System and method for dynamically utilizing cold energy of gas storage of million-cube compressed air energy storage system

Abstract

The invention provides a system and a method for dynamically utilizing cold energy of a gas storage of a million-cube compressed air energy storage system. The method comprises the following steps: acquiring multi-source monitoring data to construct a basic data set; establishing an internal space-time dynamic model of the gas storage by applying a multi-scale thermal flow constraint space reconstruction method; a hierarchical control framework is developed to realize cooperative control of a throttle valve and a cold energy recovery device; and a working condition similarity enhancement method based on cold energy characteristics and a cold energy recovery control strategy self-adaptive generation method are constructed. According to the method, the technical problems of temperature field accurate modeling in the large gas storage, cold energy characteristic working condition identification, throttling-cold energy recovery cooperative control and the like are solved, the cold energy recovery efficiency is improved by 20-30%, and turbine inlet temperature fluctuation is reduced.

Description

Technical Field

[0001] The present invention relates to CAES technology, in particular to a system and method for dynamically utilizing the cold energy of a gas storage reservoir in a million-cubic-meter compressed air energy storage system. Background Technique

[0002] With the continuous increase in the grid connection ratio of renewable energy, the power system's demand for large-scale energy storage technology has become increasingly urgent. Compressed air energy storage (CAES), with its advantages of large capacity, long life, and low cost, has become an important technical route for large-scale energy storage. In a million-cubic-meter CAES system, a large amount of low-temperature cold energy is generated during the gas discharge process of the gas storage reservoir. If this cold energy can be effectively recovered and utilized, it can not only improve the overall energy utilization efficiency of the system but also reduce the thermal shock of the primary heat exchanger in the turbine section, enhancing the overall safety and economy of the system.

[0003] Currently, the research on cold energy utilization in compressed air energy storage systems mainly focuses on small systems, using conventional throttling control and simple cold energy recovery methods. Traditional methods usually adopt a throttle valve positioning system based on PID control, which measures the temperature and pressure at the turbine inlet and adjusts the throttle valve opening to ensure the stable operation of the turbine. In terms of cold energy recovery, common technical means include directly using cold air for refrigeration, coupling with a heat pump system, and using it as an industrial cold source, etc. In terms of spatial modeling, the temperature field calculation mainly adopts methods based on finite element or finite volume methods, and the control strategy is mainly based on rule-based control algorithms. The working condition identification mostly uses the feature comparison method under a single time scale.

[0004] However, when a cold energy recovery device is introduced into a million-cubic-meter CAES system, traditional methods face multiple technical challenges. For example, there is a complex coupling relationship between the throttle valve and the cold energy recovery device. The traditional independent control strategy cannot balance the dual goals of stable turbine inlet parameters and maximizing cold energy recovery efficiency, especially performing poorly in the dynamic process of rapid system state changes, seriously restricting the improvement of cold energy recovery efficiency and stable operation. Summary of the Invention

[0005] Object of the Invention: To provide a system and method for dynamically utilizing the cold energy of a gas storage reservoir in a million-cubic-meter compressed air energy storage system, in order to solve at least one technical problem existing in the prior art.

[0006] Technical Solution: A method for dynamically utilizing the cold energy of a gas storage reservoir in a million-cubic-meter compressed air energy storage system includes: Obtaining multi-source monitoring data of the compressed air energy storage system to form a system basic data set; accordingly, performing multi-scale heat flow constraint space reconstruction, and generating a dynamic cold energy distribution prediction map through a space-time dynamic model; Based on the system basic dataset, a hierarchical control framework is constructed through the coupling influence model of throttling and cold energy recovery to generate a hierarchical control strategy; Based on the system basic dataset and the dynamic cold energy distribution prediction map, a working condition feature fingerprint library is constructed, and an optimized control strategy for cold energy recovery is generated through similar working condition identification; it is fused with the hierarchical control strategy to output a comprehensive control instruction set.

[0007] According to one aspect of the present application, generating a dynamic cold energy distribution prediction map includes: Based on the system basic dataset, the full-space temperature field and pressure field in the gas storage reservoir are initially reconstructed; accordingly, multi-scale heat flow constraint space reconstruction is carried out to optimize the full-space temperature field and pressure field; Based on the optimized full-space temperature field and pressure field, time-sequence - space coupling evolution prediction is carried out to generate a dynamic cold energy distribution prediction map.

[0008] According to one aspect of the present application, optimizing the full-space temperature field and pressure field includes: The initial reconstruction results of the full-space temperature field and pressure field are subjected to hierarchical space segmentation to obtain a multi-scale spatial domain; accordingly, physical constraint models for different scales are constructed to form a multi-scale physical constraint set; Using the data of the temperature sensors on the gas storage reservoir wall in the system basic dataset, a dynamic boundary condition set is constructed through adaptive boundary constraints; it is combined with the measured data of the sensors in the system basic dataset and the multi-scale physical constraint set to perform physics-guided interpolation reconstruction to optimize the full-space temperature field and pressure field.

[0009] According to one aspect of the present application, time-sequence - space coupling evolution prediction is carried out to generate a dynamic cold energy distribution prediction map, including: Based on the optimized full-space temperature field and pressure field, adaptive time window analysis is applied to identify the time dynamic characteristics in different stages to generate a time characteristic segmentation set; for each stage among them, a space evolution model sensitive to temperature gradient is constructed to form a segmented space evolution model set; Read the gas release parameters of the throttle valve operation state sequence in the system basic dataset, analyze the influence of the throttle valve opening change on the temperature field evolution inside the gas storage reservoir, and establish a gas release - temperature coupling model; it is combined with the segmented space evolution model set, and a dynamic cold energy distribution prediction map is generated through multi-step progressive prediction.

[0010] According to one aspect of the present application, generating a time characteristic segmentation set includes: Based on the optimized full-space temperature field and pressure field, calculate the statistical characteristics of the full-space temperature change rate, including mean value, variance, and kurtosis; Classify statistical features through time series clustering to automatically identify rapid change, transition, and slow change phases; adjust the time step according to the characteristics of different phases: adopt the first small time step in the rapid change phase to improve prediction accuracy, and adopt the second time step in the slow change phase to improve calculation efficiency; finally generate a time characteristic segmented set; where the first time step is smaller than the second time step.

[0011] According to one aspect of the present application, generate an optimized control strategy for cold energy recovery, including: Based on the system basic data set, construct a working condition characteristic fingerprint library for the compressed air energy storage system; accordingly, generate a working condition similarity vector from the current system state through working condition identification; Combine the dynamic cold energy distribution prediction map to enhance the working condition similarity based on cold energy characteristics for the working condition similarity vector, and obtain an optimized working condition similarity vector; accordingly, generate an adaptive cold energy recovery control strategy and output an optimized control strategy for cold energy recovery.

[0012] According to one aspect of the present application, obtaining the optimized working condition similarity vector includes: Based on the system basic data set and the dynamic cold energy distribution prediction map, extract cold energy characteristic indicators, construct a similarity calculation formula for cold energy recovery, and generate a thermodynamic similarity index set; Construct a stage similarity weighting strategy according to the characteristics of the cold energy recovery process, and generate a stage weighting function set; Read the operation data of the cold energy recovery device in the system basic data set, analyze the relationship between working condition similarity and cold energy recovery efficiency, and establish an efficiency correlation mapping; Combine the thermodynamic similarity index set, the stage weighting function set, and the efficiency correlation mapping to calculate the enhanced similarity between the current and historical working condition similarity vectors, and output the optimized working condition similarity vector.

[0013] According to one aspect of the present application, the thermodynamic similarity index set includes: Temperature gradient distribution similarity index, obtained by calculating the cosine similarity of the temperature gradient fields of two working conditions; Similarity index of the movement mode of the cold and hot interface, obtained by comparing the moving speed and direction of the cold and hot dividing interfaces; Cold energy generation rate similarity index, obtained by comparing the change mode of the cold energy generation amount per unit time.

[0014] According to one aspect of the present application, perform adaptive generation of the cold energy recovery control strategy and output an optimized control strategy for cold energy recovery, including: Retrieve the control strategies of similar operating conditions from the historical control database based on the optimized operating condition similarity vector, extract the key control decisions, and generate a candidate control strategy set; identify the control decision elements related to the cold energy recovery efficiency from them, and construct a cold energy efficiency impact factor set. Based on the coupling influence model and the cold energy efficiency impact factor set, construct a throttling-recovery dual-objective optimization framework, regard the throttle valve control and the cold energy recovery device control as a coupled optimization problem, and construct a multi-objective optimization model; combine it with the current system state, perform real-time optimization calculations, and generate the cold energy recovery optimization control strategy under the current operating condition.

[0015] A cold energy dynamic utilization system for a gas storage reservoir of a million cubic meter-level compressed air energy storage system, including: A compression system for compressing air; A gas storage reservoir connected to the compression system for storing the compressed gas; An exhaust pipeline connected to the gas storage reservoir for discharging the gas in the gas storage reservoir; A throttle valve arranged in the exhaust pipeline for controlling the gas flow rate and pressure; A cold energy recovery device arranged at the rear end of the exhaust port of the throttle valve in the exhaust pipeline for recovering the cold energy in the gas; A heat exchanger arranged at the rear end of the cold energy recovery device for performing heat exchange between the cold energy and the gas; A turbine system arranged after the heat exchanger for performing turbine work on the gas after heat exchange to generate electricity; A control module, including: At least one processor; and, A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method of the above technical solution.

[0016] Beneficial effects: The present invention ensures that the cold energy recovery device can be effectively arranged and controlled according to the actual cold energy distribution position, avoiding energy waste and inefficient operation of equipment; improving the cold energy recovery efficiency and the system operation stability; ensuring the stable operation of the turbine system, solving the contradiction between cold energy recovery and the stable operation of the turbine, and improving the overall energy utilization efficiency of the system. Description of the Drawings

[0017] Figure 1 It is a step flow chart of a method for dynamically utilizing the cold energy of a gas storage reservoir of a million cubic meter-level compressed air energy storage system provided by an embodiment of the present application.

[0018] Figure 2 It is a step flow chart of optimizing the full-space temperature field and pressure field provided by an embodiment of the present application.

[0019] Figure 3 This is a flowchart of the steps for obtaining an optimized operating condition similarity vector provided by the embodiments of the present application.

[0020] Figure 4 This is a flowchart of the steps for adaptively generating an output cold energy recovery optimization control strategy for cold energy recovery control strategy provided by the embodiments of the present application. Detailed implementation manners

[0021] In order to enable those skilled in the art of the present technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] It should be particularly noted that, for the convenience of clearly showing the step flow of the present application, serial numbers are marked for each step in the specification. These serial numbers are only for the convenience of explanation and do not limit the execution order of the steps. In actual operation, according to the technical requirements of the specific implementation scenario, the steps can be executed in an order different from that shown in the specification, and in some cases, parallel processing between steps can also be achieved.

[0023] It is found in the research process that there are obvious temperature stratification and uneven distribution phenomena inside large gas storage caverns. Traditional spatial modeling methods cannot accurately capture such complex temperature fields under the condition of sparse sampling points, resulting in inaccurate prediction of the cold energy position; in addition, conventional operating condition identification methods lack special consideration for the characteristics of cold energy and are difficult to identify operating conditions that are similar in terms of cold energy recovery efficiency, affecting the effective transfer of control experience.

[0024] Specifically, in the currently constructed and studied compressed air energy storage system, the air with a lower temperature discharged from the gas storage cavern is directly heated and then enters the turbine for turbine work, without targeting the cold energy of the gas discharged from the gas storage cavern. For small-scale compressed air energy storage power stations, the cold energy recovery of the gas storage cavern is of low economic value and application value, but for a compressed air energy storage power station with a capacity of millions of cubic meters, it has both economic value and application value. In order to make full use of the cold energy of the gas storage cavern of the compressed air energy storage power station and improve the comprehensive energy utilization efficiency of the system, a dynamic cold energy utilization system for the gas storage cavern of a compressed air energy storage system with a capacity of millions of cubic meters is proposed.

[0025] It mainly includes a gas storage reservoir, an exhaust pipeline, a throttle valve, a cold energy recovery device, a heat exchanger, a turbine system, and a compression system. During the power generation process of a compressed air energy storage power station, high-pressure gas is discharged through the exhaust pipeline. The discharged high-pressure air is injected into the cold energy recovery device after passing through the throttle valve, and then the high-pressure air is injected into the heat exchanger after passing through the cold energy recovery system, and then enters the turbine system to perform turbine work and generate electricity. The cold energy recovery device is connected to the exhaust port of the throttle valve at the rear end of the exhaust pipeline of the gas storage reservoir, uses a medium to recover the cold energy in the air discharged from the gas storage reservoir, and stores it in the cold energy storage tank. And the recovered cold energy can be regulated by adjusting the throttle valve and the exhaust flow rate of the gas storage reservoir. Specifically, the compression system is used to compress air; the gas storage reservoir is connected to the compression system and is used to store the compressed gas; the exhaust pipeline is connected to the gas storage reservoir and is used to discharge the gas in the gas storage reservoir; the throttle valve is arranged in the exhaust pipeline and is used to control the gas flow rate and pressure; the cold energy recovery device is arranged at the rear end of the exhaust port of the throttle valve in the exhaust pipeline and is used to recover the cold energy in the gas; the heat exchanger is arranged at the rear end of the cold energy recovery device and exchanges heat with the gas through cold energy; the turbine system is arranged after the heat exchanger and is used to perform turbine work and generate electricity on the gas after heat exchange.

[0026] As Figure 1 shown, for the above system, a method for dynamically utilizing the cold energy of a gas storage reservoir in a million-cubic-meter compressed air energy storage system is provided, including: S1. Obtain multi-source monitoring data of the compressed air energy storage system to form a system basic data set; Specifically, the multi-source monitoring data includes multi-point temperature and pressure data inside the gas storage reservoir, the operating status data of the throttle valve, and the working parameters of the cold energy recovery device.

[0027] S2. Based on the system basic data set, perform multi-scale heat flow constraint space reconstruction, and generate a dynamic cold energy distribution prediction map through a space-time dynamic model; Specifically, multi-scale heat flow constraint space reconstruction means that the system will consider data at different levels, such as microscopic temperature changes and macroscopic overall energy trends, analyze and optimize the spatial heat flow, and more accurately simulate the temperature and energy flow in the compressed air energy storage system. The space-time dynamic model combines the change factors of space (each region inside the gas storage reservoir) and time (different operating stages) to predict the flow and distribution of cold energy dynamically.

[0028] S3. Based on the system basic data set, construct a hierarchical control framework through a coupling influence model of throttling and cold energy recovery, and generate a hierarchical control strategy; Specifically, the throttling process refers to the restriction experienced by air when flowing in the energy storage system, which affects the changes in pressure and temperature; cold energy recovery refers to recovering low-temperature energy from the system to improve the overall energy utilization rate. There is a complex interaction between the two, so a mathematical model is needed to analyze how they affect each other to ensure that the system can operate efficiently under different working conditions. Since the operation of the system involves multiple variables such as pressure, temperature, and flow rate, direct control would be complex, so a hierarchical control architecture needs to be established. This framework may include multiple levels, such as: Bottom layer control: Adjust the operation of individual devices, such as the opening and closing state of the throttle valve; Middle layer control: Comprehensively regulate the interaction between multiple devices, such as the coordinated operation of the gas storage reservoir and the cold energy recovery device; Top layer control: Conduct overall optimization to ensure that the entire system meets the expected energy efficiency target.

[0029] S4. Based on the system basic data set and the dynamic cold energy distribution prediction map, construct a working condition characteristic fingerprint library, and generate an optimized control strategy for cold energy recovery through similar working condition identification; Integrate the optimized control strategy for cold energy recovery with the hierarchical control strategy, and output a comprehensive control instruction set, including the throttle valve control signal and the cold energy recovery device control signal.

[0030] Specifically, the working condition characteristic fingerprint library refers to a database of different states of system operation. Through the fingerprint library system, it can be identified whether the current working condition is similar to some known states in the past. In this way, based on historical experience, the cold energy flow situation of the current system can be predicted, and an optimized control strategy can be generated.

[0031] According to one aspect of the present application, the steps of forming the system basic data set include: S11. Collect the temperature and pressure data at multiple points inside the gas storage reservoir to construct a spatial distribution monitoring data set; S12. Obtain the operation state data of the throttle valve to form a throttle control data set; S13. Monitor the working parameters of the cold energy recovery device to establish a cold energy recovery characteristic data set; S14. Integrate the spatial distribution monitoring data set, the throttle control data set, and the cold energy recovery characteristic data set, perform time alignment and cleaning, and output the system comprehensive monitoring data set as the system basic data set.

[0032] Specifically, collect multi-point temperature and pressure data inside the gas storage reservoir to construct a spatial distribution monitoring data set. Read the data of the distributed temperature sensor array (T1, T2, ..., Tn) and the pressure sensor array (P1, P2, ..., Pm) inside the gas storage reservoir, and combine the sensor spatial position information (X1, Y1, Z1, ..., Xn, Yn, Zn) to generate the original spatial distribution data containing time stamps. Obtain the operating status data of the throttle valve to form a throttle control data set. Collect the throttle valve opening signal (Vo), the pressure difference before and after the throttle valve (ΔP), the temperature difference before and after the throttle valve (ΔT), and the flow rate data (Q), and combine the time stamp information to generate the operating status sequence of the throttle valve. Monitor the working parameters of the cold energy recovery device to establish a cold energy recovery characteristic data set. Collect key parameters such as the inlet temperature (Tci), the outlet temperature (Tco), the working fluid flow rate (Qc), and the heat exchange efficiency (η) of the recovery device of the cold energy recovery working fluid to generate the operating data of the cold energy recovery device. Integrate multi-source data and perform time alignment and cleaning to output the standardized system comprehensive monitoring data set D. Apply the time stamp alignment algorithm to handle the sampling frequency differences of different data sources to ensure the time consistency and availability of the data.

[0033] According to one aspect of the present application, the steps of generating a dynamic cold energy distribution prediction map include: S21. Based on the system basic data set, preliminarily reconstruct the full-space temperature field and pressure field inside the gas storage reservoir; S22. Based on the preliminary reconstruction result, aiming at the sparse problem of monitoring points in a gas storage reservoir with a scale of millions of cubic meters, perform multi-scale heat flow constraint spatial reconstruction to optimize the full-space temperature field and pressure field; S23. Based on the optimized full-space temperature field and pressure field, perform time-series - space coupling evolution prediction to generate a dynamic cold energy distribution prediction map.

[0034] Specifically, based on the system comprehensive monitoring dataset D, the kernel function interpolation method is applied to preliminarily reconstruct the full-space temperature field Ts(x, y, z, t) and pressure field Ps(x, y, z, t) in the gas storage reservoir. The standard kernel function interpolation is used to construct the basic spatial distribution model, providing an initial value for subsequent optimization. Aiming at the sparse monitoring points problem in the million-cubic-level gas storage reservoir, a multi-scale heat flow constraint space reconstruction method is developed to optimize the full-space temperature field Ts and pressure field Ps. Specifically: introducing the fluid thermodynamics theory constraint, using the prediction based on physical laws to replace pure mathematical interpolation in the data-sparse region, to solve the "false distribution" problem caused by insufficient sampling points in the traditional spatial reconstruction method in large gas storage reservoirs; constructing a multi-scale analysis framework, applying different physical models at different spatial scales: the heat diffusion model is used at the micro-scale to describe the local temperature change, the convective heat transfer model is used at the meso-scale to describe the heat transfer caused by air flow, and the stratified flow model is used at the macro-scale to describe the overall temperature stratification phenomenon; constructing an adaptive boundary constraint mechanism, using the data of the temperature sensors on the gas storage reservoir wall to construct dynamic boundary conditions, to solve the problem of unclear boundary conditions in the traditional method when dealing with large closed spaces.

[0035] Construct a time-sequence - space coupling evolution prediction algorithm to generate the prediction map G of the internal dynamic cold energy distribution in the gas storage reservoir. Aiming at the unique requirements of the cold energy recovery scenario, the deficiencies of the traditional spatio-temporal model are solved: constructing a non-uniform time-step prediction mechanism, adaptively adjusting the prediction time step according to the dynamic characteristics of different stages in the cold energy recovery process, using a small time step in the fast-changing stage (such as the initial stage of gas discharge) to improve the accuracy, and using a large time step in the slow-changing stage to improve the efficiency. Construct a temperature-gradient-sensitive spatial evolution model, especially strengthening the prediction accuracy for low-temperature regions and regions with large temperature gradients, providing accurate spatial positioning support for cold energy recovery; introducing the coupling analysis of gas discharge flow and temperature field, establishing an influence model of the throttle valve opening change on the evolution of the internal temperature field in the gas storage reservoir, accurately predicting the change of cold energy distribution under different gas discharge strategies.

[0036] As Figure 2 shown, according to one aspect of the present application, the steps of optimizing the full-space temperature field and pressure field include: Performing hierarchical spatial segmentation on the preliminary reconstruction results of the full-space temperature field and pressure field to obtain a multi-scale spatial domain; Based on the multi-scale spatial domain, constructing physical constraint models for different scales to form a multi-scale physical constraint set; Using the data of the temperature sensors on the gas storage reservoir wall in the system basic dataset, constructing a dynamic boundary condition set through adaptive boundary constraints; Combining the dynamic boundary condition set with the sensor measured data and the multi-scale physical constraint set in the system basic dataset, performing physics-guided interpolation reconstruction to optimize the full-space temperature field and pressure field.

[0037] According to one aspect of the present application, the steps of obtaining a multi-scale spatial domain and forming a multi-scale physical constraint set include: Dividing the space of a million-cubic-meter gas storage reservoir into three scale levels: microscale, mesoscale, and macroscale; Creating adaptive grids for each level, forming fine grids in areas with large temperature gradients and coarse grids in areas with gentle temperature changes; Applying the heat diffusion equation at the microscale to describe local heat conduction, using the convective heat transfer model at the mesoscale to depict the heat transfer caused by airflows, and using the stratified flow model at the macroscale to capture the vertical temperature stratification phenomenon; Setting boundary matching conditions for each physical model to ensure the consistency of different scale models in the overlapping regions, thereby forming a multi-scale physical constraint set.

[0038] Specifically, reading the preliminary reconstruction results of the full-space temperature field Ts and pressure field Ps, applying the hierarchical space segmentation algorithm to generate the multi-scale spatial domain division D_scale. First, divide the space of the million-cubic-meter gas storage reservoir into three scale levels: microscale (0.1 - 1m 3 ), mesoscale (1 - 100m 3 ), and macroscale (> 100m 3 ). For each level, use the octree segmentation method to create adaptive grids, forming fine grids in areas with large temperature gradients and coarse grids in areas with gentle temperature changes to ensure efficient allocation of computing resources. Based on the multi-scale spatial domain division D_scale, construct physical constraint models for different scales to form the multi-scale physical constraint set P_constraint. At the microscale, apply the heat diffusion equation (ΨT / Ψt = α ▽ 2 T) to describe local heat conduction; at the mesoscale, use the convective heat transfer model (ΨT / Ψt + u· ▽ T = α ▽ 2 T) to depict the heat transfer caused by airflows; at the macroscale, use the stratified flow model to capture the vertical temperature stratification phenomenon. Set boundary matching conditions for each physical model to ensure the consistency of different scale models in the overlapping regions.

[0039] Combining the measured data of sensors in the system comprehensive monitoring dataset D with the multi-scale physical constraint set P_constraint, perform physics-guided interpolation reconstruction to optimize the full-space temperature field Ts and pressure field Ps. Different from traditional pure data-driven interpolation methods, physical constraints are introduced during the interpolation process, which is specifically implemented as: first, construct a finite element grid with sensor positions as nodes, discretize the physical constraint equations into a system of linear equations, and then use the measured data as boundary conditions to solve the constrained least-squares problem (min‖Ax - b‖ 2+λ‖Lx‖ 2 ) Obtain an interpolation result that satisfies physical constraints, where A is the observation matrix, L is the physical constraint operator, and λ is the balance parameter.

[0040] Construct an adaptive boundary constraint mechanism, and use the data of the temperature sensors on the gas storage reservoir wall in the original spatial distribution data to construct a set of dynamic boundary conditions B_condition. Aiming at the problem that it is difficult to accurately determine the boundary conditions of a million-cubic-meter gas storage reservoir, construct a boundary learning algorithm based on measured data: first extract the time-varying temperature distribution characteristics from the wall sensor data, then use the heat conduction theory to deduce the heat exchange relationship between the wall and the internal fluid, construct a time-varying non-uniform Neumann boundary condition, and finally apply this boundary condition to the reconstruction process, solving the problem of inaccurate boundary condition setting in traditional methods when dealing with large closed spaces. In this embodiment, the accuracy of the reconstruction result is improved from 85% of the traditional method to more than 95%, especially in the area near the wall, and the improvement effect is more significant.

[0041] According to one aspect of the present application, the steps of performing time-sequence and space-coupled evolution prediction to generate a dynamic cold energy distribution prediction map include: Based on the optimized full-space temperature field and pressure field, apply adaptive time window analysis to identify the time dynamic characteristics of different stages and generate a time characteristic segmentation set; For each stage in the time characteristic segmentation set, construct a space evolution model sensitive to temperature gradients to form a segmented space evolution model set; Read the gas release parameters of the throttle valve operation state sequence in the system basic dataset, analyze the influence of the throttle valve opening change on the evolution of the temperature field inside the gas storage reservoir, and establish a gas release-temperature coupling model; Combine the gas release-temperature coupling model with the segmented space evolution model set, and generate a dynamic cold energy distribution prediction map through multi-step progressive prediction.

[0042] According to one aspect of the present application, the steps of generating a time characteristic segmentation set include: Based on the optimized full-space temperature field and pressure field, calculate the statistical characteristics of the full-space temperature change rate, including mean, variance, and kurtosis; Classify the statistical characteristics through time series clustering to automatically identify the fast-changing, transitional, and slow-changing stages; Adjust the time step according to the characteristics of different stages: adopt a small time step within a preset range to improve the prediction accuracy in the fast-changing stage, and adopt a large time step within a preset range to improve the calculation efficiency in the slow-changing stage; finally generate a time characteristic segmentation set.

[0043] Calculate the spatial temperature gradient distribution based on the current temperature field, identify the high-gradient regions, and construct a partial differential equation solver with adaptive weights. Use a finer spatial discretization and a more complex flow model for the high-gradient regions.

[0044] Specifically, based on the optimized full-space temperature field Ts and pressure field Ps, apply adaptive time-window analysis to identify the time-dynamic characteristics of different stages and generate a time-characteristic segmentation set T_segments. First, calculate the statistical characteristics (mean, variance, kurtosis, etc.) of the full-space temperature change rate, and then automatically identify the rapid change stage (such as the initial stage of gas discharge), the transition stage, and the slow change stage (such as the final stage of gas discharge) through the time-series clustering of the change rate characteristics. Assign appropriate time steps to each stage to ensure the prediction accuracy of the rapid change stage and improve the calculation efficiency of the slow change stage. For each stage in the time-characteristic segmentation set T_segments, develop a temperature-gradient-sensitive spatial evolution model to form a segmented spatial evolution model set E_models. Different from the traditional spatio-temporal prediction model that treats all spatial regions equally, this embodiment particularly strengthens the evolution prediction of low-temperature regions and regions with large temperature gradients. The specific implementation is as follows: First, calculate the spatial temperature gradient distribution based on the current temperature field and identify the high-gradient regions; then construct a partial differential equation solver with adaptive weights and use a finer spatial discretization and a more complex flow model for the high-gradient regions; finally, particularly consider the influence of phase change effects and condensation phenomena on temperature evolution in low-temperature regions to ensure the prediction accuracy of the cold energy generation regions.

[0045] Read the gas discharge parameters in the throttle valve operation state sequence, analyze the influence of the throttle valve opening change on the evolution of the internal temperature field of the gas storage reservoir, and establish a gas discharge-temperature coupling model C_model. Treat the throttling process and the evolution of the internal temperature field of the reservoir as a coupled system: First, construct a mathematical model of the throttle valve and calculate the gas discharge flow rate and temperature drop based on the throttle valve opening and the pressure difference before and after; then track the changes in the flow field and temperature field around the gas discharge port and establish a quantitative relationship between the gas discharge flow rate and the local temperature drop; finally, integrate this relationship into the overall temperature field evolution equation to accurately capture the changes in the cold energy distribution under different gas discharge strategies. This embodiment solves the problem that the traditional method ignores the influence of the gas discharge dynamic process on the internal temperature field and provides a more accurate prediction basis for cold energy recovery.

[0046] Integrating the comprehensive segmented space evolution model set E_models and the degassing-temperature coupling model C_model, a multi-step progressive prediction algorithm is adopted to generate the prediction map G of the internal dynamic cold energy distribution in the gas storage reservoir. First, based on the current temperature field and the preset degassing strategy, short-term prediction is carried out using the most suitable time step; then the prediction result is used as the new initial condition, and the prediction process is iteratively executed; finally, by identifying the regions where the temperature is lower than the ambient temperature and their temperature differences, the cold energy density and total amount of each spatial unit are calculated to form a spatiotemporal-resolved cold energy distribution prediction map. Compared with the traditional prediction method, in this embodiment, when predicting the cold energy distribution of a gas storage reservoir with a capacity of millions of cubic meters, the prediction error is reduced from 15 - 20% of the traditional method to 5 - 8%, and especially the prediction accuracy of the position of the cold energy concentration region is increased by more than 40%, providing key support for the precise control of the cold energy recovery device.

[0047] According to one aspect of the present application, the steps of generating a hierarchical control strategy include: S31. Based on the throttle valve operation state sequence and the cold energy recovery device operation data in the system basic dataset, establish a coupling influence model of throttling and cold energy recovery; S32. According to the coupling influence model, design a hierarchical control framework and generate a hierarchical control strategy; S33. Based on the hierarchical control strategy, develop a cooperative control algorithm and output the optimized control signal for the throttle valve and the control signal for the working medium flow rate to form a comprehensive control instruction set.

[0048] Specifically, based on the throttle valve operation state sequence and the cold energy recovery device operation data, establish a coupling influence model C of throttling and cold energy recovery. Analyze the dynamic influence of the change of throttle valve parameters on the performance of the cold energy recovery device to provide a theoretical basis for cooperative control. According to the coupling influence model C, construct a hierarchical control framework and generate a hierarchical control strategy P. Decompose the control task into control layers with different time scales, and reasonably allocate control objectives according to the different dynamic characteristics of the throttle valve and the cold energy recovery device. Construct a cooperative control algorithm and output the optimized control signal Vo for the throttle valve and the control signal Qc for the working medium flow rate. The algorithm comprehensively considers the operation requirements of the turbine and the cold energy recovery efficiency to achieve the balance optimization of the two.

[0049] The specific process of constructing the coupling influence model of throttling and cold energy recovery is as follows: Read the operation status sequence of the throttle valve and the operation data of the cold energy recovery device, and apply time alignment and data preprocessing techniques to generate the coupling analysis dataset C_data. First, perform time standardization processing to ensure the time consistency of different data sources; then use the sliding window method to divide the time series into multiple operating condition segments; finally, perform normalization and outlier detection on each operating condition segment to provide a high-quality data basis for subsequent coupling analysis. Based on the coupling analysis dataset C_data, use the Granger causality analysis method to identify the causal relationship between the throttle valve parameters and the cold energy recovery performance, and construct the dynamic causal network D_causal. Different from traditional correlation analysis, in this embodiment, the causal direction between variables is determined through time lag analysis: First, construct a multi-variable time series model including factors such as throttle valve opening, pressure difference before and after, and flow rate, and indicators such as the inlet temperature of the cold energy recovery device, working medium temperature difference, and recovery efficiency; then calculate the Granger causality index by comparing the prediction errors of the restricted model and the complete model; finally, construct a directed weight network based on the causality index to reveal the dynamic causal relationship between throttle control and cold energy recovery. According to the dynamic causal network D_causal, construct the coupling influence model C of throttling and cold energy recovery. Quantify the identified key causal paths into a mathematical model: First, for each main causal path, extract the transfer function from the input variable (such as the change in throttle valve opening) to the output variable (such as the change in cold energy recovery efficiency); then integrate multiple paths into a state space model to describe the dynamic characteristics of the entire system; finally, verify through historical data and optimize parameters to ensure the accuracy and generalization ability of the model. This embodiment breaks through the limitations of traditional independent control, provides a theoretical basis for the coordinated control of the throttle valve and the cold energy recovery device, and improves the cold energy recovery efficiency by 15 - 25%.

[0050] According to one aspect of the present application, the steps of generating an optimized control strategy for cold energy recovery include: S41. Based on the system basic dataset, construct a fingerprint library of the operating condition characteristics of the compressed air energy storage system; S42. Based on the fingerprint library of the operating condition characteristics, generate an operating condition similarity vector from the current system state through operating condition identification; S43. Combine the dynamic cold energy distribution prediction map to enhance the operating condition similarity based on cold energy characteristics for the operating condition similarity vector to obtain an optimized operating condition similarity vector; S44. Based on the optimized operating condition similarity vector, adaptively generate a cold energy recovery control strategy and output an optimized control strategy for cold energy recovery.

[0051] Specifically, based on the system comprehensive monitoring dataset D, a condition characteristic fingerprint library H of the compressed air energy storage system is constructed. Feature extraction and clustering analysis methods are applied to identify typical condition categories from historical operation data. A condition recognition algorithm is constructed to generate a condition similarity vector S from the current system state. The similarity between the current condition and each typical condition in the condition characteristic fingerprint library H is calculated to provide a basis for subsequent control strategy generation. A method for enhancing condition similarity based on cold energy characteristics is constructed to improve the pertinence of similar condition recognition. The specific implementation is as follows: aiming at the deficiencies of traditional similar condition recognition in cold energy recovery applications, a thermodynamic similarity index system is constructed, and feature dimensions centered on cold energy generation and transfer, such as temperature gradient distribution similarity, cold and heat interface movement pattern similarity, and cold energy generation rate similarity, are introduced into traditional condition similarity calculations to solve the problem of the mismatch between the focus of traditional methods and the cold energy recovery requirements; a staged similarity weighting strategy is constructed to dynamically adjust the weights of each feature in similarity calculations according to different stages of the cold energy recovery process (initial rapid cooling stage, stable recovery stage, and end-of-stage inefficient recovery stage) to improve the recognition accuracy of key stages; cold energy recovery efficiency correlation analysis is introduced to establish a mapping relationship between condition similarity and cold energy recovery efficiency, so that similarity calculations directly point to the goal of optimizing cold energy recovery efficiency.

[0052] Based on the condition similarity vector S, a method for adaptively generating a cold energy recovery control strategy is constructed, and an optimized cold energy recovery control strategy O is output. The specific implementation is as follows: aiming at the limitations of traditional control strategy generation methods based on similar conditions, a key factor extraction technology for cold energy recovery is constructed to extract control decision-making elements highly correlated with cold energy recovery efficiency from the control experience of historical similar conditions, avoiding suboptimal control caused by undifferentiated migration in traditional methods; a throttling-recovery dual-objective optimization framework is constructed, treating the throttle valve control and the cold energy recovery device control as a coupled optimization problem, and maximizing the cold energy recovery amount under the premise of meeting the turbine inlet parameter requirements; a cold energy quality evaluation mechanism is established to comprehensively evaluate according to the temperature level (quality) and quantity of cold energy, guiding the control strategy generation process to give priority to the recovery of high-quality cold energy.

[0053] The optimized cold energy recovery control strategy O and the hierarchical control strategy P are fused, and the final comprehensive control instruction set I is output, including the throttle valve control signal and the cold energy recovery device control signal. The two types of strategies are integrated through a model predictive control framework to ensure the stable operation of the system while maximizing the cold energy recovery efficiency.

[0054] As Figure 3 shown, according to one aspect of the present application, the steps for obtaining an optimized condition similarity vector include: Based on the system basic dataset and the dynamic cold energy distribution prediction map, cold energy characteristic indicators are extracted to construct a cold energy feature vector set; Based on the cold energy feature vector set, construct a similarity calculation formula for cold energy recovery through thermodynamic similarity indicators, and generate a thermodynamic similarity indicator set; Construct a stage similarity weighting strategy according to the characteristics of the cold energy recovery process, and generate a stage weighting function set; Read the operation data of the cold energy recovery device in the system basic dataset, analyze the relationship between operating condition similarity and cold energy recovery efficiency, and establish an efficiency correlation mapping; Combine the thermodynamic similarity indicator set, the stage weighting function set and the efficiency correlation mapping, calculate the enhanced similarity of the current and historical operating condition similarity vectors, and output the optimized operating condition similarity vector.

[0055] According to one aspect of the present application, the thermodynamic similarity indicator set includes: The temperature gradient distribution similarity indicator, obtained by calculating the cosine similarity of the temperature gradient fields of two operating conditions; The cold and hot interface movement pattern similarity indicator, obtained by comparing the moving speed and direction of the cold and hot dividing interfaces; The cold energy generation rate similarity indicator, obtained by comparing the change patterns of the cold energy generation amounts per unit time.

[0056] Divide the cold energy recovery process into an initial rapid cooling stage, a stable recovery stage and a final low-efficiency recovery stage, and design specific weight functions for each stage, so that the similarity calculation can automatically adjust the importance of each feature according to the current stage, and form a stage weighting function set.

[0057] Specifically, based on the system comprehensive monitoring dataset D and the dynamic cold energy distribution prediction map G, extract cold energy characteristic indicators and construct a cold energy feature vector set F_cold. Aiming at the problem that traditional operating condition characteristics cannot fully express cold energy characteristics, a cold energy feature extraction method is constructed: first calculate the spatial temperature gradient distribution index to quantify the clarity and position of the cold and hot interfaces; then analyze the time derivative distribution of the temperature field to characterize the rate and stability of cold energy generation; finally, combine the spatial distribution pattern of cold energy density and the total change trend to form a multi-dimensional feature vector that comprehensively depicts cold energy characteristics.

[0058] Construct a thermodynamic similarity indicator system, design a similarity calculation formula for cold energy recovery, and generate a thermodynamic similarity indicator set H_index. Traditional operating condition similarity mainly focuses on general operating characteristics (such as pressure, flow rate, etc.). In this embodiment, a feature dimension with cold energy as the core is introduced: first define the "temperature gradient distribution similarity" and calculate the cosine similarity of the temperature gradient fields of two operating conditions; then design the "cold and hot interface movement pattern similarity" to compare the moving speed and direction of the cold and hot dividing interfaces; finally create the "cold energy generation rate similarity" to compare the change patterns of the cold energy generation amounts per unit time. These indicators directly target the core characteristics of cold energy recovery and improve the pertinence of similar operating condition identification.

[0059] Based on the characteristics of the cold energy recovery process, a stage similarity weighted strategy is developed to generate a stage weighted function set W_phase. A dynamically adjusted weight allocation scheme is constructed for the characteristic differences in different stages of the cold energy recovery process: First, the cold energy recovery process is divided into an initial rapid cooling stage, a stable recovery stage, and a final inefficient recovery stage; then, the key influencing factors of each stage are analyzed. For example, the temperature drop rate is focused on in the initial stage, the heat exchange efficiency is focused on in the stable stage, and the energy recovery balance point is focused on in the final stage; finally, specific weight functions are designed for each stage, enabling the similarity calculation to automatically adjust the importance of each feature according to the current stage.

[0060] Read the historical operation data of the cold energy recovery device, analyze the relationship between operating condition similarity and cold energy recovery efficiency, and establish an efficiency correlation mapping R_map. Directly associate the similar operating condition identification with the recovery efficiency target: First, extract the corresponding relationship between the operating condition characteristics and the actual recovery efficiency from the historical data; then, use the supervised learning method to train the mapping model from the operating condition characteristics to the recovery efficiency; finally, further adjust the feature weights in the similarity calculation according to the efficiency sensitive factors predicted by the model, making the similarity calculation result more directly point to the direction of the optimal cold energy recovery efficiency. Combining the thermodynamic similarity index set H_index, the stage weighted function set W_phase, and the efficiency correlation mapping R_map, calculate the enhanced similarity between the current operating condition and the historical operating conditions, and output the operating condition similarity vector S. This embodiment improves the accuracy of cold energy recovery operating condition identification from 70% of the traditional method to over 90%, providing a high-quality reference for similar operating conditions for subsequent control strategy generation.

[0061] As Figure 4 shown, according to one aspect of the present application, the steps for adaptively generating a cold energy recovery control strategy and outputting an optimized cold energy recovery control strategy include: Based on the optimized operating condition similarity vector, retrieve the control strategies of similar operating conditions from the historical control database, extract the key control decisions, and generate a candidate control strategy set; Identify the control decision elements related to the cold energy recovery efficiency from the candidate control strategy set, and construct a cold energy efficiency impact factor set; Based on the coupling impact model and the cold energy efficiency impact factor set, construct a throttling-recovery dual-objective optimization framework, taking the throttle valve control and the cold energy recovery device control as a coupled optimization problem, and construct a multi-objective optimization model; Combine the multi-objective optimization model with the current system state, perform real-time optimization calculations, and generate an optimized cold energy recovery control strategy for the current operating condition, which is used to fuse with the hierarchical control strategy to form the comprehensive control instruction set.

[0062] Specifically, based on the operating condition similarity vector S, retrieve the control strategies of similar operating conditions from the historical control database, extract the key control decisions, and generate the candidate control strategy set C_strategies. Different from the traditional method of simply applying the control strategy of the most similar operating condition, this embodiment adopts a more refined strategy extraction method: first, select the top N historical operating conditions in terms of similarity; then decompose the control strategy of each operating condition to identify key control decisions such as the throttle valve opening trajectory and the working fluid flow regulation mode; finally, evaluate the effect of each decision in the historical scenario, and retain the control elements that have a significant positive impact on the cold energy recovery efficiency to form the candidate strategy set.

[0063] Construct a key factor extraction technology for cold energy recovery, identify the control decision elements highly related to the cold energy recovery efficiency from the candidate control strategy set C_strategies, and construct the cold energy efficiency impact factor set E_factors. Directly associate the control strategy with the cold energy recovery efficiency: first, conduct a fine-grained analysis of the implementation effect of the historical strategy to quantify the influence degree of different control parameters on the cold energy recovery efficiency; then use the sensitivity analysis method to identify the key control parameters and their optimal value ranges; finally, construct a response surface model of the control parameters and the recovery efficiency to provide a theoretical basis for subsequent optimization.

[0064] Construct a throttling-recovery dual-objective optimization framework, treat the throttle valve control and the cold energy recovery device control as a coupled optimization problem, and construct a multi-objective optimization model M_opt. Aiming at the limitation of single-objective optimization in the traditional method, construct a collaborative optimization method: first, based on the coupling influence model C and the cold energy efficiency impact factor set E_factors, establish a mathematical model describing the relationship between the throttle valve parameters, the cold energy recovery device parameters and the system performance (power generation efficiency, cold energy recovery efficiency); then define a multi-objective function including the stability of the turbine inlet parameters and the maximization of cold energy recovery; finally, design a solution algorithm based on Pareto optimality to maximize the cold energy recovery amount on the premise of meeting the requirements of the turbine inlet parameters.

[0065] Based on the current system state and the multi-objective optimization model M_opt, perform real-time optimization calculations to generate the cold energy recovery optimization control strategy O for the current operating condition. Combine the theoretical model with the current actual operating condition: first, read the real-time system state and update the initial conditions and constraint conditions in the optimization model; then perform model predictive control (MPC) calculations to predict the effects of different control strategies in the future for a period of time; finally, select the control sequence that can maximize the cold energy recovery while meeting the requirements of turbine operation to form a complete control strategy. This embodiment realizes the synergistic effect of throttling control and cold energy recovery through coupled optimization. Compared with the traditional separate control, the cold energy recovery efficiency is increased by 20 - 30%, and at the same time, the stable operation of the turbine system is ensured.

[0066] In a specific embodiment of the present application, it is based on the dynamic utilization of cold energy of a certain million-cubic-level compressed air energy storage system. This embodiment is based on a certain underground rock salt cavern gas storage reservoir with a gas storage capacity of 1.2×10 6 m 3 , a working pressure range of 4.5 - 8.0 MPa, a system rated power of 100 MW, and a designed gas release duration of 10 hours. The specific steps are as follows: Step 1: Obtain multi-source monitoring data of the compressed air energy storage system and construct a system basic data set.

[0067] S11: Collect multi-point temperature and pressure data inside the gas storage reservoir to construct a spatial distribution monitoring data set.

[0068] 120 measuring points are arranged inside the gas storage reservoir, including 86 temperature sensors (T1, T2,..., T 86 ) and 34 pressure sensors (P1, P2,..., P 34 ). The position of each sensor is represented by spatial coordinates (X, Y, Z), and the sampling frequency is 1 time per 10 seconds. Taking part of the temperature data at t = 0 during a certain gas release process as an example: T1(50, 30, 20) = 25.3 °C (coordinate unit: m, the origin is located at the center of the bottom of the gas storage reservoir); T2(50, 30, 40) = 24.8 °C; T3(50, 30, 60) = 24.1 °C...; T 86 (750, 500, 300) = 23.6 °C; Example of pressure data: P1(50, 50, 50) = 7.85 MPa; P2(150, 150, 50) = 7.84 MPa...; P 34 (750, 750, 300) = 7.82 MPa.

[0069] S12: Obtain the operating status data of the throttle valve to form a throttle control data set.

[0070] The system is configured with 3 parallel throttle valves, and their opening signals (Vo1, Vo2, Vo3), pressure differences before and after the throttle valves (ΔP1, ΔP2, ΔP3), temperature differences (ΔT1, ΔT2, ΔT3), and flow rate data (Q1, Q2, Q3) are collected, with a sampling frequency of 1 time per second. Example of data at the initial moment: Vo1 = 35%, ΔP1 = 3.42 MPa, ΔT1 = 68.5 °C, Q1 = 48.3 kg / s; Vo2 = 30%, ΔP2 = 3.40 MPa, ΔT2 = 67.8 °C, Q2 = 41.5 kg / s; Vo3 = 0%, ΔP3 = 0 MPa, ΔT3 = 0 °C, Q3 = 0 kg / s.

[0071] S13. Monitor the operating parameters of the cold energy recovery device and establish a dataset of cold energy recovery characteristics.

[0072] The cold energy recovery device uses an ethylene glycol aqueous solution as the working medium, and parameters such as the inlet temperature (Tci), outlet temperature (Tco), working medium flow rate (Qc), and heat transfer efficiency (η) are collected. The sampling frequency is 1 time per 5 seconds. Example data at the initial moment: Tci = 0°C, Tco = -20.5°C, Qc = 25.0 kg / s, η = 0.85.

[0073] S14. Integrate multi-source data, perform time alignment and cleaning, and output a standardized system comprehensive monitoring dataset D.

[0074] Perform time alignment on data with different sampling frequencies. Using 1 time per 10 seconds as the standard sampling frequency, perform average downsampling on the throttling control dataset and use linear interpolation to complete missing data. Finally, form the system comprehensive monitoring dataset D.

[0075] Step 2. Establish an internal space-time dynamic model of the gas storage reservoir to accurately depict the cold energy distribution characteristics.

[0076] S21. Apply the kernel function interpolation method to preliminarily reconstruct the temperature field and pressure field in the entire space.

[0077] Use the radial basis function (RBF) interpolation method to preliminarily reconstruct the temperature field Ts(x, y, z, t) and pressure field Ps(x, y, z, t) in the entire space. The calculation method is as follows: The calculation formula for preliminarily reconstructing the temperature field: Ts(x, y, z, t) = ∑[i = 1 to n] wi × φ(||r - ri||); where φ(r) = (r 2 + c 2 ) 1 / 2 is the radial basis function; r is the coordinate vector (x, y, z) of the target point; ri is the coordinate vector of the i-th sensor; wi is the weight coefficient; c is the shape parameter, with a value of 1.5 times the average sensor spacing, which is 25 m in this example; n is the number of sensors. Solve the linear equation system to find the weight coefficient wi so that the value of the interpolation function at the sensor position is equal to the measured value. At a specific moment (t = 0), the values of the preliminarily reconstructed temperature field at some grid points are: Ts(100, 100, 50, 0) = 24.6°C; Ts(200, 200, 100, 0) = 24.2°C; Ts(300, 300, 150, 0) = 23.9°C.

[0078] S22. Apply the multi-scale heat flow constraint space reconstruction method to optimize the temperature field and pressure field in the entire space.

[0079] S221. Apply the hierarchical space partitioning algorithm to divide the gas storage space into multi-scale domains. According to the magnitude of the temperature gradient, the gas storage space is divided into three scales: micro-scale (0.1 - 1m 3 ): mainly distributed near the gas storage wall surface and the throttle valve outlet area; meso-scale (1 - 100m 3 ): mainly distributed in the area with a large vertical temperature gradient; macro-scale (> 100m 3 ): the area with a gentle temperature change. Use the octree algorithm for space partitioning, with an initial grid size of 50m, and adaptively refine according to the temperature gradient. The area with a temperature gradient greater than 0.1 °C / m is refined to a 5m grid size.

[0080] S222. Construct physical constraint models for different scales to form a multi-scale physical constraint set. For different scale regions, apply different physical models: for the micro-scale, apply the heat diffusion equation: ΨT / Ψt = α ▽ 2 T; where T is the temperature (°C); t is the time (s); Ψ is the partial derivative; α is the heat diffusion coefficient, which is approximately 2.2×10 -5 m 2 / s for air at normal pressure; ▽ 2 is the Laplace operator, ▽ 2 T = Ψ 2 T / Ψx 2 + Ψ 2 T / Ψy 2 + Ψ 2 T / Ψz 2 . For the meso-scale, use the convective heat transfer model: ΨT / Ψt + u· ▽ T = α ▽ 2 T; where u is the velocity vector (m / s), estimated from the pressure gradient; ▽ T is the temperature gradient, ▽ T = (ΨT / Ψx, ΨT / Ψy, ΨT / Ψz). For the macro-scale, use the stratified flow model. The vertical temperature distribution equation in the stratified flow model: T(z) = Tb + (Tt - Tb) × [1 - exp(-βz)]; where Tb is the bottom temperature (°C); Tt is the top temperature (°C); z is the normalized height; β is the stratification parameter, obtained by fitting the measured data. In this example, β = 2.3. At the scale junction, set the continuity boundary conditions to ensure the continuity of the temperature field and the heat flux density.

[0081] S223. Perform physically-guided interpolation reconstruction to optimize the temperature field and pressure field in the entire space. Discretize the physical constraint equation into a linear equation system and jointly solve the interpolation problem: Physically-guided interpolation reconstruction: min‖Ax - b‖ 2 + λ‖Lx‖ 2 ; where A is the observation matrix with dimensions n×m, n is the number of sensors, and m is the number of reconstructed grid points; x is the temperature field vector to be solved; b is the vector of measured temperatures by the sensors; L is the physical constraint operator matrix constructed according to the physical models of each region; λ is the balance parameter with a value of 0.15. Solve the above least squares problem using the conjugate gradient method to obtain the optimized temperature field. Compared with the preliminary reconstruction, the temperature reconstruction accuracy in the throttle valve outlet area is improved by about 12%.

[0082] S224. Construct an adaptive boundary constraint mechanism and construct a set of dynamic boundary conditions. Use the data of the temperature sensors on the gas storage wall to construct the dynamic boundary condition: -k ▽ T·n = h(Tw - T); where k is the thermal conductivity of the medium (W / m·K); ▽ T is the temperature gradient; n is the wall normal vector; h is the convective heat transfer coefficient (W / m 2 ·K), obtained by fitting historical data with a value of 5.8; Tw is the wall temperature (°C); T is the fluid temperature (°C). After applying this boundary condition, the reconstruction accuracy of the temperature field near the wall is improved from 85% to 96%.

[0083] S23. Construct a time-space coupled evolution prediction algorithm to generate a prediction map of the dynamic cold energy distribution.

[0084] S231. Apply adaptive time window analysis to identify the time dynamic characteristics of different stages. Calculate the statistical characteristics of the temperature change rate in the entire space: mean (μ): the average temperature change rate of each grid point; variance (σ 2 ): the spatial dispersion degree of the temperature change rate; kurtosis (κ): reflecting the distribution of extreme temperature changes. The gas discharge process is divided into three stages: rapid change stage (0 - 1.5 h): μ = -0.52 °C / min, σ 2 = 0.125, κ = 3.8; transition stage (1.5 - 4 h): μ = -0.28 °C / min, σ 2 = 0.063, κ = 2.6; slow change stage (4 - 10 h): μ = -0.11 °C / min, σ 2 = 0.024, κ = 2.1. According to the characteristics of each stage, assign different prediction time steps: rapid change stage: time step is 30 seconds; transition stage: time step is 2 minutes; slow change stage: time step is 5 minutes.

[0085] S232. Develop a spatially evolving model sensitive to temperature gradients to form a set of segmented spatially evolving models. For different temperature gradient regions, set different model parameters and grid densities: Temperature gradient sensitive weight function: w( ▽ T) = 1 + α ×| ▽ T| 2 ; where w is the weight factor; ▽ T is the temperature gradient; α is the sensitivity parameter, with a value of 100 (°C / m) -2 . Adjust the spatially evolving model according to the weight factor: In high-gradient regions (near the gas outlet, | ▽ T| > 0.2 °C / m): Use a fine grid (grid size 3m) and a fully coupled convection model; In medium-gradient regions (0.05 °C / m < | ▽ T| < 0.2 °C / m): Use a medium grid (grid size 10m) and a simplified convection model; In low-gradient regions (| ▽ T| < 0.05 °C / m): Use a coarse grid (grid size 30m) and a heat diffusion model.

[0086] S233. Analyze the influence of the throttle valve opening change on the evolution of the internal temperature field of the gas storage reservoir and establish a gas release-temperature coupling model. Throttle valve dynamic model: ΔT = f(P1, P2, Vo); where ΔT is the throttle temperature drop (°C); P1 is the pressure before throttling (MPa); P2 is the pressure after throttling (MPa); Vo is the valve opening (%); f is the coupling function, obtained by fitting experimental data. In this embodiment, the throttle temperature drop calculation formula is: ΔT = k × (P1 - P2) × [1 - 0.08 × (1 - Vo / 100) 2 ; where k is the proportionality coefficient, approximately 20 °C / MPa; the throttle valve opening Vo ranges from 0 to 100%. Based on this model, when the throttle valve opening is adjusted from 35% to 40%, the predicted temperature drop changes from 65.2 °C to 62.8 °C, with a difference from the measured value of less than 2 °C.

[0087] S234. Adopt a multi-step progressive prediction algorithm to generate a dynamic cold energy distribution prediction map. Couple the temperature field prediction, gas release flow, and temperature drop model, and perform multi-step progressive prediction: Based on the current temperature field and throttle parameters, predict the temperature distribution at the next time step; Update the boundary conditions and internal heat sources; Use the prediction result as the initial state for the next step and continue the prediction. Cold energy density calculation: E_cold(x, y, z, t) = ρ × cp × max(0, T_amb - T(x, y, z, t)); where E_cold is the cold energy density (J / m 3 ); ρ is the air density (kg / m 3) ; cp is the specific heat capacity of air at constant pressure (J / kg·K); Tamb is the ambient reference temperature, which is 20°C; T(x, y, z, t) is the predicted temperature field. The generated dynamic cold energy distribution prediction map shows that the cold energy is mainly concentrated near the throttle valve outlet and the top area of ​​the reservoir, with the maximum cold energy density reaching 8.5×10 5 J / m 3 The average error between the predicted cold energy distribution and the actual measured value is 7.6%, which is better than the 18.5% error of the traditional method.

[0088] Step 3: Develop a multi-time scale hierarchical control framework to achieve coordinated control of the throttling valve and the cold energy recovery device.

[0089] S31. Establish a coupling impact model of throttling and cold energy recovery.

[0090] S311. Generate coupling analysis data set. Time alignment and preprocessing of throttle valve operation state sequence and cold energy recovery device operation data: standardize the data to the interval [-1, 1]; use sliding window method (window width is 10 minutes, step length is 1 minute) to split the time series; use 3σ criterion to eliminate outliers.

[0091] S312. Use Granger causal analysis method to identify causal relationship. Establish causal relationship between throttle valve parameters and cold energy recovery performance: Granger causal index calculation: GCI(X→Y) = ln[var(ε_r) / var(ε_ur)]; GCI is Granger causal index; var(ε_r) is the prediction error variance of the restricted model; var(ε_ur) is the prediction error variance of the complete model; X is the dependent variable, such as throttle valve opening; Y is the result variable, such as cold energy recovery efficiency. The analysis results show that the Granger causal index of throttle valve opening to cold energy recovery inlet temperature is 2.36, indicating that the change in opening has a significant causal relationship with the change in inlet temperature after 3-5 minutes.

[0092] S313. Construct a coupling influence model of throttling and cold energy recovery. Based on the results of causal analysis, a state space model is established: State space model: X(t+1) = A·X(t) + B·U(t), Y(t) = C·X(t) + D·U(t); where X is the state vector, including cold energy density, temperature, etc.; U is the control vector, including throttle valve opening, working fluid flow, etc.; Y is the output vector, including recovery efficiency, recovery power, etc.; A, B, C, D are system matrices, obtained through historical data identification. Model verification shows that the coupling model has a prediction accuracy of 92% for cold energy recovery efficiency, which is 15% higher than the traditional independent model.

[0093] S32. Build a hierarchical control framework and generate a hierarchical control strategy.

[0094] Based on the different time scales of the system's dynamic characteristics, a three-layer control framework is designed: Policy layer (time scale: hourly level): responsible for overall operation strategy planning; Coordination layer (time scale: minute level): coordinates the cooperation between throttling and cold energy recovery; Execution layer (time scale: second level): realizes the basic PID control function. The output of each layer of controller serves as the input or constraint of the lower layer of controller, forming a hierarchical control strategy.

[0095] S33. Construct a cooperative control algorithm and output a control signal.

[0096] The model predictive control (MPC) method is used to achieve cooperative control: Objective function: J = w1·J_throttle + w2·J_recovery; where J_throttle is the throttling control objective function to ensure the stability of the turbine inlet parameters; J_recovery is the cold energy recovery objective function to maximize the cold energy recovery; w1 and w2 are weight coefficients, which are dynamically adjusted according to the operation stage. Initially, w1 = 0.7 and w2 = 0.3. In the middle stage, w1 = w2 = 0.5. In the later stage, w1 = 0.3 and w2 = 0.7. The constraint conditions include: Turbine inlet temperature range: -5°C ≤ T_in ≤ 5°C; Turbine inlet pressure fluctuation: |dP / dt| ≤ 0.05 MPa / min; Working medium flow rate range: 10 kg / s ≤ Qc ≤ 40 kg / s; Safety constraint of the cold energy recovery device: Tco ≥ -35°C. By solving this optimization problem, the optimized control signal of the throttle valve and the control signal of the working medium flow rate are obtained.

[0097] Step 4. Construct a method for identifying similar working conditions and generating control strategies to optimize the cold energy recovery efficiency under different working conditions.

[0098] S41. Construct a fingerprint library of the working condition characteristics of the compressed air energy storage system.

[0099] Based on historical operation data, using the principal component analysis (PCA) and clustering methods, 8 typical working conditions are identified, including: High-pressure rapid air release working condition; High-pressure stable air release working condition; Medium-pressure rapid air release working condition; Medium-pressure stable air release working condition; Low-pressure rapid air release working condition; Low-pressure stable air release working condition; Pressure fluctuation air release working condition; End-stage low-pressure air release working condition. Feature vectors are extracted for each working condition, including 20 feature parameters, to form a fingerprint library of working condition characteristics.

[0100] S42. Develop a working condition identification algorithm to generate a working condition similarity vector.

[0101] The cosine similarity method is used to calculate the similarity between the current working condition and the typical working conditions in the library: Cosine similarity calculation: sim(A, B) = (A·B) / (||A||·||B||); where A is the feature vector of the current working condition; B is the feature vector of the typical working condition; · represents the dot product of vectors; ||A|| represents the Euclidean norm of vector A. For a certain real-time working condition, an example of the calculated similarity vector: S = [0.78, 0.92, 0.45, 0.33, 0.26, 0.21, 0.15, 0.09] indicates that the current working condition is most similar to the second type of "high-pressure stable gas discharge working condition".

[0102] S43. Design a method to enhance the similarity of working conditions based on cold energy characteristics.

[0103] S431. Extract cold energy characteristic indicators and construct a cold energy feature vector set. From the system data and the cold energy distribution prediction map, extract the following cold energy characteristic indicators: Spatial temperature gradient distribution index: μ_grad = 0.15 °C / m, σ_grad = 0.04 °C / m; Clarity of the cold and hot interface: CI = 0.83 (range 0 - 1, the larger the value, the clearer the interface); Cold energy generation rate: CR = 2.8 MW; Spatial distribution pattern of cold energy density: mainly concentrated in the top and outlet areas; Total cold energy change trend: stable increase, rate about 1.5 MJ / s; These indicators form the cold energy feature vector F_cold.

[0104] S432. Develop a thermodynamic similarity index system. Construct a similarity calculation formula for cold energy recovery: Temperature gradient distribution similarity: Sim_grad(A, B) = ( ▽ TA· ▽ TB) / (|| ▽ TA||·|| ▽ TB||); where ▽ TA, ▽ TB are the temperature gradient fields of working conditions A and B. Cold and hot interface movement pattern similarity: Sim_interface(A, B) = exp(-||VA - VB|| 2 / σ 2 ); where VA, VB are the cold and hot interface movement speed vectors of working conditions A and B; σ is the normalization parameter, with a value of 0.5 m / s. Cold energy generation rate similarity: Sim_rate(A, B) = 1 - |CRA - CRB| / max(CRA, CRB); where CRA, CRB are the cold energy generation rates of working conditions A and B.

[0105] S433. Develop a similarity weighting strategy for different development stages. Divide the cold energy recovery process into three stages and construct different weight functions: Initial rapid cooling stage (t < 1.5 h): W_init(t) = [0.5, 0.15, 0.35]; the weights correspond to temperature gradient, interface movement, and production rate similarity respectively. Stable recovery stage (1.5 h ≤ t < 8 h): W_stable(t) = [0.2, 0.3, 0.5]; pay more attention to the similarity of cold energy production rate. Final low-efficiency recovery stage (t ≥ 8 h): W_end(t) = [0.15, 0.5, 0.35]; pay more attention to the similarity of interface movement pattern.

[0106] S434. Establish an efficiency correlation mapping and calculate the enhanced similarity. Analyze the relationship between historical operating conditions and recovery efficiency, and establish a mapping model: Efficiency prediction model: η_pred = Σ[i = 1 to 3] βi · Simi(A, B); where η_pred is the predicted recovery efficiency; Simi is the i-th similarity index; βi is the regression coefficient, obtained by training with historical data, which are [0.35, 0.25, 0.4] respectively. Based on this model, adjust the final weight of similarity calculation and output the operating condition similarity vector S. Practice shows that the accuracy of enhanced operating condition recognition has been improved from 70% to 91%.

[0107] S44. Develop an adaptive generation method for the cold energy recovery control strategy.

[0108] S441. Retrieve the control strategies of similar operating conditions and generate a candidate control strategy set. Retrieve the top 5 operating conditions with the highest similarity from the historical control database and extract their control strategies: Throttle valve opening trajectory: Vo(t) = f1(t); Working fluid flow rate adjustment mode: Qc(t) = f2(t); Working fluid temperature set point: Tset(t) = f3(t).

[0109] S442. Extract the control decision-making elements highly correlated with the cold energy recovery efficiency. Through sensitivity analysis, identify the key control parameters: Sensitivity calculation: S(η, p) = (Ψη / Ψp)·(p / η); where S is the sensitivity; η is the cold energy recovery efficiency; p is the control parameter; Ψη / Ψp is the partial derivative of efficiency with respect to the parameter. The analysis results show that the sensitivity of the working fluid flow rate to the recovery efficiency is 0.78, which is the most critical parameter; the sensitivity of the throttle valve opening change rate is 0.65, which is the second most critical parameter.

[0110] S443. Construct a throttling-recovery dual-objective optimization framework. Build a multi-objective optimization model: objective function: min[-w1·η_recovery - w2·s_stability]; where η_recovery is the cold energy recovery efficiency, ranging from 0 to 1; s_stability is the stability index of the turbine inlet parameters, ranging from 0 to 1; w1 and w2 are weight coefficients, taking 0.6 and 0.4 respectively. The constraint conditions include technical constraints and safety constraints, such as valve action speed limits, working fluid temperature safety limits, etc.

[0111] S444. Perform real-time optimization calculations to generate an optimized control strategy for cold energy recovery. Apply the genetic algorithm to solve the multi-objective optimization problem to obtain the Pareto optimal solution set, and then select the most suitable control strategy point according to the current working conditions. The finally obtained optimized control strategy for cold energy recovery includes: the throttling valve opening adjustment curve; the cold energy recovery working fluid flow rate adjustment curve; the working fluid temperature set point curve.

[0112] S45. Integrate the control strategies and output a comprehensive control instruction set.

[0113] Integrate the hierarchical control strategy and the optimized control strategy for cold energy recovery to generate a comprehensive control instruction set: directly adopt their respective strategies when the objectives are consistent; when the objectives conflict, use the weighted average method to resolve the conflict: U_final = α·U_layer + (1-α)·U_recovery; where U_final is the final control instruction; U_layer is the output of the hierarchical control strategy; U_recovery is the output of the optimized control strategy for cold energy recovery; α is the weight coefficient, defaulting to 0.5. The comprehensive control instruction set outputs the throttling valve control signal and the cold energy recovery device control signal at a frequency of 1 second / time.

[0114] This embodiment has been applied and verified on an actual million-cubic-meter compressed air energy storage system, achieving remarkable results compared with traditional methods: the average error of the temperature field reconstruction accuracy of the traditional pure interpolation method is 18.5%, and the average error of this embodiment is reduced to 7.6%. Especially in the area with sparse measurement points, the reconstruction accuracy is improved by more than 40%. The cold energy recovery efficiency under the traditional independent control method is 62%, and after adopting collaborative control in this embodiment, the cold energy recovery efficiency is increased to 81%, an increase of 19 percentage points. The turbine inlet temperature fluctuation under the traditional method is ±4.2°C, and this embodiment reduces it to ±1.8°C, with the stability improved by 57%. The system cycle efficiency under the traditional method is 55.6%, and this embodiment improves the system cycle efficiency to 59.8% through efficient cold energy recovery, with a relative increase of 7.5%. The accuracy of the traditional working condition identification based on general characteristics is 70%, and the accuracy of the working condition identification based on cold energy characteristics in this embodiment reaches 91%. During the 10-hour air release period, the average deviation of the predicted cold energy position by the traditional method is 35m, and this embodiment reduces it to 12m, improving the positioning accuracy by 65%.

[0115] The following shows the comparison of cold energy recovery efficiency during a certain air release process: during the air release period from 0 to 2 hours, the efficiency of the traditional method is 58.5%; the efficiency of this embodiment is 76.2%; the improvement amplitude is 17.7%; during the air release period from 2 to 5 hours, the efficiency of the traditional method is 64.3%; the efficiency of this embodiment is 85.7%; the improvement amplitude is 21.4%; during the air release period from 5 to 8 hours, the efficiency of the traditional method is 66.1%; the efficiency of this embodiment is 83.8%; the improvement amplitude is 17.7%; during the air release period from 8 to 10 hours, the efficiency of the traditional method is 55.2%; the efficiency of this embodiment is 72.6%; the improvement amplitude is 17.4%; the average efficiency of the traditional method is 62.0%; the average efficiency of this embodiment is 81.0%; the evaluation improvement amplitude is 19.0%; According to one aspect of the present application, the technical implementation of the multi-scale heat flow constraint space reconstruction method is specifically as follows: in practical applications, only 120 measurement points are arranged in a million-cubic-meter gas storage reservoir, with only one measurement point per 10,000 cubic meters on average, and the spatial measurement point distribution is extremely sparse. It is difficult for traditional interpolation methods to accurately reconstruct the temperature distribution of such a large space, especially in areas with large temperature gradients. By introducing multi-scale physical constraints, the detailed implementation steps are as follows: Octree space segmentation method: The space is divided by using the recursive octant method. The initial cube side length is 100m, and it is adaptively subdivided according to the temperature gradient. The subdivision conditions are: if the temperature gradient | ▽ T| > 0.2°C / m, then it is subdivided to a side length of 5m; if 0.05°C / m < | ▽ T| < 0.2°C / m, then it is subdivided to a side length of 20m; if | ▽If T < 0.05℃ / m, then maintain the large-scale grid (side length ≥ 50m). Method for applying multi-physics models: Use physical models with different complexities for different scale regions and dynamically balance calculation accuracy and efficiency during the calculation process. The heat flow constraint is solved using the finite difference method, and the discrete format is: Discrete format of the micro-scale heat diffusion equation: (T n +1_i，j，k - T n_i，j，k ) / Δt = α[(T n_i+1，j，k + T n_i-1，j，k - 2T n_i，j，k ) / (Δx) 2 + (T n_i，j+1，k + T n_i，j-1，k -2T n_i，j，k ) / (Δy) 2 + (T n_i，j，k+1 + T n_i，j，k-1 - 2T n _i,j,k) / (Δz) 2 ; where T n_i，j，k is the temperature at time step n and spatial point (i, j, k); Δt is the time step size; Δx, Δy, Δz are the spatial step sizes; α is the thermal diffusivity. Core algorithm for physics-guided interpolation: Construct a global objective function considering physical constraints: J(T) = w1·J_data(T) + w2·J_phys(T) + w3·J_smooth(T); where J_data is the data fitting term, J_data(T) = Σ||T(x_i) - T_i|| 2 ; J_phys is the physical constraint term, constructed using different physical equations for different regions; J_smooth is the smoothing regularization term; w1, w2, w3 are weight coefficients, with values [0.6, 0.3, 0.1] in the example. By solving this optimization problem, a temperature field that satisfies physical constraints and fits the measurement data is obtained. In this embodiment, the reconstruction accuracy in the region where the temperature gradient is greater than 0.1℃ / m is improved by 40%.

[0116] The specific implementation of the time-series - space coupling evolution prediction algorithm is as follows: Traditional prediction methods often use a unified time step size, making it difficult to balance calculation efficiency and prediction accuracy. This embodiment constructs a non-uniform time step size mechanism, and the specific implementation is as follows: Method for determining the adaptive time step size: Automatically adjust the time step size according to the temperature change rate, and the calculation formula is: Δt_adaptive = Δt_base × min(1, (|dT / dt|_threshold / |dT / dt|_max) β) where Δt_adaptive is the adaptive time step; Δt_base is the base time step, set to 300 seconds; |dT / dt|_threshold is the threshold temperature change rate, set to 0.2 °C / min; |dT / dt|_max is the current maximum temperature change rate; β is the adjustment coefficient, set to 0.5. Specific implementation of the temperature gradient sensitive model: Construct an adaptive weight coefficient and adopt a more refined calculation model in the high temperature gradient region. The formula for the weight coefficient is: W_grad(i, j, k) = 1 + γ·(| ▽ T|_i,j,k / | ▽ T|_avg) Δ ; where W_grad is the gradient weight coefficient; | ▽ T|_i,j,k is the temperature gradient at point (i, j, k); | ▽ T|_avg is the average temperature gradient; γ is the amplification coefficient, set to 2.0; Δ is the exponential coefficient, set to 1.5. Precise modeling of the deflation-temperature coupling model: Establish the coupling relationship between the throttling deflation process and the temperature field, and adopt the principles of mass conservation and energy conservation. The specific formula is: dT / dt = -v· ▽ T + α· ▽ 2 T + S_T; where v is the velocity field, calculated from the deflation flow rate and pressure gradient; S_T is the temperature source term, related to the temperature drop caused by the throttle valve. Throttle valve temperature drop calculation formula: ΔT_throttle = k·(P1 ((κ-1) / κ) - P2 ((κ-1) / κ) )·(P1 / P2); where ΔT_throttle is the throttle temperature drop; P1 and P2 are the pressures before and after throttling respectively; k is the coefficient, approximately 358 K; κ is the air adiabatic index, approximately 1.4. The comprehensive application of these technologies has greatly improved the accuracy of the cold energy distribution prediction map, and the average prediction error has decreased from 18.5% of the traditional method to 7.6%.

[0117] The specific implementation of the method for enhancing the similarity of operating conditions based on cold energy characteristics is as follows: Traditional operating condition identification methods mainly rely on general characteristics such as pressure and flow rate, and pay insufficient attention to cold energy characteristics. In this embodiment, a similarity calculation method specifically for cold energy characteristics is constructed: Specific method for cold energy feature extraction: Extract the following cold energy features from the temperature field data: Temperature gradient vector field: ▽ T = (ΨT / Ψx, ΨT / Ψy, ΨT / Ψz); Position of the cold and hot interface: Defined by the temperature isosurface T = T_amb; Clarity of the cold and hot interface: CI = | ▽ T|_interface / | ▽ T|_avg; where | ▽T|_interface is the magnitude of the temperature gradient at the cold and hot interface,| ▽ T|_avg is the magnitude of the average temperature gradient in the entire temperature field; Cold energy spatio-temporal distribution pattern: Principal component analysis is used to extract eigenvectors. Precise calculation of the thermodynamic similarity index: Considering both the static and dynamic characteristics of the temperature field, the similarity is calculated as: S_thermo = w1·S_gradient + w2·S_interface + w3·S_rate + w4·S_pattern; where S_gradient is the temperature gradient similarity; S_interface is the interface characteristic similarity; S_rate is the cold energy generation rate similarity; S_pattern is the distribution pattern similarity; w1, w2, w3, and w4 are weight coefficients, which are dynamically adjusted according to the system stage. Specific algorithm for the stage-based weighting strategy: Define a time-dependent weight function to automatically adjust the weights according to the system stage: W(t) = W_init·f_init(t) + W_stable·f_stable(t) + W_end·f_end(t); where W_init, W_stable, and W_end are the weight vectors for the initial, stable, and final stages respectively; f_init, f_stable, and f_end are time-dependent stage functions that satisfy f_init(t) + f_stable(t) + f_end(t) = 1. This embodiment makes the working condition identification more accurate, and the recovery strategy for similar working conditions is also more targeted, which is a key factor in improving the cold energy recovery efficiency.

[0118] The specific implementation of the cold energy recovery control strategy adaptive generation method is as follows: This embodiment breaks through the limitations of traditional independent control strategies and constructs a collaborative control mechanism for throttling and cold energy recovery: Cold energy recovery key factor extraction technology: Using data mining methods, extract the key factors affecting cold energy recovery efficiency from historical data: The matching degree of working fluid flow rate and throttling temperature drop: MD = Qc / (ΔT·Q_throttle), the optimal value is 0.12 - 0.15, where Q_throttle is the heat flow rate during the throttling process; The temperature gradient at the inlet of the cold energy recovery device: |dTci / dt|, the optimal range is <0.8℃ / min; The working fluid flow rate adjustment rate: |dQc / dt|, the optimal range is <0.5kg / s·min. Throttling-recovery dual-objective optimization framework: Construct an optimization model that comprehensively considers the stability of the turbine and the cold energy recovery efficiency: f_obj = -[α·η_recovery + (1-α)·(1-σT / σT_max)]; where η_recovery is the cold energy recovery efficiency; σT is the standard deviation of the turbine inlet temperature fluctuation; σT_max is the maximum allowable standard deviation; α is the trade-off coefficient, default value is 0.6. Cold energy quality evaluation mechanism: Introduce temperature level to evaluate the cold energy quality, and the calculation formula is: Q_cold = m_cold·cp·(T_ref - T_cold)·(1 + β·(T_ref - T_cold) / T_ref); where Q_cold is the cold energy value considering quality; m_cold is the mass flow rate of the cold working fluid; cp is the specific heat capacity; T_ref is the reference temperature; T_cold is the cold energy temperature; β is the quality correction coefficient, and the value is 0.15. This embodiment enables the system to more effectively utilize the cold energy resources in the gas storage reservoir and realizes the collaborative control with the turbine power generation process, effectively improving the overall system efficiency.

[0119] This embodiment details the specific implementation process of the dynamic utilization method of the cold energy in the gas storage reservoir of the million-cubic-level compressed air energy storage system. It solves technical problems such as accurate modeling of the temperature field in large gas storage reservoirs, identification of cold energy characteristic working conditions, and collaborative control of throttling-cold energy recovery, improves the cold energy recovery efficiency, and at the same time ensures the stable operation of the turbine system. Experimental verification shows that compared with traditional technologies, the cold energy recovery efficiency of this embodiment has increased by 19 percentage points, the stability of the turbine inlet parameters has increased by 57%, and the overall system efficiency has increased by 7.5%, providing important technical support for the large-scale application of compressed air energy storage systems.

[0120] The present invention divides the gas storage space of millions of cubic meters into three scale levels: microscale, mesoscale, and macroscale, and creates adaptive grids for each level using the octree segmentation method. At the same time, different physical models are applied at different scales (the heat diffusion equation at the microscale, the convective heat transfer model at the mesoscale, and the stratified flow model at the macroscale), solving the "false distribution" problem caused by sparse sampling points in traditional space reconstruction methods in large gas storage reservoirs. Through physically guided interpolation reconstruction and an adaptive boundary constraint mechanism, the reconstruction accuracy of the temperature field is improved from 85% of traditional methods to over 95%, especially in areas with sparse measurement points, where the reconstruction accuracy is increased by 40%, providing precise spatial positioning support for cold energy recovery, ensuring that the cold energy recovery device can be effectively arranged and controlled according to the actual cold energy distribution position, and avoiding energy waste and inefficient operation of equipment. By developing a non-uniform time step prediction mechanism, a temperature gradient-sensitive spatial evolution model, and a coupled analysis of gas discharge flow and temperature field, accurate prediction of the dynamic evolution of the temperature field inside the gas storage reservoir is achieved. According to the statistical characteristics of the temperature change rate, the rapid change stage, transition stage, and slow change stage are automatically identified, and appropriate time steps are allocated, while particularly strengthening the prediction accuracy in low-temperature regions and regions with large temperature gradients. By treating the throttling process and the temperature field evolution inside the reservoir as a coupled system, the change in cold energy distribution under different gas discharge strategies is accurately captured. The prediction error is reduced from 15 - 20% of traditional methods to 5 - 8%, especially the position prediction accuracy of cold energy concentration regions is increased by over 40%, providing a reliable basis for the feedforward control of cold energy recovery devices and achieving the ability of advance planning and dynamic adjustment. By developing a thermodynamic similarity index system (similarity of temperature gradient distribution, similarity of cold and hot interface movement patterns, similarity of cold energy generation rate) and a stage similarity weighting strategy, the calculation of operating condition similarity is directly associated with the cold energy recovery efficiency target. A similarity calculation formula is specifically designed for the characteristics of cold energy, and the weights of each feature are dynamically adjusted according to different stages of the cold energy recovery process (initial rapid cooling stage, stable recovery stage, and late inefficient recovery stage). By introducing a cold energy recovery efficiency correlation mapping, the similarity calculation directly points to the optimization target of cold energy recovery efficiency. The accuracy of cold energy recovery operating condition identification is increased from 70% of traditional methods to over 90%, providing a high-quality reference of similar operating conditions for subsequent control strategy generation, improving the cold energy recovery efficiency and system operation stability, and avoiding the inefficient control problems caused by only focusing on general operating characteristics in traditional methods. By developing cold energy recovery key factor extraction technology, a throttling-recovery dual-objective optimization framework, and a cold energy quality evaluation mechanism, the adaptive generation of cold energy recovery control strategies is achieved. Control decision elements highly correlated with cold energy recovery efficiency are extracted from the control experience of historical similar operating conditions, and the control of the throttle valve and the cold energy recovery device is treated as a coupled optimization problem to maximize the cold energy recovery amount under the premise of meeting the requirements of turbine inlet parameters.By constructing a cold energy quality evaluation mechanism, a comprehensive evaluation is carried out according to the temperature level (quality) and quantity of cold energy to guide the control strategy to prioritize the recovery of high-quality cold energy. Through coupling optimization, the synergistic effect of throttle control and cold energy recovery is achieved. Compared with the traditional separate control, the cold energy recovery efficiency is increased by 20 - 30%, while ensuring the stable operation of the turbine system, solving the contradiction between cold energy recovery and the stable operation of the turbine, and improving the overall energy utilization efficiency of the system. By using the Granger causality analysis method, the causal relationship between throttle valve parameters and cold energy recovery performance is identified, and a coupling influence model is established. First, a time-aligned coupling analysis data set is extracted from multi-source data, then the causal direction between variables is determined through time lag analysis, a directed weight network is constructed, and finally the identified key causal paths are quantified into a mathematical model. This model breaks through the limitations of traditional independent control and provides a theoretical basis for the coordinated control of the throttle valve and the cold energy recovery device. Model verification shows that the prediction accuracy of this coupling model for cold energy recovery efficiency reaches 92%, which is 15% higher than that of the traditional independent model. By understanding how throttle parameter changes affect cold energy generation and distribution, the system can predictively adjust the control strategy, avoiding the cold energy waste and system fluctuations caused by control lag in traditional methods. By designing a three-layer control framework including a strategy layer (hourly level), a coordination layer (minute level), and an execution layer (second level), the coordinated control of the throttle valve and the cold energy recovery device is achieved. This framework reasonably distributes control tasks according to different time scales of the system's dynamic characteristics, and the output of the upper-layer controller serves as the input or constraint of the lower-layer controller. The model predictive control method is used to achieve coordinated control, and by dynamically adjusting the weight coefficients, the throttle control target and the cold energy recovery target are balanced at different operating stages. This hierarchical control framework improves the system response characteristics, reduces the turbine inlet temperature fluctuation from ±4.2°C in the traditional method to ±1.8°C, and the stability is increased by 57%. At the same time, it fully utilizes the predictability and controllability of the system, solves the time-delay problem and stability problem in the cold energy recovery process, and improves the system's resistance and adaptability to disturbances.

[0121] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A method for dynamically utilizing the cold energy of a gas storage reservoir in a million-cubic-meter compressed air energy storage system, characterized in that Including: Obtain multi-source monitoring data of the compressed air energy storage system to form a system basic data set; Based on this, perform multi-scale thermal flow constraint space reconstruction, and generate a dynamic cold energy distribution prediction map through a space-time dynamic model; Based on the system basic data set, construct a hierarchical control framework through a coupling influence model of throttling and cold energy recovery, and generate a hierarchical control strategy; Based on the system basic data set and the dynamic cold energy distribution prediction map, construct a working condition characteristic fingerprint library, and generate an optimized control strategy for cold energy recovery through similar working condition identification; Fuse it with the hierarchical control strategy and output a comprehensive control instruction set.

2. The method according to claim 1, characterized in that, Generating a dynamic cold energy distribution prediction map includes: Based on the system basic data set, preliminarily reconstruct the full-space temperature field and pressure field in the gas storage reservoir; based on this, perform multi-scale thermal flow constraint space reconstruction and optimize the full-space temperature field and pressure field; Based on the optimized full-space temperature field and pressure field, perform time-series-space coupling evolution prediction to generate a dynamic cold energy distribution prediction map.

3. The method according to claim 2, wherein Optimizing the full-space temperature field and pressure field includes: Perform hierarchical space segmentation on the preliminary reconstruction results of the full-space temperature field and pressure field to obtain a multi-scale spatial domain; based on this, construct physical constraint models for different scales to form a multi-scale physical constraint set; Utilize the gas storage reservoir wall temperature sensor data in the system basic data set to construct a dynamic boundary condition set through adaptive boundary constraints; combine it with the measured sensor data and the multi-scale physical constraint set in the system basic data set, and perform physics-guided interpolation reconstruction to optimize the full-space temperature field and pressure field.

4. The method according to claim 2, wherein Performing time-series-space coupling evolution prediction to generate a dynamic cold energy distribution prediction map includes: Based on the optimized full-space temperature field and pressure field, apply adaptive time window analysis to identify the time dynamic characteristics of different stages to generate a time characteristic segmentation set; for each stage among them, construct a space evolution model sensitive to temperature gradient to form a segmented space evolution model set; Read the gas release parameters of the throttle valve operation state sequence in the system basic data set, analyze the influence of the throttle valve opening change on the temperature field evolution inside the gas storage reservoir, and establish a gas release-temperature coupling model; combine it with the segmented space evolution model set, and generate a dynamic cold energy distribution prediction map through multi-step progressive prediction.

5. The method according to claim 4, wherein Generating a time characteristic segmentation set includes: Based on the optimized full-space temperature field and pressure field, calculate the statistical characteristics of the full-space temperature change rate, including mean, variance, and kurtosis; Classify the statistical characteristics through time series clustering to automatically identify the rapid change, transition, and slow change stages; adjust the time step according to the characteristics of different stages: adopt the first small time step in the rapid change stage to improve the prediction accuracy, and adopt the second time step in the slow change stage to improve the calculation efficiency; finally generate a time characteristic segmentation set; where the first time step is less than the second time step.

6. The method according to claim 1, wherein Generating an optimized control strategy for cold energy recovery includes: Based on the system basic data set, construct a working condition characteristic fingerprint library of the compressed air energy storage system; based on this, generate a working condition similarity vector from the current system state through working condition identification; Enhance the operating condition similarity based on cold energy characteristics for the operating condition similarity vector in combination with the dynamic cold energy distribution prediction map, and obtain the optimized operating condition similarity vector; accordingly, adaptively generate the cold energy recovery control strategy and output the optimized cold energy recovery control strategy.

7. The method according to claim 6, wherein Obtaining the optimized operating condition similarity vector includes: Based on the system basic dataset and the dynamic cold energy distribution prediction map, extract the cold energy characteristic indexes, construct a similarity calculation formula for cold energy recovery, and generate a set of thermodynamic similarity indexes; Construct a stage similarity weighting strategy according to the characteristics of the cold energy recovery process and generate a set of stage weighting functions; Read the operating data of the cold energy recovery device in the system basic dataset, analyze the relationship between operating condition similarity and cold energy recovery efficiency, and establish an efficiency correlation mapping; Combine the set of thermodynamic similarity indexes, the set of stage weighting functions and the efficiency correlation mapping, calculate the enhanced similarity between the current and historical operating condition similarity vectors, and output the optimized operating condition similarity vector.

8. The method according to claim 7, wherein The set of thermodynamic similarity indexes includes: The temperature gradient distribution similarity index, obtained by calculating the cosine similarity of the temperature gradient fields of two operating conditions; The cold and hot interface movement pattern similarity index, obtained by comparing the moving speed and direction of the cold and hot dividing interfaces; The cold energy generation rate similarity index, obtained by comparing the change patterns of the cold energy generation amount per unit time.

9. The method according to claim 6, wherein Adaptively generate the cold energy recovery control strategy and output the optimized cold energy recovery control strategy, including: Based on the optimized operating condition similarity vector, retrieve the control strategies of similar operating conditions from the historical control database, extract the key control decisions, and generate a set of candidate control strategies; identify the control decision elements related to the cold energy recovery efficiency from them and construct a set of cold energy efficiency impact factors; Based on the coupling impact model and the set of cold energy efficiency impact factors, construct a throttling-recovery dual-objective optimization framework, take the throttle valve control and the cold energy recovery device control as a coupled optimization problem, and construct a multi-objective optimization model; combine it with the current system state and perform real-time optimization calculations to generate the optimized cold energy recovery control strategy under the current operating condition.

10. Cold energy dynamic utilization system for gas storage in a million-cubic-meter compressed air energy storage system, characterized in that, Including: A compression system for compressing air; A gas storage tank connected to the compression system for storing the compressed gas; An exhaust pipeline connecting the gas storage tank for discharging the gas in the gas storage tank; A throttle valve arranged in the exhaust pipeline for controlling the gas flow and pressure; A cold energy recovery device arranged at the rear of the exhaust port of the throttle valve in the exhaust pipeline for recovering the cold energy in the gas; A heat exchanger arranged at the rear of the cold energy recovery device for heat exchange between the cold energy and the gas; A turbine system arranged after the heat exchanger for performing turbine work on the gas after heat exchange to generate electricity; A control module, including: At least one processor; and, A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method according to any one of claims 1 to 9.

Images