Distribution transformer load prediction method and system considering population flow of communication base station

Abstract

The invention discloses a distribution transformer load prediction method and system considering population flow of a communication base station, and relates to the technical field of power system load prediction, and the method comprises the steps: obtaining dynamic data in a target power distribution region, carrying out the analysis and summarization of a first prediction model, carrying out the first prediction of the dynamic data through the first prediction model, and obtaining a first prediction result; obtaining static data in the target power distribution area, analyzing and summarizing a second prediction model, and performing second prediction on the static data through the second prediction model to obtain a second prediction result; and obtaining a third prediction result in combination with the first prediction result and the second prediction result, and completing distribution transformer load prediction of population flow of the communication base station. According to the method, more comprehensive and accurate load prediction is realized by combining the prediction results of the dynamic data and the static data, and the trend and characteristics of load change can be better captured.

Description

Technical Field

[0001] The present invention relates to the technical field of power system load forecasting, and in particular to a distribution transformer load forecasting method and system taking into account population flow at communication base stations. Background Art

[0002] With the rapid development and widespread application of smart grid technology, distribution network load forecasting, as an important basis for power system planning and operation, has attracted more and more attention for its accuracy and reliability. Traditional distribution transformer load forecasting methods mainly rely on static factors such as historical electricity consumption data and meteorological data, and use statistical methods such as time series analysis and regression analysis to build models. However, these methods often ignore the impact of dynamic factors such as population mobility on load changes, especially in the areas covered by communication base stations, where population mobility is strong and load fluctuations are large, resulting in a large deviation between the prediction results and the actual situation. In recent years, with the development of big data technology and artificial intelligence algorithms, some researchers have begun to try to apply machine learning methods to the field of load forecasting, such as support vector machines (SVM) and artificial neural networks (ANN). These methods have improved the prediction accuracy to a certain extent, but still face problems such as difficulty in feature selection and insufficient model generalization ability.

[0003] Existing distribution transformer load forecasting methods have obvious deficiencies in dealing with the impact of population mobility on load at communication base stations. First, most methods fail to fully consider the synergy of dynamic data and static data, resulting in the inability of the forecasting model to effectively capture the complexity and diversity of load changes. Secondly, traditional statistical methods and single machine learning algorithms often show limitations when processing high-dimensional, nonlinear load data, and are difficult to adapt to complex and changeable load change patterns. In addition, existing methods lack the organic combination of empirical knowledge and data-driven models, and it is difficult to fully utilize the experience and insights of domain experts. Finally, when faced with emergencies or abnormal situations, the existing methods lack robustness and adaptability, and are prone to large forecasting errors. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to provide a distribution transformer load forecasting method and system that takes into account the population flow of communication base stations, so as to solve the problems of insufficient consideration of dynamic factors, low prediction accuracy, poor model adaptability, etc. in the prior art. The present invention combines dynamic data and static data, takes advantage of mathematical statistics analysis and machine learning models, and proposes a new load forecasting method, which is expected to improve the prediction accuracy and generalization ability of the model.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a distribution transformer load prediction method taking into account the population flow of communication base stations, comprising: obtaining dynamic data within the target distribution area, analyzing and summarizing a first prediction model, performing a first prediction on the dynamic data through the first prediction model, and obtaining a first prediction result; obtaining static data within the target distribution area, analyzing and summarizing a second prediction model, performing a second prediction on the static data through the second prediction model, and obtaining a second prediction result; combining the first prediction result and the second prediction result to obtain a third prediction result, thereby completing the distribution transformer load prediction of the population flow of communication base stations.

[0007] As a preferred solution of the distribution transformer load prediction method considering the population flow of communication base stations described in the present invention, the dynamic data includes real-time load, temperature, humidity, number of base station users, time, date type, special event mark, and power system operation status; the static data includes substation capacity, distribution area area, voltage stability, and user type.

[0008] As a preferred solution of the distribution transformer load prediction method considering the population flow of communication base stations described in the present invention, wherein: the first prediction model is a preliminary prediction of the real-time load of the distribution area based on the LSTM network, and the first revised prediction value is obtained according to the influence of historical data and the second revised prediction value is obtained according to the influence of dynamic factors, and periodic changes are added to obtain, specifically: based on the output of the LSTM network, the real-time load of the distribution area is preliminarily predicted to obtain a preliminary load prediction value Based on preliminary load forecasts According to the influence of the historical data, the first revised forecast value is obtained Based on the first revised prediction value According to the influence of dynamic factors, the second revised forecast value is obtained Based on the second revised prediction value Introducing the periodic function, we get the first prediction result L t .

[0009] As a preferred solution of the distribution transformer load forecasting method considering the population flow of communication base stations described in the present invention, the preliminary load forecast value It is expressed by the following formula:

[0010]

[0011] in, is the initial load forecast value; h t is the hidden state output of the LSTM network at time t; C t is the cell state output of the LSTM network at time t.

[0012] The first corrected predicted value It is expressed by the following formula:

[0013]

[0014] Among them, x i is a historical data point, used to calculate the similarity with the current moment; x t is the feature vector at the current moment; n is the number of historical data points used to calculate the similarity.

[0015] The second corrected predicted value It is expressed by the following formula:

[0016]

[0017] Among them, P t is the population mobility factor at time t, reflecting the population mobility captured by the communication base station; T t is the temperature factor at time t, reflecting the impact of temperature on load; F t is the holiday factor at time t, reflecting the impact of holidays on load; S t is the system state at time t; S t-1 is the system state at time t-1; |S t -S t-1 | is the change amplitude of the system state; t is the time variable, the unit is hour.

[0018] The first prediction result L t It is expressed by the following formula:

[0019]

[0020] Wherein, π is the pi constant; the first prediction result L t The value range is a positive real number. When the first prediction result L t When the second preset threshold is between the first preset threshold and the second preset threshold, it indicates that the load is in a normal range, and the specific value represents the ratio relative to the historical average load. t When it is greater than the first preset threshold, it means that the load is higher than the historical average level, and the specific value represents the multiple of the excess; the second preset threshold is less than the first preset threshold.

[0021] As a preferred solution of the distribution transformer load prediction method considering the population flow of communication base stations described in the present invention, wherein: the second prediction model is obtained by obtaining a preliminary load estimate based on the substation capacity and the distribution area area, combined with the voltage change rate and the user type, specifically: obtaining a preliminary load estimate L1 based on the substation capacity C and the target distribution area area A; obtaining a preliminary load estimate L2 after considering voltage stability based on the preliminary load estimate L1 and the voltage change rate in the target distribution area; obtaining a second prediction result L based on the preliminary load estimate L2 after considering voltage stability and the user type in the target distribution area base .

[0022] As a preferred solution of the distribution transformer load prediction method considering the population flow of communication base stations described in the present invention, the preliminary load estimate L1 is expressed by the following formula:

[0023]

[0024] Where C is the substation capacity; A is the target distribution area; A ref is the reference area.

[0025] The preliminary load estimation L2 after considering voltage stability is expressed by the following formula:

[0026]

[0027] in, It is the voltage change rate, which is used to reflect the change of voltage over time.

[0028] The second prediction result L base It is expressed by the following formula:

[0029]

[0030] Among them, i, c, and r represent industrial, commercial, and residential user types respectively; P i is the proportion of each type of users, P i +P c +P r =1; E i is the electricity efficiency factor of each type of user, reflecting the electricity consumption characteristics of different types of users; L i,hist is the historical load data of various users; L ref is the reference load value.

[0031] As a preferred solution of the distribution transformer load prediction method considering the population flow of communication base stations described in the present invention, wherein: combining the first prediction result and the second prediction result to obtain a third prediction result includes: using a weighted average method to fuse the first prediction result and the second prediction result:

[0032] L final =γ·L t +(1-γ)·L base K;

[0033] Among them, γ is the dynamic weight factor, 0≤γ≤1; K is the adjustment factor; L final This is the third prediction result.

[0034] To further solve the above technical problems, the present invention provides the following technical solutions: a distribution transformer load prediction system taking into account the population flow of communication base stations, comprising: a dynamic data prediction module, used to obtain dynamic data in a target distribution area, analyze and summarize a first prediction model, and perform a first prediction on the dynamic data through the first prediction model to obtain a first prediction result; a static data prediction module, used to obtain static data in the target distribution area, analyze and summarize a second prediction model, and perform a second prediction on the static data through the second prediction model to obtain a second prediction result; a result fusion module, used to combine the first prediction result and the second prediction result to obtain a third prediction result, thereby completing the distribution transformer load prediction of the population flow of communication base stations.

[0035] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the distribution transformer load prediction method considering the population flow of communication base stations as described above are implemented.

[0036] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the distribution transformer load prediction method considering the population flow of communication base stations as described above are implemented.

[0037] Beneficial effects of the present invention: The present invention realizes a more comprehensive and accurate load forecast by combining the prediction results of dynamic data and static data, and can better capture the trend and characteristics of load changes. Through the collaborative work of the dynamic data prediction module and the static data prediction module, and the intelligent integration of the result fusion module, the present invention significantly improves the accuracy and reliability of distribution transformer load forecasting, provides more accurate decision support for power grid operation and management, and helps to improve the operating efficiency and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0039] Figure 1 An overall flow chart of a distribution transformer load forecasting method considering population flow in communication base stations provided by one embodiment of the present invention.

[0040] Figure 2 This is a computer equipment diagram of the distribution transformer load prediction method considering the population flow of communication base stations of the present invention. DETAILED DESCRIPTION

[0041] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0042] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0043] Example 1

[0044] Reference Figure 1 , which is an embodiment of the present invention, provides a distribution transformer load prediction method considering the population flow of communication base stations, comprising:

[0045] S1: Acquire dynamic data in a target power distribution area, analyze and summarize a first prediction model, perform a first prediction on the dynamic data through the first prediction model, and obtain a first prediction result.

[0046] Specifically, the dynamic data within the target distribution area includes real-time load, temperature, humidity, number of base station users, time, date type, special event markers, and power system operation status.

[0047] According to the dynamic data in the target distribution area, the real-time load of the distribution area is initially predicted based on the LSTM network. The influence of historical data and dynamic factors is considered at the same time. Finally, periodic changes are added to analyze and summarize the first prediction model. The specific process is as follows:

[0048] Step 1: Make a preliminary prediction of the real-time load in the distribution area based on the output of the LSTM network:

[0049]

[0050] in, is the preliminary load forecast value, which represents the preliminary estimate of the real-time load by the LSTM network based on historical data and current input; h t is the hidden state output of the LSTM network at time t; C t is the cell state output of the LSTM network at time t.

[0051] In a preferred embodiment, the output mode of the standard LSTM network is improved. The present invention uses the softplus function Process the hidden state, ensure that the output is positive, and use the error function erf(C t ) processing unit state and mapping it to the [-1,1] interval, which can better capture the nonlinear characteristics of load changes than directly using the output of LSTM. Such improvements have better prediction capabilities for load mutations caused by extreme weather or special events, show better stability in long-term predictions, and are less likely to have divergent prediction values; at the same time, they show faster convergence speed and better generalization ability during training, and are more robust to noise and outliers in input data; in addition, they can perform well in load forecasting tasks of different scales (such as communities, cities, and provincial power grids) and have better portability.

[0052] As shown in Table 1, it is a performance comparison table of the standard LSTM network and the LSTM network of the present invention in different load forecasting scenarios, which shows the superiority of the LSTM network of the present invention over the standard LSTM in various load forecasting scenarios. The LSTM network of the present invention shows lower prediction errors in all cases, especially when dealing with complex and drastically changing situations. For example, in extreme weather, holidays, emergencies, and fluctuations in new energy power generation, the prediction error reduction of the LSTM network of the present invention is usually between 20% and 50%. This performance improvement is not only reflected in the reduction of the absolute value of the error, but also in the ability to quickly adapt to abnormal situations, such as shortening the adaptation time on holidays from 1 to 2 days to 0.5 days. In short, the LSTM network of the present invention not only improves the prediction accuracy, but also enhances the adaptability and robustness of the model to various complex scenarios, which is of great significance for improving the operating efficiency and reliability of the power system.

[0053] Table 1 Performance comparison between the standard LSTM network and the LSTM network of the present invention in different load forecasting scenarios

[0054]

[0055]

[0056] Step 2: Based on the initial load forecast According to the influence of historical data, the first revised forecast value is obtained

[0057]

[0058] Among them, x i is a historical data point, used to calculate the similarity with the current moment; x t is the feature vector at the current moment; n is the number of historical data points used to calculate the similarity.

[0059] Preferably, the first revised prediction value The calculation improves the traditional similarity calculation method. The present invention uses a form similar to the radial basis function kernel, but uses it in the denominator. In this way, the prediction value can be reduced when the similarity between the current moment and the historical data is high, and vice versa, the prediction value can be increased, thereby capturing abnormal situations.

[0060] Step 3: Based on the first revised forecast value According to the influence of dynamic factors, the second revised forecast value is obtained

[0061]

[0062] Among them, P t is the population mobility factor at time t, reflecting the population mobility captured by the communication base station; T t is the temperature factor at time t, reflecting the impact of temperature on load; F t is the holiday factor at time t, reflecting the impact of holidays on load; S t is the system state at time t; S t-1 is the system state at time t-1; |S t -S t-1 | is the change amplitude of the system state; t is the time variable, the unit is hour.

[0063] Step 4: Based on the second revised forecast value Introducing the periodic function, we get the first prediction result L t :

[0064]

[0065] Based on the initial load forecast First revised forecast value Second revised forecast value Get the complete first prediction model:

[0066]

[0067] Among them, L t is the predicted load value at time t, i.e., the first prediction result; h t is the hidden state output of the LSTM network at time t; C t is the cell state output of the LSTM network at time t; x i is a historical data point, used to calculate the similarity with the current moment; x t is the feature vector at the current moment; n is the number of historical data points used to calculate similarity; P t is the population mobility factor at time t, reflecting the population mobility captured by the communication base station; T t is the temperature factor at time t, reflecting the impact of temperature on load; F t is the holiday factor at time t, reflecting the impact of holidays on load; S t is the system state at time t; S t-1 is the system state at time t-1; |S t -S t-1 | is the amplitude of change of the system state; t is the time variable, in hours; π is the pi constant.

[0068] It should be noted that the real-time load L t The output of the first prediction model refers to the actual power load of the target distribution area at time t. It is the most direct dynamic data, usually in kilowatts (kW) or megawatts (MW). Temperature and humidity are included in the temperature factor T t The number of base station users is used to calculate the population mobility factor P t , refers to the number of users captured by the communication base station at a certain moment, reflecting the population activity in the area. Time t is the last term in the first prediction model Although time is continuous, it is usually expressed as discrete time points in the collection, such as one data point every hour or every 15 minutes. The date type includes information on whether it is a weekday, weekend, or holiday, which is used to calculate the holiday factor F. t Special event markers may affect the holiday factor F t The calculation of S is used to mark whether there are special events such as large-scale activities and events at the current moment. The operating status of the power system is reflected in S t and S t-1 (System status), such as the load rate of major substations, line status, etc.

[0069] L t The range of is positive real numbers. t The closer to the second preset threshold, the lower the load; when Lt When the load is between the second preset threshold and the first preset threshold, it means that the load is in the normal range, and the specific value represents the ratio relative to the historical average load; when L t When it is greater than the first preset threshold, it means that the load is higher than the historical average level, and the value indicates the multiple of the excess. t =0.5 means the current load is 50% of the historical average load, L t =1.2 indicates that the current load is 120% of the historical average load. In this embodiment, the second preset threshold value is less than the first preset threshold value, the first preset threshold value is 1, and the second preset threshold value is 0.

[0070] Preferably, the first prediction model in the present invention provides a more accurate and flexible method for load forecasting in the distribution area. Compared with the prior art, the first prediction model solves the following important technical problems: 1. Effective integration of multi-source data. The prior art usually finds it difficult to effectively integrate multiple different types of data. The first prediction model of the present invention organically combines multi-source data such as LSTM network output, historical data similarity, population flow, temperature, holiday factors, and system status. This integration is not a simple linear superposition, but captures the complex interaction between various factors through nonlinear functions and product relationships, thereby improving the accuracy of the prediction. 2. Sensitivity to abnormal situations. Traditional methods often perform poorly in dealing with abnormal situations. The first prediction model in the present invention significantly improves the sensitivity of the model to abnormal situations by introducing similarity calculation based on kernel functions. When the current input is significantly different from the historical data, the prediction result can be adjusted in time, which improves the prediction ability of emergencies or abnormal loads. 3. Unified processing of timing characteristics and periodicity. The prior art often treats timing characteristics and periodicity as independent components, resulting in increased model complexity and possible inconsistency. This formula captures the timing characteristics through the LSTM network and directly introduces periodic functions into the model, which realizes the unified processing of timing characteristics and periodicity, simplifies the model structure and improves the consistency of prediction. 4. Interpretability and adjustability of the model. Although many advanced machine learning models such as deep neural networks have excellent performance, they often lack interpretability. Each component of the first prediction model in the present invention has a clear physical meaning, so that those skilled in the art can understand the decision-making process of the model and make targeted adjustments according to actual conditions.

[0071] S2: Obtain static data in the target power distribution area, analyze and summarize a second prediction model, perform a second prediction on the static data using the second prediction model, and obtain a second prediction result.

[0072] Specifically, static data includes substation capacity, distribution area, voltage stability, and user type.

[0073] According to the static data in the target distribution area, the preliminary load estimation is obtained according to the substation capacity and the area of ​​the distribution area, and the second prediction model is obtained by combining the voltage change rate and user type. The specific process is as follows:

[0074] Step 1: Get the initial load estimate L1 based on the substation capacity and the target distribution area:

[0075]

[0076] Where C is the substation capacity; A is the target distribution area; A ref is the reference area.

[0077] Step 2: Based on the preliminary load estimate and the voltage change rate in the target distribution area, obtain the preliminary load estimate L2 after considering voltage stability:

[0078]

[0079] in, It is the voltage change rate, which is used to reflect the change of voltage over time.

[0080] Step 3: Based on the preliminary load estimation after considering voltage stability and the user types in the target distribution area, obtain the basic load forecast, that is, the second forecast result L base :

[0081]

[0082] Among them, i, c, and r represent industrial, commercial, and residential user types respectively; P i is the proportion of each type of users, P i +P c +P r =1; E i is the electricity efficiency factor of each type of user, reflecting the electricity consumption characteristics of different types of users; L i,hist is the historical load data of various users; L ref is the reference load value.

[0083] The complete second prediction model is obtained based on the preliminary load estimate L1 and the preliminary load estimate L2 after considering voltage stability:

[0084]

[0085] Among them, L base is the predicted basic load value, i.e., the second prediction result; C is the substation capacity, which may be in MVA or kVA; A is the target distribution area; A refis the reference area, which is used to normalize the area impact. It can be selected according to the actual situation. For example, the average area of ​​the distribution network can be used. λ is the voltage stability influencing factor, which is used to adjust the impact of voltage changes on the load. is the voltage change rate, which reflects the change of voltage over time; the unit can be V / s or kV / s; i, c, r represent industrial, commercial and residential user types respectively; P i is the proportion of each type of users, P i +P c +P r =1; E i is the electricity efficiency factor of each type of user, reflecting the electricity consumption characteristics of different types of users; L i,hist is the historical load data of various users; L ref is the reference load value, which is used to normalize the historical load data. It can be selected as the average or maximum value of the historical load. π is a mathematical constant, which is approximately equal to 3.14159.

[0086] Preferably, the second prediction model has the following improvements over the prior art: by introducing the voltage change rate to reflect the impact of grid state changes on the load in real time, voltage stability is a key factor in the distribution network that takes into account the population flow of communication base stations; the load characteristics of different types of users are distinguished in detail, allowing more accurate load analysis, and the inverse tangent function is used to simulate the load characteristics of different types of users, taking into account the saturation effect of load growth, which is closer to the actual situation than the traditional linear weighted method; the square root relationship is used to reflect the impact of the distribution area area, taking into account the marginal effect; through the inverse tangent function, the historical load data is cleverly integrated into the model, which not only takes into account the historical trend, but also avoids over-reliance on historical data.

[0087] S3: Combining the first prediction result and the second prediction result to obtain a third prediction result, and completing the distribution transformer load prediction of the population flow of the communication base station.

[0088] Specifically, the weighted average method is used to fuse the first prediction result and the second prediction result, and some dynamic adjustment factors are introduced to optimize the fusion effect. The third prediction result is defined as L final :

[0089] L final =γ·L t +(1-γ)·L base K;

[0090] Among them, K is the adjustment factor; γ is the dynamic weight factor, 0≤γ≤1.

[0091] Furthermore, the dynamic weight factor γ is dynamically adjusted based on the historical performance of the first prediction model and the second prediction model and is expressed by the following formula:

[0092]

[0093] Among them, RMSE t and RMSE base are the root mean square errors of the first prediction model and the second prediction model in the recent period of time.

[0094] Furthermore, the adjustment factor K is used to adjust L base Make it consistent with L t It is more closely matched in magnitude, and can also introduce some additional dynamic features:

[0095]

[0096] Among them, β is a small adjustment parameter (such as 0.1); L t,avg and L base,avg is the average prediction value of the first prediction model and the second prediction model within a preset time; The term introduces daily periodic fluctuations.

[0097] In summary, the present invention achieves a more comprehensive and accurate load forecast by combining the prediction results of dynamic data and static data, and can better capture the trends and characteristics of load changes. Through the collaborative work of the dynamic data prediction module and the static data prediction module, and the intelligent integration of the result fusion module, the present invention significantly improves the accuracy and reliability of distribution transformer load forecasting, provides more accurate decision support for power grid operation and management, and helps to improve the operating efficiency and stability of the power system.

[0098] Example 2

[0099] Reference Figure 2 , is an embodiment of the present invention, which provides a distribution transformer load prediction system that takes into account the population flow of communication base stations, including: a dynamic data prediction module, used to obtain dynamic data in a target distribution area, analyze and summarize a first prediction model, perform a first prediction on the dynamic data through the first prediction model, and obtain a first prediction result; a static data prediction module, used to obtain static data in a target distribution area, analyze and summarize a second prediction model, perform a second prediction on the static data through the second prediction model, and obtain a second prediction result; a result fusion module, used to combine the first prediction result and the second prediction result to obtain a third prediction result, and complete the distribution transformer load prediction of the population flow of the communication base station.

[0100] Example 3

[0101] It is an embodiment of the present invention, which is different from the previous embodiment in that:

[0102] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0103] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0104] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0105] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0106] Example 4

[0107] As an embodiment of the present invention, a distribution transformer load forecasting method considering the population flow of communication base stations is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0108] The distribution transformer load data of a central business district of a provincial capital city for one year (January 1, 2023 to December 31, 2023) is selected. The area is about 25 square kilometers, including 5 10kV substations and 50 distribution transformers.

[0109] The dynamic data obtained include: real-time load (once every 15 minutes), temperature, humidity, wind speed (once every hour), number of communication base station users (once every hour, from 3 major operators), time, date type (weekdays / weekends / holidays), special event markers (such as large-scale events, major competitions, etc.), and power system operation status (equipment maintenance, failures, etc.).

[0110] The static data obtained include: substation capacity, distribution area, voltage stability index, and user type distribution (industrial, commercial, and residential proportions).

[0111] In this example, the present invention is compared with ARIMA: Autoregressive Integrated Moving Average Model, SVR: Support Vector Regression and LSTM: Long Short-Term Memory Network. The evaluation indicators are mean absolute percentage error (MAPE), root mean square error (RMSE), determination coefficient (R 2 ), and the maximum absolute error (MaxAE). The prediction period is ultra-short term: 1 hour, short term: 24 hours, medium term: 7 days, and long term: 30 days. The experimental scenarios are regular scenarios: randomly selecting 20 time points in the annual data for prediction, holiday scenarios: selecting important holidays such as the Spring Festival and National Day for prediction, extreme weather scenarios: selecting high temperatures in summer and cold waves in winter for prediction, and large-scale event scenarios: selecting large-scale events such as commercial promotions and sports events for prediction.

[0112] Then, the data from January 1 to November 30, 2023 is used for model training, and the data from December 1 to December 31, 2023 is used for model testing and evaluation.

[0113] Table 2 Performance comparison of different methods in each prediction cycle

[0114]

[0115] Table 2 Prediction error of different methods in special scenarios (MAPE, %)

[0116] method Holidays Extreme Weather Large events ARIMA 12.35 15.67 18.92 SVR 10.78 13.42 16.35 LSTM 8.64 10.89 13.27 Method of the present invention 6.43 8.21 9.76

[0117] Table 3 Comparison of computational efficiency (average single prediction time, seconds)

[0118] method Ultra short term (1h) Short term (24h) Mid-term (7 days) Long term (30 days) ARIMA 0.15 0.32 0.68 1.25 SVR 0.23 0.47 0.95 1.86 LSTM 0.31 0.62 1.28 2.45 Method of the present invention 0.28 0.56 1.15 2.23

[0119] As can be seen from Table 1, the present invention is significantly better than other methods in all forecasting periods. Especially in the short-term (24h) forecast, the MAPE of the present invention is 3.21%, which is 44.8%, 35.4% and 22.1% lower than ARIMA, SVR and LSTM respectively, indicating that the present invention can more accurately capture the trend and characteristics of load changes, especially when considering the population flow of communication base stations. As the forecast period increases, the forecast errors of all methods increase, but the error growth of the present invention is the smallest. From ultra-short-term to long-term forecasts, the MAPE of the present invention increased by 4.27 percentage points, while ARIMA, SVR and LSTM increased by 6.18, 5.90 and 5.22 percentage points respectively, indicating that the present invention has better long-term forecasting ability and stability. The R of the present invention in each forecast period 2 The values ​​of R are significantly higher than those of other methods, indicating that the present invention has a stronger ability to explain load changes and a higher degree of model fitting, especially in short-term prediction. 2 The value reaches 0.945, which is much higher than other methods. By comparing the MaxAE index, we can see that the performance of the present invention in predicting load peak or valley is also better than other methods, which shows that the present invention can not only accurately predict general conditions, but also better capture extreme load conditions.

[0120] As shown in Table 2, the prediction error of the present invention is significantly lower than that of other methods in special cases such as holidays, extreme weather and large-scale events. For example, in the large-scale event scenario, the MAPE of the present invention is 9.76%, which is 48.4%, 40.3% and 26.4% lower than ARIMA, SVR and LSTM, respectively, proving that the present invention takes into account dynamic factors such as population mobility of communication base stations and can better adapt to complex and changing load environments.

[0121] As can be seen from Table 3, although the calculation time of the present invention is slightly longer than that of ARIMA and SVR, it is comparable to that of LSTM, and considering the significant improvement in prediction accuracy, this slight increase in time is acceptable. In particular, in practical applications, the improvement in prediction accuracy may bring greater economic and social benefits.

[0122] It should be noted that the above embodiments are only used to illustrate the technical solutions 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 solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A distribution transformer load prediction method considering the population flow of communication base stations, characterized in that: include: Acquire dynamic data in a target power distribution area, analyze and summarize a first prediction model, perform a first prediction on the dynamic data using the first prediction model, and obtain a first prediction result; Obtaining static data in the target power distribution area, analyzing and summarizing a second prediction model, and performing a second prediction on the static data using the second prediction model to obtain a second prediction result; The third prediction result is obtained by combining the first prediction result and the second prediction result to complete the distribution transformer load prediction of population flow in the communication base station.

2. The method for predicting distribution transformer load considering population flow of communication base stations according to claim 1, characterized in that: The dynamic data includes real-time load, temperature, humidity, number of base station users, time, date type, special event markers, and power system operating status; the static data includes substation capacity, distribution area, voltage stability, and user type.

3. The method for load forecasting of distribution transformers considering population flow of communication base stations as claimed in claim 2, characterized in that: The first prediction model is a preliminary prediction of the real-time load in the distribution area based on the LSTM network. The first revised prediction value is obtained according to the influence of historical data and the second revised prediction value is obtained according to the influence of dynamic factors. The periodic changes are added to obtain the following: Based on the output of the LSTM network, a preliminary forecast of the real-time load in the distribution area is made to obtain a preliminary load forecast value. Based on preliminary load forecasts According to the influence of the historical data, the first revised forecast value is obtained Based on the first revised prediction value According to the influence of dynamic factors, the second revised forecast value is obtained Based on the second revised prediction value Introducing the periodic function, we get the first prediction result L t .

4. The method for load forecasting of distribution transformers considering population flow of communication base stations as claimed in claim 3, characterized in that: The initial load forecast value It is expressed by the following formula: in, is the initial load forecast value; h t is the hidden state output of the LSTM network at time t; C t is the cell state output of the LSTM network at time t; The first corrected predicted value It is expressed by the following formula: Among them, x i is a historical data point, used to calculate the similarity with the current moment; x t is the feature vector at the current moment; n is the number of historical data points used to calculate similarity; The second corrected predicted value It is expressed by the following formula: Among them, P t is the population mobility factor at time t, reflecting the population mobility captured by the communication base station; T t is the temperature factor at time t, reflecting the impact of temperature on load; F t is the holiday factor at time t, reflecting the impact of holidays on load; S t is the system state at time t; S t-1 is the system state at time t-1; |S t -S t-1 | is the change amplitude of the system state; t is the time variable, the unit is hour; The first prediction result L t It is expressed by the following formula: Wherein, π is the pi constant; the first prediction result L t The value range is a positive real number. When the first prediction result L t When the second preset threshold is between the first preset threshold and the second preset threshold, it indicates that the load is in a normal range, and the specific value represents the ratio relative to the historical average load. t When it is greater than the first preset threshold, it means that the load is higher than the historical average level, and the specific value represents the multiple of the excess; the second preset threshold is less than the first preset threshold.

5. The method for load forecasting of distribution transformers considering population flow of communication base stations as claimed in claim 4, characterized in that: The second prediction model is to obtain a preliminary load estimate based on the substation capacity and the area of ​​the distribution area, combined with the voltage change rate and user type, specifically: The preliminary load estimate L1 is obtained based on the substation capacity C and the target distribution area area A; Based on the preliminary load estimate L1 and the voltage change rate in the target distribution area, a preliminary load estimate L2 is obtained after considering voltage stability; Based on the preliminary load estimate L2 after considering voltage stability and the user types in the target distribution area, a second prediction result L is obtained. base .

6. The method for load forecasting of distribution transformers considering population flow of communication base stations as claimed in claim 5, characterized in that: The preliminary load estimate L1 is expressed by the following formula: Where C is the substation capacity; A is the target distribution area area; A ref is the reference area; The preliminary load estimation L2 after considering voltage stability is expressed by the following formula: in, is the voltage change rate, which is used to reflect the change of voltage over time; The second prediction result L base It is expressed by the following formula: Among them, i, c, and r represent industrial, commercial, and residential user types respectively; P i is the proportion of each type of users, P i +P c +P r =1; E i is the electricity efficiency factor of each type of user, reflecting the electricity consumption characteristics of different types of users; L i,hist is the historical load data of various users; L ref is the reference load value.

7. The method for load forecasting of distribution transformers considering population flow of communication base stations according to claim 6, characterized in that: Combining the first prediction result and the second prediction result to obtain a third prediction result includes: The first prediction result and the second prediction result are fused using a weighted average method: L final =γ·L t +(1-γ)·L base ·K; Among them, γ is the dynamic weight factor, 0≤γ≤1; K is the adjustment factor; L final This is the third prediction result.

8. A system using the method for predicting distribution transformer load considering the population flow of communication base stations as claimed in any one of claims 1 to 7, characterized in that: include: A dynamic data prediction module, used to obtain dynamic data in a target power distribution area, analyze and summarize a first prediction model, perform a first prediction on the dynamic data through the first prediction model, and obtain a first prediction result; A static data prediction module, used for acquiring static data in the target power distribution area, analyzing and summarizing a second prediction model, and performing a second prediction on the static data by using the second prediction model to obtain a second prediction result; The result fusion module is used to combine the first prediction result and the second prediction result to obtain a third prediction result, so as to complete the distribution transformer load prediction of the population flow of the communication base station.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the distribution transformer load forecasting method considering the population flow of communication base stations described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the distribution transformer load forecasting method considering the population flow of communication base stations described in any one of claims 1 to 7 are implemented.

Images