Compressed air energy storage frequency modulation method based on neural network

Abstract

The invention relates to the technical field of new energy, in particular to a compressed air energy storage frequency modulation method based on a neural network, and the method comprises the steps: obtaining the air inlet mass flow rate of a compressor unit in a compressed air energy storage system, the frequency deviation of a power system, and the frequency deviation change rate; judging whether the frequency deviation meets a certain frequency modulation starting condition of the compressed air energy storage system or not; and when the frequency deviation meets a certain frequency modulation starting condition, the output power of the compressed air energy storage system is adjusted based on the air inlet mass flow rate and a pre-trained neural network, so that the frequency of the power system is adjusted based on the adjusted output power until the frequency meets a preset frequency stability condition. Therefore, the technical problems that the intelligent processing capacity of a neural network for multi-dimensional parameters cannot be fully played, the advantages of high response speed, environment-friendly high-quality transient frequency modulation power supply and the like of the compressed air energy storage system cannot be reflected, and the active power control precision of the compressed air energy storage system is relatively low in the prior art are solved.

Description

Technical Field

[0001] The present application relates to the field of new energy technology, and in particular to a compressed air energy storage frequency modulation method based on a neural network. Background Art

[0002] In recent years, with the aggravation of environmental pollution and resource shortage, renewable energy has gradually received attention, and the traditional power system is shifting towards sustainable development. However, with the large-scale integration of renewable energy into the power grid, the randomness of its output will cause problems such as grid frequency imbalance, posing challenges to the economic operation and power quality of the power grid. In the field of frequency regulation, compressed air energy storage has large energy storage capacity, wide frequency regulation range, and natural electromechanical transient and rotational inertia characteristics, so it can be well coupled with the power grid.

[0003] In the related technology, the acquired grid frequency deviation, the energy storage operation data and energy release operation data of the air energy storage unit, the operation data of the wind turbine set and the wind information of the area where the wind turbine set is located can be input into the neural network model, and then the energy storage power of the air energy storage unit, the energy release power of the air energy storage unit and the electromagnetic power of the wind turbine set can be adjusted; it is also possible to obtain the compressed air energy storage frequency modulation droop control coefficient by establishing a gas storage model and a heat storage tank heat storage model, and then adjust the valve mass flow rate adjustment instruction.

[0004] However, in the related technologies, the research on compressed air energy storage mainly involves the analysis of system structure and system efficiency from a thermodynamic perspective. It cannot give full play to the intelligent processing capabilities of neural networks for multi-dimensional parameters, and cannot reflect the advantages of compressed air energy storage systems such as fast response speed, environmentally friendly high-quality transient frequency-modulated power supply, etc. The accuracy of active power control of compressed air energy storage systems is low and urgently needs to be improved. Summary of the invention

[0005] The present application provides a compressed air energy storage frequency modulation method based on a neural network to solve the problems in the related technology that the neural network's intelligent processing capability for multi-dimensional parameters cannot be fully utilized, the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply cannot be reflected, and the accuracy of active power control of the compressed air energy storage system is low.

[0006] The first aspect of the present application provides a compressed air energy storage frequency modulation method based on a neural network, comprising the following steps: obtaining the intake mass flow rate of a compressor group in a compressed air energy storage system, the frequency deviation of a power system, and the frequency deviation change rate; determining whether the frequency deviation satisfies a preset frequency modulation start-up condition of the compressed air energy storage system; if the frequency deviation satisfies the preset frequency modulation start-up condition, adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and a pre-trained neural network, so as to adjust the frequency of the power system based on the adjusted output power until the frequency satisfies a preset frequency stability condition.

[0007] Optionally, in one embodiment of the present application, the output power of the compressed air energy storage system is adjusted based on the intake mass flow rate and a pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets a preset frequency stability condition, including: inputting the intake mass flow rate, the frequency deviation and the frequency deviation change rate into a pre-trained neural network to obtain a target proportional parameter and a target integral parameter of the compressed air energy storage system, and controlling the compressed air energy storage system to enter a short-term increase phase based on the target proportional parameter and the target integral parameter; when the compressed air energy storage system is in the short-term increase phase, judging the frequency deviation The difference is greater than a first preset frequency deviation of the power system, or it is determined whether the frequency deviation is less than a second preset frequency deviation of the power system, wherein the first preset frequency deviation is greater than the second preset frequency deviation; if the frequency deviation is greater than the first preset frequency deviation, the target proportional parameter is the maximum value of the proportional parameter acceptable to the compressed air energy storage system, and the output power is adjusted based on the maximum value and the target integral parameter; if the frequency deviation is less than the second preset frequency deviation, the target integral parameter is the minimum value of the integral parameter acceptable to the compressed air energy storage system, and the output power is adjusted based on the target proportional parameter and the minimum value.

[0008] Optionally, in one embodiment of the present application, the output power of the compressed air energy storage system is adjusted based on the intake mass flow rate and a pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets a preset frequency stability condition, and also includes: determining the frequency in the power system for maintaining frequency stability based on the output power, and controlling the compressed air energy storage system to enter a power transition stage based on the frequency; when the compressed air energy storage system is in the power transition stage, when the intake mass flow rate is less than or equal to an upper limit of the intake mass flow rate, determining whether the frequency meets the preset frequency stability condition; if the frequency meets the preset frequency stability condition, prohibiting the use of the neural network to adjust the output power.

[0009] Optionally, in one embodiment of the present application, before adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and the pre-trained neural network, it also includes: constructing an input layer of a preset neural network based on the intake mass flow rate neurons, frequency deviation neurons and frequency deviation change rate neurons; constructing an output layer of the preset neural network based on proportional parameter neurons and integral parameter neurons; and training the preset neural network using a target loss function to obtain a trained neural network.

[0010] Optionally, in one embodiment of the present application, the calculation formula of the target loss function may be but is not limited to:

[0011]

[0012] Among them, K p With K i is the target proportional parameter and target integral parameter of the neural network output, K′ p and K′ i are the ideal proportional parameter and ideal integral parameter of the output, m is the sample size, and L is the loss function.

[0013] Optionally, in one embodiment of the present application, the calculation formula of the target proportion parameter and the target integral parameter may be, but is not limited to,:

[0014]

[0015] Among them, h j , b j , C jare the basis vector, Gaussian basis function, and center point vector value of the j-th neuron, respectively. ||·|| is the Euclidean norm. w is the weight vector from the hidden layer to the output layer. X is a ternary vector, wherein the ternary vector can be but is not limited to being represented by X=(β, Δf, Δf′), wherein β is the intake mass flow rate of the compressor, Δf is the frequency deviation of the power system, and Δf′ is the rate of change of the frequency deviation of the power system.

[0016] The second aspect of the present application provides a compressed air energy storage frequency modulation device based on a neural network, including: an acquisition module, used to obtain the intake mass flow rate of the compressor group in the compressed air energy storage system, the frequency deviation of the power system and the frequency deviation change rate; a judgment module, used to judge whether the frequency deviation meets the preset frequency modulation start-up condition of the compressed air energy storage system; a frequency modulation module, used to adjust the output power of the compressed air energy storage system based on the intake mass flow rate and a pre-trained neural network when the frequency deviation meets the preset frequency modulation start-up condition, so as to adjust the frequency of the power system based on the adjusted output power until the frequency meets the preset frequency stability condition.

[0017] Optionally, in one embodiment of the present application, the frequency modulation module includes: an input unit, which is used to input the intake mass flow rate, the frequency deviation and the frequency deviation change rate into a pre-trained neural network to obtain a target proportional parameter and a target integral parameter of the compressed air energy storage system, and control the compressed air energy storage system to enter a short-term increase phase based on the target proportional parameter and the target integral parameter; a first judgment unit, which is used to judge whether the frequency deviation is greater than a first preset frequency deviation of the power system, or judge whether the frequency deviation is less than a second preset frequency deviation of the power system when the compressed air energy storage system is in the short-term increase phase, wherein the first preset frequency deviation is greater than the second preset frequency deviation; a first frequency modulation unit, which is used to, when the frequency deviation is greater than the first preset frequency deviation, the target proportional parameter is the maximum value of the proportional parameter that the compressed air energy storage system can accept, and adjust the output power based on the maximum value and the target integral parameter; a second frequency modulation unit, which is used to, when the frequency deviation is less than the second preset frequency deviation, the target integral parameter is the minimum value of the integral parameter that the compressed air energy storage system can accept, and adjust the output power based on the target proportional parameter and the minimum value.

[0018] Optionally, in one embodiment of the present application, the frequency modulation module further includes: a determination unit, used to determine the frequency for maintaining frequency stability in the power system based on the output power, and control the compressed air energy storage system to enter a power transition stage based on the frequency; a second judgment unit, used to determine whether the frequency meets the preset frequency stability condition when the compressed air energy storage system is in the power transition stage and the intake mass flow rate is less than or equal to the upper limit of the intake mass flow rate; a third frequency modulation unit, used to prohibit using the neural network to adjust the output power when the frequency meets the preset frequency stability condition.

[0019] Optionally, in one embodiment of the present application, it also includes: a first construction module, used to construct an input layer of a preset neural network based on intake mass flow rate neurons, frequency deviation neurons and frequency deviation change rate neurons before adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and a pre-trained neural network; a second construction module, used to construct an output layer of the preset neural network based on proportional parameter neurons and integral parameter neurons; a training module, used to train the preset neural network using a target loss function to obtain a trained neural network.

[0020] Optionally, in one embodiment of the present application, the calculation formula of the target loss function may be but is not limited to:

[0021]

[0022] Among them, K p With K i is the target proportional parameter and target integral parameter of the neural network output, K′ p and K′ i are the ideal proportional parameter and ideal integral parameter of the output, m is the sample size, and L is the loss function.

[0023] Optionally, in one embodiment of the present application, the calculation formula of the target proportion parameter and the target integral parameter may be, but is not limited to,:

[0024]

[0025] Among them, h j , b j , C jare the basis vector, Gaussian basis function, and center point vector value of the j-th neuron, respectively. ||·|| is the Euclidean norm. w is the weight vector from the hidden layer to the output layer. X is a ternary vector, wherein the ternary vector can be but is not limited to being represented by X=(β, Δf, Δf′), wherein β is the intake mass flow rate of the compressor, Δf is the frequency deviation of the power system, and Δf′ is the rate of change of the frequency deviation of the power system.

[0026] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the compressed air energy storage frequency modulation method based on a neural network as described in the above embodiment.

[0027] The fourth aspect embodiment of the present application provides a computer-readable storage medium, which stores a computer program, and when the program is executed by a processor, it implements the above-mentioned neural network-based compressed air energy storage frequency modulation method.

[0028] The fifth aspect of the present application provides a computer program product, including a computer program, which, when executed, implements the above-mentioned neural network-based compressed air energy storage frequency modulation method.

[0029] In the embodiment of the present application, when the frequency deviation satisfies the preset frequency modulation start-up conditions of the compressed air energy storage system, the obtained intake mass flow rate of the compressor unit, the frequency deviation of the power system and the frequency deviation change rate are input into the pre-trained neural network to adjust the output power of the compressed air energy storage system until a certain frequency stability condition is met, which gives full play to the intelligent processing ability of the neural network for multi-dimensional parameters, realizes the optimal real-time adaptive setting of the frequency modulation controller parameters in the power system, and maximizes the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply, and improves the accuracy of active power control of the compressed air energy storage system and the frequency stability of the power system. Thus, the problems in the related art that the intelligent processing ability of the neural network for multi-dimensional parameters cannot be fully utilized, the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply cannot be reflected, and the accuracy of active power control of the compressed air energy storage system is low are solved.

[0030] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0032] Figure 1 A block diagram of a system for implementing a neural network-based compressed air energy storage frequency modulation method according to an embodiment of the present application;

[0033] Figure 2 A flowchart of a compressed air energy storage frequency modulation method based on a neural network provided according to an embodiment of the present application;

[0034] Figure 3 A flowchart of the working principle of a compressed air energy storage frequency modulation method based on a neural network provided according to an embodiment of the present application;

[0035] Figure 4 A block diagram of a compressed air energy storage frequency modulation device based on a neural network provided according to an embodiment of the present application;

[0036] Figure 5 It is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0037] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0038] The following describes the compressed air energy storage frequency modulation method based on a neural network in the embodiment of the present application with reference to the accompanying drawings. In view of the problem mentioned in the above background technology that the intelligent processing ability of the neural network for multi-dimensional parameters cannot be fully utilized, the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply cannot be reflected, and the accuracy of the active power control of the compressed air energy storage system is low, the present application provides a compressed air energy storage frequency modulation method based on a neural network, in which, when the frequency deviation meets the preset frequency modulation start-up condition of the compressed air energy storage system, the obtained intake mass flow rate of the compressor unit, the frequency deviation of the power system and the frequency deviation change rate are input into the pre-trained neural network to adjust the output power of the compressed air energy storage system until a certain frequency stability condition is met, the intelligent processing ability of the neural network for multi-dimensional parameters is fully utilized, and the optimal real-time adaptive setting of the frequency modulation controller parameters in the power system is realized, and the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply are reflected to the greatest extent, and the accuracy of the active power control of the compressed air energy storage system and the frequency stability of the power system are improved. As a result, problems in related technologies such as the inability to give full play to the intelligent processing capabilities of neural networks for multi-dimensional parameters, the inability to reflect the advantages of compressed air energy storage systems such as fast response speed, environmentally friendly high-quality transient frequency-modulated power supply, and low accuracy of active power control of compressed air energy storage systems are solved.

[0039] Before introducing the compressed air energy storage frequency modulation method based on neural network proposed in an embodiment of the present application, a system for implementing the compressed air energy storage frequency modulation method based on neural network in an embodiment of the present application is first introduced.

[0040] Specifically, Figure 1 A block diagram of a system for implementing a neural network-based compressed air energy storage frequency modulation method according to an embodiment of the present application.

[0041] like Figure 1 As shown, 101 is a collection device, 102 is a computing device, 103 is a control device, 104 is a frequency detector, 105 is a frequency differential detector, 106 is an air mass flow meter, 107 is an input layer, 108 is a hidden layer, 109 is an output layer, 110 is a proportional link, 111 is an integral link, and 112 is a compressor inlet valve.

[0042] The acquisition device 101 may include, but is not limited to, a frequency detector 104 , a frequency differential detector 105 , and an air mass flow meter 106 .

[0043] Furthermore, in the embodiment of the present application, the frequency detector 104 can detect the real-time frequency of the power system at a sampling frequency of ten times per second, and the collected sample points are smoothed and denoised by a five-point cubic filtering algorithm.

[0044] The frequency differential detector 105 is responsible for calculating the frequency change rate by finite difference approximation based on the smoothed data of the frequency detector.

[0045] The air mass flow meter 106 detects the air mass flow rate at the compressor inlet valve at the same sampling frequency of ten times per second.

[0046] For example, the power system of the embodiment of the present application can select a reference frequency of 50 Hz. If the frequency deviation of the power system meets certain frequency modulation start conditions of the compressed air energy storage system, such as the frequency deviation is greater than the maximum value Δf of the frequency modulation dead zone of the compressed air energy storage unit gate And the duration is greater than the threshold time t gate , then all the information of the three is sent to the computing device 102 for further processing.

[0047] The computing device 102 adopts a neural network model, that is, includes a pre-trained neural network, which may include, but is not limited to, an input layer 107, a hidden layer 108 and an output layer 109 of the neural network.

[0048] Furthermore, in the embodiment of the present application, the input layer 107 includes three neurons, which respectively represent the input compressor intake mass flow rate neuron, the power system frequency deviation neuron and the power system frequency deviation change rate neuron.

[0049] The output layer 109 includes two neurons, which represent the proportional parameter neuron and the integral parameter neuron of the output respectively.

[0050] The hidden layer 108 is a superposition of several layers of intermediate neurons, wherein in the embodiment of the present application, the number of neurons, the connection mode and the weight parameters are unknown. However, the parameters can be made closer to the actual power system through training. The training method adopts the loss function method, that is, the historical operation data of the power station is used as the training set, and the intake mass flow rate of the β compressor, the frequency deviation of the Δf power system and the frequency deviation change rate of the Δf′ power system are input each time, and then the output K is compared. p Scale parameter and K i The difference between the integral parameter and the actual value is used to define the loss function as the Euclidean distance between the output value and the ideal value.

[0051] It can be understood that the training goal of the neural network in the embodiment of the present application is to minimize the value of the loss function, and the solution algorithm can adopt the radial basis function method.

[0052] The control device may include, but is not limited to, a proportional link 110 and an integral link 111, which are directly connected to the compressor inlet valve 112, and the valve opening is controlled to control the compressor intake mass flow rate, thereby achieving the purpose of setting power and adjusting the power system frequency.

[0053] Furthermore, in the embodiment of the present application, the main function of the proportional link 110 is to amplify the frequency deviation, generate a control signal proportional to the frequency deviation, and apply it to the controlled object to reduce the deviation, but at the same time it will inevitably introduce lag and overshoot.

[0054] The main function of the integral link 111 is to accumulate the frequency deviation history and generate a control signal to eliminate the frequency deviation, thereby achieving error-free regulation of the power system.

[0055] It should be noted that in order to achieve rapid frequency modulation and reduce the impact on the power system as much as possible, the control process of the embodiment of the present application can be divided into two stages:

[0056] The first stage is a short-term additional power generation stage, during which the compressor releases kinetic energy for power support. The starting time of this stage is when the compressed air energy storage unit participates in frequency modulation, and the ending time is when it enters the power tracking stage;

[0057] Among them, the embodiment of the present application follows the following principles in the first stage of frequency regulation: when the frequency deviation is large, the largest proportional coefficient is adopted to accelerate the timely power response of the frequency-regulated power system; when the frequency deviation is small, the power system has tended to a stable state, and the smallest proportional coefficient is adopted to maintain the frequency stability of the power system.

[0058] The second stage is the power transition stage, which maintains the frequency of the power system at a constant value while ensuring that the intake mass flow rate does not exceed the limit. When the frequency of the power system meets certain frequency stability conditions, such as the frequency is stable at Δf gate Within and the duration is greater than t gate When , it can be considered that the frequency modulation process is completed and the compressed air energy storage unit resumes maximum power point tracking operation.

[0059] Specifically, Figure 2 The present invention is a flowchart of a compressed air energy storage frequency modulation method based on a neural network provided according to an embodiment of the present application.

[0060] like Figure 2 As shown, the compressed air energy storage frequency modulation method based on neural network includes the following steps:

[0061] In step S201, the intake mass flow rate of the compressor group in the compressed air energy storage system, the frequency deviation of the power system and the frequency deviation change rate are obtained.

[0062] As a possible implementation method, the embodiment of the present application can obtain the intake mass flow rate of the compressor group in the compressed air energy storage system, the frequency deviation of the power system and the frequency deviation change rate.

[0063] For example, Figure 1 As shown, the embodiment of the present application can utilize a frequency detector to detect the real-time frequency of the power system in real time, and then a frequency differential detector calculates the frequency deviation and the frequency deviation change rate, while an air mass flow meter detects the intake mass flow rate at the compressor inlet valve.

[0064] In step S202, it is determined whether the frequency deviation meets the preset frequency modulation start-up condition of the compressed air energy storage system.

[0065] In some embodiments, the embodiments of the present application can determine whether the frequency deviation meets the preset frequency regulation start-up conditions of the compressed air energy storage system, thereby confirming whether the compressed air energy storage system participates in the frequency regulation of the power system.

[0066] Among them, certain frequency modulation start-up conditions can be understood as whether the frequency deviation of the power system is greater than the maximum value of the frequency modulation dead zone of the compressed air energy storage unit and whether the duration is greater than the threshold time. It can be specifically set by technical personnel in this field according to actual conditions, and this application does not make specific restrictions.

[0067] For example, the power system of the present application embodiment can select a reference frequency of 50 Hz, and then determine whether the frequency deviation of the power system is greater than Δf gate And whether the duration is greater than t gate .

[0068] Optionally, in one embodiment of the present application, before adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and the pre-trained neural network, it also includes: constructing an input layer of a preset neural network based on the intake mass flow rate neurons, frequency deviation neurons, and frequency deviation change rate neurons; constructing an output layer of a preset neural network based on proportional parameter neurons and integral parameter neurons; and training the preset neural network using a target loss function to obtain a trained neural network. The calculation formula of the target loss function may be, but is not limited to,:

[0069]

[0070] Among them, K p With K i is the target proportional parameter and target integral parameter of the neural network output, K′ p and K′ i are the ideal proportional parameter and ideal integral parameter of the output, m is the sample size, and L is the loss function.

[0071] The calculation formulas of the target proportional parameter and the target integral parameter may be, but are not limited to,:

[0072]

[0073] Among them, hj 、b j , C j are the basis vector, Gaussian basis function, and center point vector value of the j-th neuron, respectively. ||·|| is the Euclidean norm. w is the weight vector from the hidden layer to the output layer. X is a ternary vector, wherein the ternary vector can be but is not limited to being represented by X=(β, Δf, Δf′), wherein β is the intake mass flow rate of the compressor, Δf is the frequency deviation of the power system, and Δf′ is the rate of change of the frequency deviation of the power system.

[0074] It should be noted that, before using a pre-trained neural network to adjust the output power of the compressed air energy storage system, the embodiment of the present application may construct a neural network and perform training.

[0075] For example, the embodiment of the present application can construct an input layer based on the intake mass flow rate neuron, the frequency deviation neuron and the frequency deviation change rate neuron, and construct an output layer based on the proportional parameter neuron and the integral parameter neuron, and then use the target loss function to train the preset neural network to obtain a trained neural network. The calculation formula of the target loss function can be but is not limited to:

[0076]

[0077] Among them, K p With K i is the target proportional parameter and target integral parameter of the neural network output, K′ p and K′ i are the ideal proportional parameter and ideal integral parameter of the output, m is the sample size, and L is the loss function.

[0078] Further, in the embodiment of the present application, the solution algorithm may adopt the radial basis function method, and the solution process may be described as follows:

[0079]

[0080] Among them, x is located in the input layer of the neural network and is represented by a three-element vector X = (β, Δf, Δf′), where β is the intake mass flow rate of the compressor, Δf is the frequency deviation of the power system, Δf′ is the frequency deviation change rate of the power system, and h j 、b j , C j are the basis vector, Gaussian basis function, and center point vector value of the j-th neuron, respectively. ||·|| is the Euclidean norm, and w is the weight vector from the hidden layer to the output layer.

[0081] Therefore, the embodiment of the present application can use the target loss function method and the gradient descent algorithm to train the neural network.

[0082] In step S203, if the frequency deviation meets the preset frequency regulation start-up condition, the output power of the compressed air energy storage system is adjusted based on the intake mass flow rate and the pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets the preset frequency stability condition.

[0083] As a possible implementation method, in the embodiment of the present application, when the frequency deviation meets certain frequency modulation start-up conditions, the output power of the compressed air energy storage system can be adjusted based on the intake mass flow rate and a pre-trained neural network, thereby obtaining a frequency that meets certain frequency stability conditions.

[0084] Among them, a certain frequency stability condition can be understood as the frequency being stable within the maximum value of the frequency regulation dead zone of the compressed air energy storage unit and lasting longer than the threshold time. It can be specifically set by technicians in this field according to actual conditions, and this application does not impose any specific restrictions.

[0085] It can be understood that the embodiment of the present application mainly adjusts the output power by adjusting the mass flow rate of the expander. The expression of the outlet air temperature and pressure of the i-th stage expander can be, but is not limited to, expressed as:

[0086]

[0087]

[0088] in, represents the inlet air temperature of the expander, p e ,i in represents the inlet air pressure of the i-th stage expander, represents the rated expansion ratio of the i-th stage expander, represents the rated isentropic efficiency of the i-th stage expander, and k is the air adiabatic index.

[0089] Further, in the embodiment of the present application, the output power of the i-th stage expander can be expressed as, but not limited to:

[0090]

[0091] Then the total output power of each stage of expanders is obtained, which can be expressed as but not limited to:

[0092]

[0093] in, is the mass flow rate of the expander, is the specific heat capacity of air at constant pressure, N e is the number of expander stages.

[0094] It can be seen from this that the embodiments of the present application can appropriately increase the mass flow rate of the expander, thereby improving the output power of the compressed air energy storage, thereby achieving the purpose of stabilizing the frequency in the power system.

[0095] For example, in the embodiment of the present application, the frequency deviation of the power system is greater than Δf gate And the duration is greater than t gate In the case of , it can be determined that the compressed air energy storage system participates in the frequency regulation of the power system, and then the power system can adjust the proportional parameter and the integral parameter according to the load power command through the intake mass flow rate and the pre-trained neural network to control the output power of the compressed air energy storage system until the frequency stabilizes at Δf gate Within and the duration is greater than t gate , the frequency of the power system is stable.

[0096] For example, Figure 1 As shown, in the embodiment of the present application, the frequency deviation of the power system is greater than Δf gate And the duration is greater than t gate In this case, the intake mass flow rate of the compressor unit, the frequency deviation of the power system and the frequency deviation change rate are input into a pre-trained neural network, and the neural network outputs the target proportional parameter and the target integral parameter to the control device, and the control device is connected to the compressor inlet valve, and the compressor inlet valve opening is adjusted through the proportional link and the integral link, thereby adjusting the output power.

[0097] Optionally, in one embodiment of the present application, the output power of the compressed air energy storage system is adjusted based on the intake mass flow rate and a pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets a preset frequency stability condition, including: inputting the intake mass flow rate, the frequency deviation and the frequency deviation change rate into the pre-trained neural network to obtain a target proportional parameter and a target integral parameter of the compressed air energy storage system, and controlling the compressed air energy storage system to enter a short-term increase phase based on the target proportional parameter and the target integral parameter; when the compressed air energy storage system is in the short-term increase phase, determining whether the frequency deviation is greater than a first preset frequency deviation of the power system, or determining whether the frequency deviation is less than a second preset frequency deviation of the power system, wherein the first preset frequency deviation is greater than the second preset frequency deviation; if the frequency deviation is greater than the first preset frequency deviation, the target proportional parameter is the maximum value of the proportional parameter that the compressed air energy storage system can accept, and the output power is adjusted based on the maximum value and the target integral parameter; if the frequency deviation is less than the second preset frequency deviation, the target integral parameter is the minimum value of the integral parameter that the compressed air energy storage system can accept, and the output power is adjusted based on the target proportional parameter and the minimum value.

[0098] It will be understood by those skilled in the art that, in the process of adjusting the output power of the compressed air energy storage system using a pre-trained neural network in the embodiment of the present application, in order to achieve rapid frequency modulation as much as possible and reduce the impact of the compressed air energy storage system on the power system, it is also possible to determine whether the frequency deviation is greater than a preset frequency deviation of the power system.

[0099] Exemplarily, the control process is divided into two stages in the embodiment of the present application, wherein Δf 1 is the first certain frequency deviation of the power system, Δf 2 The second certain frequency deviation of the power system, and the first certain frequency deviation is greater than the second certain frequency deviation, can be specifically set by technicians in this field according to actual conditions, and this application does not make specific restrictions. It should be noted that the specific content of the first stage of the embodiment of this application is:

[0100] In the embodiment of the present application, the first stage is a short-term additional power stage, in which the compressor releases kinetic energy for power support. The starting time is the time when the compressed air energy storage unit participates in frequency modulation, and the end time is the time when the power tracking stage is entered. The frequency modulation in this stage follows the following principles:

[0101] Principle 1: Determine whether the frequency deviation is greater than the first preset frequency deviation of the power system. When the frequency deviation is greater than the first preset frequency deviation, the target proportional parameter is the maximum value of the proportional parameter that the compressed air energy storage system can accept, that is:

[0102] When Δf>Δf 1 When K p =K pmax ,

[0103] In other words, when the frequency deviation is large, the maximum proportional parameter K is used. p To accelerate the timely power response of the frequency-regulated power system.

[0104] Principle 2: Determine whether the frequency deviation is less than the second preset frequency deviation of the power system. When the frequency deviation is less than the second preset frequency deviation, the target integral parameter is the minimum value of the integral parameter that the compressed air energy storage system can accept, that is:

[0105] When Δf<Δf 2 When K i =K imin ,

[0106] In other words, in the embodiment of the present application, when the frequency deviation is small, the power system has tended to be stable, and the minimum integral parameter K can be used. i To maintain frequency stability in the power system.

[0107] Optionally, in one embodiment of the present application, the output power of the compressed air energy storage system is adjusted based on the intake mass flow rate and a pre-trained neural network, so as to adjust the frequency of the power system based on the adjusted output power until the frequency meets a preset frequency stability condition, and further includes: determining the frequency in the power system for maintaining frequency stability based on the output power, and controlling the compressed air energy storage system to enter a power transition stage based on the frequency; when the compressed air energy storage system is in the power transition stage, when the intake mass flow rate is less than or equal to an upper limit of the intake mass flow rate, determining whether the frequency meets a preset frequency stability condition; if the frequency meets the preset frequency stability condition, it is prohibited to adjust the output power using the neural network.

[0108] Through the above analysis, it can be known that the embodiment of the present application can determine in the first stage that the frequency of the power system has been initially stabilized, and then enter the second stage, that is, when the intake mass flow rate is less than or equal to the upper limit of the intake mass flow rate, it is judged whether the frequency meets certain frequency stability conditions, and when the frequency meets the preset frequency stability conditions, it is prohibited to use the neural network to adjust the output power.

[0109] Exemplarily, the specific contents of the second stage of the embodiment of the present application are:

[0110] In the embodiment of the present application, the second stage is the power transition stage. Under the condition that the intake mass flow rate is less than or equal to the upper limit of the intake mass flow rate, it is determined whether the frequency meets a certain frequency stability condition, that is, the frequency of the power system is maintained at a constant value. For example, when the frequency of the power system is stable at Δf gate Within and the duration is greater than t gate When , it can be considered that the frequency modulation process is over, the compressed air energy storage unit resumes maximum power point tracking operation, and it is prohibited to use the neural network to adjust the output power.

[0111] The embodiment of the present application adopts a neural network architecture with three-dimensional system state quantity input, which overcomes the shortcomings of the previous frequency regulation controller parameters being fixed or only considering one input system state variable (frequency deviation) in parameter setting. It can give full play to the intelligent processing ability of the neural network for multi-dimensional parameters, realize the optimal real-time adaptive setting of the parameters of the power system frequency regulation controller, and give full play to the advantages of the compressed air energy storage system's fast response speed, environmentally friendly high-quality transient frequency regulation power supply, while improving the accuracy of the compressed air energy storage system's active power control and the frequency stability of the power system.

[0112] Combine the following Figure 1 and Figure 3 , the working principle of the compressed air energy storage frequency modulation method based on neural network proposed in the embodiment of the present application is introduced in detail with a specific embodiment.

[0113] in, Figure 3The present invention is a flowchart of the working principle of a compressed air energy storage frequency modulation method based on a neural network provided according to an embodiment of the present application.

[0114] Step S301: The collection device collects the intake air mass flow rate of the compressor unit, the frequency deviation of the power system and the frequency deviation change rate, and then proceeds to step S302.

[0115] Step S302: When the frequency deviation of the power system is greater than the maximum value of the frequency regulation dead zone of the compressed air energy storage unit and the duration is greater than the threshold time, it is determined that the compressed air energy storage system participates in the frequency regulation of the power system, and step S303 is continued.

[0116] Step S303: input the collected intake mass flow rate, frequency deviation and frequency deviation change rate into a pre-trained neural network.

[0117] Step S304: Obtain target proportional parameters and target integral parameters of the compressed air energy storage system.

[0118] Step S305: The control device controls the opening of the compressor air inlet valve to adjust the output power of the compressed air energy storage system.

[0119] Step S306: Enter the short-term additional issuance phase.

[0120] Step S307: determine whether the frequency duration of the power system is greater than the threshold time, if it is greater than the threshold time, execute step S308, otherwise, execute step S306.

[0121] Step S308: Enter the power transition phase.

[0122] According to the compressed air energy storage frequency modulation method based on neural network proposed in the embodiment of the present application, when the frequency deviation meets the preset frequency modulation start-up condition of the compressed air energy storage system, the obtained intake mass flow rate of the compressor unit, the frequency deviation of the power system and the frequency deviation change rate are input into the pre-trained neural network to adjust the output power of the compressed air energy storage system until a certain frequency stability condition is met, which gives full play to the intelligent processing ability of the neural network for multi-dimensional parameters, realizes the optimal real-time adaptive setting of the frequency modulation controller parameters in the power system, and maximizes the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply, and improves the accuracy of active power control of the compressed air energy storage system and the frequency stability of the power system. Thus, the problems in the related art that the intelligent processing ability of the neural network for multi-dimensional parameters cannot be fully utilized, the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply cannot be reflected, and the accuracy of active power control of the compressed air energy storage system is low, etc.

[0123] Next, a compressed air energy storage and frequency modulation device based on a neural network proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0124] Figure 4 A block diagram of a compressed air energy storage frequency modulation device based on a neural network provided according to an embodiment of the present application.

[0125] like Figure 4 As shown, the compressed air energy storage frequency modulation device 10 based on neural network includes: an acquisition module 100, a judgment module 200 and a frequency modulation module 300.

[0126] Among them, the acquisition module 100 is used to obtain the intake mass flow rate of the compressor group in the compressed air energy storage system, the frequency deviation of the power system and the frequency deviation change rate.

[0127] The judgment module 200 is used to judge whether the frequency deviation meets the preset frequency modulation start condition of the compressed air energy storage system.

[0128] The frequency modulation module 300 is used to adjust the output power of the compressed air energy storage system based on the intake mass flow rate and a pre-trained neural network when the frequency deviation meets the preset frequency modulation start condition, so as to adjust the frequency of the power system based on the adjusted output power until the frequency meets the preset frequency stability condition.

[0129] Optionally, in an embodiment of the present application, the frequency modulation module 300 includes: an input unit, a first judgment unit, a first frequency modulation unit and a second frequency modulation unit.

[0130] Among them, the input unit is used to input the intake mass flow rate, frequency deviation and frequency deviation change rate into a pre-trained neural network to obtain the target proportional parameter and target integral parameter of the compressed air energy storage system, and control the compressed air energy storage system to enter the short-term increase phase based on the target proportional parameter and target integral parameter.

[0131] The first judgment unit is used to judge whether the frequency deviation is greater than a first preset frequency deviation of the power system, or to judge whether the frequency deviation is less than a second preset frequency deviation of the power system when the compressed air energy storage system is in a short-term increase in power generation stage, wherein the first preset frequency deviation is greater than the second preset frequency deviation.

[0132] The first frequency modulation unit is used to adjust the output power based on the maximum value and the target integral parameter when the frequency deviation is greater than the first preset frequency deviation and the target proportional parameter is the maximum value of the proportional parameter acceptable to the compressed air energy storage system.

[0133] The second frequency modulation unit is used to adjust the output power based on the target proportional parameter and the minimum value when the frequency deviation is less than the second preset frequency deviation, the target integral parameter is the minimum value of the integral parameter acceptable to the compressed air energy storage system.

[0134] Optionally, in an embodiment of the present application, the frequency modulation module 300 further includes: a determination unit, a second judgment unit and a third frequency modulation unit.

[0135] Among them, the determination unit is used to determine the frequency for maintaining frequency stability in the power system based on the output power, and control the compressed air energy storage system to enter a power transition stage based on the frequency.

[0136] The second judgment unit is used to judge whether the frequency meets the preset frequency stability condition when the compressed air energy storage system is in the power transition stage and the intake mass flow rate is less than or equal to the intake mass flow rate upper limit.

[0137] The third frequency modulation unit is used to prohibit the use of a neural network to adjust the output power when the frequency meets a preset frequency stability condition.

[0138] Optionally, in one embodiment of the present application, it also includes: a first construction module, a second construction module and a training module.

[0139] Among them, the first construction module is used to construct an input layer of a preset neural network based on intake mass flow rate neurons, frequency deviation neurons and frequency deviation change rate neurons before adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and a pre-trained neural network.

[0140] The second construction module is used to construct an output layer of a preset neural network based on proportional parameter neurons and integral parameter neurons.

[0141] The training module is used to train the preset neural network using the target loss function to obtain a trained neural network.

[0142] Optionally, in one embodiment of the present application, the calculation formula of the target loss function may be, but is not limited to,:

[0143]

[0144] Among them, K p With K i is the target proportional parameter and target integral parameter of the neural network output, K′ p and K′ i are the ideal proportional parameter and ideal integral parameter of the output, m is the sample size, and L is the loss function.

[0145] Optionally, in one embodiment of the present application, the calculation formulas of the target proportion parameter and the target integral parameter may be, but are not limited to,:

[0146]

[0147] Among them, h j , b j , C j are the basis vector, Gaussian basis function, and center point vector value of the j-th neuron, respectively. ||·|| is the Euclidean norm. w is the weight vector from the hidden layer to the output layer. X is a ternary vector, wherein the ternary vector can be but is not limited to being represented by X=(β, Δf, Δf′), wherein β is the intake mass flow rate of the compressor, Δf is the frequency deviation of the power system, and Δf′ is the rate of change of the frequency deviation of the power system.

[0148] It should be noted that the aforementioned explanation of the embodiment of the compressed air energy storage frequency modulation method based on a neural network is also applicable to the compressed air energy storage frequency modulation device based on a neural network in this embodiment, and will not be repeated here.

[0149] According to the compressed air energy storage frequency modulation device based on neural network proposed in the embodiment of the present application, when the frequency deviation meets the preset frequency modulation start-up condition of the compressed air energy storage system, the obtained intake mass flow rate of the compressor group, the frequency deviation of the power system and the frequency deviation change rate are input into the pre-trained neural network to adjust the output power of the compressed air energy storage system until a certain frequency stability condition is met, which gives full play to the intelligent processing ability of the neural network for multi-dimensional parameters, realizes the optimal real-time adaptive setting of the frequency modulation controller parameters in the power system, and maximizes the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply, and improves the accuracy of active power control of the compressed air energy storage system and the frequency stability of the power system. Thus, the problems in the related art that the intelligent processing ability of the neural network for multi-dimensional parameters cannot be fully utilized, the advantages of the compressed air energy storage system such as fast response speed and environmentally friendly high-quality transient frequency modulation power supply cannot be reflected, and the accuracy of active power control of the compressed air energy storage system is low, etc.

[0150] Figure 5 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. The electronic device may include:

[0151] A memory 501 , a processor 502 , and a computer program stored in the memory 501 and executable on the processor 502 .

[0152] When the processor 502 executes the program, the compressed air energy storage frequency modulation method based on neural network provided in the above embodiment is implemented.

[0153] Furthermore, the electronic device further comprises:

[0154] The communication interface 503 is used for communication between the memory 501 and the processor 502 .

[0155] The memory 501 is used to store computer programs that can be executed on the processor 502 .

[0156] The memory 501 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0157] If the memory 501, the processor 502 and the communication interface 503 are implemented independently, the communication interface 503, the memory 501 and the processor 502 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0158] Optionally, in a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 can communicate with each other through an internal interface.

[0159] The processor 502 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0160] An embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned neural network-based compressed air energy storage frequency modulation method is implemented.

[0161] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed, implements the above-mentioned neural network-based compressed air energy storage frequency modulation method.

[0162] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0163] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0164] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0165] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0166] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one or a combination of multiple of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0167] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0168] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0169] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A compressed air energy storage frequency modulation method based on neural network, characterized in that: The following steps are involved: Obtaining the air intake mass flow rate of the compressor unit in the compressed air energy storage system, the frequency deviation of the power system, and the frequency deviation change rate; Determining whether the frequency deviation meets a preset frequency modulation start condition of the compressed air energy storage system; If the frequency deviation meets the preset frequency regulation start-up condition, the output power of the compressed air energy storage system is adjusted based on the intake mass flow rate and the pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets the preset frequency stability condition.

2. The method according to claim 1, characterized in that The adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and the pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets a preset frequency stability condition includes: Inputting the intake mass flow rate, the frequency deviation and the frequency deviation change rate into a pre-trained neural network to obtain a target proportional parameter and a target integral parameter of the compressed air energy storage system, and controlling the compressed air energy storage system to enter a short-term increase phase based on the target proportional parameter and the target integral parameter; When the compressed air energy storage system is in the short-term additional generation stage, judging whether the frequency deviation is greater than a first preset frequency deviation of the power system, or judging whether the frequency deviation is less than a second preset frequency deviation of the power system, wherein the first preset frequency deviation is greater than the second preset frequency deviation; If the frequency deviation is greater than the first preset frequency deviation, the target proportional parameter is the maximum value of the proportional parameter acceptable to the compressed air energy storage system, and the output power is adjusted based on the maximum value and the target integral parameter; If the frequency deviation is less than the second preset frequency deviation, the target integral parameter is the minimum value of the integral parameter acceptable to the compressed air energy storage system, and the output power is adjusted based on the target proportional parameter and the minimum value.

3. The method according to claim 2, characterized in that The adjusting the output power of the compressed air energy storage system based on the intake mass flow rate and the pre-trained neural network to adjust the frequency of the power system based on the adjusted output power until the frequency meets a preset frequency stability condition, further comprising: Determine a frequency in the power system to maintain frequency stability based on the output power, and control the compressed air energy storage system to enter a power transition phase based on the frequency; When the compressed air energy storage system is in the power transition stage, when the intake air mass flow rate is less than or equal to the intake air mass flow rate upper limit, determining whether the frequency meets the preset frequency stability condition; If the frequency satisfies the preset frequency stability condition, it is prohibited to adjust the output power using the neural network.

4. The method according to claim 1, characterized in that: Before adjusting the output power of the compressed air energy storage system based on the intake air mass flow rate and the pre-trained neural network, the method further includes: An input layer of a preset neural network is constructed based on intake mass flow rate neurons, frequency deviation neurons, and frequency deviation change rate neurons; Constructing the output layer of the preset neural network based on the proportional parameter neurons and the integral parameter neurons; The preset neural network is trained using the target loss function to obtain a trained neural network.

5. The method according to claim 4, characterized in that The calculation formula of the objective loss function is: Among them, K p With K i are the target proportional parameters and target integral parameters of the neural network output, K p ′ With K i ′ are the ideal proportional parameter and ideal integral parameter of the output, m is the sample size, and L is the loss function.

6. The method according to claim 2, characterized in that The calculation formulas of the target proportional parameter and the target integral parameter are: Among them, h j 、b j , C j are the basis vector, Gaussian basis function, and center point vector value of the jth neuron, respectively. ∥·∥ is the Euclidean norm. w is the weight vector from the hidden layer to the output layer. X is a ternary vector.

7. A compressed air energy storage frequency modulation device based on a neural network, characterized in that: include: An acquisition module is used to obtain the intake mass flow rate of the compressor unit in the compressed air energy storage system, the frequency deviation of the power system and the frequency deviation change rate; A judgment module, used to judge whether the frequency deviation meets the preset frequency modulation start condition of the compressed air energy storage system; A frequency modulation module is used to adjust the output power of the compressed air energy storage system based on the intake mass flow rate and a pre-trained neural network when the frequency deviation meets the preset frequency modulation start condition, so as to adjust the frequency of the power system based on the adjusted output power until the frequency meets the preset frequency stability condition.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the neural network-based compressed air energy storage frequency modulation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the compressed air energy storage frequency modulation method based on a neural network as described in any one of claims 1 to 6.

10. A computer program product, characterized in that It includes a computer program, which, when executed, is used to implement the compressed air energy storage frequency modulation method based on neural network as described in any one of claims 1-6.

Images