Remote monitoring system of distributed photovoltaic power station

Abstract

The embodiment of the invention provides a remote monitoring system of a distributed photovoltaic power station. The method comprises the following steps: firstly, acquiring output voltage values and output current values of a to-be-analyzed photovoltaic cell assembly in a distributed photovoltaic power station at a plurality of preset time points in a preset time period, then, transmitting the output voltage values and the output current values at the plurality of preset time points to a background monitoring server through a wireless communication module, and then, transmitting the output voltage values and the output current values to the background monitoring server through a wireless communication module. And the background monitoring server analyzes correlation characteristics between the output voltage values and the output current values at the plurality of preset time points, and finally, based on the correlation characteristics, whether a fault early warning prompt is generated is determined. Therefore, the power generation efficiency and the economic benefit of the whole distributed photovoltaic power station can be improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of intelligent monitoring, and more specifically, to a remote monitoring system for a distributed photovoltaic power station. Background Art

[0002] Distributed photovoltaic power station (distributed photovoltaic power generation system) refers to a photovoltaic power generation system installed on the user side or the distribution network side, which can directly supply electricity to the user, or transmit electricity to the distribution network, or achieve a balance between the two. The basic equipment of a distributed photovoltaic power station includes photovoltaic cell components, photovoltaic array brackets, DC combiner boxes, DC distribution cabinets, grid-connected inverters, AC distribution cabinets, etc.

[0003] Under conditions of solar radiation, the solar cell module array of the photovoltaic power station converts the output electricity from solar energy and sends it to the DC distribution cabinet through the DC combiner box. The grid-connected inverter then converts it into AC power to supply the building's own load. Excess or insufficient power is regulated by connecting to the grid.

[0004] Among them, photovoltaic cell components are the core components of photovoltaic power stations, and their performance and life directly affect the power generation efficiency and economic benefits of the system. Therefore, monitoring photovoltaic cell components in distributed photovoltaic power stations is of great significance. Summary of the invention

[0005] In order to solve the above technical problems, the present invention is proposed. The embodiment of the present invention provides a remote monitoring system for a distributed photovoltaic power station, which can use a deep learning algorithm to remotely analyze the output voltage data and output current data of a photovoltaic cell assembly, and thereby achieve real-time detection and judgment of the operating status of the photovoltaic cell assembly, so as to timely issue a fault warning prompt when the operating status is not good, reminding relevant technical personnel to take remedial measures and maintenance work, thereby improving the power generation efficiency and economic benefits of the entire distributed photovoltaic power station.

[0006] An embodiment of the present invention provides a remote monitoring system for a distributed photovoltaic power station, which includes:

[0007] A data acquisition module is used to obtain output voltage values ​​and output current values ​​of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period;

[0008] A wireless transmission module, used for transmitting the output voltage values ​​and output current values ​​at the plurality of predetermined time points to a background monitoring server via a wireless communication module;

[0009] a correlation feature analysis module, used for analyzing, in the background monitoring server, correlation features between the output voltage values ​​and the output current values ​​at the plurality of predetermined time points; and

[0010] The fault warning judgment module is used to determine whether to generate a fault warning prompt based on the associated features.

[0011] In some possible embodiments, the correlation feature analysis module includes:

[0012] A data preprocessing unit, configured to perform data preprocessing on the output voltage values ​​and the output current values ​​at the plurality of predetermined time points to obtain an output voltage timing input vector and an output current timing input vector;

[0013] An association coding unit, used for calculating a voltage-current output timing association coding matrix between the output voltage timing input vector and the output current timing input vector;

[0014] A feature extraction unit, configured to extract features from the voltage-current output timing correlation coding matrix using a deep learning network model to obtain a voltage-current timing correlation feature vector; and

[0015] The correlation feature acquisition unit is used to use the voltage-current time series correlation feature vector as the correlation feature.

[0016] In some possible embodiments, the data preprocessing unit is used to:

[0017] The output voltage values ​​and the output current values ​​at the plurality of predetermined time points are arranged according to the time dimension into the output voltage timing input vector and the output current timing input vector.

[0018] In some possible embodiments, the deep learning network model is an electrical property-related feature extractor comprising a first convolutional layer and a second convolutional layer;

[0019] Wherein, the feature extraction unit is used to:

[0020] The voltage-current output timing correlation coding matrix is ​​passed through the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer to obtain the voltage-current timing correlation feature vector.

[0021] In some possible embodiments, the first convolution layer and the second convolution layer respectively use two-dimensional convolution kernels with different scales.

[0022] In some possible embodiments, the feature extraction unit is further used to:

[0023] Use each layer of the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer to perform two-dimensional convolution processing, global pooling processing and non-linear activation processing on the input data in the forward pass of the layer to generate the voltage-current timing correlation feature vector from the last layer of the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer, wherein the input of the first layer of the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer is the voltage-current output timing correlation coding matrix.

[0024] In some possible embodiments, the fault warning judgment module includes:

[0025] a characteristic distribution correction unit, configured to perform characteristic distribution correction on the voltage-current time series correlation characteristic vector to obtain a corrected voltage-current time series correlation characteristic vector; and

[0026] The multi-task classification unit is used to pass the corrected voltage-current time series correlation feature vector through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated.

[0027] In some possible embodiments, the multi-task classification unit includes:

[0028] A fine-grained classification subunit, used for passing the corrected voltage-current time series correlation feature vector through a fine-grained classifier to obtain multiple fault category probability values;

[0029] A coarse-grained classification subunit, used for passing the corrected voltage-current time series correlation feature vector through a coarse-grained classifier to obtain a first probability value and a second probability value;

[0030] a probability weighted fusion subunit, configured to fuse the plurality of fault category probability values, the first probability value, and the second probability value in a probability weighted fusion manner to obtain a comprehensive expression probability value; and

[0031] The classification result acquisition subunit is used to obtain the classification result based on the comprehensive expression probability value.

[0032] Compared with the prior art, the remote monitoring system of the distributed photovoltaic power station provided by the present invention first obtains the output voltage value and output current value of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period, then transmits the output voltage value and output current value of the multiple predetermined time points to the background monitoring server through a wireless communication module, and then, in the background monitoring server, analyzes the correlation characteristics between the output voltage value and the output current value of the multiple predetermined time points, and finally, based on the correlation characteristics, determines whether to generate a fault warning prompt. In this way, the power generation efficiency and economic benefits of the entire distributed photovoltaic power station can be improved. . BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0034] Figure 1 FIG. is a block diagram of a remote monitoring system for a distributed photovoltaic power station according to an embodiment of the present invention;

[0035] Figure 2 FIG. is a block diagram of the associated feature analysis module in the remote monitoring system of the distributed photovoltaic power station according to an embodiment of the present invention;

[0036] Figure 3 FIG. is a block diagram of the fault warning and judgment module in the remote monitoring system of the distributed photovoltaic power station according to an embodiment of the present invention;

[0037] Figure 4 FIG. is a block diagram of the multi-task classification unit in the remote monitoring system of the distributed photovoltaic power station according to an embodiment of the present invention;

[0038] Figure 5 FIG. is a flowchart of a remote monitoring method for a distributed photovoltaic power station according to an embodiment of the present invention;

[0039] Figure 6 FIG. is a schematic diagram of the system architecture of a remote monitoring method for a distributed photovoltaic power station according to an embodiment of the present invention;

[0040] Figure 7 FIG. is an application scenario diagram of the remote monitoring system of the distributed photovoltaic power station according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0042] Unless otherwise specified, the technical terms or scientific terms used in the embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. "Including" or "comprising" used in the embodiments of the present invention neither limit the shapes, numbers, steps, actions, operations, components, originals and / or their groups mentioned, nor exclude the appearance or addition of one or more other different shapes, numbers, steps, actions, operations, components, originals and / or their groups, or the addition of these. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number and order of the indicated technical features. Thus, the features defined as "first" and "second" can explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0043] Unless otherwise specifically stated, the relative arrangement of the components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention. At the same time, it should be understood that for ease of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship, and the techniques, methods and devices known to ordinary technicians in the relevant fields may not be discussed in detail, but where appropriate, the techniques, methods and devices shown should be considered as part of the authorized specification. In all examples shown and discussed here, any specific other examples may have different values. It should be noted that similar symbols and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0044] In the description of the embodiments of the present invention, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in the embodiments of the present invention and the features of the different embodiments or examples, without contradiction.

[0045] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0046] It should be understood that the technical problem that the present invention is expected to solve is: how to monitor and analyze in real time the operating status of photovoltaic cell assemblies in distributed photovoltaic power stations. In this regard, the technical concept of the present invention is: using a deep learning algorithm to remotely analyze the output voltage data and output current data of photovoltaic cell assemblies, and thereby achieve real-time detection and judgment of the operating status of photovoltaic cell assemblies, so as to issue a fault warning prompt in time when the operating status is not good, to remind relevant technicians to take remedial measures and maintenance work, and improve the power generation efficiency and economic benefits of the entire distributed photovoltaic power station.

[0047] Based on this, Figure 1 FIG. 1 is a block diagram of a remote monitoring system for a distributed photovoltaic power station according to an embodiment of the present invention. Figure 1 As shown, the remote monitoring system 100 of a distributed photovoltaic power station according to an embodiment of the present invention includes: a data acquisition module 110, which is used to obtain the output voltage value and the output current value of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period; a wireless transmission module 120, which is used to transmit the output voltage value and the output current value at the multiple predetermined time points to the background monitoring server through the wireless communication module; a correlation feature analysis module 130, which is used to analyze the correlation features between the output voltage values ​​and the output current values ​​at the multiple predetermined time points in the background monitoring server; and a fault warning judgment module 140, which is used to determine whether to generate a fault warning prompt based on the correlation features.

[0048] Specifically, in the technical solution of the present invention, the output voltage value and output current value of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period are first obtained; and the output voltage value and output current value of the multiple predetermined time points are transmitted to the background monitoring server through the wireless communication module.

[0049] Next, in the background monitoring server, data preprocessing is performed on the output voltage values ​​and output current values ​​at the plurality of predetermined time points to obtain an output voltage timing input vector and an output current timing input vector. That is, the timing discrete distribution of the output voltage value and the output current value is converted into a structured vector representation to facilitate the subsequent reading and recognition of the model.

[0050] In a specific example of the present invention, data preprocessing of the output voltage values ​​and output current values ​​at the multiple predetermined time points to obtain output voltage timing input vectors and output current timing input vectors is implemented by arranging the output voltage values ​​and output current values ​​at the multiple predetermined time points into output voltage timing input vectors and output current timing input vectors according to the time dimension.

[0051] Then, a voltage-current output timing correlation coding matrix between the output voltage timing input vector and the output current timing input vector is calculated; and the voltage-current output timing correlation coding matrix is ​​passed through an electrical performance correlation feature extractor including a first convolution layer and a second convolution layer to obtain a voltage-current timing correlation feature vector, wherein the first convolution layer and the second convolution layer respectively use two-dimensional convolution kernels with different scales. That is, two-dimensional convolution kernels of different scales are used to capture the correlation feature distribution in different spatial neighborhoods in the voltage-current output timing correlation coding matrix.

[0052] Accordingly, if Figure 2 As shown, the correlation feature analysis module 130 includes: a data preprocessing unit 131, which is used to perform data preprocessing on the output voltage values ​​and the output current values ​​at the multiple predetermined time points to obtain the output voltage timing input vector and the output current timing input vector; an association coding unit 132, which is used to calculate the voltage-current output timing correlation coding matrix between the output voltage timing input vector and the output current timing input vector; a feature extraction unit 133, which is used to perform feature extraction on the voltage-current output timing correlation coding matrix using a deep learning network model to obtain a voltage-current timing correlation feature vector; and an association feature acquisition unit 134, which is used to use the voltage-current timing correlation feature vector as the association feature.

[0053] It should be understood that the correlation feature analysis module 130 includes four units: a data preprocessing unit 131, an association coding unit 132, a feature extraction unit 133, and an association feature acquisition unit 134. In one example, in the data preprocessing unit 131, data preprocessing may include operations such as data cleaning, normalization, and filtering to ensure the accuracy and consistency of the input data. The voltage-current output timing correlation coding matrix between the output voltage timing input vector and the output current timing input vector calculated by the association coding unit 132 represents the timing correlation between the voltage and the current, which can be used for subsequent feature extraction and analysis. The feature extraction unit 133 uses a deep learning network model to extract features from the voltage-current output timing correlation coding matrix to obtain a voltage-current timing correlation feature vector. The deep learning network model can extract useful features by learning patterns and correlations in the data, and these features can be used for subsequent association analysis and prediction tasks. The correlation feature acquisition unit 134 outputs the voltage-current time series correlation feature vector as the final correlation feature, which can be used for further data analysis, model building, fault detection and other tasks to extract information about the correlation between voltage and current. In general, the role of the correlation feature analysis module is to analyze and extract features from the correlation between output voltage and output current for subsequent applications and tasks. Each unit plays a different role in this process, responsible for specific data processing, correlation coding, feature extraction and feature output tasks.

[0054] Among them, the data preprocessing unit 131 is used to: arrange the output voltage values ​​and output current values ​​at the multiple predetermined time points according to the time dimension into the output voltage time series input vector and the output current time series input vector. It should be understood that the function of the data preprocessing unit 131 is to arrange the output voltage values ​​and output current values ​​at the multiple predetermined time points according to the time dimension into the output voltage time series input vector and the output current time series input vector. This arrangement can keep the data in order in time and can reflect the changes of voltage and current over time. Such an arrangement has the following uses: 1. Time series analysis: By arranging the output voltage and output current according to the time dimension, the change trend of voltage and current over time can be better observed and analyzed. This is very important for understanding the dynamic characteristics of the power system, detecting anomalies and faults, etc. 2. Feature extraction: After arranging into a time series input vector, various statistical features, frequency domain features or time domain features can be extracted using the time series analysis method. These features can be used for subsequent model training, fault diagnosis, state estimation and other tasks. 3. Model input: The time series input vector arranged according to the time dimension can be used as the input of the deep learning model. Deep learning models can learn the complex relationship between voltage and current and perform tasks such as fault detection and load forecasting. 4. Data visualization: The arrangement of the time series input vector can easily visualize the data as a time series graph, so that the changes in voltage and current can be more intuitively observed and anomalies or trends can be found. In general, arranging the output voltage value and the output current value as a time series input vector according to the time dimension is helpful for tasks such as time series analysis, feature extraction, model input and data visualization.

[0055] The deep learning network model is an electrical performance correlation feature extractor including a first convolution layer and a second convolution layer; wherein the feature extraction unit 133 is used to: pass the voltage-current output timing correlation coding matrix through the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer to obtain the voltage-current timing correlation feature vector. Furthermore, the first convolution layer and the second convolution layer respectively use two-dimensional convolution kernels with different scales.

[0056] It is worth mentioning that the two-dimensional convolution kernel is a filter commonly used in deep learning. It is a two-dimensional matrix used to perform convolution operations on input data. In the electrical performance correlation feature extractor, the first convolution layer and the second convolution layer use two-dimensional convolution kernels with different scales. Two-dimensional convolution kernels with different scales can capture the features of different spatial scales of the input data. Smaller convolution kernels can capture details and local features, while larger convolution kernels can capture broader contextual information and global features. By using two-dimensional convolution kernels with different scales, multi-scale features can be extracted, thereby more comprehensively representing the features of the input data.

[0057] In the electrical performance correlation feature extractor, the first convolution layer and the second convolution layer use two-dimensional convolution kernels with different scales, which can capture different features in the voltage-current output timing correlation encoding matrix through multi-level feature extraction. The first convolution layer can extract lower-level features, such as local patterns and edge features, while the second convolution layer can further extract higher-level features, such as more complex patterns and combination features. Through such hierarchical feature extraction, a richer and more expressive voltage-current timing correlation feature vector can be obtained. In summary, the role of two-dimensional convolution kernels with different scales in the electrical performance correlation feature extractor is to capture features of different scales through multi-level feature extraction to obtain a more comprehensive and rich voltage-current timing correlation feature vector.

[0058] More specifically, the feature extraction unit 133 is further used to: use each layer of the electrical performance association feature extractor including the first convolution layer and the second convolution layer to perform two-dimensional convolution processing, global pooling processing and non-linear activation processing on the input data in the forward pass of the layer to generate the voltage-current timing correlation feature vector from the last layer of the electrical performance association feature extractor including the first convolution layer and the second convolution layer, wherein the input of the first layer of the electrical performance association feature extractor including the first convolution layer and the second convolution layer is the voltage-current output timing correlation coding matrix.

[0059] Then, the voltage-current time series correlation feature vector is passed through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated.

[0060] Accordingly, if Figure 3As shown, the fault warning judgment module 140 includes: a feature distribution correction unit 141, which is used to perform feature distribution correction on the voltage-current time series correlation feature vector to obtain a corrected voltage-current time series correlation feature vector; and a multi-task classification unit 142, which is used to pass the corrected voltage-current time series correlation feature vector through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated. It should be understood that the function of the feature distribution correction unit 141 is to perform feature distribution correction on the voltage-current time series correlation feature vector to obtain a corrected voltage-current time series correlation feature vector. This unit can adjust the distribution of the feature vector through various statistical methods, normalization methods or other data processing techniques to better adapt to subsequent classification tasks. Feature distribution correction can help improve the reliability, stability and robustness of the feature. The function of the multi-task classification unit 142 is to pass the corrected voltage-current time series correlation feature vector through a multi-task classification head module to obtain a classification result. This unit uses the multi-task learning ability of the deep learning model to classify multiple related tasks at the same time. In the fault warning judgment module, the multi-task classification unit can classify and judge different types of faults according to the input voltage-current time series correlation feature vector. The classification result indicates whether to generate a fault warning prompt, that is, to judge whether there is a possibility of fault in the current power system. In summary, the feature distribution correction unit and the multi-task classification unit are used to correct the feature distribution and perform multi-task classification in the fault warning judgment module, respectively. Feature distribution correction can improve the reliability and robustness of the feature, while multi-task classification can judge whether there is a fault in the power system and generate corresponding warning prompts. The combination of these two units can improve the accuracy and reliability of the fault warning system.

[0061] In a specific example of the present invention, the voltage-current timing association feature vector is passed through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether to generate a fault warning prompt encoding process, including: first, the voltage-current timing association feature vector is passed through a fine-grained classifier to obtain multiple fault category probability values; at the same time, the voltage-current timing association feature vector is passed through a coarse-grained classifier to obtain a first probability value and a second probability value; then, the multiple fault category probability values, the first probability value and the second probability value are fused in a probability weighted fusion manner to obtain a comprehensive expression probability value; and then the classification result is obtained based on the comprehensive expression probability value.

[0062] Accordingly, if Figure 4As shown, the multi-task classification unit 142 includes: a fine-grained classification subunit 1421, used to pass the corrected voltage-current timing association feature vector through a fine-grained classifier to obtain multiple fault category probability values; a coarse-grained classification subunit 1422, used to pass the corrected voltage-current timing association feature vector through a coarse-grained classifier to obtain a first probability value and a second probability value; a probability weighted fusion subunit 1423, used to fuse the multiple fault category probability values, the first probability value and the second probability value in a probability weighted fusion manner to obtain a comprehensive expression probability value; and a classification result acquisition subunit 1424, used to obtain the classification result based on the comprehensive expression probability value.

[0063] It should be understood that the fine-grained classification subunit 1421 passes the corrected voltage-current time series correlation feature vector through a fine-grained classifier to obtain probability values ​​of multiple fault categories. This subunit is used to classify the multiple fault types that may exist in the power system and output the probability of each fault category. The coarse-grained classification subunit 1422 passes the corrected voltage-current time series correlation feature vector through a coarse-grained classifier to obtain a first probability value and a second probability value. This subunit is used to perform a rougher classification, usually used to distinguish between normal and fault states, or to perform a preliminary and rough classification of faults. The probability weighted fusion subunit 1423 fuses multiple fault category probability values, the first probability value and the second probability value in a probability weighted fusion manner to obtain a comprehensive expression probability value. This subunit is used to comprehensively consider the output probabilities of each classifier and perform weighted fusion according to their importance to obtain a more comprehensive and comprehensive probability expression. The classification result acquisition subunit 1424 obtains the final classification result based on the comprehensive expression probability value. This subunit can convert the comprehensive expression probability value into the final classification judgment result according to the set threshold or other decision rules, indicating whether to generate a fault warning prompt. In summary, the subunits in the multi-task classification unit 142 are used for fine-grained classification, coarse-grained classification, probability weighted fusion and final classification result acquisition. Through the collaborative work of these subunits, the classification judgment of different fault types in the power system can be realized, and the corresponding fault warning prompts can be generated.

[0064] Here, the fault categories of the fine-grained classifier are hot spot effect, crack, desoldering, debonding and PID effect, and the probability values ​​of multiple fault categories respectively represent the probability values ​​of the occurrence of the above faults; the first probability value of the coarse-grained classifier represents the probability of the occurrence of a fault, and the second probability value represents the probability of no fault; obtaining the classification result based on the comprehensive expression probability value refers to determining the classification result based on the comprehensive expression probability value and the threshold obtained by training.

[0065] Among them, the reason for setting the label category of the fine-grained classifier to the above-mentioned fault categories is that when actually judging and classifying the faults of photovoltaic cell modules, the faults of photovoltaic cell modules are usually divided into the above-mentioned categories according to different causes and manifestations, namely hot spot effect, cracks, desoldering, debonding and PID effect. Specifically, the hot spot effect refers to the fact that one or some cells in the photovoltaic cell module have excessive local impedance, which causes the temperature to rise and form hot spots. The hot spot effect will reduce the output power of the cell and even cause an open circuit or short circuit, resulting in serious performance loss and safety hazards; cracks refer to the fact that one or some cells in the photovoltaic cell module have cracks on the surface or inside due to mechanical stress, temperature changes or other external forces. The cracks will affect the photoelectric conversion efficiency of the cell, increase the series resistance and reduce the output power; desoldering refers to the disconnection of the solder joints between one or some cells in the photovoltaic cell module and the backplane or connecting belt, resulting in Circuit discontinuity and desoldering will lead to a decrease in output power or even an open circuit or short circuit; debonding refers to the failure of the adhesive between the glass, EVA, backplane and other packaging materials in the photovoltaic cell module, resulting in separation between the layers. Debonding will affect the mechanical strength and waterproof performance of the photovoltaic cell module, and increase the risk of water seepage and oxidation; the PID effect refers to the reverse leakage current between the positive and negative electrodes of the photovoltaic cell module under high temperature, high humidity, high pressure and other conditions due to the polarization effect, so that a negatively charged barrier layer is formed on the surface of the cell. The PID effect will reduce the output power and stability of the photovoltaic cell module, causing permanent performance degradation.

[0066] That is, the multi-task classification head module can use a fine-grained classifier to learn the operating state feature expression of the photovoltaic cell assembly under abnormal conditions such as hot spot effect, crack, desoldering, debonding and PID effect. In this way, the model can learn the difference between the feature information of the photovoltaic cell assembly in normal operation and the photovoltaic cell assembly with abnormal or faulty conditions, and clarify the range boundary of the abnormal or faulty conditions and the range boundary of the abnormal or faulty conditions. In actual application scenarios, it is only necessary to clarify whether the photovoltaic cell assembly has an abnormal or faulty condition. Therefore, in the technical solution of the present invention, a coarse-grained classifier is used to learn the general feature distribution of abnormal or faulty conditions, so as to judge whether the operating state of the photovoltaic cell assembly has an abnormal or faulty condition, and generate a fault warning prompt in time. In this way, the capabilities of the fine-grained classifier and the coarse-grained classifier are complemented, thereby increasing the generalization ability of the network model. Even if a fault category that has not been specifically classified appears, it can still be effectively identified as long as it is not within the boundary of the feature distribution of the photovoltaic cell assembly under normal operation.

[0067] Here, the voltage-current output time series correlation coding matrix is ​​passed through an electrical performance correlation feature extractor including a first convolution layer and a second convolution layer, and the eigenvalues ​​of each position of the voltage-current time series correlation feature vector are used to express the local correlation characteristics of the time series-sample cross dimension of the voltage-current at different scales, and the eigenvalues ​​follow the channel distribution of the electrical performance correlation feature extractor. However, considering that the scale difference of the two-dimensional convolution kernel with different scales will bring significant distribution differences to the feature representation of each eigenvalue, the voltage-current time series correlation feature vector will still have a sparse local feature distribution sparseness, that is, a sparse sub-manifold outside the distribution relative to the overall high-dimensional feature manifold, so that when the voltage-current time series correlation feature vector is subjected to class probability regression mapping through a multi-task classification head module, the convergence of the voltage-current time series correlation feature vector to the predetermined class probability category representation in the probability space is poor, affecting the accuracy of the classification result. Therefore, preferably, when the voltage-current time series correlation feature vector is classified through a classifier, the voltage-current time series correlation feature vector is optimized for position-by-position eigenvalues.

[0068] Accordingly, in one example, the characteristic distribution correction unit 141 is further used to: perform characteristic distribution correction on the voltage-current time series correlation characteristic vector using the following optimization formula to obtain the corrected voltage-current time series correlation characteristic vector; wherein the optimization formula is:

[0069]

[0070] Where V is the voltage-current time series correlation characteristic vector, v i is the eigenvalue of the ith position of the voltage-current time series associated eigenvector V, exp(·) represents the exponential operation of a value, and the exponential operation of the value represents the calculation of the natural exponential function value with the value as the power, v′ i is the eigenvalue of the i-th position of the corrected voltage-current time series correlation eigenvector.

[0071] That is, the sparse distribution in the high-dimensional feature space is processed by regularization based on heavy probability to activate the natural distribution transfer of the geometric manifold of the voltage-current timing association feature vector V in the high-dimensional feature space to the probability space, thereby improving the class convergence of complex high-dimensional feature manifolds with high spatial sparsity under predetermined class probabilities by performing heavy-probability-based smoothing regularization on the distributed sparse submanifolds of the high-dimensional feature manifold of the voltage-current timing association feature vector V, thereby improving the accuracy of the classification results obtained by the multi-task classification head module of the voltage-current timing association feature vector V.

[0072] In summary, the remote monitoring system 100 of the distributed photovoltaic power station based on the embodiment of the present invention is explained, which can realize real-time detection and judgment of the operating status of the photovoltaic cell assembly, so as to promptly issue a fault warning prompt when its operating status is not good, reminding relevant technical personnel to take remedial measures and maintenance work, thereby improving the power generation efficiency and economic benefits of the entire distributed photovoltaic power station.

[0073] As described above, the remote monitoring system 100 of the distributed photovoltaic power station based on the embodiment of the present invention can be implemented in various terminal devices, such as a server having a remote monitoring algorithm of the distributed photovoltaic power station based on the embodiment of the present invention. In one example, the remote monitoring system 100 of the distributed photovoltaic power station based on the embodiment of the present invention can be integrated into the terminal device as a software module and / or a hardware module. For example, the remote monitoring system 100 of the distributed photovoltaic power station based on the embodiment of the present invention can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the remote monitoring system 100 of the distributed photovoltaic power station based on the embodiment of the present invention can also be one of the many hardware modules of the terminal device.

[0074] Alternatively, in another example, the remote monitoring system 100 of the distributed photovoltaic power station based on the embodiment of the present invention and the terminal device may also be separate devices, and the remote monitoring system 100 of the distributed photovoltaic power station may be connected to the terminal device via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0075] Figure 5 The figure is a flow chart of a remote monitoring method for a distributed photovoltaic power station according to an embodiment of the present invention. Figure 6 FIG. 1 is a schematic diagram of a system architecture of a remote monitoring method for a distributed photovoltaic power station according to an embodiment of the present invention. Figure 5 and 6 As shown, the remote monitoring method of a distributed photovoltaic power station according to an embodiment of the present invention includes: S110, obtaining the output voltage value and the output current value of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period; S120, transmitting the output voltage value and the output current value at the multiple predetermined time points to a background monitoring server through a wireless communication module; S130, at the background monitoring server, analyzing the correlation characteristics between the output voltage values ​​and the output current values ​​at the multiple predetermined time points; and, S140, based on the correlation characteristics, determining whether to generate a fault warning prompt.

[0076] In a specific example, in the remote monitoring method of the above-mentioned distributed photovoltaic power station, the background monitoring server analyzes the correlation characteristics between the output voltage values ​​and the output current values ​​at the multiple predetermined time points, including: performing data preprocessing on the output voltage values ​​and the output current values ​​at the multiple predetermined time points to obtain an output voltage timing input vector and an output current timing input vector; calculating a voltage-current output timing correlation coding matrix between the output voltage timing input vector and the output current timing input vector; using a deep learning network model to perform feature extraction on the voltage-current output timing correlation coding matrix to obtain a voltage-current timing correlation feature vector; and using the voltage-current timing correlation feature vector as the correlation feature.

[0077] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned remote monitoring method of distributed photovoltaic power station have been described in the above reference. Figures 1 to 4 The remote monitoring system 100 for a distributed photovoltaic power station has been described in detail, and therefore, its repeated description will be omitted.

[0078] Figure 7 FIG. 1 is an application scenario diagram of a remote monitoring system for a distributed photovoltaic power station according to an embodiment of the present invention. Figure 7 As shown, in this application scenario, first, the output voltage values ​​of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period are obtained (for example, Figure 7 D1) and output current value (e.g., Figure 7 Then, the output voltage values ​​and output current values ​​at the plurality of predetermined time points are input to a server (for example, Figure 7 In S) shown in , the server can use the remote monitoring algorithm of the distributed photovoltaic power station to process the output voltage values ​​and output current values ​​of the multiple predetermined time points to obtain a classification result for indicating whether a fault warning prompt is generated.

[0079] It is worth mentioning that distributed photovoltaic power stations refer to a power generation mode in which photovoltaic power generation systems are distributed in multiple locations, which is opposite to traditional centralized photovoltaic power stations. Traditional centralized photovoltaic power stations usually build photovoltaic power generation facilities in a large area, while distributed photovoltaic power stations install photovoltaic power generation systems in various locations, which can be roofs, building surfaces, parking lots, industrial areas, etc. Distributed photovoltaic power stations are characterized by flexibility and scalability. Since distributed photovoltaic power stations can be installed in various locations, they can better adapt to the energy needs and resource distribution in different regions. At the same time, distributed photovoltaic power stations can be flexibly expanded according to demand, and the power generation capacity can be gradually increased according to the growth of electricity demand without large-scale investment and construction. The advantages of distributed photovoltaic power stations include: 1. Increase the reliability of energy supply: Distributed photovoltaic power stations can disperse the power generation system in multiple locations, reduce the risk of single point failure, and improve the reliability of energy supply. 2. Reduce transmission loss: Distributed photovoltaic power stations can be closer to energy consumption points, reduce energy loss in the transmission process, and improve energy utilization efficiency. 3. Reduce dependence on traditional power grids: Distributed photovoltaic power stations can operate in off-grid or semi-off-grid modes, reduce dependence on traditional power grids, and improve the stability of power supply. 4. Utilize local resources: Distributed photovoltaic power stations can utilize available space at various locations, such as roofs, building surfaces, etc., to maximize the use of local resources for power generation. Distributed photovoltaic power stations play an important role in renewable energy generation and can contribute to energy transformation and sustainable development. It can not only provide clean energy, but also promote the decentralization and decentralization of energy, and promote the sustainability and reliability of energy supply.

[0080] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, including:

[0081] one or more processors;

[0082] A storage unit is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors can implement the method described in the above text. For details, please refer to the relevant records in the above text and will not be repeated here.

[0083] The present invention uses specific words to describe the embodiments of the present invention. For example, "first / second embodiment", "one embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic associated with at least one embodiment of the present invention. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different locations in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of the present invention may be appropriately combined.

[0084] In addition, those skilled in the art will appreciate that various aspects of the present invention may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present invention may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present invention may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0085] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0086] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A remote monitoring system for a distributed photovoltaic power station, characterized in that: include: A data acquisition module is used to obtain output voltage values ​​and output current values ​​of the photovoltaic cell assembly to be analyzed in the distributed photovoltaic power station at multiple predetermined time points within a predetermined time period; A wireless transmission module, used for transmitting the output voltage values ​​and output current values ​​at the plurality of predetermined time points to a background monitoring server via a wireless communication module; A correlation feature analysis module, used for analyzing, in the background monitoring server, correlation features between the output voltage values ​​and the output current values ​​at the plurality of predetermined time points; as well as The fault warning judgment module is used to determine whether to generate a fault warning prompt based on the associated features.

2. The remote monitoring system for a distributed photovoltaic power station according to claim 1, characterized in that: The correlation feature analysis module comprises: A data preprocessing unit, configured to perform data preprocessing on the output voltage values ​​and the output current values ​​at the plurality of predetermined time points to obtain an output voltage timing input vector and an output current timing input vector; An association coding unit, used for calculating a voltage-current output timing association coding matrix between the output voltage timing input vector and the output current timing input vector; A feature extraction unit, configured to extract features from the voltage-current output timing correlation coding matrix using a deep learning network model to obtain a voltage-current timing correlation feature vector; and The correlation feature acquisition unit is used to use the voltage-current time series correlation feature vector as the correlation feature.

3. The remote monitoring system for a distributed photovoltaic power station according to claim 2, characterized in that: The data preprocessing unit is used for: The output voltage values ​​and the output current values ​​at the plurality of predetermined time points are arranged according to the time dimension into the output voltage timing input vector and the output current timing input vector.

4. The remote monitoring system for a distributed photovoltaic power station according to claim 3, characterized in that: The deep learning network model is an electrical performance related feature extractor comprising a first convolutional layer and a second convolutional layer; Wherein, the feature extraction unit is used to: The voltage-current output timing correlation coding matrix is ​​passed through the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer to obtain the voltage-current timing correlation feature vector.

5. The remote monitoring system for a distributed photovoltaic power station according to claim 4, characterized in that: The first convolution layer and the second convolution layer respectively use two-dimensional convolution kernels with different scales.

6. The remote monitoring system for a distributed photovoltaic power station according to claim 5, characterized in that: The feature extraction unit is further used for: Use each layer of the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer to perform two-dimensional convolution processing, global pooling processing and non-linear activation processing on the input data in the forward pass of the layer to generate the voltage-current timing correlation feature vector from the last layer of the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer, wherein the input of the first layer of the electrical performance correlation feature extractor including the first convolution layer and the second convolution layer is the voltage-current output timing correlation coding matrix.

7. The remote monitoring system for a distributed photovoltaic power station according to claim 6, characterized in that: The fault warning judgment module comprises: a characteristic distribution correction unit, configured to perform characteristic distribution correction on the voltage-current time series correlation characteristic vector to obtain a corrected voltage-current time series correlation characteristic vector; and The multi-task classification unit is used to pass the corrected voltage-current time series correlation feature vector through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated.

8. The remote monitoring system for a distributed photovoltaic power station according to claim 7, characterized in that: The multi-task classification unit comprises: A fine-grained classification subunit, used for passing the corrected voltage-current time series correlation feature vector through a fine-grained classifier to obtain multiple fault category probability values; A coarse-grained classification subunit, used for passing the corrected voltage-current time series correlation feature vector through a coarse-grained classifier to obtain a first probability value and a second probability value; a probability weighted fusion subunit, configured to fuse the plurality of fault category probability values, the first probability value, and the second probability value in a probability weighted fusion manner to obtain a comprehensive expression probability value; and The classification result acquisition subunit is used to obtain the classification result based on the comprehensive expression probability value.

Images