Compressed air energy storage chamber fatigue life prediction method, system and storage device

Abstract

The present invention discloses a method, system, and storage device for predicting the fatigue life of a compressed air energy storage cavern, belonging to the field of compressed air energy storage technology. The method comprises: preprocessing the compressed air storage cavern data to be analyzed to obtain a sample data set; inputting the sample data set into a pretrained prediction model for key parameters of the gas storage cavern structure, analyzing the data to obtain analysis results, and predicting the fatigue life of the compressed air storage cavern structure based on the analysis results. This method can accurately predict the fatigue life of the compressed air energy storage cavern, identify potential safety and efficiency issues in advance, effectively reduce the risk of fatigue damage during long-term operation of underground caverns, and ensure the stability of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of compressed air energy storage, and in particular to a method, system and storage device for predicting fatigue life of a compressed air energy storage chamber. Background Art

[0002] As the installed capacity of renewable energy sources like solar and wind continues to expand, the proportion of renewable energy generation in the energy supply system is increasing. However, due to the significant volatility and intermittency of renewable energy sources like solar and wind, their power generation is significantly affected by natural factors such as weather and seasons, making it difficult to achieve the stable and continuous power output achieved by traditional fossil fuel generation. This instability in power generation poses a significant challenge to the stable operation of the power grid. When renewable energy generation fluctuates significantly, the supply and demand balance of the power grid is disrupted, potentially leading to a series of problems such as frequency deviation and voltage fluctuations. In severe cases, it can even cause power grid collapse, threatening the safe and reliable operation of the entire power system.

[0003] Against this backdrop, compressed air energy storage technology, a highly promising energy storage solution, has garnered widespread attention and entered a golden period of rapid development. This technology cleverly leverages the compressibility and storage properties of air. During periods of low electricity demand, excess power in the grid is used to compress air, which is then stored in specialized spaces such as underground caverns. When electricity demand peaks, the stored compressed air is released, and a device like an expander converts the air's internal energy into mechanical energy, which in turn drives a generator. This achieves both energy storage and energy release, effectively balancing the power grid's supply and demand at different times, and provides crucial technical support for addressing the intermittent and volatile nature of renewable energy generation.

[0004] However, the application of compressed air energy storage technology in underground caverns faces a critical and thorny issue: fatigue damage to the underground cavern structure. During the long-term cycle of inflation, storage, and deflation, the underground cavern structure is subjected to cyclical pressure fluctuations. This repeated stress can cause fatigue damage to the cavern surrounding rock and lining. As damage accumulates, the cavern may experience crack expansion, reduced sealing performance, and even air leakage. This not only significantly reduces the efficiency of the energy storage system and increases operating costs, but can also cause serious safety accidents, posing a threat to the surrounding environment and personnel safety. Summary of the Invention

[0005] In response to existing problems with renewable energy generation fluctuations causing grid instability, and the long-term fatigue damage of compressed air energy storage underground caverns, which can lead to efficiency and safety issues, the present invention provides a method for predicting the fatigue life of compressed air energy storage caverns. This method can accurately predict the fatigue life of compressed air energy storage caverns, identify potential safety and efficiency issues in advance, effectively reduce the risk of fatigue damage during long-term operation of underground caverns, and ensure grid stability.

[0006] In order to achieve the above objectives, the present invention provides the following technical solutions.

[0007] In a first aspect, the present invention provides a method for predicting fatigue life of a compressed air energy storage chamber, comprising:

[0008] Preprocess the compressed air storage cavity data to be analyzed to obtain a sample data set;

[0009] The sample data set is input into a pre-trained prediction model of key parameters of the gas storage cavern structure, and the analysis is performed to obtain analysis results. The fatigue life of the compressed air storage cavern structure is predicted based on the analysis results;

[0010] The training method of the pre-trained prediction model of key parameters of the gas storage cavern structure includes:

[0011] Obtain compressed air storage cavern data and preprocess it to obtain key parameter data sets;

[0012] Use the CNN spatial feature extraction model to extract spatial features from the key parameter data set to obtain spatial feature data;

[0013] The spatial feature data is input into the LSTM-based prediction model for training and processing to obtain a pre-trained prediction model for key parameters of the gas storage cavern structure.

[0014] As a further improvement of the present invention, the compressed air storage cavity data is obtained and preprocessed to obtain a key parameter data set, including:

[0015] Based on the engineering geological data, design parameters, and operating parameters of the compressed air storage caverns in the data, a numerical simulation method was used to simulate the filling and degassing process of the compressed air energy storage underground caverns, and the key parameters of the cavern steel lining, concrete lining, and surrounding rock were obtained.

[0016] The key parameters of the cavern steel lining, concrete lining and surrounding rock include the maximum tensile strain of the steel lining , minimum tensile strain of steel lining , maximum tensile stress of concrete lining , the first principal stress of surrounding rock , the second principal stress of surrounding rock and the third principal stress of surrounding rock .

[0017] As a further improvement of the present invention, the spatial feature extraction of the key parameter data set using the CNN spatial feature extraction model to obtain spatial feature data includes:

[0018] The CNN spatial feature extraction model is used to process the key parameter data set layer by layer to extract feature maps;

[0019] The mean square error loss function is used to calculate the loss function between the output features and the true features in the feature map ;

[0020] in, is the loss value; is the sample size; It is True values; It is predicted values, ;

[0021] The parameters of CNN are updated according to the gradient of the loss function. CNN is optimized through the back-propagation algorithm to adjust the convolution kernel and weights. After that, the loss function is recalculated until the loss value L of the loss function meets the set conditions, and the feature sequence is output to obtain spatial feature data.

[0022] As a further improvement of the present invention, the spatial feature data is input into the LSTM-based prediction model for training to obtain a pre-trained prediction model of key parameters of the gas storage cavern structure, including:

[0023] Input the training data set in the spatial feature data into the LSTM-based prediction model for training to obtain the training results and the trained LSTM-based prediction model;

[0024] Input the test data set in the spatial feature data into the trained LSTM-based prediction model to obtain the test results;

[0025] The training results are compared with the test results to determine whether the LSTM-based prediction model has converged. If so, the pre-trained prediction model of the key parameters of the gas storage cavern structure is obtained.

[0026] As a further improvement of the present invention, the training data set in the spatial feature data is input into the LSTM-based prediction model for training to obtain the training results, including:

[0027] Build an LSTM-based prediction model based on the LSTM network, and set the input layer, LSTM layer and output layer in the LSTM-based prediction model;

[0028] The input layer receives the training data set in the spatial feature data, extracts the feature sequence in the training data set in the spatial feature data, and sends the extracted feature sequence in the training data set to the LSTM layer;

[0029] The LSTM layer receives the feature sequence in the training data set from the input layer and trains the LSTM layer. After training, it captures the changing trend of key parameters in several cycles and sends the changing trend to the output layer.

[0030] The output layer receives the change trend from the LSTM layer and generates training results.

[0031] As a further improvement of the present invention, the LSTM layer receives the feature sequence in the training set transmitted from the input layer, and trains the LSTM layer, including:

[0032] The LSTM layer receives the feature sequence in the training set from the input layer, calculates the output of the feature sequence in the training set at each time step in order of time steps, obtains the predicted cavern structure state parameters, compares the predicted cavern structure state parameters with the true values, and calculates the loss function;

[0033] The LSTM layer is updated according to the gradient of the loss function. The gradient of the loss function in each time step is reversely calculated using the BPTT algorithm. The parameters of the LSTM network are updated using the gradient descent method until the parameters of the LSTM network meet the set conditions, completing the training of the LSTM layer.

[0034] As a further improvement of the present invention, the sample data set is input into a pre-trained prediction model of key parameters of the gas storage cavern structure, and analyzed to obtain analysis results, and the fatigue life of the compressed air storage cavern structure is predicted based on the analysis results, including:

[0035] The sample data set is input into the trained prediction model of key parameters of the gas storage cavern structure to obtain the prediction results;

[0036] The prediction results are analyzed according to fatigue criteria to obtain the fatigue life prediction results of compressed air energy storage caverns;

[0037] The cavern structural parameters are optimized based on the fatigue life prediction results of the compressed air energy storage cavern and the engineering design service life.

[0038] In a second aspect, the present invention provides a system for predicting fatigue life of a compressed air energy storage chamber, comprising:

[0039] Dataset acquisition module: used to pre-process the compressed air storage cavern data to be analyzed to obtain a sample data set;

[0040] Analysis result module: used to input the sample data set into the pre-trained prediction model of key parameters of the gas storage cavern structure, perform analysis, obtain analysis results, and predict the fatigue life of the compressed air storage cavern structure based on the analysis results;

[0041] The training method of the pre-trained prediction model of key parameters of the gas storage cavern structure includes:

[0042] Obtain compressed air storage cavern data and preprocess it to obtain key parameter data sets;

[0043] Use the CNN spatial feature extraction model to extract spatial features from the key parameter data set to obtain spatial feature data;

[0044] The spatial feature data is input into the LSTM-based prediction model for training and processing to obtain a pre-trained prediction model for key parameters of the gas storage cavern structure.

[0045] In a third aspect, the present invention 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 implements the method for predicting fatigue life of a compressed air energy storage chamber when executing the computer program.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting fatigue life of a compressed air energy storage chamber is implemented.

[0047] In a fifth aspect, the present invention provides a computer program product comprising computer instructions, which, when executed by a processor, implement the method for predicting fatigue life of a compressed air energy storage chamber.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] This application accurately predicts the fatigue life of the compressed air energy storage cavern, and can predict in advance the fatigue damage that may occur to the cavern structure during long-term operation, thereby providing a scientific basis for grid dispatching, enabling the grid to reasonably allocate power resources according to the health status of the energy storage cavern, effectively buffering the impact of new energy power generation fluctuations on the grid, greatly enhancing the stability and reliability of the grid, and ensuring the continuity of power supply. In addition, this application, through the organic combination of preprocessing technology, CNN spatial feature extraction model and LSTM algorithm, can deeply mine the key information in the cavern operation data, accurately analyze the fatigue evolution law of the cavern structure, and discover potential safety and efficiency problems in advance. Based on accurate fatigue life prediction results, operation and maintenance personnel can take targeted maintenance and repair measures in a timely manner to avoid further deterioration of the cavern structure, effectively reduce the risk of fatigue damage in the long-term operation of the underground cavern, extend the service life of the cavern, and ensure that the energy storage system is always in an efficient operation state, thereby improving energy utilization efficiency and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are for illustration purposes only and are not intended to limit the scope of the present disclosure in any way. In the drawings:

[0051] Figure 1 This is a flow chart of a method for predicting fatigue life of a compressed air energy storage chamber according to the present invention;

[0052] Figure 2 A schematic flow chart of a method for training a pre-trained prediction model of key parameters of a gas storage cavern structure according to the present invention;

[0053] Figure 3 This is a schematic diagram of a specific process of a method for predicting fatigue life of a compressed air energy storage chamber according to the present invention;

[0054] Figure 4 A schematic diagram of numerical simulation of the operation of a compressed air energy storage cavern in a method for predicting fatigue life of a compressed air energy storage cavern according to the present invention;

[0055] Figure 5 This is a schematic diagram of the variation rules of key parameters of the cavern steel lining, concrete lining and surrounding rock in a method for predicting fatigue life of a compressed air energy storage cavern according to the present invention;

[0056] Figure 6 This is a structural schematic diagram of a fatigue life prediction system for a compressed air energy storage chamber according to the present invention;

[0057] Figure 7 A schematic structural diagram of a training system for a pre-trained prediction model of key parameters of a gas storage cavern structure according to the present invention;

[0058] Figure 8Schematic diagram of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention. The embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0061] Aiming at the existing problems in the prior art, the fluctuation of new energy power generation causes grid instability, and the long-term operation of compressed air energy storage underground caverns is prone to fatigue damage, which will cause efficiency, safety and other problems. The present invention provides a method for predicting the fatigue life of compressed air energy storage caverns, such as Figure 1 Shown, including:

[0062] S100: pre-processing the compressed air storage cavity data to be analyzed to obtain a sample data set;

[0063] S200: Inputting the sample data set into a pre-trained prediction model of key parameters of the gas storage cavern structure, performing analysis, obtaining analysis results, and predicting the fatigue life of the cavern structure of the compressed air storage cavern based on the analysis results.

[0064] like Figure 2 As shown, a training method for the pre-trained prediction model of key parameters of the gas storage cavern structure includes:

[0065] S201: Obtain compressed air storage cavern data and perform preprocessing to obtain a key parameter data set;

[0066] S202: Using a CNN spatial feature extraction model to extract spatial features from the key parameter data set to obtain spatial feature data;

[0067] S203: Inputting the spatial feature data into the LSTM-based prediction model for training and processing to obtain a pre-trained prediction model of key parameters of the gas storage cavern structure.

[0068] This method can accurately predict the fatigue life of compressed air energy storage caverns, discover potential safety and efficiency issues in advance, effectively reduce the risk of fatigue damage in long-term operation of underground caverns, and ensure the stability of the power grid.

[0069] The present application is further explained below with reference to the accompanying drawings.

[0070] like Figure 3 As shown, a method for predicting fatigue life of a compressed air energy storage chamber includes:

[0071] S1: Obtain engineering geological data, design parameters and operating parameters of the gas storage cavern.

[0072] Specifically, obtain the engineering geological data, design parameters and operation parameters of the gas storage cavern, among which the engineering geological data of the gas storage cavern includes the rock density of the stratum where the compressed air energy storage cavern is located. , cohesion , internal friction angle , rock elastic modulus , Poisson's ratio of rock , where rock density Measured by wax seal method, cohesion , internal friction angle The rock elastic modulus is measured by triaxial compression test of rock. , Poisson's ratio of rock Measured by uniaxial compression test; design parameters include cavity volume , chamber surface area , chamber radius , steel lining thickness , steel lining density , steel lining elastic modulus , Poisson's ratio of steel lining , concrete lining thickness , concrete lining density , concrete lining elastic modulus , Poisson's ratio of concrete lining ; Operation parameters include the initial temperature in the cavern , initial air pressure in the cave , gas injection temperature , Gas injection time for one cycle , Storage time after gas injection , gas extraction time , Storage time after gas extraction , gas injection flow .

[0073] S2: Conduct numerical simulation of the cyclic charging and degassing of compressed air energy storage underground caverns to obtain the key parameters of the cavern steel lining, concrete lining and surrounding rock.

[0074] Among them, examples of engineering geology, design parameters and operating parameters for numerical simulation of cyclic charging and discharging of compressed air energy storage underground caverns are shown in Tables 1 to 3 respectively.

[0075] Table 1 Examples of engineering geological parameters for numerical simulation of cyclic charging and discharging of compressed air energy storage underground caverns

[0076]

[0077] Table 2 Example of cavern design parameters for numerical simulation of cyclic charging and discharging of compressed air energy storage underground caverns

[0078]

[0079] Table 3 Examples of cavern operation parameters for numerical simulation of cyclic charging and discharging of compressed air energy storage underground caverns

[0080]

[0081] The COMSOL Multiphysics software is used to carry out numerical simulation of the cyclic charging and discharging of compressed air energy storage underground caverns. Figure 4 As shown, Figure 4 In the figure, the vertical axis is the depth of the compressed air energy storage underground cavern, and the horizontal axis is the width on both sides of the center point of the cross section of the compressed air energy storage underground cavern. The maximum tensile strain of the steel lining in each cycle is obtained. and minimum tensile strain , maximum tensile stress of concrete lining and the first principal stress of the surrounding rock , the second principal stress and the third principal stress .

[0082] In COMSOL Multiphysics (a multi-physics simulation platform), the computational domain is numerically discretized using a discretized mesh model, and a local mesh refinement strategy is used to encrypt the cells in the area around the cave wall to accurately capture the stress concentration effect. The rock mechanics behavior adopts the Moore-Coulomb elastic-plastic constitutive model to simulate the yield, plastic deformation and shear failure characteristics of rock and soil under cyclic loading. The model applies a gravity field and the initial displacement is set to a zero-value boundary. The boundary conditions are set as follows: the upper boundary is free, the lower boundary is fixed, the left and right boundaries are roller-supported, and the initial air pressure in the cave is 0. , the load is applied by the air pressure boundary condition on the inner boundary of the steel lining; the initial value of the temperature field is A convection heat transfer boundary is set on the inner wall of the steel lining to simulate the gas-structure heat exchange. In COMSOL Multiphysics software, the ground stress balance is first performed, and then the cyclic filling and degassing simulation is performed for 30 cycles. The key parameters of the cavern steel lining, concrete lining and surrounding rock are obtained as follows: Figure 5 shown.

[0083] Therefore, the key parameters of the cavern steel lining, concrete lining and surrounding rock are the maximum tensile strain of the steel lining in each cycle and minimum tensile strain , maximum tensile stress of concrete lining , the first principal stress of surrounding rock , the second principal stress and the third principal stress .

[0084] S3: Process the acquired key parameters and use the CNN (Convolutional Neural Network) algorithm to extract spatial features to obtain the distribution pattern of the key parameters in each inflation and deflation cycle.

[0085] S31: CNN feature sequence extraction

[0086] The stress and strain values ​​in each cycle obtained by simulation are organized into a format suitable for CNN input. Assume that the simulated space area is divided into Grid points, where represents the number of grid points in the first spatial dimension; represents the number of grid points in the second spatial dimension; Represents the number of grid points in the third spatial dimension, and each grid point has a corresponding stress and strain value. , organize the key parameters of each cave structure into a three-dimensional matrix ,in, Indicates in Grid positions in cycles In order to meet the input requirements of CNN, the three-dimensional matrix is ​​further processed into a suitable tensor form to form a A tensor of , where is the number of channels, each channel corresponds to a physical quantity. , the input data can be expressed as:

[0087]

[0088] in, For the Input data for a cycle; It is In the cycle A three-dimensional matrix of physical quantities.

[0089] Construct a convolution layer and use the convolution kernel to slide on the input data to extract local features and capture the concentrated areas of stress and strain. The mathematical expression of the convolution layer is:

[0090]

[0091] in, Is the output feature map at the grid position ,aisle The value at is the input data; For the input feature map at grid position Department, channel The value of is the current position coordinate of the output feature map; is the offset inside the convolution kernel, The value range is 0 to , The value range is 0 to , The value range of is a positive integer; is a fixed channel index, indicating that the operation is only for the first channels; is the convolution kernel; is the convolution kernel at the grid position Department, channel The weight value of is the bias term; To add to the output channel The bias value is used to adjust the output value distribution.

[0092] Select the activation function. After the convolution operation, the activation function is used to introduce nonlinearity, alleviate the gradient vanishing problem, and speed up the training. The activation function used is , the expression is:

[0093]

[0094] in, is the activated feature value, that is, the convolution layer of the value passed to the next layer; is the original input value before the neuron is activated; Indicates taking 0 and The larger value among them, when the original input value before the neuron is activated When it is a positive number, the output is , when the original input value before the neuron is activated When it is negative or zero, the output is 0.

[0095] Construct a pooling layer and use the maximum pooling method to reduce the spatial size of the feature map, reduce the amount of calculation and retain important features. The mathematical expression of maximum pooling is:

[0096]

[0097] in, The feature map after pooling is at the grid position ,aisle The value at is the size of the pooling window; Is the output feature map at the grid position ,aisle The value at To traverse all positions within the pooling window, calculate the maximum value of the local area.

[0098] After multiple layers of convolution and pooling, the resulting feature maps contain the spatial characteristics of the input data. These feature maps serve as input to the LSTM (Long Short-Term Memory) recurrent neural network (RNN) to predict the state of the cavern structure under long-term inflation and deflation cycles.

[0099] S32: CNN training

[0100] The training process of CNN is divided into two stages: forward propagation and back propagation.

[0101] In the forward propagation stage, the input data passes through the convolution layer, activation function, pooling layer and other operations to finally obtain the output feature map. The mean square error loss function is used to calculate the loss between the CNN output features and the true features. The loss function expression is:

[0102]

[0103] in, is the loss value; is the sample size; It is True values; It is predicted values, .

[0104] In the backpropagation stage, the parameters of the CNN are updated according to the gradient of the loss function to minimize the prediction error. The values ​​of the convolution kernel and the bias term are adjusted by calculating the gradients of the output layer, the fully connected layer, the convolution layer, and the pooling layer. The specific steps and formulas are as follows:

[0105] Calculate the output layer gradient:

[0106]

[0107] Calculate the gradient of the fully connected layer:

[0108]

[0109]

[0110] in, is the weight matrix of the fully connected layer; is the activation output of the previous layer; Transpose the activation output of the previous layer; It is a full vector of is the partial differential symbol.

[0111] Calculate the convolutional layer gradient:

[0112] Assume that the input of the convolutional layer is , the output is , the convolution kernel is , the bias term is ,but:

[0113]

[0114]

[0115] in, Represents the convolution operation; is the gradient of the loss function with respect to the output of the convolutional layer; The output feature map is at position The gradient at is the convolutional layer gradient; is the bias gradient.

[0116] Calculate the pooling layer gradient:

[0117]

[0118] in, is an indicator function only if It is 1 if it is the maximum value, otherwise it is 0.

[0119] According to the calculated gradient, the gradient descent method is used to update the values ​​of the convolution kernel and bias terms:

[0120]

[0121] According to the convolutional layer gradient The information of the convolution kernel Adjust in the direction of the loss function downward until the loss converges to a satisfactory level, completing the convolution kernel Update iteration.

[0122]

[0123] According to the bias gradient The information will be biased Adjust in the direction of the loss function downward until the loss converges to a satisfactory level, completing the bias term Update iteration.

[0124] in, is the learning rate, which controls the step size of parameter updates; Represents an assignment operation.

[0125] Repeat the forward propagation and back propagation process until the loss function reaches a smaller value or no longer decreases significantly, and output the feature sequence.

[0126] S4: Input the spatial feature data into the prediction model of the key parameters of the gas storage cavern structure. Use the LSTM algorithm to train the extracted spatial feature data to obtain the change pattern of the key parameters with the number of filling and deflation times, and establish a prediction model for the key parameters of the gas storage cavern structure.

[0127] The LSTM algorithm trains the extracted spatial feature data, including the following steps:

[0128] The feature sequence extracted by CNN is used as the input data of LSTM. The input feature of each t time step can be the feature vector of a certain cycle extracted by CNN. At the same time, S2 is used to directly obtain the state parameters of the cave structure at the next t+1 time step, and prepare the corresponding target value, which is the true value.

[0129] According to the LSTM network, a prediction model based on LSTM is constructed, including input layer, LSTM layer, output layer, etc. Among them, the input layer receives the feature sequence extracted by CNN, including the maximum tensile strain of the steel lining in each cycle and minimum tensile strain , maximum tensile stress of concrete lining , the first principal stress of surrounding rock , the second principal stress and the third principal stress Spatial characteristics of equal parameters.

[0130] The LSTM layer receives the feature sequence from the input layer and models the time series data through its internal gating mechanism to capture the changing trends of key parameters in multiple cycles.

[0131] The output layer converts the output of the LSTM layer into specific prediction values.

[0132] Initialize the weight and bias parameters of the LSTM network using the Glorot initialization method. The specific formula is:

[0133] Uniform distribution initialization:

[0134] Normal distribution initialization:

[0135] Among them, U is uniform distribution; N is normal distribution; is the number of input neurons; is the number of output neurons.

[0136] The LSTM training process is divided into two stages: forward propagation and backpropagation.

[0137] In the forward propagation phase, at each time step, the input feature vector is passed into the LSTM unit, and the activation values ​​of the forget gate, input gate, and output gate are calculated, as well as the updates of the cell state and hidden state. The specific formula is as follows:

[0138] Forget Gate:

[0139]

[0140] Where, is the output of the forget gate; is the weight matrix of the forget gate; is the bias term of the forget gate; is the sigmoid function; is the hidden state at the current moment; is the hidden state of the previous time step; is the input vector for the current time step.

[0141] Input Gate:

[0142]

[0143] Where, is the output of the input gate; is the weight matrix of the input gate; is the bias term of the input gate.

[0144]

[0145] Where, is the candidate cell state; is the cell state weight matrix; is the cell state bias term.

[0146] Cell status update:

[0147]

[0148] Where, is the cell state at the current moment; is the candidate cell state at the previous time step.

[0149] Output Gate:

[0150]

[0151] Where, is the output of the output gate; is the weight matrix of the output gate; is the bias term of the output gate.

[0152]

[0153] Where, is the hidden state at the current moment, ⊙ represents element-by-element multiplication; is the activation function; ( ) for the general Each element in is compressed to the range [-1, 1].

[0154] The hidden state at the current moment After being processed by the output layer, it is converted into the predicted cavern structure state parameters.

[0155] At each time step, the predicted cavern structure state parameters are compared with the true values ​​and the loss function is calculated.

[0156] In the backpropagation phase, the parameters of the LSTM network are updated according to the gradient of the loss function to minimize the prediction error. Backpropagation uses the BPTT (Backpropagation Through Time) algorithm to calculate the gradient of the loss function with respect to each parameter at each time step. The specific process is as follows:

[0157] Calculate the hidden state gradient:

[0158]

[0159] in, is the loss value at the current moment; is the hidden state at the next time step.

[0160] Calculate the cell state gradient:

[0161]

[0162] in, is the candidate cell state for the next time step; is the output of the forget gate for the next time step; is the derivative of tanh; For The derivative is calculated independently for each element of ; is the multiplication operation in the chain rule.

[0163] when hour, ,at this time The gradient disappears.

[0164] when hour, ,at this time The gradient is fully transferred.

[0165] Compute the gated gradient:

[0166] Output gradient of the forget gate:

[0167] Output gradient of the input gate:

[0168] Candidate cell state gradients:

[0169] Calculate weight and bias gradients:

[0170] The weight matrix gradient of the forget gate:

[0171] The bias gradient of the forget gate:

[0172] Gradient of the weight matrix of the input gate:

[0173] The bias gradient of the input gate:

[0174] Cell state weight matrix gradient:

[0175] Cell state bias gradient:

[0176] The gradient of the weight matrix of the output gate:

[0177] The bias gradient of the output gate:

[0178] According to the calculated gradient, the parameters of the LSTM network are updated using the gradient descent method:

[0179]

[0180] According to the weight matrix gradient of the forget gate The information of the forget gate weight matrix Adjust in the direction of the loss function until the loss converges to a satisfactory level and complete the weight matrix of the forget gate Update iteration.

[0181]

[0182] According to the weight matrix gradient of the forget gate The information of the forget gate will be Adjust in the direction of the loss function until the loss converges to a satisfactory level, completing the bias term of the forget gate Update iteration.

[0183]

[0184] According to the weight matrix gradient of the input gate The information of the input gate weight matrix Adjust in the direction of the loss function until the loss converges to a satisfactory level and complete the weight matrix of the input gate Update iteration.

[0185]

[0186] According to the bias gradient of the input gate The information will be input into the bias term of the gate Adjust in the direction of the loss function until the loss converges to a satisfactory level, completing the bias term of the input gate Update iteration.

[0187]

[0188] According to the cell state weight matrix gradient The cell state weight matrix Adjust in the direction of the loss function until the loss converges to a satisfactory level, completing the cell state weight matrix Update iteration.

[0189]

[0190] According to the cell state bias gradient The information of the cell state bias Adjust in the direction of the loss function until the loss converges to a satisfactory level, completing the cell state bias term Update iteration.

[0191]

[0192] According to the weight matrix gradient of the output gate The information of the output gate weight matrix Adjust in the direction of the loss function until the loss converges to a satisfactory level and complete the weight matrix of the output gate Update iteration.

[0193]

[0194] According to the bias gradient of the output gate The information of the output gate will be biased Adjust in the direction of the loss function until the loss converges to a satisfactory level, completing the bias term for the output gate Update iteration.

[0195] 70% of the simulated data is used for training and 30% is used for testing. The forward propagation and backpropagation processes are repeated until the model converges.

[0196] S5: Use the trained model to predict key parameters and predict the fatigue life of the cavern structure based on fatigue criteria;

[0197] The method for predicting the fatigue life of cavern structures based on fatigue criteria is:

[0198] (1) Steel lining fatigue criterion:

[0199]

[0200] Where, is the maximum tensile stress of the steel lining in each stress cycle; is the minimum tensile stress of the steel lining in each stress cycle, To calculate the allowable strain, the following formula can be used:

[0201]

[0202] Where, The elongation at break of the material; is the destructive strength; is the elastic modulus; is the fatigue life, that is, the number of cycles to failure.

[0203] (2) Criteria for concrete lining fatigue:

[0204]

[0205] Where, is the maximum tensile stress of the concrete lining, is the tensile strength of concrete.

[0206] (3) Criteria for surrounding rock fatigue:

[0207]

[0208] Where, is the first principal stress of the surrounding rock; The second principal stress of the surrounding rock; is the third principal stress of the surrounding rock; is the tensile strength of rock.

[0209] When predicting the life of a cavern structure, if one of the above three criteria is met, the cavern is considered to be damaged. The corresponding number of cycles n is the maximum number of cycles of the cavern. The calculation formula for the operating life of the cavern is:

[0210]

[0211] In one cycle, is the gas injection time, is the storage time after gas injection, is the gas sampling time, is the storage time after gas extraction; The operating life of the cavern.

[0212] S6: Optimize the cavern structure parameters based on the prediction results and the engineering design service life until the predicted service life meets the design requirements.

[0213] The second purpose of the present invention is to propose a system for predicting the fatigue life of compressed air energy storage chambers. Figure 6 Shown, including:

[0214] The data set acquisition module 100 is used to pre-process the compressed air storage cavity data to be analyzed to obtain a sample data set;

[0215] Analysis result module 200: used to input the sample data set into the pre-trained prediction model of key parameters of the air storage cavern structure, perform analysis, obtain analysis results, and predict the fatigue life of the cavern structure of the compressed air storage cavern based on the analysis results.

[0216] like Figure 7 As shown, a training system for a pre-trained prediction model of key parameters of a gas storage cavern structure includes:

[0217] Key parameter data module 201: used to obtain compressed air storage cavern data and perform preprocessing to obtain a key parameter data set;

[0218] Spatial feature data module 202: used to extract spatial features from key parameter data sets using a CNN spatial feature extraction model to obtain spatial feature data;

[0219] The training prediction model module 203 is used to input the spatial feature data into the LSTM-based prediction model for training and processing to obtain a pre-trained prediction model of key parameters of the gas storage cavern structure.

[0220] like Figure 8 As shown, a third object of the present invention is to provide an electronic device, comprising: a processor 301, a memory 302, and a display screen 303. The memory 302 and the display screen 303 are both connected to the processor 301, for example, via a bus 304. Optionally, the electronic device may further include a transceiver 305. It should be noted that in actual applications, the number of transceivers 305 is not limited to one, and the structure of the electronic device does not constitute a limitation on the embodiments of the present application.

[0221] Processor 301 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0222] Bus 304 may include a path for transmitting information between the aforementioned components. Bus 304 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, for example. Bus 304 may be divided into an address bus, a data bus, a control bus, and other components.

[0223] The memory 302 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0224] The memory 302 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 302 to implement the content shown in the above method embodiment.

[0225] Figure 8 The electronic device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0226] The fourth object of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program is stored thereon, and when the program is executed by a processor, the computer program is realized as described above. Figures 1 to 3 The various processes of the illustrated method embodiment include, for example, a memory including instructions, and the instructions can be executed by a processor of an electronic device to perform the above method.

[0227] A computer-readable storage medium may be a tangible device that holds and stores instructions for use by an instruction execution device. A computer-readable storage medium may be, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof. Specifically, a computer-readable storage medium may be a portable computer disk, a hard disk, a USB flash drive, a random access memory, a read-only memory, an erasable programmable read-only memory, a podium random access memory, a portable compact disc read-only memory, a digital versatile disk, a memory stick, a floppy disk, an optical disc, a magnetic disk, a mechanical encoding device, or any combination thereof.

[0228] The fifth object of the present invention is to provide a computer program product comprising computer instructions, which, when executed by a processor, implement the above Figures 1 to 3The various processes of the method embodiment shown can achieve the same technical effect, and to avoid repetition, they will not be described here.

[0229] Many embodiments and applications beyond the examples provided will be apparent to those skilled in the art upon reading the foregoing description. Therefore, the scope of the present teachings should be determined not with reference to the foregoing description, but rather with reference to the preceding claims, along with the full scope of equivalents to which such claims are entitled. For the purpose of completeness, all articles and references, including the disclosures of patent applications and publications, are incorporated herein by reference. The omission of any aspect of the subject matter disclosed herein from the preceding claims is not a disclaimer of such subject matter, nor should it be interpreted that the applicants did not consider such subject matter to be part of the disclosed inventive subject matter.

[0230] The above content is a further detailed description of the present invention, and it cannot be considered that the specific implementation methods of the present invention are limited to these. For ordinary technicians in the technical field to which the present invention belongs, they can make several simple deductions or substitutions without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. A method for predicting fatigue life of a compressed air energy storage chamber, characterized in that: include: Preprocess the compressed air storage cavity data to be analyzed to obtain a sample data set; The sample data set is input into a pre-trained prediction model of key parameters of the gas storage cavern structure, and the analysis is performed to obtain analysis results. The fatigue life of the compressed air storage cavern structure is predicted based on the analysis results; The training method of the pre-trained prediction model of key parameters of the gas storage cavern structure includes: Obtain compressed air storage cavern data and preprocess it to obtain key parameter data sets; Use the CNN spatial feature extraction model to extract spatial features from the key parameter data set to obtain spatial feature data; The spatial feature data is input into the LSTM-based prediction model for training and processing to obtain a pre-trained prediction model for the key parameters of the gas storage cavern structure. The compressed air storage cavern data is obtained and pre-processed to obtain a key parameter data set, including: Based on the engineering geological data, design parameters, and operating parameters of the compressed air storage caverns in the data, a numerical simulation method was used to simulate the filling and degassing process of the compressed air energy storage underground caverns, and the key parameters of the cavern steel lining, concrete lining, and surrounding rock were obtained. The key parameters of the cavern steel lining, concrete lining and surrounding rock include the maximum tensile strain of the steel lining , minimum tensile strain of steel lining , maximum tensile stress of concrete lining , the first principal stress of surrounding rock , the second principal stress of surrounding rock and the third principal stress of surrounding rock ; The CNN spatial feature extraction model is used to extract spatial features from the key parameter data set to obtain spatial feature data, including: The CNN spatial feature extraction model is used to process the key parameter data set layer by layer to extract feature maps; The mean square error loss function is used to calculate the loss function between the output features and the true features in the feature map ; in, is the loss value; is the sample size; It is True values; It is predicted values, ; The parameters of the CNN are updated according to the gradient of the loss function. The CNN is optimized through the back-propagation algorithm to adjust the convolution kernel and weights. After that, the loss function is recalculated until the loss value L of the loss function meets the set conditions, and the feature sequence is output to obtain the spatial feature data. The LSTM prediction model captures the temporal variation trend of key parameters in the charging and discharging cycle through the BPTT algorithm; based on the steel lining fatigue criterion , concrete lining fatigue criterion and surrounding rock fatigue criterion Analyze the prediction results to determine the fatigue life of the cavern; Where, To allow for strain; is the tensile strength of concrete; is the tensile strength of rock.

2. A method for predicting fatigue life of a compressed air energy storage chamber according to claim 1, characterized in that: The spatial feature data is input into the LSTM-based prediction model for training to obtain a pre-trained prediction model of key parameters of the gas storage cavern structure, including: Input the training data set in the spatial feature data into the LSTM-based prediction model for training to obtain the training results and the trained LSTM-based prediction model; Input the test data set in the spatial feature data into the trained LSTM-based prediction model to obtain the test results; The training results are compared with the test results to determine whether the LSTM-based prediction model has converged. If so, the pre-trained prediction model of the key parameters of the gas storage cavern structure is obtained.

3. A method for predicting fatigue life of a compressed air energy storage chamber according to claim 2, characterized in that: The training data set in the spatial feature data is input into the LSTM-based prediction model for training to obtain training results, including: Build an LSTM-based prediction model based on the LSTM network, and set the input layer, LSTM layer and output layer in the LSTM-based prediction model; The input layer receives the training data set in the spatial feature data, extracts the feature sequence in the training data set in the spatial feature data, and sends the extracted feature sequence in the training data set to the LSTM layer; The LSTM layer receives the feature sequence in the training data set from the input layer and trains the LSTM layer. After training, it captures the changing trend of key parameters in several cycles and sends the changing trend to the output layer. The output layer receives the change trend from the LSTM layer and generates training results.

4. A method for predicting fatigue life of a compressed air energy storage chamber according to claim 3, characterized in that: The LSTM layer receives the feature sequence in the training set from the input layer and trains the LSTM layer, including: The LSTM layer receives the feature sequence in the training set from the input layer, calculates the output of the feature sequence in the training set at each time step in order of time steps, obtains the predicted cavern structure state parameters, compares the predicted cavern structure state parameters with the true values, and calculates the loss function; The LSTM layer is updated according to the gradient of the loss function. The gradient of the loss function in each time step is reversely calculated using the BPTT algorithm. The parameters of the LSTM network are updated using the gradient descent method until the parameters of the LSTM network meet the set conditions, completing the training of the LSTM layer.

5. A method for predicting fatigue life of a compressed air energy storage chamber according to claim 1, characterized in that: The sample data set is input into a pre-trained prediction model of key parameters of the gas storage cavern structure, analyzed, and analysis results are obtained. The fatigue life of the compressed air storage cavern structure is predicted based on the analysis results, including: The sample data set is input into the trained prediction model of key parameters of the gas storage cavern structure to obtain the prediction results; The prediction results are analyzed according to fatigue criteria to obtain the fatigue life prediction results of compressed air energy storage caverns; The cavern structural parameters are optimized based on the fatigue life prediction results of the compressed air energy storage cavern and the engineering design service life.

6. A system for predicting fatigue life of a compressed air energy storage chamber, characterized in that: include: Dataset acquisition module: used to pre-process the compressed air storage cavern data to be analyzed to obtain a sample data set; Analysis result module: used to input the sample data set into the pre-trained prediction model of key parameters of the gas storage cavern structure, perform analysis, obtain analysis results, and predict the fatigue life of the compressed air storage cavern structure based on the analysis results; The training method of the pre-trained prediction model of key parameters of the gas storage cavern structure includes: Obtain compressed air storage cavern data and preprocess it to obtain key parameter data sets; Use the CNN spatial feature extraction model to extract spatial features from the key parameter data set to obtain spatial feature data; The spatial feature data is input into the LSTM-based prediction model for training and processing to obtain a pre-trained prediction model for the key parameters of the gas storage cavern structure. The compressed air storage cavern data is obtained and pre-processed to obtain a key parameter data set, including: Based on the engineering geological data, design parameters, and operating parameters of the compressed air storage caverns in the data, a numerical simulation method was used to simulate the filling and degassing process of the compressed air energy storage underground caverns, and the key parameters of the cavern steel lining, concrete lining, and surrounding rock were obtained. The key parameters of the cavern steel lining, concrete lining and surrounding rock include the maximum tensile strain of the steel lining , minimum tensile strain of steel lining , maximum tensile stress of concrete lining , the first principal stress of surrounding rock , the second principal stress of surrounding rock and the third principal stress of surrounding rock ; The CNN spatial feature extraction model is used to extract spatial features from the key parameter data set to obtain spatial feature data, including: The CNN spatial feature extraction model is used to process the key parameter data set layer by layer to extract feature maps; The mean square error loss function is used to calculate the loss function between the output features and the true features in the feature map ; in, is the loss value; is the sample size; It is True values; It is predicted values, ; The parameters of the CNN are updated according to the gradient of the loss function. The CNN is optimized through the back-propagation algorithm to adjust the convolution kernel and weights. After that, the loss function is recalculated until the loss value L of the loss function meets the set conditions, and the feature sequence is output to obtain the spatial feature data. The LSTM prediction model captures the temporal variation trend of key parameters in the charging and discharging cycle through the BPTT algorithm; based on the steel lining fatigue criterion , concrete lining fatigue criterion and surrounding rock fatigue criterion Analyze the prediction results to determine the fatigue life of the cavern; Where, To allow for strain; is the tensile strength of concrete; is the tensile strength of rock.

7. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for predicting fatigue life of a compressed air energy storage chamber as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method for predicting fatigue life of a compressed air energy storage chamber as described in any one of claims 1 to 5.

Images