Industrial park production plan decision-making method and system based on load prediction

Abstract

The invention discloses an industrial park production plan decision-making method and system based on load prediction, and the method comprises the steps: constructing a day-ahead load prediction model and an intra-day load prediction model respectively, inputting meteorological data, holiday and festival data and historical load data in the industrial park input data into the day-ahead load prediction model for prediction to obtain day-ahead load prediction data, and inputting the day-ahead load prediction data into the intra-day load prediction model for prediction to obtain intra-day load prediction data; generating a day-ahead production plan decision through a production plan adjustment decision algorithm by using electricity price prediction data and day-ahead load prediction data in the input data of the industrial park, or wind-light power prediction data and day-ahead load prediction data in the input data of the industrial park; and generating an intra-day production plan decision through a production plan adjustment decision algorithm by using electricity price prediction data and intra-day load prediction data in the input data of the industrial park, or wind-light power prediction data and intra-day load prediction data in the input data of the industrial park.

Description

Technical Field

[0001] The present invention belongs to the technical field of source-grid-load-storage integration, and in particular relates to a method and system for making production plans for an industrial park based on load forecasting. Background Art

[0002] The integrated power generation, grid, load and storage technology optimizes and integrates the resources on the power supply side, grid side and load side, and explores the development path of building a new power system with a highly integrated power generation, grid, load and storage, supported by advanced technological breakthroughs and institutional and mechanism innovations.

[0003] In large-scale industrial parks with high energy consumption and high carbon emissions, how to adjust the load balance in the power system, especially how to deal with sudden load changes or failures, is a key challenge. To ensure the stable operation of the system, coordination and linkage between power generation, power supply, load and energy storage networks are required. However, the power system is a system that needs to achieve instantaneous balance. The traditional power system relies on the rotational inertia and frequency regulation capabilities of the power generation side to adjust the power generation according to the changes on the load side to maintain power balance, that is, the "source follows the load" mode. In this mode, when the load of the industrial park suddenly increases, if the power generation capacity is insufficient, it is easy to cause an imbalance between supply and demand and an increase in the power abandonment rate, which seriously threatens the safety of the power grid.

[0004] In view of the above situation, the production plan adjustment strategy can be given with the help of wind and solar power forecast data and load forecast data, that is, the production load or production plan can be appropriately adjusted according to the output of new energy, so as to realize "source-load interaction" under the premise of meeting the production plan. However, this strategy has certain application limitations, because the strategy aims to maximize the consumption of new energy under the premise of meeting the production plan, but does not consider the impact of production plan adjustment on the energy cost of the park. In addition, in actual applications, industrial parks are not necessarily equipped with wind and solar power generation equipment, and wind and solar power forecast data are sometimes missing. Summary of the invention

[0005] The purpose of the present invention is to overcome the problem of missing wind and solar power forecast data in existing production plan adjustment strategies, and propose an industrial park production plan decision-making method and system based on load forecasting.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: A production planning decision method for an industrial park based on load forecasting includes the following steps: Collect and obtain effective data of the industrial park, which includes meteorological data, holiday data, historical load data, electricity price forecast data and wind and solar power forecast data; Preprocessing the acquired valid data of the industrial park to obtain the input data of the industrial park; A day-ahead load forecasting model and an intraday load forecasting model are constructed respectively, and the meteorological data, holiday data and historical load data in the industrial park input data are input into the day-ahead load forecasting model to obtain the day-ahead load forecasting data; and the meteorological data, holiday data and historical load data in the industrial park input data are input into the intraday load forecasting model to obtain the intraday load forecasting data; Using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, a day-ahead production plan decision is generated through a production plan adjustment decision algorithm; By using the electricity price forecast data and the intraday load forecast data in the industrial park input data, or by using the wind and solar power forecast data and the intraday load forecast data in the industrial park input data, the intraday production plan decision is generated through the production plan adjustment decision algorithm.

[0007] Furthermore, the acquisition of effective data of the industrial park includes acquiring historical loads, measured values ​​of meteorological data and holiday data of the industrial park in the past three months, as well as predicted values ​​of meteorological data and holiday data, electricity price prediction data, and wind and solar power prediction data for the next 15 days. The data is collected every 15 minutes, and there are 96 valid data collection values ​​at each time point in a day. The meteorological data includes temperature characteristic data, humidity characteristic data, wind speed characteristic data and irradiance characteristic data.

[0008] Furthermore, the preprocessing of the acquired industrial park valid data to obtain the industrial park input data comprises the following steps: The outliers in the valid data of the industrial park are detected by the 3σ criterion. Whether the valid data of the industrial park contains 96 data every day is detected. If there are less than 96 data, there are missing values. The detected outliers and missing values ​​are processed for missing values ​​and outliers. The holiday data is quantified. The maximum and minimum temperature values ​​in the 96 time point data of each day are calculated respectively, and the 96 time point data are filled with the maximum and minimum temperature values. The missing values ​​and abnormal values ​​are specifically processed as follows: if a single point in time is missing or abnormal, the data is filled with the arithmetic mean of the data before and after the abnormal moment; if there are missing or abnormal data for consecutive time periods on a certain day, and the amount of missing or abnormal data accounts for more than 2 / 3 of the total data volume on that day, all the data on that day will be removed. If it does not exceed 2 / 3, the data is filled with the value of the corresponding time period of the previous day; The quantization processing is specifically as follows: the data of national statutory holidays is quantized as 1, the data of Saturday and Sunday is quantized as 2, and the data from Monday to Friday is quantized as 3.

[0009] Further, the meteorological data, holiday data and historical load data in the industrial park input data are input into the day-ahead load forecasting model for forecasting to obtain the day-ahead load forecasting data, and the meteorological data, holiday data and historical load data in the industrial park input data are input into the intraday load forecasting model for forecasting to obtain the intraday load forecasting data, specifically: the meteorological data, holiday data and historical load data in the industrial park input data are screened, the correlation between the meteorological data, holiday data and historical load data in the industrial park input data is calculated, the meteorological data, holiday data and historical load data in the industrial park input data with high correlation are selected to construct a data set, the data set is input into the day-ahead load forecasting model, the day-ahead load forecasting model outputs 15-minute load forecast values ​​for a total of 288 moments in the next 3 days, the data set is input into the intraday load forecasting model, and the intraday load forecasting model outputs 15-minute load forecast values ​​for a total of 32 moments in the next 8 hours; The data set is divided into training set and test set to train and evaluate the day-ahead load forecasting model and the intraday load forecasting model. The mean absolute error and root mean square error are used to calibrate the day-ahead load forecasting model and the intraday load forecasting model.

[0010] Further, the power price forecast data and the day-ahead load forecast data in the industrial park input data, or the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, are used to generate a day-ahead production plan decision through a production plan adjustment decision algorithm, including taking the lowest total energy cost at 96 moments in the 24 hours before the day as the goal or taking the maximum wind and solar power consumption at 96 moments in the 24 hours before the day as the goal, to generate a day-ahead production plan decision; The goal is to minimize the total energy cost at 96 hours in the previous 24 hours to generate the previous production plan decision, specifically: Using the day-ahead load forecast data and the day-ahead electricity price forecast data, with the goal of minimizing the total energy cost at 96 moments in the 24 hours the day before, and with the constraint that the total forecast load value at 96 moments remains unchanged, the linear programming algorithm is used to solve the load value adjusted at each moment to obtain the minimum total energy cost within 24 hours, and the day-ahead production plan decision is made based on the unchanged total load; The above-mentioned production plan decision is generated with the goal of maximizing the consumption of wind and solar power at 96 moments in the 24 hours before the day, specifically as follows: For parks equipped with wind and solar power generation equipment, the day-ahead load forecast data and the day-ahead wind and solar power forecast data are used to calculate the ratio of the day-ahead wind and solar power forecast value to the day-ahead load forecast value; If the ratio is greater than 50%, the load forecast value is adjusted to obtain an adjustment value, which is:

[0011] in, To adjust the value, is the ratio of the wind and solar power forecast value to the load forecast value on the previous day, is the total predicted load value at 96 moments, for t Load value at the moment; The load forecast value at each moment plus the adjustment value are used to obtain the day-ahead production plan decision.

[0012] Further, the electricity price forecast data and the intraday load forecast data in the industrial park input data, or the wind and solar power forecast data and the intraday load forecast data in the industrial park input data, are used to generate the intraday production plan decision through the production plan adjustment decision algorithm, including the goal of minimizing the total energy cost at 4 hours and 16 moments in the day or the goal of maximizing the consumption of wind and solar power at 4 hours and 16 moments in the day, to generate the intraday production plan decision; The goal is to minimize the total energy cost at 16 o'clock in 4 hours a day, and generate a daily production plan decision, specifically: Using the daily load forecast data and the daily electricity price forecast data, with the goal of minimizing the total energy cost at 16 moments within 4 hours, and the constraint of the constant total forecast load value at 16 moments, the linear programming algorithm is used to solve the load value adjusted at each moment to obtain the minimum total energy cost within 4 hours, and the daily production plan decision is made based on the constant total load; The above-mentioned goal is to maximize the consumption of wind and solar power at 16 o'clock in 4 hours within a day, and generate a daily production plan decision, specifically: For parks equipped with wind and solar power generation equipment, the daily load forecast data and the daily wind and solar power forecast data are used to calculate the ratio of the daily wind and solar power forecast value to the daily load forecast value. If the ratio is greater than 50%, the load forecast value is adjusted to obtain the adjusted value, which is:

[0013] in, To adjust the value, is the ratio of the wind and solar power forecast value to the load forecast value on the previous day, is the total predicted load value at 16 moments, for t Load value at the moment; The load forecast value at each moment plus the adjustment value are used to obtain the daily production plan decision.

[0014] Furthermore, the day-ahead load forecasting model and the intraday load forecasting model are both constructed through the LSTM model.

[0015] A production planning decision system for an industrial park based on load forecasting, comprising: A data acquisition module is used to acquire effective data of the industrial park, including meteorological data, holiday data, historical load data, electricity price forecast data and wind and solar power forecast data; A data processing module is used to pre-process the acquired valid data of the industrial park to obtain the input data of the industrial park; The load forecasting module is used to construct a day-ahead load forecasting model and an intraday load forecasting model respectively, input the meteorological data, holiday data and historical load data in the industrial park input data into the day-ahead load forecasting model to predict and obtain the day-ahead load forecasting data, and input the meteorological data, holiday data and historical load data in the industrial park input data into the intraday load forecasting model to predict and obtain the intraday load forecasting data; The day-ahead production plan decision module is used to generate a day-ahead production plan decision by using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or by using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, through a production plan adjustment decision algorithm; The intraday production plan decision module is used to utilize the electricity price forecast data and intraday load forecast data in the industrial park input data, or utilize the wind and solar power forecast data and intraday load forecast data in the industrial park input data, and generate intraday production plan decisions through the production plan adjustment decision algorithm.

[0016] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for making production plans for an industrial park based on load forecasting is implemented.

[0017] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for making a production plan for an industrial park based on load forecasting is implemented.

[0018] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention proposes a method for making production plans for industrial parks based on load forecasting, which is a comprehensive load auxiliary decision-making scheme that comprehensively considers load forecast data, wind and solar power forecast data, and electricity price forecast data. First, meteorological data, holiday data, historical load data, etc. are used to predict the load of the park before and during the day. Then, combined with the load forecast data, wind and solar power forecast data, and electricity price forecast data, decision suggestions are given for the production plans of the park before and during the day. This is a complete load decision-making comprehensive solution that weakens the dependence on wind and solar power forecast data, and has a wider range of application scenarios and higher application value.

[0019] Furthermore, starting from the load forecast, decision suggestions are given for the production plan of the park in the day and the day from the perspective of maximizing the consumption of new energy and minimizing the energy cost. A linear programming algorithm with optimal energy cost is designed in combination with electricity price data. Load decisions can be made from the perspective of maximizing the consumption of wind and solar power, as well as from the perspective of optimizing energy cost. Load decisions are made without relying on the linear programming algorithm of wind and solar power forecast. Regardless of whether the industrial park is equipped with wind and solar power generation equipment, constructive load decision suggestions can be given, which has a wide range of application scenarios and higher application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are for explanation purposes only and are not intended to limit the scope of the present invention in any way. In addition, the shapes and proportional dimensions of the components in the drawings are only for illustration purposes to help understand the present invention, and are not intended to specifically limit the shapes and proportional dimensions of the components of the present invention. In the drawings: Figure 1 The present invention is a flow chart of a method for making production plans for an industrial park based on load forecasting.

[0021] Figure 2 This is a structural diagram of an industrial park production planning decision system based on load forecasting according to the present invention.

[0022] Figure 3 This is an electronic equipment diagram of an industrial park production plan decision method based on load forecasting according to the present invention.

[0023] Figure 4 A schematic diagram of a flow chart of the implementation of the load forecasting algorithm in the fifth embodiment of the present invention; Figure 5 This is a schematic diagram of the overall flow of the load auxiliary decision algorithm in the fifth embodiment of the present invention; Figure 6 This is a system module diagram in Embodiment 5 of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0025] Embodiment 1 See also Figure 1 , a production planning decision method for an industrial park based on load forecasting, comprising the following steps: Collect and obtain effective data of industrial parks, including meteorological data, holiday data, historical load data, electricity price forecast data and wind and solar power forecast data; Preprocessing the acquired valid data of the industrial park to obtain the input data of the industrial park; A day-ahead load forecasting model and an intraday load forecasting model are constructed respectively, and the meteorological data, holiday data and historical load data in the industrial park input data are input into the day-ahead load forecasting model to obtain the day-ahead load forecasting data; and the meteorological data, holiday data and historical load data in the industrial park input data are input into the intraday load forecasting model to obtain the intraday load forecasting data; Using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, a day-ahead production plan decision is generated through a production plan adjustment decision algorithm; By using the electricity price forecast data and the intraday load forecast data in the industrial park input data, or by using the wind and solar power forecast data and the intraday load forecast data in the industrial park input data, the intraday production plan decision is generated through the production plan adjustment decision algorithm.

[0026] The load forecast-based industrial park production planning decision-making method provided in this embodiment integrates wind and solar power forecast data with electricity price forecast data, and predicts future loads, electricity prices and wind and solar power through multiple data sources and forecasting models. It can improve production efficiency, optimize resource allocation, reduce operating costs, enhance energy sustainability, improve decision-making accuracy, enhance risk response capabilities and promote intelligent management.

[0027] The collection and acquisition of effective data of the industrial park includes the collection and acquisition of the historical load of the industrial park in the past three months, the measured values ​​of meteorological data and holiday data, as well as the predicted values ​​of meteorological data and holiday data, electricity price forecast data, and wind and solar power forecast data for the next 15 days. The data is collected every 15 minutes, and there are 96 valid data collection values ​​in one day; Meteorological data include temperature characteristic data, humidity characteristic data, wind speed characteristic data and irradiance characteristic data.

[0028] Preprocessing the acquired valid data of the industrial park to obtain the input data of the industrial park includes the following steps: The outliers in the valid data of the industrial park are detected by the 3σ criterion. Whether the valid data of the industrial park contains 96 data every day is detected. If there are less than 96 data, there are missing values. The detected outliers and missing values ​​are processed for missing values ​​and outliers. The holiday data is quantified. The maximum and minimum temperature values ​​in the 96 time point data of each day are calculated respectively, and the 96 time point data are filled with the maximum and minimum temperature values. The specific processing of missing values ​​and outliers is as follows: if a single point in time is missing or abnormal, it is filled with the arithmetic mean of the data before and after the abnormal moment; if there are missing or abnormal data for consecutive time periods on a certain day, and the amount of missing or abnormal data accounts for more than 2 / 3 of the total data volume on that day, all the data on that day will be removed. If it does not exceed 2 / 3, it will be filled with the value of the corresponding time period of the previous day; The specific quantification processing is as follows: national statutory holidays data is quantified as 1, Saturday and Sunday data is quantified as 2, and Monday to Friday data is quantified as 3.

[0029] When calculating and analyzing time series data, especially in meteorology, environmental monitoring or other applications that require continuous data recording, it is often necessary to process multiple data points within a day. 24 hours a day, recording every 15 minutes, that is 96 time points, each time point has a corresponding temperature data record. For each day's data, find the highest and lowest temperatures in these 96 time points, traverse these 96 data points and record the maximum and minimum values ​​of the 96 data points. These two values ​​need to be "filled" or "mapped" back to the original 96 time points.

[0030] The meteorological data, holiday data and historical load data in the input data of the industrial park are input into the day-ahead load forecasting model for forecasting to obtain the day-ahead load forecasting data. The meteorological data, holiday data and historical load data in the input data of the industrial park are input into the intraday load forecasting model for forecasting to obtain the intraday load forecasting data. Specifically, the meteorological data, holiday data and historical load data in the input data of the industrial park are screened, and the correlation between the meteorological data, holiday data and historical load data in the input data of the industrial park is calculated. The meteorological data, holiday data and historical load data in the input data of the industrial park with high correlation are selected to construct a data set. The data set is input into the day-ahead load forecasting model. The day-ahead load forecasting model outputs 15-minute load forecast values ​​for a total of 288 moments in the next 3 days. The data set is input into the intraday load forecasting model. The intraday load forecasting model outputs 15-minute load forecast values ​​for a total of 32 moments in the next 8 hours. The data set is divided into training set and test set to train and evaluate the day-ahead load forecasting model and the intraday load forecasting model. The mean absolute error and root mean square error are used to calibrate the day-ahead load forecasting model and the intraday load forecasting model.

[0031] The training set is used in the model training process. During the training phase, the model will learn the data features and patterns in the training set to build a prediction model. The training set contains the selected high-relevance features and historical load data, which are input into the model. Through the learning and optimization of the algorithm, the model can gradually fit the law of load changes.

[0032] The test set is used to evaluate the prediction performance of the model. After training, the model will use the test set data to make predictions and compare the prediction results with the true load values ​​of the test set. By calculating indicators such as prediction error and accuracy, the prediction ability and generalization performance of the model can be evaluated.

[0033] Using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, a day-ahead production plan decision is generated through a production plan adjustment decision algorithm, including taking the lowest total energy cost at 96 moments in the 24 hours before the day as the goal or taking the maximum consumption of wind and solar power at 96 moments in the 24 hours before the day as the goal, to generate a day-ahead production plan decision; With the goal of minimizing the total energy cost at 96 hours in the previous 24 hours, the previous day production plan decision is generated, specifically: Using the day-ahead load forecast data and the day-ahead electricity price forecast data, with the goal of minimizing the total energy cost at 96 moments in the 24 hours the day before, and with the constraint that the total forecast load value at 96 moments remains unchanged, the linear programming algorithm is used to solve the load value adjusted at each moment to obtain the minimum total energy cost within 24 hours, and the day-ahead production plan decision is made based on the unchanged total load; The linear programming algorithm mainly solves the problem of multivariable functions with linear relationships, and seeks the optimal (maximum, minimum) objective function under certain linear constraints. The steps of using the linear programming algorithm to establish a mathematical model usually include: determining the variables that affect the target, which are factors that need to be considered in the decision-making process and are also unknowns in the objective function and constraints; establishing the objective function, converting the target into a mathematical expression, namely the objective function, according to the actual needs of the problem; adding constraints, and listing all conditions that limit the values ​​of decision variables according to the actual situation of the problem, namely constraints; solving the specific values ​​of various unknown parameters, and on the premise of maximizing the target, finding a set of optimal combinations based on the specific restrictions of the constraints of various unknown parameters, namely the optimal solution of linear programming.

[0034] With the goal of maximizing the consumption of wind and solar power at 96 moments in the 24 hours before the day, the day-ahead production plan decision is generated, specifically: For parks equipped with wind and solar power generation equipment, the day-ahead load forecast data and the day-ahead wind and solar power forecast data are used to calculate the ratio of the day-ahead wind and solar power forecast value to the day-ahead load forecast value; If the ratio is greater than 50%, the load forecast value is adjusted to obtain an adjustment value, which is:

[0035] in, To adjust the value, is the ratio of the wind and solar power forecast value to the load forecast value on the previous day, is the total predicted load value at 96 moments, for t Load value at the moment; The load forecast value at each moment plus the adjustment value are used to obtain the day-ahead production plan decision.

[0036] Using the electricity price forecast data and the intraday load forecast data in the industrial park input data, or using the wind and solar power forecast data and the intraday load forecast data in the industrial park input data, the intraday production plan decision is generated through the production plan adjustment decision algorithm, including the goal of minimizing the total energy cost within 4 hours and 16 moments in a day or the goal of maximizing the consumption of wind and solar power within 4 hours and 16 moments in a day, to generate the intraday production plan decision; With the goal of minimizing the total energy cost at 16 o'clock in 4 hours a day, the daily production plan decision is generated, specifically: Using the daily load forecast data and the daily electricity price forecast data, with the goal of minimizing the total energy cost at 16 moments within 4 hours, and the constraint of the constant total forecast load value at 16 moments, the linear programming algorithm is used to solve the load value adjusted at each moment to obtain the minimum total energy cost within 4 hours, and the daily production plan decision is made based on the constant total load; With the goal of maximizing the consumption of wind and solar power at 16 o'clock in 4 hours a day, the daily production plan decision is generated, specifically: For parks equipped with wind and solar power generation equipment, the daily load forecast data and the daily wind and solar power forecast data are used to calculate the ratio of the daily wind and solar power forecast value to the daily load forecast value. If the ratio is greater than 50%, the load forecast value is adjusted to obtain the adjusted value, which is:

[0037] in, To adjust the value, is the ratio of the wind and solar power forecast value to the load forecast value on the previous day, is the total predicted load value at 16 moments, for t Load value at the moment; The load forecast value at each moment plus the adjustment value are used to obtain the daily production plan decision.

[0038] Both the day-ahead load forecasting model and the intraday load forecasting model are constructed using the LSTM model.

[0039] Embodiment 2 See also Figure 2 , an industrial park production planning decision system based on load forecasting, including: A data acquisition module is used to acquire effective data of the industrial park, including meteorological data, holiday data, historical load data, electricity price forecast data and wind and solar power forecast data; A data processing module is used to pre-process the acquired valid data of the industrial park to obtain the input data of the industrial park; The load forecasting module is used to construct a day-ahead load forecasting model and an intraday load forecasting model respectively, input the meteorological data, holiday data and historical load data in the industrial park input data into the day-ahead load forecasting model to predict and obtain the day-ahead load forecasting data, and input the meteorological data, holiday data and historical load data in the industrial park input data into the intraday load forecasting model to predict and obtain the intraday load forecasting data; The day-ahead production plan decision module is used to generate a day-ahead production plan decision by using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or by using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, through a production plan adjustment decision algorithm; The intraday production plan decision module is used to utilize the electricity price forecast data and intraday load forecast data in the industrial park input data, or utilize the wind and solar power forecast data and intraday load forecast data in the industrial park input data, and generate intraday production plan decisions through the production plan adjustment decision algorithm.

[0040] This embodiment provides an industrial park production planning decision system based on load forecasting, which brings many benefits such as accurate prediction, data-driven, flexible response to market changes, optimization of resource allocation, improvement of decision-making efficiency, and promotion of digital transformation, which helps to improve the overall operation level and market competitiveness of the industrial park.

[0041] Embodiment 3 See also Figure 3 , an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for making production plans for an industrial park based on load forecasting is implemented.

[0042] Embodiment 4 A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for making a production plan for an industrial park based on load forecasting is implemented.

[0043] Embodiment 5 A production planning decision method for an industrial park based on load forecasting includes the following steps: S1. Obtain park meteorological data, holiday data, historical load data, electricity price forecast data, wind and solar power forecast data, and pre-process the data, including outlier and missing value processing, holiday data quantification, etc.; S2. Use meteorological data, holiday data, and historical load data to design models to predict the day-ahead and intraday loads; S3: For parks equipped with wind and solar power generation equipment, based on the day-ahead load predicted by S2 and the day-ahead electricity price forecast data and wind and solar power forecast data obtained by S1, production plan decision suggestions for the day-ahead and the day-ahead are given from the perspectives of minimizing energy costs and maximizing the consumption of new energy. S4: For industrial parks that are not equipped with wind and solar power generation equipment, production planning decision suggestions for the day and within the day are given from the perspective of minimizing energy costs based on the day-ahead load predicted by S2 and the day-ahead electricity price forecast data obtained by S1.

[0044] The above steps are specifically performed as follows: Step 1: Obtain data. Obtain the historical load, meteorological measured data, and holiday data of an industrial park in the past three months, as well as the meteorological forecast data and holiday data, electricity price forecast data, and wind and solar power forecast data for the next 15 days. All data are valid data at 96 times a day.

[0045] Step 2: Process the acquired data. First, The criterion detects whether there are abnormal values ​​in various types of data. Subsequently, it is detected whether the data set contains 96 data every day. If there are less than 96 data, there are missing data. The detected abnormal values ​​and missing values ​​are processed by substitution and removal methods. That is, if a single point of time data is missing or abnormal, it is filled with the arithmetic mean of the data before and after the abnormal moment. If there are missing or abnormal data in consecutive time periods on a certain day, and the proportion of missing or abnormal data to the total data volume of the day exceeds 2 / 3, all data on the day are removed. If it does not exceed 2 / 3, it is filled with the value of the corresponding time period of the previous day. The holiday data is quantified. If a certain day is a national statutory holiday, such as Qingming Festival, Labor Day, etc., the data of that day is quantified to 1, and Saturday and Sunday are quantified to 2, and Monday to Friday are quantified to 3 for the remaining days. In addition, the maximum and minimum temperature values ​​in the 96 time point data of each day are calculated respectively, and filled into 96 time point data, and then used as new input features of the model respectively.

[0046] Step 3: Perform load forecasting. First, perform feature screening, calculate the correlation between all features in step 2 and historical loads, select features with higher correlation and historical loads to construct a data set, and divide the data into training and test sets. For day-ahead load forecasting, the model output is the 15-minute load forecast value for 288 moments in the next 3 days. For intraday load forecasting, the model output is the 15-minute load forecast value for 32 moments in the next 8 hours. Build LSTM models for day-ahead load forecasting and intraday load forecasting respectively. Use mean absolute error and root mean square error to calibrate the model. The load forecasting algorithm process is as follows: Figure 4 shown.

[0047] Step 4: Generate a day-ahead production plan adjustment strategy. For parks that are not equipped with wind and solar power generation equipment, obtain the day-ahead load forecast data and the day-ahead electricity price forecast data, with the goal of minimizing the total energy cost at 96 moments in the 24 hours before the day. The objective function is as follows:

[0048] in for t The electricity price forecast at the time, For the optimized t The load value at each moment is constrained by the constant total predicted load value at 96 moments, that is:

[0049] in For the original t In addition, the load value at a single moment must be within a reasonable range, that is:

[0050] in and It is the upper and lower limits of the load value at each moment, and the specific value is determined according to the actual working conditions. By solving the load value adjusted at each moment through the linear programming algorithm, the minimum total energy cost within 24 hours can be obtained, while keeping the total load unchanged, so as to obtain the production plan decision recommendation for the day ahead.

[0051] For parks equipped with wind and solar power generation equipment, the same method can be used to adjust the production plan with the lowest energy cost as the goal, or the maximum consumption of wind and solar can be used as the goal. At this time, the core idea is to produce as much as possible on the basis of meeting the basic requirement that wind and solar account for 50% of the total load. First, obtain the day-ahead load forecast data and the day-ahead wind and solar power forecast data, and calculate the ratio of the day-ahead wind and solar power forecast value to the day-ahead load forecast value, that is:

[0052] like k >50%, the load value is adjusted. The load forecast value at each moment plus the adjustment value can be used to obtain the adjustment result of the production plan on the day before. At this time, photovoltaic power can be absorbed as much as possible and the production plan can be improved.

[0053] Step 5: Generate a daily production plan adjustment strategy. The daily production plan adjustment strategy is consistent with the day-ahead strategy. For parks that are not equipped with photovoltaic power generation equipment, the goal is to minimize the 4-hour energy cost. A linear programming algorithm is used to obtain the load adjustment value, and then the daily production plan decision strategy is obtained. For parks equipped with photovoltaic power generation equipment, the goal can be either to minimize the energy cost or to maximize the consumption of wind and solar power. The algorithm consistent with the day-ahead strategy can be used to obtain the production plan adjustment result. The industrial park dispatcher makes a trade-off between the two goals based on the park's working conditions to decide which strategy to use. The production plan adjustment decision algorithm process is as follows: Figure 5 shown.

[0054] like Figure 6 As shown, this embodiment provides an industrial park production plan decision system based on load forecasting, including: The data acquisition module is used to obtain the industrial park's historical load, meteorological data, holiday data, electricity price forecast data, and wind and solar power forecast data. The data is recorded every 15 minutes, and there are 96 valid data collection values ​​at each moment every day. The meteorological data includes characteristic data such as temperature, humidity, wind speed, and irradiance. The data processing module processes the acquired data, including missing value and outlier processing, quantitative processing, etc. The load forecasting module inputs the processed historical load data, holiday data, and meteorological data as input features into the Long Short-Term Memory (LSTM) model. The model outputs the day-ahead load forecasting results and the intraday load forecasting results. The day-ahead load forecasting and the intraday load forecasting are implemented using different LSTM models. The day-ahead load auxiliary decision-making module obtains the day-ahead load forecast data and the day-ahead electricity price forecast data for parks that are not equipped with wind and solar power generation equipment. With the goal of minimizing the total energy cost at 96 moments in the 24 hours the day before and the constraint of the constant total forecast load value at 96 moments, the linear programming algorithm is used to solve the load adjustment value at each moment, and the day-ahead production plan decision recommendations can be obtained. For parks equipped with wind and solar power generation equipment, with the goal of maximizing the consumption of wind and solar power, the day-ahead load forecast data and the day-ahead wind and solar power forecast data are obtained. On the basis of satisfying the basic requirement that wind and solar power accounts for 50% of the total load, as much production as possible is produced, and the day-ahead production plan decision recommendations can be obtained.

[0055] For parks that are not equipped with wind and solar power generation equipment, the daily load auxiliary decision-making module obtains daily load forecast data and daily electricity price forecast data, takes the lowest total energy cost at 16 moments within 4 hours a day as the goal, and takes the constant total predicted load value at 16 moments as the constraint. The load value adjusted at each moment is solved through a linear programming algorithm to obtain daily production plan decision suggestions; for parks equipped with wind and solar power generation equipment, with the goal of maximizing the consumption of wind and solar power, the daily load forecast data and daily wind and solar power forecast data are obtained, and as much production as possible is produced on the basis of satisfying the basic requirement that wind and solar power accounts for 50% of the total load, to obtain daily production plan decision suggestions.

[0056] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0057] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0058] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0059] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1A process or multiple processes and / or boxes Figure 1 Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention is described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation of the present invention can still be modified or replaced by equivalents, and any modification or equivalent replacement that does not deviate from the spirit and scope of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A production planning decision method for an industrial park based on load forecasting, characterized in that: The following steps are involved: Collect and obtain effective data of the industrial park, which includes meteorological data, holiday data, historical load data, electricity price forecast data and wind and solar power forecast data; Preprocessing the acquired valid data of the industrial park to obtain the input data of the industrial park; A day-ahead load forecasting model and an intraday load forecasting model are constructed respectively, and the meteorological data, holiday data and historical load data in the industrial park input data are input into the day-ahead load forecasting model to obtain the day-ahead load forecasting data; and the meteorological data, holiday data and historical load data in the industrial park input data are input into the intraday load forecasting model to obtain the intraday load forecasting data; Using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, a day-ahead production plan decision is generated through a production plan adjustment decision algorithm; By using the electricity price forecast data and the intraday load forecast data in the industrial park input data, or by using the wind and solar power forecast data and the intraday load forecast data in the industrial park input data, the intraday production plan decision is generated through the production plan adjustment decision algorithm.

2. The method for industrial park production planning decision-making based on load forecasting according to claim 1 is characterized in that: The acquisition of effective data of the industrial park includes acquiring the historical load of the industrial park in the past three months, the measured values ​​of meteorological data and holiday data, as well as the predicted values ​​of meteorological data and holiday data, electricity price prediction data, and wind and solar power prediction data for the next 15 days. The data is collected every 15 minutes, and there are 96 valid data collection values ​​at each time point in a day. The meteorological data includes temperature characteristic data, humidity characteristic data, wind speed characteristic data and irradiance characteristic data.

3. The method for industrial park production planning decision-making based on load forecasting according to claim 2 is characterized in that: The step of preprocessing the acquired valid industrial park data to obtain the industrial park input data comprises the following steps: The outliers in the valid data of the industrial park are detected by the 3σ criterion. Whether the valid data of the industrial park contains 96 data every day is detected. If there are less than 96 data, there are missing values. The detected outliers and missing values ​​are processed for missing values ​​and outliers. The holiday data is quantified. The maximum and minimum temperature values ​​in the 96 time point data of each day are calculated respectively, and the 96 time point data are filled with the maximum and minimum temperature values. The missing values ​​and abnormal values ​​are specifically processed as follows: if a single point in time is missing or abnormal, the data is filled with the arithmetic mean of the data before and after the abnormal moment; if there are missing or abnormal data for consecutive time periods on a certain day, and the amount of missing or abnormal data accounts for more than 2 / 3 of the total data volume on that day, all the data on that day will be removed. If it does not exceed 2 / 3, the data is filled with the value of the corresponding time period of the previous day; The quantization processing is specifically as follows: the data of national statutory holidays is quantized as 1, the data of Saturday and Sunday is quantized as 2, and the data from Monday to Friday is quantized as 3.

4. The method for industrial park production planning decision-making based on load forecasting according to claim 3 is characterized in that: The meteorological data, holiday data and historical load data in the industrial park input data are input into the day-ahead load forecasting model for forecasting to obtain the day-ahead load forecasting data, and the meteorological data, holiday data and historical load data in the industrial park input data are input into the intraday load forecasting model for forecasting to obtain the intraday load forecasting data, specifically: the meteorological data, holiday data and historical load data in the industrial park input data are screened, the correlation between the meteorological data, holiday data and historical load data in the industrial park input data is calculated, the meteorological data, holiday data and historical load data in the industrial park input data with high correlation are selected to construct a data set, the data set is input into the day-ahead load forecasting model, and the day-ahead load forecasting model outputs 15-minute load forecast values ​​for a total of 288 moments in the next 3 days; the data set is input into the intraday load forecasting model, and the intraday load forecasting model outputs 15-minute load forecast values ​​for a total of 32 moments in the next 8 hours; The data set is divided into training set and test set to train and evaluate the day-ahead load forecasting model and the intraday load forecasting model. The mean absolute error and root mean square error are used to calibrate the day-ahead load forecasting model and the intraday load forecasting model.

5. The method for industrial park production planning decision-making based on load forecasting according to claim 4 is characterized in that: The method uses the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or uses the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, to generate a day-ahead production plan decision through a production plan adjustment decision algorithm, including taking the lowest total energy cost at 96 moments in the 24 hours before the day as a goal or taking the maximum consumption of wind and solar power at 96 moments in the 24 hours before the day as a goal, to generate a day-ahead production plan decision; The goal is to minimize the total energy cost at 96 hours in the previous 24 hours to generate the production plan decision for the previous day, specifically: Using the day-ahead load forecast data and the day-ahead electricity price forecast data, with the goal of minimizing the total energy cost at 96 moments in the 24 hours the day before, and with the constraint that the total forecast load value at 96 moments remains unchanged, the linear programming algorithm is used to solve the load value adjusted at each moment to obtain the minimum total energy cost within 24 hours, and the day-ahead production plan decision is made based on the unchanged total load; The above-mentioned production plan decision is generated with the goal of maximizing the consumption of wind and solar power at 96 moments in the 24 hours before the day, specifically as follows: For parks equipped with wind and solar power generation equipment, the day-ahead load forecast data and the day-ahead wind and solar power forecast data are used to calculate the ratio of the day-ahead wind and solar power forecast value to the day-ahead load forecast value; If the ratio is greater than 50%, the load forecast value is adjusted to obtain an adjustment value, which is: in, To adjust the value, is the ratio of the wind and solar power forecast value to the load forecast value on the previous day, is the total predicted load value at 96 moments, for t Load value at the moment; The load forecast value at each moment plus the adjustment value are used to obtain the day-ahead production plan decision.

6. The method for industrial park production planning decision-making based on load forecasting according to claim 4 is characterized in that: The method uses the electricity price forecast data and the daily load forecast data in the industrial park input data, or uses the wind and solar power forecast data and the daily load forecast data in the industrial park input data to generate a daily production plan decision through a production plan adjustment decision algorithm, including taking the minimum total energy cost at 4 hours and 16 moments in a day as the goal or taking the maximum consumption of wind and solar power at 4 hours and 16 moments in a day as the goal to generate a daily production plan decision; The goal is to minimize the total energy cost at 16 o'clock in 4 hours a day, and generate a daily production plan decision, specifically: Using the daily load forecast data and the daily electricity price forecast data, with the goal of minimizing the total energy cost at 16 moments within 4 hours, and the constraint of the constant total forecast load value at 16 moments, the linear programming algorithm is used to solve the load value adjusted at each moment to obtain the minimum total energy cost within 4 hours, and the daily production plan decision is made based on the constant total load; The above-mentioned goal is to maximize the consumption of wind and solar power at 16 o'clock in 4 hours within a day, and generate a daily production plan decision, specifically: For parks equipped with wind and solar power generation equipment, the daily load forecast data and the daily wind and solar power forecast data are used to calculate the ratio of the daily wind and solar power forecast value to the daily load forecast value. If the ratio is greater than 50%, the load forecast value is adjusted to obtain the adjusted value, which is: in, To adjust the value, is the ratio of the wind and solar power forecast value to the load forecast value on the previous day, is the total predicted load value at 16 moments, for t Load value at the moment; The load forecast value at each moment plus the adjustment value are used to obtain the daily production plan decision.

7. The method for industrial park production planning decision-making based on load forecasting according to claim 1 is characterized in that: The day-ahead load forecasting model and the intraday load forecasting model are both constructed through the LSTM model.

8. An industrial park production planning decision system based on load forecasting, characterized in that: include: A data acquisition module is used to acquire effective data of the industrial park, including meteorological data, holiday data, historical load data, electricity price forecast data and wind and solar power forecast data; A data processing module is used to pre-process the acquired valid data of the industrial park to obtain the input data of the industrial park; The load forecasting module is used to construct a day-ahead load forecasting model and an intraday load forecasting model respectively, input the meteorological data, holiday data and historical load data in the industrial park input data into the day-ahead load forecasting model to predict and obtain the day-ahead load forecasting data, and input the meteorological data, holiday data and historical load data in the industrial park input data into the intraday load forecasting model to predict and obtain the intraday load forecasting data; The day-ahead production plan decision module is used to generate a day-ahead production plan decision by using the electricity price forecast data and the day-ahead load forecast data in the industrial park input data, or by using the wind and solar power forecast data and the day-ahead load forecast data in the industrial park input data, through a production plan adjustment decision algorithm; The intraday production plan decision module is used to utilize the electricity price forecast data and intraday load forecast data in the industrial park input data, or utilize the wind and solar power forecast data and intraday load forecast data in the industrial park input data, and generate intraday production plan decisions through the production plan adjustment decision algorithm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for making production plans for an industrial park based on load forecasting as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the method for making production plans for an industrial park based on load forecasting as described in any one of claims 1 to 7.

Images