Micro-grid quasi-periodic energy management method based on time sequence prediction and linear programming

Abstract

The invention belongs to the technical field of electric power, and discloses a micro-grid quasi-periodic energy management method based on time sequence prediction and linear programming, and a micro-grid system comprises an energy storage system, renewable energy sources, a load and an external power grid. The invention provides an economic energy exchange method between a micro-grid and a power grid. Based on the quasi-periodic hypothesis, the load power, the internal source power and the electricity price are accurately predicted, the energy distribution is optimized by using the linear programming method, the model training calculation complexity is low, and the possibility is provided for the use of edge intelligent equipment. The energy utilization efficiency is effectively improved, dependence on traditional energy is reduced, and sustainability and economic benefits are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid energy management, and in particular to a microgrid quasi-periodic energy management method based on time series prediction and linear programming. Background Art

[0002] In recent years, there has been a growing interest in developing smart and sustainable energy systems that effectively integrate renewable energy and energy storage technologies. A microgrid is a localized energy distribution network that operates independently or in conjunction with the main grid and integrates a variety of renewable energy sources, energy storage, and intelligent control systems. These systems offer many benefits, including enhanced reliability, lower energy costs, and reduced greenhouse gas emissions. Coupled with effective energy management practices such as load management and demand response, microgrids have the potential to revolutionize the energy landscape by providing sustainable, resilient, and cost-effective power solutions. The energy management system (EMS) is the brain of the microgrid, coordinating the production, consumption, storage, and distribution of electricity across the various microgrids and power plants within the system to ensure a reliable supply. By optimizing the use of renewable energy, storage systems, and grid interconnections, an effective EMS can enable microgrids to maximize efficiency, reduce dependence on traditional energy sources, and contribute to a greener and decentralized energy future. Traditional energy management systems are designed to integrate different energy sources, regulate power flows, ensure smooth operation within the microgrid, and improve grid stability. However, with the development of renewable energy sources such as wind power and photovoltaics, as well as power storage technologies, the share of wind power and photovoltaic power generation is increasing, and power storage capacity is also rising. In this context, the requirements for the economic operation of microgrids have been raised. The economic efficiency of microgrids is mainly reflected in the energy exchange with the grid. In the case of power balance, microgrids can sell excess energy to the grid when the internal power supply is sufficient, or obtain power from the grid to provide loads and charge batteries when the internal power supply is insufficient. The ability to sell energy when the electricity price is high and to obtain electricity when the electricity price is low is the key to the economic operation of the grid.

[0003] With the rapid development of renewable energy, microgrids, as a localized energy distribution network that integrates multiple energy forms, have gradually become an important solution for achieving sustainable energy development. Traditional microgrid energy management systems mainly regulate the flow of different energy sources to ensure the power balance within the system. However, due to the instability of renewable energy generation and the fluctuation of electricity demand, the economic operation of microgrids faces new challenges. The shortcomings of existing technologies are the lack of effective prediction and optimization methods, which makes it difficult for energy management methods to adapt to fluctuations in electricity prices and changes in electricity demand, and the energy exchange between microgrids and power grids cannot be optimized. Therefore, there is an urgent need for an energy management method based on time series prediction and linear programming to improve the economy and reliability of microgrids. Summary of the invention

[0004] In view of the above problems existing in the prior art, the present invention provides a microgrid quasi-periodic energy management method based on time series prediction and linear programming.

[0005] In this application, the main parameters are defined as follows: y is the predicted value from the Holt-Winters method, l is the smoothing value of the Holt-Winters method, t represents the trend value of the Holt-Winters method, s represents the seasonal value of the Holt-Winters method, k represents the length of the season in the Holt-Winters method, h represents the number of prediction points of the Holt-Winters method, a represents the level parameter of the Holt-Winters method, b represents the trend parameter of the Holt-Winters method, c represents the seasonal parameter of the Holt-Winters method, p represents the linear programming power vector, f represents the total cost of linear programming, A price represents the hourly electricity price, P pv Represents the power of renewable energy such as photovoltaics, P load Indicates the power of the load, P battery Indicates the battery power. SOC stands for state of charge. P grid Indicates the external grid power.

[0006] The technical solution adopted by the present invention is: A microgrid quasi-periodic energy management method based on time series prediction and linear programming, characterized in that it includes the following steps: Step 1: Establish a microgrid quasi-periodic energy management system model based on the capacity of the energy storage system, load power and renewable energy power in the microgrid; Step 2: Based on the quasi-periodicity assumption, time series forecasting of load power, internal power and electricity price is performed; Step 3: Use linear programming method to optimize the energy exchange method between the microgrid and the grid; Step 4: Considering the fluctuation of electricity prices and system constraints, adjust the energy distribution of the microgrid to minimize the energy cost.

[0007] The above-mentioned microgrid quasi-periodic energy management method based on time series prediction and linear programming is characterized in that the time series prediction adopts the Holt-Winters method to deal with the periodic fluctuations of load and endogenous power, especially in the scenario where daily load changes and electricity price fluctuations are significant, the data of the previous cycle is smoothed to generate prediction results including short-term trends, long-term trends and periodic changes. The model can adapt to the dynamic fluctuations of load power and endogenous power, effectively improve the accuracy of prediction, and thus provide accurate input data for subsequent linear programming.

[0008] The Holt-Winters method introduces the Winters period term (also called the seasonal term) based on the Holt model to deal with the fixed-period fluctuation behavior in time series such as monthly data (period 12), quarterly data (period 4), and weekly data (period 7). Multiple Winters terms are introduced to deal with the coexistence of multiple periods. Specifically, the Holt-Winters method is used to predict the data for the day. The prediction can be expressed as follows: y i+h = l i +ht i +s i-k+h (1) Where y is the forecast value, l is the smoothed value, t is the trend value, s is the seasonal value, k is the length of the season, and h is the number of forecast points. The calculation of the above values ​​can be expressed as follows: l i =a(y i -s i-k )+(1-a)(l i-1 -t i-1 ) (2) t i =b(l i -l i-k )+(1-b)t i-1 (3) s i =c(y i -l i )+(1-c)s i-1 (4) Where a, b and c represent the level, trend and seasonal parameters respectively. The parameters can be chosen arbitrarily and the forecast performance will vary.

[0009] The above-mentioned microgrid quasi-periodic energy management method based on time series prediction and linear programming is characterized in that linear programming is a method for finding the minimum value of an objective function under specific constraints. The weighted arithmetic mean of the observed values ​​is calculated by using the observation values ​​of the same variable arranged in chronological order in the past and the number of times the chronological variable appears as the weight, and this number is used as a trend prediction method to predict the predicted value of the variable in the future. Linear programming predicts future electricity prices and power fluctuations by weighted moving average method and optimizes power exchange. The weighted moving average method processes historical electricity prices and power data by introducing different weights, giving the latest data a higher weight, so that the prediction results of future electricity prices and power fluctuations can be dynamically adjusted. The variable of the linear programming solution is the power exchanged between the microgrid and the power grid at each moment.

[0010] When performing linear programming, in order to simplify the problem and make the solution of linear programming more stable and reliable, the time granularity of the analysis is considered to be 1 hour, assuming that the power exchanged between the grid and the microgrid remains the same during 1 hour. The load power, photovoltaic power and electricity price used in the linear programming are taken as the average value of each 1 hour. The power vector p and the total cost f can be expressed as follows: P=[x1,x2,x3…,x 24 ] T (5) f=A price P (6) Among them, A price represents the hourly electricity price, and f is the objective function. On this basis, the energy exchange method between the microgrid and the grid can obtain electricity when the electricity price is low and sell energy when the electricity price is high, thereby optimizing power dispatch and reducing operating costs.

[0011] The above-mentioned microgrid quasi-periodic energy management method based on time series prediction and linear programming is characterized in that the method further takes into account electricity price fluctuations and system constraints, and optimizes the use and storage of electricity during the energy exchange process while ensuring battery life, thereby maximizing the economic benefits of the microgrid.

[0012] The present invention effectively improves energy utilization efficiency, reduces dependence on traditional energy, and has sustainability and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is the microgrid model in the present invention.

[0014] Figure 2 This is a schematic diagram of the operation results of the microgrid system in an island state.

[0015] Figure 3It is a comparison chart of the measured value of load power and the predicted value of Holt-Winters method.

[0016] Figure 4 It is the result diagram of the optimal method for solving linear programming.

[0017] Figure 5 It is a comparison chart of the total cost of uniform energy conversion and linear programming energy conversion.

[0018] Figure 6 yes Figure 1 Further model diagrams. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand and implement this patent, the specific implementation methods of this application are now described in detail in conjunction with the drawings in the specification.

[0020] Please see Figures 1 to 6 The present invention provides a microgrid quasi-periodic energy management method based on time series prediction and linear programming. The specific microgrid model schematic diagram is as follows: Figure 1 , 6 As shown, the method comprises the following steps: Step 1: According to the capacity of the energy storage system, load power and renewable energy power in the microgrid, a microgrid quasi-periodic energy management system model is established.

[0021] The step 1 specifically includes the following steps: the four main components include an energy storage system, renewable energy, a load and an external power grid. Specifically, the energy storage system is represented by a battery. Renewable energy is represented by photovoltaic power generation. When photovoltaic power is sufficient, it generates electricity to supply the load or charge the battery, but when photovoltaic power is insufficient, the battery or the external power grid is required to provide energy to supply the load. In the present invention, the capacity of the battery is 1000Ah, the maximum power of the battery is 10kW, and the capacity of the power grid is considered to be unlimited. The photovoltaic power adopts a typical photovoltaic power curve, with a power generation of 0W from 20h to 4h and a peak value from 14h to 15h. The load adopts a typical residential load power curve, in which the power load reaches a peak at 9 hours, 19 hours and 22 hours. The island operation status of the microgrid is as follows Figure 2 , where P pv and P load Represent the power of renewable energy such as photovoltaic and load, P battery is the battery power, SOC is the state of charge, P grid Indicates the external grid power.

[0022] Step 2: Based on the quasi-periodicity assumption, time series forecasting of load power, internal power and electricity price is performed.

[0023] The step 2 specifically includes the following steps: predicting the result by the Holt-Winters method as follows: Figure 3 As shown, the bold line represents the measured load power of the load. The dashed line represents the predicted value. The prediction is performed at 6 time points, corresponding to the 6 dashed lines. The prediction times are 1h, 5h, 9h, 13h, 17h and 21h, represented by closed marks. The data of the previous day are used in the prediction process. The adjustable parameters a, b and c of the Holt-Winters method are determined by minimizing the error between measurement and prediction at the first prediction (1h). Figure 3 It is obvious that the prediction performance is good. As the number of measurement points increases, the prediction error becomes smaller and smaller. The results show that the use of real-time measurement data can indeed improve the accuracy of the full-day data estimation, thereby improving the energy exchange method.

[0024] Step 3: Use linear programming method to optimize the energy exchange method between the microgrid and the grid.

[0025] The step 3 specifically includes the following steps: When performing linear programming, in order to simplify the problem and make the solution of the linear programming more stable and reliable, consider taking the time granularity of the analysis as 1 hour, assuming that the power exchanged between the grid and the microgrid remains the same during 1 hour. The load power, photovoltaic power and electricity price used in the linear programming are taken as the average value of each 1 hour. The power vector p and the total cost f can be expressed as (5) and (6). At this time, the variable of the linear programming solution is the power exchanged between the microgrid and the grid at each moment.

[0026] Step 4: Considering the fluctuation of electricity prices and system constraints, adjust the energy distribution of the microgrid to minimize the energy cost.

[0027] The step 4 specifically includes the following steps: Considering the fluctuation of electricity prices and system constraints, the power vector needs to satisfy the constraint that the net power can keep the battery SOC restored to its initial value in one cycle of operation as shown in formula (7), and the power vector P is required to ensure that the battery SOC is between 0 and 100 at each moment. According to power balance, this condition is expressed as formula (8). In addition, the instantaneous power of the battery cannot exceed its maximum power of 10kW as shown in formula (9): Through discretization and normalization, we obtain a 24-dimensional linear programming problem containing 1 equality constraint obtained from (7), 46 inequality constraints obtained from (8), and 24 boundary constraints obtained from (9). Based on these constraints, we can perform linear programming to solve the minimum cost f.

[0028] In order to verify the quasi-periodic energy management method of a microgrid based on time series prediction and linear programming of the present invention, a power exchange method is planned within a day by simulating an electricity price curve that fluctuates over time. Figure 4 The simulation results of the optimal method solved by linear programming. Price is the electricity price curve for a simulated day, which fluctuates between 0.45 yuan / kWh and 1.45 yuan / kWh, reaching a low point in 4 hours and a peak in 16 hours, which is similar to the photovoltaic power curve. grid The optimal solution of grid-to-grid power exchange solved by linear programming method is shown. When the electricity price is low, the microgrid tends to obtain energy from the grid, while when the electricity price is high, the microgrid tends to provide energy to the grid. pv Image, due to the increase in photovoltaic power, the battery does not need to discharge to provide a large amount of power to the load, and the maximum capacity of the battery limits its further absorption of electricity from the grid. And after one day of operation of the microgrid, the battery capacity has recovered to 80%, meeting the requirements of sustainable operation. At the same time, the electricity price is higher in 12 hours, but the microgrid still absorbs electricity from the grid. This is because the electricity price will be higher later, and in order to keep the final SOC unchanged, it is necessary to absorb part of the electricity in advance to charge the battery.

[0029] In order to more clearly show the advantages of linear programming energy conversion, the total cost of uniform energy conversion and linear programming energy conversion is compared. Figure 5 As shown. In the first few hours, the former costs less than the latter, but after 11 hours, the latter is always less than the former. When the linear programming method is used, the total cost does not rise all the time, but there is a downward interval, corresponding to the situation where the microgrid is discharging to the grid. After one day of operation, the total cost of unified charging is 0.0364 yuan, which is 59% lower than the total cost of energy conversion by the linear programming solution method of 0.0149 yuan.

[0030] The present invention effectively improves energy utilization efficiency, reduces dependence on traditional energy, and has sustainability and economic benefits.

[0021] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including the equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A microgrid quasi-periodic energy management method based on time series prediction and linear programming, characterized in that: The steps include: Step 1: Establish a microgrid quasi-periodic energy management system model based on the capacity of the energy storage system, load power and renewable energy power in the microgrid; Step 2: Based on the quasi-periodicity assumption, time series forecasting of load power, internal power and electricity price is performed; Step 3: Use linear programming method to optimize the energy exchange method between the microgrid and the grid; Step 4: Considering the fluctuation of electricity prices and system constraints, adjust the energy distribution of the microgrid to minimize the energy cost.

2. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 1, characterized in that: The microgrid quasi-periodic energy management system model includes four main components: energy storage system, renewable energy, load and external grid. These components cooperate with each other to achieve efficient management and utilization of energy.

3. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 1 is characterized in that: The time series prediction of the quasi-periodic assumption in step 2 adopts the Holt-Winters method to deal with the periodic fluctuations of load and endogenous power. Especially in the scenario where daily load changes and electricity price fluctuations are significant, the data of the previous period is smoothed to generate prediction results including short-term trends, long-term trends and periodic changes. The model can adapt to the dynamic fluctuations of load power and endogenous power, effectively improve the accuracy of the prediction, and thus provide accurate input data for subsequent linear programming.

4. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 1 is characterized by: The linear programming described in step 3 predicts future electricity prices and power fluctuations through the weighted moving average method, and optimizes power exchange. The weighted moving average method processes historical electricity prices and power data by introducing different weights, giving the latest data a higher weight, so that the prediction results of future electricity prices and power fluctuations can be dynamically adjusted. On this basis, the energy exchange method between the microgrid and the power grid can obtain electricity when the electricity price is low and sell energy when the electricity price is high, thereby optimizing power dispatch and reducing operating costs.

5. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 1, characterized in that: The method further considers electricity price fluctuations and system constraints, optimizes the use and storage of electricity during the energy exchange process while ensuring battery life, and maximizes the economic benefits of the microgrid.

6. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 3 is characterized by: The Holt-Winters method introduces the Winters period term on the basis of the Holt model to deal with the fixed-period fluctuation behavior in the time series of monthly, quarterly, and weekly data. Multiple Winters terms are introduced to deal with the coexistence of multiple periods. The Holt-Winters method is used to predict the data of the day. The prediction is expressed as follows: y i+h =l i +ht i +s i-k+h (1) Where y is the forecast value, l is the smoothed value, t is the trend value, s is the seasonal value, k is the length of the season, and h is the number of forecast points; the calculation of the above values ​​is expressed as follows: l i =a(y i -s i-k )+(1-a)(l i-1 -t i-1 ) (2) t i =b(l i -l i-k )+(1-b)t i-1 (3) s i =c(y i -l i )+(1-c)s i-1 (4) where a, b, and c represent level, trend, and seasonal parameters, respectively; the parameters can be chosen arbitrarily and the forecast performance will vary.

7. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 4 is characterized by: When performing linear programming, in order to simplify the problem and make the solution of linear programming more stable and reliable, the time granularity of the analysis is considered to be 1 hour. It is assumed that the power exchanged between the grid and the microgrid remains the same during 1 hour. The load power, photovoltaic power and electricity price used in the linear programming are taken as the average value of each 1 hour. The power vector p and the total cost f are expressed as follows: P=[x1,x2,x3…,x 24 ] T (5) f=A price P (6) Among them, A price represents the electricity price per hour, and f is the objective function.

8. The microgrid quasi-periodic energy management method based on time series prediction and linear programming according to claim 1, characterized in that: Step 4 specifically includes the following steps: Considering the fluctuation of electricity prices and system constraints, the power vector needs to satisfy the constraint that the net power can keep the battery SOC restored to its initial value in one cycle of operation as shown in formula (7), and the power vector P is required to ensure that the battery SOC is between 0 and 100 at each moment; according to power balance, this condition is expressed as formula (8), in addition, the instantaneous power of the battery cannot exceed its maximum power of 10kW as shown in formula (9): Through discretization and normalization, a 24-dimensional linear programming problem is obtained, which includes an equality constraint obtained from (7), 46 inequality constraints obtained from (8), and 24 boundary constraints obtained from (9); based on these constraints, linear programming is performed to solve the minimum cost f.

Images