Working regulation and control method and system for high-temperature adiabatic compressed air energy storage system

Abstract

The invention provides an acting regulation and control method and system for a high-temperature adiabatic compressed air energy storage system. The method comprises the steps that S1, a power grid load instruction is obtained; s2, inputting a power grid load instruction to obtain a salt cavern outlet pressure prediction value by using an LSTM-based neural network prediction model; s3, according to the power grid load instruction and the salt cavern outlet pressure prediction value, the salt-gas heat exchanger fused salt flow, the water-gas heat exchanger water flow and the air flow corresponding to the optimal efficiency of the unit are determined; s4, determining corresponding given values of the opening degree of the fused salt regulating valve, the rotating speed of the water pump and the opening degree of the air valve; and S5, according to the given values and the measured values of the molten salt regulating valve opening, the water pump rotating speed and the air valve opening, the deviation DV of the molten salt regulating valve opening, the water pump rotating speed and the air valve opening is determined, and the power grid load instruction requirement is met through control. The problem that the system efficiency is difficult to predict is solved, and a basis is provided for regulation and control of the high-temperature adiabatic compressed air energy storage system.

Description

Technical Field

[0001] The present invention relates to the field of high-temperature adiabatic compressed air energy storage systems, and in particular to a work control method and system for high-temperature adiabatic compressed air energy storage systems. Background Art

[0002] my country's renewable energy sector is developing rapidly, with its development and utilization gradually deepening. However, renewable energy sources such as solar and wind energy have poor energy supply stability and are prone to energy loss and waste, which has a certain impact on the power grid. At the same time, the need to improve the power grid's ability to accept, configure, and regulate clean energy is also gaining attention. Energy storage, as an important solution to the mismatch between energy supply and demand, is a key support for building a new power system, and its role and value are becoming increasingly prominent. Against this backdrop, compressed air energy storage technology has made significant progress. Based on traditional regenerative compressed air energy storage technology, the high-temperature adiabatic compressed air energy storage system uses a heat exchanger to recover the heat generated during the compression process and uses this heat to heat the air entering the expander inlet. This achieves heat exchange between energy storage and release, improves the overall efficiency of the system, and does not rely on fossil fuel regeneration, avoiding carbon dioxide emissions. It provides a low-carbon and effective solution for large-scale wind and solar power grid integration.

[0003] Due to fluctuations in renewable energy power input, changes in gas storage chamber pressure, load regulation needs, and changes in ambient temperature, high-temperature adiabatic compressed air energy storage systems often need to operate under non-design conditions. The discharge work capacity of the underground salt caverns in the system as gas storage equipment is highly uncertain, making the system efficiency difficult to predict. It is difficult to obtain the system efficiency before the end of a complete cycle, making it difficult to maintain the optimal efficiency of the unit through regulation. Summary of the Invention

[0004] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a method and device for controlling the work performance of a high-temperature adiabatic compressed air energy storage system based on machine learning prediction.

[0005] The technical solution adopted in the present invention is:

[0006] A method for controlling the work of a high-temperature adiabatic compressed air energy storage system comprises the following steps:

[0007] S1. Obtaining the grid load instruction L;

[0008] S2. Using the LSTM-based neural network prediction model, input the grid load instruction L to obtain the salt cavern outlet pressure prediction value P y ;

[0009] The construction of the LSTM-based neural network prediction model includes:

[0010] S2-1, perform data preprocessing on the time series working parameters of the high temperature adiabatic compressed air energy storage system. The time series working parameters include load L i and salt cavern outlet pressure P i The data preprocessing steps include time series reconstruction, Min-Max normalization processing and data set division.

[0011] The time series reconstruction includes correcting outliers, deleting missing values, and correctly matching the number of samples, time steps, and number of features of the time series working parameters;

[0012] The method for correcting outliers includes:

[0013] a) Identify outliers: Data that exceeds the upper and lower limits of the preset load and salt cavern outlet pressure, or data that deviates from the approximate function fitted by normal data by more than 5% is identified as an outlier;

[0014] b) Correcting outliers with average values: Select the working parameters of the two time steps before and after the outlier and take the average value to cover the outlier; the normalization process adopts Min-Max normalization, and the normalization process formula is as follows:

[0015]

[0016] Among them, x' is the normalized value after normalization, x is the input timing working parameter, x max is the maximum value of the timing working parameter, x min is the minimum value of the temporal working parameter, and the normalized data set is obtained after processing;

[0017] The data set is divided into a training set, a validation set, and a test set in proportion. The ratio of the training set to the test set is 7:3 during the first training. The division ratio is changed in each subsequent training, and the model with the highest accuracy is finally selected.

[0018] S2-2. Network structure construction. The network structure includes an input layer, multiple LSTM layers, and an output layer; the input layer receives load data, and the output layer outputs salt cavern outlet pressure data.

[0019] S2-3. Initialize the hyperparameters of the initial model. The initial model parameters include time step, number of neurons, etc.

[0020] S2-4. Initial model training. Use the training set to train the initial model. Based on the size of the training data and the complexity of the system, adjust the number of LSTM layers in the neural network model and repeatedly train until the model accuracy meets the requirements.

[0021] S3, according to the grid load instruction L and the salt cavern outlet pressure prediction value P y , using genetic algorithm as the optimization algorithm to solve the efficiency function ηmax=f0(L,P y ), and then by the mechanism model Q Y,SV =f1(ηmax),Q S,SV =f2(ηmax),Q K,SV =f3(ηmax), determine the molten salt flow rate Q of the salt-gas heat exchanger corresponding to the unit's optimal efficiency ηmax Y,SV , water flow rate Q of water-gas heat exchanger S,SV With air flow Q K,SV ;

[0022] S4, according to the molten salt flow Q of the salt-gas heat exchanger Y,SV , water flow rate Q of water-gas heat exchanger S,SV With air flow Q K,SV , through the given nonlinear relationship I between the system molten salt flow and the molten salt regulating valve opening, the water flow and the water pump speed, and the air flow and the air valve opening Y,SV =g1(Q Y,SV ), N SV =g2(Q S,SV ), I K,SV =g3(Q K,SV ), determine the corresponding molten salt regulating valve opening, water pump speed and air valve opening given value I Y,SV 、N SV and I K,SV ;

[0023] S5, according to the given value I of the molten salt regulating valve opening, water pump speed and air valve opening Y,SV 、N SV and I K,SV , and the measured values of the molten salt regulating valve opening, water pump speed and air valve opening at the current moment I Y,PV 、N PV , I K,PV By taking the difference between the given value and the measured value as the valve opening of the molten salt regulating valve, the water pump speed and the air valve opening deviation DV, the grid load instruction requirements are met through control.

[0024] The present invention also provides a work control system for a high-temperature adiabatic compressed air energy storage system, comprising a basic data extraction module, a prediction operation module, a parameter control module, and a system output module;

[0025] The basic data extraction module is used to obtain the grid load instructions and the status data of the unit, including the molten salt flow of the salt-gas heat exchanger, the water flow of the water-gas heat exchanger, and the corresponding molten salt regulating valve opening, water pump speed, and air valve opening measurement values;

[0026] The prediction operation module is used to determine the molten salt flow rate of the salt-gas heat exchanger, the water flow rate of the water-gas heat exchanger, and the air flow rate corresponding to the unit achieving optimal efficiency based on the grid load instruction and the predicted value of the salt cavern outlet pressure, thereby determining the valve opening of the molten salt regulating valve, the water pump speed, and the air valve opening set value;

[0027] The parameter control module is used to: 1. determine the deviation of the molten salt regulating valve valve opening, water pump speed and air valve opening based on the given values and measured values of the molten salt regulating valve valve opening, water pump speed and air valve opening; 2. output control instructions based on the deviation value and PID control parameters to implement control so as to meet the power grid load instruction requirements.

[0028] The system output module is used to output the status data of the unit, the variable load operation data of the unit, parameter control data, etc. according to the comprehensive control needs of the energy storage system.

[0029] Each module in the above-mentioned high-temperature adiabatic compressed air energy storage system work control device based on machine learning prediction can be fully or partially implemented through software, hardware and their combination.

[0030] The above modules may be embedded in or independent of the processor in the computer device in the form of hardware, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] This invention uses an LSTM-based neural network model to predict the operating capacity of a high-temperature adiabatic compressed air energy storage system. By inputting grid load instructions to obtain a predicted value for the salt cavern outlet pressure, this method addresses the difficulty in predicting system efficiency and provides a basis for regulating the high-temperature adiabatic compressed air energy storage system. Based on the grid load instructions and the predicted value for the salt cavern outlet pressure, the molten salt flow rate in the salt-gas heat exchanger, the water flow rate in the water-gas heat exchanger, and the air flow rate required for the unit to achieve optimal efficiency are determined. The corresponding molten salt regulating valve opening, water pump speed, and air valve opening are used as given values. Based on the deviation between the given values and the measured values, the unit load is adjusted to achieve optimal efficiency through control, thereby improving the grid's ability to absorb renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of the method provided by the present invention;

[0034] Figure 2 This is a control structure diagram of the molten salt regulating valve opening, water pump speed and air valve opening in the present invention;

[0035] Figure 3It is a schematic diagram of the system structure provided by the present invention. DETAILED DESCRIPTION

[0036] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0037] Figure 1 This is a flow chart of the work control method for the high-temperature adiabatic compressed air energy storage system provided by the present invention. Taking a 300MW unit with a total energy storage capacity of 1200MWh as an example, the process is as follows:

[0038] At a certain moment in the discharge process, the grid load instruction is obtained from the dispatching center, requiring that the unit load reach L = 250MW after 30 minutes;

[0039] The load instruction L = 250MW is input into the LSTM-based neural network prediction model, and the predicted value of the salt cavern outlet pressure P after 30 minutes is output. y ;

[0040] According to L and P y Determine the molten salt flow rate Q of the salt-gas heat exchanger corresponding to the unit's optimal efficiency ηmax Y,SV , water flow rate Q of water-gas heat exchanger S,SV With air flow Q K,SV ;

[0041] According to the molten salt flow rate Q of the salt-gas heat exchanger Y,SV , water flow rate Q of water-gas heat exchanger S,SV With air flow Q K,SV , and the nonlinear relationship between the unit's molten salt flow rate and the molten salt regulating valve opening, the water flow rate and the water pump speed, and the air flow rate and the air valve opening, determine the corresponding molten salt regulating valve opening I Y,SV , water pump speed N SV and air valve opening I K,SV is a given value;

[0042] According to I Y,SV 、N SV with I K,SV and the current moment molten salt regulating valve opening, water pump speed and air valve opening measurement value I Y,PV 、N PV , I K,PV Determine the deviation DV, input DV into the PID controller, output the molten salt regulating valve opening, water pump speed and air valve opening instruction MV, and achieve the grid load instruction requirement L through feedback control.

[0043] Figure 2This is a control structure diagram of the molten salt regulating valve opening, water pump speed, and air valve opening in the present invention. To maintain the optimal operation of the energy storage system under a given load, when the grid load instruction requirement increases, the system power generation is lower than the load. The system implements control based on the molten salt regulating valve opening, water pump speed, and the deviation of the air valve opening, reducing the molten salt regulating valve opening, lowering the water pump speed, and increasing the air valve opening; when the grid load instruction requirement decreases, the system power generation is higher than the load. The system implements control based on the molten salt regulating valve opening, water pump speed, and the deviation of the air valve opening, increasing the molten salt regulating valve opening, increasing the water pump speed, and reducing the air valve opening;

[0044] Figure 3 It is a schematic diagram of the system structure provided by the present invention. The system consists of a basic data extraction module, a prediction operation module, a parameter control module, and a system output module. Among them, the basic data extraction module obtains the grid load instruction and the status data of the unit from the high-temperature adiabatic compressed air energy storage system scheduling system, including the molten salt flow of the salt-gas heat exchanger, the water flow of the water-gas heat exchanger and the air flow to determine the corresponding molten salt regulating valve valve opening, water pump speed and air valve opening measurement values, etc., which are used for closed-loop control of system operation; the prediction operation module is used to determine the salt-gas heat exchanger molten salt flow, water flow and air flow corresponding to the unit achieving optimal efficiency according to the grid load instruction and the salt cavern outlet pressure prediction value, and determine the corresponding molten salt regulating valve valve opening. , water pump speed and air valve opening are given values; the parameter control module is used to: ① determine the valve opening of the molten salt regulating valve, the water pump speed and the air valve opening deviation according to the given values and measured values of the valve opening of the molten salt regulating valve, the water pump speed and the air valve opening; ② output control instructions to implement control according to the valve opening of the molten salt regulating valve, the water pump speed and the air valve opening deviation and the PID control parameters to meet the grid load instruction requirements; the system output module is used to output the unit status data, unit variable load operation data, parameter control data, etc. according to the comprehensive control needs of the energy storage system.

[0045] The above is a detailed introduction to the method and specific implementation of the present invention. Of course, in addition to the above examples, the present invention can also have other implementations, and any technical solutions formed by equivalent replacement or equivalent transformation fall within the scope of protection of the present invention.

Claims

1. A method for controlling the work of a high-temperature adiabatic compressed air energy storage system, characterized in that: The following steps are involved: S1. Obtaining the grid load instruction L; S2. Using the LSTM-based neural network prediction model, the grid load command L is input to obtain the predicted value Py of the salt cavern outlet pressure; S3, according to the grid load instruction L and the salt cavern outlet pressure prediction value P y , using genetic algorithm as the optimization algorithm to solve the efficiency function ηmax=f0(L,P y ), and then by the mechanism model Q Y,SV =f1(ηmax),Q S,SV =f2(ηmax),Q K,SV =f3(ηmax), determine the molten salt flow rate Q of the salt-gas heat exchanger corresponding to the unit's optimal efficiency ηmax Y,SV , water flow rate Q of water-gas heat exchanger S,SV With air flow Q K,SV ; S4, according to the molten salt flow Q of the salt-gas heat exchanger Y,SV , water flow rate Q of water-gas heat exchanger S,SV With air flow Q K,SV , through the given nonlinear relationship I between the system molten salt flow and the molten salt regulating valve opening, the water flow and the water pump speed, and the air flow and the air valve opening Y,SV =g1(Q Y,SV ), N SV =g2(Q S,SV ), I K,SV =g3(Q K,SV ), determine the corresponding molten salt regulating valve opening, water pump speed and air valve opening as a given value I Y,SV 、N SV and I K,SV ; S5, according to the given value I of the molten salt regulating valve opening, water pump speed and air valve opening Y,SV 、N SV and I K,SV , and the measured values of the molten salt regulating valve opening, water pump speed and air valve opening at the current moment I Y,PV 、N PV and I K,PV By taking the difference between the given value and the measured value as the valve opening of the molten salt regulating valve, the water pump speed and the air valve opening deviation DV, the grid load instruction requirements are met through control.

2. The method for controlling the work of a high-temperature adiabatic compressed air energy storage system according to claim 1, characterized in that: In step S2, the method for constructing the LSTM-based neural network prediction model includes: S2-1. Preprocess the data of the time series working parameters of the high temperature adiabatic compressed air energy storage system; wherein, the time series working parameters include load L i and salt cavern outlet pressure P i ;Data preprocessing includes time series reconstruction, normalization and data set partitioning; The time series reconstruction includes correcting outliers, deleting missing values, and correctly matching the number of samples, time steps, and number of features of the time series working parameters; The method for correcting outliers includes: a) Identify outliers: Data that exceeds the upper and lower limits of the preset load and salt cavern outlet pressure, or data that deviates from the approximate function fitted by normal data by more than 5% is identified as an outlier; b) Correct outliers with average values: select the working parameters of the two time steps before and after the outlier and take the average value to cover the outlier; The normalization process adopts Min-Max normalization process, and the normalization process formula is as follows: Among them, x' is the normalized value after normalization, x is the input timing working parameter, x max is the maximum value of the timing working parameter, x min is the minimum value of the temporal working parameter, and the normalized data set is obtained after processing; The data set is divided into a training set, a validation set, and a test set in proportion. The ratio of the training set to the test set is 7:3 during the first training. The division ratio is changed in each subsequent training, and the model with the highest accuracy is finally selected; S2-2. Network structure construction; the network structure includes an input layer, multiple LSTM layers, and an output layer; the input layer receives load data, and the output layer outputs salt cavern outlet pressure data; S2-3. Initialize the hyperparameters of the initial model; the initial model parameters include the time step and the number of neurons; S2-4. Initial model training: Use the training set to train the initial model; adjust the number of LSTM layers of the neural network model according to the scale of training data and system complexity and train repeatedly until the model accuracy meets the requirements.

3. A work control system for a high-temperature adiabatic compressed air energy storage system, characterized in that: Used to implement the method according to any one of claims 1-2, comprising a basic data extraction module, a prediction operation module, a parameter control module, and a system output module; The basic data extraction module is used to obtain grid load instructions and unit status data, including the molten salt flow rate of the salt-gas heat exchanger, the water flow rate and air flow rate of the water-gas heat exchanger, and the corresponding molten salt regulating valve opening, water pump speed and air valve opening measurement values; The prediction operation module is used to determine the molten salt flow rate of the salt-gas heat exchanger, the water flow rate of the water-gas heat exchanger, and the air flow rate corresponding to the unit achieving optimal efficiency based on the grid load instruction and the predicted value of the salt cavern outlet pressure, thereby determining the valve opening of the molten salt regulating valve, the water pump speed, and the air valve opening set value; The parameter control module is used to determine the deviation of the molten salt regulating valve opening, the water pump speed and the air valve opening according to the given values and measured values of the molten salt regulating valve opening, the water pump speed and the air valve opening; and output a control instruction to implement control according to the deviation value and the PID control parameter to meet the power grid load instruction requirement; The system output module is used to output the status data of the unit, the variable load operation data of the unit, and the parameter control data according to the comprehensive control needs of the energy storage system.

Images