Power dispatching method and system for distributed photovoltaic system

Abstract

The present application relates to the field of power management, and provides a power dispatching method and system for distributed photovoltaic systems, which collects the historical power consumption data of the target area, and uses data analysis and processing technology based on deep learning to perform time series cutting and local time series feature encoding of the historical power consumption data, so as to automatically evaluate the power consumption forecast demand of the target area according to the spatiotemporal propagation aggregation representation of the local time series features of each historical power consumption data in the full time domain. In this way, it is possible to automatically learn and extract complex time series features from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, it is possible to capture the complex time series relationships and dynamic interactions between data, thereby improving the accuracy of prediction.

Description

Technical Field

[0001] The present application relates to the field of power management, and more specifically, to a power dispatching method and system for a distributed photovoltaic system. Background Art

[0002] Distributed photovoltaic systems are solar power generation facilities built at the user end or in the adjacent area. They usually have a small installed capacity and are directly connected to the power system on the user side. Due to the growing global demand for clean energy, distributed photovoltaic systems are rapidly becoming popular and applied around the world. In such systems, power dispatch plays a vital role.

[0003] Patent CN118523285A proposes a power dispatching method and system for distributed photovoltaic systems, which analyzes the predicted demand of electrical equipment in the target area, arranges monitoring devices to collect data, integrates photovoltaic power generation information, calculates the power balance coefficient, evaluates the safety status, and performs dispatching to ensure stable operation of the power system.

[0004] Electricity demand forecasting is crucial in the management and dispatch of power systems. Especially in distributed photovoltaic systems, accurate forecasting becomes even more important due to the uneven temporal and spatial distribution of electricity production and consumption. However, the above-mentioned patent predicts the demand for power equipment in the target area by collecting historical power consumption data in the target area and classifying and analyzing the power equipment, and then combining historical data, equipment power consumption patterns and external factors to establish a forecasting model to derive future power demand. However, this method based on a forecasting model usually needs to be performed manually, including steps such as data cleaning, feature selection and feature combination. These processes are not only time-consuming and labor-intensive, but also prone to human errors. In addition, this type of forecasting method is usually based on statistics or simple mathematical models, which makes it difficult to capture complex nonlinear relationships and the interactive correlations between various data features, resulting in limited forecasting accuracy.

[0005] Therefore, an optimized power dispatching scheme for distributed photovoltaic systems is needed. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present application provides a power dispatching method and system for a distributed photovoltaic system.

[0007] According to one aspect of the present application, a power dispatching method for a distributed photovoltaic system is provided, comprising: determining a predicted power demand of a target area; installing a power monitoring device in the target area according to the predicted power demand to perform real-time monitoring and feedback on the target area; acquiring photovoltaic power generation data through a distributed photovoltaic power generation system; determining a power balance coefficient of the target area based on power feedback information and the photovoltaic power generation data; generating a judgment result based on the power balance coefficient, the judgment result being the power safety operation status of the target area, and performing power dispatching on the target area based on the judgment result; wherein determining the predicted power demand of the target area comprises:

[0008] Collecting historical electricity consumption data of the target area;

[0009] The power consumption historical data is sequence-cut to obtain a set of local power consumption historical data, and then local power consumption time series sequence encoding is performed to obtain a set of local power consumption time series forward propagation representation vectors;

[0010] The set of the local time series forward propagation representation vectors of the power consumption is subjected to propagation aggregation based on spatiotemporal potential energy to obtain the full-time domain propagation aggregation representation of the power consumption, including: extracting the current local time series forward propagation representation vector of the power consumption from the set of the local time series forward propagation representation vectors of the power consumption, and defining other local time series forward propagation representation vectors of the power consumption in the set of the local time series forward propagation representation vectors of the power consumption as the historical local time series forward propagation representation vectors to obtain a sequence of the historical local time series forward propagation representation vectors of the power consumption; and performing feature aggregation based on energy propagation on the sequence of the historical local time series forward propagation representation vectors of the power consumption and the current local time series forward propagation representation vector of the power consumption to obtain the full-time domain propagation aggregation representation of the power consumption;

[0011] Based on the aggregated representation of the power consumption propagation in the entire time domain, the predicted power consumption demand of the target area is obtained.

[0012] In the above-mentioned power dispatching method for distributed photovoltaic systems, the electricity consumption historical data is sequence-cut to obtain a set of local electricity consumption historical data, and then the local electricity consumption time series sequence encoding is performed to obtain a set of local electricity consumption time series forward propagation representation vectors, including: based on a predetermined time scale, the electricity consumption historical data is sequence-cut to obtain the set of local electricity consumption historical data; each local electricity consumption historical data in the set of local electricity consumption historical data is input into the electricity consumption local time series sequence encoder based on the forward LSTM model to obtain the set of local electricity consumption time series forward propagation representation vectors.

[0013] In the above-mentioned power dispatching method for distributed photovoltaic systems, the sequence of the local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption are subjected to feature aggregation based on energy propagation to obtain the full time domain propagation aggregation representation of electricity consumption, including: calculating the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption in the sequence of the local time series forward propagation representation vectors of historical electricity consumption; based on the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption, the sequence of the local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption are subjected to feature message passing aggregation to obtain the full time domain propagation aggregation representation of electricity consumption.

[0014] In the above-mentioned power dispatching method for distributed photovoltaic systems, the static energy value of the power consumption of each local time series forward propagation representation vector of historical power consumption in the sequence of the local time series forward propagation representation vector of historical power consumption is calculated, including: respectively calculating the mean and variance of the local time series forward propagation representation vector of historical power consumption to obtain the mean of the local time series characteristics of historical power consumption and the variance of the local time series characteristics of historical power consumption; subtracting the local time series forward propagation representation vector of historical power consumption from the mean of the local time series characteristics of historical power consumption by position, and calculating the local time series difference of historical power consumption The fourth power of the characteristic value of each position in the vector is used to obtain the local time series differential modulation vector of historical electricity consumption; the expected value of the local time series differential modulation vector of historical electricity consumption is calculated, and the expected value is divided by the square of the variance of the local time series characteristic of historical electricity consumption to obtain the local time series energy value of historical electricity consumption; the maximum value of the local time series forward propagation representation vector of historical electricity consumption is extracted, and the maximum value of the local time series characteristic of historical electricity consumption is added to the local time series energy value of historical electricity consumption to obtain the static energy value of electricity consumption corresponding to the local time series forward propagation representation vector of historical electricity consumption.

[0015] In the above-mentioned power dispatching method for distributed photovoltaic systems, based on the static energy value of the electricity consumption of each historical electricity consumption local time series forward propagation representation vector, the sequence of the historical electricity consumption local time series forward propagation representation vector and the current electricity consumption local time series forward propagation representation vector are subjected to feature message transmission aggregation to obtain the full time domain propagation aggregation representation of the electricity consumption, including: based on the static energy value of the electricity consumption of each historical electricity consumption local time series forward propagation representation vector, calculating the dynamic energy spatiotemporal transmission of electricity consumption of each historical electricity consumption local time series forward propagation representation vector relative to the current electricity consumption local time series forward propagation representation vector value; input the dynamic energy spatiotemporal transfer value of each historical electricity consumption local time series forward propagation representation vector into the energy transfer gating module based on the mask function to obtain a sequence of dynamic energy spatiotemporal transfer weights of electricity consumption; based on the sequence of dynamic energy spatiotemporal transfer weights of electricity consumption, calculate the weighted sum of the sequence of historical electricity consumption local time series forward propagation representation vectors to obtain the historical electricity consumption transfer aggregation modulation vector; calculate the positional sum of the historical electricity consumption transfer aggregation modulation vector and the current electricity consumption local time series forward propagation representation vector to obtain the full-time domain forward propagation aggregation representation vector of electricity consumption as the full-time domain propagation aggregation representation of electricity consumption.

[0016] In the above-mentioned power dispatching method for distributed photovoltaic systems, based on the static energy value of the historical local time series forward propagation representation vector, the dynamic energy spatiotemporal transfer value of the historical local time series forward propagation representation vector relative to the current local time series forward propagation representation vector is calculated, including: determining the spatial attenuation factor of energy of electricity consumption based on the spatial span coefficient between the historical local time series forward propagation representation vector and the current local time series forward propagation representation vector; determining the time attenuation factor of energy of electricity consumption based on the time span coefficient between the historical local time series forward propagation representation vector and the current local time series forward propagation representation vector; based on the spatial attenuation factor of energy of electricity consumption and the time attenuation factor of energy of electricity consumption, performing spatiotemporal modulation on the static energy value of the historical local time series forward propagation representation vector to obtain the dynamic energy spatiotemporal transfer value of energy of electricity consumption of the historical local time series forward propagation representation vector relative to the current local time series forward propagation representation vector.

[0017] In the above-mentioned power dispatching method for distributed photovoltaic systems, the power consumption energy spatial attenuation factor is determined based on the spatial span coefficient between the local time series forward propagation representation vectors of each historical power consumption and the local time series forward propagation representation vector of the current power consumption, including: calculating the square of the spatial span value between the local time series forward propagation representation vector of the historical power consumption and the local time series forward propagation representation vector of the current power consumption to obtain the power consumption spatial span coefficient; dividing the power consumption static energy value corresponding to the local time series forward propagation representation vector of the historical power consumption by the power consumption spatial span coefficient to obtain the power consumption energy spatial attenuation factor.

[0018] In the above-mentioned power dispatching method for distributed photovoltaic systems, the power consumption energy time decay factor is determined based on the time span coefficient between the local time series forward propagation representation vectors of each historical power consumption and the local time series forward propagation representation vector of the current power consumption, including: subtracting the power consumption timestamp of the local time series forward propagation representation vector of the current power consumption from the power consumption timestamp of the local time series forward propagation representation vector of the historical power consumption and rounding down to obtain the power consumption time span value; dividing the power consumption time span value by the time decay parameter to obtain the power consumption time series span energy decay coefficient; using the power consumption time series span energy decay coefficient as the exponential power, calculating the natural exponential function value with the natural constant e as the base to obtain the power consumption time span coefficient; dividing the power consumption static energy value corresponding to the local time series forward propagation representation vector of the historical power consumption by the power consumption time span coefficient to obtain the power consumption energy time decay factor.

[0019] In the above-mentioned power dispatching method for distributed photovoltaic systems, the predicted power demand of the target area is obtained based on the full-time domain propagation aggregation representation of the power consumption, including: inputting the full-time domain forward propagation aggregation representation vector of the power consumption into a decoder-based power demand predictor to obtain the predicted power demand of the target area.

[0020] According to another aspect of the present application, a power dispatching system for a distributed photovoltaic system is provided, comprising: a power forecast demand determination module, used to determine the power forecast demand of a target area; an actual power demand data monitoring module, used to install a power monitoring device in the target area according to the power forecast demand to perform real-time monitoring and feedback on the target area; a photovoltaic power generation data acquisition module, used to obtain photovoltaic power generation data through a distributed photovoltaic power generation system; a power balance coefficient determination module, used to determine the power balance coefficient of the target area based on power feedback information and the photovoltaic power generation data; a power dispatching module, used to generate a judgment result based on the power balance coefficient, the judgment result being the power safety operation status of the target area, and to perform power dispatching on the target area based on the judgment result; wherein the power forecast demand determination module comprises:

[0021] A power consumption history data acquisition unit, used to collect the power consumption history data of the target area;

[0022] A power consumption historical data time series encoding unit, used for performing sequence cutting on the power consumption historical data to obtain a set of local power consumption historical data, and then performing local power consumption time series encoding to obtain a set of local power consumption time series forward propagation representation vectors;

[0023] A unit for generating a full-time-domain propagation aggregation representation of electricity consumption, used for performing propagation aggregation based on spatiotemporal potential energy on the set of the local time series forward propagation representation vectors of the electricity consumption to obtain a full-time-domain propagation aggregation representation of electricity consumption, including: a subunit for dividing the local time series forward propagation representation of current electricity consumption and historical electricity consumption, used for extracting the local time series forward propagation representation vector of current electricity consumption from the set of the local time series forward propagation representation vectors of electricity consumption, and defining other local time series forward propagation representation vectors of electricity consumption in the set of the local time series forward propagation representation vectors of electricity consumption as the local time series forward propagation representation vectors of historical electricity consumption to obtain a sequence of local time series forward propagation representation vectors of historical electricity consumption; a subunit for aggregating the characteristics of the local time series of current electricity consumption and historical electricity consumption, used for performing feature aggregation based on energy propagation on the sequence of the local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption to obtain the full-time-domain propagation aggregation representation of electricity consumption;

[0024] The power consumption forecast demand result generating unit is used to obtain the power consumption forecast demand of the target area based on the full time domain propagation aggregation representation of the power consumption.

[0025] This application has significant technical effects due to the adoption of the above technical solutions:

[0026] The power dispatching method and system for distributed photovoltaic systems provided by the present application collects the historical power consumption data of the target area, and uses data analysis and processing technology based on deep learning to perform time series cutting and local time series feature encoding of the historical power consumption data, so as to automatically evaluate the power consumption forecast demand of the target area according to the spatiotemporal propagation aggregation representation of the local time series features of each historical power consumption data in the full time domain. In this way, it is possible to automatically learn and extract complex time series features from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, it is possible to capture the complex time series relationships and dynamic interactions between data, thereby improving the accuracy of prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 The present invention is a flowchart of a method for dispatching electric power for a distributed photovoltaic system according to an embodiment of the present application.

[0029] Figure 2 The present invention is a flowchart for determining the predicted electricity demand of a target area in a power dispatching method for a distributed photovoltaic system according to an embodiment of the present application.

[0030] Figure 3 A schematic diagram of data flow for determining predicted electricity demand in a target area in a power dispatching method for a distributed photovoltaic system according to an embodiment of the present application.

[0031] Figure 4 In a power dispatching method for a distributed photovoltaic system according to an embodiment of the present application, a flow chart is provided for performing propagation aggregation based on spatiotemporal potential energy on a set of local time series forward propagation representation vectors of power consumption to obtain a full time domain propagation aggregation representation of power consumption.

[0032] Figure 5 In the power dispatching method for distributed photovoltaic systems according to an embodiment of the present application, a flowchart is provided for performing energy propagation-based feature aggregation on the sequence of local time series forward propagation representation vectors of historical power consumption and the local time series forward propagation representation vector of current power consumption to obtain the full time domain propagation aggregation representation of the power consumption.

[0033] Figure 6 4 is a system block diagram of a power dispatching system for a distributed photovoltaic system according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0035] Distributed photovoltaic systems are solar power generation facilities close to the user end. Due to their small installed capacity and direct access to the user's power system, they are rapidly gaining popularity around the world. Patent CN118523285A proposes a power dispatching method and system for distributed photovoltaic systems, which ensures the stable operation of the power system by analyzing the target area equipment to predict power demand, collecting monitoring data, integrating power generation information, calculating power balance coefficients, and evaluating safety status.

[0036] Electricity demand forecasting is crucial for the management of power systems, especially in distributed photovoltaic systems, where accurate forecasting is particularly critical due to the uneven temporal and spatial distribution of electricity production and consumption. The above patent analyzes the historical electricity consumption data of the target area, classifies the power-consuming equipment, and establishes a forecasting model based on the power consumption patterns of the equipment and external factors to predict future electricity demand. However, this method based on the forecasting model has limitations because it relies on manual operations, including steps such as data cleaning and feature selection, which are not only time-consuming but also prone to errors. In addition, these methods are usually based on statistics or simple mathematical models, which make it difficult to capture complex nonlinear relationships and interactions between data features, thereby limiting the accuracy of electricity consumption forecasts.

[0037] Based on this, the present application proposes a power dispatching method for a distributed photovoltaic system. Figure 1 FIG. 1 is a flow chart of a method for dispatching electric power in a distributed photovoltaic system according to an embodiment of the present application. Figure 1 As shown, according to the embodiment of the present application, the power dispatching method for a distributed photovoltaic system includes: S110, determining the predicted power demand of a target area; S120, installing a power monitoring device in the target area to perform real-time monitoring and feedback on the target area based on the predicted power demand; S130, acquiring photovoltaic power generation data through a distributed photovoltaic power generation system; S140, determining the power balance coefficient of the target area based on power feedback information and the photovoltaic power generation data; S150, generating a judgment result based on the power balance coefficient, the judgment result being the safe power operation status of the target area, and performing power dispatch on the target area based on the judgment result.

[0038] It should be understood that the target area refers to a specific geographical area where power dispatching is required, which can be a city, an industrial park or a specific building complex. The predicted power demand refers to the power demand of the target area in the future, that is, the power that needs to be consumed in a specific time period. In this application, it is first necessary to determine the predicted power demand of the target area. Generally, the historical power consumption data of the target area in the past period of time (including daily or weekly power consumption, peak power consumption period, etc.) can be collected, and then the historical power consumption data in the target area can be analyzed to understand the power consumption law of the target area, and then the predicted power demand can be obtained. Then, according to the predicted power demand, a power monitoring device is installed in the target area to monitor and feedback the target area in real time. Here, the power monitoring device refers to a device for monitoring the operating status, power quality, fault detection and other aspects of the power system. The power feedback information of the target area can be collected by the power monitoring device, and the power feedback information refers to various information of the power system during operation, including current, voltage, power factor, power consumption, fault information, etc., which can reflect the operating status, power quality, faults and other conditions of the power system. For example, by monitoring current and voltage data, we can understand the load and stability of the power system; by monitoring power factor data, we can understand the reactive compensation and energy utilization efficiency of the power system; by monitoring power consumption data, we can understand the power demand and energy consumption of the power system. That is, by obtaining the power feedback information of the target area through the power monitoring device, we can timely discover the problems in the power system, and then provide an important reference for power dispatching. Then, the photovoltaic power generation data is obtained through the distributed photovoltaic power generation system. Specifically, photovoltaic power generation data refers to the power data emitted by the distributed photovoltaic power generation system, including power generation, power generation, power generation time, etc. The extraction of photovoltaic power generation data can be achieved by directly transmitting the data collected by sensors and monitoring equipment to the dispatching center, or directly grabbing data from the distributed photovoltaic power generation system through the software interface and sending it to the dispatching center. Next, based on the power feedback information and the photovoltaic power generation data, the power balance coefficient of the target area is determined. It should be noted that the power balance coefficient refers to the proportional relationship between the power demand and the photovoltaic power generation in the target area within a specific time period, reflecting the balance of power supply and demand in the target area, which can be a percentage or a specific value. Here, the collected power feedback information and photovoltaic power generation data are processed and analyzed to obtain the power balance coefficient of the target area. Finally, a judgment result is generated based on the power balance coefficient, the judgment result is the power safety operation status of the target area, and power dispatch is performed on the target area based on the judgment result.In detail, first, a reference value is set for the power balance coefficient as a reference for determining the balance of power supply and demand. If the power balance coefficient is close to the set reference value, it is considered that the power supply and demand in the target area is basically balanced, and the operation status of the power system is safe and reliable. If the power balance coefficient is less than the set reference value, it means that the power load is less than the power generation capacity, there is a certain amount of power waste, and it is necessary to adjust the power load or reduce the power generation capacity. If the power balance coefficient is greater than the set reference value, it means that the power load is greater than the power generation capacity, there is a certain amount of power shortage, and it is necessary to adjust the power generation power or limit the power load, thereby realizing reasonable power dispatch in the target area.

[0039] Accordingly, in determining the forecast electricity demand of the target area, the technical concept of the present application is to collect the historical electricity consumption data of the target area, and use data analysis and processing technology based on deep learning to perform time series cutting and local time series feature encoding of the historical electricity consumption data, so as to automatically evaluate the forecast electricity demand of the target area according to the spatiotemporal propagation aggregation representation of the local time series features of each historical electricity consumption data in the whole time domain. In this way, it is possible to automatically learn and extract complex time series features from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, it is possible to capture the complex time series relationships and dynamic interactions between data, thereby improving the accuracy of prediction.

[0040] Figure 2 The present invention is a flowchart for determining the predicted electricity demand of a target area in a power dispatching method for a distributed photovoltaic system according to an embodiment of the present application. Figure 3 The following is a schematic diagram of data flow for determining the predicted power demand of a target area in a power dispatching method for a distributed photovoltaic system according to an embodiment of the present application. Figure 2 and Figure 3 As shown, determining the predicted electricity demand of the target area includes: S111, collecting the historical electricity consumption data of the target area; S112, performing sequence cutting on the historical electricity consumption data to obtain a set of local historical electricity consumption data, and then performing local time series sequence encoding on the electricity consumption to obtain a set of local time series forward propagation representation vectors of electricity consumption; S113, performing propagation aggregation based on spatiotemporal potential energy on the set of local time series forward propagation representation vectors to obtain a full time domain propagation aggregation representation of electricity consumption; S114, obtaining the predicted electricity demand of the target area based on the full time domain propagation aggregation representation of electricity consumption.

[0041] In step S111, the historical electricity consumption data of the target area is collected. It should be understood that the historical electricity consumption data of the target area is the basis for prediction. By analyzing the past electricity consumption data, the patterns and trends of electricity consumption can be identified. These patterns include seasonal fluctuations, differences in electricity consumption during the day and at night, changes in electricity consumption during holidays, etc. In order to make the collected electricity consumption data more accurate and thus improve the accuracy of electricity consumption forecasting needs, it is necessary to collect historical electricity consumption data according to different times. It is worth noting that the historical electricity consumption data in this application is obtained based on the electrical equipment (including various types of electrical equipment, industrial equipment, etc.) in the target area. The electrical equipment is scattered in different locations in the target area, so the electricity consumption data will also be collected according to different geographical locations.

[0042] In step S112, the electricity consumption historical data is sequence-cut to obtain a set of local electricity consumption historical data, and then local electricity consumption time series sequence encoding is performed to obtain a set of local electricity consumption time series forward propagation representation vectors. Specifically, in an embodiment of the present application, the electricity consumption historical data is sequence-cut to obtain a set of local electricity consumption historical data, and then local electricity consumption time series sequence encoding is performed to obtain a set of local electricity consumption time series forward propagation representation vectors, including: sequence-cutting the electricity consumption historical data based on a predetermined time scale to obtain the set of local electricity consumption historical data; inputting each local electricity consumption historical data in the set of local electricity consumption historical data into a local electricity consumption time series sequence encoder based on a forward LSTM model to obtain a set of local electricity consumption time series forward propagation representation vectors.

[0043] It should be understood that the electricity consumption historical data is constantly changing with the passage of time, and the time series characteristics exhibited in different local time periods are different, such as short-term fluctuations or outliers. Based on this, in order to better capture and understand the changing laws and fluctuations of electricity consumption in different local time scales, in the technical solution of the present application, the electricity consumption historical data is sequenced based on a predetermined time scale to obtain a set of local electricity consumption historical data, so that the electricity consumption pattern and trend in a specific time period can be identified, providing important data support for the prediction and scheduling of power resources. Then, considering that each local electricity consumption historical data in the set of local electricity consumption historical data has dynamic change information in time series, based on this, in order to further capture and extract the time series characteristic information of electricity consumption in each specific local time period, in the technical solution of the present application, each local electricity consumption historical data in the set of local electricity consumption historical data is input into the local electricity consumption time series sequence encoder based on the forward LSTM model to obtain a set of local electricity consumption time series forward propagation representation vectors.

[0044] In step S113, the set of the local time series forward propagation representation vectors of the power consumption is subjected to propagation aggregation based on the space-time potential energy to obtain the full-time domain propagation aggregation representation of the power consumption. Accordingly, considering that the user's power demand is not only affected by time factors (such as diurnal changes and seasonal fluctuations), but also by spatial factors (such as the interaction between different geographical locations in the target area). Therefore, in order to capture the dynamic changes in time and the interdependence in space at the same time, so as to integrate the power consumption characteristics at different time points and different locations to fully reflect the changing laws of power demand, in the technical solution of the present application, the set of the local time series forward propagation representation vectors of the power consumption is subjected to propagation aggregation based on the space-time potential energy to obtain the full-time domain forward propagation aggregation representation vector of the power consumption as the full-time domain propagation aggregation representation of the power consumption.

[0045] In detail, first, the current node and the sequence of historical nodes are extracted from the set of the local time series forward propagation representation vector of the power consumption to obtain the sequence of the local time series forward propagation representation vector of the current power consumption and the local time series forward propagation representation vector of the historical power consumption. That is, the current node is the power demand feature at the current moment, reflecting the power consumption pattern and demand status at the current moment. The historical node can help identify the periodic pattern or seasonal changes in power demand, thereby providing a basis for subsequent feature propagation aggregation. Next, the static energy value of the power consumption of each historical power consumption local time series forward propagation representation vector is calculated to help evaluate the importance of different historical power consumption characteristics. For example, the static energy value in certain time periods is high, indicating that the power consumption pattern in this time period has a greater impact on the overall power demand. In particular, the static energy value of the power consumption is based on the maximum eigenvalue, mean and variance of each historical node, providing a numerical measure to measure the inherent attributes in space, and setting a reference standard for the decay of potential energy in time and space, which helps the model grasp the relative stability and influence of each node in space and time. Next, the power consumption energy space attenuation factor and the power consumption energy time attenuation factor are determined based on the space span coefficient and time span coefficient between the local time series forward propagation representation vector of each historical power consumption and the local time series forward propagation representation vector of the current power consumption. It should be understood that the space attenuation factor is used to describe how the influence of power demand weakens as the spatial distance increases. The time attenuation factor is used to describe how the influence of the historical power consumption pattern on the current power consumption pattern weakens over time, which enhances the model's sensitivity to the locality and timeliness of power consumption and makes information transmission more consistent with the actual physical process. Subsequently, based on the power consumption energy space attenuation factor and the power consumption energy time attenuation factor, the static energy value of the power consumption of each historical node is spatiotemporally modulated to obtain a sequence of the dynamic energy spatiotemporal transfer value of power consumption. That is, the spatiotemporal modulation can dynamically adjust the degree of influence of historical data, so that the prediction model can better adapt to the actual power demand changes at the current moment, and strengthen the model's ability to capture dynamic changes in time series. Then, the dynamic energy spatiotemporal transfer value of the power consumption of each historical node is input into the energy transfer gating module based on the mask function to obtain a sequence of dynamic energy spatiotemporal transfer weights of power consumption. In particular, the mask function is to filter out the most valuable historical data for the current power consumption forecast, highlight those historical data that are highly relevant to the current power consumption pattern, and ignore those historical data that are less relevant. The gating module dynamically adjusts the weight of historical data by controlling the flow of information, which means that some historical data may be given a greater weight, while other data will be suppressed or ignored.Afterwards, the sequence of the local time series forward propagation representation vector of historical power consumption is transferred and aggregated based on the masked weights to integrate different historical features and form an overall historical power consumption transfer aggregation modulation vector containing historical information, thereby capturing important features in the historical power consumption pattern. Finally, the historical modulation vector and the local time series forward propagation representation vector of current power consumption are added to obtain the full-time domain forward propagation aggregation representation vector of power consumption that integrates the historical and current full-time domains.

[0046] Specifically, Figure 4 The present invention is a flow chart of performing propagation aggregation based on spatiotemporal potential energy on the set of local time series forward propagation representation vectors of power consumption in the power dispatching method for distributed photovoltaic system according to the embodiment of the present application to obtain the full time domain propagation aggregation representation of power consumption. Figure 4 As shown, the set of the local time series forward propagation representation vectors of the electricity consumption is subjected to propagation aggregation based on spatiotemporal potential energy to obtain the full time domain propagation aggregation representation of the electricity consumption, including: S1131, extracting the current local time series forward propagation representation vector of the electricity consumption from the set of the local time series forward propagation representation vectors of the electricity consumption, and defining other local time series forward propagation representation vectors of the electricity consumption in the set of the local time series forward propagation representation vectors of the electricity consumption as historical local time series forward propagation representation vectors to obtain a sequence of historical local time series forward propagation representation vectors of the electricity consumption; S1132, performing feature aggregation based on energy propagation on the sequence of the historical local time series forward propagation representation vectors of the electricity consumption and the current local time series forward propagation representation vector to obtain the full time domain propagation aggregation representation of the electricity consumption.

[0047] More specifically, Figure 5 The present invention is a flow chart of performing energy propagation-based feature aggregation on the sequence of the local time series forward propagation representation vectors of the historical power consumption and the local time series forward propagation representation vectors of the current power consumption in the power dispatching method for the distributed photovoltaic system according to the embodiment of the present application to obtain the full time domain propagation aggregation representation of the power consumption. Figure 5 As shown, the sequence of the local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vectors of current electricity consumption are subjected to feature aggregation based on energy propagation to obtain the aggregated representation of the full time domain propagation of electricity consumption, including: S11321, calculating the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption in the sequence of the local time series forward propagation representation vectors of historical electricity consumption; S11322, based on the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption, feature message passing aggregation is performed on the sequence of the local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption to obtain the aggregated representation of the full time domain propagation of electricity consumption.

[0048] More specifically, in an embodiment of the present application, the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption in the sequence of the local time series forward propagation representation vector of historical electricity consumption is calculated, including: respectively calculating the mean and variance of the local time series forward propagation representation vector of historical electricity consumption to obtain the mean of the local time series characteristics of historical electricity consumption and the variance of the local time series characteristics of historical electricity consumption; subtracting the local time series forward propagation representation vector of historical electricity consumption from the mean of the local time series characteristics of historical electricity consumption by position, and calculating the static energy value of each local time series difference vector of historical electricity consumption. The fourth power of the position characteristic value is used to obtain the local time series differential modulation vector of historical electricity consumption; the expected value of the local time series differential modulation vector of historical electricity consumption is calculated, and the expected value is divided by the square of the variance of the local time series characteristic of historical electricity consumption to obtain the local time series energy value of historical electricity consumption; the maximum value of the local time series forward propagation representation vector of historical electricity consumption is extracted, and the maximum value of the local time series characteristic of historical electricity consumption is added to the local time series energy value of historical electricity consumption to obtain the static energy value of electricity consumption corresponding to the local time series forward propagation representation vector of historical electricity consumption.

[0049] More specifically, in an embodiment of the present application, based on the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption, the sequence of the local time series forward propagation representation vector of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption are subjected to feature message transmission aggregation to obtain the full-time domain propagation aggregation representation of electricity consumption, including: based on the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption, calculating the dynamic energy spatiotemporal transfer value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption relative to the local time series forward propagation representation vector of current electricity consumption; The dynamic energy spatiotemporal transfer values ​​of each historical electricity consumption local time series forward propagation representation vector are input into the energy transfer gating module based on the mask function to obtain a sequence of dynamic energy spatiotemporal transfer weights of electricity consumption; based on the sequence of dynamic energy spatiotemporal transfer weights of electricity consumption, the weighted sum of the sequence of historical electricity consumption local time series forward propagation representation vectors is calculated to obtain the historical electricity consumption transfer aggregation modulation vector; the positional sum of the historical electricity consumption transfer aggregation modulation vector and the current electricity consumption local time series forward propagation representation vector is calculated to obtain the full-time domain forward propagation aggregation representation vector of electricity consumption as the full-time domain transmission aggregation representation of electricity consumption.

[0050] More specifically, in an embodiment of the present application, based on the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption, the dynamic energy spatiotemporal transfer value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption relative to the local time series forward propagation representation vector of current electricity consumption is calculated, including: determining the spatial attenuation factor of electricity consumption energy based on the spatial span coefficient between each local time series forward propagation representation vector of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption; determining the temporal attenuation factor of electricity consumption energy based on the temporal span coefficient between each local time series forward propagation representation vector of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption; and performing spatiotemporal modulation on the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption based on the spatial attenuation factor of electricity consumption and the temporal attenuation factor of electricity consumption to obtain the dynamic energy spatiotemporal transfer value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption relative to the local time series forward propagation representation vector of current electricity consumption.

[0051] More specifically, in an embodiment of the present application, based on the spatial span coefficient between the local time series forward propagation representation vector of each historical power consumption and the local time series forward propagation representation vector of the current power consumption, the power consumption energy spatial attenuation factor is determined, including: calculating the square of the spatial span value between the local time series forward propagation representation vector of the historical power consumption and the local time series forward propagation representation vector of the current power consumption to obtain the power consumption spatial span coefficient; dividing the power consumption static energy value corresponding to the local time series forward propagation representation vector of the historical power consumption by the power consumption spatial span coefficient to obtain the power consumption energy spatial attenuation factor.

[0052] More specifically, in an embodiment of the present application, based on the time span coefficient between the local time series forward propagation representation vectors of each historical power consumption and the local time series forward propagation representation vector of the current power consumption, the power consumption energy time decay factor is determined, including: subtracting the power consumption timestamp of the current power consumption local time series forward propagation representation vector from the power consumption timestamp of the historical power consumption local time series forward propagation representation vector and rounding down to obtain the power consumption time span value; dividing the power consumption time span value by the time decay parameter to obtain the power consumption time series span energy decay coefficient; using the power consumption time series span energy decay coefficient as an exponential power, calculating the natural exponential function value with the natural constant e as the base to obtain the power consumption time span coefficient; dividing the power consumption static energy value corresponding to the historical power consumption local time series forward propagation representation vector by the power consumption time span coefficient to obtain the power consumption energy time decay factor.

[0053] More specifically, in an embodiment of the present application, based on the spatial attenuation factor of the electricity consumption energy and the temporal attenuation factor of the electricity consumption energy, the static energy value of the electricity consumption of each historical electricity consumption local time series forward propagation representation vector is temporally modulated to obtain the dynamic energy spatiotemporal transfer value of the electricity consumption of each historical electricity consumption local time series forward propagation representation vector relative to the current electricity consumption local time series forward propagation representation vector, including: multiplying the spatial attenuation factor of the electricity consumption energy and the temporal attenuation factor of the electricity consumption energy of the historical electricity consumption local time series forward propagation representation vector with the first modulation parameter and the second modulation parameter respectively, and adding the obtained spatial attenuation modulation factor of the electricity consumption energy and the temporal attenuation modulation factor of the electricity consumption energy to obtain the dynamic energy spatiotemporal transfer value of the electricity consumption.

[0054] More specifically, in an embodiment of the present application, the dynamic spatiotemporal energy transfer values ​​of the local time series forward propagation representation vectors of each historical power consumption are input into an energy transfer gating module based on a mask function to obtain a sequence of dynamic spatiotemporal energy transfer weights of power consumption, including: in response to the dynamic spatiotemporal energy transfer value of power consumption being greater than a predetermined threshold, the dynamic spatiotemporal energy transfer value of power consumption is input into a sigmoid function to obtain a value as the dynamic spatiotemporal energy transfer weight of power consumption.

[0055] In the embodiment of the present application, specifically, the set of the forward propagation representation vectors of the local time series of the power consumption is subjected to propagation aggregation based on the spatiotemporal potential energy, which can be expressed by the formula:

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] in, is a set of vectors representing the local time series forward propagation of power consumption, are the first, second, ..., and third vectors in the set of the local time series forward propagation representation vectors of the power consumption. ,..., The local time series forward propagation representation vector of power consumption, It is The local time series of power consumption is forward propagated to represent the eigenvalues ​​of each position in the vector. To calculate the expected value, and They are The mean and variance of For extraction The maximum value in yes The corresponding static energy value of electricity consumption, express and The space span value between yes and The energy spatial attenuation factor of electricity consumption between and Distribution represents and The timestamp of the local time series of power consumption forward propagation vector, To round down, is the time decay parameter, represents the exponential function value with the natural constant e as the base, yes and The energy time decay factor of the electricity consumption between and is the corresponding modulation parameter, yes and The dynamic energy space-time transfer value of electricity consumption between For masking, yes function, is a predetermined threshold, is the number of vectors in the set of vectors representing the local time series forward propagation of power consumption, It is the aggregate representation vector of the power consumption in the entire time domain forward propagation.

[0064] In step S114, based on the full-time domain propagation aggregation representation of the electricity consumption, the electricity demand forecast of the target area is obtained. Specifically, in an embodiment of the present application, based on the full-time domain propagation aggregation representation of the electricity consumption, the electricity demand forecast of the target area is obtained, including: inputting the full-time domain forward propagation aggregation representation vector of the electricity consumption into a decoder-based electricity demand predictor to obtain the electricity demand forecast of the target area. That is, the set of the local time series forward propagation representation vectors of the electricity consumption is subjected to a decoding process based on the full-time domain propagation aggregation representation of the electricity consumption obtained by propagation aggregation based on spatiotemporal potential energy, so as to automatically evaluate the electricity demand forecast of the target area. In this way, complex time series features can be automatically learned and extracted from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, the complex time series relationships and dynamic interactions between data can be captured, thereby improving the accuracy of the prediction.

[0065] In particular, in the technical solution of the present application, each local time series forward propagation representation vector of the power consumption in the set of local time series forward propagation representation vectors respectively represents the time series implicit correlation characteristics of the local historical data of the power consumption. Therefore, when the set of the local time series forward propagation representation vectors of the power consumption is subjected to propagation aggregation based on spatiotemporal potential energy, it is considered that each local historical data of the power consumption in the set of the local historical data of the power consumption has time series distribution differences and possible noise information in the data source domain, which makes the local time series forward propagation representation vectors in the set of the local time series forward propagation representation vectors of the power consumption have more significant feature distribution heterogeneity, which will cause weight mismatch and distortion when determining the spatiotemporal transfer weight when performing propagation aggregation based on spatiotemporal potential energy, so that the obtained full-time domain forward propagation aggregation representation vector of the power consumption has spatiotemporal aggregation differences due to the difference in spatiotemporal transfer weights, so it is expected to improve the detail semantic aggregation expression effect of the full-time domain forward propagation aggregation representation vector of the power consumption based on the difference in time series semantic representation.

[0066] Preferably, the whole-time-domain forward propagation aggregation representation vector of the electricity consumption is input into the decoder-based electricity demand predictor to obtain the predicted electricity demand of the target area, including: calculating the sum of the absolute values ​​of each eigenvalue of the whole-time-domain forward propagation aggregation representation vector of the electricity consumption to obtain the first whole-time-domain forward propagation aggregation and modulation value of the electricity consumption, and calculating the square root of the sum of the squares of each eigenvalue of the whole-time-domain forward propagation aggregation representation vector of the electricity consumption to obtain the second whole-time-domain forward propagation aggregation and modulation value of the electricity consumption; after point-wise subtraction of the whole-time-domain forward propagation aggregation representation vector of the electricity consumption and the second whole-time-domain forward propagation aggregation and modulation value of the electricity consumption, respectively, point-wise multiplication is performed with the number of eigenvalues ​​of the whole-time-domain forward propagation aggregation representation vector of the electricity consumption and the reciprocal of the first whole-time-domain forward propagation aggregation and modulation value of the electricity consumption, and the reciprocal of each eigenvalue is taken to obtain the first whole-time-domain forward propagation aggregation and modulation value of the electricity consumption. forward propagation aggregation phase conversion vector; after subtracting the full-time domain forward propagation aggregation representation vector of the power consumption from the first full-time domain forward propagation aggregation and modulation value, respectively perform dot multiplication with the square root of the number of eigenvalues ​​of the full-time domain forward propagation aggregation representation vector of the power consumption and the inverse of the second full-time domain forward propagation aggregation and modulation value of the power consumption, and take the inverse of each eigenvalue to obtain the second full-time domain forward propagation aggregation phase conversion vector of the power consumption; subtract the dot product vector of the second full-time domain forward propagation aggregation phase conversion vector of the first full-time domain forward propagation aggregation phase conversion vector from the dot product vector of the weighted hyperparameter to obtain the optimized full-time domain forward propagation aggregation representation vector of the power consumption; input the optimized full-time domain forward propagation aggregation representation vector of the power consumption into the decoder-based power demand predictor to obtain the predicted power demand of the target area.

[0067] Here, the optimized representation of the full-time domain forward propagation aggregation representation vector of the power consumption is:

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074] in, is the total time domain forward propagation aggregation representation vector of the power consumption, represents the first The eigenvalues ​​at the positions, represents the number of eigenvalues ​​of the forward propagation aggregation representation vector of the power consumption in the entire time domain, is the first power consumption full time domain forward propagation aggregation and modulation value, is the second power consumption full time domain forward propagation aggregation and modulation value, It means point multiplication by position. It means to subtract by position point. To take the inverse of each eigenvalue of a vector, represents the first power consumption full time domain forward propagation aggregate phase conversion vector, represents the second power consumption full time domain forward propagation aggregate phase conversion vector, represents the weighted hyperparameter, A full-time domain forward propagation aggregation representation vector representing the optimized power consumption.

[0075] In this preferred embodiment, the difference of the characteristic value of the full-time domain forward propagation aggregation representation vector relative to the vector feature set of the full-time domain forward propagation aggregation representation vector is used as the semantic change intensity information, and the phase-like conversion corresponding to the position-based intensity modulation is performed through different and modulation representation forms, so that the aggregation enhancement of the semantic change phase perception can improve the axial aggregation receptive field along the feature aggregation direction by performing a spatial translation operation based on alternating stacking under the vector set scale balance of the full-time domain forward propagation aggregation representation vector of the power consumption. , So as to improve the perception effect of the aggregation semantics of the full-time domain forward propagation aggregation representation vector for the semantic changes of the details of the time series, so as to improve the expression effect of the full-time domain forward propagation aggregation representation vector of the power consumption, and improve the accuracy of the power demand forecasting demand of the target area obtained by inputting it into the power demand forecaster based on the decoder. In this way, it is possible to automatically learn and extract complex time series features from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, it is possible to capture the complex time series relationship and dynamic interaction between data, thereby improving the accuracy of prediction.

[0076] In summary, the power dispatching method for distributed photovoltaic systems based on the embodiment of the present application is explained, which collects the historical power consumption data of the target area, and uses data analysis and processing technology based on deep learning to perform time series cutting and local time series feature encoding of the historical power consumption data, so as to automatically evaluate the power consumption forecast demand of the target area according to the spatiotemporal propagation aggregation representation of the local time series features of each historical power consumption data in the whole time domain. In this way, it is possible to automatically learn and extract complex time series features from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, it is possible to capture the complex time series relationships and dynamic interactions between data, thereby improving the accuracy of prediction.

[0077] Figure 6 1 is a system block diagram of a power dispatching system for a distributed photovoltaic system according to an embodiment of the present application. Figure 6 As shown, according to an embodiment of the present application, the power dispatching system 100 for a distributed photovoltaic system includes: a power forecast demand determination module 110, which is used to determine the power forecast demand of the target area; an actual power demand data monitoring module 120, which is used to install a power monitoring device in the target area according to the power forecast demand to perform real-time monitoring and feedback on the target area; a photovoltaic power generation data acquisition module 130, which is used to obtain photovoltaic power generation data through a distributed photovoltaic power generation system; a power balance coefficient determination module 140, which is used to determine the power balance coefficient of the target area based on power feedback information and the photovoltaic power generation data; a power dispatching module 150, which is used to generate a judgment result based on the power balance coefficient, the judgment result is the power safety operation status of the target area, and the target area is dispatched based on the judgment result; wherein the power forecast demand determination module includes: a power consumption history data acquisition unit, which is used to collect the power consumption history data of the target area; a power consumption history data time series sequence encoding unit, which is used to perform sequence cutting on the power consumption history data to obtain a set of local power consumption history data and then perform local power consumption time series Sequence coding to obtain a set of local time series forward propagation representation vectors of power consumption; a power consumption full time domain propagation aggregation representation generation unit, which is used to perform propagation aggregation based on spatiotemporal potential energy on the set of local time series forward propagation representation vectors of power consumption to obtain a full time domain propagation aggregation representation of power consumption, including: a current power consumption and historical power consumption local time series forward propagation representation division subunit, which is used to extract the current power consumption local time series forward propagation representation vector from the set of local time series forward propagation representation vectors of power consumption, and to separate other power consumption local time series forward propagation representation vectors in the set of local time series forward propagation representation vectors of power consumption. The time series forward propagation representation vector is defined as the local time series forward propagation representation vector of historical electricity consumption to obtain a sequence of local time series forward propagation representation vectors of historical electricity consumption; the current electricity consumption historical electricity consumption local time series feature aggregation subunit is used to perform energy propagation-based feature aggregation on the sequence of local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption to obtain the full time domain propagation aggregation representation of electricity consumption; the electricity consumption forecast demand result generation unit is used to obtain the electricity consumption forecast demand of the target area based on the full time domain propagation aggregation representation of electricity consumption.

[0078] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the power dispatching system 100 for a distributed photovoltaic system have been described in detail above. Figures 1 to 5The description of the power dispatching method for a distributed photovoltaic system has been described in detail, and therefore, its repeated description will be omitted.

[0079] In summary, the power dispatching system 100 for distributed photovoltaic systems based on the embodiment of the present application is explained, which collects the historical power consumption data of the target area, and uses data analysis and processing technology based on deep learning to perform time series cutting and local time series feature encoding of the historical power consumption data, so as to automatically evaluate the power consumption forecast demand of the target area according to the spatiotemporal propagation aggregation representation of the local time series features of each historical power consumption data in the whole time domain. In this way, it is possible to automatically learn and extract complex time series features from a large amount of historical data, reducing the tedious work and possibility of errors in manual feature engineering. At the same time, it is possible to capture the complex time series relationships and dynamic interactions between data, thereby improving the accuracy of the prediction.

[0080] As described above, the power dispatching system 100 for a distributed photovoltaic system according to an embodiment of the present application can be implemented in various wireless terminals, such as a server for power dispatching of a distributed photovoltaic system. In one example, the power dispatching system 100 for a distributed photovoltaic system according to an embodiment of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the power dispatching system 100 for a distributed photovoltaic system can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the power dispatching system 100 for a distributed photovoltaic system can also be one of the many hardware modules of the wireless terminal.

[0081] Alternatively, in another example, the power dispatching system 100 for a distributed photovoltaic system and the wireless terminal may also be separate devices, and the power dispatching system 100 for a distributed photovoltaic system may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

Claims

1. A power dispatching method for a distributed photovoltaic system, comprising: Determine the forecast electricity demand in the target area; According to the predicted power demand, a power monitoring device is installed in the target area to perform real-time monitoring and feedback on the target area; photovoltaic power generation data is obtained through a distributed photovoltaic power generation system; Based on the power feedback information and the photovoltaic power generation data, determine the power balance coefficient of the target area; generate a judgment result based on the power balance coefficient, the judgment result is the power safety operation status of the target area, and perform power dispatch on the target area based on the judgment result; characterized in that determining the predicted power demand of the target area includes: Collecting historical electricity consumption data of the target area; The power consumption historical data is sequence-cut to obtain a set of local power consumption historical data, and then local power consumption time series sequence encoding is performed to obtain a set of local power consumption time series forward propagation representation vectors; The set of the local time series forward propagation representation vectors of the power consumption is subjected to propagation aggregation based on spatiotemporal potential energy to obtain the full-time domain propagation aggregation representation of the power consumption, including: extracting the current local time series forward propagation representation vector of the power consumption from the set of the local time series forward propagation representation vectors of the power consumption, and defining other local time series forward propagation representation vectors of the power consumption in the set of the local time series forward propagation representation vectors of the power consumption as the historical local time series forward propagation representation vectors to obtain a sequence of the historical local time series forward propagation representation vectors of the power consumption; and performing feature aggregation based on energy propagation on the sequence of the historical local time series forward propagation representation vectors of the power consumption and the current local time series forward propagation representation vector of the power consumption to obtain the full-time domain propagation aggregation representation of the power consumption; Based on the full-time domain propagation aggregation representation of the power consumption, the predicted power demand of the target area is obtained; The power consumption historical data is sequence cut to obtain a set of local power consumption historical data, and then local power consumption time series sequence encoding is performed to obtain a set of local power consumption time series forward propagation representation vectors, including: Sequentially cutting the power consumption historical data based on a predetermined time scale to obtain a set of the power consumption local historical data; Input each local historical data of electricity consumption in the set of local historical data of electricity consumption into the local time series encoder of electricity consumption based on the forward LSTM model to obtain a set of forward propagation representation vectors of the local time series of electricity consumption; The sequence of the local time series forward propagation representation vectors of the historical power consumption and the local time series forward propagation representation vector of the current power consumption are subjected to feature aggregation based on energy propagation to obtain the full time domain propagation aggregation representation of the power consumption, including: Calculating the static energy value of electricity consumption of each local time series forward propagation representation vector of historical electricity consumption in the sequence of local time series forward propagation representation vectors of historical electricity consumption; Based on the static energy value of the power consumption of each historical power consumption local time series forward propagation representation vector, the sequence of the historical power consumption local time series forward propagation representation vector and the current power consumption local time series forward propagation representation vector are aggregated by feature message passing to obtain the full time domain propagation aggregation representation of the power consumption; Calculating the static energy value of the power consumption of each local time series forward propagation representation vector of the historical power consumption in the sequence of the local time series forward propagation representation vector of the historical power consumption includes: Respectively calculate the mean and variance of the local time series forward propagation representation vector of the historical electricity consumption to obtain the mean of the local time series characteristics of the historical electricity consumption and the variance of the local time series characteristics of the historical electricity consumption; subtract the local time series forward propagation representation vector of the historical electricity consumption from the mean of the local time series characteristics of the historical electricity consumption by position, calculate the fourth power of the characteristic value of each position in the local time series difference vector of the historical electricity consumption to obtain the local time series difference modulation vector of the historical electricity consumption; calculate the expected value of the local time series difference modulation vector of the historical electricity consumption, and divide the obtained expected value by the square of the variance of the local time series characteristics of the historical electricity consumption to obtain the local time series energy value of the historical electricity consumption; Extract the maximum value of the local time series forward propagation representation vector of the historical power consumption, and add the obtained maximum value of the local time series feature of the historical power consumption to the local time series energy value of the historical power consumption to obtain the static energy value of the power consumption corresponding to the local time series forward propagation representation vector of the historical power consumption; Based on the static energy value of the power consumption of each historical power consumption local time series forward propagation representation vector, the sequence of the historical power consumption local time series forward propagation representation vector and the current power consumption local time series forward propagation representation vector are aggregated by feature message passing to obtain the full time domain propagation aggregation representation of the power consumption, including: Based on the static energy value of the power consumption of each historical power consumption local time series forward propagation representation vector, calculate the dynamic energy spatiotemporal transfer value of the power consumption of each historical power consumption local time series forward propagation representation vector relative to the current power consumption local time series forward propagation representation vector; Inputting the power consumption dynamic energy spatiotemporal transfer value of each historical power consumption local time series forward propagation representation vector into the energy transfer gating module based on the mask function to obtain a sequence of power consumption dynamic energy spatiotemporal transfer weights; Based on the sequence of the dynamic energy spatiotemporal transfer weights of the power consumption, the weighted sum of the sequence of the local time series forward propagation representation vectors of the historical power consumption is calculated to obtain the historical power consumption transfer aggregation modulation vector; Calculate the positional sum of the historical power consumption transfer aggregation modulation vector and the current power consumption local time series forward propagation representation vector to obtain the power consumption full time domain forward propagation aggregation representation vector as the power consumption full time domain propagation aggregation representation; Based on the static energy value of the power consumption of each historical power consumption local time series forward propagation representation vector, the dynamic energy spatiotemporal transfer value of the power consumption of each historical power consumption local time series forward propagation representation vector relative to the current power consumption local time series forward propagation representation vector is calculated, including: Determine the energy spatial attenuation factor of power consumption based on the spatial span coefficient between the local time series forward propagation representation vector of each historical power consumption and the local time series forward propagation representation vector of the current power consumption; Determine the energy time decay factor of the power consumption based on the time span coefficient between the local time series forward propagation representation vector of each historical power consumption and the local time series forward propagation representation vector of the current power consumption; Based on the power consumption energy spatial attenuation factor and the power consumption energy temporal attenuation factor, the power consumption static energy value of each historical power consumption local time series forward propagation representation vector is temporally modulated to obtain the power consumption dynamic energy spatiotemporal transfer value of each historical power consumption local time series forward propagation representation vector relative to the current power consumption local time series forward propagation representation vector; Based on the spatial span coefficient between the local time series forward propagation representation vectors of each historical power consumption and the local time series forward propagation representation vector of the current power consumption, determining the power consumption energy spatial attenuation factor includes: Calculate the square of the spatial span value between the local time series forward propagation representation vector of the historical power consumption and the local time series forward propagation representation vector of the current power consumption to obtain the power consumption spatial span coefficient; Divide the power consumption static energy value corresponding to the historical power consumption local time series forward propagation representation vector by the power consumption spatial span coefficient to obtain the power consumption energy spatial attenuation factor; Based on the time span coefficient between the local time series forward propagation representation vector of each historical power consumption and the local time series forward propagation representation vector of the current power consumption, determining the power consumption energy time attenuation factor includes: The power consumption timestamp of the current power consumption local time series forward propagation representation vector is subtracted from the power consumption timestamp of the historical power consumption local time series forward propagation representation vector, and then rounded down to obtain a power consumption time span value; Divide the power consumption time span value by the time decay parameter to obtain the power consumption time series span energy decay coefficient; Taking the energy attenuation coefficient of the power consumption time series span as the exponential power, calculating the value of the natural exponential function with the natural constant e as the base to obtain the power consumption time span coefficient; The power consumption static energy value corresponding to the historical power consumption local time series forward propagation representation vector is divided by the power consumption time span coefficient to obtain the power consumption energy time attenuation factor.

2. The power dispatching method for a distributed photovoltaic system according to claim 1, characterized in that: Based on the full-time domain propagation aggregation representation of the electricity consumption, the electricity consumption forecast demand of the target area is obtained, including: inputting the full-time domain forward propagation aggregation representation vector of the electricity consumption into a decoder-based electricity consumption forecaster to obtain the electricity consumption forecast demand of the target area.

3. A power dispatching system for a distributed photovoltaic system, used to execute the power dispatching method for a distributed photovoltaic system according to claim 1, characterized in that: include: A power consumption forecast demand determination module, used to determine the power consumption forecast demand of the target area; An actual power demand data monitoring module, used to install a power monitoring device in the target area according to the predicted power demand to monitor and feedback the target area in real time; a photovoltaic power generation data acquisition module, used to obtain photovoltaic power generation data through a distributed photovoltaic power generation system; a power balance coefficient determination module, used to determine the power balance coefficient of the target area based on power feedback information and the photovoltaic power generation data; A power dispatching module, used to generate a judgment result based on the power balance coefficient, the judgment result is the power safety operation status of the target area, and perform power dispatching for the target area based on the judgment result; characterized in that the power consumption forecast demand determination module includes: A power consumption history data acquisition unit, used to collect the power consumption history data of the target area; A power consumption historical data time series encoding unit, used for performing sequence cutting on the power consumption historical data to obtain a set of local power consumption historical data, and then performing local power consumption time series encoding to obtain a set of local power consumption time series forward propagation representation vectors; A unit for generating a full-time-domain propagation aggregation representation of electricity consumption, used for performing propagation aggregation based on spatiotemporal potential energy on the set of the local time series forward propagation representation vectors of the electricity consumption to obtain a full-time-domain propagation aggregation representation of electricity consumption, including: a subunit for dividing the local time series forward propagation representation of current electricity consumption and historical electricity consumption, used for extracting the local time series forward propagation representation vector of current electricity consumption from the set of the local time series forward propagation representation vectors of electricity consumption, and defining other local time series forward propagation representation vectors of electricity consumption in the set of the local time series forward propagation representation vectors of electricity consumption as the local time series forward propagation representation vectors of historical electricity consumption to obtain a sequence of local time series forward propagation representation vectors of historical electricity consumption; a subunit for aggregating the characteristics of the local time series of current electricity consumption and historical electricity consumption, used for performing feature aggregation based on energy propagation on the sequence of the local time series forward propagation representation vectors of historical electricity consumption and the local time series forward propagation representation vector of current electricity consumption to obtain the full-time-domain propagation aggregation representation of electricity consumption; The power consumption forecast demand result generating unit is used to obtain the power consumption forecast demand of the target area based on the full time domain propagation aggregation representation of the power consumption.

Images