A power distribution network power situation awareness prediction method and prediction system

Abstract

The present invention relates to the field of artificial intelligence technology, and discloses a power situation perception prediction method and prediction system for a distribution network, including: calculating multidimensional power data of each node in the distribution network according to electrical data; extracting multidimensional power features corresponding to the multidimensional power data, and analyzing the power operation state of the distribution network through the multidimensional power features; using a time series model to perceive the power time series, and obtaining a situation prediction value of the distribution network; updating the power time series within a moving time window, determining the power situation triggering condition of the distribution network according to the updated power time series, and updating the model parameters in the time series model according to the power situation triggering condition; optimizing the situation prediction value using the updated time series model, analyzing the power situation perception of the distribution network using the situation optimization value, and obtaining a power situation perception prediction result. The present invention can improve the accuracy of the power situation perception prediction of the distribution network.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a power situation awareness prediction method and prediction system for a distribution network. Background Art

[0002] With the development of social economy, the types of loads connected to the distribution network are becoming increasingly diverse. In addition to traditional residential, commercial and industrial loads, a large number of new loads such as electric vehicle charging piles and distributed energy have emerged. The power consumption characteristics of these loads are complex and are affected by many factors, such as residents' electricity consumption habits, weather changes, economic activities, etc., which leads to a significant increase in load uncertainty and brings huge challenges to the power prediction and operation management of the distribution network.

[0003] The existing power situation awareness prediction technology of distribution network predicts future power values ​​by analyzing historical power data and establishing a time series model. In practical applications, it is assumed that the power data is stable, which has poor adaptability to the complex and changeable power characteristics of modern distribution networks and is difficult to accurately capture sudden changes and nonlinear changes in power, resulting in low accuracy in power situation awareness prediction of distribution networks. Summary of the invention

[0004] The present invention provides a power distribution network power situation awareness prediction method and prediction system, the main purpose of which is to solve the problem of low accuracy when making product recommendations.

[0005] To achieve the above object, the present invention provides a power distribution network power situation awareness prediction method, comprising: Collecting electrical data corresponding to each node in the distribution network, and calculating multi-dimensional power data of each node in the distribution network based on the electrical data; Extracting multidimensional power features corresponding to the multidimensional power data, and analyzing the power operation state of the distribution network through the multidimensional power features; Constructing a power time series of the distribution network according to the power operation state, and using a pre-constructed time series model to perform situation awareness on the power time series to obtain a situation prediction value of the distribution network; The power time series is updated within a preset moving time window, a power situation triggering condition of the distribution network is determined according to the updated power time series, and a model parameter in the time series model is updated according to the power situation triggering condition; The situation prediction value is optimized using the updated time series model to obtain a situation optimization value, and the power situation awareness of the distribution network is analyzed using the situation optimization value to obtain a power situation awareness prediction result.

[0006] Optionally, extracting the multi-dimensional power features corresponding to the multi-dimensional power data includes: Divide the multi-dimensional power data of each node into power data sets according to the same dimension; Extracting time domain features and frequency domain features from the power data set; converting the power data set into a sequence set, and extracting sequence pattern features of the sequence set; The time domain features, the frequency domain features and the sequence pattern features are aggregated into a multi-dimensional power feature.

[0007] Optionally, analyzing the power operation state of the distribution network by using the multi-dimensional power characteristics includes: Constructing an operation state matrix according to the power feature dimension in the multi-dimensional power feature and the power operation state dimension of the distribution network acquired in advance; Determining a power operation state value of the distribution network by using the multi-dimensional power characteristics and the multi-dimensional power history characteristics acquired in advance; Filling the operation state matrix according to the power operation state value; The populated operation state matrix is ​​used to calculate the power operation state quantization value of the distribution network, where the power operation state quantization value calculation formula is: ; in, is the quantized value of the power operation state, is the state standard value, is the first Line The matrix values ​​of the columns, is the row dimension in the running status matrix, is the column dimension in the running status matrix, is the number of nodes in the distribution network, is the node number in the distribution network; The power operation state of the power distribution network is determined according to the power operation state quantization value.

[0008] Optionally, constructing a power time series of a power distribution network according to the power operation state includes: Extracting the operation time point corresponding to the power operation state; Mapping the operation time point with the power operation state to obtain a mapping sequence; The mapping sequence is converted into a power time series of the power distribution network.

[0009] Optionally, the using a pre-built time series model to perform situation awareness on the power time series to obtain a situation prediction value of the distribution network includes: converting the power time series into model sequence input data; Extract model parameters of pre-built time series models; Initializing the hidden state and cell state of the time series model; Calculating the target hidden state corresponding to the model sequence input data using the initialized time series model and the model parameters; Output data is determined according to the target hidden state, and a linear transformation is performed on the output data to obtain a state prediction value of the distribution network.

[0010] Optionally, determining a power situation triggering condition of the distribution network according to the updated power time series includes: Detect mutation factors in the updated power time series; Determining a triggering factor of the distribution network according to the mutation factor; The power situation triggering condition of the power distribution network is generated by the triggering factor.

[0011] Optionally, updating the model parameters in the time series model according to the power situation triggering condition includes: Collecting power sequence data within a preset time period according to the power situation trigger condition; Extracting the data change degree and data change direction in the power sequence data; Determining an adjustment factor according to the degree of change of the data and the direction of change of the data; The learning rate in the time series model is updated by the adjustment factor, wherein the learning rate update formula is: ; in, For the The learning rate of the iteration, is the minimum value of the learning rate, is the maximum value of the learning rate, is the current iteration number, is the total number of iterations in one cycle, is a constant, is the cosine function, is the regulating factor; The updated model parameters in the time series model are determined by the learning rate.

[0012] Optionally, the optimizing the situation prediction value by using the updated time series model to obtain the situation optimization value includes: Using the updated time series model to perform situation awareness on the power time series, and obtain a situation update prediction value of the distribution network; Calculating an error value between the situation update prediction value and the situation prediction value; When the error value is less than or equal to a preset error threshold, the situation prediction value is used as the situation optimization value; When the error value is greater than a preset error threshold, the situation update prediction value is used as the situation optimization value.

[0013] Optionally, the using the situation optimization value to analyze the power situation awareness of the distribution network to obtain a power situation awareness prediction result includes: Extract the target power situation value at the target time; Comparing the situation optimization value with the target power situation value to obtain a comparison result; Determine the power situation awareness level of the distribution network according to the comparison result; The power situation awareness prediction result is determined according to the awareness level.

[0014] In order to solve the above problems, the present invention also provides a power distribution network power situation awareness prediction system, the system comprising: A multi-dimensional power data calculation module, used to collect electrical data corresponding to each node in the distribution network, and calculate the multi-dimensional power data of each node in the distribution network according to the electrical data; A power operation status analysis module, used to extract the multi-dimensional power features corresponding to the multi-dimensional power data, and analyze the power operation status of the distribution network through the multi-dimensional power features; A situation prediction value analysis module is used to construct a power time series of the distribution network according to the power operation state, and use a pre-constructed time series model to perform situation perception on the power time series to obtain a situation prediction value of the distribution network; A model parameter updating module, used to update the power time series within a preset moving time window, determine the power situation triggering condition of the distribution network according to the updated power time series, and update the model parameters in the time series model according to the power situation triggering condition; The power situation awareness analysis module is used to optimize the situation prediction value using the updated time series model to obtain the situation optimization value, and use the situation optimization value to analyze the power situation awareness of the distribution network to obtain the power situation awareness prediction result.

[0015] The embodiment of the present invention collects electrical data of each node and calculates multi-dimensional power data, which can comprehensively cover different aspects of power distribution network; constructs power time series according to power operation status, and uses pre-constructed time series model for situation awareness, which can effectively capture the change law and trend of power over time; updates the power time series within the preset moving time window, so that the system can track the latest changes of power distribution network in real time. According to the updated time series, the power situation triggering condition is determined, and the abnormal fluctuation or mutation of power can be captured in time; the parameters of the time series model are updated based on the power situation triggering condition, so that the model can dynamically adjust itself according to the real-time operation status of the distribution network, and always maintain the ability to accurately predict the power situation; the situation prediction value is optimized using the updated time series model, and by comparing the predicted value before and after the update with the actual power situation value, a better situation optimization value is selected, which effectively corrects the prediction deviation, improves the matching degree between the predicted value and the actual power situation, and makes the prediction result more accurate and reliable. Therefore, the power situation perception prediction method and prediction system of the distribution network proposed by the present invention can solve the problem of low accuracy when making product recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a flow chart of a power distribution network power situation awareness and prediction method provided by an embodiment of the present invention; Figure 2 A schematic diagram of a process for extracting multi-dimensional power features provided by an embodiment of the present invention; Figure 3 A schematic diagram of a process for constructing a power time series according to an embodiment of the present invention; Figure 4 A functional module diagram of a power distribution network power situation awareness and prediction system provided by an embodiment of the present invention; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0017] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0018] The embodiment of the present application provides a distribution network power situation awareness prediction method. The execution subject of the distribution network power situation awareness prediction method includes but is not limited to at least one of the electronic devices such as the server, the terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the distribution network power situation awareness prediction method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0019] Reference Figure 1 FIG. 1 is a flow chart of a power distribution network power situation awareness prediction method provided by an embodiment of the present invention. In this embodiment, the power distribution network power situation awareness prediction method includes: S1. Collect electrical data corresponding to each node in the distribution network, and calculate multi-dimensional power data of each node in the distribution network based on the electrical data.

[0020] In the embodiment of the present invention, the electrical data refers to electrical parameters, which include but are not limited to the effective value, phase, and frequency of voltage; the effective value and phase of current; active power, reactive power, apparent power, and power factor.

[0021] In detail, the electrical data of each node in the distribution network can be collected through the sensors configured at each node in the distribution network. In the distribution network, the outgoing line terminal of the substation, the branch point of the feeder, the access point of important users and the access point of the distributed power source are all key collection nodes. The current transformer and the voltage transformer are used to accurately measure the current and voltage values ​​of high-voltage and high-current lines to provide accurate data for power calculation; the power sensor can directly measure the size and direction of the power, with high measurement accuracy and fast response speed, and can be installed at key nodes to monitor the changes in power in real time.

[0022] Furthermore, the power balance of the distribution network can be monitored in real time based on the electrical parameters of the distribution network. In order to monitor the power status of the distribution network, it is necessary to analyze a variety of power data in the distribution network, and then accurately analyze the power status of the distribution network to improve the operating efficiency of the power grid.

[0023] In an embodiment of the present invention, the multi-dimensional power data refers to active power, reactive power and apparent power, wherein active power refers to the power consumed by the resistance element in the AC circuit. It is the part of power that converts electrical energy into other forms of energy (such as heat energy, mechanical energy, etc.). In the distribution network, active power is used to drive various electrical equipment to work normally, such as motor operation, light bulb lighting, electric furnace heating, etc.; reactive power refers to the power occupied by the inductor and capacitor elements in the energy conversion process in the AC circuit. It does not do work to the outside, but exchanges energy between the power supply and the inductor and capacitor to establish the magnetic field and electric field; apparent power refers to the product of the effective value of voltage and current in the AC circuit. It represents the total power provided by the power supply, including active power and reactive power. The apparent power reflects the total electrical load capacity of the equipment in the distribution network.

[0024] In an embodiment of the present invention, the step of calculating the multi-dimensional power data of each node in the distribution network according to the electrical data includes: Extract the circuit type corresponding to each node in the distribution network; quantizing the circuit type to obtain a type quantization value; The power data of each node in the distribution network is calculated according to the type quantization value, wherein the power data calculation formula is: ; in, For the power data, is the effective value of the voltage at the node in the first circuit type, is the effective value of the current at the node in the first circuit type, is the phase difference between the voltage and current at the node in the first circuit type, is a type quantized value, is the effective value of the line voltage at the node in the second circuit type, is the effective value of the line current at the node in the second circuit type, is the phase difference between the phase voltage and the phase current at the node in the second circuit type, is the first power variation factor, is the second power variation factor, is the third power variation factor, is the cosine function, is a sine function; The power data is aggregated into multi-dimensional power data of each node in the distribution network according to the power change factor.

[0025] In detail, there are different types of circuits in the distribution network, and the circuit types include single-phase AC circuits and three-phase AC circuits. The power calculation methods for different circuit types are different. Accurately extracting the circuit type corresponding to each node is the basis for the subsequent correct calculation of power data, and using intelligent monitoring equipment to obtain the circuit connection information of the node. For example, if the node is connected to a single-phase electrical device of a residential user, it can be determined that the node is a single-phase AC circuit; if the node is connected to a three-phase load such as a three-phase motor, it is a three-phase AC circuit. In order to reflect the impact of different circuit types on power calculation in a unified calculation formula, it is necessary to quantify the circuit type and use specific numerical values ​​to represent different circuit types, such as setting the type quantization value of a single-phase AC circuit. , the type quantization value of the three-phase AC circuit .

[0026] Specifically, when When , it indicates the first circuit type (single-phase AC circuit). The power change factor can be set according to different physical meanings and calculation purposes. If the active power is calculated, then , if the reactive power is calculated, then , if the apparent power is calculated, then ;when When , it indicates the second circuit type (three-phase AC circuit). If the active power is calculated, then , if the reactive power is calculated, then , if the apparent power is calculated, then , and then calculate the power data of each node after considering different factors according to different circuit types. However, in order to fully reflect the power characteristics of distribution network nodes, these power data need to be aggregated into multi-dimensional power data.

[0027] Furthermore, multi-dimensional power data contains active power, reactive power and apparent power. It is difficult to intuitively judge the overall operating status of the distribution network by only viewing the raw data. By extracting multi-dimensional power features, such as power change rate, power fluctuation amplitude, etc., the operation of the distribution network can be comprehensively described from multiple dimensions. For example, the power factor can reflect the utilization efficiency of electric energy, and the power change rate can reflect the changing trend of the load. Through power features, we can fully understand whether the distribution network is in a stable and efficient operating state.

[0028] S2. Extracting multidimensional power features corresponding to the multidimensional power data, and analyzing the power operation status of the distribution network through the multidimensional power features.

[0029] In the embodiment of the present invention, the multi-dimensional power characteristics include characteristic information in multiple aspects such as time domain, frequency domain and sequence pattern, which can comprehensively and deeply describe the changing law, fluctuation characteristics, periodicity and time correlation of power in the distribution network.

[0030] In the embodiment of the present invention, refer to Figure 2 As shown, the extracting the multi-dimensional power features corresponding to the multi-dimensional power data includes: S21, dividing the multidimensional power data of each node into power data sets according to the same dimension; S22, extracting time domain features and frequency domain features from the power data set; S23, converting the power data set into a sequence set, and extracting sequence pattern features of the sequence set; S24, aggregating the time domain features, the frequency domain features and the sequence pattern features into a multi-dimensional power feature.

[0031] In detail, the multi-dimensional power data of each node in the distribution network contains different types of data such as active power, reactive power and apparent power, and these data are constantly changing at different time points. In order to facilitate subsequent analysis, it is necessary to classify and organize the power data of the same type (i.e., the same dimension). Based on the power type, the active power data of each node is classified into one data set, the reactive power data into another data set, and the apparent power data into a third data set.

[0032] Specifically, the time domain characteristics reflect the changes of power data over time. By analyzing the time domain characteristics, we can understand the changing trend, fluctuation and extreme value of power, which is helpful to judge the operating status and load characteristics of the distribution network. The time domain characteristics include mean, maximum, minimum, standard deviation, rate of change, etc. For example, calculating the mean of the active power data set can understand the average level of power in this dimension; calculating the standard deviation can measure the discreteness of the power data and reflect the fluctuation of power; calculating the rate of change of power can observe the speed of power change over time; frequency domain characteristics are used to analyze the frequency components of power data, which can reveal the periodic information hidden in the power data. Different load types and operating conditions may produce power fluctuations of different frequencies. The power data in the time domain is converted to the frequency domain using Fourier transform to obtain the spectrum of the power data. The frequency domain characteristics such as the main frequency components and harmonic content can be extracted from the spectrum.

[0033] Furthermore, the power data set is converted into a sequence set, that is, the power data arranged in chronological order is regarded as a sequence, and the data in the power data set are arranged in chronological order to form a time series. For example, the active power data of a certain node is arranged in chronological order to obtain an active power time series. Such a conversion is performed on each power data set to obtain multiple power sequence sets, and then the sequence pattern features are extracted. The sequence pattern features can reflect the temporal order and correlation of the power data, which is helpful to discover the law and trend of power changes and predict the future power status. For example, through analysis, it is found that the active power of a certain node will have a peak value between 7 and 9 o'clock every night, that is, the sequence pattern feature. Then, the extracted time domain features, frequency domain features and sequence pattern features are integrated into a feature vector or feature matrix to form a multi-dimensional power feature, which describes the characteristics of the multi-dimensional power data from different angles and more accurately reflects the power operation status of the distribution network.

[0034] Furthermore, multi-dimensional power characteristics contain information in multiple dimensions such as time domain, frequency domain and sequence pattern. Combining the characteristics of different dimensions, it can comprehensively and meticulously characterize the power operation status of the distribution network from multiple angles, avoiding the limitations of single-dimensional analysis.

[0035] In the embodiment of the present invention, the power operation state refers to the power-related working condition of the distribution network at a certain moment, including a normal operation state, in which the power distribution in the distribution network is reasonable, the equipment operates stably, and various power indicators are within the normal range; an overload state, that is, the load in the distribution network exceeds the rated capacity of the equipment, which may cause problems such as heating and damage of the equipment; an underload state, which means that the load of the distribution network is too low, the equipment is not fully utilized, resulting in a waste of resources; a fault state, such as a short circuit fault, a ground fault, etc.

[0036] In an embodiment of the present invention, the step of analyzing the power operation state of the distribution network by using the multi-dimensional power characteristics includes: Constructing an operation state matrix according to the power feature dimension in the multi-dimensional power feature and the power operation state dimension of the distribution network acquired in advance; Determining a power operation state value of the distribution network by using the multi-dimensional power characteristics and the multi-dimensional power history characteristics acquired in advance; Filling the operation state matrix according to the power operation state value; The populated operation state matrix is ​​used to calculate the power operation state quantization value of the distribution network, where the power operation state quantization value calculation formula is: ; in, is the quantized value of the power operation state, is the state standard value, is the first Line The matrix values ​​of the columns, is the row dimension in the running status matrix, is the column dimension in the running status matrix, is the number of nodes in the distribution network, is the node number in the distribution network; The power operation state of the power distribution network is determined according to the power operation state quantization value.

[0037] In detail, the multidimensional power features include features extracted from multiple angles such as the time domain, frequency domain and sequence patterns, such as the mean, standard deviation, main frequency components, periodic patterns, etc. of power. Different types of features constitute the power feature dimension; various operating states of the distribution network, such as normal operation, overload, underload, short circuit fault, ground fault, etc., different operating states constitute the power operating state dimension. A matrix is ​​constructed with the power feature dimension as the column and the power operating state dimension as the row. Each element of the matrix represents the association information between a certain power feature and a certain operating state.

[0038] Specifically, the multidimensional power historical characteristics refer to the multidimensional power characteristic data of the distribution network collected and analyzed in the past period of time. The historical characteristics record the power characteristic performance of the distribution network under different operating conditions, and then compare and analyze the currently acquired multidimensional power characteristics with the multidimensional power historical characteristics. By comparing the current power fluctuation, frequency component and other characteristics with the characteristics under historical normal operation, overload and other states, the current operating state of the distribution network can be determined; according to the determined power operating state value, the corresponding matrix elements are assigned values. If it is determined that the current distribution network is in an overload state, then the rows corresponding to the overload state in the operating state matrix are assigned corresponding values ​​according to different power characteristic dimensions.

[0039] Furthermore, there are n multidimensional power features and m power operation status dimensions. When the data corresponding to the normal operating status are all 1, it indicates normal operation, otherwise it is abnormal. The value 3 in the power operation status quantization value calculation formula refers to active power, reactive power, and apparent power. Only when the three powers are normal values, it means that at time T, nodes A, B, and C are normal operating powers. If the characteristic value in each multidimensional power feature belongs to the normal range, it corresponds to the normal operating state, that is, the corresponding value of the normal operating state corresponding to each multidimensional power feature is 1, and overload, underload, short circuit fault, and ground fault are regarded as abnormal operating states. As long as multiple If the characteristic value in the multi-dimensional power feature does not fall within the normal range, it corresponds to an abnormal operating state, that is, the corresponding value of the operating state corresponding to each multi-dimensional power feature is 0, that is, only when all the values ​​in the normal operating state row in the operating state matrix are 1, and all the values ​​in the normal operating state row in the three types of power data are 1, the distribution network is in a normal operating state, otherwise, in any other case, the distribution network is in an abnormal operating state. When the power operating state quantization value is 1, it indicates that the power operating state of the distribution network is in a normal operating state, and when the power operating state quantization value is 0, it indicates that the power operating state of the distribution network is in an abnormal operating state.

[0040] Furthermore, the power operation state of the distribution network is not constant, but fluctuates over time. In order to clearly capture the real-time fluctuations of power and intuitively display the changing trend of power over time, it is necessary to construct a power time series and arrange the power operation states at different times in chronological order, thereby providing real-time information for the dispatch and control of the power grid and realizing the intelligent operation and management of the power grid.

[0041] S3. Construct a power time series of the distribution network according to the power operation status, and use a pre-constructed time series model to perform situation awareness on the power time series to obtain a situation prediction value of the distribution network.

[0042] In an embodiment of the present invention, the power time series is arranged in chronological order, and is a sequence reflecting changes in information related to the power status of the distribution network, with time as the horizontal axis and the power operation status (such as normal, abnormal, etc.) as the vertical axis. Through the power time series, the change trend, fluctuation and periodic characteristics of the power of the distribution network over time can be intuitively observed.

[0043] In the embodiment of the present invention, refer to Figure 3 As shown, the power time series of the power distribution network is constructed according to the power operation state, including: S31, extracting the operation time point corresponding to the power operation state; S32, mapping the operation time point with the power operation state to obtain a mapping sequence; S33: Convert the mapping sequence into a power time series of the distribution network.

[0044] In detail, during the operation of the distribution network, the monitoring of the power operation status is carried out at different time points, which record the moment when the power operation status data is generated. For example, through the monitoring equipment installed at each node, power-related data is collected at fixed time intervals (such as every minute, every hour, etc.). These collection moments are the operation time points, and the power operation status includes normal operation, overload, underload, etc. A correspondence is established between each operation time point and the power operation status information corresponding to the moment, that is, one row records the time and the other row records the power operation status data at the same moment to form a mapping, and then the mapping sequence represents the association of time and power operation status to obtain the power time series of the distribution network. It is an ordered sequence with time as the index and power-related data or status as the value, which can be used for subsequent in-depth analysis, such as load forecasting, fault diagnosis, etc.

[0045] Furthermore, in order to predict the electricity demand in different time periods, for example, during the high temperature period in summer, if the increase in air-conditioning load is predicted to lead to an increase in power demand, the power department can adjust the power generation plan in advance, increase power generation, ensure that the power supply meets the demand, and avoid power outages or power restrictions. Therefore, it is necessary to conduct situational awareness forecasts based on power time series to know the electricity demand in different time periods in advance.

[0046] In an embodiment of the present invention, the situation prediction value of the distribution network refers to the predicted value of power-related conditions in the future operating state of the distribution network obtained after analyzing and processing the power time series using a time series model. This value can represent a variety of power-related situations, such as the predicted values ​​of active power and reactive power, or the load prediction value and voltage prediction value obtained based on power analysis.

[0047] In an embodiment of the present invention, the method of using a pre-built time series model to perform situation awareness on the power time series to obtain a situation prediction value of the distribution network includes: converting the power time series into model sequence input data; Extract model parameters of pre-built time series models; Initializing the hidden state and cell state of the time series model; Calculating the target hidden state corresponding to the model sequence input data using the initialized time series model and the model parameters; Output data is determined according to the target hidden state, and a linear transformation is performed on the output data to obtain a state prediction value of the distribution network.

[0048] In detail, the pre-built time series model is an LSTM model, and the LSTM model requires the input data to be a three-dimensional array in the format of [number of samples, time steps, number of features]. Assume that the input power time series has samples, each containing The input data has the power value of time steps and only has one feature, power. The shape of the input data is If the input data does not meet this format, corresponding adjustments and reshaping operations are required. After loading the pre-trained LSTM model, the weight and bias parameters of the model are initialized. These parameters have been learned in the training phase and are used to control the input gate, forget gate and output gate of the LSTM unit, and reset the hidden state and cell state of the LSTM model. At the beginning of prediction, the hidden state and cell state are usually initialized to zero vectors.

[0049] Specifically, the input data is input into the LSTM unit in sequence according to the time step. Each LSTM unit updates the hidden state and cell state at the current moment through the calculation of the input gate, forget gate and output gate according to the current input, the hidden state and cell state at the previous moment. , the Forget Gate , cell state candidate value , cell state update , output gate , hide status updates ,in, is the input at the current moment, is the hidden state at the previous moment, is the cell state at the previous moment, is the weight matrix, is the bias vector, is the sigmoid function, is the hyperbolic tangent function, Represents element-by-element multiplication, the hidden state of the last time step Input to the output layer and get the predicted value through linear transformation ,in is the weight matrix of the output layer, It is the bias vector of the output layer. In addition, it is possible to predict iteratively, that is, predict the value of one time point each time, and then add the predicted value as a new input to the input sequence to continue predicting the next time point. Another method is direct prediction, that is, training an LSTM model that can directly output predicted values ​​for multiple time points, and obtaining the prediction results for multiple time points at one time during prediction.

[0050] Furthermore, the load in the distribution network is in continuous dynamic change, and it is necessary to capture the power load change information in real time. When making power situation forecasts, the newer the data, the more it can reflect the actual operating trend of the current distribution network. For example, in short-term load forecasts, the newly included real-time power data may contain load change information caused by sudden weather changes or special activities. Forecasting based on the updated time series can enable the forecasting model to better capture these changes, thereby improving the forecast accuracy.

[0051] S4. Update the power time series within a preset moving time window, determine the power situation triggering condition of the distribution network according to the updated power time series, and update the model parameters in the time series model according to the power situation triggering condition.

[0052] In an embodiment of the present invention, a moving time window is a time period with a fixed time length, which slides on the power time series and is used to select data within a specific time range for analysis. For example, a moving time window is set to 1 hour, that is, power data for 1 hour is selected for processing each time. This time window is like a window, which moves continuously on the time axis of the power time series. Each movement will cover a new section of data. The moving time window continuously updates the power time series to ensure that the time series used for prediction always contains the latest power data.

[0053] Furthermore, the power time series records the changes in power over time. By continuously updating the sequence, these dynamic changes can be captured in real time. By determining the trigger conditions based on the updated sequence, it is possible to keenly perceive when the operating status of the distribution network changes significantly.

[0054] In the embodiment of the present invention, the power situation trigger condition is a regular setting for monitoring and managing the operation status of the distribution network. It is formulated based on the analysis of the power time series of the distribution network, by detecting abnormal situations such as power mutation, and determining possible triggering factors.

[0055] In the embodiment of the present invention, the step of determining the power situation triggering condition of the distribution network according to the updated power time series includes: Detect mutation factors in the updated power time series; Determining a triggering factor of the distribution network according to the mutation factor; The power situation triggering condition of the power distribution network is generated by the triggering factor.

[0056] In detail, the power time series reflects the change of distribution network power over time. In actual operation, the power of the distribution network may mutate due to various reasons. This mutation may indicate a change in the operating state of the distribution network or a potential problem. The mutation factor is a parameter used to quantify the degree and characteristics of power mutation. The first-order derivative or second-order derivative of the power time series can be calculated to measure the rate and acceleration of power change. If the derivative shows a large numerical change at a certain moment, it means that the power has mutated at that moment. In addition, the sliding window technology can be used to calculate the mean, standard deviation and other statistics of the power data in the window. When the statistical characteristics of the data in the window are significantly different from those of the previous and next windows, it also indicates that there may be a mutation. For example, in a sliding window with a length of 10 minutes, the mean and standard deviation of the power are calculated. If the mean of the current window suddenly increases by 50% compared with the mean of the previous window, and the standard deviation also increases significantly, this may be a mutation signal, so that the mutation factor in the updated power time series can be accurately detected, providing key information for subsequent analysis.

[0057] Specifically, once the mutation factor is detected, the reasons behind the mutation are analyzed to determine the triggering factors of the distribution network. Different types of mutation factors often correspond to different triggering factors. For example, if the mutation factor is manifested as a sudden and substantial increase in power, the possible triggering factors are the startup of large power-consuming equipment, the access of new loads, or the failure of distributed power sources; if the mutation factor shows a sharp drop in power, it may be caused by line failure, equipment tripping, or power outages of important users. The updated power time series is analyzed to determine the power situation triggering conditions. The triggering conditions can be set based on factors such as the power change amplitude, change rate, and deviation from historical data. For example, when the change rate of the active power of a node exceeds a preset threshold in a short period of time, or the deviation of the power value from the historical data of the same period exceeds a certain range, the corresponding conditions are triggered. When the power situation triggering conditions are met, it means that the operating state of the distribution network has changed significantly, and the parameters in the time series model need to be updated. By re-inputting the updated power time series into the model for training, the parameters of the model are adjusted so that the model can better adapt to the new operating state and improve the accuracy of the prediction. If the real-time value of the active power exceeds 120% of its historical average, or the power change rate exceeds 30% in a short period of time, the triggering conditions are met.

[0058] Furthermore, as the distribution network operates, the model may deviate during the prediction process. The abnormal or newly emerged power situation reflected by the power situation triggering condition may indicate that the model prediction is biased. For example, when the actual power exceeds the model prediction range and the power situation condition is triggered, it indicates that the model's estimation of the current power change is inaccurate. By updating the model parameters, these deviations can be corrected, making the model prediction more consistent with the actual power change and improving the prediction accuracy.

[0059] In an embodiment of the present invention, updating model parameters refers to adjusting the parameters used for calculation and prediction within the model according to newly acquired data features (such as the degree and direction of change reflected in the power series data collected according to the power status trigger conditions this time) during the training or operation of the time series model. The model parameters include but are not limited to the weight matrix and bias vector in the neural network. When predicting the future power value of the distribution network, the updated model parameters can enable the model to better capture the complex patterns and trends in the power data and improve the accuracy of the prediction.

[0060] In the embodiment of the present invention, updating the model parameters in the time series model according to the power situation triggering condition includes: Collecting power sequence data within a preset time period according to the power situation trigger condition; Extracting the data change degree and data change direction in the power sequence data; Determining an adjustment factor according to the degree of change of the data and the direction of change of the data; The learning rate in the time series model is updated by the adjustment factor, wherein the learning rate update formula is: ; in, For the The learning rate of the iteration, is the minimum value of the learning rate, is the maximum value of the learning rate, is the current iteration number, is the total number of iterations in one cycle, is a constant, is the cosine function, is the regulating factor; The updated model parameters in the time series model are determined by the learning rate.

[0061] In detail, the power situation triggering conditions are rules set based on the power changes of the distribution network. When these conditions are triggered, it means that the power state of the distribution network has changes that need attention, such as power mutations, abnormal fluctuations, etc. At this time, in order to more accurately understand the operating status of the distribution network and optimize the time series model, it is necessary to collect power series data within a preset time period. The power data within the preset time period contains sufficient information to reflect the characteristics of power changes, which can represent the process of power from stability to mutation and then recovery, and then extract the degree of data change and data change direction in the power series data. The degree of data change is used to measure the drastic degree of power change in the time period, such as by calculating the standard deviation, range or difference between adjacent data points of the power series data to quantify the degree of change. The larger the standard deviation, the greater the discreteness of the power data, that is, the greater the degree of change; the range directly reflects the difference between the maximum and minimum values ​​of the data, reflecting the fluctuation range of power in the time period; the direction of data change describes whether the power is rising or falling. It can be determined by observing the size relationship of adjacent data points. If the latter data point is larger than the previous one, the direction of change is upward; otherwise, it is downward. Accurately extracting the degree and direction of data change helps to understand the specific characteristics of power change and provide a basis for subsequent adjustment of the time series model.

[0062] Specifically, the degree and direction of data change will affect the adjustment of the learning rate. If the power data changes greatly, it means that the model needs to adapt to the new data pattern more quickly. At this time, it is hoped that the learning rate is relatively large so that the model parameters can be updated faster. On the contrary, if the degree of change is small, the learning rate can be appropriately reduced to make the model parameter update more stable. Different directions of change may imply different power change mechanisms and may also affect the strategy for adjusting the learning rate. For example, if the power continues to rise and the degree of change is large, it may be necessary to adjust the learning rate more actively. In this way, the adjustment factor is set to 1.5 to keep up with the rhythm of power change. In this way, the learning rate is dynamically adjusted according to the characteristics of the power sequence data, so that the model can better adapt to data changes during training. For example, if the power continues to decrease and the degree of change is large, the adjustment factor is set to 0.5. If the power continues to decrease and the degree of change is small or the power continues to rise and the degree of change is small, the adjustment factor is set to 1, and then the learning rate of the model is calculated according to the learning rate update formula. The learning rate changes in the form of a cosine function with the number of iterations, gradually decreasing from the maximum value to the minimum value within a cycle, and then the process is repeated periodically. This method helps the model escape from the local optimal solution and find a better global optimal solution.

[0063] Furthermore, the learning rate is an important hyperparameter in the training process of the time series model. It determines the step size of the model when updating parameters in each iteration. The appropriate learning rate is crucial to the convergence speed and prediction accuracy of the model. The learning rate plays a role in controlling the parameter update step size in this process. When the learning rate is determined, the model moves the model parameters in the opposite direction of the gradient by a certain distance in each iteration according to the calculated gradient (the gradient represents the rate of change of the error to the model parameters). This distance is the product of the learning rate and the gradient. By repeating this process, the model parameters are gradually adjusted using the updated learning rate, so that the model can better fit the power series data and improve the prediction ability of the power situation of the distribution network. The model parameters determine how the model transforms and combines the input data and finally generates the prediction results. By updating the model parameters, the model can adapt to the dynamic changes of the data and continuously optimize its own prediction ability to more accurately reflect the changing law of the power situation of the distribution network.

[0064] Furthermore, as time goes by, the time series model may deviate in the prediction process. The abnormal power situation reflected by the power situation triggering condition often suggests that there are deficiencies in the model prediction. For example, when the actual power deviates greatly from the predicted value and triggers the corresponding condition, updating the model parameters and optimizing the situation prediction value can correct the model deviation, make the predicted value closer to the actual power situation, and improve the prediction accuracy.

[0065] S5. Utilize the updated time series model to optimize the situation prediction value to obtain a situation optimization value, and utilize the situation optimization value to analyze the power situation awareness of the distribution network to obtain a power situation awareness prediction result.

[0066] In an embodiment of the present invention, the situation optimization value refers to a prediction value that can more accurately reflect the future power situation of the distribution network, which is finally determined by updating the time series model during the prediction process of the power situation of the distribution network and comparing the prediction results of the updated model with those of the original model.

[0067] In the embodiment of the present invention, the method of optimizing the situation prediction value using the updated time series model to obtain the situation optimization value includes: Using the updated time series model to perform situation awareness on the power time series, and obtain a situation update prediction value of the distribution network; Calculating an error value between the situation update prediction value and the situation prediction value; When the error value is less than or equal to a preset error threshold, the situation prediction value is used as the situation optimization value; When the error value is greater than a preset error threshold, the situation update prediction value is used as the situation optimization value.

[0068] In detail, the updated time series model is obtained by updating the original model parameters based on the power situation triggering conditions. The updated model has the ability to better adapt to the latest power change characteristics of the distribution network. Using this model to perceive the power time series situation means that the model uses the updated parameters and algorithms based on the current and historical power data to predict the future power situation of the distribution network, thereby obtaining the situation update prediction value. According to the data patterns and rules in the power time series, combined with the updated model parameters, the power situation related values ​​at the next time point or in the future period are predicted, such as predicting the active power value in the next hour. This prediction value is the situation update prediction value.

[0069] Specifically, the situation prediction value is the prediction result of the power situation of the distribution network obtained before the model is updated. In order to evaluate whether the updated model has improved the accuracy of the prediction, it is necessary to calculate the difference between the situation update prediction value and the situation prediction value. This difference is the error value. The methods for calculating the error value include mean square error (MSE), mean absolute error (MAE), etc., and the preset error threshold is a standard value set in advance based on actual needs and the requirements for prediction accuracy. When the calculated error value is less than or equal to this threshold, it means that although the updated model has adjusted the prediction value, it has not significantly improved the accuracy of the prediction, or the difference between the updated prediction value and the original prediction value is within an acceptable range. In this case, it is believed that the original situation prediction value can already meet the current prediction needs, so the situation prediction value is directly used as the situation optimization value, that is, in some situations In this case, over-reliance on the updated forecast value may introduce unnecessary fluctuations or instability, while the original forecast value may already be a relatively reliable result; if the error value is greater than the preset error threshold, it means that the updated time series model has made a large change to the forecast value, and this change exceeds the acceptable range, indicating that the updated model has captured some information that the original model did not consider, causing the forecast result to change significantly. In this case, it is believed that the updated model has better prediction effect and can more accurately reflect the power situation of the distribution network. Therefore, the situation update forecast value is used as the situation optimization value. In this way, after the model is updated, the better forecast value can be flexibly selected as the final situation optimization value according to the actual forecast effect, so as to improve the accuracy and reliability of the forecast.

[0070] Furthermore, the situation optimization value is based on the updated time series model and is obtained by comprehensively considering the latest change characteristics of the distribution network power. Compared with the initial situation prediction value, it can more accurately reflect the current actual operating status and future development trend of the distribution network. The situation optimization value obtained by updating the model can capture these changes in time, providing a more accurate data basis for subsequent power situation awareness analysis, thereby improving the accuracy of the prediction.

[0071] In an embodiment of the present invention, the power situation perception prediction result is an information set that comprehensively reflects the accuracy of the power situation prediction of the distribution network and related conclusions. It is not just a simple prediction value, but a power situation prediction effect of the distribution network obtained through a series of analysis steps (such as comparing with the target power situation value, determining the perception level, etc.).

[0072] In an embodiment of the present invention, the method of analyzing the power situation awareness of the distribution network by using the situation optimization value to obtain a power situation awareness prediction result includes: Extract the target power situation value at the target time; Comparing the situation optimization value with the target power situation value to obtain a comparison result; Determine the power situation awareness level of the distribution network according to the comparison result; The power situation awareness prediction result is determined according to the awareness level.

[0073] In detail, when analyzing the power situation of the distribution network, it is necessary to pay attention to the power-related status at a specific time point. This specific time point is the target moment, and the target power situation value refers to the actual or expected power-related value of the distribution network at the target moment. Comparing the situation optimization value with the target power situation value can evaluate the accuracy of the prediction and the actual changes in the power situation of the distribution network. The difference between the two (predicted value - actual value) is calculated to measure the degree of deviation of the prediction, or the error rate (|predicted value - actual value| / actual value) is calculated to more intuitively represent the proportion of the prediction deviation from the actual. This difference or error rate is the comparison result.

[0074] Specifically, in order to more intuitively evaluate the effect of power situation awareness in the distribution network, different perception levels are divided according to the comparison results. The perception levels are usually pre-set and determined according to actual needs and the requirements for the accuracy of power situation prediction. For example, the error rate within 5% can be set as a high perception level, indicating that the predicted value is very close to the actual value, and the perception of the power situation is relatively accurate; the error rate between 5%-15% is a medium perception level, indicating that the prediction has a certain deviation, but is still within an acceptable range; the error rate exceeds 15% for a low perception level, which means that the prediction deviation is large, and the perception of the power situation is not accurate enough. By comparing the comparison results (such as the error rate) with these preset level ranges, the power state of the distribution network can be determined. The perception level of power situation perception reflects the closeness between the predicted value and the actual value and the perception quality of the power situation. The power situation perception prediction result is then determined based on the perception level. For example, when the perception level is high, it means that the prediction is more reliable. The power situation perception prediction result can be considered as an accurate prediction of the power situation of the distribution network. Relevant personnel can perform routine power grid dispatching, equipment maintenance and other work based on this prediction. If the perception level is low, it means that the prediction deviation is large, and it may be necessary to review the prediction model, check the data accuracy or consider more influencing factors. At this time, the power situation perception prediction result indicates that the prediction process needs to be adjusted and optimized to improve the prediction ability of the distribution network power situation.

[0075] The embodiment of the present invention collects electrical data of each node and calculates multi-dimensional power data, which can comprehensively cover different aspects of power distribution network; constructs power time series according to power operation status, and uses pre-constructed time series model for situation awareness, which can effectively capture the change law and trend of power over time; updates the power time series within the preset moving time window, so that the system can track the latest changes of power distribution network in real time. According to the updated time series, the power situation triggering condition is determined, and the abnormal fluctuation or mutation of power can be captured in time; the parameters of the time series model are updated based on the power situation triggering condition, so that the model can dynamically adjust itself according to the real-time operation status of the distribution network, and always maintain the ability to accurately predict the power situation; the situation prediction value is optimized using the updated time series model, and by comparing the predicted value before and after the update with the actual power situation value, a better situation optimization value is selected, which effectively corrects the prediction deviation, improves the matching degree between the predicted value and the actual power situation, and makes the prediction result more accurate and reliable. Therefore, the power situation perception prediction method and prediction system of the distribution network proposed by the present invention can solve the problem of low accuracy when making product recommendations.

[0076] like Figure 4 , which is a functional module diagram of a distribution network power situation awareness and prediction system provided by an embodiment of the present invention.

[0077] The power distribution network power situation awareness prediction system 100 of the present invention can be installed in an electronic device. According to the functions to be implemented, the power distribution network power situation awareness prediction system 100 can include a multi-dimensional power data calculation module 101, a power operation status analysis module 102, a situation prediction value analysis module 103, a model parameter update module 104 and a power situation awareness analysis module 105. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0078] In this embodiment, the functions of each module / unit are as follows: The multi-dimensional power data calculation module 101 is used to collect electrical data corresponding to each node in the distribution network, and calculate the multi-dimensional power data of each node in the distribution network according to the electrical data; The power operation status analysis module 102 is used to extract the multi-dimensional power characteristics corresponding to the multi-dimensional power data, and analyze the power operation status of the distribution network through the multi-dimensional power characteristics; The situation prediction value analysis module 103 is used to construct a power time series of the distribution network according to the power operation state, and use a pre-constructed time series model to perform situation perception on the power time series to obtain a situation prediction value of the distribution network; The model parameter updating module 104 is used to update the power time series within a preset moving time window, determine the power situation triggering condition of the distribution network according to the updated power time series, and update the model parameters in the time series model according to the power situation triggering condition; The power situation awareness analysis module 105 is used to optimize the situation prediction value using the updated time series model to obtain a situation optimization value, and use the situation optimization value to analyze the power situation awareness of the distribution network to obtain a power situation awareness prediction result.

[0079] In detail, each module described in the power distribution network power situation awareness prediction system 100 in the embodiment of the present invention is used in the same manner as described above. Figures 1 to 3 The same technical means are used as the distribution network power situation awareness and prediction method described in, and can produce the same technical effects, so they will not be repeated here.

[0080] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0081] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0082] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0083] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0084] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is not limited only according to the above description, and it is intended that all changes within the meaning and scope of equivalent elements within the scope of protection are included in the present invention.

[0085] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0086] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units or systems stated in the system embodiment can also be implemented by one unit or system through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any specific order.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A power distribution network power situation awareness prediction method, characterized in that: The method comprises: Collecting electrical data corresponding to each node in the distribution network, and calculating multi-dimensional power data of each node in the distribution network based on the electrical data; Extracting multidimensional power features corresponding to the multidimensional power data, and analyzing the power operation state of the distribution network through the multidimensional power features; Constructing a power time series of the distribution network according to the power operation state, and using a pre-constructed time series model to perform situation awareness on the power time series to obtain a situation prediction value of the distribution network; The power time series is updated within a preset moving time window, a power situation triggering condition of the distribution network is determined according to the updated power time series, and a model parameter in the time series model is updated according to the power situation triggering condition; The situation prediction value is optimized using the updated time series model to obtain a situation optimization value, and the power situation awareness of the distribution network is analyzed using the situation optimization value to obtain a power situation awareness prediction result.

2. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The extracting the multi-dimensional power features corresponding to the multi-dimensional power data includes: Divide the multi-dimensional power data of each node into power data sets according to the same dimension; Extracting time domain features and frequency domain features from the power data set; converting the power data set into a sequence set, and extracting sequence pattern features of the sequence set; The time domain features, the frequency domain features and the sequence pattern features are aggregated into a multi-dimensional power feature.

3. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The analyzing the power operation state of the distribution network by using the multi-dimensional power characteristics includes: Constructing an operation state matrix according to the power feature dimension in the multi-dimensional power feature and the power operation state dimension of the distribution network acquired in advance; Determining a power operation state value of the distribution network by using the multi-dimensional power characteristics and the multi-dimensional power history characteristics acquired in advance; Filling the operation state matrix according to the power operation state value; The populated operation state matrix is ​​used to calculate the power operation state quantization value of the distribution network, where the power operation state quantization value calculation formula is: ； in, is the quantized value of the power operation state, is the state standard value, is the first Line The matrix values ​​of the columns, is the row dimension in the running status matrix, is the column dimension in the running status matrix, is the number of nodes in the distribution network, is the node number in the distribution network; The power operation state of the power distribution network is determined according to the power operation state quantization value.

4. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The step of constructing a power time series of a power distribution network according to the power operation state includes: Extracting the operation time point corresponding to the power operation state; Mapping the operation time point with the power operation state to obtain a mapping sequence; The mapping sequence is converted into a power time series of the power distribution network.

5. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The method of using a pre-built time series model to perform situation awareness on the power time series to obtain a situation prediction value of the distribution network includes: converting the power time series into model sequence input data; Extract model parameters of pre-built time series models; Initializing the hidden state and cell state of the time series model; Calculating the target hidden state corresponding to the model sequence input data using the initialized time series model and the model parameters; Output data is determined according to the target hidden state, and a linear transformation is performed on the output data to obtain a state prediction value of the distribution network.

6. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The step of determining the power situation triggering condition of the distribution network according to the updated power time series includes: Detect mutation factors in the updated power time series; Determining a triggering factor of the distribution network according to the mutation factor; The power situation triggering condition of the power distribution network is generated by the triggering factor.

7. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The updating of the model parameters in the time series model according to the power situation triggering condition includes: Collecting power sequence data within a preset time period according to the power situation trigger condition; Extracting the data change degree and data change direction in the power sequence data; Determining an adjustment factor according to the degree of change of the data and the direction of change of the data; The learning rate in the time series model is updated by the adjustment factor, wherein the learning rate update formula is: ； in, For the The learning rate of the iteration, is the minimum value of the learning rate, is the maximum value of the learning rate, is the current iteration number, is the total number of iterations in one cycle, is a constant, is the cosine function, is the regulating factor; The updated model parameters in the time series model are determined by the learning rate.

8. The power distribution network power situation awareness prediction method according to claim 4, characterized in that: The method of optimizing the situation prediction value by using the updated time series model to obtain the situation optimization value includes: Using the updated time series model to perform situation awareness on the power time series, and obtain a situation update prediction value of the distribution network; Calculating an error value between the situation update prediction value and the situation prediction value; When the error value is less than or equal to a preset error threshold, the situation prediction value is used as the situation optimization value; When the error value is greater than a preset error threshold, the situation update prediction value is used as the situation optimization value.

9. The power distribution network power situation awareness prediction method according to claim 1, characterized in that: The method of analyzing the power situation awareness of the distribution network by using the situation optimization value to obtain a power situation awareness prediction result includes: Extract the target power situation value at the target time; Comparing the situation optimization value with the target power situation value to obtain a comparison result; Determine the power situation awareness level of the distribution network according to the comparison result; The power situation awareness prediction result is determined according to the awareness level.

10. A power distribution network power situation awareness prediction system, characterized in that: The system comprises: A multi-dimensional power data calculation module, used to collect electrical data corresponding to each node in the distribution network, and calculate the multi-dimensional power data of each node in the distribution network according to the electrical data; A power operation status analysis module, used to extract the multi-dimensional power features corresponding to the multi-dimensional power data, and analyze the power operation status of the distribution network through the multi-dimensional power features; A situation prediction value analysis module is used to construct a power time series of the distribution network according to the power operation state, and use a pre-constructed time series model to perform situation perception on the power time series to obtain a situation prediction value of the distribution network; A model parameter updating module, used to update the power time series within a preset moving time window, determine the power situation triggering condition of the distribution network according to the updated power time series, and update the model parameters in the time series model according to the power situation triggering condition; The power situation awareness analysis module is used to optimize the situation prediction value using the updated time series model to obtain the situation optimization value, and use the situation optimization value to analyze the power situation awareness of the distribution network to obtain the power situation awareness prediction result.

Images