Public Transformer Load Forecasting Method Based on New Business Expansion Data

Abstract

The present invention discloses a public transformer load forecasting method based on service expansion data, which includes step S1: cleaning the obtained preliminary service expansion data through first information to screen out the substantial service expansion data that affects electricity consumption, so as to analyze the characteristics of public transformer load changes; step S2: establishing a public transformer load forecasting model based on the load rate characteristics of various types of users according to the substantial service expansion data, so as to obtain public transformer load forecasting data through the public transformer load forecasting model. The public transformer load forecasting method based on service expansion data disclosed by the present invention, through in-depth mining of public transformer historical and existing data, strives to achieve accurate and effective forecasting of public transformer load, better ensure equipment safety, assist in the accurate operation and maintenance of public transformers and the accurate investment of distribution networks, improve the utilization rate of distribution transformers, and enhance the reliability of customer electricity consumption.

Description

Technical Field

[0001] The present invention belongs to the technical field of public transformer load forecasting, and particularly relates to a public transformer load forecasting method based on service expansion data. Background Art

[0002] Affected by factors such as the improvement of people's living standards, rapid economic development, industrial distribution, industry transformation, policy changes, urban relocation and renovation, the growth and change of electricity load show diversification, and it is difficult to grasp the regular characteristics; inaccurate load forecasting will bring relatively serious consequences. For example, if the growth rate forecast is too high, it will lead to many public transformer capacity expansion projects, insufficient utilization rate of newly added transformers, and investment waste; if the growth rate forecast is too low, it will cause insufficient power supply capacity of public transformers, overloading or even overloading operation, which not only threatens the safety of equipment, but also leads to frequent power outages, seriously affecting the lives of residents.

[0003] With the advancement of the informatization construction of the distribution network, a large amount of power consumption and distribution data has been generated during the daily operation of the distribution network. However, these data have not been fully mined and effectively utilized. How to effectively utilize these data to promote the precise operation and maintenance construction of the distribution network and customer personalized services has become the focus of current data mining work.

[0004] Therefore, in view of the above problems, further improvements are made. Summary of the Invention

[0005] The main purpose of the present invention is to provide a public transformer load forecasting method based on service expansion data. Through in-depth mining of the historical and existing data of public transformers, it strives to achieve accurate and effective forecasting of public transformer loads, better ensure equipment safety, assist in the precise operation and maintenance of public transformers and the precise investment of the distribution network, improve the utilization rate of distribution transformers, and enhance the reliability of customer power consumption.

[0006] To achieve the above object, the present invention provides a public transformer load forecasting method based on service expansion data for forecasting the load of public transformers, including the following steps:

[0007] Step S1: Clean the obtained preliminary service expansion data through the first information to screen out the substantial service expansion data that affects power consumption, so as to analyze the change characteristics of the public transformer load;

[0008] Step S2: Establish a public transformer load forecasting model based on the load rate characteristics of various users according to the substantial service expansion data, so as to obtain public transformer load forecasting data through the public transformer load forecasting model.

[0009] As a further preferred technical solution of the above technical solution, step S2 is specifically implemented as the following steps:

[0010] Step S2.1: Calculate the historical load rates of different users under the same public transformer;

[0011] Step S2.2: Perform empirical fitting based on the obtained historical load rates of users to obtain the predicted public transformer load data for the predicted year.

[0012] As a further preferred technical solution of the above technical solution, step S2.1 is specifically implemented as the following steps:

[0013] Step S2.1.1: Calculate the load rates of users of different electricity consumption property types under a single public transformer according to the historical public transformer data over the years, the load data of the user acquisition system, and the business expansion data of the marketing system through the following formula:

[0014]

[0015] Among them, there are N users of the i-th type among all categories of users, the average load of the j-th user on a certain day is P ij , the electricity consumption capacity of this user is S ij , the electricity consumption capacity of the i-th type of user is S i , and the load rate of this user is obtained as K ij . Summing up all users in this category, the load rate of the i-th type of user can be obtained as K i ;

[0016] Step S2.1.2: Accumulate the load rates of this category of users in one year to obtain the annual electricity load rate curve of this category of users.

[0017] As a further preferred technical solution of the above technical solution, step S2.2 is specifically implemented as the following steps:

[0018] Step S2.2.1: Select the initial key frames of the annual electricity load rate curve through the twenty-four solar terms table, select and set the temperature key frames (preferably 5) symbolizing temperature changes, the precipitation key frames (preferably 7) reflecting precipitation changes, and the Spring Festival key frame (New Year's Eve of the lunar calendar) reflecting the load changes during the Spring Festival, and apply the temperature key frames, precipitation key frames, and Spring Festival key frames to the preliminary matching mapping in the historical electricity load rate curves over the years to reduce the oscillation differences in the load rates in the same historical period;

[0019] Step S2.2.2: Establish a corresponding relationship between the load rates between key frames through interpolation to obtain the predicted load rate curve;

[0020] Step S2.2.3: Optimize the corresponding relationship between the load rates between key frames through the genetic algorithm, verify the selection of key frames by calculating the variance between the corresponding fitting points, establish the mapping relationship between the optimized fitting points, and determine the empirical key frames for the predicted year by taking the average of the historical key frames over the years, and then fit to obtain the empirical load rate table for the predicted year;

[0021] Step S2.2.4: Perform public transformer load prediction for the corresponding period through the empirical load rate table to obtain the public transformer load prediction value matrix including each day in the prediction year.

[0022] As a further preferred technical solution of the above technical solution, in step S2.2.4, according to the known public transformer load data in the prediction year, the calculation error rate in the public transformer load prediction model is verified and compared, so as to feedback and modify the public transformer load prediction model.

[0023] To achieve the above object, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the public transformer load prediction method based on service expansion data are implemented.

[0024] To achieve the above object, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the public transformer load prediction method based on service expansion data are implemented.

[0025] The beneficial effects of the present invention are as follows:

[0026] 1. Problem identification

[0027] 1.1 The prediction model is simple and effective

[0028] For the same type of distribution transformer load prediction model based on service expansion data, it is necessary to use growth curves, etc. to simulate the impact of service expansion increments on the load, and at the same time, multi-dimensional factor variables such as service expansion types and industry characteristics need to be considered. However, the present invention uses a prediction method based on fitting the existing service expansion capacity and historical load rate, effectively reducing the uncertain factors and uncontrollable variables in model prediction. Since the selected load research object is the user, the impacts of multi-factors such as meteorological factors, holidays, and production cycles on electricity consumption are actually reflected in the daily load of users, simplifying the prediction model and handing over a large amount of data operation and analysis work to the machine for processing.

[0029] 1.2 Facilitate short-term prediction of public transformer load

[0030] The load prediction of the present invention is based on service expansion data and requires real-time service expansion data as support. And the actual daily service expansion data is constantly changing, and the increase or decrease of low-voltage users under each public transformer directly affects the load of the public transformer. Usually, the change degree of the increase or decrease, capacity increase or decrease, etc. of low-voltage users under the public transformer is limited in the short term, which is conducive to predicting short-term load.

[0031] 1.3 The prediction object is refined

[0032] Different from traditional distribution transformer load prediction methods, the present invention does not start from predicting the overall change trend of the distribution transformer load. Instead, it focuses on the load of classified users with different electricity consumption characteristics under the distribution transformer, "differentiates" the distribution transformer load into "elementary loads", and then classifies and predicts the "elementary loads" of different types of users and performs "integration" to achieve more refined and accurate distribution transformer load prediction.

[0033] 1.4 Facilitate the integration of operation and distribution data

[0034] By using big data analysis means, integrate operation and distribution data. Through analyzing information such as the electricity consumption category, industry classification, electricity consumption capacity, and metering point of the collected users, combined with the distribution transformer equipment capacity (distribution) information in the metering point, build a distribution transformer load prediction model based on historical and new data to achieve prediction and analysis of the size, change law, cycle characteristics, persistence, etc. of the distribution transformer load, and identify and reflect the problems of non-integration of operation and distribution such as information asymmetry and asynchronous work between business installation and distribution network operation and maintenance.

[0035] 2. Risk control

[0036] Apply the prediction model, use big data algorithms to predict the distribution transformer load in advance, realize the transformation from manual analysis to intelligent analysis, and provide scientific support for the precise operation and maintenance of the distribution transformer. It can be combined with the early warning model to establish a hierarchical early warning mechanism, promote prediction and early warning simultaneously, reduce the occurrence of heavy overload of the distribution transformer, avoid faults caused by heavy overload of the distribution transformer, and improve the production safety factor and the level of quality service.

[0037] 3. Auxiliary decision-making

[0038] The most ideal data for this invention is load data, that is, a load value every 15 minutes collected by the terminal, which can better reflect the load distribution at different times of a day such as peaks and power outages. However, since the low-voltage user terminals are not fully covered at present and some have been installed for a short time, the load data volume of low-voltage users in the power consumption acquisition system is incomplete and cannot be compared and analyzed with historical values. In the case of limited data volume, the present invention calculates the daily average load using the daily electricity consumption of users, and the load peaks of actual users are averaged, which cannot clearly reflect the predicted load peak of the public transformer. As the data is gradually improved and accumulated, models such as maximum load prediction and real-time load prediction can be gradually incorporated to optimize the model and make the prediction more accurate.

[0039] Under the background of the deepening of the current power system reform and marketization process, as an enterprise, the profitability of power supply companies has become an important indicator to measure the operation level of enterprises. The accurate prediction of distribution transformer load helps to provide accurate data support and decision-making basis for the project establishment of power grid invention transformation and the overhaul of distribution network technical transformation, so that operation and maintenance personnel can take measures such as load segmentation and new point layout to improve the capacity utilization rate of distribution transformers, realize the reasonable and effective allocation of distribution transformer capacity resources, and lay a foundation for the accurate investment of distribution network funds.

[0040] 4. Popularization and application

[0041] Distribution transformers are the most direct power consumption equipment facing users and also important nodes connecting the power grid and the user side. The load prediction model of the present invention selects research individuals covering households, and the prediction object penetrates into each type of user. It is based on the in-depth analysis and mining of a large amount of historical and current data, and is suitable for the popularization and application of all public transformer load predictions in the province, with the universality of data value mining and the practicability of load prediction.

[0042] The accurate prediction of distribution transformer load data can provide an effective basis for power supply enterprises to take measures such as serving small and micro enterprises, optimizing power consumption services, and helping to create a good business environment in advance. At the same time, extending from the distribution transformer to the line and from the line to the substation (office), the accurate prediction of distribution transformer load can provide a more accurate basis for line load prediction and operation mode adjustment, and also lay a foundation for the accuracy of main grid load prediction. Brief description of the drawings

[0043] Figure 1 It is a model schematic diagram of the public transformer load prediction method based on business expansion data of the present invention.

[0044] Figure 2 It is the twenty-four solar terms diagram of the public transformer load prediction method based on business expansion data of the present invention.

[0045] Figure 3 It is the set initial key frame diagram of the public transformer load prediction method based on business expansion data of the present invention.

[0046] Figure 4 It is the correspondence diagram established by interpolation method of the public transformer load prediction method based on business expansion data of the present invention.

[0047] Figure 5 It is the schematic diagram of determining the key frame by using the minimum variance between corresponding initial key frames of the public transformer load prediction method based on business expansion data of the present invention. Detailed implementation manners

[0048] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other implementation schemes, variation schemes, improvement schemes, equivalent schemes, and other technical schemes that do not depart from the spirit and scope of the present invention.

[0049] In the preferred embodiment of the present invention, those skilled in the art should note that the users, electronic devices, etc. involved in the present invention can be regarded as the prior art.

[0050] Preferred embodiment.

[0051] The electricity consumption of any electricity customer must be based on a certain capacity, that is, the capacity is an important factor restricting the customer's electricity demand. The capacity includes the existing capacity and the newly added capacity (capacity expansion for new business). The change in electricity demand is jointly composed of the change in the utilization of the existing capacity and the utilization of the capacity expansion for new business. The existing capacity is the basis of the electricity market, and its demand directly determines the base value of the load demand; while the capacity expansion for new business has an absolute impact on the future load. Based on the prediction method combining the existing capacity and the application for capacity expansion, the present invention aims to start from the root cause of electricity consumption, connect the data of the distribution network marketing and operation and maintenance systems, analyze the impact of the user's electricity consumption nature on the public transformer load through the mining of the data of the customers applying for capacity expansion, consider summarizing the historical data of the public transformers in recent years, and carry out the mining and prediction research on the data of the public transformer load, so as to provide a scientific judgment basis for the changing trend of the public transformer load.

[0052] In this embodiment, 2017 and 2018 are used as historical years, and 2019 is used as the prediction year.

[0053] The present invention discloses a method for predicting the load of a public transformer based on the data of capacity expansion for new business, which is used to predict the load of the public transformer and includes the following steps:

[0054] Step S1: Clean the obtained preliminary data of capacity expansion for new business through the first information to screen out the substantial data of capacity expansion for new business that affects electricity consumption, so as to analyze the change characteristics of the load of the public transformer;

[0055] Preferably, the data in the present invention comes from the system data table, and the data such as user information and daily electricity consumption really exist, with fewer breakpoints and bad data, so there is no need for too much preprocessing work. However, for the data of capacity expansion for new business which is mainly studied in the present invention, considering that not all the capacity expansion processes that occur, such as meter rotation and meter relocation, will affect the load, so the information such as the cancellation time, applied contract capacity, original contract capacity, and total contract capacity in the above data table will be combined to clean the massive data of capacity expansion for new business, leaving the part of the data of capacity expansion for new business that really affects electricity consumption, and analyzing the load change characteristics.

[0056] Step S2: Establish a public transformer load prediction model based on the load rate characteristics of various types of users according to the actual business expansion data, so as to obtain public transformer load prediction data through the public transformer load prediction model.

[0057] Preferably, the present invention considers the impact of business expansion and installation on the load, calculates and fits the load rates of 23 types of users under the same public variable, considers the real-time operating capacity of each type of user, and establishes a load prediction model based on the load rate characteristics of each type of user.

[0058] Specifically, step S2 is implemented as follows:

[0059] Step S2.1: Calculate the historical load rates of different users under the same public variable;

[0060] Step S2.2: Perform empirical fitting based on the historical load rates of the users, thereby obtaining the public variable load forecast data for the forecast year.

[0061] It is worth mentioning that for the empirical K value fitting, the present invention adopts a genetic algorithm model with a "key frame". After calculating the historical K value matrix of each year and establishing a preliminary one-to-one correspondence, the initial key frame is set, the initial key frame is optimized by the genetic algorithm, and the empirical key frame is determined according to the adjustment plan of the historical key frame to improve the operating efficiency of the genetic algorithm. Then, the interpolation method is used to perform the optimal corresponding mapping of the historical K values ​​between the corresponding key frames, and finally the trend analysis of the K values ​​of the past years with the mapping relationship established is performed to obtain the empirical K value matrix.

[0062] Specifically, step S2.1 is implemented as follows:

[0063] Step S2.1.1: Calculate the load rate of users with different types of electricity consumption under a single public transformer using the following formula based on the historical data of public transformers over the years, the load data of the user collection system, and the business expansion data of the marketing system:

[0064]

[0065] Among all categories of users, there are N users in the i-th category, and the average load of the j-th user on a certain day is P ij , the user's power consumption capacity is S ij , the power consumption capacity of the i-th user is S i , and the user load rate is K ij , summing up all users in this category, we can get the load rate of the i-th category user to be K i ;

[0066] Step S2.1.2: Accumulate the load rate of the users of this category for one year to obtain the annual electricity load rate curve of the users of this category.

[0067] It is worth mentioning that for the calculation of historical K values, even under the same industry category and the same type of electricity consumption, the electricity loads reflected by individual low-voltage users still have significant randomness and differences. However, the production, life, and other social activities of users in the same category will tend to be the same. Therefore, the distribution characteristics reflected by the electricity load within the annual cycle will tend to be similar. According to the public transformer historical data, user acquisition system, and marketing system in 2017 and 2018 (preferably the two years before the prediction year as historical years), calculate the K values (the K value is the load rate of the user, that is, the ratio of the actual load of the user's meter to the electricity application capacity of the user) of 23 types of users selected according to the electricity consumption nature type under a single public transformer.

[0068] More specifically, step S2.2 is specifically implemented as the following steps:

[0069] Step S2.2.1: Select the initial key frames of the annual electricity load rate curve through the twenty-four solar terms table, select and set the temperature key frames (preferably 5) symbolizing temperature changes, the precipitation key frames (preferably 7) reflecting precipitation changes, and the Spring Festival key frame (Lunar New Year's Eve) reflecting the load changes during the Spring Festival, and apply the temperature key frames, precipitation key frames, and Spring Festival key frames to the preliminary matching mapping in the historical electricity load rate curves of previous years to reduce the oscillation differences of the load rate in the same historical period;

[0070] Among them, as Figure 2 and Figure 3 shown, in the present invention, the K value on the same day will be affected by factors such as temperature, solar terms, and holidays. If the K values on the same day in three years are fitted to predict the K value of the next year, there will be a large difference. Therefore, in this model, the method of setting the main key frames of the K value curve is adopted, which is defined by using the twenty-four solar terms of the lunar calendar that can reflect human production and life, to find and set 5 "temperature key frames" symbolizing temperature changes, 7 "precipitation key frames" reflecting precipitation changes, and the "Spring Festival key frame" (Lunar New Year's Eve) reflecting the load changes during the Spring Festival, and apply them to the preliminary matching mapping in the historical K value curves of previous years to reduce the oscillation differences of the K value in the same historical period.

[0071] Step S2.2.2: (As Figure 4 shown) Establish the corresponding relationship between the load rates between key frames through interpolation to obtain the predicted load rate curve;

[0072] Step S2.2.3: Optimize the corresponding relationship between the load rates between key frames through the genetic algorithm, verify the selection of key frames by calculating the variance between the corresponding fitting points, establish the mapping relationship between the optimized fitting points, and determine the empirical key frames of the prediction year by taking the average of the historical key frames of previous years, and then fit to obtain the empirical load rate table of the prediction year;

[0073] Preferably, as Figure 5 shown, by calculating the variance between corresponding fitting points to verify the selection of key frames, and finally find the global optimal solution to establish the mapping relationship between the fitting points closest to the trend change. And determine the empirical key frames for 2019 by taking the mean of the adjustment schemes of historical key frames.

[0074]

[0075] Based on the above historical K-value tables for 2017 and 2018, taking 2018 as the benchmark, obtain the matrix M' of 2017 after key frame fitting 2017 , after calculation, the empirical K-value table for the predicted year 2019 can be fitted.

[0076] Step S2.2.4: Perform the public variable load prediction for the corresponding period through the empirical load rate table to obtain the matrix of public variable load prediction values including each day in the predicted year.

[0077] Furthermore, in step S2.2.4, according to the known public variable load data in the predicted year, verify and compare the calculation error rate in the public variable load prediction model, and then perform feedback modification on the public variable load prediction model.

[0078] According to the fitted historical K-value empirical table, carry out the public variable load prediction for the corresponding period, and a matrix of public variable load prediction values for 365 points in a year can be obtained.

[0079]

[0080]

[0081] According to the known public variable load data in the first half of 2019, the error rate of the prediction model calculation can be verified and compared, and the model can be corrected by feedback.

[0082] The present invention also discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the public variable load prediction method based on service expansion data are implemented.

[0083] The present invention also discloses a non-transitory computer-readable storage medium, on which a computer program is stored. The computer program is characterized in that when it is executed by a processor, the steps of the public variable load prediction method based on service expansion data are implemented.

[0084] It is worth mentioning that technical features such as users and electronic devices involved in this invention patent application should be regarded as prior art. For the specific structures, working principles, possible control methods, and spatial arrangement methods of these technical features, conventional selections in the art can be adopted, and they should not be regarded as the inventive points of this invention patent. This invention patent will not be further elaborated specifically.

[0085] For those skilled in the art, it is still possible to modify the technical solutions described in the foregoing embodiments or perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A public transformer load forecasting method based on service expansion data, which is used to forecast the public transformer load, is characterized in that Including the following steps: Step S1: Clean the obtained preliminary service expansion data through the first information to screen out the substantial service expansion data that affects power consumption, so as to analyze the characteristics of public transformer load changes; Step S2: Establish a public transformer load prediction model based on the load rate characteristics of various types of users according to the substantial service expansion data, so as to obtain public transformer load prediction data through the public transformer load prediction model; The specific implementation of Step S2 is as follows: Step S2.1: Calculate the historical load rates of different users under the same public transformer; Step S2.2: Perform empirical fitting through the obtained historical load rates of users to obtain the public transformer load prediction data for the predicted year; The specific implementation of Step S2.2 is as follows: Step S2.2.1: Select the initial key frames of the annual power consumption load rate curve through the twenty-four solar terms table, select and set the temperature key frames symbolizing temperature changes, precipitation key frames reflecting precipitation changes, and Spring Festival key frames reflecting Spring Festival load changes, and apply the temperature key frames, precipitation key frames, and Spring Festival key frames to the preliminary matching mapping in the historical annual power consumption load rate curve to reduce the oscillation difference of the load rate in the same historical period; Step S2.2.2: Establish a corresponding relationship between the load rates between key frames through interpolation to obtain the predicted load rate curve; Step S2.2.3: Optimize the corresponding relationship between the load rates between key frames through the genetic algorithm, calculate the variance between the corresponding fitting points, so as to verify the selection of key frames, establish the mapping relationship between the optimized fitting points, and determine the empirical key frames for the predicted year by taking the average of the historical key frames over the years, and then fit to obtain the empirical load rate table for the predicted year; Step S2.2.4: Perform public transformer load prediction for the corresponding period through the empirical load rate table to obtain the public transformer load prediction value matrix including each day in the predicted year.

2. The public transformer load forecasting method based on service expansion data according to claim 1, wherein The specific implementation of Step S2.1 is as follows: Step S2.1.1: Calculate the load rates of users of different electricity consumption property types under a single public transformer according to the historical public transformer data over the years, the load data of the user acquisition system, and the service expansion data of the marketing system through the following formula: Among them, there are N users in the i-th category among all categories of users, and the average load of the j-th user on a certain day is P ij , and the electricity consumption capacity of this user is S ij , and the electricity consumption capacity of the i-th category of users is S i , and the load factor of this user is obtained as K ij , summing up all users in this category, the load factor of the i-th category of users can be obtained as K i ; Step S2.1.2: Accumulate the load rates of this category of users in one year to obtain the annual power consumption load rate curve of this category of users.

3. A public transformer load forecasting method based on service expansion data according to claim 2, characterized in that, In Step S2.2.4, according to the known public transformer load data in the predicted year, verify and compare the calculation error rate in the public transformer load prediction model, so as to make feedback modifications to the public transformer load prediction model.

4. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the public transformer load prediction method based on service expansion data as described in any one of claims 1 to 3.

5. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the public transformer load prediction method based on service expansion data as described in any one of claims 1 to 3.

Images