Industrial and commercial energy storage system demand prediction and control method and system

Abstract

The present application relates to the field of industrial and commercial energy storage technology, and in particular to a method and system for predicting and controlling demand for industrial and commercial energy storage systems. By constructing an electric meter demand prediction model based on an LSTM neural network, the future demand of the energy storage system is accurately predicted, and the time period when the electric meter demand prediction value is higher than a first preset threshold is automatically identified. These time periods are often critical periods for demand management. Then, within the time period, the output power of the energy storage system within this time period is intelligently and smoothly increased based on the electric meter demand prediction value within the time period and the output power of the energy storage converter predicted by the MPC algorithm, thereby lowering the instantaneous demand peak of the power grid. This control method not only helps the factory reduce electricity costs, but also maximizes the use of energy storage resources by planning the charging and discharging behavior of the energy storage system in advance, significantly improving the overall economic benefits of the energy storage system.

Description

Technical Field

[0001] The present application relates to the field of industrial and commercial energy storage technology, and in particular to a method and system for predicting and controlling demand for industrial and commercial energy storage systems. Background Art

[0002] When discussing advanced strategies for energy management, demand management has become an indispensable part of the industrial and commercial fields. Demand, that is, the maximum load demand of the power system in a specific time period, is an important indicator for measuring the balance between energy consumption and supply. With the widespread application of renewable energy and the profound transformation of the energy structure, how to effectively reduce demand has become a key challenge to improve energy utilization efficiency and promote sustainable energy development. In this context, industrial and commercial energy storage systems, with their unique energy storage and release capabilities, have become an important tool for reducing demand and optimizing energy allocation.

[0003] Industrial and commercial energy storage control systems can release stored electrical energy during peak load periods by intelligently managing the charging and discharging process of energy storage equipment, alleviating grid pressure and effectively reducing peak demand. At the same time, they can charge during low load periods to achieve reasonable storage and reuse of electrical energy. The existing patent application with publication number CN110148957A provides a demand control method based on an energy storage system, which controls the energy storage system by combining the real-time load demand of the target user, the real-time load power, and the charge state of the energy storage system, thereby improving the demand reduction effect. However, this control method mainly relies on real-time data for decision-making, and therefore has certain limitations in terms of energy storage resource utilization. Summary of the invention

[0004] Based on this, it is necessary to provide a method and system for demand prediction and control of industrial and commercial energy storage systems to address the problem that existing demand control of energy storage systems has certain limitations in terms of energy storage resource utilization.

[0005] In a first aspect, the present application provides a method for predicting and controlling demand for an industrial and commercial energy storage system. The method comprises:

[0006] Step S1, obtaining the grid voltage measurement value, the grid current measurement value and the energy storage converter output voltage measurement value at time k;

[0007] Step S2, based on the grid voltage measurement value, the grid-connected current measurement value and the energy storage converter output voltage measurement value at time k, combined with the MPC algorithm, obtain the energy storage converter output power prediction value at time k+1;

[0008] Step S3, obtaining the historical value of the electricity meter demand;

[0009] Step S4, obtaining an electric meter demand prediction value according to the electric meter demand historical value and an electric meter demand prediction model based on an LSTM neural network;

[0010] Step S5, obtaining a peak demand value during a peak period from the electricity meter demand prediction value; the peak demand value during a peak period is a time period when the electricity meter demand prediction value is higher than a first preset threshold value;

[0011] Step S6, smoothly increasing the output power of the energy storage system within the peak demand value during the peak period according to the meter demand forecast value within the peak demand value during the peak period and the energy storage converter output power forecast value.

[0012] Furthermore, the method further comprises:

[0013] S7, controlling the output power of the energy storage system outside the peak demand value during the peak period to be the predicted value of the output power of the energy storage converter.

[0014] Furthermore, the step S5 comprises:

[0015] Step S51, using an IQR statistical tool to obtain a peak demand value during a peak period from the electricity meter demand forecast value; the peak demand value during a peak period is a time period when the electricity meter demand forecast value is higher than Q3+1.5*IQR;

[0016] The step S6 comprises:

[0017] Step S61, within the peak demand value during the peak period, if the meter demand prediction value is within the Q3-Q1 interval, the output power of the energy storage system is controlled to be the sum of the energy storage converter output power prediction value and the average energy storage demand prediction value within the Q3-Q1 interval; if the meter demand prediction value is within the Q3+1.5*IQR interval, the output power of the energy storage system is controlled to be the sum of the energy storage converter output power prediction value and the maximum energy storage demand prediction value within the Q3+1.5*IQR interval.

[0018] Furthermore, the step S2 comprises:

[0019] Step S20, constructing a topological structure based on the energy storage system KVL equation and power equation in the coordinate system; wherein the KVL equation is as follows:

[0020] In the formula, represents the grid-connected equivalent inductance, Represents the grid-connected equivalent resistance, Indicates that the grid-connected current measurement value is The weight on the axis, Indicates that the grid-connected current measurement value is The weight on the axis, Indicates that the grid voltage measurement value is The weight on the axis, Indicates that the grid voltage measurement value is The weight on the axis, Indicates that the output voltage measurement value of the energy storage converter is The weight on the axis, Indicates that the output voltage measurement value of the energy storage converter is Components on the axis;

[0021] The power equation is as follows:

[0022] In the formula, represents the active power of the energy storage converter, Indicates the reactive power of the energy storage converter;

[0023] Step S21, discretizing the KVL equation using the forward Euler formula to obtain a predicted value of the output current of the energy storage converter at time k+1; the expression of the predicted value of the output current of the energy storage converter at time k+1 is as follows:

[0024] In the formula, The predicted value of the output current of the energy storage converter at the k+1 moment is The weight on the axis, The predicted value of the output current of the energy storage converter at the k+1 moment is The weight on the axis, is the sampling period, The measured value of the grid voltage at time k is The weight on the axis, The measured value of the grid voltage at time k is The weight on the axis, The output voltage measurement value of the energy storage converter at time k is The weight on the axis, The output voltage measurement value of the energy storage converter at time k is The weight on the axis, The grid-connected current measurement value at time k is The weight on the axis, The grid-connected current measurement value at time k is Components on the axis;

[0025] Step S22, constructing an energy storage system output power tracking error cost function; the expression of the energy storage system output power tracking error cost function is as follows:

[0026] In the formula, represents the output power of the energy storage system, represents the actual active power of the energy storage converter at time k+1, represents the predicted active power of the energy storage converter at time k+1, represents the actual reactive power of the energy storage converter at time k+1, represents the predicted reactive power of the energy storage converter at time k+1;

[0027] Step S23, with the goal of minimizing the energy storage system output power tracking error value function, obtain the energy storage converter output voltage prediction value at time k+1 according to the energy storage converter output current prediction value at time k+1; the expression of the energy storage converter output voltage prediction value at time k+1 is as follows:

[0028] In the formula, The predicted value of the output voltage of the energy storage converter at the k+1 moment is The weight on the axis, The predicted value of the output voltage of the energy storage converter at the k+1 moment is Components on the axis;

[0029] Step S24, obtaining the predicted value of the output power of the energy storage converter at the k+1 moment according to the predicted value of the output current of the energy storage converter at the k+1 moment and the predicted value of the output voltage of the energy storage converter at the k+1 moment.

[0030] Furthermore, the prediction results of the electricity meter demand prediction model are evaluated using mean absolute error, root mean square error, mean absolute percentage error and relative error.

[0031] Furthermore, after step S3 and before step S4, the method further includes:

[0032] Step S31, using a box plot to analyze the historical value of the electricity meter demand to obtain abnormal data;

[0033] Step S32, correcting the abnormal data according to the average value of the upper and lower adjacent points of the abnormal data.

[0034] Furthermore, the step S4 further includes:

[0035] Step S41, obtaining the historical output power value and the historical energy storage SOC value of the energy storage converter;

[0036] Step S42, obtaining the electricity meter demand prediction value according to the electricity meter demand historical value, the energy storage converter output power historical value, the energy storage SOC historical value and the electricity meter demand prediction model.

[0037] Furthermore, the step S3 comprises:

[0038] Step S33, obtaining the historical value of the electricity meter demand;

[0039] Step S34: using the SED mechanism to encode and decode the historical value of the electricity meter demand.

[0040] In a second aspect, the present application also provides a demand prediction and control system for an industrial and commercial energy storage system. The system includes a data acquisition and display module, a demand prediction module, and a demand optimization control module;

[0041] The data acquisition and display module is used to obtain the grid voltage measurement value, grid-connected current measurement value and energy storage converter output voltage measurement value at time k, as well as the meter demand history value;

[0042] The demand prediction module is used to obtain the output power prediction value of the energy storage converter at time k+1 based on the grid voltage measurement value, grid-connected current measurement value, and energy storage converter output voltage measurement value at time k, combined with the MPC algorithm; obtain the meter demand prediction value according to the meter demand history value and the meter demand prediction model based on the LSTM neural network;

[0043] The demand optimization control module is used to obtain the peak demand value during the peak period from the electricity meter demand prediction value; the peak demand value during the peak period is the time period when the electricity meter demand prediction value is higher than the first preset threshold value; and the output power of the energy storage system within the peak demand value during the peak period is smoothly increased according to the electricity meter demand prediction value within the peak demand value during the peak period and the output power prediction value of the energy storage inverter.

[0044] Furthermore, the data collection and display module is designed with Hadoop distributed architecture, and processes the real-time collected grid voltage measurement values, grid-connected current measurement values, and energy storage converter output voltage measurement values ​​based on the Flink platform and Storm data pipeline model.

[0045] The above-mentioned industrial and commercial energy storage system demand prediction and control method and system accurately predicts the future demand of the energy storage system by constructing an electricity meter demand prediction model based on the LSTM neural network, and automatically identifies the time period when the electricity meter demand prediction value is higher than the first preset threshold. These time periods are often critical periods for demand management. Then, within this time period, the output power of the energy storage system within this time period is further adjusted intelligently and smoothly based on the electricity meter demand prediction value within this time period and the output power of the energy storage inverter predicted by the MPC algorithm, thereby lowering the instantaneous demand peak of the power grid. This control method not only helps the factory reduce electricity costs, but also maximizes the use of energy storage resources by planning the charging and discharging behavior of the energy storage system in advance, significantly improving the overall economic benefits of the energy storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a flow chart of a method for predicting and controlling demand for an industrial and commercial energy storage system in one embodiment;

[0047] Figure 2 A structural topological diagram of an energy storage system in one embodiment;

[0048] Figure 3 A control block diagram of an MPC algorithm in one embodiment;

[0049] Figure 4 A schematic diagram of the structure of an industrial and commercial energy storage system demand prediction and control system in one embodiment;

[0050] Figure 5 A schematic diagram of historical data of an energy storage system based on a demand prediction and control method for an industrial and commercial energy storage system in one embodiment;

[0051] Figure 6 A schematic diagram of energy storage system prediction data based on an industrial and commercial energy storage system demand prediction and control method in one embodiment;

[0052] Figure 7 Schematic diagram of real-time control of an energy storage system based on a demand prediction and control method for an industrial and commercial energy storage system in one embodiment. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0054] Example 1

[0055] like Figure 1 As shown, the present application provides a method for demand prediction and control of an industrial and commercial energy storage system, comprising:

[0056] Step S1, obtaining the grid voltage measurement value, grid-connected current measurement value and energy storage converter output voltage measurement value at time k.

[0057] Among them, the grid voltage measurement value reflects the current voltage state of the grid, the grid-connected current measurement value represents the current exchange between the energy storage system and the grid, and the energy storage converter (PCS, Power Conversion System) is a key device connecting the energy storage battery and the grid, and its output voltage directly affects the output power and efficiency of the energy storage system.

[0058] Step S2, based on the grid voltage measurement value, grid-connected current measurement value and energy storage converter output voltage measurement value at time k, combined with the MPC algorithm, obtain the energy storage converter output power prediction value at time k+1.

[0059] Where k is a positive integer, k=1, 2, 3... Model Predictive Control (MPC) is an advanced control strategy that predicts future system behavior based on the current system state and prediction model, and formulates control strategies accordingly. In a preferred embodiment, firstly, a topological structure of the energy storage system is constructed. KVL equation and power equation in the coordinate system. Specifically, taking MMHC-BESS (modular multilevel hybrid converter-battery energy storage system) as an example, its topology is as follows Figure 2 As shown in the figure, the system consists of n half-bridge submodules cascaded to form a single-phase bridge arm. Each phase bridge arm is connected to an external H bridge to flip the output voltage. The output end is connected to an L-type filter and connected to the grid. The working state of the submodule is determined by the power switch devices D1, D2 and the bypass switch K; in normal operation, the bypass switch K is in the disconnected state; when the submodule fails, the bypass switch K is immediately closed to isolate the faulty submodule. When D1 is closed and D2 is disconnected, the submodule is in the on state, and the output voltage is the voltage of the submodule energy storage battery. (i=1, 2, 3, ..., n, represents the submodule serial number); when D1 is disconnected and D2 is closed, the submodule is in the cut-off state and outputs 0 voltage. Figure 2 middle represents the grid-connected equivalent inductance, Represents the grid-connected equivalent resistance, Indicates the measured value of the grid-connected current. The KVL equation in the coordinate system is shown in formula (1), and the power equation is shown in formula (2): (1)

[0060] In the formula, Indicates the grid-connected current measurement value is The weight on the axis, Indicates the grid-connected current measurement value is The weight on the axis, Indicates the measured value of the grid voltage The weight on the axis, Indicates the measured value of the grid voltage The weight on the axis, Indicates that the output voltage measurement value of the energy storage converter is The weight on the axis, Indicates that the output voltage measurement value of the energy storage converter is Components on the axis.

[0061] (2)

[0062] In the formula, represents the active power of the energy storage converter, Indicates the reactive power of the energy storage converter;

[0063] Then, the forward Euler formula is used to discretize the KVL equation shown in formula (1) to obtain the predicted value of the energy storage converter output current at time k+1 as shown in formula (3): (3)

[0064] In the formula, The predicted value of the energy storage converter output current at time k+1 is The weight on the axis, The predicted value of the energy storage converter output current at time k+1 is The weight on the axis, is the sampling period, is the measured value of the grid voltage at time k. The weight on the axis, is the measured value of the grid voltage at time k. The weight on the axis, The measured value of the output voltage of the energy storage converter at time k is The weight on the axis, The measured value of the output voltage of the energy storage converter at time k is The weight on the axis, is the measured value of the grid-connected current at time k. The weight on the axis, is the measured value of the grid-connected current at time k. The components on the axis. Compared with the backward Euler method, the forward Euler method has smaller computational complexity and higher computational efficiency.

[0065] The principle of the classic MPC algorithm is: first, establish a value function according to the control target, then calculate the corresponding value function value g(i) under all switching states of the converter, and finally compare the value function values ​​of all states, and select the switching state S(i) that minimizes the value function value as the input state of the converter at the next moment. By minimizing the error between the measured output and the predicted output, feedback correction is performed to achieve the optimal control under the current control target. The specific control block diagram is as follows: Figure 3 The value function of this embodiment is the energy storage system output power tracking error value function as shown in formula (4), which is used to quantify the difference between the actual output power of the energy storage system and the predicted output power: (4)

[0066] In the formula, represents the output power of the energy storage system, represents the actual active power of the energy storage converter at time k+1, represents the predicted active power of the energy storage converter at time k+1, represents the actual reactive power of the energy storage converter at time k+1, represents the predicted reactive power of the energy storage converter at time k+1. and Corresponds to Figure 3 In , and Corresponds to Figure 3 In , Corresponds to Figure 3 In .

[0067] Taking the minimum value function of the energy storage system output power tracking error as the goal, the predicted value of the energy storage converter output voltage at time k+1 can be obtained according to the predicted value of the energy storage converter output current at time k+1 as shown in formula (5), which is the optimal output voltage value: (5)

[0068] In the formula, The predicted value of the output voltage of the energy storage converter at time k+1 is The weight on the axis, The predicted value of the output voltage of the energy storage converter at time k+1 is Components on the axis.

[0069] Finally, the output power prediction value of the energy storage converter at time k+1 is obtained according to the output current prediction value of the energy storage converter at time k+1 and the output voltage prediction value of the energy storage converter at time k+1.

[0070] Specifically, the predicted value of the output current of the energy storage converter shown in formula (3) is multiplied by the predicted value of the output voltage of the energy storage converter shown in formula (5) to obtain the predicted value of the output power of the energy storage converter at time k+1.

[0071] Step S3, obtaining the historical value of the electricity meter demand.

[0072] Among them, the meter demand is the meter-measured power, including the power directly consumed by the power grid and the active power of the energy storage converter interacting with the power grid. Specifically, after obtaining the historical value of the meter demand, the box plot can be used to analyze the historical value of the meter demand to obtain abnormal data, and then the abnormal data can be corrected according to the average value of the upper and lower adjacent points of the abnormal data. Box Plot, also known as box and whisker plot or box plot, is a statistical chart used to show the distribution of data. It can provide five key statistics about the data set (minimum, first quartile, median, third quartile and maximum value), and show the discreteness and abnormal value of the data by drawing boxes and whiskers. Exemplarily, based on the collected historical values ​​of the electricity meter demand, the statistics required for the box plot are calculated, including the minimum value, the first quartile, the median, the third quartile and the maximum value, and then a number axis is drawn to represent the range of data values. Next, two line segments are drawn above and below the box, called box whiskers, whose lengths correspond to the maximum and minimum values ​​of the data, respectively. The box represents the quartile range of the data, and the median is represented by a horizontal line. In the box plot, points outside the box whiskers are considered outliers, which may represent extreme values ​​of the data or erroneous records. After that, two adjacent data points above and below the outlier are determined, and the average of the two adjacent data points is calculated, and then the outlier is replaced with the calculated average.

[0073] Furthermore, a SED (Simple Encoder-Decoder) mechanism can be introduced to encode and decode the corrected historical values ​​of the electricity meter demand, so that the electricity meter demand prediction model can better learn the global temporal characteristics and long-range dependencies of the time series data.

[0074] Step S4, obtaining the electricity meter demand prediction value according to the electricity meter demand historical value and the electricity meter demand prediction model based on the LSTM neural network.

[0075] Among them, the historical values ​​of electricity meter demand can include monthly electricity meter demand historical values, quarterly electricity meter demand historical values, annual electricity meter demand historical values, etc. LSTM neural network is good at processing time series data and can capture long-term dependencies in the data, so it is suitable for predicting the future demand of energy storage systems. The internal structure of LSTM consists of three parts: forget gate, input gate, and output gate. The calculation formula of the forget gate is shown in formula (6), the calculation formula of the input gate is shown in formula (7), and the calculation formula of the output gate is shown in formulas (8)-(9): (6) (7) (8) (9)

[0076] In equations (6)-(8), σ is the Sigmoid function, W f is the weight matrix of the forget gate, b f is the bias term of the forget gate, h t-1 is the output value of LSTM at the previous moment, x t is the input value of the network at the current moment, W i is the weight matrix of the input gate, b i is the bias term of the input gate, W o is the weight matrix of the output gate, b o is the bias term of the output gate, h t is the output value of LSTM at the current moment, C t is the unit state at the current moment.

[0077] In addition, the prediction results of the electric meter demand prediction model in this embodiment are evaluated by the mean absolute error, root mean square error, mean absolute percentage error and relative error. The calculation formula of the mean absolute error MAE is shown in formula (10), the calculation formula of the root mean square error MAPE is shown in formula (11), the calculation formula of the mean absolute percentage error RMSE is shown in formula (12), and the calculation formula of the relative error RE is shown in formula (13): (10) (11) (12) (13)

[0078] In the formula, n represents the length of the input sequence, Represents the real value at the current moment, represents the predicted meter demand, Indicates the actual meter demand.

[0079] Furthermore, in order to improve the accuracy of the meter demand prediction value obtained by the meter demand prediction model, the historical value of the energy storage converter output power and the historical value of the energy storage SOC can also be obtained, and the meter demand prediction value is obtained according to the historical value of the meter demand, the historical value of the energy storage converter output power, the historical value of the energy storage SOC and the meter demand prediction model. By comprehensively considering the historical data of the meter demand, the output power of the energy storage converter and the energy storage SOC, the meter demand prediction model can more accurately capture the interaction between the energy storage system and the power grid, thereby generating a more accurate meter demand prediction value. Similarly, after obtaining the historical value of the meter demand, the historical value of the output power of the energy storage converter and the historical value of the energy storage SOC, a box plot can also be used to analyze these data to obtain abnormal data, and the abnormal data can be corrected according to the average value of the upper and lower adjacent points of the abnormal data, and then the corrected data is introduced into the SED mechanism for encoding and decoding.

[0080] Step S5, obtaining the peak demand value during the peak period from the electricity meter demand forecast value.

[0081] The peak demand value during the peak period is the time period when the meter demand forecast value is higher than the first preset threshold. By comparing the meter demand forecast value with the preset first threshold, the peak demand value during the peak period with higher demand is identified. These periods are usually periods with high grid load and high energy costs.

[0082] Step S6, smoothly increasing the output power of the energy storage system within the peak demand period according to the meter demand forecast value within the peak demand period and the energy storage converter output power forecast value.

[0083] Specifically, within the peak demand value during the peak period, the output power of the energy storage system can be controlled to be the sum of the energy storage inverter output power prediction value and the meter demand prediction value, so as to reduce the demand within the peak demand value during the peak period. Outside the peak demand value during the peak period, the output power of the energy storage system can be controlled to be the energy storage inverter output power prediction value at time k+1.

[0084] In a preferred embodiment, the IQR (Interquartile Range) statistical tool is used to obtain the peak demand value during the peak period from the electricity meter demand forecast value. IQR is a concept in statistics, which is used to describe the degree of dispersion of data distribution. It is equal to the difference between the third quartile (Q3) and the first quartile (Q1). Specifically, the interquartile range IQR is calculated based on the electricity meter demand forecast value predicted by the electricity meter demand forecast model. Based on the IQR, the first preset threshold is set to Q3+1.5*IQR to identify the peak demand value during the peak period. This threshold is determined based on the distribution characteristics of the data, and it represents a value that is higher but still within a reasonable range in the data set. Then traverse the electricity meter demand forecast values ​​to find those time periods that are higher than the first preset threshold. These time periods are identified as peak demand peaks during the peak period because they represent the peak periods in the electricity meter demand forecast value. In addition, IQR can also be used as a reference for judging whether the output power forecast value of the energy storage inverter deviates from the electricity meter demand forecast value. If the predicted value of the energy storage converter output power falls within the IQR interval, it is considered that the predicted value is relatively reliable; if it exceeds the IQR interval, it may be necessary to further analyze the accuracy and reliability of the prediction model in the MPC algorithm and consider corresponding adjustment measures.

[0085] After obtaining the peak demand value during the peak period, if the meter demand forecast value is within the Q3-Q1 interval, the output power of the energy storage system is controlled to be the sum of the energy storage converter output power forecast value and the average energy storage demand forecast value within the Q3-Q1 interval; if the meter demand forecast value is within the Q3+1.5*IQR interval, the output power of the energy storage system is controlled to be the sum of the energy storage converter output power forecast value and the maximum energy storage demand forecast value within the Q3+1.5*IQR interval, which means that within the peak demand value during the peak period, the energy storage system will output as much power as possible to meet the demand, so as to reduce the demand during this period. Outside the peak demand value during the peak period, the output power of the energy storage system is set according to the output power forecast value of the energy storage converter at time k+1. This control strategy combines the real-time energy storage inverter output power prediction based on the MPC algorithm and the non-real-time meter demand prediction based on the LSTM neural network, and uses the IQR statistical method to identify the peak demand during peak hours, and then flexibly adjusts the output power of the energy storage system. This allows the energy storage system to adjust power output more smoothly to adapt to different demand conditions and reduce the demand within the peak demand during peak hours.

[0086] The demand prediction and control method for industrial and commercial energy storage systems in this embodiment accurately predicts the future demand of the energy storage system by constructing an electricity meter demand prediction model based on an LSTM neural network, and automatically identifies time periods when the electricity meter demand prediction value is higher than a first preset threshold value. These time periods are often critical periods for demand management. Then, within the time period, the output power of the energy storage system within this time period is intelligently and smoothly increased based on the electricity meter demand prediction value within the time period and the output power of the energy storage converter predicted by the MPC algorithm, thereby lowering the instantaneous demand peak of the power grid. This control method not only helps the factory reduce electricity costs, but also maximizes the use of energy storage resources by planning the charging and discharging behavior of the energy storage system in advance, thereby significantly improving the overall economic benefits of the energy storage system.

[0087] Example 2

[0088] like Figure 4 As shown, the present application also provides a demand prediction and control system for industrial and commercial energy storage systems, including a data acquisition and display module, a demand prediction module, and a demand optimization control module. The system can be integrated on the local EMS of the energy storage system, and the local EMS sends commands to the PCS, and the PCS responds to the commands to control the charging and discharging of the energy storage system.

[0089] The data collection and display module is used to obtain the grid voltage measurement value, grid-connected current measurement value, energy storage converter output voltage measurement value, and meter demand history value at time k. Among them, the data collection and display module adopts the Hadoop distributed architecture design, and processes the real-time collected grid voltage measurement value, grid-connected current measurement value, and energy storage converter output voltage measurement value based on the Flink platform and Storm data pipeline model. The Storm data pipeline model is used to process continuous and unbounded data streams in real time to ensure the real-time and accuracy of data. The Flink platform provides a unified processing platform that supports the combination of stream processing and batch processing, and seamlessly connects to Hadoop batch processing jobs. In addition, the data collection and display module can also display the acquired data (including but not limited to single-month or multi-month electricity bill information, maximum meter demand, PCS active power, etc.) synchronously on the local control page, which is convenient for users to monitor and analyze data in real time.

[0090] The demand prediction module is used to obtain the output power prediction value of the energy storage converter at time k+1 based on the grid voltage measurement value, grid-connected current measurement value, and energy storage converter output voltage measurement value at time k, combined with the MPC algorithm; the meter demand prediction value is obtained according to the meter demand history value and the meter demand prediction model based on the LSTM neural network. Among them, the specific calculation process of the energy storage converter output power prediction value and the meter demand prediction value can refer to Example 1, which will not be repeated here.

[0091] The demand optimization control module is used to obtain the peak demand value during the peak period from the electricity meter demand prediction value. Among them, the peak demand value during the peak period is the time period when the electricity meter demand prediction value is higher than the first preset threshold value. In order to effectively respond to the demand for the peak demand value during the peak period, the demand optimization control module dynamically adjusts the output power of the energy storage system within the peak demand value during the peak period according to the electricity meter demand prediction value within the peak demand value during the peak period and the output power prediction value of the energy storage inverter, so as to achieve smooth increase. Specifically, the demand optimization control module accurately identifies the peak demand value during the peak period through the IQR statistical tool, and combines the output power prediction value of the energy storage inverter obtained by the MPC algorithm, and adopts a dynamic algorithm to smoothly increase the output power of the energy storage system during this period, thereby effectively reducing the electricity meter demand for the peak demand value during the peak period. Among them, the specific process of adjusting the output power of the energy storage system using a dynamic algorithm can refer to the above embodiment.

[0092] The demand prediction and control system for the industrial and commercial energy storage system of this embodiment can automatically predict the peak demand period of each month and each day and complete the corresponding power control, without the need for manual real-time monitoring of demand fluctuations, greatly reducing the difficulty of operation and maintenance. In addition, since the system is integrated on the local EMS of the energy storage system, when the network is interrupted, the system can also stably and effectively control the demand within a reasonable range.

[0093] Example 3

[0094] In order to better understand the industrial and commercial energy storage system demand prediction and control method and system in the above embodiments, based on Embodiment 1 and Embodiment 2, the following explanation is given using the Mexican factory energy storage power station as an application scenario.

[0095] The capacity of the energy storage power station in the Mexican factory is 344kWh, the rated power is 200kW, the discharge depth is 0.5C, the demand control is 2 hours, and the billing cycle is 31 days (i.e., monthly). Due to the high data density, the peak maximum demand, PCS total active power and SOC from 2024 / 10 / 07 to 2024 / 10 / 11 are selected here to represent the historical values ​​of the meter demand, the historical values ​​of the energy storage converter output power and the historical values ​​of the energy storage SOC, respectively. Figure 5 As shown. Figure 5 It can be seen that the peak maximum demand is at 20:06:24 on 2024 / 10 / 08, and the peak maximum demand is 133.65kW (obtained from meter measurement). Subsequently, the meter demand history values, energy storage converter output power history values, energy storage SOC history values, and meter demand prediction model are used to obtain the following: Figure 6 The peak maximum demand forecast data shown is the meter demand forecast value mentioned above. At the same time, in order to intuitively understand the forecast process, Figure 6The meter demand prediction model also shows the PCS total active power predicted by the energy storage converter output power history value, and the SOC value predicted by the energy storage SOC history value. It is worth noting that the final meter demand prediction value is the result of comprehensive consideration of the PCS total active power prediction data and SOC prediction data. Figure 6 It can be seen that the predicted peak demand during the peak period occurs at 20:00-20:15 on 2024 / 10 / 14, and the peak value can reach 142.38kW.

[0096] Then, based on the demand optimization control method in the above embodiment, the output power prediction value of the energy storage converter at time k+1 and the meter demand prediction value are combined to obtain the following: Figure 7 The actual energy storage output shown includes the actual total active power of the PCS, the actual SOC and the actual peak maximum demand. Figure 7 It can be seen that the actual peak demand is generated at 2024 / 10 / 14 20:02, and the actual peak demand is 132.42kW.

[0097] Compared with the local control method of energy storage system that some integrators use to periodically set the discharge power, the demand prediction and control method of industrial and commercial energy storage system of this application can not only achieve more timely and effective demand control, but also greatly reduce the burden of operation and maintenance management. Specifically, Table 1 shows the cost reduction of charging and discharging with constant power. Table 2 shows the cost reduction of industrial and commercial energy storage system demand prediction and control method of this application.

[0098] Table 1 Actual maximum peak demand generation time Actual maximum peak demand value kW Factory restored value (energy storage without power generation) kW Savings (peso) October 14, 20:02 174.70 240.45 28277.64

[0099] Table 2 Predict the time when the maximum peak demand will occur Predicted maximum peak demand value kW Actual maximum peak demand generation time Actual maximum peak demand value kW Factory restored value (energy storage without power generation) kW Savings (peso) October 14, 20:00-20:15 142.38 October 14, 20:09 132.42 240.45 42,170.1

[0100] It can be seen from Tables 1 and 2 that the industrial and commercial energy storage system demand prediction and control method of the present application maximizes the use of energy storage resources by planning the charging and discharging behavior of the energy storage system in advance, reduces a lot of electricity costs, and significantly improves the overall economic benefits of the energy storage system.

[0101] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for predicting and controlling demand for industrial and commercial energy storage systems, characterized in that: The method comprises: Step S1, obtaining the grid voltage measurement value, the grid current measurement value and the energy storage converter output voltage measurement value at time k; Step S2, based on the grid voltage measurement value, the grid-connected current measurement value and the energy storage converter output voltage measurement value at time k, combined with the MPC algorithm, obtain the energy storage converter output power prediction value at time k+1; Step S3, obtaining the historical value of the electricity meter demand; Step S4, obtaining an electric meter demand prediction value according to the electric meter demand historical value and an electric meter demand prediction model based on an LSTM neural network; Step S5, obtaining a peak demand value during a peak period from the electricity meter demand prediction value; the peak demand value during a peak period is a time period when the electricity meter demand prediction value is higher than a first preset threshold value; Step S6, smoothly increasing the output power of the energy storage system within the peak demand value during the peak period according to the meter demand forecast value within the peak demand value during the peak period and the energy storage converter output power forecast value.

2. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 1, characterized in that: The method further comprises: S7, controlling the output power of the energy storage system outside the peak demand value during the peak period to be the predicted value of the output power of the energy storage converter.

3. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 2, characterized in that: The step S5 comprises: Step S51, using an IQR statistical tool to obtain a peak demand value during a peak period from the electricity meter demand forecast value; the peak demand value during a peak period is a time period when the electricity meter demand forecast value is higher than Q3+1.5*IQR; The step S6 comprises: Step S61, within the peak demand value during the peak period, if the meter demand prediction value is within the Q3-Q1 interval, the output power of the energy storage system is controlled to be the sum of the energy storage converter output power prediction value and the average energy storage demand prediction value within the Q3-Q1 interval; if the meter demand prediction value is within the Q3+1.5*IQR interval, the output power of the energy storage system is controlled to be the sum of the energy storage converter output power prediction value and the maximum energy storage demand prediction value within the Q3+1.5*IQR interval.

4. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 1, characterized in that: The step S2 comprises: Step S20, constructing a topological structure based on the energy storage system KVL equation and power equation in the coordinate system; wherein the KVL equation is as follows: ； In the formula, represents the grid-connected equivalent inductance, Represents the grid-connected equivalent resistance, Indicates that the grid-connected current measurement value is The weight on the axis, Indicates that the grid-connected current measurement value is The weight on the axis, Indicates that the grid voltage measurement value is The weight on the axis, Indicates that the grid voltage measurement value is The weight on the axis, Indicates that the output voltage measurement value of the energy storage converter is The weight on the axis, Indicates that the output voltage measurement value of the energy storage converter is Components on the axis; The power equation is as follows: ； In the formula, represents the active power of the energy storage converter, Indicates the reactive power of the energy storage converter; Step S21, discretizing the KVL equation using the forward Euler formula to obtain a predicted value of the output current of the energy storage converter at time k+1; the expression of the predicted value of the output current of the energy storage converter at time k+1 is as follows: ； In the formula, The predicted value of the output current of the energy storage converter at the k+1 moment is The weight on the axis, The predicted value of the output current of the energy storage converter at the k+1 moment is The weight on the axis, is the sampling period, The measured value of the grid voltage at time k is The weight on the axis, The measured value of the grid voltage at time k is The weight on the axis, The output voltage measurement value of the energy storage converter at time k is The weight on the axis, The output voltage measurement value of the energy storage converter at time k is The weight on the axis, The grid-connected current measurement value at time k is The weight on the axis, The grid-connected current measurement value at time k is Components on the axis; Step S22, constructing an energy storage system output power tracking error cost function; the expression of the energy storage system output power tracking error cost function is as follows: ； In the formula, represents the output power of the energy storage system, represents the actual active power of the energy storage converter at time k+1, represents the predicted active power of the energy storage converter at time k+1, represents the actual reactive power of the energy storage converter at time k+1, represents the predicted reactive power of the energy storage converter at time k+1; Step S23, with the goal of minimizing the energy storage system output power tracking error value function, obtain the energy storage converter output voltage prediction value at time k+1 according to the energy storage converter output current prediction value at time k+1; the expression of the energy storage converter output voltage prediction value at time k+1 is as follows: ； In the formula, The predicted value of the output voltage of the energy storage converter at the k+1 moment is The weight on the axis, The predicted value of the output voltage of the energy storage converter at the k+1 moment is Components on the axis; Step S24, obtaining the predicted value of the output power of the energy storage converter at the k+1 moment according to the predicted value of the output current of the energy storage converter at the k+1 moment and the predicted value of the output voltage of the energy storage converter at the k+1 moment.

5. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 1, characterized in that: The prediction results of the electricity meter demand prediction model are evaluated using mean absolute error, root mean square error, mean absolute percentage error and relative error.

6. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 1, characterized in that: After step S3 and before step S4, the method further includes: Step S31, using a box plot to analyze the historical value of the electricity meter demand to obtain abnormal data; Step S32, correcting the abnormal data according to the average value of the upper and lower adjacent points of the abnormal data.

7. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 1, characterized in that: The step S4 further comprises: Step S41, obtaining the historical output power value and the historical energy storage SOC value of the energy storage converter; Step S42, obtaining the electricity meter demand prediction value according to the electricity meter demand historical value, the energy storage converter output power historical value, the energy storage SOC historical value and the electricity meter demand prediction model.

8. The method for predicting and controlling demand of industrial and commercial energy storage systems according to claim 1, characterized in that: The step S3 comprises: Step S33, obtaining the historical value of the electricity meter demand; Step S34: using the SED mechanism to encode and decode the historical value of the electricity meter demand.

9. An industrial and commercial energy storage system demand prediction and control system, characterized in that: It includes data collection and display module, demand prediction module and demand optimization control module; The data acquisition and display module is used to obtain the grid voltage measurement value, grid-connected current measurement value and energy storage converter output voltage measurement value at time k, as well as the meter demand history value; The demand prediction module is used to obtain the output power prediction value of the energy storage converter at time k+1 based on the grid voltage measurement value, grid-connected current measurement value, and energy storage converter output voltage measurement value at time k, combined with the MPC algorithm; Obtaining an electric meter demand prediction value according to the electric meter demand historical value and an electric meter demand prediction model based on an LSTM neural network; The demand optimization control module is used to obtain the peak demand value during the peak period from the electricity meter demand prediction value; the peak demand value during the peak period is the time period when the electricity meter demand prediction value is higher than the first preset threshold value; and the output power of the energy storage system within the peak demand value during the peak period is smoothly increased according to the electricity meter demand prediction value within the peak demand value during the peak period and the output power prediction value of the energy storage inverter.

10. The industrial and commercial energy storage system demand prediction and control system according to claim 9, characterized in that: The data collection and display module is designed with Hadoop distributed architecture, and processes the real-time collected grid voltage measurement values, grid-connected current measurement values, and energy storage converter output voltage measurement values ​​based on the Flink platform and Storm data pipeline model.

Images