A multi-step prediction method and system for city gas hourly load based on proportional splitting

Abstract

A multi-step forecasting method for hourly urban gas load based on proportional splitting includes: step S1: acquiring and organizing historical hourly gas load data for each supply point in a certain area; step S2: summing the gas loads of each supply point at each moment to obtain the total gas load for the area at each moment; step S3: recording the proportion of the load of each supply point to the total load at each moment; step S4: data preprocessing: outlier detection and elimination and missing value supplementation; step S5: constructing total gas load characteristics; step S6: training a single-step forecasting model for the total hourly gas load of the area; step S7: recursively predicting the total hourly gas load data for a certain period in the future; and step S8: proportionally splitting to obtain the hourly gas load data for each gas supply point for a certain period in the future. The present invention uses load data to simulate and perform calculations and status monitoring on the gas pipeline network at future moments, and is of great significance for guiding the work of gas personnel.

Description

Technical Field

[0001] The present invention relates to the field of urban gas transmission and distribution, and in particular to a method and system for multi-step prediction of urban gas hourly load based on proportional splitting. Background Art

[0002] Gas supply is closely linked to people's lives, and ensuring the stability and reliability of urban gas supply is a crucial livelihood issue. Gas pipeline network load forecasting is crucial for both planning and optimized operation. Specifically, within urban gas transmission and distribution systems, gas load forecasting provides an important basis for peak load regulation, resource allocation, and the management of gas storage facilities.

[0003] Gas load forecasting methods are developing rapidly. Domestic and international scholars have conducted extensive research and proposed numerous forecasting methods, ranging from early traditional mathematical models (such as time series models and regression forecasting models) to artificial intelligence forecasting methods (such as artificial neural networks and long-short-term memory networks), as well as optimized combinations of multiple intelligent algorithms. Despite this, achieving online gas load forecasting at low cost and high efficiency, and using gas load data to simulate and monitor future gas pipeline networks, remains a common vision for gas industry professionals.

[0004] City gas system networks are typically circular, with multiple gas supply points sharing responsibility for the same area. Existing gas forecasting methods are mostly based on scheduling purposes, predicting the total daily gas load for a specific area while ignoring the load of individual gas supply points. However, for real-time simulation purposes, the hourly load of individual gas supply points over a period of time is crucial for monitoring the pressure of the entire gas network. This real-time load data can be used to simulate and calculate the gas network status at future times, providing important information for gas dispatchers.

[0005] Therefore, developing a gas load prediction method for a ring network, predicting the hourly load of a single gas supply point in the future, and simulating and monitoring the status of the gas network at future moments based on the load data are of great significance for guiding the work of gas workers. Summary of the Invention

[0006] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a multi-step prediction method for city gas hourly load based on proportional splitting.

[0007] Step S1: Obtain and organize the historical hourly gas load data of each supply point in a certain area, including the time t and the load data Q at the corresponding moment t,n , and sort them in chronological order; where: t is the tth moment; n is the supply point number;

[0008] Step S2: Add the gas load of each supply point at each moment to obtain the total gas load Q of the area at each moment t =Q t,1 +Q t,2 +…+Q t,n ; Among them: Q t,1 is the gas load of supply point 1 at time t, and so on, Q t,n is the gas load at supply point n at time t. There are n gas supply points in the area. The total gas load at time t is the sum of the gas loads at each point at time t.

[0009] Step S3: Calculate and record the proportion of the load of each supply point to the total load at each moment

[0010] Step S4: performing data preprocessing on the collected load data;

[0011] Step S5: Construct the total gas load characteristics; wherein: the total gas load characteristics include: temperature, day of the week, hour, whether it is a holiday, daily maximum temperature, daily minimum temperature, load data Q of the previous hour t-1 , the load data Q at this time in the past two weeks t-24j and the load data Q at the previous moment t-24j-1 (j=1,2,…,14);

[0012] Step S6: training a single-step prediction model for the total hourly gas load of the area based on historical data;

[0013] Step S7: Use recursion to predict the total gas hourly load data (Q t+1 ,Q t+2 ,…,Q t+k ); where: k is the length of time for predicting future load;

[0014] Step S8: Split the predicted load data according to the proportion of the load data at the corresponding moment in the historical data to obtain the hourly gas load data of each gas supply point at the corresponding moment in the future t+k period.

[0015] Furthermore, in steps S1-S3, the historical hourly gas load data of each supply point in a certain area is obtained and sorted, and then added to obtain the total gas load, and then the proportion of the load of each supply point to the total load is calculated and recorded.

[0016] Furthermore, the data preprocessing is specifically: detecting and eliminating abnormal gas load data, and filling in missing values.

[0017] Furthermore, the box plot is used to detect abnormal data and delete them, and the linear interpolation method is used to fill the missing values ​​of gas load.

[0018] Furthermore, the load data is clustered and abnormal data is deleted based on the distance of the data from the cluster center.

[0019] A city gas hourly load multi-step forecasting system based on proportional splitting, the system is used to implement the city gas hourly load multi-step forecasting method based on proportional splitting.

[0020] A cloud computing server is used to run a program, wherein the program executes the multi-step prediction method for city gas hourly load based on proportional splitting when running.

[0021] A processor is used to run a program, wherein when the program is run, the multi-step prediction method for city gas hourly load based on proportional splitting is executed.

[0022] An electronic device comprising a processor, a memory, and a communication bus;

[0023] The communication bus is used to connect the processor and the memory;

[0024] The processor is configured to execute the computer program stored in the memory to implement the method.

[0025] A computer-readable storage medium stores a computer program, wherein the computer program is used to cause the computer to describe the method.

[0026] Compared with the prior art, the advantages and positive effects of the present invention are:

[0027] (1) Based on the prediction of the total gas load of the entire area, it is divided into each gas supply point according to the historical proportion to obtain the gas load of each gas supply point, which can meet the needs of real-time simulation and pressure monitoring of the gas pipeline network.

[0028] (2) The difference between weekdays and weekends is fully considered when splitting the ratio. On Mondays or Saturdays, the ratio of the same time in the previous week is used for splitting; in other periods, the ratio of the same time in the previous day is used for splitting, which is in line with the characteristics of gas use by users.

[0029] (3) The LightGBM model is used to construct a single-step prediction model for total gas hourly load. It has fast training and calculation speeds and is highly efficient. When the user load characteristics change, the prediction model can be quickly trained and updated based on the latest real-time load data to ensure the accuracy of the model.

[0030] (4) The recursive method is used to predict the load data of each supply point of the gas pipeline network in a certain period of time in the future. Based on the hourly load data in this period of time, the overall monitoring of the gas pipeline network in this period of time in the future can be completed, which has broad application prospects.

[0031] (5) The load forecasting model does not consider many features and is relatively simple and convenient to implement. The proportional splitting method is easy to understand and can be easily promoted in actual engineering projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The flowchart of a multi-step forecasting method for city gas hourly load based on proportional splitting is shown in the figure.

[0033] Figure 2 This figure shows the comparison between the predicted value and the actual value of the single-step prediction model of the total gas hourly load in the embodiment on the test set.

[0034] Figure 3 This is a comparison between the prediction result of the embodiment in which the MAPE of the gas load for the next 24 hours is predicted to be 4.3% at a certain moment on the test set and the actual result. DETAILED DESCRIPTION

[0035] The present invention will be described in more detail below with reference to the accompanying drawings, in which preferred embodiments of the present invention are shown. It should be understood that those skilled in the art may modify the present invention described herein while still achieving the beneficial effects of the present invention. Therefore, the following description should be understood as a general guide for those skilled in the art and not as a limitation of the present invention.

[0036] For the sake of clarity, not all features of the actual embodiments are described. In the following description, well-known functions and structures are not described in detail because they would obscure the present invention with unnecessary detail. It should be understood that in the development of any actual embodiment, numerous implementation details must be made to achieve the developer's specific goals.

[0037] In order to make the purpose and features of the present invention more obvious and easy to understand, the specific embodiments of the present invention are further described below with reference to the accompanying drawings. This application provides a multi-step prediction method for city gas hourly load based on proportional splitting, comprising the following steps: Figure 1 As shown:

[0038] Step S1: Obtain and organize the historical hourly gas load data of each supply point in a certain area, including the time t and the load data Q at the corresponding moment t,n and sort them in chronological order; where: t refers to the tth moment; n is the supply point number;

[0039] Step S2: Add the gas load of each supply point at each moment to obtain the total gas load Q of the area at each moment t=Q t,1 +Q t,2 +…+Q t,n ; Among them: Q t,1 is the gas load of supply point 1 at time t, and so on, Q t,n is the gas load at supply point n at time t. There are n gas supply points in the area. The total gas load at time t is the sum of the gas loads at each point at time t.

[0040] Step S3: Calculate and record the proportion of the load of each supply point to the total load at each moment

[0041] This example involves a multi-step hourly gas load forecasting task for a district in a city. Specifically, the task involves predicting gas load data for the next 24 hours. The area has two gas supply points, designated A and B. The gas data collected and organized in steps S1-S3 for 24 hours on January 1, 2021, is shown in Table 1.

[0042] Table 1 Gas load data of a certain district in a certain city on January 1, 2021, sorted by step S1

[0043]

[0044]

[0045] Step S4: Data preprocessing, specifically: detecting and eliminating abnormal gas load data, and filling in missing values.

[0046] Step S5: Construct the total gas load characteristics; wherein: the total gas load characteristics include: temperature, day of the week, hour, whether it is a holiday, daily maximum temperature, daily minimum temperature, load data Q of the previous hour t-1 , the load data Q at this time in the past two weeks t-24j and the load data Q at the previous moment t-24j-1 ; Where: t is the current time; t-1 is the previous hour; Here we can obtain the load characteristics of historical moments by accessing the historical database;

[0047] Step S6: Train a single-step forecasting model for the total gas hourly load in the region. Using the features described in step S3 as input features, the LightGBM model is used to train a single-step forecasting model for the total gas hourly load.

[0048] In the embodiment, the features obtained in step S3 are used as input features to train a single-step LightGBM prediction model for total gas hourly load. In the embodiment, the training data is the gas load data for the whole year from January 1, 2021 to December 31, 2021, and the test data is the gas load data for January 1, 2022 to January 31, 2022. The mean absolute percentage error (MAPE) of the trained single-step LightGBM prediction model for total gas hourly load on the test set is 2.90%, and the comparison between the model prediction value and the actual value is as follows: Figure 2 shown.

[0049] Here, the input of the single-step LightGBM prediction model of total gas hourly load is load characteristics; the training sample uses the load data in the historical data as the model output value, and the load characteristic data as the input value. For example, the training sample in the embodiment is based on the gas load data of the whole year from 2021.1.1 to 2021.12.31. The load characteristics supplemented on this basis (including: temperature, day of the week, hour, whether it is a holiday, daily maximum temperature, daily minimum temperature, load data Q of the previous hour) t-1 , the load data Q at this time in the past two weeks t-24j and the load data Q at the previous moment t-24j-1 (j=1,2,…,14).

[0050] Step S7: Recursive method is used to predict the total gas hourly load data (Q t+1 ,Q t+2 ,…,Q t+k ). Specifically, first, the LightGBM total gas hourly load single-step prediction model trained in step S4 is used to predict Q t+1 , and then Q t+1 As the input feature, the LightGBM total gas hourly load single-step forecasting model is used to predict Q t+2 , and so on, until the prediction is Q t+k , prediction completed.

[0051] In the embodiment, the total gas hourly load data (Q t+1 ,Q t+2 ,…,Q t+24 ). Specifically, first, the LightGBM total gas hourly load single-step prediction model trained in step S4 is used to predict Q t+1 , and then Q t+1 As the input feature, the LightGBM total gas hourly load single-step forecasting model is used to predict Q t+2 , and so on, until the prediction is Q t+24, prediction is completed. In the embodiment, the test data is the gas load data from January 1, 2022 to January 31, 2022. The trained total gas hourly load single-step LightGBM prediction model predicts the gas load data for the next 24 hours by recursion. The mean absolute percentage error (MAPE) of the test results on the test set is 4.39%. Figure 3 The figure shows the comparison between the predicted result of 4.3% MAPE at a certain moment on the test set and the actual result. It can be seen that through the prediction model, Q t+1 The input used is the characteristic data at time t+1 (including: temperature, day of the week, hour, whether it is a holiday, daily maximum temperature, daily minimum temperature, load data Q of the previous hour t , the load data Q at this time in the past two weeks t+1-24j and the load data Q at the previous moment t+1-24j-1 (j=1,2,…,14) and other data).

[0052] Step S8: Proportionally split and obtain the hourly gas load data of each gas supply point within the period t+k in the future. Where k is the time length for predicting the future load; for example, if the load data is predicted for the next 12 hours, k is set to 12; if the load data is predicted for the next 24 hours, k is set to 24;

[0053] Specifically, if the current day is Monday or Saturday, the hourly gas load data of each gas supply point at the corresponding time is obtained using the ratio of the corresponding time in the previous week, which can be expressed as:

[0054] Q t+i,n =Q t+i ×R t+i-24*7,n (i=1,2,…,k)

[0055] If the current day is not Monday or Saturday, the hourly gas load data of each gas supply point at the corresponding time is obtained using the ratio of the previous day's corresponding time, which can be expressed as:

[0056] Q t+i,n =Q t+i ×R t+i-24,n (i=1,2,…,k)

[0057] In the embodiment, a box plot is used to detect abnormal data and delete them, and a linear interpolation method is used to fill in the missing values ​​of the gas load.

[0058] Alternatively, cluster the load data and delete abnormal data based on the distance the data deviates from the cluster center.

[0059] In this embodiment, the hourly gas load data of each gas supply point during the future period t+24 is obtained by proportional splitting. Specifically, if the current day is Monday or Saturday, the hourly gas load data of each gas supply point at the corresponding time is obtained using the ratio of the corresponding time of the previous week, which can be expressed as:

[0060] Q t+i,n =Q t+i ×R t+i-24*7,n (i=1,2,…,24)

[0061] If the current day is not Monday or Saturday, the hourly gas load data of each gas supply point at the corresponding time is obtained using the ratio of the previous day's corresponding time, which can be expressed as:

[0062] Q t+i,n =Q t+i ×R t+i-24,n (i=1,2,…,24)

[0063] After splitting, the mean absolute percentage errors (MAPE) of the test results of area A and area B on the test set were 5.42% and 4.74% respectively.

[0064] In the above embodiments, references in the specification to "this embodiment," "one embodiment," "another embodiment," "in some exemplary embodiments," or "other embodiments" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least some, but not necessarily all, embodiments. Multiple occurrences of "this embodiment," "one embodiment," or "another embodiment" do not necessarily refer to the same embodiment.

[0065] In the above embodiments, although the invention has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory structures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed. The embodiments of the present invention are intended to encompass all such alternatives, modifications, and variations that fall within the broad scope of the appended claims.

[0066] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0067] The present invention can be used in a wide variety of general-purpose or special-purpose computing system environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments that include any of the above.

[0068] The present invention may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0069] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, any equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A multi-step forecasting method for city gas hourly load based on proportional decomposition, characterized in that: The following steps are involved: Step S1: Obtain and organize the historical hourly gas load data of each supply point in a certain area, including the time t and the load data Q at the corresponding moment t,n , and sort them in chronological order; where: t is the tth moment; n is the supply point number; Step S2: Add the gas load of each supply point at each moment to obtain the total gas load Q of the area at each moment t =Q t,1 +Q t,2 +…+Q t,n ; Among them: Q t,1 is the gas load of supply point 1 at time t, and so on, Q t,n is the gas load at supply point n at time t. There are n gas supply points in the area. The total gas load at time t is the sum of the gas loads at each point at time t. Step S3: Calculate and record the proportion of the load of each supply point to the total load at each moment Step S4: performing data preprocessing on the collected load data; Step S5: Construct the total gas load characteristics; wherein: the total gas load characteristics include: temperature, day of the week, hour, whether it is a holiday, daily maximum temperature, daily minimum temperature, load data Q of the previous hour t-1 , the load data Q at this time in the past two weeks t-24j and the load data Q at the previous moment t-24j-1 (j=1,2,…,14); Step S6: training a single-step prediction model for the total hourly gas load of the area based on historical data; Step S7: Use recursion to predict the total gas hourly load data (Q t+1 ,Q t+2 ,…,Q t+k ); where: k is the length of time for predicting future load; Step S8: Split the predicted load data according to the proportion of the load data at the corresponding moment in the historical data to obtain the hourly gas load data of each gas supply point at the corresponding moment in the future t+k period.

2. The method for multi-step forecasting of city gas hourly load based on proportional decomposition according to claim 1 is characterized by: In steps S1-S3, the historical hourly gas load data of each supply point in a certain area is obtained and sorted, and then added to obtain the total gas load, and then the proportion of the load of each supply point to the total load is calculated and recorded.

3. The multi-step forecasting method for hourly city gas load based on proportional decomposition according to claim 1 is characterized by: The data preprocessing specifically includes: detecting and eliminating abnormal gas load data, and filling in missing values.

4. The method for multi-step forecasting of city gas hourly load based on proportional decomposition according to claim 3 is characterized by: The box plot is used to detect abnormal data and delete them, and the linear interpolation method is used to fill the missing values ​​of gas load.

5. The method for multi-step forecasting of city gas hourly load based on proportional decomposition according to claim 1 is characterized by: The load data is clustered and abnormal data is deleted based on the distance of the data from the cluster center.

6. A multi-step forecasting system for city gas hourly load based on proportional splitting, characterized in that: The system is used to implement the method according to any one of claims 1 to 5.

7. A cloud computing server, characterized in that: The cloud computing server is used to run a program, wherein the program, when running, executes the multi-step prediction method for city gas hourly load based on proportional splitting according to any one of claims 1 to 5.

8. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the multi-step prediction method for city gas hourly load based on proportional splitting according to any one of claims 1 to 5.

9. An electronic device, characterized in that: Includes processor, memory and communication bus; The communication bus is used to connect the processor and the memory; The processor is configured to execute the computer program stored in the memory to implement the method according to any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and the computer program is used to make the computer execute the method according to any one of claims 1 to 5.

Images