Intelligent monitoring system and method for realizing dynamic performance adjustment for distributed photovoltaic network

Abstract

The invention discloses an intelligent monitoring system and method for realizing dynamic performance adjustment for a distributed photovoltaic network, and relates to the technical field of intelligent power grids. The system comprises a data acquisition module used for acquiring image data of a photovoltaic panel and operation data of a photovoltaic power station in real time; the task scheduling module is used for coordinating the dynamic transmission of the image data and the operation data between the edge computing node and the cloud; the data processing module is used for carrying out preprocessing and feature extraction on the image data and the operation data; the dynamic estimation module is used for inputting the extracted features into a sparse regression model and predicting the dynamic state of the photovoltaic network by using an unscented Kalman filter; and the monitoring feedback module is used for remotely and dynamically adjusting the angle or position of the photovoltaic panel, starting and stopping equipment and / or adjusting parameters according to the prediction result of the dynamic state. The overall operation efficiency of the photovoltaic network can be remarkably improved, the operation and maintenance cost can be reduced, and the service life of equipment is prolonged.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to an intelligent monitoring system and method for implementing dynamic performance adjustment for a distributed photovoltaic network. Background Art

[0002] Distributed photovoltaic networks are an important part of modern renewable energy solutions. They mainly refer to photovoltaic power generation systems built near users, such as rooftops, communities, factories, etc. They use solar energy as energy and improve the directness and efficiency of energy use by reducing losses during power transmission.

[0003] At present, photovoltaic power generation networks have been widely used due to their clean and renewable characteristics. However, with the expansion of the scale of distributed photovoltaic power stations, the monitoring and maintenance of photovoltaic networks have become increasingly complex and arduous. Traditional photovoltaic monitoring systems mainly rely on a single central node for data collection and processing. This approach not only increases the burden on the central server, but also easily leads to delay problems. Especially in the scenario of large-scale networks, the efficiency of data processing is difficult to guarantee. At the same time, as the dynamic state of photovoltaic panels changes, traditional monitoring systems cannot respond quickly and adjust the operating state of the equipment, resulting in the dynamic performance of photovoltaic panels cannot be optimally adjusted, affecting power generation efficiency and equipment life. In addition, intelligent systems that lack real-time adjustment often require more manpower input and do not have the ability to automatically optimize. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes an intelligent monitoring system and method for realizing dynamic performance adjustment of distributed photovoltaic networks. By adjusting the hierarchical processing rules of data in real time, the dynamic transmission of image data and operation data between edge computing nodes and the cloud is optimized, which significantly improves data processing efficiency and network response speed.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows:

[0006] An intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network, the system comprising:

[0007] A data acquisition module is used to collect image data of photovoltaic panels in the target area in real time through high-resolution remote sensing technology, collect operation data of photovoltaic power stations in real time through distributed sensors, and transmit them to edge computing nodes or the cloud through wired or wireless means; the operation data includes voltage, current, power and temperature information;

[0008] The task scheduling module is used to coordinate the dynamic transmission of image data and operation data between edge computing nodes and the cloud;

[0009] A data processing module, used for preprocessing and feature extraction of the image data and operation data;

[0010] The dynamic estimation module is used to input the extracted features into the sparse regression model and use the unscented Kalman filter to predict the dynamic state of the photovoltaic network;

[0011] A monitoring feedback module is used to remotely and dynamically adjust the angle or position of the photovoltaic panel, start and stop the equipment and / or adjust the parameters according to the prediction results of the dynamic state, and feed back the adjustment results to the system to continuously optimize the state estimation model and adjustment strategy;

[0012] The report statistics module is used to generate daily reports and statistical data to help users understand the operation status and economic benefits of the photovoltaic network.

[0013] As a preferred solution of the present invention, the task scheduling module includes:

[0014] The intermediate data buffer is used to temporarily store the collected image data and operation data, waiting for the hierarchical processing decision of the edge-cloud collaborative optimization model;

[0015] A hierarchical processing unit is used to hierarchically process image data and operation data based on the edge-cloud collaborative optimization model to reduce the computing load of the central node;

[0016] The dynamic adjustment unit is used to adjust the data classification processing rules in real time according to the load conditions of the edge computing nodes and the differences in cloud network delays, prevent server overload and make full use of cloud computing resources for task processing.

[0017] As a preferred solution of the present invention, the edge-cloud collaborative optimization model is specifically:

[0018] Set the coordination factor D between edge computing nodes and cloud processing capabilities to globally balance the distribution of data processing tasks between edge computing nodes and the cloud;

[0019] in:

[0020]

[0021] In the formula, G represents the data processing efficiency of a single edge computing node; M represents the average delay of data exchange between multiple edge computing nodes; C is the response time of the cloud for processing data per unit time; N is the number of edge computing nodes in the photovoltaic network;

[0022] When the coordination factor D is greater than a preset first threshold, the image data or operation data is preferentially transmitted from the intermediate data buffer to the cloud for processing, otherwise it is transmitted to the edge computing node for subsequent processing.

[0023] As a preferred solution of the present invention, the hierarchical processing rules for adjusting data in real time according to the load of the edge computing node and the difference in cloud network delay are specifically:

[0024] Detect whether the difference ΔLoad between the current data processing load of the edge computing node and the preset safety threshold is greater than the preset difference δ;

[0025] Set a standard value η, and evaluate the urgency coefficient E of sending data to the cloud based on the performance parameters of the edge computing node and the current network delay conditions;

[0026] When ΔLoad>δ and E≥η, increase the storage time of local data and reduce the frequency of data transmission to edge computing nodes;

[0027] When ΔLoad≤δ or E<η, increase the frequency of data upload to the cloud.

[0028] As a preferred solution of the present invention, the data processing module includes:

[0029] An image preprocessing unit, used for preprocessing the image data, including cropping irrelevant areas, adjusting the resolution to unify the input image size, color normalization, image noise removal and image enhancement;

[0030] An operation data preprocessing unit, used for performing preprocessing operations on the operation data, including data cleaning, missing value processing and data standardization;

[0031] The feature extraction unit is used to extract high-dimensional features from the preprocessed image data through a deep learning model, and to extract time series features from the preprocessed running data through an LSTM algorithm.

[0032] As a preferred solution of the present invention, the image enhancement method comprises:

[0033] Perform the following steps independently for each color channel of the image:

[0034] The input image is subjected to Gaussian blur processing to simulate different blur conditions in the imaging process, and the standard deviation σ of the Gaussian blur is set to 2 to remove high-frequency details of a specific scale;

[0035] Perform discrete wavelet transform on the blurred image to decompose the image into coefficients of multiple scales and directions. The wavelet transform formula is:

[0036]

[0037] Where W(f)(x,s) is the wavelet transform result of the input image f at position x under scale s; f(u) represents the pixel value of the input image f at position u; ψ is the wavelet basis function; s is the scale parameter; u is the spatial position;

[0038] The wavelet coefficients of the image are randomly perturbed, and 20% of the coefficients are randomly set to 0. The original unperturbed wavelet coefficient vector z s Multiplying with the mask matrix M of the control coefficient perturbation is expressed as:

[0039] z′ s =z s M;

[0040] In the formula, z′ s is the wavelet coefficient vector after perturbation; the elements of M are 0 or 1, where 0 indicates that the coefficient is perturbed and 1 indicates that the coefficient remains unchanged;

[0041] The inverse discrete wavelet transform is performed on the perturbed wavelet coefficients to reconstruct the image. The reconstruction formula is:

[0042]

[0043] Where f1(x,y) is the value of the reconstructed image at the pixel point (x,y); W -1 is the inverse wavelet transform operator.

[0044] As a preferred solution of the present invention, the deep learning model includes:

[0045] The preprocessed image data is fed into the deep learning model and trained using a convolutional neural network. The convolution operation is defined as:

[0046]

[0047] In the formula, f2(x,y) is the output value of the image at pixel (x,y) after the convolution operation; f(xi,yj) is the value of the input image at pixel (xi,yj); k is the convolution kernel; k(i,j) represents the weight value of the convolution kernel at position (i,j);

[0048] For each scale s of the image, the wavelet scale attribution method is used to analyze the response of the model to the details of images of different scales, and the Sobol index Ss is used to calculate the degree of dependence of the model on scale details. The formula is:

[0049]

[0050] Where Var(f(x,y)) is the variance of the original image f(x,y), and Var(f′(x,y,s)) is the variance of the perturbed image f′(x,y,s) after removing the details of scale s;

[0051] The dependency of each scale s is summarized into a scale embedding vector z = (z1, z2, ..., z L )∈R L , where z L Indicates that the model is at scale s L the degree of dependence on

[0052] The weights of the model at different scales are adjusted according to the value in the scale embedding vector z. Specifically, the loss function of the model is modified during the training phase so that the model's dependence on features of different scales matches their importance. The modified loss function L is:

[0053]

[0054] In the formula, w s is the weight of scale s, L s is the loss value at scale s.

[0055] As a preferred solution of the present invention, the dynamic estimation module includes:

[0056] The sparse matrix construction unit is used to construct a sparse regression model and calculate the sparse coefficient matrix Π, which is expressed as:

[0057]

[0058] in, Represents the system state; Θ(X,U) is a feature function library based on the feature matrix X and the input matrix U. The feature matrix X includes high-dimensional features extracted by the deep learning model and time series features extracted by the LSTM algorithm;

[0059] A sparse processing unit, used to perform sparse processing on the sparse coefficient matrix Π according to a preset sparse threshold value using a sequence threshold least square method, and set non-critical coefficients to zero;

[0060] The state prediction unit is used to predict the system state using the unscented Kalman filter and predict the state of the system at time k+1 according to the state transfer equation. The formula is:

[0061]

[0062] In the formula, represents the estimated state of the system at time k; u kis the input control signal at time k; g represents the nonlinear state transfer function of the system; w k is the process noise;

[0063] The covariance update unit is used to update the covariance matrix according to the state prediction results, which is expressed as:

[0064] P k+1 =FP k F T +Q;

[0065] Where P k is the state covariance matrix of the current time k; P k+1 is the state covariance matrix at time k+1; F is the state transfer matrix; T represents the matrix transpose; Q is the process noise covariance matrix;

[0066] The measurement update unit is used to correct the predicted state according to the Kalman gain K. The calculation formula of the Kalman gain is:

[0067] K=P k+1 H T (HP k+1 H T +R) -1 ;

[0068] Where H is the measurement matrix; R is the measurement noise covariance matrix;

[0069] Then update the state estimate:

[0070]

[0071] In the formula, y k+1 is the actual measurement value at time k+1.

[0072] An intelligent monitoring method for implementing dynamic performance adjustment for a distributed photovoltaic network, the method comprising:

[0073] The image data of the photovoltaic panels in the target area are collected in real time by high-resolution remote sensing technology, the operation data of the photovoltaic power station are collected in real time by distributed sensors, and the image data and the operation data are temporarily stored in an intermediate data buffer;

[0074] Based on the edge-cloud collaborative optimization model, image data and operation data are processed in a hierarchical manner. The hierarchical processing rules of data are adjusted in real time according to the load of edge computing nodes and the difference in cloud network latency, and the dynamic transmission of image data and operation data between edge computing nodes and the cloud is coordinated.

[0075] Preprocessing the image data includes cropping irrelevant areas, adjusting the resolution to unify the input image size, color normalization, image noise removal, and image enhancement;

[0076] Performing preprocessing operations on the operating data, including data cleaning, missing value processing and data standardization;

[0077] Extract high-dimensional features from preprocessed image data through deep learning models, and extract time series features from preprocessed running data through LSTM algorithms;

[0078] The extracted high-dimensional features and time series features are input into the sparse regression model, and the unscented Kalman filter is used to predict the dynamic state of the photovoltaic network;

[0079] According to the prediction results of the dynamic state, the angle or position of the photovoltaic panel can be adjusted remotely and dynamically, the equipment can be started and stopped, and / or the parameters can be adjusted. The adjustment results are fed back to the system to continuously optimize the state estimation model and adjustment strategy.

[0080] The beneficial effects of the present invention are:

[0081] The edge-cloud collaborative optimization model is used to perform hierarchical processing of data, which effectively disperses the computing load of the central node and avoids the risk of server overload. At the same time, it makes full use of cloud computing resources to ensure the efficient operation of the system. The data transmission frequency is dynamically adjusted according to the load of the edge computing node and the difference in cloud network latency, making data transmission more flexible and reducing the possibility of network congestion. Using the image preprocessing module, the system can crop, enhance and denoise the photovoltaic panel image data, improve the accuracy of image recognition, and provide high-quality input for subsequent feature extraction. The dynamic state of the photovoltaic network is predicted in real time through deep learning models and unscented Kalman filters. The system can remotely and dynamically adjust parameters such as the angle of the photovoltaic panel and the start and stop of the equipment, thereby optimizing the efficiency of photovoltaic power generation. The monitoring feedback module automatically adjusts according to the prediction results, reducing the need for manual intervention, improving the intelligence level of the system, and can continuously optimize the state estimation model and adjustment strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0083] in:

[0084] Figure 1 It is a system modular structure diagram of the present invention;

[0085] Figure 2 4 is a flow chart of a method in an embodiment of the present invention. DETAILED DESCRIPTION

[0086] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0087] like Figure 1 As shown, it is an embodiment of the present invention, which provides an intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network, including:

[0088] (1) Data acquisition module

[0089] It is used to collect image data of photovoltaic panels in the target area in real time through high-resolution remote sensing technology, collect operation data of photovoltaic power stations in real time through distributed sensors, and transmit them to edge computing nodes or the cloud through wired or wireless means; the operation data includes voltage, current, power, temperature and other information;

[0090] By rationally distributing data to central servers and edge computing nodes, the overall efficiency of the system can be improved, ensuring real-time requirements without sacrificing the accuracy of complex computing tasks.

[0091] (2) Task Scheduling Module

[0092] Used to coordinate the dynamic transmission of image data and operation data between edge computing nodes and cloud servers, including an intermediate data buffer, a hierarchical processing unit, and a dynamic adjustment unit;

[0093] The intermediate data buffer is used to temporarily store the collected image data and operation data, waiting for the hierarchical processing decision of the edge-cloud collaborative optimization model;

[0094] Before the edge-cloud collaborative optimization model decides whether the data is processed by the cloud or the edge node, the collected data needs an intermediate storage or transmission waiting mechanism to ensure that the data will not be lost before the edge-cloud processing path is determined. This mechanism is completed by the intermediate data buffer or task scheduling system.

[0095] The hierarchical processing unit is used to hierarchically process image data and operation data based on the edge-cloud collaborative optimization model to reduce the computing load of the central node;

[0096] The edge-cloud collaborative optimization model is as follows:

[0097] Set the coordination factor D between edge computing nodes and cloud processing capabilities to globally balance the distribution of data processing tasks between edge computing nodes and the cloud;

[0098] in:

[0099]

[0100] In the formula, G represents the data processing efficiency of a single edge computing node, representing the effectiveness of each edge device in independently processing local or near-ground data. The value range is 0-10, the optimal value is 8, and the ideal value is the maximum, which reflects the highest processing level; M represents the average delay of data exchange between multiple edge computing nodes, ranging from 0-150ms, with an optimal value of 2ms, representing the communication quality of the network. The lower the delay, the more efficient the data transmission; C is the response time of the cloud to process data per unit time, that is, the time the central server takes to receive and analyze information bundles, ranging from 50-1000ms, with an optimal value of 65ms. The smaller the C value, the faster the cloud processing speed and the ability to quickly respond to large amounts of data; N is the number of edge computing nodes in the photovoltaic network. The larger the value of N, the stronger the edge computing capability of the system and the more computing tasks it can share;

[0101] When the score of coordination factor D increases, it means that the cloud can continue to provide strong support while bearing high data load, and reduces the possibility of end-to-end delay problems, and enhances the self-regulation function in dynamically changing situations and the tolerance and compatibility with a variety of different hardware interfaces or services. Therefore, when the coordination factor D is greater than the preset first threshold (determined according to different application scenarios or network status), the image data or operation data is transmitted from the intermediate data buffer to the cloud for processing, otherwise it is transmitted to the edge computing node for subsequent processing.

[0102] For example, imagine a photovoltaic power generation site with 300 edge devices, each of which has a fairly good performance, such as G of 7 points; in addition, assume that the communication speed between all machines in the area is very fast, taking only about 2 milliseconds (M); and the cloud computing service is also very agile and can respond within 65 milliseconds (C). Now substitute the actual numbers into the formula to calculate: D = (7 + 2 + 65) / 300 × 100 = 24.67%. Since this number is relatively low, it shows that even with a large number of sensor nodes and a large increase in data volume, it can still maintain good interactive response performance and direct visual presentation, which verifies that the edge-cloud collaborative optimization model of the present invention can successfully operate in real application fields and meet the preset standards.

[0103] The dynamic adjustment unit is used to adjust the data classification processing rules in real time according to the load of the edge computing node and the difference in cloud network delay, to prevent server overload and make full use of cloud computing resources for task processing, specifically:

[0104] Detect whether the difference ΔLoad between the current data processing load (such as CPU occupancy, bandwidth utilization or data processing task volume) of the edge computing node and the preset safety threshold is greater than the preset difference δ;

[0105] The key to this step is to monitor and quantify the computing power burden of edge computing nodes in real time and the actual distance of this burden from the safe operation boundary; for example, if a 10GB / s safety data processing speed is pre-defined for an industrial automation scenario and the current measured value is 9GB / s, δ can be pre-set to 0.7GB / s according to system requirements; if the actual load reaches 9.4GB / s, ΔLoad is 0.4GB / s. When δ is set to 0.7GB / s, ΔLoad < δ, which indicates that the equipment is operating well and does not exceed the expected acceptable load range;

[0106] The urgency coefficient E of sending data to the cloud is evaluated based on the performance parameters of the edge computing nodes and the current network delay conditions;

[0107] Based on the specific resource conditions of the edge computing node, such as the speed or storage capacity of the processor, the number and activity status of the connected IoT devices, and the communication delay to the cloud at this moment, comprehensively evaluate the demand level and importance of uploading data to the cloud. The evaluation criteria can consider the remaining processing capacity of the edge device and the real-time communication time from the device to the central management cloud platform to determine the urgency coefficient E;

[0108] For example, Where: P represents the remaining processing power of the edge device, in GHz; S represents the remaining storage capacity, in GB; O represents the communication delay between the device and the cloud, in ms; A represents the number of currently connected IoT devices and the proportion of active devices; α, β, γ, ξ are the weight coefficients of each factor, which can be adjusted dynamically according to the application scenario; the higher the E value, the more urgent it is to upload data to the cloud;

[0109] When the difference ΔLoad exceeds the preset difference (ΔLoad>δ) and the urgency coefficient E reaches the emergency threshold (E≥η, η is a standard value predetermined to divide high and low priorities), the data transmission frequency of the edge device is reduced and the storage time of local data is increased to reduce the amount of data transmitted to the cloud. For example, assuming that η is set at 0.6, data below this value can be processed later. When the system calculates that the urgency coefficient is 0.7 and the edge computing node is already very busy (assuming that ΔLoad=1GB / s>δ=0.7GB / s at this time), it will execute measures to slow down data reporting and temporarily store it.

[0110] But on the other hand, once the calculated data transmission intensity is insufficient, that is, when the gap ΔLoad is within the preset difference range (ΔLoad≤δ) or the urgency coefficient E is lower than the emergency threshold (E<η), it means that there is room to upload more data as soon as possible for use by higher-level algorithms. The frequency of data upload to the cloud can be increased to use the powerful computing power of the cloud for more advanced data analysis and strategy generation. As mentioned earlier, when the E calculation result shows that it is only 0.4 below the limit of 0.6, although there may be a little light burden, it will still tend to be sent to the remote platform immediately. This method not only takes advantage of the instant response on the spot but also balances the value of centralized deep data analysis. The purpose of maximizing the overall efficiency of the system is achieved by adjusting the computing load sharing scheme at the edge and cloud.

[0111] By using the coordination factor D, the distribution of data processing tasks can be intelligently adjusted according to factors such as the processing power of edge devices and the cloud, network latency, etc., thereby improving the response speed of the system. By detecting the edge device load and network latency, the data upload frequency and the distribution of processing tasks can be adjusted in real time to ensure the stability and adaptability of the system under different operating conditions.

[0112] (3) Data processing module

[0113] Used to preprocess and extract features from image data and operation data, including an image preprocessing unit, an operation data preprocessing unit and a feature extraction unit;

[0114] The image preprocessing unit is used to preprocess the image data, including cropping irrelevant areas, adjusting the resolution to unify the input image size, color normalization, image noise removal and image enhancement;

[0115] Specifically, the image enhancement method includes:

[0116] Perform the following steps independently for each color channel of the image:

[0117] The input image is Gaussian blurred to simulate different blur conditions in the imaging process. The standard deviation σ of the Gaussian blur is set to 2 to remove high-frequency details of a specific scale.

[0118] Perform discrete wavelet transform on the blurred image to decompose the image into coefficients of multiple scales and directions. The wavelet transform formula is:

[0119]

[0120] Where W(f)(x,s) is the wavelet transform result of the input image f at position x under scale s; f(u) represents the pixel value of the input image f at position u; ψ is the wavelet basis function; s is the scale parameter; u is the spatial position;

[0121] The wavelet coefficients of the image are randomly perturbed, and 20% of the coefficients are randomly set to 0. The original unperturbed wavelet coefficient vector z s Multiplying with the mask matrix M of the control coefficient perturbation is expressed as:

[0122] z′ s =z s M;

[0123] In the formula, z′ s is the wavelet coefficient vector after perturbation; the elements of M are 0 or 1, where 0 indicates that the coefficient is perturbed and 1 indicates that the coefficient remains unchanged;

[0124] The inverse discrete wavelet transform is performed on the perturbed wavelet coefficients to reconstruct the image. The reconstruction formula is:

[0125]

[0126] Where f1(x,y) is the value of the reconstructed image at the pixel point (x,y); W -1 is the inverse wavelet transform operator.

[0127] The operation data preprocessing unit is used to perform preprocessing operations on the operation data, including data cleaning, missing value processing and data standardization;

[0128] The feature extraction unit is used to extract high-dimensional features from the preprocessed image data through a deep learning model, and to extract time series features from the preprocessed running data through an LSTM algorithm;

[0129] Deep learning models include:

[0130] The preprocessed image data is fed into the deep learning model and trained using a convolutional neural network. The convolution operation is defined as:

[0131]

[0132] In the formula, f2(x,y) is the output value of the image at pixel (x,y) after the convolution operation; f(xi,yj) is the value of the input image at pixel (xi,yj); k is the convolution kernel; k(i,j) represents the weight value of the convolution kernel at position (i,j);

[0133] For each scale s of the image, the wavelet scale attribution method is used to analyze the response of the model to the details of images of different scales, and the Sobol index Ss is used to calculate the degree of dependence of the model on scale details. The formula is:

[0134]

[0135] Where Var(f(x,y)) is the variance of the original image f(x,y), and Var(f′(x,y,s)) is the variance of the perturbed image f′(x,y,s) after removing the details of scale s;

[0136] The dependency of each scale s is summarized into a scale embedding vector z = (z1, z2, ..., z L )∈R L , where z L Indicates that the model is at scale s L the degree of dependence on

[0137] The weights of the model at different scales are adjusted according to the value in the scale embedding vector z. Specifically, the loss function of the model is modified during the training phase so that the model's dependence on features of different scales matches their importance. The modified loss function L is:

[0138]

[0139] In the formula, w s is the weight of scale s, L s is the loss value at scale s.

[0140] (4) Dynamic Estimation Module

[0141] It is used to input the extracted features into the sparse regression model and use the unscented Kalman filter to predict the dynamic state of the photovoltaic network;

[0142] In one embodiment, the dynamic estimation module includes:

[0143] The sparse matrix construction unit is used to construct a sparse regression model and calculate the sparse coefficient matrix Π, which is expressed as:

[0144]

[0145] in, Represents the system state; Θ(X,U) is a feature function library based on the feature matrix X and the input matrix U. The feature matrix X includes high-dimensional features extracted by the deep learning model and time series features extracted by the LSTM algorithm;

[0146] A sparse processing unit, used to perform sparse processing on the sparse coefficient matrix Π according to a preset sparse threshold value using a sequence threshold least square method, and set non-critical coefficients to zero;

[0147] The state prediction unit is used to predict the system state using the unscented Kalman filter and predict the state of the system at time k+1 according to the state transfer equation. The formula is:

[0148]

[0149] In the formula, represents the estimated state of the system at time k; u k is the input control signal at time k; g represents the nonlinear state transfer function of the system; w k is the process noise;

[0150] The covariance update unit is used to update the covariance matrix according to the state prediction results, which is expressed as:

[0151] P k+1 =FP k F T +Q;

[0152] Where P k is the state covariance matrix of the current time k; P k+1 is the state covariance matrix at time k+1; F is the state transfer matrix; T represents the matrix transpose; Q is the process noise covariance matrix;

[0153] The measurement update unit is used to correct the predicted state according to the Kalman gain K. The calculation formula of the Kalman gain is:

[0154] K=P k+1 H T (HP k+1 H T +R) -1 ;

[0155] Where H is the measurement matrix; R is the measurement noise covariance matrix;

[0156] Then update the state estimate:

[0157]

[0158] In the formula, y k+1 is the actual measurement value at time k+1.

[0159] (5) Monitoring feedback module

[0160] It is used to remotely and dynamically adjust the angle or position of photovoltaic panels, start and stop equipment and / or adjust parameters according to the prediction results of dynamic states, and feed back the adjustment results to the system to continuously optimize the state estimation model and adjustment strategy.

[0161] (6) Report statistics module

[0162] Used to generate daily reports and statistical data to help users understand the operation status and economic benefits of the PV network.

[0163] like Figure 2 FIG. 1 is another embodiment of the present invention, which provides an intelligent monitoring method for implementing dynamic performance adjustment for a distributed photovoltaic network, comprising the following steps:

[0164] The image data of photovoltaic panels in the target area are collected in real time through high-resolution remote sensing technology, the operation data of photovoltaic power stations are collected in real time through distributed sensors, and the image data and operation data are temporarily stored in the intermediate data buffer;

[0165] Based on the edge-cloud collaborative optimization model, image data and operation data are processed in a hierarchical manner. The hierarchical processing rules of data are adjusted in real time according to the load of edge computing nodes and the difference in cloud network latency, and the dynamic transmission of image data and operation data between edge computing nodes and the cloud is coordinated.

[0166] Preprocess the image data, including cropping irrelevant areas, adjusting the resolution to unify the input image size, color normalization, image noise removal, and image enhancement;

[0167] Perform preprocessing operations on running data, including data cleaning, missing value processing and data standardization;

[0168] Extract high-dimensional features from preprocessed image data through deep learning models, and extract time series features from preprocessed running data through LSTM algorithms;

[0169] The extracted high-dimensional features and time series features are input into the sparse regression model, and the unscented Kalman filter is used to predict the dynamic state of the photovoltaic network;

[0170] Remotely and dynamically adjust the angle or position of the photovoltaic panel, start and stop the equipment and / or adjust the parameters according to the prediction results of the dynamic state, and feed the adjustment results back to the system to continuously optimize the state estimation model and adjustment strategy

[0171] To sum up, the present invention optimizes the dynamic transmission of image data and operation data between edge computing nodes and the cloud by adjusting the hierarchical processing rules of data in real time, significantly improving data processing efficiency and network response speed. It can not only significantly improve the overall operating efficiency of the photovoltaic network, but also reduce operation and maintenance costs and extend the service life of the equipment. It has broad application prospects.

[0172] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network, characterized in that: The system includes: a data acquisition module, which is used to collect image data of photovoltaic panels in a target area in real time through high-resolution remote sensing technology, collect operation data of photovoltaic power stations in real time through distributed sensors, and transmit them to edge computing nodes or the cloud through wired or wireless means; the operation data includes voltage, current, power and temperature information; The task scheduling module is used to coordinate the dynamic transmission of image data and operation data between edge computing nodes and the cloud; A data processing module, used for preprocessing and feature extraction of the image data and operation data; The dynamic estimation module is used to input the extracted features into the sparse regression model and use the unscented Kalman filter to predict the dynamic state of the photovoltaic network; A monitoring feedback module is used to remotely and dynamically adjust the angle or position of the photovoltaic panel, start and stop the equipment and / or adjust the parameters according to the prediction results of the dynamic state, and feed back the adjustment results to the system to continuously optimize the state estimation model and adjustment strategy; The report statistics module is used to generate daily reports and statistical data to help users understand the operation status and economic benefits of the photovoltaic network.

2. According to claim 1, an intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network is characterized in that: The task scheduling module includes: The intermediate data buffer is used to temporarily store the collected image data and operation data, waiting for the hierarchical processing decision of the edge-cloud collaborative optimization model; A hierarchical processing unit is used to hierarchically process image data and operation data based on the edge-cloud collaborative optimization model to reduce the computing load of the central node; The dynamic adjustment unit is used to adjust the data classification processing rules in real time according to the load conditions of the edge computing nodes and the differences in cloud network delays, prevent server overload and make full use of cloud computing resources for task processing.

3. The intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to claim 2, characterized in that: The edge-cloud collaborative optimization model is specifically as follows: Set the coordination factor D between edge computing nodes and cloud processing capabilities to globally balance the distribution of data processing tasks between edge computing nodes and the cloud; in: In the formula, G represents the data processing efficiency of a single edge computing node; M represents the average delay of data exchange between multiple edge computing nodes; C is the response time of the cloud for processing data per unit time; N is the number of edge computing nodes in the photovoltaic network; When the coordination factor D is greater than a preset first threshold, the image data or operation data is preferentially transmitted from the intermediate data buffer to the cloud for processing, otherwise it is transmitted to the edge computing node for subsequent processing.

4. The intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to claim 2, characterized in that: The hierarchical processing rules of data are adjusted in real time according to the load of edge computing nodes and the difference in cloud network delay, specifically: Detect whether the difference ΔLoad between the current data processing load of the edge computing node and the preset safety threshold is greater than the preset difference δ; Set a standard value η, and evaluate the urgency coefficient E of sending data to the cloud based on the performance parameters of the edge computing node and the current network delay conditions; When ΔLoad>δ and E≥η, increase the storage time of local data and reduce the frequency of data transmission to the edge computing node; when ΔLoad≤δ or E<η, increase the frequency of data upload to the cloud.

5. The intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to claim 1, characterized in that: The data processing module comprises: An image preprocessing unit, used for preprocessing the image data, including cropping irrelevant areas, adjusting the resolution to unify the input image size, color normalization, image noise removal and image enhancement; An operation data preprocessing unit, used for performing preprocessing operations on the operation data, including data cleaning, missing value processing and data standardization; The feature extraction unit is used to extract high-dimensional features from the preprocessed image data through a deep learning model, and to extract time series features from the preprocessed running data through an LSTM algorithm.

6. The intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to claim 5, characterized in that: The image enhancement method comprises: Perform the following steps independently for each color channel of the image: The input image is subjected to Gaussian blur processing to simulate different blur conditions in the imaging process, and the standard deviation σ of the Gaussian blur is set to 2 to remove high-frequency details of a specific scale; Perform discrete wavelet transform on the blurred image to decompose the image into coefficients of multiple scales and directions. The wavelet transform formula is: Where W(f)(x,s) is the wavelet transform result of the input image f at position x under scale s; f(u) represents the pixel value of the input image f at position u; ψ is the wavelet basis function; s is the scale parameter; u is the spatial position; The wavelet coefficients of the image are randomly perturbed, and 20% of the coefficients are randomly set to 0. The original unperturbed wavelet coefficient vector z s Multiplying with the mask matrix M of the control coefficient perturbation is expressed as: from' s =from s ·M; In the formula, z′ s is the wavelet coefficient vector after perturbation; the elements of M are 0 or 1, where 0 indicates that the coefficient is perturbed and 1 indicates that the coefficient remains unchanged; The inverse discrete wavelet transform is performed on the perturbed wavelet coefficients to reconstruct the image. The reconstruction formula is: Where f1(x, y) is the value of the reconstructed image at the pixel point (x, y); W -1 is the inverse wavelet transform operator.

7. The intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to claim 6, characterized in that: The deep learning model includes: The preprocessed image data is fed into the deep learning model and trained using a convolutional neural network. The convolution operation is defined as: In the formula, f2(x,y) is the output value of the image at pixel (x,y) after the convolution operation; f(xi,yj) is the value of the input image at pixel (xi,yj); k is the convolution kernel; k(i,j) represents the weight value of the convolution kernel at position (i,j); For each scale s of the image, the wavelet scale attribution method is used to analyze the response of the model to the details of images of different scales, and the Sobol index Ss is used to calculate the degree of dependence of the model on scale details. The formula is: Where Var(f(x,y)) is the variance of the original image f(x,y), and Var(f′(x,y,s)) is the variance of the perturbed image f′(x,y,s) after removing the details of scale s; The dependency of each scale s is summarized into a scale embedding vector z = (z1, z2, ..., z L )∈R L , where z L Indicates that the model is at scale s L the degree of dependence on The weights of the model at different scales are adjusted according to the value in the scale embedding vector z. Specifically, the loss function of the model is modified during the training phase so that the model's dependence on features of different scales matches their importance. The modified loss function L is: In the formula, w s is the weight of scale s, L s is the loss value at scale s.

8. The intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to claim 1, characterized in that: The dynamic estimation module comprises: The sparse matrix construction unit is used to construct a sparse regression model and calculate the sparse coefficient matrix Π, which is expressed as: in, Represents the system state; Θ(X,U) is a feature function library based on the feature matrix X and the input matrix U. The feature matrix X includes high-dimensional features extracted by the deep learning model and time series features extracted by the LSTM algorithm; A sparse processing unit, used to perform sparse processing on the sparse coefficient matrix Π according to a preset sparse threshold value using a sequence threshold least square method, and set non-critical coefficients to zero; The state prediction unit is used to predict the system state using the unscented Kalman filter and predict the state of the system at time k+1 according to the state transfer equation. The formula is: In the formula, represents the estimated state of the system at time k; u k is the input control signal at time k; g represents the nonlinear state transfer function of the system; w k is the process noise; The covariance update unit is used to update the covariance matrix according to the state prediction results, which is expressed as: P k+1 =FP k F T +Q; Where P k is the state covariance matrix of the current time k; P k+1 is the state covariance matrix at time k+1; F is the state transfer matrix; T represents the matrix transpose; Q is the process noise covariance matrix; The measurement update unit is used to correct the predicted state according to the Kalman gain K. The calculation formula of the Kalman gain is: K=P k+1 H T (HP k+1 H T +R) -1 ; Where H is the measurement matrix; R is the measurement noise covariance matrix; Then update the state estimate: In the formula, y k+1 is the actual measurement value at time k+1.

9. A monitoring method for an intelligent monitoring system for implementing dynamic performance adjustment for a distributed photovoltaic network according to any one of claims 1 to 8, characterized in that: The method comprises: The image data of the photovoltaic panels in the target area are collected in real time by high-resolution remote sensing technology, the operation data of the photovoltaic power station are collected in real time by distributed sensors, and the image data and the operation data are temporarily stored in an intermediate data buffer; Based on the edge-cloud collaborative optimization model, image data and operation data are processed in a hierarchical manner. The hierarchical processing rules of data are adjusted in real time according to the load of edge computing nodes and the difference in cloud network latency, and the dynamic transmission of image data and operation data between edge computing nodes and the cloud is coordinated. Preprocessing the image data includes cropping irrelevant areas, adjusting the resolution to unify the input image size, color normalization, image noise removal, and image enhancement; Performing preprocessing operations on the operating data, including data cleaning, missing value processing and data standardization; Extract high-dimensional features from preprocessed image data through deep learning models, and extract time series features from preprocessed running data through LSTM algorithms; The extracted high-dimensional features and time series features are input into the sparse regression model, and the unscented Kalman filter is used to predict the dynamic state of the photovoltaic network; According to the prediction results of the dynamic state, the angle or position of the photovoltaic panel can be adjusted remotely and dynamically, the equipment can be started and stopped, and / or the parameters can be adjusted. The adjustment results are fed back to the system to continuously optimize the state estimation model and adjustment strategy.

Images