Cloud-based industrial park micro-grid regulation and control system and method

Abstract

The invention relates to the technical field of micro-grid regulation and control, and discloses an industrial park micro-grid regulation and control system and method based on a cloud end, and the system employs an edge terminal to monitor the node voltage in real time, and compares the node voltage with a preset voltage threshold value which is received from the cloud end and is dynamically adjusted according to meteorological time sequence data. And if the node voltage exceeds the threshold value, the edge terminal sends out a local alarm and adopts a voltage-reactive droop control strategy to adjust the reactive power output of the photovoltaic inverter so as to absorb redundant reactive power to reduce the node voltage. All monitoring data, results and alarm signals are uploaded to the cloud for storage, and centralized management and global optimization are realized. Therefore, by combining real-time monitoring, dynamic adjustment and an automatic response mechanism, the stability and efficiency of the micro-grid of the industrial park are effectively improved, and the problems of poor coordination and untimely response in a traditional system are solved.

Description

Technical Field

[0001] This application relates to the technical field of microgrid control, and more specifically, to a cloud-based microgrid control system and method for industrial parks. Background Art

[0002] With the growing global demand for sustainable energy and the development of smart grid technology, distributed energy systems (such as solar photovoltaics, energy storage devices, etc.) in industrial parks are becoming increasingly popular. However, the integration and management of these systems pose new challenges, especially in terms of voltage stability and energy distribution.

[0003] Traditional microgrid management systems typically rely on local controllers to monitor and regulate power parameters, such as node voltage. Although this approach is effective in small-scale applications, its limitations become particularly evident in large-scale or complex industrial environments. First, local controllers often lack sufficient computing power and storage resources for complex data analysis and prediction, which limits their ability to respond to dynamic changes. Second, since each controller operates independently, it is difficult to achieve global optimization, resulting in suboptimal coordination between different regions. In addition, in the face of changing meteorological conditions (such as light intensity and temperature), statically set voltage thresholds may no longer be applicable, thus affecting the efficiency and stability of the system.

[0004] Therefore, an optimized cloud-based microgrid control solution for industrial parks is expected. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. Embodiments of this application provide a cloud-based microgrid control system and method for industrial parks, which effectively improve the stability and efficiency of the microgrid in industrial parks by combining real-time monitoring, dynamic adjustment, and automated response mechanisms, and solve the problems of poor coordination and untimely response existing in traditional systems.

[0006] According to one aspect of this application, a cloud-based microgrid control method for industrial parks is provided, including: an edge terminal monitors the node voltage in real time; the edge terminal compares the node voltage with a preset voltage threshold to obtain a monitoring result, and the preset voltage threshold is dynamically adjusted based on meteorological time series data received from the cloud; if the monitoring result is that the node voltage exceeds the preset voltage threshold, the edge terminal issues a local alarm signal; the edge terminal adjusts the reactive power output of the photovoltaic inverter using a voltage-reactive power droop control strategy to absorb reactive power and reduce the node voltage; the node voltage, the monitoring result, and the local alarm signal are uploaded to the cloud for storage.

[0007] In the above cloud-based industrial park microgrid control method, the edge terminal adopts a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb reactive power and reduce the node voltage, including: calculating the reactive power that the photovoltaic inverter needs to output using the following formula; the formula is: ; where is the droop coefficient, is the node voltage, is the preset voltage threshold, is the reactive power that the photovoltaic inverter needs to output.

[0008] In the above cloud-based industrial park microgrid control method, the preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud, including: grouping the meteorological time series data to obtain the light intensity time series data and the temperature time series data; performing time series encoding on the light intensity time series data and the temperature time series data based on the forward LTM model to obtain the light intensity time series feature and the temperature time series feature; fusing the light intensity time series feature and the temperature time series feature to obtain the light-temperature time series collaborative feature; decoding the light-temperature time series collaborative feature to obtain the voltage threshold adaptive adjustment factor; dynamically adjusting the preset voltage threshold based on the voltage threshold adaptive adjustment factor.

[0009] In the above cloud-based industrial park microgrid control method, dynamically adjusting the preset voltage threshold based on the voltage threshold adaptive adjustment factor includes: dynamically adjusting the preset voltage threshold using the following formula, the formula is: ; where is the voltage threshold adaptive adjustment factor, is the preset voltage threshold, represents the adjusted preset voltage threshold.

[0010] In the above cloud-based industrial park microgrid control method, fusing the light intensity time series feature and the temperature time series feature to obtain the light-temperature time series collaborative feature includes: constructing a feature cross-fusion network through a multi-layer perceptron, expanding the channel dimension of the light intensity time series feature, and performing an element-wise multiplication operation with the temperature time series feature to generate an initial light-temperature time series fusion feature; using a multi-head attention mechanism to extract spatio-temporal correlation features from the initial light-temperature time series fusion feature to generate the final light-temperature time series collaborative feature.

[0011] In the above-mentioned cloud-based industrial park microgrid control method, decoding the light-temperature time-series collaborative feature to obtain a voltage threshold adaptive adjustment factor includes: inputting the light-temperature time-series collaborative feature into a decoder for the voltage threshold adaptive adjustment factor to obtain the voltage threshold adaptive adjustment factor.

[0012] In the above-mentioned cloud-based industrial park microgrid control method, inputting the light-temperature time-series collaborative feature into a decoder for the voltage threshold adaptive adjustment factor to obtain the voltage threshold adaptive adjustment factor includes: the decoder for the voltage threshold adaptive adjustment factor decodes the light-temperature time-series collaborative feature according to the following formula: ; where is the voltage threshold adaptive adjustment factor, is matrix multiplication, is the decoding weight matrix for the voltage threshold adaptive adjustment factor, is the regression function.

[0013] In the above-mentioned cloud-based industrial park microgrid control method, it further includes: determining the decoding weight matrix for the droop coefficient through the structural distribution transfer application of the decoding weight matrix for the voltage threshold adaptive adjustment factor, and decoding the light-temperature time-series collaborative feature based on the decoding weight matrix for the droop coefficient to obtain the droop coefficient.

[0014] In the above-mentioned cloud-based industrial park microgrid control method, determining the decoding weight matrix for the droop coefficient through the structural distribution transfer application of the decoding weight matrix for the voltage threshold adaptive adjustment factor includes: performing a full-parameter domain space structural expansion on the decoding weight matrix for the voltage threshold adaptive adjustment factor to obtain an expansion matrix , where represents the transpose matrix of the decoding weight matrix.

[0015] Based on the regression function, a dynamic transfer matrix is obtained through the transfer response of the spatial hierarchical sub-parameter distribution structure of each row and each column of the expansion matrix, where the dynamic transfer matrix of the eigenvalue is: ; where and are respectively the th row and the th column of the expansion matrix

[0016] By based on the dynamic transfer matrix The decoding weight matrix for the voltage threshold adaptive adjustment factor is subjected to transfer projection modulation to obtain the decoding weight matrix for the droop coefficient : ; wherein is the weight coefficient represents dot product represents matrix multiplication

[0017] According to another aspect of the present application, there is also provided a cloud-based industrial park microgrid regulation system. The cloud-based industrial park microgrid regulation system includes: a node voltage real-time monitoring module for the edge terminal to monitor the node voltage in real time; a node voltage monitoring module for the edge terminal to compare the node voltage with a preset voltage threshold to obtain a monitoring result, and the preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud; a local alarm module for the edge terminal to send a local alarm signal if the monitoring result is that the node voltage exceeds the preset voltage threshold; a voltage-reactive power droop control unit for the edge terminal to adopt a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb reactive power and reduce the node voltage; and a cloud storage unit for uploading the node voltage, the monitoring result, and the local alarm signal to the cloud for storage

[0018] Compared with the prior art, the cloud-based industrial park microgrid regulation system and method provided by the present application utilize the edge terminal to monitor the node voltage in real time and compare it with a preset voltage threshold received from the cloud and dynamically adjusted according to the meteorological time series data. If the node voltage exceeds the threshold, the edge terminal issues a local alarm and adopts a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb the excess reactive power and reduce the node voltage. All monitoring data, results, and alarm signals are uploaded to the cloud for storage, realizing centralized management and global optimization. In this way, by combining real-time monitoring, dynamic adjustment, and an automated response mechanism, the stability and efficiency of the industrial park microgrid are effectively improved, and the problems of poor coordination and untimely response existing in the traditional system are solved BRIEF DESCRIPTION OF THE DRAWINGS

[0019] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification, and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps

[0020] Figure 1 is a schematic flowchart of a cloud-based industrial park microgrid regulation method according to an embodiment of the present application

[0021] Figure 2 It is a schematic flowchart of S2 in the cloud-based industrial park microgrid regulation method according to an embodiment of the present application.

[0022] Figure 3 It is a schematic flowchart of S23 in the cloud-based industrial park microgrid regulation method according to an embodiment of the present application.

[0023] Figure 4 It is a schematic flowchart of S6 in the cloud-based industrial park microgrid regulation method according to an embodiment of the present application.

[0024] Figure 5 It is a schematic block diagram of a cloud-based industrial park microgrid regulation system according to an embodiment of the present application. Detailed implementation manners

[0025] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0026] Figure 1 It is a schematic flowchart of a cloud-based industrial park microgrid regulation method according to an embodiment of the present application. As Figure 1 shown, the cloud-based industrial park microgrid regulation method includes: S1, the edge terminal monitors the node voltage in real time; S2, the edge terminal compares the node voltage with a preset voltage threshold to obtain a monitoring result, and the preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud; S3, if the monitoring result is that the node voltage exceeds the preset voltage threshold, the edge terminal issues a local alarm signal; S4, the edge terminal adjusts the reactive power output of the photovoltaic inverter using a voltage-reactive power droop control strategy to absorb reactive power and reduce the node voltage; S5, uploads the node voltage, the monitoring result, and the local alarm signal to the cloud for storage.

[0027] Specifically, in step S1, the edge terminal monitors the node voltage in real time. It should be understood that voltage monitoring in traditional power systems often relies on a centralized control center, and this approach has significant limitations. First, in large-scale or complex industrial environments, the response speed of the centralized system is slow and it is difficult to cope with the rapidly changing power grid conditions. Second, due to the latency of data transmission and processing, key decisions may not be executed in a timely manner, thus affecting the stability and security of the system. By deploying edge terminals to monitor the node voltage in real time, the above problems can be effectively overcome. The edge computing architecture allows data to be pre-processed close to the data source, reducing the amount of data transmission and latency, and greatly improving the response speed of the system. This means that once a voltage anomaly is detected, the edge terminal can take immediate action, such as issuing a local alarm signal or initiating a corresponding adjustment strategy, without waiting for instructions from the central server. This immediate response mechanism is crucial for maintaining the stability of the power grid. Especially in the face of emergencies such as sudden voltage increases or decreases, a quick response can prevent equipment damage or power grid collapse.

[0028] In a specific embodiment, edge terminal devices generally include high-precision voltage sensors, current sensors, and embedded computing modules, etc. Taking an industrial park as an example, multiple distributed photovoltaic systems are installed in the park, and each system is equipped with an inverter and an energy storage device. To monitor the voltage conditions of each node in real time, voltage sensors can be installed near each inverter or key distribution box, and the data is transmitted to the edge terminal via the RS485 bus or other communication protocols. These edge terminal devices are built with high-performance processors and sufficient storage resources, and can pre-process and analyze the collected data.

[0029] In the specific implementation process, the edge terminal not only needs to be responsible for data collection, but also needs to have a certain data pre-processing ability. For example, noise interference is removed through filtering algorithms to ensure that the collected voltage data is as accurate as possible. At the same time, the edge terminal calibrates the sensor regularly to ensure the measurement accuracy during long-term operation. In addition, considering factors such as electromagnetic interference in the industrial environment, the design of the edge terminal needs to follow strict anti-interference standards to ensure that it can work stably in a complex environment.

[0030] Furthermore, in order to achieve real-time monitoring, the edge terminal needs to maintain a continuous connection with the cloud. This is usually achieved through a wireless communication module (such as 4G / 5G or Wi-Fi). In some cases, a dedicated network or encrypted communication protocol can also be used to improve the security and reliability of data transmission. When the edge terminal receives the node voltage data, it immediately uploads this data to the cloud for storage and further analysis. This design not only reduces the burden on local devices, but also makes global data integration and optimization possible.

[0031] Specifically, in step S2, the edge terminal compares the node voltage with a preset voltage threshold to obtain a monitoring result, and the preset voltage threshold is dynamically adjusted based on the meteorological time-series data received from the cloud. It should be understood that in an industrial environment, the power demand and supply situation are often affected by various factors, among which the change of meteorological conditions is an important factor that cannot be ignored. For example, changes in light intensity and temperature directly affect the output power of a photovoltaic power generation system. When the light is sufficient, the photovoltaic system will generate more electrical energy, while high temperature may lead to a decline in equipment performance and affect the power generation efficiency. Therefore, simply relying on a statically set voltage threshold cannot meet the dynamic changing requirements, and it is easy to cause the voltage fluctuation to exceed the safe range, thus affecting the stability of the power grid. By introducing a dynamic adjustment mechanism, the voltage threshold can be flexibly adjusted according to real-time meteorological data, enabling the system to better cope with the challenges brought by environmental changes.

[0032] Specifically, while the edge terminal monitors the node voltage in real time, it also needs to compare it with a preset voltage threshold. This process not only helps to detect abnormal situations in a timely manner but also provides a basis for subsequent adjustment measures. If it is detected that the node voltage exceeds the preset voltage threshold, the edge terminal can immediately take corresponding measures, such as sending a local alarm signal or starting a reactive power compensation strategy to reduce the voltage. However, to ensure the effectiveness of these measures, the preset voltage threshold must be accurate and adaptable enough. Based on this, in this application, the preset voltage threshold is dynamically adjusted based on the meteorological time-series data received from the cloud.

[0033] In one embodiment, as Figure 2 shown, the dynamic adjustment of the preset voltage threshold based on the meteorological time-series data received from the cloud includes: S21, grouping the meteorological time-series data to obtain light intensity time-series data and temperature time-series data; S22, performing time-series encoding on the light intensity time-series data and the temperature time-series data based on the forward LTM model to obtain light intensity time-series features and temperature time-series features; S23, fusing the light intensity time-series features and the temperature time-series features to obtain light-temperature time-series collaborative features; S24, decoding the light-temperature time-series collaborative features to obtain a voltage threshold adaptive adjustment factor; S25, dynamically adjusting the preset voltage threshold based on the voltage threshold adaptive adjustment factor.

[0034] Specifically, in step S21, the meteorological time-series data is grouped to obtain the light intensity time-series data and the air temperature time-series data. It should be understood that meteorological conditions such as light intensity and air temperature have a significant impact on the output power of a photovoltaic power generation system. For example, under sufficient light conditions, the photovoltaic system will generate more electrical energy, while high temperatures may cause a decline in equipment performance and affect the power generation efficiency. Therefore, grouping the meteorological time-series data can better capture the relationships between these variables, thus providing a more accurate basis for dynamically adjusting the voltage threshold. By grouping the meteorological time-series data, the changing trends of different meteorological parameters and their impacts on the power grid can be separately focused on. For example, the light intensity directly affects the power generation of the photovoltaic system, while the air temperature may indirectly affect the power generation efficiency by influencing the operating temperature of the equipment. Separating these parameters for processing can not only reduce the interference between data but also improve the accuracy of feature extraction.

[0035] In a specific embodiment, a variety of meteorological parameters such as light intensity, air temperature, and humidity are collected in real time through high-precision meteorological monitoring equipment. The data obtained from the meteorological monitoring equipment is usually stored in the form of a time series, and each row of data contains all the meteorological parameter values at a certain moment. For ease of subsequent analysis, the data can be sorted according to the time stamp and imported into a database or a data processing platform (such as the Pandas library in Python). Next, according to specific analysis requirements, the data can be split by parameter type. For example, the light intensity data is separately extracted to form an independent time series, and similarly, the air temperature data is also extracted to form another time series. To ensure the quality of the data, data cleaning can also be performed before grouping. This includes steps such as removing outliers and filling in missing values. For example, if the light intensity data at a certain time point significantly deviates from the normal range, it can be considered an outlier and corrected. For missing values, interpolation methods or other appropriate filling methods can be used to complete the filling to ensure the continuity and integrity of the data.

[0036] Specifically, in step S22, temporal encoding based on the forward LTM model is performed on the light intensity temporal data and the air temperature temporal data to obtain light intensity temporal features and air temperature temporal features. It should be understood that light intensity and air temperature, as key factors affecting the output power of a photovoltaic power generation system, have significant time dependence. For example, the change in light intensity is not only affected by the daily cycle but may also be influenced by various factors such as weather changes and seasonal changes. Similarly, the change in air temperature also has certain regularity and randomness. By encoding these temporal data, these complex patterns and relationships can be effectively captured. As a deep learning model specialized in processing time series data, the forward LTM model can remember past information through its internal memory units and apply it to the prediction at the current moment, thereby improving the accuracy of feature extraction.

[0037] At the same time, meteorological temporal data usually contains a large amount of noise and redundant information. Directly using the original data for analysis may lead to overfitting or underfitting of the model. Through temporal encoding based on the forward LTM model, the data can be dimensionally reduced and denoised, and the most representative features can be extracted. These features not only contain the main information of the original data but also can better reflect the potential patterns and trends in the data. For example, in the light intensity temporal data, the forward LTM model can capture the change trend of light intensity during a day through its memory units and convert this information into a more representative feature vector. Similarly, for the air temperature temporal data, the forward LTM model can also extract the main patterns of temperature change.

[0038] In a specific embodiment, a forward LTM model is constructed and trained. The forward LTM model consists of multiple LSTM (Long Short-Term Memory) units, and each unit contains components such as an input gate, a forget gate, and an output gate. These components work together to enable the model to effectively capture the long-term dependence relationships in the time series. Specifically, the input of the model is the preprocessed light intensity and air temperature temporal data, and the output is the corresponding feature vector. During the training process, a supervised learning method can be adopted, using historical data as the input and the known future light intensity and air temperature changes as the labels. By continuously adjusting the weight and bias parameters of the model, the model can accurately predict future meteorological changes. Once the model training is completed, it can be applied to new temporal data to extract light intensity temporal features and air temperature temporal features.

[0039] Specifically, in step S23, the light intensity time-series feature and the air temperature time-series feature are fused to obtain a light-temperature time-series collaborative feature. It should be understood that the changes in light intensity and air temperature have a certain spatio-temporal correlation. The light intensity usually shows an obvious daily periodic change, while the air temperature is affected by seasons, diurnal variations, and local climate conditions. By fusing these two features, the spatio-temporal dependence relationship between them can be captured, thereby improving the prediction ability of the model. For example, during high-temperature periods in summer, the light intensity is high, but the actual power generation efficiency of the photovoltaic system may decrease due to the increase in temperature; while during low-temperature periods in winter, although the light intensity is low, the efficiency of the photovoltaic system is relatively high due to the low temperature. Therefore, fusing these features can help predict the output power of the photovoltaic system more accurately, and then adjust the voltage threshold. Moreover, the changes in light intensity and air temperature are often not independent but interact with each other. For example, a change in light intensity may cause the air temperature to rise or fall, and vice versa. By fusing these two features, this complex interaction can be captured, thereby improving the robustness and generalization ability of the model. This is particularly important for dynamically adjusting the voltage threshold because only by fully considering all relevant factors can the effectiveness and timeliness of the control measures be ensured. In one embodiment, the light intensity time-series feature and the air temperature time-series feature can be fused by concatenation or cascading to obtain the light-temperature time-series collaborative feature.

[0040] In a preferred embodiment, as Figure 3 shown, in step S23, fusing the light intensity time-series feature and the air temperature time-series feature to obtain a light-temperature time-series collaborative feature includes: S231, constructing a feature cross-fusion network through a multi-layer perceptron to expand the channel dimension of the light intensity time-series feature and perform an element-wise multiplication operation with the air temperature time-series feature to generate an initial light-temperature time-series fusion feature; S232, using a multi-head attention mechanism to extract spatio-temporal correlation features from the initial light-temperature time-series fusion feature to generate a final light-temperature time-series collaborative feature.

[0041] First, starting from the light intensity time-series feature and the air temperature time-series feature encoded by the forward LTM model, a multi-layer perceptron (MLP) network is constructed. This network expands the channel dimension of the light intensity time-series feature and performs an element-wise multiplication operation with the air temperature time-series feature to generate an initial light-temperature time-series fusion feature. This process can not only enhance the representation ability of the features but also reduce the redundant information between the data, making the generated fusion feature more representative.

[0042] Next, a multi-head attention mechanism is adopted to extract spatio-temporal correlation features from the initial light-temperature time-series fusion features, generating the final light-temperature time-series collaborative features. As a powerful feature extraction tool, the multi-head attention mechanism can effectively capture complex patterns and dependencies in time series. Specifically, the multi-head attention mechanism processes the input features through multiple parallel attention heads, and each attention head can focus on different subsets of features, thus capturing richer information. For example, in one attention head, it may focus on the change trend of light intensity and its impact on temperature; while in another attention head, it may focus on the impact of temperature change on light intensity.

[0043] Specifically, in step S24, the light-temperature time-series collaborative features are decoded to obtain a voltage threshold adaptive adjustment factor. In a specific embodiment, decoding the light-temperature time-series collaborative features to obtain a voltage threshold adaptive adjustment factor includes: inputting the light-temperature time-series collaborative features into a decoder for the voltage threshold adaptive adjustment factor to obtain the voltage threshold adaptive adjustment factor.

[0044] In a specific embodiment, inputting the light-temperature time-series collaborative features into a decoder for the voltage threshold adaptive adjustment factor to obtain the voltage threshold adaptive adjustment factor includes: the decoder for the voltage threshold adaptive adjustment factor decodes the light-temperature time-series collaborative features according to the following formula, and the formula is: ; where is the voltage threshold adaptive adjustment factor, is matrix multiplication, is the decoding weight matrix for the voltage threshold adaptive adjustment factor, is the regression function.

[0045] Specifically, in the training stage, a historical data set is used to optimize the decoding weight matrix. Using a historical data set containing light intensity and temperature data at multiple time points, each time point corresponds to a known voltage threshold adaptive adjustment factor. By minimizing the error between the predicted value and the true value (such as using the mean square error loss function), the decoding weight matrix can be iteratively updated, enabling the decoder to accurately map the light-temperature time-series collaborative features to the voltage threshold adaptive adjustment factor.

[0046] Specifically, in step S25, the preset voltage threshold is dynamically adjusted based on the voltage threshold adaptive adjustment factor. It should be understood that once the voltage threshold adaptive adjustment factor is generated, the cloud platform will send it to each edge terminal. After receiving this data, the edge terminal will dynamically adjust the preset voltage threshold according to the current node voltage situation and the received adjustment factor. For example, during a period with high light intensity, since the power generation of the photovoltaic system increases, it may cause a local voltage increase. At this time, the edge terminal can appropriately increase the voltage threshold according to the adjustment factor provided by the cloud to avoid unnecessary alarms and adjustment measures. On the contrary, in the case of insufficient light or low temperature, it may be necessary to lower the voltage threshold to ensure that the voltage will not be too low to affect the normal operation of electrical equipment.

[0047] In a specific embodiment, dynamically adjusting the preset voltage threshold based on the voltage threshold adaptive adjustment factor includes: dynamically adjusting the preset voltage threshold according to the following formula, and the formula is: ; where is the voltage threshold adaptive adjustment factor, is the preset voltage threshold, represents the adjusted preset voltage threshold.

[0048] Specifically, in step S3, if the monitoring result shows that the node voltage exceeds the preset voltage threshold, the edge terminal issues a local alarm signal. It should be understood that voltage fluctuation is one of the important factors affecting the stability of the power system. In an industrial environment, due to frequent load changes, unstable output power of distributed energy systems (such as photovoltaic power generation systems), etc., the node voltage may fluctuate. If these fluctuations exceed the safe range, it may cause a decline in the performance of electrical equipment or even damage. By setting a preset voltage threshold and issuing an alarm signal when the node voltage exceeds this threshold, it can timely remind the operation and maintenance personnel to take corresponding measures to prevent the problem from deteriorating further.

[0049] Here, considering that traditional centralized control systems often rely on a central server for data processing and decision-making, which may lead to delays, especially in the case of limited network bandwidth or interrupted communication links. The edge computing architecture allows data to be preliminarily processed and analyzed close to the data source, reducing the data transmission volume and delay. Once a voltage anomaly is detected, the edge terminal can immediately issue a local alarm signal without waiting for instructions from the central server. This instant response mechanism is crucial for maintaining the stability of the power grid. Especially in the face of emergencies such as sudden voltage increases or decreases, a quick response can avoid equipment damage or power grid collapse.

[0050] Specifically, once it is detected that the node voltage exceeds the preset threshold, the edge terminal will immediately send out a local alarm signal. The alarm signal can be transmitted to relevant personnel in various ways, including but not limited to audible and visual alarms, SMS notifications, email, or push notifications from a mobile application, etc. For example, when the node voltage in a certain area exceeds the upper limit value, the edge terminal can emit an alarm sound and a flashing light through the connected audible and visual alarm, while sending an alarm message to the central management system through the wireless communication module, and notifying the operation and maintenance personnel via SMS or email. In this way, even in the case of a network interruption, the local alarm signal can ensure that relevant personnel are informed of the abnormal situation in a timely manner.

[0051] Specifically, in step S4, the edge terminal adopts a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb reactive power and reduce the node voltage. It should be understood that by adopting the voltage-reactive power droop control strategy, the reactive power output of the photovoltaic inverter can be dynamically adjusted according to the real-time monitored node voltage situation, so as to maintain the node voltage within a safe range. Here, the voltage-reactive power droop control is a decentralized reactive power control strategy, and its core idea is to simulate the adjustment characteristics of a traditional synchronous generator and establish the relationship between the reactive power output of distributed power sources and the local voltage. Among them, the droop characteristic usually shows a "downward heavy characteristic curve", that is, the higher the voltage, the less reactive power of the inductor needs to be emitted (or the more reactive power of the capacitor needs to be absorbed), and the lower the voltage, the more reactive power of the inductor needs to be emitted (or the less reactive power of the capacitor needs to be absorbed).

[0052] In a specific embodiment, the edge terminal adopts a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb reactive power and reduce the node voltage, including: calculating the reactive power that the photovoltaic inverter needs to output according to the following formula; The formula is: ; where is the droop coefficient, is the node voltage, is the preset voltage threshold, is the reactive power that the photovoltaic inverter needs to output. Here, in a preferred embodiment, the preset voltage threshold in the above formula uses an adjusted preset voltage threshold.

[0053] In particular, considering that the light intensity directly affects the power generation of the photovoltaic system. When the light intensity is high, the photovoltaic system will generate more electrical energy, which may lead to a local voltage increase; while in the case of insufficient light, the power generation of the photovoltaic system decreases, which may lead to a voltage drop. Therefore, under different light conditions, it is necessary to adjust the droop coefficient to adapt to different voltage fluctuations. For example, during periods of sufficient light, the droop coefficient can be appropriately reduced so that the inverter can regulate reactive power more smoothly; while during periods of weak light, the droop coefficient can be increased to respond more quickly to voltage changes.

[0054] In a preferred example, while dynamically adjusting the preset voltage threshold based on the meteorological time-series data received from the cloud, preferably, the droop coefficient for reactive power is adjusted simultaneously. That is, when decoding the light-temperature time-series collaborative feature to obtain the voltage threshold adaptive adjustment factor, the droop coefficient can also be decoded according to the light-temperature time-series collaborative feature. And, compared with the scheme of training two decoders simultaneously to separately decode the voltage threshold adaptive adjustment factor and the droop coefficient, considering the correlation between the voltage threshold adaptive adjustment factor and the droop coefficient based on the meteorological time-series data characteristics, the decoding parameter mapping space for the droop coefficient can be determined by applying the structural distribution transfer of the decoding parameter mapping space of the decoder for the voltage threshold adaptive adjustment factor.

[0055] In a specific embodiment, it further includes: S6, determining the decoding weight matrix for the droop coefficient through the structural distribution transfer application of the decoding weight matrix for the voltage threshold adaptive adjustment factor, and decoding the light-temperature time-series collaborative feature based on the decoding weight matrix for the droop coefficient to obtain the droop coefficient.

[0056] In a specific embodiment, as Figure 4 shown, in step S6, determining the decoding weight matrix for the droop coefficient through the structural distribution transfer application of the decoding weight matrix for the voltage threshold adaptive adjustment factor includes: S61, performing a full-parameter domain space structural expansion on the decoding weight matrix for the voltage threshold adaptive adjustment factor to obtain an expansion matrix; S62, based on the regression function, obtaining a dynamic transfer matrix through the transfer response of the spatial hierarchical sub-parameter distribution structure of each row and each column of the expansion matrix; S63, obtaining the decoding weight matrix for the droop coefficient through transfer projection modulation of the decoding weight matrix for the voltage threshold adaptive adjustment factor based on the dynamic transfer matrix.

[0057] Specifically, assuming that the light-temperature time-series collaborative feature is represented as , the decoding parameter mapping space of the voltage threshold adaptive adjustment factor decoder can be expressed as , where is matrix multiplication, is the decoding weight matrix for the voltage threshold adaptive adjustment factor, and is a regression function, such as a weighted sum regression function. Thus, the decoding weight matrix for the voltage threshold adaptive adjustment factor can be first extended in the full parameter domain space structure to obtain , where represents the transpose matrix of the decoding weight matrix; then, based on the regression function , the dynamic transfer matrix is obtained by transferring the response of the spatial hierarchical sub-parameter distribution structure of each row and each column of the extended matrix . Among them, the -th eigenvalue of the dynamic transfer matrix is: ; where and are the -th row and the -th column of the extended matrix , respectively.

[0058] Finally, the decoding weight matrix for the droop coefficient is obtained by performing transfer projection modulation on the decoding weight matrix for the voltage threshold adaptive adjustment factor based on the dynamic transfer matrix : ; where is the weight coefficient, represents dot product, represents matrix multiplication.

[0059] Thus, through the structured mapping and transfer of the sub-level parameter distribution under the predetermined parameter mapping space system, the hierarchical representation learning of the decoding weight matrix for the voltage threshold adaptive adjustment factor based on the same functional parameter space is realized, thereby reducing the computational burden.

[0060] Specifically, in step S5, the node voltage, the monitoring result, and the local alarm signal are uploaded to the cloud for storage. It should be understood that by uploading the node voltage and the monitoring result to the cloud, the data of all edge terminals can be centrally managed and analyzed, so as to better identify potential problem points and take corresponding measures. For example, when the node voltage in a certain area is abnormal, the problem can be quickly located through the cloud platform, and the resources in other areas can be coordinated for support. At the same time, uploading the local alarm signal to the cloud helps to improve the response speed and reliability of the system. Although the edge terminal can immediately send out a local alarm signal when detecting a voltage anomaly, the scope of action of this alarm signal is limited and may not be able to notify all relevant personnel in time. By uploading the alarm signal to the cloud, it can be ensured that the relevant information can be quickly transmitted to the central management system and the relevant operation and maintenance personnel. For example, when the node voltage in a certain area exceeds the preset threshold, the edge terminal can send an alarm message to the cloud through the wireless communication module and notify the operation and maintenance personnel by text message or email. In this way, even in the case of network interruption, the local alarm signal can ensure that the relevant personnel are informed of the anomaly in time.

[0061] In summary, the cloud-based industrial park microgrid regulation method provided by this application uses edge terminals to monitor the node voltage in real time and compares it with a preset voltage threshold received from the cloud and dynamically adjusted according to meteorological time series data. If the node voltage exceeds the threshold, the edge terminal issues a local alarm and adopts a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb the excess reactive power and reduce the node voltage. All monitoring data, results, and alarm signals are uploaded to the cloud for storage to achieve centralized management and global optimization. In this way, by combining real-time monitoring, dynamic adjustment, and an automated response mechanism, the stability and efficiency of the industrial park microgrid are effectively improved, and the problems of poor coordination and untimely response existing in the traditional system are solved.

[0062] This application also provides a cloud-based industrial park microgrid regulation system, as Figure 5As shown, the cloud-based industrial park microgrid control system 500 includes: a node voltage real-time monitoring module 510 for the edge terminal to monitor the node voltage in real time; a node voltage monitoring module 520 for the edge terminal to compare the node voltage with a preset voltage threshold to obtain a monitoring result, where the preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud; a local alarm module 530 for the edge terminal to send a local alarm signal if the monitoring result is that the node voltage exceeds the preset voltage threshold; a voltage-reactive power droop control unit 540 for the edge terminal to adjust the reactive power output of the photovoltaic inverter using a voltage-reactive power droop control strategy to absorb reactive power and reduce the node voltage; and a cloud storage unit 550 for uploading the node voltage, the monitoring result, and the local alarm signal to the cloud for storage.

[0063] An embodiment of the present application also provides a computer-readable storage medium in which computer program code is stored. When the computer program code runs on a computer, it causes the computer to execute the above-related method steps to implement a cloud-based industrial park microgrid control method provided in the above embodiment.

[0064] An embodiment of the present application also provides a computer program product. When the computer program product runs on a computer, it causes the computer to execute the above-related steps to implement a cloud-based industrial park microgrid control method provided in the above embodiment.

[0065] Among them, the system, computer-readable storage medium, or computer program product provided in the embodiments of the present application are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be elaborated here.

[0066] It should be noted that: the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous. Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A cloud-based industrial park microgrid control method, characterized in that: include: The edge terminal monitors the node voltage in real time; The edge terminal compares the node voltage with a preset voltage threshold to obtain a monitoring result, and the preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud; If the monitoring result is that the node voltage exceeds the preset voltage threshold, the edge terminal sends a local alarm signal; The edge terminal adopts a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb reactive power and reduce the node voltage; The node voltage, the monitoring result and the local alarm signal are uploaded to the cloud for storage.

2. The cloud-based industrial park microgrid control method according to claim 1 is characterized in that: The edge terminal uses a voltage-reactive power droop control strategy to adjust the reactive output of the photovoltaic inverter to absorb reactive power and reduce the node voltage, including: calculating the reactive power that the photovoltaic inverter needs to output using the following formula; the formula is: ;in, is the droop coefficient, is the node voltage, is the preset voltage threshold, It is the reactive power that the photovoltaic inverter needs to output.

3. The cloud-based industrial park microgrid control method according to claim 2 is characterized in that: The preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud, including: grouping the meteorological time series data to obtain light intensity time series data and air temperature time series data; performing time series encoding based on the forward LTM model on the light intensity time series data and the air temperature time series data to obtain light intensity time series characteristics and air temperature time series characteristics; fusing the light intensity time series characteristics and the air temperature time series characteristics to obtain light-air temperature time series synergistic characteristics; decoding the light-air temperature time series synergistic characteristics to obtain a voltage threshold adaptive adjustment factor; and dynamically adjusting the preset voltage threshold based on the voltage threshold adaptive adjustment factor.

4. The cloud-based industrial park microgrid control method according to claim 3 is characterized in that: Based on the voltage threshold adaptive adjustment factor, dynamically adjusting the preset voltage threshold includes: dynamically adjusting the preset voltage threshold according to the following formula, wherein the formula is: ;in, is the voltage threshold adaptive adjustment factor, is the preset voltage threshold, Indicates the adjusted preset voltage threshold.

5. The cloud-based industrial park microgrid control method according to claim 4 is characterized in that: The light intensity time series feature and the temperature time series feature are integrated to obtain the light-temperature time series coordinated feature, including: constructing a feature cross-fusion network through a multi-layer perceptron, expanding the channel dimension of the light intensity time series feature, and performing an element-by-element multiplication operation with the temperature time series feature to generate an initial light-temperature time series fusion feature; and using a multi-head attention mechanism to extract spatiotemporal correlation features from the initial light-temperature time series fusion feature to generate a final light-temperature time series coordinated feature.

6. The cloud-based industrial park microgrid control method according to claim 5 is characterized in that: The light-air temperature timing cooperative feature is decoded to obtain a voltage threshold adaptive adjustment factor, including: inputting the light-air temperature timing cooperative feature into a decoder for the voltage threshold adaptive adjustment factor to obtain the voltage threshold adaptive adjustment factor.

7. The cloud-based industrial park microgrid control method according to claim 6 is characterized in that: Inputting the light-air temperature timing cooperative feature into a decoder for voltage threshold adaptive adjustment factor to obtain the voltage threshold adaptive adjustment factor, including: the decoder for voltage threshold adaptive adjustment factor decodes the light-air temperature timing cooperative feature according to the following formula, wherein the formula is: ;in, is the voltage threshold adaptive adjustment factor, is matrix multiplication, is the decoding weight matrix for the voltage threshold adaptive adjustment factor, is the regression function.

8. The cloud-based industrial park microgrid control method according to claim 7 is characterized in that: Also includes: The decoding weight matrix for the droop coefficient is determined by applying the structural distribution transfer of the decoding weight matrix for the voltage threshold adaptive adjustment factor, and the light-air temperature timing cooperative feature is decoded based on the decoding weight matrix for the droop coefficient to obtain the droop coefficient.

9. The cloud-based industrial park microgrid control method according to claim 8 is characterized in that: Determining a decoding weight matrix for a droop coefficient by applying a structural distribution transfer of the decoding weight matrix for the voltage threshold adaptive adjustment factor, including: performing a full parameter domain spatial structural expansion on the decoding weight matrix for the voltage threshold adaptive adjustment factor to obtain an expanded matrix ;in, Represents the transposed matrix of the decoding weight matrix; Based on the regression function, the dynamic transfer matrix is ​​obtained by expanding the transfer response of the spatial hierarchical sub-parameter distribution structure of each row and column of the matrix , where the dynamic transfer matrix No. Eigenvalue for: ;in, and The expansion matrix No. Row and Column; Based on the dynamic transfer matrix The decoding weight matrix for the voltage threshold adaptive adjustment factor Perform transfer projection modulation to obtain a decoding weight matrix for the droop coefficient : ;in, is the weight coefficient, represents dot product, Represents matrix multiplication.

10. A cloud-based industrial park microgrid control system, characterized in that: include: Node voltage real-time monitoring module, used for edge terminals to monitor node voltage in real time; A node voltage monitoring module, used for the edge terminal to compare the node voltage with a preset voltage threshold to obtain a monitoring result, wherein the preset voltage threshold is dynamically adjusted based on the meteorological time series data received from the cloud; A local alarm module, used for the edge terminal to send a local alarm signal if the monitoring result is that the node voltage exceeds a preset voltage threshold; a voltage-reactive power droop control unit, used for the edge terminal to use a voltage-reactive power droop control strategy to adjust the reactive power output of the photovoltaic inverter to absorb reactive power and reduce the node voltage; The cloud storage unit is used to upload the node voltage, the monitoring result and the local alarm signal to the cloud for storage.

Images