to adjust the timing of control command issuance to the photovoltaic nodes in advance, thereby achieving proactive power matching control. The specific implementation is as follows: The historical latency data sequence is denoted as The data ti data of each edge node within a specified time period is retrieved from the data storage module and arranged in chronological order. The real-time operating status feature vector of the photovoltaic cluster is denoted as... Extract the real-time operating characteristics of the current photovoltaic cluster, including photovoltaic output. Environmental conditions (irradiance, temperature, from the sensing layer environmental parameter acquisition unit), link load These are formed after feature standardization. Let the prediction delay of the i-th edge node be denoted as . ; Introducing a time series prediction model, denoted as (such as LSTM, GRU, etc.), used to mine the temporal correlation between historical data and real-time status, and to achieve latency prediction; The following formula is used to... Make a prediction: ; The processing logic is: to and Input the preset time series prediction model The model learns the historical latency variation patterns as the photovoltaic cluster operates, and outputs the predicted latency of the i-th edge node within a future time period. ; Will Transmitted to the delay monitoring and power matching coordination module, which combines... Adjust the coordinated control zones in advance and This enables forward-looking power matching and regulation; for example, predicting the power distribution of a certain region. If the time offset of the partition is increased, the time offset of the partition will be increased in advance to avoid the control response lag caused by the increase in time delay, and further improve the accuracy and foresight of the control.

[0037] The complete workflow of the IoT-based intelligent control and management system for photovoltaic power generation of the present invention is as follows: The sensing layer collects multi-dimensional data through photovoltaic status acquisition units, environmental parameter acquisition units, and power grid parameter acquisition units, categorized and processed accordingly. After being categorized, the labels are formed Uploaded to the edge network layer; The edge network layer uses an intelligent data hierarchical submodule and a sampling frequency dynamic adaptation submodule to... Preprocessing is performed to reduce the amount of data transmitted; then, the link differentiation scheduling module allocates independent links and dynamically allocates bandwidth, and the instruction priority allocation module divides instruction priorities to achieve efficient transmission of data and instructions. The cloud platform layer collects data from each edge node through the latency data acquisition submodule. And generate Divided by photovoltaic node region clustering submodule Timing control submodule calculation Adjusting the command issuance timing, the power matching submodule... Verify and optimize the timing sequence, and construct a closed-loop optimization control parameter feedback iteration submodule to control the parameters. Intelligent decision-making module integration generate After the delay prediction module is combined After forward-looking adjustments, the plan was issued to photovoltaic nodes for implementation. Data storage module stores data throughout the entire process. This provides data support for latency prediction and strategy optimization.

[0038] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0039] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A photovoltaic power generation intelligent control and management system based on the Internet of Things, characterized in that, It includes a perception layer, an edge network layer, and a cloud platform layer. The perception layer is communicatively connected to the edge network layer, and the edge network layer is communicatively connected to the cloud platform layer. Its characteristic is: The edge network layer is equipped with a link differentiation scheduling module and an instruction priority allocation module. The link differentiation scheduling module logically separates the sensing data uploaded by the sensing layer from the control instructions issued by the cloud platform layer and allocates independent transmission links. It dynamically allocates link bandwidth resources according to the real-time load status of the links. The instruction priority allocation module assigns transmission priorities to control instructions according to control requirements, with higher priority control instructions taking priority in occupying transmission resources. The cloud platform layer is equipped with a latency monitoring and power matching coordination module, which collects the transmission latency data of the control commands of each edge node in real time, and adjusts the timing of the control command issuance of each photovoltaic node according to the latency data, so that the synchronization of the control response of each photovoltaic node matches the power output of the photovoltaic array.

2. The intelligent control and management system for photovoltaic power generation based on the Internet of Things as described in claim 1 is characterized in that, The sensing layer includes a photovoltaic status acquisition unit, an environmental parameter acquisition unit, and a power grid parameter acquisition unit. Each acquisition unit classifies the acquired data according to preset rules and uploads the classified sensing data to the edge network layer. The sensing data includes photovoltaic operation status data, environmental monitoring data, and power grid operation parameter data.

3. The intelligent control and management system for photovoltaic power generation based on the Internet of Things as described in claim 1 is characterized in that, The edge network layer also includes an edge data preprocessing module, which filters, aggregates, and reduces noise in the sensing data uploaded by the sensing layer, eliminating redundant and invalid data.

4. The intelligent control and management system for photovoltaic power generation based on the Internet of Things as described in claim 1 is characterized in that, The link differentiation scheduling module is equipped with a bandwidth adaptive adjustment submodule, which dynamically and adaptively adjusts the bandwidth of the separated transmission links based on the real-time transmission volume of sensing data and control commands.

5. The intelligent control and management system for photovoltaic power generation based on the Internet of Things as described in claim 1 is characterized in that, The delay monitoring and power matching coordination module includes a delay data acquisition submodule, a timing control submodule, and a power matching submodule. The delay data acquisition submodule acquires the delay data of each edge node in real time and generates a delay distribution map. The timing control submodule adjusts the timing of the control command issuance for the corresponding photovoltaic node according to the delay distribution map. The power matching submodule verifies and optimizes the timing of the control command issuance according to the real-time power output status of the photovoltaic array, thereby achieving dynamic matching of power output.

6. The intelligent control and management system for photovoltaic power generation based on the Internet of Things as described in claim 1 is characterized in that, The cloud platform layer is also equipped with an intelligent decision-making module and a data storage module. The intelligent decision-making module generates the optimal control instructions based on the sensing data, latency data and grid dispatching requirements. The data storage module stores the sensing data, latency data, control instructions and photovoltaic array power data throughout the entire lifecycle.

7. The intelligent control and management system for photovoltaic power generation based on the Internet of Things according to claim 1 is characterized in that, The cloud platform layer is also equipped with an intelligent decision-making module and a data storage module. The intelligent decision-making module generates the optimal control instructions based on the sensing data, latency data and grid dispatching requirements. The data storage module stores the sensing data, latency data, control instructions and photovoltaic array power data throughout the entire lifecycle.

8. The intelligent control and management system for photovoltaic power generation based on the Internet of Things according to claim 1 is characterized in that, The cloud platform layer is also equipped with a latency prediction module, which predicts the transmission latency of control commands for each edge node based on historical latency data and the operating status of the photovoltaic cluster. The latency monitoring and power matching coordination module combines the predicted latency data to adjust the timing of the control commands issued by the photovoltaic nodes in advance, thereby achieving forward-looking power matching control.

9. The intelligent control and management system for photovoltaic power generation based on the Internet of Things according to claim 3 is characterized in that, The edge data preprocessing module includes a machine learning-based intelligent data classification submodule and a sampling frequency dynamic adaptation submodule. The intelligent data classification submodule combines the dynamic characteristics of the photovoltaic cluster operation scenario and the contribution weight of the sensing data to the control decision. Through a preset multi-dimensional feature recognition model, the sensing data is divided into core decision data, key operation data and redundant auxiliary data. Among them, photovoltaic module fault characteristic data and grid frequency / voltage change data are classified as core decision data, photovoltaic array conventional operation parameter data are classified as key operation data, and repetitive redundant data of non-core environmental monitoring are classified as redundant auxiliary data. The sampling frequency dynamic adaptation submodule dynamically adjusts the sampling period of the corresponding sensing and acquisition unit based on the priority ranking results of each data level and the real-time load utilization of the link in the edge network layer. Core decision data maintains a preset maximum sampling frequency to ensure the timeliness of the adjustment response, key operating data is sampled using adaptive frequency conversion according to the link load utilization, and redundant auxiliary data uses a preset minimum sampling frequency to transmit data volume.

10. The intelligent control and management system for photovoltaic power generation based on the Internet of Things according to claim 5 is characterized in that, The delay monitoring and power matching coordination module also includes a photovoltaic node region clustering submodule and a control parameter feedback iteration submodule; The photovoltaic node regional clustering submodule divides the photovoltaic cluster into several independent collaborative control zones based on the geographical deployment characteristics of photovoltaic nodes, the network topology of IoT communication links, and historical transmission delay fluctuation characteristics. Photovoltaic nodes within the same collaborative control zone share a unified delay calibration benchmark and control timing benchmark. The regulation parameter feedback iteration submodule collects the power output matching error value of each collaborative regulation zone in real time, and corrects the time delay regulation threshold and the timing offset of the regulation command issuance in reverse according to the error amplitude, thus constructing a closed-loop collaborative regulation mechanism, including real-time time delay acquisition, dynamic adjustment of regulation timing, power output matching verification, and iterative optimization of regulation parameters.

Images