Dynamic simulation method for multi-stage compressor of compressed air energy storage system

Abstract

The application relates to the technical field of new energy, in particular to a dynamic simulation method for a multistage compressor of a compressed air energy storage system, which comprises the following steps: constructing a static thermodynamic model framework of the compressed air energy storage system compression side according to a static thermodynamic model of the multistage compressor and a static thermodynamic model of an interstage heat exchanger; embedding a pre-trained LSTM neural network into the static thermodynamic model framework of the compressed air energy storage system compression side, and carrying out real-time variable working condition nonlinear characteristic simulation in a system dynamic process after embedding is completed, wherein the pre-trained LSTM neural network is obtained by training a preset LSTM neural network through a compressor set characteristic data set. Therefore, the problem that the multistage compressor of the compressed air energy storage system is difficult to accurately simulate and optimize dynamic performance under a variable working condition in the background technology is solved, and the performance prediction accuracy of the compressed air energy storage system under a variable working condition is improved through the dynamic simulation method.

Description

Technical Field

[0001] This application relates to the field of new energy technology, and in particular to a dynamic simulation method for a multi-stage compressor in a compressed air energy storage system. Background Technology

[0002] With the introduction of the "dual carbon" target, the large-scale grid connection of renewable energy sources, represented by wind and solar power, has brought significant volatility and uncertainty, posing new challenges to the stable and efficient operation of the new power system. Energy storage technology can store excess energy when renewable energy is abundant and release the stored energy to supply loads during peak demand periods, effectively mitigating the fluctuations in renewable energy levels and serving as an effective way to improve the power supply quality of the new power system. In recent years, compressed air energy storage technology, as one of the commercially viable large-scale energy storage technologies, is not limited by geographical conditions and is green, safe, and efficient, making it a hot topic of common concern in industry and academia. The construction of related demonstration projects and commercial power plants is also progressing steadily.

[0003] Currently, compressed air energy storage power stations are mainly used on the power supply side for renewable energy consumption and volatility mitigation, aiming to improve the external characteristics of renewable energy power plants such as wind and solar power and reduce their impact on the power grid. On the grid side, in addition to peak shaving, they also need to provide ancillary services such as frequency regulation and phase regulation. In these application scenarios, because the compressor side of the compressed air energy storage system needs to track fluctuating power output commands, the operating conditions of multi-stage compressor units often deviate from their rated operating conditions. When the compressor operates outside its rated conditions, its pressure ratio and isentropic efficiency will change with mass flow rate and speed. Furthermore, the coupling relationship between pipeline network characteristics and different compressors further exacerbates the variable operating condition characteristics. The variable operating condition characteristics caused by the coupling of multiple physics fields and multiple components are difficult to accurately characterize, making it difficult to optimize the dynamic performance of the compressor side of the compressed air energy storage system, which urgently needs to be addressed. Summary of the Invention

[0004] This application provides a dynamic simulation method for multi-stage compressors in compressed air energy storage systems to solve the problem in the background art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions. The dynamic simulation method improves the accuracy of performance prediction for compressed air energy storage systems under varying operating conditions.

[0005] The first aspect of this application provides a dynamic simulation method for a multi-stage compressor in a compressed air energy storage system, comprising the following steps:

[0006] Construct static thermodynamic models of multi-stage compressors and interstage heat exchangers;

[0007] Based on the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the inter-stage heat exchanger, a static thermodynamic model framework for the compression side of the compressed air energy storage system is constructed.

[0008] A pre-trained LSTM (Long Short-Term Memory) neural network is embedded into the static thermodynamic model framework of the compressed air energy storage system on the compression side. After the embedding is completed, real-time simulation of the nonlinear characteristics of the system under varying operating conditions is performed. The pre-trained LSTM neural network is obtained by training the preset LSTM neural network with the compressor unit characteristic dataset.

[0009] According to one embodiment of this application, after performing real-time variable operating condition nonlinear characteristic simulation, the method further includes:

[0010] Acquire multiple sets of time-series data on mass flow rate, rotational speed, and pressure; set initial time constant parameter estimates and corresponding error covariance matrices; and obtain the predicted data for the current time step based on the system model and the time constant parameter estimates from the previous time step.

[0011] Based on the time-series acquired data and the predicted data, calculate the Kalman gain, and update the estimated value of the time constant parameter and the error covariance matrix based on the Kalman gain;

[0012] If the updated time constant parameter estimate does not meet the preset convergence condition, the step of acquiring multiple sets of time-series data of mass flow rate, rotational speed and pressure is repeated until the updated time constant parameter estimate meets the preset convergence condition. An inertial element is constructed based on the converged time constant parameter estimate, and the inertial element is embedded into the static thermodynamic model framework of the compressed air energy storage system on the compression side for dynamic process simulation fitting.

[0013] According to one embodiment of this application, before embedding the pre-trained LSTM neural network into the compression-side static thermodynamic model framework of the compressed air energy storage system, the method further includes:

[0014] The pressure ratio, inlet and outlet temperatures, mass flow rate and speed of the multi-stage compressor under various operating conditions are collected, and a compressor unit characteristic dataset is constructed based on the pressure ratio, inlet and outlet temperatures, mass flow rate and speed of the multi-stage compressor under various operating conditions.

[0015] The compressor unit characteristic dataset is preprocessed to obtain a preprocessed compressor unit characteristic dataset;

[0016] The preprocessed compressor unit characteristic dataset is divided into a training set and a validation set according to a preset partitioning ratio;

[0017] The preset LSTM neural network is trained using the training set, and after training, the preset LSTM neural network is validated using the validation set to obtain validation results. If the validation results do not meet the preset validation conditions, the network structure and / or hyperparameters of the preset LSTM neural network are optimized and adjusted until the validation results of the preset LSTM neural network meet the preset validation conditions, thus obtaining the pre-trained LSTM neural network.

[0018] According to one embodiment of this application, the construction of the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the interstage heat exchanger includes:

[0019] Calculate the pressure ratio, outlet temperature, and isentropic efficiency of the multi-stage compressor, and call the first preset property library to construct a static thermodynamic model of the multi-stage compressor;

[0020] Based on the compressor interstage heat exchanger equations and by calling the second preset property library, a static thermodynamic model of the interstage heat exchanger is constructed.

[0021] According to one embodiment of this application, the equation for the compressor interstage heat exchanger is:

[0022]

[0023] in, This refers to the outlet air temperature of the heat exchanger. Let ΔT be the inlet air temperature of the heat exchanger, and ΔT be the temperature change. The outlet heat exchange medium temperature. The inlet heat exchange medium temperature. τ is the derivative of the temperature change. r ε is the heat transfer inertia time constant. r The coefficient of performance (COP) of the heat exchanger.

[0024] According to the dynamic simulation method for multi-stage compressors in compressed air energy storage systems according to embodiments of this application, a static thermodynamic model framework for the compression side of the compressed air energy storage system is constructed based on the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the interstage heat exchanger. A pre-trained LSTM neural network is embedded into the static thermodynamic model framework of the compression side of the compressed air energy storage system, and real-time simulation of the nonlinear characteristics under varying operating conditions is performed after embedding. The pre-trained LSTM neural network is obtained by training a preset LSTM neural network using a compressor unit characteristic dataset. This solves the problem in the prior art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions, and improves the accuracy of performance prediction for compressed air energy storage systems under varying operating conditions through dynamic simulation.

[0025] A second aspect of this application provides a dynamic simulation device for a multi-stage compressor of a compressed air energy storage system, comprising:

[0026] The first building module is used to build a static thermodynamic model of a multi-stage compressor and a static thermodynamic model of an inter-stage heat exchanger.

[0027] The second construction module is used to construct a static thermodynamic model framework for the compression side of the compressed air energy storage system based on the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the inter-stage heat exchanger.

[0028] The simulation module is used to embed a pre-trained LSTM neural network into the static thermodynamic model framework of the compressed air energy storage system on the compression side, and to perform real-time simulation of the nonlinear characteristics of the system under varying operating conditions during the dynamic process after embedding. The pre-trained LSTM neural network is obtained by training the preset LSTM neural network with the compressor unit characteristic dataset.

[0029] According to one embodiment of this application, after performing real-time variable operating condition nonlinear characteristic simulation, the simulation module is further configured to:

[0030] Acquire multiple sets of time-series data on mass flow rate, rotational speed, and pressure; set initial time constant parameter estimates and corresponding error covariance matrices; and obtain the predicted data for the current time step based on the system model and the time constant parameter estimates from the previous time step.

[0031] Based on the time-series acquired data and the predicted data, calculate the Kalman gain, and update the estimated value of the time constant parameter and the error covariance matrix based on the Kalman gain;

[0032] If the updated time constant parameter estimate does not meet the preset convergence condition, the step of acquiring multiple sets of time-series data of mass flow rate, rotational speed and pressure is repeated until the updated time constant parameter estimate meets the preset convergence condition. An inertial element is constructed based on the converged time constant parameter estimate, and the inertial element is embedded into the static thermodynamic model framework of the compressed air energy storage system on the compression side for dynamic process simulation fitting.

[0033] According to one embodiment of this application, before embedding the pre-trained LSTM neural network into the compression-side static thermodynamic model framework of the compressed air energy storage system, the simulation module is further configured to:

[0034] The pressure ratio, inlet and outlet temperatures, mass flow rate and speed of the multi-stage compressor under various operating conditions are collected, and a compressor unit characteristic dataset is constructed based on the pressure ratio, inlet and outlet temperatures, mass flow rate and speed of the multi-stage compressor under various operating conditions.

[0035] The compressor unit characteristic dataset is preprocessed to obtain a preprocessed compressor unit characteristic dataset;

[0036] The preprocessed compressor unit characteristic dataset is divided into a training set and a validation set according to a preset partitioning ratio;

[0037] The preset LSTM neural network is trained using the training set, and after training, the preset LSTM neural network is validated using the validation set to obtain validation results. If the validation results do not meet the preset validation conditions, the network structure and / or hyperparameters of the preset LSTM neural network are optimized and adjusted until the validation results of the preset LSTM neural network meet the preset validation conditions, thus obtaining the pre-trained LSTM neural network.

[0038] According to one embodiment of this application, the first construction module is configured to:

[0039] Calculate the pressure ratio, outlet temperature, and isentropic efficiency of the multi-stage compressor, and call the first preset property library to construct a static thermodynamic model of the multi-stage compressor;

[0040] Based on the compressor interstage heat exchanger equations and by calling the second preset property library, a static thermodynamic model of the interstage heat exchanger is constructed.

[0041] According to one embodiment of this application, the equation for the compressor interstage heat exchanger is:

[0042]

[0043] in, This refers to the outlet air temperature of the heat exchanger. Let ΔT be the inlet air temperature of the heat exchanger, and ΔT be the temperature change. The outlet heat exchange medium temperature. The inlet heat exchange medium temperature. τ is the derivative of the temperature change. r ε is the heat transfer inertia time constant. r The coefficient of performance (COP) of the heat exchanger.

[0044] According to the embodiments of this application, a dynamic simulation device for a multi-stage compressor in a compressed air energy storage system constructs a static thermodynamic model framework for the compression side of the compressed air energy storage system based on a static thermodynamic model of the multi-stage compressor and a static thermodynamic model of the interstage heat exchanger. A pre-trained LSTM neural network is embedded into this static thermodynamic model framework, and real-time simulation of the nonlinear characteristics under varying operating conditions is performed during the dynamic process of the system. The pre-trained LSTM neural network is obtained by training a pre-set LSTM neural network using a compressor unit characteristic dataset. This solves the problem in the prior art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions, and improves the accuracy of performance prediction for compressed air energy storage systems under varying operating conditions through dynamic simulation.

[0045] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the multi-stage compressor dynamic simulation method for a compressed air energy storage system as described in the above embodiments.

[0046] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the multi-stage compressor dynamic simulation method for a compressed air energy storage system as described in the above embodiments.

[0047] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0048] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0049] Figure 1 This is a flowchart illustrating a dynamic simulation method for a multi-stage compressor in a compressed air energy storage system according to an embodiment of this application.

[0050] Figure 2 This is a schematic diagram of a compressed-side simulation framework for an embedded LSTM neural network according to an embodiment of this application;

[0051] Figure 3 This is a block diagram of a dynamic simulation device for a multi-stage compressor in a compressed air energy storage system according to an embodiment of this application.

[0052] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0053] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0054] The following describes a dynamic simulation method for a multi-stage compressor in a compressed air energy storage system according to embodiments of this application, with reference to the accompanying drawings. Addressing the problem mentioned in the background art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions, this application provides a dynamic simulation method for a multi-stage compressor in a compressed air energy storage system. A static thermodynamic model framework for the compression side of the compressed air energy storage system is constructed based on a static thermodynamic model of the multi-stage compressor and a static thermodynamic model of the interstage heat exchanger. A pre-trained LSTM neural network is embedded into the static thermodynamic model framework for the compression side of the compressed air energy storage system, and real-time simulation of the nonlinear characteristics under varying operating conditions is performed after embedding. The pre-trained LSTM neural network is obtained by training a preset LSTM neural network using a compressor unit characteristic dataset. This solves the problem in the background art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions, and improves the accuracy of performance prediction for compressed air energy storage systems under varying operating conditions through dynamic simulation.

[0055] Specifically, Figure 1 This is a flowchart illustrating a dynamic simulation method for a multi-stage compressor in a compressed air energy storage system, as provided in an embodiment of this application.

[0056] like Figure 1 As shown, the dynamic simulation method for the multi-stage compressor of this compressed air energy storage system includes the following steps:

[0057] In step S101, a static thermodynamic model of a multi-stage compressor and a static thermodynamic model of an interstage heat exchanger are constructed.

[0058] Furthermore, in some embodiments, constructing a static thermodynamic model of a multi-stage compressor and a static thermodynamic model of an interstage heat exchanger includes: calculating the pressure ratio, outlet temperature, and isentropic efficiency of the multi-stage compressor, and calling a first preset property library to construct a static thermodynamic model of the multi-stage compressor; and constructing a static thermodynamic model of the interstage heat exchanger based on the compressor interstage heat exchanger equation and calling a second preset property library.

[0059] Among them, the first preset property library is a real property library, and the second preset property library is a heat exchange working fluid property library.

[0060] Specifically, firstly, a static thermodynamic model of a multi-stage compressor is constructed, where the pressure ratio of the k-th stage compressor is:

[0061]

[0062] Where, β c,k Let k be the pressure ratio of the k-th stage compressor. Let K be the inlet pressure of the k-th stage compressor. Let be the outlet pressure of the k-th stage compressor.

[0063] Furthermore, the ratio of the outlet temperature to the inlet temperature of the k-th stage compressor can be approximated based on the pressure ratio:

[0064]

[0065] in, Let K be the inlet air temperature of the k-th stage compressor. Let γ be the outlet air temperature of the k-th stage compressor, and γ be the polytropic coefficient.

[0066] Furthermore, treating air as an ideal gas, the isentropic efficiency of the k-th stage compressor can be obtained as:

[0067]

[0068] Where, η ci,k Let be the isentropic efficiency of the k-th stage compressor. Let be the outlet temperature of the isentropic process in the k-th stage compressor. Let be the inlet temperature of the isentropic process in the k-th stage compressor. Let K be the outlet air temperature of the k-th stage compressor. Let be the inlet air temperature of the k-th stage compressor.

[0069] Combining equations (1), (2), and (3) above, we can obtain the outlet air temperature and power of the k-th stage compressor as follows:

[0070]

[0071] in, Let η be the outlet air temperature of the k-th stage compressor. ci,k Let be the isentropic efficiency of the k-th stage compressor. Let β be the inlet air temperature of the k-th stage compressor. c,k Let P be the pressure ratio of the k-th stage compressor, γ be the polytropic coefficient, and P be the pressure ratio of the k-th stage compressor. c,k Let η be the power of the k-th stage compressor. cm,k For mechanical efficiency, c p,air The specific heat capacity of air at constant pressure. For air mass flow rate.

[0072] In the process of constructing a static thermodynamic model of a multi-stage compressor, the specific heat capacity involved can be calculated using REFPROP by inputting the temperature and pressure at the corresponding time.

[0073] Secondly, a static thermodynamic model of the interstage heat exchanger is constructed.

[0074] In some embodiments, the equations for the compressor interstage heat exchanger are as follows:

[0075]

[0076] in, This refers to the outlet air temperature of the heat exchanger. Let ΔT be the inlet air temperature of the heat exchanger, and ΔT be the temperature change. The outlet heat exchange medium temperature. The inlet heat exchange medium temperature. τ is the derivative of the temperature change. r ε is the heat transfer inertia time constant. r The coefficient of performance (COP) of the heat exchanger.

[0077] In the process of constructing the static thermodynamic model of the interstage heat exchanger, the specific heat capacity involved can be calculated using REFPROP. By inputting the temperature and pressure at the corresponding time, the specific heat capacity of air and heat exchange medium at the corresponding time can be calculated respectively.

[0078] Therefore, in this embodiment, the pressure and temperature functions at the inlet and outlet of the compressor are constructed based on the pressure ratio and isentropic efficiency, respectively, and the pressure ratio and isentropic efficiency are expressed as implicit functions of rotational speed and mass flow rate. Furthermore, a real property library is introduced to correct the ideal gas process. A thermodynamic static model of the interstage heat exchanger is constructed based on the ε-NTU method, and a heat exchange working fluid property library is introduced to correct the thermal parameters.

[0079] In step S102, a static thermodynamic model framework for the compression side of the compressed air energy storage system is constructed based on the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the interstage heat exchanger.

[0080] Specifically, the embodiments of this application can combine multi-stage compressor and heat exchanger models based on the principle of consistency of boundary mass flow rate and pressure to construct a static thermodynamic model framework for the compression side of the compressed air energy storage system. That is, based on the principle that the mass flow rate of the air at the outlet of each stage compressor is the same as the mass flow rate at the inlet of the interstage heat exchanger, and the pressure at the interface is equal, the compressors and interstage heat exchangers at each stage are connected to complete the construction of the static thermodynamic model for the compression side of the compressed air energy storage system.

[0081] In step S103, a pre-trained LSTM neural network is embedded into the static thermodynamic model framework of the compressed air energy storage system on the compression side, and after the embedding is completed, a real-time simulation of the nonlinear characteristics of the system under varying operating conditions is performed during the dynamic process. The pre-trained LSTM neural network is obtained by training the preset LSTM neural network with the compressor unit characteristic dataset.

[0082] In some embodiments, before embedding the pre-trained LSTM neural network into the static thermodynamic model framework of the compressed air energy storage system, the method further includes: collecting the pressure ratio, inlet and outlet temperatures, mass flow rates, and rotational speeds of the multi-stage compressor under various operating conditions, and constructing a compressor unit characteristic dataset based on the pressure ratio, inlet and outlet temperatures, mass flow rates, and rotational speeds of the multi-stage compressor under various operating conditions; preprocessing the compressor unit characteristic dataset to obtain a preprocessed compressor unit characteristic dataset; dividing the preprocessed compressor unit characteristic dataset into a training set and a validation set according to a preset partitioning ratio; training the preset LSTM neural network using the training set, and validating the preset LSTM neural network using the validation set after training to obtain validation results; if the validation results do not meet the preset validation conditions, optimizing and adjusting the network structure and / or hyperparameters of the preset LSTM neural network until the validation results of the preset LSTM neural network meet the preset validation conditions, thus obtaining the pre-trained LSTM neural network.

[0083] The preset division ratio can be a division ratio pre-set by those skilled in the art, such as 7:3, and is not specifically limited here. The preset verification conditions can also be verification conditions pre-set by those skilled in the art, such as accuracy reaching a preset threshold, and are not specifically limited here.

[0084] Specifically, this application embodiment can collect measured values ​​of compressor pressure ratio, inlet and outlet temperatures, mass flow rate, and rotational speed under various operating conditions provided by the manufacturer to construct a compressor unit characteristic dataset. The operating conditions must cover start-up and shutdown, rated power operation, wide-condition sliding pressure operation, surge, and blockage. The data in the compressor unit characteristic dataset is preprocessed, including data cleaning, outlier removal, and data standardization, to ensure data quality and suitability for training neural networks. The preprocessed compressor unit characteristic dataset is then divided into training and validation sets according to a preset ratio (e.g., 7:3) for subsequent use.

[0085] Furthermore, to accurately represent the nonlinear characteristics of a multi-stage compressor, a pre-defined LSTM neural network is constructed to nonlinearly fit the coupling relationship between compressor speed, mass flow rate, pressure ratio, and isentropic efficiency. This network includes an input layer, multiple hidden layers, and an output layer. The input layer receives the time-series signals of compressor speed and mass flow rate under different operating conditions as input quantities. Each hidden layer contains 64 hidden units and employs the ReLU (Rectified Linear Unit) activation function. The output layer outputs the pressure ratio and isentropic efficiency of each compressor stage. Thus, using this LSTM structure, its powerful nonlinear time-series fitting capability can translate the nonlinear characteristics of the compressor and the coupling relationship between its stages.

[0086] Furthermore, LSTM neural networks involve multiple hyperparameters. To determine suitable hyperparameters, a grid search method can be used for hyperparameter optimization. Specifically, the number of network layers, the number of hidden units, and the learning rate are combined to form different network structures. Then, each network structure is trained using a training set, and its performance is evaluated using a validation set. Based on the validation results, the network structure with the best performance is selected as the final model. Additionally, a learning rate decay strategy can be employed, gradually reducing the learning rate during training to improve the model's convergence and stability.

[0087] Furthermore, after completing the pre-set LSTM neural network construction, a pre-set LSTM neural network reflecting the nonlinear characteristics of the compressor under varying operating conditions is trained. In this embodiment, the pre-set LSTM neural network is trained using a constructed training set. During training, the network's weights and biases are iterated and adjusted multiple times through forward and backward propagation. In each iteration, the difference between the network's predicted and actual values ​​is calculated, and the network's weights and biases are updated using the backpropagation algorithm. In addition, batch gradient descent is employed, using a small batch of samples for calculation each time the weights are updated, to accelerate the training process.

[0088] Furthermore, after training, the pre-trained LSTM neural network model is evaluated using a validation set. If the performance of the pre-trained LSTM neural network model does not meet the pre-defined validation conditions, the network structure and hyperparameters need to be adjusted and optimized so that the neural network can accurately represent the pressure ratio and isentropic efficiency based on mass flow rate and rotational speed. Specifically, the number of layers or hidden units in the network can be increased, or the learning rate can be adjusted. In addition, techniques such as regularization can be used to prevent overfitting. Through continuous iteration and optimization, until the performance of the pre-trained LSTM neural network on the validation set meets the pre-defined validation conditions, a pre-trained LSTM neural network is obtained, thereby achieving an accurate representation of the nonlinear characteristics of the compressor under varying operating conditions.

[0089] Furthermore, after obtaining the pre-trained LSTM neural network, the trained LSTM neural network model is integrated into the thermodynamic model framework of the compressed air energy storage system to replace the implicit pressure ratio and isentropic efficiency functions. This allows for real-time matching of nonlinear operating conditions by inputting discrete signals of mass flow rate and rotational speed. During iterative calculations, the trained neural network is called in each loop to achieve real-time updates of compressor operating conditions and nonlinear real characteristics.

[0090] Furthermore, in some embodiments, after performing real-time variable operating condition nonlinear characteristic simulation, the method further includes: acquiring multiple sets of time-series data of mass flow rate, rotational speed, and pressure; setting initial time constant parameter estimates and corresponding error covariance matrices; and obtaining the predicted data for the current time based on the system model and the time constant parameter estimates of the previous time step; calculating the Kalman gain based on the time-series data and the predicted data; and updating the time constant parameter estimates and error covariance matrix based on the Kalman gain; if the updated time constant parameter estimates do not meet the preset convergence conditions, the step of acquiring multiple sets of time-series data of mass flow rate, rotational speed, and pressure is repeated until the updated time constant parameter estimates meet the preset convergence conditions; constructing an inertial element based on the converged time constant parameter estimates; and embedding the inertial element into the static thermodynamic model framework of the compressed air energy storage system on the compression side for dynamic process simulation fitting.

[0091] The preset convergence condition can be a convergence condition pre-set by those skilled in the art, and is not specifically limited here.

[0092] Specifically, in this embodiment, after embedding the trained LSTM neural network into the pressure ratio and isentropic efficiency function in the thermodynamic model framework of the compressed air energy storage system to realize the real-time simulation of the nonlinear characteristics of the system under varying operating conditions during the dynamic process, the system dynamic process time constant parameters are identified by the Kalman filter algorithm based on the actual measured parameters. Based on the identified constants, a first-order or higher-order inertial element is constructed and embedded in the iterative process of pressure, temperature and mass flow rate to achieve high-precision fitting modeling of the system dynamic response performance. Finally, a nonlinear dynamic simulation system for the compression side of the compressed air energy storage system under varying operating conditions is constructed.

[0093] For example, firstly, this application embodiment can collect multiple sets of time-series data on mass flow rate, rotational speed, and pressure based on experimental data provided by the equipment manufacturer and measured data from the field unit. Then, it initializes the Kalman filter algorithm, setting initial time constant parameter estimates and corresponding error covariance matrices. Subsequently, it performs a Kalman filter prediction step, predicting the system state and parameters at the current moment based on the system model and the time constant parameter estimate from the previous moment. Following this, it performs a Kalman filter update step, calculating the Kalman gain based on the actual observed data and predicted data, and updating the estimated time constant parameters and error covariance matrix.

[0094] Furthermore, the prediction and update steps of the Kalman filter are repeated until the estimated value of the time constant parameter satisfies the preset convergence condition. Finally, based on the estimated value of the time constant parameter that meets the preset convergence condition, a first-order or higher-order inertial element is constructed. The inertial element is embedded into the previously constructed static thermodynamic model framework of the compressed air energy storage system on the compression side. In each cycle of updating the state parameters, the inertial element must be passed through, thereby achieving simulation fitting of the dynamic process.

[0095] Secondly, a nonlinear dynamic model of the compressed air energy storage system under varying operating conditions on the compression side is constructed.

[0096] Specifically, based on the constructed static thermodynamic model framework of the compressed air energy storage system, a pre-set LSTM neural network is embedded to reflect the fitting relationship between rotational speed, mass flow rate, pressure ratio, and isentropic efficiency. The inertial time constant of the dynamic process of pressure, rotational speed, and mass flow rate is fitted by the Kalman filter algorithm. By combining the above steps, a nonlinear dynamic model of the compressed air energy storage system under varying operating conditions can be constructed and stored in a hardware medium for use.

[0097] Therefore, this application, based on the static thermodynamic model framework of the compressed air energy storage system, maps the implicit functions of pressure ratio and isentropic efficiency with rotational speed and mass flow rate through a trained LSTM neural network, and identifies the inertial time constant of the dynamic process through the Kalman filter algorithm, thereby achieving accurate simulation of the nonlinear coupled dynamic process of the compressed air energy storage system under different operating conditions on the compressed side.

[0098] To facilitate a clearer and more intuitive understanding of the multi-stage compressor dynamic simulation method for compressed air energy storage systems proposed in this application, the following section combines... Figure 2 Please provide a detailed explanation.

[0099] Specifically, such as Figure 2As shown, firstly, a static thermodynamic model framework including the compressor and interstage heat exchanger is established, where the pressure ratio and isentropic efficiency are modeled as implicit functions of mass flow rate and rotational speed. Then, based on the measured values ​​(i.e., test data) of pressure ratio and isentropic efficiency under various operating conditions provided by the manufacturer, a compressor unit characteristic dataset is constructed, and an LSTM neural network representing the nonlinear characteristics of a multi-stage compressor is trained using this dataset. Finally, based on the thermodynamic model framework of the compressed air energy storage system, the trained LSTM neural network is embedded to reflect the fitting relationship between rotational speed, mass flow rate, pressure ratio, and isentropic efficiency. Based on the test data, the Kalman filter algorithm is used to identify the system's dynamic process time constant parameters, and a first-order or higher-order inertial element is constructed based on the identified constant, embedded in the iterative process of mass flow rate, rotational speed, pressure, and temperature, to achieve simulation modeling of the thermodynamic steady-state and nonlinear variable operating condition dynamic characteristics of the compressed air energy storage system on the compression side.

[0100] According to the dynamic simulation method for multi-stage compressors in compressed air energy storage systems according to embodiments of this application, a static thermodynamic model framework for the compression side of the compressed air energy storage system is constructed based on the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the interstage heat exchanger. A pre-trained LSTM neural network is embedded into the static thermodynamic model framework of the compression side of the compressed air energy storage system, and real-time simulation of the nonlinear characteristics under varying operating conditions is performed after embedding. The pre-trained LSTM neural network is obtained by training a preset LSTM neural network using a compressor unit characteristic dataset. This solves the problem in the prior art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions, and improves the accuracy of performance prediction for compressed air energy storage systems under varying operating conditions through dynamic simulation.

[0101] Next, referring to the accompanying drawings, a dynamic simulation device for a multi-stage compressor of a compressed air energy storage system according to an embodiment of this application is described.

[0102] Figure 3 This is a block diagram of a multi-stage compressor dynamic simulation device for a compressed air energy storage system according to an embodiment of this application.

[0103] like Figure 3 As shown, the multi-stage compressor dynamic simulation device 10 of the compressed air energy storage system includes: a first building module 100, a second building module 200 and a simulation module 300.

[0104] The system comprises: a first construction module 100 for constructing a static thermodynamic model of a multi-stage compressor and a static thermodynamic model of an interstage heat exchanger; a second construction module 200 for constructing a static thermodynamic model framework for the compression side of a compressed air energy storage system based on the static thermodynamic models of the multi-stage compressor and the interstage heat exchanger; and a simulation module 300 for embedding a pre-trained LSTM neural network into the static thermodynamic model framework for the compression side of the compressed air energy storage system, and performing real-time simulation of the nonlinear characteristics under varying operating conditions during the dynamic process of the system after embedding. The pre-trained LSTM neural network is obtained by training a preset LSTM neural network using a compressor unit characteristic dataset.

[0105] Furthermore, in some embodiments, after performing real-time variable operating condition nonlinear characteristic simulation, the simulation module 300 is also used to: acquire multiple sets of time-series data of mass flow rate, rotational speed, and pressure; set initial time constant parameter estimates and corresponding error covariance matrices; and obtain the predicted data for the current time based on the system model and the time constant parameter estimates of the previous time step; calculate the Kalman gain based on the time-series data and the predicted data; and update the estimated value of the time constant parameter and the error covariance matrix based on the Kalman gain; if the updated estimated value of the time constant parameter does not meet the preset convergence condition, the step of acquiring multiple sets of time-series data of mass flow rate, rotational speed, and pressure is repeated until the updated estimated value of the time constant parameter meets the preset convergence condition; an inertial element is constructed based on the converged estimated value of the time constant parameter; and the inertial element is embedded into the static thermodynamic model framework of the compressed air energy storage system on the compression side for dynamic process simulation fitting.

[0106] Furthermore, in some embodiments, before embedding the pre-trained LSTM neural network into the static thermodynamic model framework of the compressed air energy storage system, the simulation module 300 is also used to: collect the pressure ratio, inlet and outlet temperatures, mass flow rates, and rotational speeds of the multi-stage compressor under various operating conditions, and construct a compressor unit characteristic dataset based on the pressure ratio, inlet and outlet temperatures, mass flow rates, and rotational speeds of the multi-stage compressor under various operating conditions; preprocess the compressor unit characteristic dataset to obtain a preprocessed compressor unit characteristic dataset; divide the preprocessed compressor unit characteristic dataset into a training set and a validation set according to a preset partitioning ratio; train the preset LSTM neural network using the training set, and after training, validate the preset LSTM neural network using the validation set to obtain validation results. If the validation results do not meet the preset validation conditions, the network structure and / or hyperparameters of the preset LSTM neural network are optimized and adjusted until the validation results of the preset LSTM neural network meet the preset validation conditions, thus obtaining the pre-trained LSTM neural network.

[0107] Furthermore, in some embodiments, the first construction module 200 is used to: calculate the pressure ratio, outlet temperature, and isentropic efficiency of the multi-stage compressor, and call a first preset property library to construct a static thermodynamic model of the multi-stage compressor; and construct a static thermodynamic model of the interstage heat exchanger based on the compressor interstage heat exchanger equation and call a second preset property library.

[0108] Furthermore, in some embodiments, the equations for the compressor interstage heat exchanger are:

[0109]

[0110] in, This refers to the outlet air temperature of the heat exchanger. Let ΔT be the inlet air temperature of the heat exchanger, and ΔT be the temperature change. The outlet heat exchange medium temperature. The inlet heat exchange medium temperature. τ is the derivative of the temperature change. r ε is the heat transfer inertia time constant. r The coefficient of performance (COP) of the heat exchanger.

[0111] It should be noted that the explanation of the above-mentioned embodiment of the dynamic simulation method for multi-stage compressors of compressed air energy storage system also applies to the dynamic simulation device for multi-stage compressors of compressed air energy storage system in this embodiment, and will not be repeated here.

[0112] According to the embodiments of this application, a dynamic simulation device for a multi-stage compressor in a compressed air energy storage system constructs a static thermodynamic model framework for the compression side of the compressed air energy storage system based on a static thermodynamic model of the multi-stage compressor and a static thermodynamic model of the interstage heat exchanger. A pre-trained LSTM neural network is embedded into this static thermodynamic model framework, and real-time simulation of the nonlinear characteristics under varying operating conditions is performed during the dynamic process of the system. The pre-trained LSTM neural network is obtained by training a pre-set LSTM neural network using a compressor unit characteristic dataset. This solves the problem in the prior art of accurately simulating and optimizing the dynamic performance of multi-stage compressors in compressed air energy storage systems under varying operating conditions, and improves the accuracy of performance prediction for compressed air energy storage systems under varying operating conditions through dynamic simulation.

[0113] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0114] The memory 401, the processor 402, and the computer program stored on the memory 401 and capable of running on the processor 402.

[0115] When the processor 402 executes the program, it implements the multi-stage compressor dynamic simulation method for the compressed air energy storage system provided in the above embodiments.

[0116] Furthermore, electronic devices also include:

[0117] Communication interface 403 is used for communication between memory 401 and processor 402.

[0118] The memory 401 is used to store computer programs that can run on the processor 402.

[0119] The memory 401 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0120] If the memory 401, processor 402, and communication interface 403 are implemented independently, then the communication interface 403, memory 401, and processor 402 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized into address buses, data buses, control buses, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0121] Optionally, in a specific implementation, if the memory 401, processor 402, and communication interface 403 are integrated on a single chip, then the memory 401, processor 402, and communication interface 403 can communicate with each other through an internal interface.

[0122] Processor 402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0123] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described dynamic simulation method for a multi-stage compressor in a compressed air energy storage system.

[0124] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0125] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0126] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method of dynamic simulation of a multistage compressor of a compressed air energy storage system, characterized in that, The method comprises the following steps: constructing a static thermodynamic model of a multi-stage compressor and a static thermodynamic model of an inter-stage heat exchanger; constructing a static thermodynamic model framework of a compressed air energy storage system compression side according to the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the inter-stage heat exchanger; embedding a pre-trained LSTM neural network into the static thermodynamic model framework of the compressed air energy storage system compression side, and performing real-time variable working condition nonlinear characteristic simulation in a system dynamic process after embedding is completed, wherein the pre-trained LSTM neural network is obtained by training a preset LSTM neural network by using a compressor unit characteristic data set; after the real-time variable working condition nonlinear characteristic simulation is performed, further comprising: acquiring a plurality of groups of time sequence acquisition data of mass flow rate, rotating speed and pressure, setting an initial time constant parameter estimation value and a corresponding error covariance matrix, and obtaining predicted data at a current time according to a system model and a time constant parameter estimation value at a previous time; calculating a Kalman gain according to the time sequence acquisition data and the predicted data, and updating the estimation value of the time constant parameter and the error covariance matrix based on the Kalman gain; if the updated estimation value of the time constant parameter does not satisfy a preset convergence condition, re-executing the step of acquiring the plurality of groups of time sequence acquisition data of mass flow rate, rotating speed and pressure until the updated estimation value of the time constant parameter satisfies the preset convergence condition, constructing an inertial link according to the converged time constant parameter estimation value, and embedding the inertial link into the static thermodynamic model framework of the compressed air energy storage system compression side to perform dynamic process simulation fitting.

2. The method of claim 1, wherein, Before embedding the pre-trained LSTM neural network into the static thermodynamic model framework of the compressed air energy storage system compression side, further comprising: acquiring pressure ratio, inlet and outlet temperature, mass flow rate and rotating speed of the multi-stage compressor under multiple working conditions, and constructing a compressor unit characteristic data set according to the pressure ratio, inlet and outlet temperature, mass flow rate and rotating speed of the multi-stage compressor under the multiple working conditions; preprocessing the compressor unit characteristic data set to obtain a preprocessed compressor unit characteristic data set; dividing the preprocessed compressor unit characteristic data set into a training set and a validation set according to a preset division ratio; training the preset LSTM neural network by using the training set, and verifying the preset LSTM neural network by using the validation set after training is completed, obtaining a verification result, if the verification result does not satisfy a preset verification condition, optimizing and adjusting a network structure and / or hyperparameter of the preset LSTM neural network until the verification result of the preset LSTM neural network satisfies the preset verification condition, and obtaining the pre-trained LSTM neural network.

3. The method of claim 1, wherein, The method of constructing the static thermodynamic model of the multi-stage compressor and the static thermodynamic model of the inter-stage heat exchanger comprises: calculating pressure ratio, outlet temperature and isentropic efficiency of the multi-stage compressor, and calling a first preset property library to construct the static thermodynamic model of the multi-stage compressor; The inter-stage heat exchanger static thermodynamic model is constructed based on a compressor inter-stage heat exchanger equation and by calling a second preset property library.

4. The method of claim 3, wherein, The compressor inter-stage heat exchanger equation is as follows: wherein, Tout is the outlet air temperature of the heat exchanger, Tin is the inlet air temperature of the heat exchanger, ΔT is the temperature change, Tout,ex is the outlet heat transfer medium temperature, Tin,ex is the inlet heat transfer medium temperature, is the derivative of the temperature change, is the heat transfer inertia time constant, is the energy efficiency coefficient of the heat exchanger.

5. A dynamic simulation apparatus of a multi-stage compressor of a compressed air energy storage system, characterized in that, The method comprises the following steps: A first construction module is configured to construct a multi-stage compressor static thermodynamic model and an inter-stage heat exchanger static thermodynamic model. A second construction module is configured to construct a compressed air energy storage system compression side static thermodynamic model framework according to the multi-stage compressor static thermodynamic model and the inter-stage heat exchanger static thermodynamic model. A simulation module is configured to embed a pre-trained LSTM neural network into the compressed air energy storage system compression side static thermodynamic model framework and perform real-time variable working condition nonlinear characteristic simulation in a system dynamic process after embedding is completed, wherein the pre-trained LSTM neural network is obtained by training a preset LSTM neural network by using a compressor unit characteristic data set. After the real-time variable working condition nonlinear characteristic simulation is performed, the simulation module is further configured to: acquire a plurality of groups of time sequence acquisition data of mass flow rate, rotating speed and pressure, set an initial time constant parameter estimation value and a corresponding error covariance matrix, and obtain predicted data at a current time according to a system model and a time constant parameter estimation value at a previous time; calculate a Kalman gain according to the time sequence acquisition data and the predicted data, and update the estimation value of the time constant parameter and the error covariance matrix based on the Kalman gain; if the updated estimation value of the time constant parameter does not satisfy a preset convergence condition, the step of acquiring the plurality of groups of time sequence acquisition data of mass flow rate, rotating speed and pressure is re-executed until the updated estimation value of the time constant parameter satisfies the preset convergence condition, an inertial link is constructed according to the converged time constant parameter estimation value, and the inertial link is embedded into the compressed air energy storage system compression side static thermodynamic model framework to perform dynamic process simulation fitting.

6. The apparatus of claim 5, wherein, Before the pre-trained LSTM neural network is embedded into the compressed air energy storage system compression side static thermodynamic model framework, the simulation module is further configured to: acquire pressure ratio, inlet and outlet temperature, mass flow rate and rotating speed of the multi-stage compressor under a plurality of working conditions, and construct a compressor unit characteristic data set according to the pressure ratio, inlet and outlet temperature, mass flow rate and rotating speed of the multi-stage compressor under the plurality of working conditions; preprocess the compressor unit characteristic data set to obtain a preprocessed compressor unit characteristic data set; divide the preprocessed compressor unit characteristic data set into a training set and a verification set according to a preset division ratio; train the preset LSTM neural network by using the training set, and verify the preset LSTM neural network by using the verification set after training is completed to obtain a verification result, if the verification result does not satisfy a preset verification condition, optimize and adjust a network structure and / or hyperparameter of the preset LSTM neural network until the verification result of the preset LSTM neural network satisfies the preset verification condition, and obtain the pre-trained LSTM neural network.

7. An electronic device, comprising: The method comprises the following steps: The method comprises the following steps: - a memory, a processor and a computer program stored on the memory and runable on the processor, the processor executing the program to implement the method of dynamic simulation of a multistage compressor of a compressed air energy storage system according to any one of claims 1-4.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by a processor for implementing the method of dynamic simulation of a multistage compressor of a compressed air energy storage system according to any one of claims 1-4.

Images