Gas storage salt cavern performance identification method and system based on multi-field coupling

Abstract

This invention belongs to the field of compressed air energy storage technology and provides a method and system for identifying the performance of gas storage salt caverns based on multi-field coupling. A dynamic graph is constructed using salt caverns, wellbores, surrounding rock, and surface process equipment as nodes, realizing dynamic characterization of the system's operating status. Based on this, a multi-source anomaly evidence identification mechanism is formed by integrating mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models using multi-source data. Furthermore, by collaboratively judging anomaly types and confidence levels using physical and data residuals, a sealing integrity index, a salt cavern stability index, and an abnormal operating condition index are constructed, achieving multi-dimensional quantitative expression of safety status. A comprehensive safety index is established by combining adaptive weights for the operating stage, realizing system safety level classification. This enables continuous monitoring, intelligent identification, and graded handling of the salt cavern gas storage system, significantly improving system operational safety.

Description

Technical Field

[0001] This invention belongs to the field of compressed air energy storage technology, and particularly relates to a method and system for identifying the performance of gas storage salt caverns based on multi-field coupling. Background Technology

[0002] In the field of compressed air energy storage power stations, salt caverns have become an important underground gas storage medium for large-scale compressed air energy storage due to their large volume, good sealing performance, high operating pressure, and relatively low construction cost. However, during long-term operation, salt caverns are affected by periodic gas injection, storage, release, and shutdown conditions. The pressure, temperature, humidity, stress state, and chemical environment within the salt cavern continuously change, which can easily lead to problems such as salt rock creep, cavity shrinkage, thermal stress damage, microcrack propagation, leakage of caprock or interlayer, and wellbore seal failure, resulting in gas leakage or reduced pressure bearing capacity of the salt cavern. Therefore, evaluating the stability of salt caverns is particularly important.

[0003] Existing methods for evaluating the stability of salt caverns are mostly based on single pressure boundaries, static mechanical analysis, or periodic testing results. These methods are insufficient to fully reflect the multi-field coupling effects of high-frequency pressure cycling, rapid pressure rise and fall, large temperature difference cycling, and long-term multi-cycle operation in compressed air energy storage power stations. In addition, complex operating conditions such as wellbore fluid accumulation, brine backflow, salt dust entrainment, corrosion blockage, and annular pressure anomalies may occur in actual operation. These anomalies are often coupled with the sealing performance and structural stability of the salt cavern, and traditional threshold alarm methods are unable to identify their evolution trends in a timely and accurate manner. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a method and system for identifying the performance of gas storage salt caverns based on multi-field coupling. This invention constructs a dynamic graph using salt caverns, wellbore, surrounding rock, and surface process equipment as nodes, achieving dynamic characterization of the system's operating status. Based on this, it integrates mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models using multi-source data to form a multi-source anomaly evidence identification mechanism. Furthermore, by collaboratively judging anomaly types and confidence levels using physical and data residuals, it constructs sealing integrity indices, salt cavern stability indices, and abnormal operating condition indices, achieving multi-dimensional quantitative expression of safety status. Combined with adaptive weights for operational stages, a comprehensive safety index is established, enabling system safety level classification. This allows for continuous monitoring, intelligent identification, and tiered handling of the salt cavern gas storage system, significantly improving system operational safety and risk control capabilities.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: In a first aspect, the present invention provides a method for identifying the performance of gas storage salt caverns based on multi-field coupling, comprising: Acquire multi-source historical data of the salt cavern gas storage system; construct a dynamic map with salt caverns, wellbores, surrounding rock, and surface process equipment as nodes; Based on multi-source historical data, and pre-set mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models, the physical model residuals and data-driven model residuals are determined; and anomaly evidence is determined based on the physical model residuals and data-driven model residuals. Abnormal evidence is transformed into a seal integrity index, a salt cavern stability index, and an abnormal operating condition index. A comprehensive safety index is determined by weighted summation based on the seal integrity index, salt cavern stability index, and abnormal operating condition index, and the weights are adaptively adjusted according to the operating stage of the salt cavern gas storage system. Anomaly types are identified and attributed based on the comprehensive safety index.

[0006] Furthermore, the multi-source historical data includes operational condition data, salt cavern body monitoring data, sealing status data, fluid medium data, and geological engineering static data; the dynamic graph includes nodes, edge sets, and a dynamic adjacency matrix; the nodes are the salt cavern gas cavity, salt cavern top plate, salt cavern sidewall, salt cavern bottom plate, caprock, interlayer, wellbore, casing, cement sheath, packer, and surface process equipment; the edge set is used to describe the pressure transmission, heat conduction, seepage, stress transmission, damage propagation, corrosion blockage propagation, and operational control coupling relationships between different nodes; the dynamic adjacency matrix is ​​updated based on pressure cycle amplitude, temperature cycle amplitude, pressure rise and fall rate, annular pressure, salt dust concentration, filter differential pressure, microseismic activity, and sealing status data to characterize the dynamic evolution of risk propagation relationships within the salt cavern gas storage system.

[0007] Furthermore, the mass balance model is as follows: ; in, The mass of gas inside the salt cavern; This refers to the mass flow rate of the gas injected into the salt cavern. This refers to the mass flow rate of the extracted gas. This represents the mass flow rate of the leaked gas.

[0008] Furthermore, the thermo-mechanical coupling model is used to determine the thermal stress and damage accumulation under the combined effects of pressure cycling and temperature cycling: ； ; in, Thermal stress; The elastic modulus of the material; The coefficient of linear expansion; This represents the amplitude of the temperature cycle. For the first Time-dependent damage variables; For the first Damage variables at any time The damage increment function, This represents the amplitude of the pressure cycle. This represents the amplitude of the temperature cycle. For the rate of increase or decrease in voltage, For the rate of temperature change, This represents the cumulative number of loops.

[0009] Furthermore, the creep closure model is used to predict changes in the effective capacity of salt caverns over long-term operation: ; in, For the first Time or number The effective storage capacity of the salt cavern per operating cycle; For the first Time or number The effective storage capacity of the salt cavern per operating cycle; This represents the volumetric shrinkage caused by salt rock creep. This represents the change in volume.

[0010] Furthermore, the corrosion clogging model is used to identify operational anomalies caused by water, brine, salt dust, and corrosion products, and a corrosion clogging index is defined: ; in, For a moment The dimensionless index corresponding to the moisture content or humidity; For a moment The dimensionless index corresponding to the chloride ion concentration; For a moment The dimensionless index corresponding to the salt dust concentration; For a moment The dimensionless index corresponding to the pressure difference of the filter or pipeline; For a moment The dimensionless index corresponding to the corrosion rate; to These are the weighting coefficients for the corresponding indicators.

[0011] Furthermore, if both the physical model residuals and the data-driven model residuals are abnormal, it is determined to be high-confidence evidence of anomalies; if only the data-driven model residuals are abnormal, it is determined to be evidence of unknown patterns; if only the physical model residuals are abnormal, it is determined to be evidence of boundary conditions or insufficient historical samples. The abnormal evidence includes evidence of sealing anomalies, evidence of structural stability anomalies, and evidence of operational condition anomalies. The sealing integrity index is used to evaluate the overall sealing status of the salt cavern, wellbore, cement sheath, packer, caprock, and fracture system. The salt cavern stability index is used to evaluate the structural stability of the salt cavern cavity, surrounding rock, and under long-term operating conditions. The abnormal operating condition index is used to evaluate the degree of anomalies in water content, brine, salt dust, corrosion, blockage, and flow response.

[0012] Furthermore, the adaptive adjustment of the weights includes: increasing the weights of the sealing integrity index and operating boundary margin during the gas injection stage; increasing the weights of the corresponding items of the salt cavern stability index and abnormal operating condition index during the gas extraction stage; increasing the weights of the sealing integrity index and salt cavern stability index during the gas storage stage; and increasing the weights of the corresponding items of the salt cavern stability index and abnormal operating condition index during the peak shaving stage.

[0013] Furthermore, the attribution identification includes: when the overall safety index decreases mainly due to a decrease in the seal integrity index, and is accompanied by continuous pressure drop, abnormal mass balance, abnormal increase in annular pressure, abnormal gas composition, or abnormal integrity of the cement ring and packer, it is determined to be a seal integrity risk; when the overall safety index decreases mainly due to a decrease in the salt cavern stability index, and is accompanied by a lower minimum pressure, increased pressure drop rate, increased reservoir shrinkage rate, increased micro-vibration events, and increased risk at the top plate or sidewall nodes, it is determined to be a salt cavern structural stability risk; when the overall safety index decreases mainly due to an increase in the abnormal operating condition index, and is accompanied by increased moisture content, abnormal brine backflow, increased salt dust concentration, increased pressure differential of filters or pipelines, sluggish valve operation, or increased corrosion rate, it is determined to be an abnormal operating condition risk; when the overall safety index decreases mainly due to a decrease in the operating boundary margin, and is accompanied by an increase in the cumulative number of cycles and a pressure increase or decrease rate exceeding the dynamic safety boundary, it is determined to be an operating boundary risk.

[0014] Secondly, the present invention also provides a gas storage salt cavern performance identification system based on multi-field coupling, comprising: The data acquisition module is configured to: acquire multi-source historical data of the salt cavern gas storage system; and construct a dynamic map with salt caverns, wellbores, surrounding rock, and surface process equipment as nodes. The anomalous evidence determination module is configured to: determine the physical model residuals and data-driven model residuals based on multi-source historical data and preset mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models; and determine anomalous evidence based on the physical model residuals and data-driven model residuals. The safety index determination module is configured to: convert abnormal evidence into a sealing integrity index, a salt cavern stability index, and an abnormal operating condition index; determine a comprehensive safety index by weighted summation based on the sealing integrity index, salt cavern stability index, and abnormal operating condition index; and adaptively adjust the weights according to the operating stage of the salt cavern gas storage system. The identification module is configured to: identify the cause of anomalies based on the comprehensive security index.

[0015] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the gas storage salt cavern performance identification method based on multi-field coupling described in the first aspect.

[0016] Fourthly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the program to implement the steps of the gas storage salt cavern performance identification method based on multi-field coupling described in the first aspect.

[0017] Fifthly, the present invention also provides a computer program product, the computer program product comprising a computer program, which, when executed by a processor, implements the steps of the gas storage salt cavern performance identification method based on multi-field coupling described in the first aspect.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention constructs a dynamic diagram using salt caverns, wellbores, surrounding rock, and surface process equipment as nodes, realizing a dynamic representation of the system's operating status. Based on this, it integrates a mass balance model, a thermo-mechanical coupling model, a creep closure model, a corrosion blockage model, and a data-driven prediction model using multi-source data to form a multi-source anomaly evidence identification mechanism. Furthermore, it constructs a sealing integrity index, a salt cavern stability index, and an abnormal operating condition index by collaboratively judging anomaly types and confidence levels using physical residuals and data residuals, realizing a multi-dimensional quantitative expression of safety status. Combined with adaptive weights for the operating stage, it establishes a comprehensive safety index, realizing the classification of system safety levels. This enables continuous monitoring, intelligent identification, and graded handling of the salt cavern gas storage system, significantly improving the system's operational safety and risk prevention capabilities. Attached Figure Description

[0019] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0020] Figure 1 This is a flowchart of the method in Embodiment 1 of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0023] Example 1: Compressed air energy storage power stations utilize off-peak electricity to compress air and store it in underground spaces. During peak electricity demand periods, they release high-pressure air to drive power generation, offering advantages such as large scale, long lifespan, and strong regulation capabilities. Salt caverns, due to their large volume, good sealing, high operating pressure, and relatively low construction costs, have become an important underground gas storage medium for large-scale compressed air energy storage. However, existing salt cavern compressed air energy storage operations show that, during long-term operation, these storage caverns are affected by periodic gas injection, storage, release, and shutdown conditions. The pressure, temperature, humidity, stress state, and chemical environment within the salt cavern continuously change, easily leading to problems such as salt rock creep, cavity shrinkage, thermal stress damage, microcrack propagation, caprock or interlayer leakage, and wellbore seal failure, resulting in gas leakage or reduced pressure bearing capacity of the salt cavern.

[0024] As described in the technical documentation, existing methods for evaluating the stability of salt caverns are mostly based on single pressure boundaries, static mechanical analysis, or periodic testing results. These methods are insufficient to fully reflect the multi-field coupling effects under high-frequency pressure cycling, rapid pressure rise and fall, large temperature difference cycling, and long-term multi-cycle operation of compressed air energy storage power stations. Furthermore, in actual operation, complex conditions such as wellbore fluid accumulation, brine backflow, salt dust entrainment, corrosion blockage, and annular pressure anomalies may occur. These anomalies are often coupled with the sealing performance and structural stability of the salt cavern, making it difficult for traditional threshold alarm methods to identify their evolution trends in a timely and accurate manner.

[0025] To solve at least one of the above problems, such as Figure 1 As shown, this embodiment provides a method for identifying the performance of gas storage salt caverns based on multi-field coupling, including: S1. Construction of a multi-source state characterization model for salt cavern gas storage systems: First, a multi-source historical database for the salt cavern gas storage system is established. This database includes operational data, salt cavern body monitoring data, sealing status data, fluid medium data, and geological engineering static data.

[0026] Among them, the operating condition data is used to describe the pressure, temperature, flow rate and circulation characteristics of the salt cavern during the gas injection, gas storage, gas production and standby stages; the salt cavern body monitoring data is used to describe the morphology of the salt cavern cavity, changes in storage capacity, deformation of the surrounding rock and microseismic activity; the sealing status data is used to describe the integrity of the wellbore, casing, cement sheath, packer, caprock and fracture system; the fluid medium data is used to describe anomalies in the operating medium such as water, brine, salt dust, corrosion and blockage; and the geological engineering static data is used to describe the salt layer structure, rock mechanical properties, thermophysical parameters and wellbore structural parameters.

[0027] After time synchronization, outlier removal, missing value imputation, and standardization, the above data forms a unified salt cave multi-field coupled state vector: ; in, For the first Time or number The state vector of the salt cavern gas storage system during its operating cycle; This is a pressure field state; This refers to the temperature field state; The state of water, humidity, and brine; The state is a mechanical field. The condition is characterized by chemical corrosion and salt dust blockage. The condition is damaged. This refers to the geometric shape of the salt cavern.

[0028] The further state vector can be represented as: ; in, This refers to the internal pressure of the salt cavern. This refers to the wellbore pressure. For annular pressure; This represents the amplitude of the pressure cycle. The rate of increase or decrease in voltage; The temperature of the gas in the salt cavern; This represents the amplitude of the temperature cycle. The rate of temperature change; This refers to the gas injection flow rate; This refers to the gas extraction flow rate; This provides the effective storage capacity of the salt caverns. This represents the stability coefficient of salt cavern morphology. Energy of microseismic events; Salt dust concentration; For corrosion rate; For damage variables.

[0029] Based on this, the salt cavern gas chamber, salt cavern top plate, salt cavern sidewall, salt cavern bottom plate, caprock, interlayer or fracture, wellbore, casing, cement sheath, packer, and surface process equipment are abstracted into a multi-node dynamic diagram: ; in, For the first Time or number Dynamic diagram of multiple nodes in the operation cycle, from salt cavern to wellbore to surrounding rock and surface equipment; A set of nodes; Let it be the set of edges; This is a dynamic adjacency matrix. The edge set... This is used to describe the coupling relationships between different nodes, including pressure transmission, heat conduction, seepage, stress transmission, damage propagation, corrosion and blockage propagation, and operational control. Dynamic adjacency matrix. The data is updated based on pressure cycle amplitude, temperature cycle amplitude, pressure rise and fall rate, annular pressure, salt dust concentration, filter differential pressure, micro-vibration activity, and sealing status to characterize the dynamic evolution of risk propagation relationships within the salt cavern gas storage system.

[0030] Therefore, step S1 partially outputs a unified state vector. and animated GIFs This serves as the foundational input for subsequent anomaly identification, index calculation, and risk classification.

[0031] S2. Anomaly evidence identification based on the fusion of physical models and data-driven approaches: Obtaining the state vector and animated GIFs Subsequently, an anomaly identification method integrating physical and data-driven models is constructed to generate basic evidence for sealing anomalies, structural stability anomalies, and operational condition anomalies.

[0032] First, a multi-field coupled physical prediction model is established to calculate the theoretical response of the salt cavern under normal operating conditions. This physical prediction model includes a mass balance model, a thermo-mechanical coupling model, a creep closure model, and a corrosion blockage model.

[0033] The mass balance model is expressed as: ; in, The mass of gas inside the salt cavern; This refers to the mass flow rate of the gas injected into the salt cavern. This refers to the mass flow rate of the extracted gas. This represents the mass flow rate of the leaked gas.

[0034] When a sustained pressure drop occurs during the gas storage or pressure stabilization phase, the equivalent leakage rate after temperature correction is calculated based on changes in pressure, temperature, and storage capacity. Specifically, at two adjacent sampling times and First, the pressure, temperature, and effective volume are converted into the mass of gas inside the salt cavern using the compressed air state equation: ; in, and They are respectively Time and The compressibility factor of constantly compressed air. air is the gas constant. This represents the time interval between adjacent sampling moments. For the mass balance equation... Integrate over the time interval and use the average of the injection mass flow rate and the sampling mass flow rate within that interval to obtain the equivalent leakage mass flow rate after temperature, effective volume, and gas compressibility correction: ; in, The equivalent leakage mass flow rate after temperature and effective volume correction; and They are respectively Average injection mass flow rate and average extraction mass flow rate within the time interval; and These are calculated using the aforementioned equations of state. This calculation first converts changes in pressure, temperature, and effective volume into changes in gas mass, then deducts the actual injection and extraction volumes based on the law of mass conservation. Therefore, it can distinguish between the apparent pressure drop caused by temperature fluctuations, storage capacity changes, and gas compressibility, and the mass loss caused by actual leakage. During the gas storage or pressure stabilization phase, when both the injection and extraction mass flow rates are approximately zero, the above equation simplifies to: ; Specifically, when the calculation result is less than 0, it is taken as 0. Compare with leakage thresholds determined from historical normal gas storage data or airtightness tests; when Evidence of sealing abnormality is generated when multiple consecutive sampling cycles exceed the leakage threshold, or when the difference between the theoretical pressure drop calculated based on the leakage-free mass conservation relationship and the measured pressure drop continuously exceeds the pressure residual threshold.

[0035] The thermo-mechanical coupling model is used to calculate the thermal stress and damage accumulation under the combined effects of pressure cycling and temperature cycling. Its thermal stress is expressed as: ; in, Thermal stress; The elastic modulus of the material; The coefficient of linear expansion; This represents the amplitude of the temperature cycle.

[0036] To ensure that the damage variable remains consistent with the subsequent safety status index in the range of 0 to 1, in one specific implementation, the damage increment at the current moment or during the current operating cycle is recorded as... And the damage variable is capped and updated according to the following formula: ; in, For the first The damage variable at any given time ranges from 0 to 1. To clarify the effects of pressure cycling, temperature cycling, pressure increase / decrease rates, temperature change rates, and cumulative cycle count on the damage increment, this embodiment defines the damage increment function as follows: ; in, to It is a non-negative damage weight, and satisfies Each input quantity is first processed to be dimensionless based on the corresponding design value or calibration reference value: ; ; in, This is a normalized index for pressure cycle amplitude, reflecting... Relative to the reference value of permissible pressure cycle amplitude The degree of closeness; This is a normalized index for temperature cycle amplitude, reflecting... Relative to the reference value of allowable temperature cycle amplitude The degree of closeness; The normalized index for the rate of pressure increase or decrease is the absolute value of the rate of pressure change. Reference value for allowable boost / buck rate Normalize the ratio; The normalized index for the rate of temperature change is the absolute value of the rate of temperature change. Reference value for allowable rate of temperature change Normalize the ratio; This is a normalized metric for the cumulative number of iterations, reflecting the cumulative number of iterations. Relative to design or rated cycle life The degree of similarity. Indicates will Limited to between 0 and 1. Because... Contains (1- The term is set to an upper limit of 1 during updates, ensuring that the damage variable monotonically increases with cyclic accumulation and remains within the range of 0 to 1. When the damage variable, microseismic activity, or local deformation continues to increase, evidence of structural instability is generated.

[0037] Creep closure models are used to predict changes in the effective capacity of salt caverns over long-term operation. ; in, For the first Time or number The effective storage capacity of the salt cavern per operating cycle; For the first Time or number The effective storage capacity of the salt cavern per operating cycle; This represents the volumetric shrinkage caused by salt rock creep. This represents the volume change caused by damage, spalling, or local instability. When the measured reservoir capacity decay rate is significantly higher than the model prediction, evidence of abnormal closure or surrounding rock deformation is generated.

[0038] The corrosion clogging model is used to identify operational anomalies caused by water content, brine, salt dust, and corrosion products. Moisture content or humidity, chloride ion concentration, salt dust concentration, filter or pipeline pressure differential, and corrosion rate are each dimensionlessly processed, and the corrosion clogging index is determined based on these dimensionlessly processed characteristic parameters. Its expression is: ; in, For a moment The dimensionless index corresponding to the moisture content or humidity; For a moment The dimensionless index corresponding to the chloride ion concentration; For a moment The dimensionless index corresponding to the salt dust concentration; For a moment The dimensionless index corresponding to the pressure difference of the filter or pipeline; For a moment The dimensionless index corresponding to the corrosion rate; to These are the weighting coefficients for the corresponding indicators.

[0039] Secondly, a data-driven prediction model is constructed to learn the nonlinear evolution of the historical operation of the salt cavern gas storage system. This embodiment adopts a stage-gated double-difference residual regression prediction model (SG-DDRR) to directly predict the future state using historical state changes, operating stages, and online residuals.

[0040] Let the continuous historical state sequence at time t be: ; in, This is the preprocessed and normalized state vector. Let this be the length of the historical backtracking window. Construct first-order and second-order differences for the historical states: ; The first and second differences within the backtracking window are weighted and merged in such a way that the closer to the current time, the greater the weight, to obtain the short-term trend quantity and the trend change quantity: ; ; in, and These are the time weights for the first-order and second-order differences, respectively, and the sum of their respective weights is 1.

[0041] Set an adaptive trend gating coefficient based on the degree of trend change: ; in, This is the trend gating coefficient. Assuming a positive value is preset; when the second-order change is relatively enhanced, Increase the size to improve the model's response to rapid pressure rises and falls, sudden temperature changes, and annular pressure variations.

[0042] Encode the current running stage as a stage indicator vector. Unique thermal coding can be used to represent gas injection, gas storage, gas harvesting, high-frequency peak shaving, or standby, and predictive features can be constructed: ; Regarding the first Each prediction step establishes a direct prediction relationship for the state increment: ; in, and For the first The regression parameters corresponding to each step. and The system is trained using least squares with L2 regularization based on historical sliding window samples, with each prediction step size trained separately to reduce the accumulation of recursive errors.

[0043] To track slow drift, the single-step prediction residual from the previous time step is introduced: ; No. The final predicted value for each prediction step is: ; in, To predict residuals online, This is a correction factor that decays with the prediction step size; it is set for the first run or when there is no valid prediction value from the previous time step. =0.

[0044] This will output the future. State prediction sequence for each time period or operating cycle: ; in, The model parameters can be calibrated based on the sampling frequency, parameter response time, and historical data. The model uses first-order differences to represent the direction of change, and second-order differences to represent the direction of change. The system characterizes accelerating trends and uses stage indicator vectors to distinguish different operating conditions, and utilizes online residuals to correct slow-varying deviations.

[0045] This model is used to identify trends that may occur in the next few cycles, such as abnormal pressure decay, abnormal temperature fluctuations, abnormal reservoir capacity shrinkage, abnormal annular pressure, increased microseismic events, increased salt dust concentration, and increased risk of corrosion and blockage; its output As the predicted state for subsequent data-driven model residual calculation, it is used for state trend prediction in the time dimension.

[0046] Based on dynamic graph A risk propagation model is constructed to facilitate the propagation of spatial or structural association risks between nodes. Specifically, this model is implemented using a Dynamically Weighted Graph Convolutional Network (DW-GCN). As the graph structure foundation of DW-GCN, the dynamic adjacency matrix A t Characterizing the coupling strength between nodes in the salt cavern, wellbore, surrounding rock, and surface process equipment as operating conditions change, based on the input characteristics of each node. As node feature input; DW-GCN is based on Risk information is aggregated and propagated among adjacent nodes to identify the risk status of each node. The node at the th The input features at time t are: ; in, For the first Time of the first Pressure field characteristics of each node; For the first Time of the first Temperature field characteristics of each node; For the first Time of the first The water, humidity, or brine state characteristics of each node; For the first Time of the first The mechanical field characteristics of each node; For the first Time of the first Characteristics of chemical corrosion or blockage at each node; For the first t Time of the first Damage state characteristics of each node.

[0047] Edge weights are represented as: ; ; in, For nodes With nodes Unnormalized coupling score between them; The normalized dynamic edge weights; , , and These are the normalized values ​​of pressure cycle amplitude, temperature cycle amplitude, permeation coupling coefficient, and node distance, respectively, from 0 to 1. Risk of interconnected fractures or interlayers; This refers to the connection relationship of the well shaft structure. and It can be represented by a continuous value from 0 to 1 or a 0 / 1 value; to These are the non-negative edge weight coefficients, and their sum is 1; For nodes The set of connected neighbor nodes. (Normalized) Composition of dynamic graph In the Dynamic adjacency matrix at time step , The graph connection weights are used for calculating risk propagation between nodes in DW-GCN. For graph convolutional propagation, let the input feature matrix of layer 0 of DW-GCN be... = , = +I, for The corresponding degree matrix, the l-th layer graph convolution update is: ; in, It is a non-linear activation function. and The first Trainable weights and biases for layer graph convolutions.

[0048] After L layers of graph convolution, a Softmax risk output layer is used to obtain the node risk vector: ; in, The components correspond to sealing risk, stability risk, creep closure risk, thermal damage risk, corrosion risk, and blockage risk, respectively. For dynamic graph convolutional networks The node feature representation obtained after layer graph convolution; For the trainable weight matrix of the risk output layer; For the trainable bias vector of the risk output layer; This is a normalized exponential function. The risk propagation model outputs the sealing risk, stability risk, creep closure risk, thermal damage risk, corrosion risk, and blockage risk for each node.

[0049] Finally, define the physical model residuals. and data-driven model residuals : ; ; in, This is the actual measured state; Predict the state for the physical model; The data-driven model predicts the state. If both the physical model residuals and the data-driven model residuals are abnormal, it is considered high-confidence evidence of anomaly; if only the data-driven model residuals are abnormal, it is considered evidence of an unknown pattern; if only the physical model residuals are abnormal, it is considered evidence of boundary conditions or insufficient historical samples.

[0050] Therefore, in step S2 of this embodiment, evidence of sealing abnormality, evidence of structural stability abnormality, and evidence of operational condition abnormality are output as inputs for subsequent indexed evaluation.

[0051] S3, Multidimensional Security State Index Identification Model: After obtaining evidence of anomalies, a multidimensional safety status indexation identification model is constructed, which transforms physical residuals, data residuals, node risk probabilities, and key monitoring parameters into a seal integrity index. Salt cave stability index and abnormal working condition index .

[0052] Sealing Integrity Index The formula for evaluating the overall sealing condition of salt caverns, wellbore, cement sheath, packers, caprock, and fracture systems is as follows:

[0053] in, , , , and All are processed according to the aforementioned unified normalization formula, corresponding to the normalized dimensionless quantities of pressure anomaly residuals, mass balance anomaly residuals, annular pressure anomalies, gas composition anomalies, and evidence of anomalies in cement sheath, packer, or wellbore integrity, respectively. The weights corresponding to the above five items are as follows: , , , and All weights are non-negative and satisfy the following conditions: . The value ranges from 0 to 1. A higher value indicates better sealing integrity, while a lower value indicates a higher risk of leakage.

[0054] Salt cavern stability index The formula for evaluating the structural stability of salt caverns, surrounding rock, and under long-term operating conditions is as follows:

[0055] in, The contribution value for normalized mechanical safety; , , , and These are the normalized reservoir capacity change rate anomaly index, damage index, creep closure rate anomaly index, microseismic activity index, and geometric distortion index, respectively. Each index is processed using the aforementioned unified normalization formula. The weights corresponding to the above six items are as follows: , , , , and All weights are non-negative and satisfy the following conditions: . The value ranges from 0 to 1. A higher value indicates a more stable structure, while a lower value indicates a higher risk of instability.

[0056] Abnormal operating condition index Used to evaluate the degree of abnormality in water content, brine, salt dust, corrosion, blockage, and flow response.

[0057] The calculation formula is as follows: ; in, For the first Time or number Abnormal operating condition index of the operating cycle; , , , , and These are the normalized abnormal amounts of water content or humidity, brine, salt dust, corrosion, filter or pipeline pressure difference, and flow response. , , , , and Let be the weight, and satisfy... . The value ranges from 0 to 1, with higher values ​​indicating more severe abnormal conditions.

[0058] Based on the three indices mentioned above, the system generates quantitative identification results for sealed state, structurally stable state, and abnormal operation state, respectively. and For safety-related indicators, the higher the value, the safer the system. This is an abnormally severe indicator; the higher the value, the greater the risk.

[0059] Therefore, in step S3 of this embodiment, the abnormal evidence output in step S2 is transformed into a quantifiable, comparable, and gradable security status index.

[0060] S4. Comprehensive security index fusion and level identification based on adaptive weights during operation phase: In obtaining the seal integrity index Salt cave stability index and abnormal working condition index Subsequently, the operational boundary margin is processed using the aforementioned unified normalization method, converting it into a dimensionless quantity within the range of 0 to 1; based on this, a comprehensive safety index is constructed. The comprehensive safety index is used to characterize the overall safety level of a salt cavern gas storage system during its current operational phase. The formula for calculating the comprehensive safety index is: ; in, The overall safety index; This is the normalized operating boundary margin, and its value ranges from 0 to 1. The larger the value, the higher the operating safety margin. , , , These are the non-negative weighting coefficients corresponding to sealing integrity, structural stability, abnormal operating conditions, and normalized operating boundary margin, respectively, and satisfy the following conditions: .

[0061] The weighting coefficients are adaptively adjusted according to the operational stage of the salt cavern gas storage system.

[0062] During the gas injection phase, the system primarily focuses on maximum pressure, pressurization rate, caprock integrity, wellbore sealing, and thermal stress risk; therefore, improving the seal integrity index is crucial. and operating boundary margin The weight.

[0063] During the gas production phase, the system primarily focuses on minimum pressure, pressure drop rate, roof stability, salt cavern closure, wellbore fluid accumulation, and pipeline blockage risks. Therefore, improving the salt cavern stability index is crucial. and abnormal working condition index The weight of the corresponding item.

[0064] During long-term gas storage, the system primarily focuses on continuous pressure drop, micro-leakage, annular pressure anomalies, caprock leakage, and creep closure risks; therefore, it simultaneously improves... and The weight.

[0065] During the high-frequency peak-shaving phase, the system primarily focuses on pressure cycle amplitude, temperature cycle amplitude, cumulative cycle count, fatigue damage, microcrack propagation, and the risk of seal performance degradation. Therefore, improving... , The weights of the terms corresponding to the running boundary margin.

[0066] according to The size of the salt cavern gas storage system determines its operating status into normal, watch, warning, severe warning, and extreme risk levels. The closer it is to 1, the higher the system's safety margin; The lower the value, the higher the system risk. When the value is below a preset threshold, the system performs further analysis. , , and The contribution ratio is used to identify the main sources of risk.

[0067] Therefore, step S4 of this embodiment completes the conversion from a single security index to a comprehensive security level, and provides a basis for the anomaly attribution in step S5.

[0068] S5. Anomaly Type Attribution Identification and Operational Control Strategy: After obtaining the overall security level, the system... , , The system considers the contribution of operational boundary margins, identifies the causes of anomalies, and outputs corresponding control strategies.

[0069] when The decline was mainly due to When a decrease in pressure is accompanied by a sustained pressure drop, abnormal mass balance, abnormal increase in annular pressure, abnormal gas composition, or abnormal cement sheath or packer integrity, the system identifies it as a seal integrity risk. If this risk occurs during high-pressure injection or pressure-stabilized storage, it is further attributed to the risk of maximum internal pressure, caprock leakage, fracture connectivity, or wellbore seal failure. Corresponding control strategies include reducing the injection rate, reducing the target maximum pressure, extending the pressure stabilization observation time, conducting pressure recovery tests, verifying the wellbore seal status, and checking the integrity of the cement sheath or packer.

[0070] when The decline was mainly due to When a decrease in pressure is accompanied by a low minimum pressure, an excessively high rate of pressure reduction, an increased reservoir shrinkage rate, a rise in microseismic events, and an increased risk at roof or sidewall nodes, the system identifies it as a risk to the stability of the salt cavern structure. If the risk is concentrated in the gas production stage, it is further attributed to minimum internal pressure risk, creep closure risk, decreased roof stability risk, or local instability risk. Corresponding control strategies include increasing the minimum operating pressure, reducing the gas production rate, decreasing the pressure cycle amplitude, adjusting the peak-shaving operation strategy, increasing the frequency of sonar cavity measurements, and updating creep model parameters.

[0071] when The decline was mainly due to When an increase in temperature is caused by factors such as increased water content, abnormal brine backflow, increased salt dust concentration, increased pressure differential in filters or pipelines, sluggish valve operation, or increased corrosion rate, the system identifies this as an abnormal operating condition risk. This risk is further attributed to wellbore fluid accumulation, brine entrainment, salt dust entrainment, corrosion blockage, deliquescence damage, or decreased pipeline flow capacity. Corresponding control strategies include initiating drainage, reducing the gas flow rate, enhancing filtration and dust removal, activating the drying system, checking the corrosion protection status, and cleaning blocked pipeline sections.

[0072] when The decline is mainly due to the operating boundary margin. When a decrease in pressure cycling amplitude, temperature cycling amplitude, cumulative cycle count, or pressure increase / decrease rate exceeding the dynamic safety boundary are all factors that cause a risk, the system identifies this as an operational boundary risk. This risk is further attributed to risks associated with the maximum pressure cycling amplitude, maximum pressure increase / decrease rate, maximum temperature cycling amplitude, or long-term multi-cycle fatigue. Corresponding control strategies include limiting the maximum pressure amplitude, limiting the pressure increase / decrease rate, reducing the temperature cycling amplitude, optimizing the compression and energy release time windows, setting cumulative fatigue damage limits, and correcting long-term operational boundaries.

[0073] Through the above process, this embodiment realizes a continuous closed loop from multi-source monitoring data to state characterization, from state characterization to anomaly evidence, from anomaly evidence to safety index, from safety index to comprehensive level, and from comprehensive level to anomaly attribution and control strategy. It can be used to improve the accuracy of identification and the targeted handling of sealing risks, structural instability risks, corrosion blockage risks and long-term cyclic operation risks in compressed air energy storage salt cavern gas storage systems.

[0074] In one specific embodiment, the method is applied to a 300 MW salt cavern compressed air energy storage power station, with an effective gas storage volume of 500,000 cubic meters. The maximum designed gas storage pressure of the salt cavern is 20 MPa, the recommended minimum normal operating pressure is 14 MPa, and the normal peak-shaving operating pressure range is set from 14.5 MPa to 19.5 MPa. The burial depth of the salt cavern is 950 m to 1250 m, the height of a single cavity is 180 m to 260 m, the maximum cavity diameter is 60 m to 85 m, the safe salt layer thickness on the top plate is not less than 60 m, and the safe salt layer thickness on the bottom plate is not less than 40 m.

[0075] The power plant comprises four phases: gas injection, gas storage, gas extraction, and high-frequency peak shaving. The target pressure increase range for the gas injection phase is 14.5 MPa to 19.5 MPa, with a recommended pressure increase rate not exceeding 0.35 MPa / h. The target pressure decrease range for the gas extraction phase is 19.5 MPa to 14.5 MPa, with a recommended pressure decrease rate not exceeding 0.50 MPa / h. During the long-term gas storage phase, the allowable natural pressure fluctuation range is ±0.15 MPa / d. During the high-frequency peak shaving phase, the number of pressure cycles per day should not exceed 2, and the preferred pressure cycle amplitude is not greater than 5.0 MPa.

[0076] Based on the gas law, initial estimations show that, under conditions of an average gas temperature of 40℃ in the salt cavern, an effective gas storage volume of 500,000 cubic meters, and a salt cavern pressure of 20 MPa, the air mass within the system is approximately 1.11 × 10⁻⁶ m³ / s. 8 kg; when the salt cavern pressure is 14 MPa, the air mass in the system is approximately 7.80 × 10 kg. 7 kg; the dispatchable air mass corresponding to pressures between 14 MPa and 20 MPa is approximately 3.34 × 10 kg. 7 kg. This estimate is used to initialize the mass balance model, and is subsequently corrected by measured pressure, temperature, storage capacity, and gas compressibility factor.

[0077] Multi-source historical database and state vector construction: First, a multi-source historical database for the salt cavern gas storage system is established, including the following data: Operating data: salt cavern internal pressure, wellhead pressure, annulus pressure, gas injection flow rate, gas production flow rate, valve opening, compressor outlet pressure, expander inlet pressure, filter differential pressure, gas injection / production time, number of cycles, and start / stop status.

[0078] Salt cavern body monitoring data: sonar cavity measurement data, effective storage capacity, salt cavern top plate height, maximum cavity diameter, cavity eccentricity, top plate deformation, side wall deformation, storage capacity shrinkage rate, and microseismic event energy.

[0079] Sealing status data: production casing pressure, technical casing pressure, annular pressure, cement sheath integrity evaluation value, packer differential pressure, wellhead sealing status, caprock leakage risk, and fracture connectivity risk.

[0080] Fluid medium data: gas moisture content, dew point temperature, brine backflow rate, chloride ion concentration, salt dust concentration, corrosion rate, drainage rate, and pipeline blockage status.

[0081] Geological engineering static data: salt rock elastic modulus, Poisson's ratio, linear expansion coefficient, creep parameters, salt layer thickness, interlayer distribution, fracture distance, wellbore structural parameters, and cement sheath cementing quality parameters.

[0082] The main monitoring parameters and sampling frequency settings are shown in Table 1: Table 1 Monitoring Parameters and Sampling Frequency Settings

[0083] After performing time synchronization, outlier removal, missing value completion, and normalization on the above data, a unified state vector is formed: ; in, The internal pressure of the salt cavern. For wellbore pressure, For annular pressure, This represents the amplitude of the pressure cycle. For the rate of increase or decrease in voltage, The temperature of the gas in the salt cavern. This represents the amplitude of the temperature cycle. For the rate of temperature change, For the gas injection flow rate, For gas extraction flow rate, For the effective storage capacity of salt caverns, This represents the stability coefficient of salt cavern morphology. For microseismic events, Salt dust concentration, For corrosion rate, For damage variables.

[0084] The salt cavern gas chamber, salt cavern roof, salt cavern sidewalls, salt cavern floor, caprock, interlayers or fractures, wellbore, casing, cement sheath, packer, filter, dryer, injection and production gas pipelines, and valve assembly are abstracted into 14 types of nodes to construct a dynamic diagram. Among them, edge weight The data is dynamically updated based on pressure cycle amplitude, temperature cycle amplitude, pressure rise and fall rate, annular pressure, salt dust concentration, filter differential pressure, micro-vibration activity, and sealing status.

[0085] For example, during the high-frequency peak shaving phase, when The pressure increased from 4.0 MPa to 5.5 MPa, and When the temperature / h is increased from 3℃ / h to 7℃ / h, the edge weights between the salt cavern gas chamber and the top plate node, the salt cavern gas chamber and the side wall node, and the wellbore and the packer node are increased by 15%, 12% and 18%, respectively, to characterize the enhanced influence of pressure cycling and temperature cycling on structural stability and seal performance degradation.

[0086] Implementation of the anomaly evidence identification model: obtaining the state vector and animated GIFs Subsequently, the system establishes a physical prediction model, a data-driven prediction model, and a dynamic graph risk propagation model, respectively.

[0087] The mass balance model adopts the following mass conservation relationship: ; in, This refers to the air quality inside the salt cavern. To inject air mass flow rate, To obtain the air mass flow rate, The leaked air mass flow rate is used. During the gas storage and pressure stabilization phase, the system uses a 6-hour identification window. When there is no gas injection or sampling and the theoretical pressure drop after temperature correction is less than 0.03 MPa / 6h, but the measured pressure drop is greater than 0.08 MPa / 6h, a Level 1 sealing anomaly is generated; when the measured pressure drop is greater than 0.15 MPa / 6h, a Level 2 sealing anomaly is generated; and when the measured pressure drop is greater than 0.25 MPa / 6h, accompanied by an increase in annular pressure exceeding 0.05 MPa / h, a Level 3 sealing anomaly is generated.

[0088] In the thermo-mechanical coupling model, the elastic modulus of salt rock Take 20 GPa, coefficient of linear expansion Pick Poisson's ratio is taken as 0.27. Thermal stress is calculated according to... Calculation. When the temperature cycle amplitude At 20℃, the calculated thermal stress is 16 MPa. If the daily temperature cycle amplitude exceeds 25℃, or the temperature change rate... If the temperature rises above 8°C / h, the evidence of thermal damage is considered stronger.

[0089] Damage variables are updated as follows: ; in, For the first Time or number The damage variable during the running cycle has a value range of 0 to 1; This represents the corresponding damage increment. To match the damage increment with the long-term cyclic accumulation scale, this embodiment introduces a single-cycle damage scale coefficient. The damage increment is defined as: ; in, The scaling factor for single-cycle damage; to It is a non-negative damage weight, and ; , , , and All are normalized dimensionless indices ranging from 0 to 1, representing the pressure cycle amplitude, temperature cycle amplitude, pressure rise / fall rate, temperature change rate, and cumulative cycle count, respectively. All of these indices are processed using the aforementioned unified normalization method; the normalization formulas will not be repeated here.

[0090] In one specific implementation, when ≤5MPa ≤20℃ ≤0.50MPa / h and When ≤5℃ / h, It can be taken as 0.0001 to 0.0003; when >5MPa >25℃ >0.50MPa / h or When any of the conditions >8℃ / h is met It can be taken as 0.0005 to 0.0010; under other transitional operating conditions, It can be determined based on experimental calibration results or interpolation relationships. When the values ​​reach 0.30, 0.50, and 0.70, the system enters the states of concern, early warning, and severe early warning, respectively.

[0091] In the creep closure model, the normal annual reservoir shrinkage rate is set to 0.3% to 1.0%. When sonar cavity measurements show that the annual reservoir shrinkage rate per cavity is greater than 1.2%, evidence of creep closure anomaly is generated; when the annual reservoir shrinkage rate is greater than 1.8% and the number of microseismic events is more than twice the baseline value for 30 consecutive days, evidence of structural stability anomaly is generated.

[0092] In the corrosion clogging model, the corrosion clogging index is... Set as: ; In this embodiment, all input terms marked with a tilde (~) are processed according to the aforementioned unified normalization formula and are dimensionless values ​​between 0 and 1. The normalization formula is not repeated in this embodiment. To ensure that the specific implementation parameters are consistent with the aforementioned normalization formula, this embodiment sets the lower normalization limit for moisture content, chloride ion concentration, salt dust concentration, filter or pipeline pressure difference, and corrosion rate to 0, and the upper normalization limits to 1000 ppm, 100 mg / L, 50 mg / Nm³, 60 kPa, and 0.10 mm / a, respectively. Therefore, the normalization value corresponding to a moisture content of 500 ppm, a chloride ion concentration of 50 mg / L, a filter pressure difference of 30 kPa, and a corrosion rate of 0.05 mm / a is 0.5, and the normalization value corresponding to a salt dust concentration of 20 mg / Nm³ is 0.4. In this embodiment, when the corrosion clogging index... When the value is greater than 0.60 for 2 consecutive hours, evidence of concern regarding corrosion and blockage is generated; when If the value is greater than 0.80 for 1 consecutive hour, evidence of corrosion blockage warning will be generated.

[0093] The data-driven forecasting model uses the continuous state vectors from the past 72 hours as input to predict the trends of pressure, temperature, annular pressure, filter differential pressure, salt dust concentration, microseismic activity, and reservoir capacity changes over the next 6 hours. (Model input window) Take 432 sampling points, prediction step size 36 sampling points were selected. If the data-driven model predicts that the salt dust concentration will exceed 50 mg / Nm³ or the filter pressure difference will exceed 60 kPa within the next 6 hours, evidence of abnormal operating conditions will be generated in advance.

[0094] The physical model residuals and the data-driven model residuals are expressed as follows: ; ; When any two of the following conditions are met simultaneously: pressure residual greater than 0.10 MPa, temperature residual greater than 5 °C, annular pressure residual greater than 0.03 MPa, and filter differential pressure residual greater than 15 kPa, it is considered high-confidence anomalous evidence. If only the data-driven model residuals are anomalous, it is considered unknown mode anomalous evidence; if only the physical model residuals are anomalous, it is considered boundary condition or insufficient historical sample anomalous evidence.

[0095] Multidimensional safety status index calculation: The system calculates the seal integrity index (SII), salt cavern stability index (CSI), and abnormal operating condition index (ACI) every 10 minutes.

[0096] The Seal Integrity Index (SII) is calculated as follows:

[0097] in, , , , and These are normalized dimensionless quantities corresponding to the residuals of pressure anomalies, residuals of mass balance anomalies, annular pressure anomalies, gas composition anomalies, and evidence of anomalies in cement sheath, packer, or wellbore integrity, respectively. The value ranges from 0 to 1, with higher values ​​indicating better sealing.

[0098] The Salt Cavern Stability Index (CSI) is calculated as follows:

[0099] in, The contribution value for normalized mechanical safety; , , , and These are the normalized reservoir capacity change rate anomaly index, damage index, creep closure rate anomaly index, microseismic activity index, and geometric distortion index. All indicators are processed according to the aforementioned unified normalization formula. The higher the value, the more stable the structure.

[0100] The Abnormal Operating Condition Index (ACI) is calculated as follows: ; in, For the first Time or number Abnormal operating condition index of the operating cycle; , , , , and These are the normalized abnormal amounts of water content or humidity, brine, salt dust, corrosion, filter or pipeline pressure difference, and flow response. The higher the value, the more severe the abnormal operating conditions.

[0101] For example, during a certain gas extraction phase, the system collected the following parameters: salt cavern pressure decreased from 18.8 MPa to 14.6 MPa at a rate of 0.58 MPa / h; filter differential pressure was 42 kPa; salt dust concentration was 35 mg / Nm³; water content was 720 ppm; 9 microseismic events occurred per day; and the annualized storage capacity shrinkage rate was 1.3%. The model calculations yielded the following results: ; ; The results indicate that the seal integrity remains in good condition, but risks to salt cavern stability and abnormal operating conditions have accumulated.

[0102] Integrated Security Index Fusion and Level Identification: Integrated Security Index Calculate using the following formula: ; in, , , , The weights are respectively for sealing integrity, structural stability, abnormal operating conditions, and operating boundary margin, and satisfy the following conditions: .

[0103] The weight settings for each operational phase are shown in Table 2: Table 2 Weight Settings

[0104] In the above gas production stage example, the gas production stage weighting is used for calculation: , ; According to the classification rules Corresponding to the "early warning" level. The comprehensive security level classification is shown in Table 3: Table 3 Comprehensive Security Level Classification

[0105] Simultaneously set a single trigger rule, even if If the threshold is not lowered, the system will still force the system into a warning or severe warning state if any of the following conditions are met: When the salt cavern pressure reaches 19.8 MPa, a high-pressure warning is triggered; when it reaches 20.0 MPa, overpressure protection is triggered. When the salt cavern pressure is below 14.0 MPa, a minimum pressure warning is triggered; when it is below 13.6 MPa, a severe warning is triggered. When the annular pressure increases at a rate greater than 0.05 MPa / h for 1 hour, a sealing warning is triggered. When the pressure drop after temperature correction during the stabilization phase is greater than 0.15 MPa / 6h, a sealing warning is triggered. When the number of micro-vibration events in a single day is greater than 15, or the energy of a single micro-vibration exceeds 5 times the baseline value, a structural stability warning is triggered. When the filter pressure difference is greater than 60 kPa, or the salt dust concentration is greater than 50 mg / Nm³, an operating condition warning is triggered. When the corrosion rate is greater than 0.10 mm / a, a corrosion risk warning is triggered.

[0106] Anomaly Attribution and Control Strategy Output: After the overall security level is determined, the system outputs... , , and We will analyze the contribution ratio to identify the main sources of risk.

[0107] In the above example of the gas extraction stage, The main source of the decline is Increase and Decrease. Among them, This is mainly caused by increased salt dust concentration, filter pressure difference, and moisture content. This is mainly caused by a high depressurization rate, an increase in micro-seismic events, and a rise in reservoir capacity contraction. Therefore, the system determines that this operating condition belongs to a coupled early warning type of abnormal operating condition risk and structural stability risk during the gas production stage, and the system outputs the following control strategy: Reduce the gas production depressurization rate from 0.58 MPa / h to no more than 0.40 MPa / h; increase the minimum target pressure for this round of gas production from 14.5 MPa to 15.0 MPa; reduce the pressure cycle amplitude from 4.2 MPa to no more than 3.5 MPa; initiate the wellbore drainage program and continuously drain for 30 to 60 minutes; put the drying system into operation and control the gas dew point below -25℃; initiate a filter dust removal bypass check, and switch to the standby filter when the filter differential pressure is still higher than 45 kPa; increase the salt dust concentration sampling frequency from once / 5 min to once / 1 min; shorten the microseismic event statistical window from 24 h to 6 h; arrange sonar cavity verification after this round of peak shaving, focusing on checking the top plate, sidewalls, and narrowing sections; update the creep closure model parameters and recalculate the minimum operating pressure boundary for the next 30 days.

[0108] When the system identifies a risk to seal integrity, the control strategies include: reducing the gas injection rate and lowering the target maximum pressure from 19.5 MPa to 18.8 MPa; extending the pressure stabilization observation time to 12 hours; conducting a pressure recovery test; reviewing the annulus pressure curve; checking the sealing status of the cement sheath, packer, and wellhead; and, if necessary, suspending gas injection and initiating wellbore integrity testing.

[0109] When the system identifies a risk to the stability of the salt cavern structure, the control strategies include: increasing the minimum operating pressure from 14.0 MPa to 14.8 MPa; reducing the gas production rate; narrowing the pressure cycle amplitude; reducing the number of cycles per day; increasing the frequency of sonar cavity measurements; conducting local stability checks on the roof, sidewalls, and interlayer development areas; and revising the salt rock creep model and damage accumulation model.

[0110] When the system identifies corrosion blockage or abnormal operating conditions as a risk, control strategies include: initiating drainage; reducing the gas flow rate; enhancing filtration and dust removal; putting the drying system into operation; checking for liquid accumulation at low points in the pipeline; cleaning blocked pipe sections; checking corrosion monitoring plates; improving gas dehydration and desalination accuracy; and reducing unit load operation if necessary.

[0111] When the system identifies an operational boundary risk, the control strategy includes: limiting the maximum pressure to no more than 19.5 MPa; limiting the minimum pressure to no less than 14.5 MPa; limiting the pressure increase rate to no more than 0.35 MPa / h; limiting the pressure decrease rate to no more than 0.50 MPa / h; limiting the number of pressure cycles per day to no more than 2; and setting an upper limit for cumulative fatigue damage. When the value exceeds 0.50, it will automatically enter conservative operation mode.

[0112] This embodiment enables multi-source monitoring data fusion, anomaly evidence identification, safety index calculation, comprehensive level determination, and control strategy output for a 300 MW-class salt cavern compressed air energy storage power station with an effective gas storage volume of 500,000 cubic meters, a maximum gas storage pressure of 20 MPa, and a recommended minimum operating pressure of 14 MPa. This method can generate quantitative identification results and graded treatment recommendations at the early stages of risks such as sealing anomalies, decreased structural stability, corrosion blockage, salt dust entrainment, wellbore fluid accumulation, and high-frequency peak-shaving fatigue damage, thereby improving the safety and reliability of the long-term cyclic operation of the salt cavern gas storage system.

[0113] This invention addresses the problems that compressed air energy storage salt cavern gas storage systems are prone to encounter during high-frequency injection and extraction, rapid pressure rise and fall, large temperature difference cycling, and long-term operation, such as air leakage, cavity shrinkage, thermal damage, corrosion blockage, and wellbore seal failure. It primarily relies on existing sensing systems to collect operating parameters such as pressure, temperature, flow rate, annular pressure, water content, salt dust concentration, and corrosion status, combined with existing geological engineering parameters for analysis. This eliminates the need for large-scale addition of new testing equipment, demonstrating strong engineering feasibility and facilitating implementation and promotion. By uniformly characterizing operating conditions, the salt cavern itself, wellbore integrity, fluid media, and geological engineering parameters, it can continuously monitor and intelligently identify salt cavern sealing performance, structural stability, and abnormal operating conditions. By constructing sealing integrity indices, salt cavern stability indices, abnormal operating condition indices, and comprehensive safety indices, it can achieve quantitative evaluation, classification, and interpretable early warning of complex multi-field coupled risks. Simultaneously, by incorporating adaptive weighting and anomaly attribution strategies during the operating phase, it can provide a basis for operational boundary correction, risk warning, and safety control, thereby reducing the risks of air leakage, instability, and equipment blockage, and improving the long-term safe and stable operation capability of compressed air energy storage power stations.

[0114] Example 2: This embodiment provides a gas storage salt cavern performance identification system based on multi-field coupling, including: The data acquisition module is configured to: acquire multi-source historical data of the salt cavern gas storage system; and construct a dynamic map with salt caverns, wellbores, surrounding rock, and surface process equipment as nodes. The anomalous evidence determination module is configured to: determine the physical model residuals and data-driven model residuals based on multi-source historical data and preset mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models; and determine anomalous evidence based on the physical model residuals and data-driven model residuals. The safety index determination module is configured to: convert abnormal evidence into a sealing integrity index, a salt cavern stability index, and an abnormal operating condition index; determine a comprehensive safety index by weighted summation based on the sealing integrity index, salt cavern stability index, and abnormal operating condition index; and adaptively adjust the weights according to the operating stage of the salt cavern gas storage system. The identification module is configured to: identify the cause of anomalies based on the comprehensive security index.

[0115] The working method of the system is the same as that of the gas storage salt cavern performance identification method based on multi-field coupling in Example 1, and will not be repeated here.

[0116] Example 3: This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the gas storage salt cavern performance identification method based on multi-field coupling described in Embodiment 1.

[0117] Example 4: This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the program, it implements the steps of the gas storage salt cavern performance identification method based on multi-field coupling described in Embodiment 1.

[0118] Example 5: This embodiment provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the steps of the gas storage salt cavern performance identification method based on multi-field coupling described in Embodiment 1.

[0119] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A method for identifying the performance of gas storage salt caverns based on multi-field coupling, characterized in that, include: Acquire multi-source historical data of salt cavern gas storage systems; A dynamic diagram is constructed using salt caverns, well shafts, surrounding rock, and surface process equipment as nodes; Based on multi-source historical data, as well as pre-set mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models, the physical model residuals and data-driven model residuals are determined; and anomaly evidence is determined based on the physical model residuals and data-driven model residuals. Abnormal evidence is transformed into a seal integrity index, a salt cavern stability index, and an abnormal operating condition index. A comprehensive safety index is determined by weighted summation based on the seal integrity index, salt cavern stability index, and abnormal operating condition index, and the weights are adaptively adjusted according to the operating stage of the salt cavern gas storage system. Anomaly types are identified and attributed based on the comprehensive safety index.

2. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The multi-source historical data includes operational condition data, salt cavern body monitoring data, sealing status data, fluid medium data, and geological engineering static data; the dynamic graph includes nodes, edge sets, and a dynamic adjacency matrix; nodes are the salt cavern gas chamber, salt cavern top plate, salt cavern sidewall, salt cavern bottom plate, caprock, interlayer, wellbore, casing, cement sheath, packer, and surface process equipment; edge sets are used to describe the pressure transmission, heat conduction, seepage, stress transmission, damage propagation, corrosion blockage propagation, and operational control coupling relationships between different nodes; the dynamic adjacency matrix is ​​updated based on pressure cycle amplitude, temperature cycle amplitude, pressure rise and fall rate, annular pressure, salt dust concentration, filter differential pressure, microseismic activity, and sealing status data to characterize the dynamic evolution of risk propagation relationships within the salt cavern gas storage system.

3. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The mass balance model is as follows: ； in, The mass of gas inside the salt cavern; This refers to the mass flow rate of the gas injected into the salt cavern. This refers to the mass flow rate of the extracted gas. This represents the mass flow rate of the leaked gas.

4. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The thermo-mechanical coupling model is used to determine the thermal stress and damage accumulation under the combined effects of pressure cycling and temperature cycling: ； ； in, Thermal stress; The elastic modulus of the material; The coefficient of linear expansion; This represents the amplitude of the temperature cycle. For the first Time-dependent damage variables; For the first Damage variables at all times The damage increment function, This represents the amplitude of the pressure cycle. This represents the amplitude of the temperature cycle. For the rate of increase or decrease in voltage, For the rate of temperature change, This represents the cumulative number of loops.

5. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The creep closure model is used to predict changes in the effective capacity of salt caverns over long-term operation. ； in, For the first Time or number The effective storage capacity of the salt cavern per operating cycle; For the first Time or number The effective storage capacity of the salt cavern per operating cycle; This represents the volumetric shrinkage caused by salt rock creep. This represents the change in volume.

6. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The corrosion clogging model is used to identify operational anomalies caused by water, brine, salt dust, and corrosion products, and defines a corrosion clogging index: ； in, For a moment The dimensionless index corresponding to the moisture content or humidity; For a moment The dimensionless index corresponding to the chloride ion concentration; For a moment The dimensionless index corresponding to the salt dust concentration; For a moment The dimensionless index corresponding to the pressure difference of the filter or pipeline; For a moment The dimensionless index corresponding to the corrosion rate; to These are the weighting coefficients for the corresponding indicators.

7. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, If both the physical model residuals and the data-driven model residuals are abnormal, it is considered high-confidence evidence of anomaly; if only the data-driven model residuals are abnormal, it is considered evidence of an unknown mode anomaly; if only the physical model residuals are abnormal, it is considered evidence of boundary conditions or insufficient historical samples. The anomaly evidence includes sealing anomaly evidence, structural stability anomaly evidence, and operating condition anomaly evidence. The sealing integrity index is used to evaluate the overall sealing status of the salt cavern, wellbore, cement sheath, packer, caprock, and fracture system. The salt cavern stability index is used to evaluate the structural stability of the salt cavern cavity, surrounding rock, and under long-term operating conditions. The anomaly operating condition index is used to evaluate the degree of anomalies in water content, brine, salt dust, corrosion, blockage, and flow response.

8. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The adaptive adjustment of the weights includes: increasing the weights of the sealing integrity index and operating boundary margin during the gas injection phase; increasing the weights of the corresponding items of the salt cavern stability index and abnormal operating condition index during the gas extraction phase; increasing the weights of the sealing integrity index and salt cavern stability index during the gas storage phase; and increasing the weights of the corresponding items of the salt cavern stability index and abnormal operating condition index during the peak shaving phase.

9. The method for identifying the performance of gas storage salt caverns based on multi-field coupling as described in claim 1, characterized in that, The attribution identification includes: when the overall safety index decreases mainly due to a decrease in the seal integrity index, and is accompanied by continuous pressure drop, abnormal mass balance, abnormal increase in annular pressure, abnormal gas composition, or abnormal integrity of the cement ring and packer, it is determined to be a seal integrity risk; when the overall safety index decreases mainly due to a decrease in the salt cavern stability index, and is accompanied by a lower minimum pressure, increased pressure drop rate, increased reservoir shrinkage rate, increased micro-vibration events, and increased risk at the top plate or sidewall nodes, it is determined to be a salt cavern structural stability risk; when the overall safety index decreases mainly due to an increase in the abnormal operating condition index, and is accompanied by increased moisture content, abnormal brine backflow, increased salt dust concentration, increased pressure differential of filters or pipelines, sluggish valve operation, or increased corrosion rate, it is determined to be an abnormal operating condition risk; when the overall safety index decreases mainly due to a decrease in the operating boundary margin, and is accompanied by an increase in the cumulative number of cycles and a pressure increase or decrease rate exceeding the dynamic safety boundary, it is determined to be an operating boundary risk.

10. A gas storage salt cavern performance identification system based on multi-field coupling, characterized in that, include: The data acquisition module is configured to: acquire multi-source historical data of the salt cavern gas storage system; and construct a dynamic map with salt caverns, wellbores, surrounding rock, and surface process equipment as nodes. The anomalous evidence determination module is configured to: determine the physical model residuals and data-driven model residuals based on multi-source historical data and preset mass balance models, thermo-mechanical coupling models, creep closure models, corrosion blockage models, and data-driven prediction models; and determine anomalous evidence based on the physical model residuals and data-driven model residuals. The safety index determination module is configured to: convert abnormal evidence into a sealing integrity index, a salt cavern stability index, and an abnormal operating condition index; determine a comprehensive safety index by weighted summation based on the sealing integrity index, salt cavern stability index, and abnormal operating condition index; and adaptively adjust the weights according to the operating stage of the salt cavern gas storage system. The identification module is configured to: identify the cause of anomalies based on the comprehensive security index.

Images