Oxygen recovery and energy storage system of air separation device under peak valley electricity condition

Abstract

The invention discloses an oxygen recovery and energy storage system of an air separation device under a peak valley electricity condition, and relates to the technical field of gas separation, the oxygen recovery and energy storage system comprises a nitrogen compression system, a liquid oxygen vaporization system, an oxygen liquefaction system and a solid bed energy storage system, the nitrogen compression system comprises a nitrogen compressor; the liquid oxygen vaporization system comprises a vaporization heat exchanger, a liquid oxygen pump, a liquid nitrogen throttle valve and a liquid nitrogen vacuum tank; the oxygen liquefaction system comprises a liquefaction heat exchanger, a liquid nitrogen pump, a nitrogen expansion machine, a liquid oxygen throttle valve and a liquid oxygen vacuum tank; and the solid-state bed energy storage system comprises a solid-state bed regenerator and a nitrogen circulating fan. Through intelligent load transfer and buffer adjustment, frequent impact and large adjustment on the main process of the air separation device are reduced, an air separation system and matched compression and liquefaction units run more continuously and stably, the running stability of the whole system is improved, and the service life of the whole system is prolonged.

Description

Technical Field

[0001] This invention relates to the field of gas separation technology, and in particular to an oxygen recovery and energy storage system for an air separation unit under peak and off-peak electricity conditions. Background Technology

[0002] In the field of industrial gas production, air separation units, as core equipment for obtaining key industrial gases such as oxygen and nitrogen, are widely used in industries such as steel, chemicals, medical, and electronics. These units typically require continuous and stable operation to meet the continuous and reliable gas supply needs of downstream users. However, in actual operation, especially in a power market environment implementing peak-valley electricity pricing policies, the operation and economics of air separation units face significant challenges.

[0003] On the one hand, during peak electricity periods when electricity prices are high, downstream gas-consuming enterprises often reduce their production load or even shut down, resulting in the oxygen produced by the air separation unit not being fully utilized. In order to avoid system overpressure, they are often forced to release the gas, which not only wastes gas resources but also reduces overall energy efficiency.

[0004] On the other hand, during off-peak electricity hours when electricity prices are low, downstream gas demand increases significantly. Even if the air separation unit operates at full capacity, it may still be difficult to meet peak gas demand, affecting the continuity and stability of production.

[0005] In addition, traditional air separation units have slow start-up and shutdown, limited load adjustment range, and difficulty in flexibly adapting to fluctuations in downstream gas consumption. The system operation mode is relatively rigid and lacks effective means of utilizing the time-varying differences in electricity.

[0006] Therefore, how to achieve efficient recovery, storage and resupply of oxygen under peak and off-peak electricity conditions, and improve the synergy between air separation units and downstream gas consumption links, has become a key issue that urgently needs to be addressed in the field of gas separation technology. Summary of the Invention

[0007] To address the aforementioned technical problems, the present invention provides an oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions, comprising a nitrogen compression system, a liquid oxygen vaporization system, an oxygen liquefaction system, and a solid bed energy storage system, wherein the nitrogen compression system includes a nitrogen compressor. The liquid oxygen vaporization system includes a vaporization heat exchanger, a liquid oxygen pump, a liquid nitrogen throttle valve, and a liquid nitrogen vacuum tank. The oxygen liquefaction system includes a liquefaction heat exchanger, a liquid nitrogen pump, a nitrogen expander, a liquid oxygen throttle valve, and a liquid oxygen vacuum tank. The solid-state bed energy storage system includes a solid-state bed cooler and a nitrogen circulation fan, characterized in that it further includes a control subsystem, which includes: The data acquisition unit is used to acquire and integrate peak and valley electricity price data from the power grid, raw material air status data from upstream of the air separation unit, operating parameters of each node in the system, and historical operating database in real time, and then transmit the integrated data status to the data analysis unit. The data analysis unit, based on fused data, uses an integrated learning prediction model to make rolling predictions of electricity price trends, system load demand, and operating efficiency for specific future periods. Coupled with the system thermodynamic model, it generates a dynamic optimization control strategy with the goal of minimizing overall operating costs or maximizing energy efficiency, while meeting the constraints of oxygen product indicators. The execution control unit coordinates the operating parameters of the nitrogen compressor, liquid oxygen pump, liquid nitrogen pump, nitrogen expander, liquid nitrogen throttle valve, liquid oxygen throttle valve, and nitrogen circulating fan according to the dynamic optimization control strategy.

[0008] Preferably, the nitrogen compressor is connected to a vaporization heat exchanger, the vaporization heat exchanger is connected to a liquid nitrogen throttle valve, one path of the liquid nitrogen throttle valve is connected to the vaporization heat exchanger, and the other path is connected to a liquid nitrogen vacuum tank, which is connected to a liquid nitrogen pump.

[0009] Preferably, the liquid nitrogen pump is connected to a liquefaction heat exchanger, the liquefaction heat exchanger is connected to a nitrogen expander, the nitrogen expander is connected to the liquefaction heat exchanger, the liquefaction heat exchanger is connected to a liquid oxygen throttle valve, the liquid oxygen throttle valve is connected to a liquid oxygen vacuum tank, the liquid oxygen vacuum tank is connected to a liquid oxygen pump, and the liquid oxygen pump is connected to a vaporization heat exchanger.

[0010] Preferably, one of the nitrogen circulating fans is connected to a solid bed accumulator, which is connected to a vaporization heat exchanger; the other is connected to a liquefaction heat exchanger, which is connected to a solid bed accumulator.

[0011] Preferably, the data acquisition unit includes: The data acquisition module is used to acquire the temperature, pressure, and humidity of the raw material air in real time through the sensor network, as well as the pressure, temperature, and flow parameters of key nodes of the nitrogen compressor, heat exchangers, vacuum tank, pump, valve, and fan. The data governance module is used to perform data cleaning, outlier removal, and timestamp synchronization on real-time collected data and historical operating databases. The data fusion module is used to fuse multi-source heterogeneous data after treatment based on the extended Kalman filter algorithm and the sliding window averaging algorithm to generate a real-time system status dataset with time consistency and high confidence. The dataset contains the real-time efficiency calculation values ​​of each key device.

[0012] Preferably, the ensemble learning prediction model in the data analysis unit is the Stacking ensemble framework; The base learners of the Stacking ensemble framework include a gradient boosting decision tree model and a gated recurrent unit network model, and the meta-learners adopt a ridge regression model. The integrated learning prediction model uses historical electricity price data, meteorological data, factory production plans and system historical operation data as training sets to make rolling predictions of time-of-use electricity prices, system oxygen demand load and overall energy efficiency of the equipment in the next 24 to 72 hours. The rolling prediction period can be configured to 15 minutes or 1 hour, and the unit can evaluate and output the confidence interval of the prediction results online.

[0013] Preferably, in the data analysis unit, the process of generating the dynamic optimization control strategy includes: Using the rolling prediction results and the real-time state dataset as input, a multi-objective optimization function is constructed; The multi-objective optimization function takes minimizing the overall system operating cost as the primary objective and maximizing the system efficiency as the secondary objective, and introduces the life loss cost, which is positively correlated with the equipment operating intensity, as a penalty term. The comprehensive operating cost includes real-time electricity costs calculated based on time-of-use pricing, as well as equivalent fatigue loss costs incurred due to equipment start-up, shutdown, and load regulation.

[0014] Preferred options include: During peak electricity price periods, the load of the nitrogen compressor is dynamically reduced to below the economic operating point or partially shut down. At the same time, the energy storage capacity of the solid bed accumulator is increased, and the liquid oxygen stored in the liquid oxygen vacuum tank is used preferentially. After being pressurized by the liquid oxygen pump and reheated and vaporized by the vaporization heat exchanger, the liquid oxygen is supplied to the oxygen product to transfer electricity consumption. During periods of low electricity prices, the load of the nitrogen compressor and the nitrogen expander is increased to the high-efficiency operating range, the liquefaction rate of the oxygen liquefaction system is increased, the excess oxygen of the air separation unit is converted into liquid oxygen and stored in the liquid oxygen vacuum tank, and the solid bed accumulator is controlled to release cold, thereby reducing the liquefaction energy consumption of the auxiliary system. Based on the predicted load and electricity price trends, a distributed model predictive control algorithm is used to perform feedforward compensation on the opening of the liquid nitrogen throttling valve and the liquid oxygen throttling valve, and to perform feedback PID control on the speed of the nitrogen circulation fan.

[0015] The present invention has at least the following beneficial effects: During peak electricity price periods, the system can proactively reduce the load on high-energy-consuming nitrogen compressors or shut them down, instead utilizing stored liquid oxygen for gas supply; during off-peak electricity price periods, it operates at full capacity to produce and store liquid oxygen. This directly leverages the electricity price difference, significantly reducing electricity costs. Simultaneously, the optimization strategy incorporates fatigue wear caused by frequent equipment start-ups and drastic load adjustments into cost calculations, protecting equipment while pursuing the lowest possible electricity costs.

[0016] By coupling an integrated learning model with a thermodynamic model for rolling prediction and optimization, coordinated dynamic regulation of the entire system (compression, liquefaction, cold storage, and vaporization) was achieved. The addition of a solid-bed energy storage system enables the storage and reuse of cold energy, releasing cold during liquefaction to reduce energy consumption and storing cold during vaporization to improve efficiency, thereby enhancing overall efficiency.

[0017] During peak electricity hours or when the air separation unit produces excess oxygen, the system can efficiently liquefy and store oxygen that might otherwise be released, effectively avoiding the waste of valuable industrial gases and achieving resource recovery and recycling. Simultaneously, this technology strongly guarantees the continuity and reliability of gas supply for downstream production. During off-peak electricity hours or peak downstream gas demand, the system can rapidly release the stored liquid oxygen, which is then vaporized and supplied stably, ensuring gas supply capacity and providing a solid guarantee for continuous production. In terms of economic benefits and operational optimization, the system cleverly utilizes the peak-valley electricity price difference for intelligent scheduling, storing energy during off-peak hours and releasing it during peak hours, achieving optimal energy costs and significantly reducing overall operating expenses. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of an oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions, provided in Embodiment 1 of the present invention. Figure 2 The flowchart is for the control subsystem provided in Embodiment 2 of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0022] Example 1

[0023] like Figure 1 As shown, this embodiment provides an oxygen recovery and energy storage system for an air separation unit under peak and off-peak electricity conditions, including a nitrogen compression system, a liquid oxygen vaporization system, an oxygen liquefaction system, and a solid bed energy storage system. The nitrogen compression system includes a nitrogen compressor. The liquid oxygen vaporization system includes a vaporization heat exchanger, a liquid oxygen pump, a liquid nitrogen throttle valve, and a liquid nitrogen vacuum tank; The oxygen liquefaction system includes a liquefaction heat exchanger, a liquid nitrogen pump, a nitrogen expander, a liquid oxygen throttle valve, and a liquid oxygen vacuum tank; Solid-state bed energy storage systems include solid-state bed accumulators and nitrogen circulation fans.

[0024] Furthermore, the nitrogen compressor is connected to the vaporization heat exchanger, the vaporization heat exchanger is connected to the liquid nitrogen throttle valve, one path of the liquid nitrogen throttle valve is connected to the vaporization heat exchanger; the other path is connected to the liquid nitrogen vacuum tank, and the liquid nitrogen vacuum tank is connected to the liquid nitrogen pump.

[0025] Secondly, the liquid nitrogen pump is connected to the liquefaction heat exchanger, the liquefaction heat exchanger is connected to the nitrogen expander, the nitrogen expander is connected to the liquefaction heat exchanger, the liquefaction heat exchanger is connected to the liquid oxygen throttle valve, the liquid oxygen throttle valve is connected to the liquid oxygen vacuum tank, the liquid oxygen vacuum tank is connected to the liquid oxygen pump, and the liquid oxygen pump is connected to the vaporization heat exchanger.

[0026] Furthermore, one of the nitrogen circulation fans is connected to the solid bed accumulator, which is connected to the vaporization heat exchanger; the other is connected to the liquefaction heat exchanger, which is connected to the solid bed accumulator.

[0027] Specifically, the various subsystems are organically connected through pipelines and equipment, forming a whole that can flexibly switch operating modes: The outlet of the nitrogen compressor is connected to the hot-side inlet of the vaporization heat exchanger. After heat exchange, the nitrogen gas is cooled and partially liquefied in the vaporization heat exchanger, then enters the liquid nitrogen throttling valve. This valve divides the fluid into two paths: one path returns to the cold side of the vaporization heat exchanger as backflow gas to recover cooling; the other path enters the liquid nitrogen vacuum tank for storage.

[0028] Liquid nitrogen in the liquid nitrogen vacuum tank can be pressurized by a liquid nitrogen pump and sent to a liquefaction heat exchanger as a cooling medium for liquefying oxygen.

[0029] In the liquefaction heat exchanger, low-pressure oxygen from the pipeline network is cooled and liquefied. After being depressurized by a liquid oxygen throttling valve, it enters the liquid oxygen vacuum tank for storage. The liquid oxygen in the liquid oxygen vacuum tank can be pressurized by a liquid oxygen pump and then sent back to the vaporization heat exchanger to absorb heat and vaporize, thus completing the supply conversion from liquid to gaseous oxygen.

[0030] The solid-bed regenerator is connected to two heat exchangers via a nitrogen circulation fan: one is connected to the vaporization heat exchanger, which provides or stores cooling energy during liquid oxygen vaporization; the other is connected to the liquefaction heat exchanger, which assists in cooling or storing excess cooling energy during oxygen liquefaction. This connection method makes the solid bed a "reservoir" for system cooling energy allocation, enhancing the system's ability to cope with load fluctuations.

[0031] This system can intelligently switch operating strategies according to the peak and valley periods of the power grid: During peak electricity periods when electricity prices are high, the system mainly activates the oxygen liquefaction system and the solid bed energy storage system to liquefy and store oxygen that the air separation unit cannot process immediately, avoiding venting and waste, while using low-priced electricity to complete the energy storage process.

[0032] During off-peak electricity hours when electricity prices are low or during peak downstream gas consumption, the system activates the nitrogen compression system and the liquid oxygen vaporization system to vaporize the stored liquid oxygen and replenish the supply to meet production needs. It also utilizes the solid bed to release cooling energy to reduce vaporization energy consumption.

[0033] Peak power operation steps 1 and 2; off-peak power operation steps 3 and 4.

[0034] Step 1: Atmospheric nitrogen from the pipeline is pressurized to a certain pressure by a nitrogen compressor and then enters a vaporization heat exchanger to exchange heat with liquid oxygen from a liquid oxygen vacuum tank that is pressurized by a liquid oxygen pump. The liquid oxygen absorbs heat and vaporizes, and is then sent to the low-pressure oxygen pipeline. The pressurized nitrogen is cooled and liquefied, and then throttled by a liquid nitrogen throttle valve. The gas phase flows back to the vaporization heat exchanger for reheating and is then sent to the low-pressure nitrogen pipeline. The liquid phase enters the liquid nitrogen vacuum tank (NT) for storage.

[0035] Step 2: Atmospheric nitrogen from the pipeline is pressurized by a nitrogen circulation fan and enters a solid bed accumulator to be cooled to a low temperature. Then it enters a vaporization heat exchanger to release its cooling capacity and is reheated before being sent to the atmospheric nitrogen pipeline.

[0036] Step 3: Low-pressure oxygen from the pipeline enters the liquefaction heat exchanger and exchanges heat with liquid nitrogen from the liquid nitrogen vacuum tank, which is pressurized by the liquid nitrogen pump. The low-pressure oxygen is cooled and liquefied, and after passing through the liquid oxygen throttling valve, it is sent to the liquid oxygen vacuum tank storage tank. A portion of the liquid oxygen is sent out as liquid oxygen product. After the pressurized liquid nitrogen absorbs heat and vaporizes, it is extracted from the middle of the liquefaction heat exchanger and sent to the nitrogen expander ET for expansion and refrigeration. The expanded low-temperature nitrogen is reheated by the liquefaction heat exchanger and then sent to the atmospheric pressure nitrogen pipeline network.

[0037] Step 4: Atmospheric nitrogen from the pipeline is pressurized by a nitrogen circulation fan, enters a liquefaction heat exchanger to be cooled to a low temperature, then enters a solid bed accumulator to store the cold energy, and after reheating, is sent to the atmospheric nitrogen pipeline.

[0038] Example 2

[0039] Combination Figure 2 As shown, based on the above-described Embodiment 1, this embodiment also proposes a control subsystem, which includes: The data acquisition unit is used to acquire and integrate peak and valley electricity price data from the power grid, raw material air status data from upstream of the air separation unit, operating parameters of each node in the system, and historical operating database in real time, and then transmit the integrated data status to the data analysis unit. The data analysis unit, based on fused data, uses an integrated learning prediction model to make rolling predictions of electricity price trends, system load demand, and operating efficiency for specific future periods. Coupled with the system thermodynamic model, it generates a dynamic optimization control strategy with the goal of minimizing overall operating costs or maximizing energy efficiency, while meeting the constraints of oxygen product indicators. The execution control unit coordinates the operating parameters of the nitrogen compressor, liquid oxygen pump, liquid nitrogen pump, nitrogen expander, liquid nitrogen throttle valve, liquid oxygen throttle valve, and nitrogen circulating fan according to a dynamic optimization control strategy.

[0040] Specifically, firstly, the system's data acquisition unit constitutes the system's "sensory nerves." This unit captures real-time and predictive peak-valley electricity price signals from the power grid, as well as the temperature, pressure, and humidity of the feed air from the upstream air separation unit. It also intensively monitors the operating parameters (pressure, temperature, and flow rate) of key nodes within the system itself (such as compressor inlets and outlets, heat exchanger ends, vacuum tanks, pumps, and valves). These multi-source, heterogeneous real-time data, along with the historical operating database, undergo data cleaning, anomaly removal, and time synchronization processing. Then, advanced algorithms such as extended Kalman filtering are used to fuse these data, generating a unified, accurate, and highly confident panoramic real-time status map of the system, laying a solid data foundation for intelligent decision-making.

[0041] Secondly, the system's "intelligent brain"—the data analysis unit—begins its work. It receives the fused panoramic data and drives an ensemble learning model composed of gradient boosting decision trees and gated recurrent unit networks to continuously predict electricity price trends, downstream oxygen demand load, and the system's own energy efficiency changes over the next 24 to 72 hours. Subsequently, these predictions, along with the real-time system status, are input into a built-in system thermodynamic model. This model accurately characterizes the energy and material coupling relationships between multiple processes, including nitrogen compression, oxygen liquefaction, liquid oxygen vaporization, and solid-bed storage / release cooling. Under constraints such as oxygen product purity and pressure, the unit performs online dynamic optimization calculations with the goal of "lowest overall operating cost" or "highest system energy efficiency," generating an optimal control strategy covering the next few hours. This strategy not only considers electricity costs but also quantifies the lifespan reduction caused by frequent equipment adjustments, thus achieving a balance between economic efficiency and equipment reliability.

[0042] Finally, the execution control unit, acting as the system's "motion center," is responsible for translating optimization strategies into precise equipment actions. Based on strategy instructions, it coordinates and adjusts the operating parameters of multiple core devices: for example, during peak electricity price periods, it reduces the load on the nitrogen compressor or even partially shuts it down, while simultaneously starting the liquid oxygen pump to vaporize the liquid oxygen stored in the liquid oxygen vacuum tank for gas supply, thus shifting the high-energy-consuming process out of peak price periods; during off-peak electricity price periods, it increases the load on the nitrogen compressor and nitrogen expander to increase liquefaction capacity for storing liquid oxygen, and controls the solid-bed accumulator to release cold to reduce liquefaction energy consumption. For key adjustment points such as liquid nitrogen and liquid oxygen throttling valves, model predictive control is used for feedforward compensation, while feedback PID control is applied to the nitrogen circulation fan to ensure smooth and dynamic matching between subsystems, achieving intelligent and stable switching from "energy storage mode" to "energy release mode."

[0043] Furthermore, the data acquisition unit in the above embodiments includes: The data acquisition module is used to acquire the temperature, pressure, and humidity of the raw material air in real time through the sensor network, as well as the pressure, temperature, and flow parameters of key nodes such as the nitrogen compressor, heat exchangers, vacuum tank, pump, valve, and fan. The data governance module is used to perform data cleaning, outlier removal, and timestamp synchronization on real-time collected data and historical operating databases. The data fusion module is used to fuse multi-source heterogeneous data after treatment based on the extended Kalman filter algorithm and the sliding window averaging algorithm to generate a real-time system status dataset with time consistency and high confidence. The dataset contains the real-time efficiency calculation values ​​of each key device.

[0044] Specifically, the system deploys a high-density sensor network at key physical nodes. Temperature, humidity, and pressure transmitters are installed in the raw air pipelines to capture raw inputs from upstream conditions. High-precision temperature sensors, pressure sensors, and mass flow meters are installed at the inlet and outlet of core equipment such as nitrogen compressors, vaporization heat exchangers, and liquefaction heat exchangers, as well as in key tube and shell sides, and at liquid oxygen / liquid nitrogen vacuum tanks. These sensors operate continuously at millisecond to second-level frequencies, capturing the system's dynamic pressure, temperature, and flow parameters in real time, forming the raw data stream of the system's operation.

[0045] Subsequently, these massive, high-speed raw data streams, which may contain noise or outliers, are fed into the data governance module for standardization. This module, running on a real-time database or industrial edge computing platform, first cleans the data, automatically identifying and removing obviously erroneous outliers (such as readings exceeding physical ranges) based on process knowledge bases or statistical rules. Next, it uses a high-precision network clock protocol to add a unified timestamp to all data sources, resolving time synchronization issues caused by sensor sampling frequencies or communication delays. Finally, it normalizes various parameters, converting them into dimensionless standard values, laying a consistent data format foundation for subsequent fusion and analysis.

[0046] This module employs a hybrid strategy combining the Extended Kalman Filter (EKF) algorithm and the Sliding Window Average (SWA) algorithm. The EKF is primarily used to handle system state variables with nonlinear characteristics (such as equipment efficiency). It utilizes a simplified mechanistic model of the system to optimally estimate observations from different sensors, effectively filtering out random noise and dynamically estimating key parameters that cannot be directly measured, such as the real-time isentropic efficiency of a nitrogen compressor and the real-time heat transfer coefficient of a heat exchanger. Simultaneously, for parameters with relatively stable states (such as stable pressure values), the SWAF algorithm is used to smooth short-term fluctuations and improve data stability.

[0047] Secondly, the ensemble learning prediction model in the aforementioned data analysis unit is the Stacking ensemble framework; The Stacking ensemble framework's base learners include gradient boosting decision tree models and gated recurrent unit network models, while the meta-learner uses a ridge regression model. The integrated learning prediction model uses historical electricity price data, meteorological data, factory production plans and historical system operation data as training sets to make rolling predictions of time-of-use electricity prices, system oxygen demand load and overall energy efficiency of the unit in the next 24 to 72 hours. The rolling forecast period can be configured to 15 minutes or 1 hour, and the unit can evaluate and output the confidence interval of the forecast results online.

[0048] Specifically, the model's implementation relies on a two-layer learning structure. In the first layer, two complementary algorithms are used as base learners: the gradient boosting decision tree model excels at capturing complex nonlinear relationships and feature importance from structured historical data (such as electricity price sequences and production plans); while the gated recurrent unit network model is specifically designed for processing time-series data, effectively learning the long-term dependencies and periodic change patterns of data such as electricity prices and loads. These two models are trained in parallel on the same historical dataset, each generating preliminary predictions. In the second layer, these preliminary results serve as new feature inputs, which are then further learned and integrated by a meta-learner (here, a ridge regression model). The introduction of ridge regression effectively prevents overfitting, ensuring the stability and generalization ability of the final prediction results.

[0049] The model's training data comes from a wide range of sources and is multi-dimensional. It includes not only long-term historical time-of-use electricity price data, but also meteorological data (such as ambient temperature, which affects plant efficiency), factory production plans (directly related to oxygen demand), and the system's own historical operating parameter database. This fusion training of multi-source data enables the model to understand the complex relationships between the external market, the natural environment, production scheduling, and internal operating status.

[0050] During deployment and operation, the model performs rolling forecasting tasks. It periodically (e.g., every 15 minutes or 1 hour) uses the latest real-time and historical data as input to dynamically predict key indicators within a 24-72 hour time window. This not only provides the expected curve for time-of-use electricity prices but also simultaneously predicts downstream oxygen demand load and the potential overall energy efficiency of the plant under these operating conditions. More importantly, this implementation has online evaluation capabilities. By analyzing the uncertainty of the model's predictions, it can calculate and output confidence intervals for the predicted values ​​(e.g., there is a 95% probability that the electricity price for a future hour will fall between X yuan and Y yuan). This confidence interval provides downstream optimization decision-making units with a risk assessment basis, enabling them to make a more informed trade-off between pursuing economic efficiency and ensuring operational reliability when formulating control strategies.

[0051] Furthermore, in the aforementioned data analysis unit, the process of generating the dynamic optimization control strategy includes: A multi-objective optimization function is constructed using rolling prediction results and real-time state datasets as inputs. The multi-objective optimization function takes minimizing the overall system operating cost as the primary objective and maximizing the system efficiency as the secondary objective, and introduces the life depreciation cost, which is positively correlated with the equipment operating intensity, as a penalty term; The total operating cost includes real-time electricity costs calculated based on time-of-use pricing, as well as the equivalent fatigue loss costs incurred due to equipment start-up, shutdown, and load regulation.

[0052] Specifically, the data analysis unit receives key information from two sources: first, rolling forecasts of electricity price trends, oxygen demand load, and system energy efficiency for specific future periods (e.g., 24-72 hours) generated by an ensemble learning prediction model; and second, a panoramic real-time status dataset reflecting the system's current instantaneous operating condition provided by the data acquisition unit. The combination of these two provides both a forward-looking perspective on the market and demand, and a precise understanding of the system's current state, laying a complete spatiotemporal information foundation for optimized calculations.

[0053] Based on this, the system constructs a structured multi-objective optimization function, which is the core of the mathematical model for achieving intelligent decision-making. This function explicitly defines two optimization directions: Main objective: Minimize the overall system operating cost. This cost is precisely defined as comprising two main components: the electricity cost fluctuating in real time based on time-of-use pricing, which directly reflects the economic efficiency of responding to peak-valley pricing and achieving "peak shaving and valley filling"; and the equivalent fatigue loss cost incurred due to equipment start-up, shutdown, and load increases / decreases. The latter is incorporated into the optimization function by converting equipment mechanical wear and lifespan reduction into economic costs, thereby automatically avoiding frequent and drastic operations that could damage the equipment while pursuing low electricity costs.

[0054] Secondary objective: Maximize system efficiency. Efficiency is a scientific indicator that measures the level of energy utilization in terms of "quality." Taking this as a secondary objective, the driving system must not only save "quantity" (cost) but also improve "quality" (energy efficiency), such as optimizing the matching of hot and cold fluids, reducing heat exchange temperature differences, and increasing the working efficiency of the expander.

[0055] To achieve the best balance among multiple potentially conflicting objectives (such as minimum cost which may require frequent equipment adjustments, while maximum energy efficiency may require stable operation), the optimization function also introduces a lifetime depreciation cost that is positively correlated with the intensity of equipment operation as a penalty term. This causes the optimization algorithm to automatically "weigh the pros and cons" during the optimization process, tending to select solutions that achieve a good trade-off between cost, energy efficiency, and equipment lifespan.

[0056] Finally, the system employs the advanced intelligent optimization algorithm, Non-Dominated Sorting Genetic Algorithm-II (NSGA-II), with an elitist strategy, to solve the aforementioned complex multi-objective problem. This algorithm encodes and iteratively evolves the nitrogen compressor speed, the frequency of the liquid oxygen / liquid nitrogen pump, the opening degree of each throttle valve, and the speed of the nitrogen circulation fan as optimization variables. Under the condition of satisfying preset process constraints such as oxygen product purity and pressure stability, the algorithm can search and generate a Pareto optimal solution set. Each solution in this set represents a non-dominated feasible operating scheme that, under given conditions, cannot simultaneously improve all objectives. The system then automatically selects the final optimal control command sequence from this solution set based on the real-time strategy (e.g., whether cost savings or energy efficiency improvement is currently prioritized) and issues it to the execution control unit.

[0057] It should be noted that the aforementioned execution control unit performs coordinated adjustments based on a dynamic optimization control strategy, including: During peak electricity price periods, the load of the nitrogen compressor is dynamically reduced to below the economic operating point or partially shut down. At the same time, the energy storage capacity of the solid bed accumulator is increased, and the liquid oxygen stored in the liquid oxygen vacuum tank is used first. After being pressurized by the liquid oxygen pump and reheated and vaporized by the vaporization heat exchanger, the oxygen product is supplied to transfer electricity consumption. During periods of low electricity prices, the load of nitrogen compressors and nitrogen expanders is increased to the high-efficiency operating range, the liquefaction rate of the oxygen liquefaction system is increased, the excess oxygen in the air separation unit is converted into liquid oxygen and stored in the liquid oxygen vacuum tank, and the solid bed accumulator is controlled to release cold, thereby reducing the liquefaction energy consumption of the auxiliary system. Based on the predicted load and electricity price trends, a distributed model predictive control algorithm is used to perform feedforward compensation on the opening of the liquid nitrogen throttling valve and the liquid oxygen throttling valve, and to perform feedback PID control on the speed of the nitrogen circulation fan.

[0058] Specifically, firstly, the execution control unit continuously receives dynamically optimized control strategies generated by the data analysis unit. This strategy clearly indicates the operating mode (peak, trough, or transition) that the system should be in during a specific future period (such as the next few hours), and provides the target values ​​for each core device (such as load rate, valve opening, speed, etc.).

[0059] During peak electricity price operation, the unit immediately issues commands to dynamically reduce the speed of the energy-intensive nitrogen compressor or shut down some operating units, lowering their load below the preset economic operating point. Simultaneously, it starts and controls the liquid oxygen pump, pressurizing the liquid oxygen stored in the liquid oxygen vacuum tank at the required flow rate, and adjusts relevant parameters of the vaporization heat exchanger to ensure efficient and stable reheating and vaporization of the liquid oxygen, replacing the oxygen produced by the nitrogen compressor-driven process. During this process, the unit also increases the speed of the nitrogen circulation fan in the solid-bed accumulator loop, enhancing the cold storage process and storing cold energy for later use, thereby shifting electricity consumption from high-price periods to internal system energy storage (liquid oxygen and cold energy).

[0060] During off-peak electricity pricing, the unit performs the opposite adjustment. It increases the load on the nitrogen compressor to its efficient operating range and increases the inlet pressure or flow rate of the nitrogen expander to enhance cooling capacity. Simultaneously, it controls the liquid nitrogen pump and liquefaction heat exchanger to maximize the liquefaction rate of the oxygen liquefaction system, converting excess oxygen from the air separation unit into liquid oxygen for storage. At this time, the unit controls the solid-bed energy storage system to switch to cold release mode, adjusting the nitrogen circulation fan and pipeline valves to allow the stored cold energy to assist the liquefaction process, thereby reducing the power consumption per unit of liquefied product.

[0061] In the power supply and transition modes, to achieve smooth switching and dynamic energy balance between subsystems, the unit employs a more refined algorithm for control. Based on model predictive control, it calculates and provides feedforward compensation commands for the opening of the liquid nitrogen and liquid oxygen throttling valves in advance, according to the predicted load and electricity price change curves, to actively counteract the effects of system inertia. Simultaneously, closed-loop PID feedback control is used for the speed of the nitrogen circulation fan, making real-time fine adjustments based on the temperature or pressure difference of the solid-state bed regenerator. Through this combination of feedforward and feedback, and the coordination of centralized optimization and decentralized execution, the system can ensure a smooth transition of pressure, temperature, and flow parameters during various operating condition transitions, such as from energy storage to energy release and from liquefaction to vaporization, maintaining continuous, efficient, and stable operation of the entire system.

[0062] Example 3

[0063] This invention provides a non-transitory computer-readable storage medium storing at least one instruction or at least one program segment, which is loaded and executed by a processor to implement the following steps: Atmospheric nitrogen from the pipeline is pressurized to a certain pressure by a nitrogen compressor and then enters a vaporization heat exchanger to exchange heat with liquid oxygen from a liquid oxygen vacuum tank that is pressurized by a liquid oxygen pump. The liquid oxygen absorbs heat and vaporizes before being sent to the low-pressure oxygen pipeline. The pressurized nitrogen is cooled and liquefied, and then throttled by a liquid nitrogen throttle valve. The gas phase flows back to the vaporization heat exchanger for reheating before being sent to the low-pressure nitrogen pipeline, while the liquid phase enters the liquid nitrogen vacuum tank (NT) for storage.

[0064] Atmospheric nitrogen from the pipeline is pressurized by a nitrogen circulation fan and enters a solid bed accumulator where it is cooled to a low temperature. It then enters a vaporization heat exchanger to release its cooling capacity and is reheated before being sent back to the atmospheric nitrogen pipeline.

[0065] Low-pressure oxygen from the pipeline enters the liquefaction heat exchanger and exchanges heat with liquid nitrogen from the liquid nitrogen vacuum tank, which is pressurized by the liquid nitrogen pump. The low-pressure oxygen is cooled and liquefied, and then sent to the liquid oxygen vacuum tank storage tank after passing through the liquid oxygen throttling valve. A portion of the liquid oxygen is sent out as liquid oxygen product. After the pressurized liquid nitrogen absorbs heat and vaporizes, it is extracted from the middle of the liquefaction heat exchanger and sent to the nitrogen expander ET for expansion and refrigeration. The expanded low-temperature nitrogen is reheated by the liquefaction heat exchanger and then sent to the atmospheric pressure nitrogen pipeline network.

[0066] Atmospheric nitrogen from the pipeline is pressurized by a nitrogen circulation fan, enters a liquefaction heat exchanger to be cooled to a low temperature, then enters a solid bed accumulator to store the cold energy, and after reheating, is sent to the atmospheric nitrogen pipeline.

[0067] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0068] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0069] Example 4

[0070] This invention provides an electronic device, including a processor and a memory, wherein the memory stores at least one instruction or at least one program segment, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the following steps: Atmospheric nitrogen from the pipeline is pressurized to a certain pressure by a nitrogen compressor and then enters a vaporization heat exchanger to exchange heat with liquid oxygen from a liquid oxygen vacuum tank that is pressurized by a liquid oxygen pump. The liquid oxygen absorbs heat and vaporizes before being sent to the low-pressure oxygen pipeline. The pressurized nitrogen is cooled and liquefied, and then throttled by a liquid nitrogen throttle valve. The gas phase flows back to the vaporization heat exchanger for reheating before being sent to the low-pressure nitrogen pipeline, while the liquid phase enters the liquid nitrogen vacuum tank (NT) for storage.

[0071] Atmospheric nitrogen from the pipeline is pressurized by a nitrogen circulation fan and enters a solid bed accumulator where it is cooled to a low temperature. It then enters a vaporization heat exchanger to release its cooling capacity and is reheated before being sent back to the atmospheric nitrogen pipeline.

[0072] Low-pressure oxygen from the pipeline enters the liquefaction heat exchanger and exchanges heat with liquid nitrogen from the liquid nitrogen vacuum tank, which is pressurized by the liquid nitrogen pump. The low-pressure oxygen is cooled and liquefied, and then sent to the liquid oxygen vacuum tank storage tank after passing through the liquid oxygen throttling valve. A portion of the liquid oxygen is sent out as liquid oxygen product. After the pressurized liquid nitrogen absorbs heat and vaporizes, it is extracted from the middle of the liquefaction heat exchanger and sent to the nitrogen expander ET for expansion and refrigeration. The expanded low-temperature nitrogen is reheated by the liquefaction heat exchanger and then sent to the atmospheric pressure nitrogen pipeline network.

[0073] Atmospheric nitrogen from the pipeline is pressurized by a nitrogen circulation fan, enters a liquefaction heat exchanger to be cooled to a low temperature, then enters a solid bed accumulator to store the cold energy, and after reheating, is sent to the atmospheric nitrogen pipeline.

[0074] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An oxygen recovery and energy storage system for an air separation unit under peak and off-peak electricity conditions, comprising a nitrogen compression system, a liquid oxygen vaporization system, an oxygen liquefaction system, and a solid bed energy storage system, wherein the nitrogen compression system comprises a nitrogen compressor; The liquid oxygen vaporization system includes a vaporization heat exchanger, a liquid oxygen pump, a liquid nitrogen throttle valve, and a liquid nitrogen vacuum tank. The oxygen liquefaction system includes a liquefaction heat exchanger, a liquid nitrogen pump, a nitrogen expander, a liquid oxygen throttle valve, and a liquid oxygen vacuum tank. The solid-bed energy storage system includes a solid-bed accumulator and a nitrogen circulation fan, characterized in that... It also includes a control subsystem, which comprises: The data acquisition unit is used to acquire and integrate peak and valley electricity price data from the power grid, raw material air status data from upstream of the air separation unit, operating parameters of each node in the system, and historical operating database in real time, and then transmit the integrated data status to the data analysis unit. The data analysis unit, based on fused data, uses an integrated learning prediction model to make rolling predictions of electricity price trends, system load demand, and operating efficiency for specific future periods. Coupled with the system thermodynamic model, it generates a dynamic optimization control strategy with the goal of minimizing overall operating costs or maximizing energy efficiency, while meeting the constraints of oxygen product indicators. The execution control unit coordinates the operating parameters of the nitrogen compressor, liquid oxygen pump, liquid nitrogen pump, nitrogen expander, liquid nitrogen throttle valve, liquid oxygen throttle valve, and nitrogen circulating fan according to the dynamic optimization control strategy.

2. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 1, characterized in that, The nitrogen compressor is connected to the vaporization heat exchanger, the vaporization heat exchanger is connected to the liquid nitrogen throttle valve, one path of the liquid nitrogen throttle valve is connected to the vaporization heat exchanger; the other path is connected to the liquid nitrogen vacuum tank, and the liquid nitrogen vacuum tank is connected to the liquid nitrogen pump.

3. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 2, characterized in that, The liquid nitrogen pump is connected to the liquefaction heat exchanger, the liquefaction heat exchanger is connected to the nitrogen expander, the nitrogen expander is connected to the liquefaction heat exchanger, the liquefaction heat exchanger is connected to the liquid oxygen throttle valve, the liquid oxygen throttle valve is connected to the liquid oxygen vacuum tank, the liquid oxygen vacuum tank is connected to the liquid oxygen pump, and the liquid oxygen pump is connected to the vaporization heat exchanger.

4. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 1, characterized in that, One of the nitrogen circulating fans is connected to the solid bed accumulator, which is connected to the vaporization heat exchanger; the other is connected to the liquefaction heat exchanger, which is connected to the solid bed accumulator.

5. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 1, characterized in that, The data acquisition unit includes: The data acquisition module is used to acquire the temperature, pressure, and humidity of the raw material air in real time through the sensor network, as well as the pressure, temperature, and flow parameters of key nodes of the nitrogen compressor, heat exchangers, vacuum tank, pump, valve, and fan. The data governance module is used to perform data cleaning, outlier removal, and timestamp synchronization on real-time collected data and historical operating databases. The data fusion module is used to fuse multi-source heterogeneous data after treatment based on the extended Kalman filter algorithm and the sliding window averaging algorithm to generate a real-time system status dataset with time consistency and high confidence. The dataset contains the real-time efficiency calculation values ​​of each key device.

6. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 1, characterized in that, The ensemble learning prediction model in the data analysis unit is the Stacking ensemble framework. The base learners of the Stacking ensemble framework include a gradient boosting decision tree model and a gated recurrent unit network model, and the meta-learners adopt a ridge regression model. The integrated learning prediction model uses historical electricity price data, meteorological data, factory production plans and system historical operation data as training sets to make rolling predictions of time-of-use electricity prices, system oxygen demand load and overall energy efficiency of the equipment in the next 24 to 72 hours. The rolling prediction period can be configured to 15 minutes or 1 hour, and the unit can evaluate and output the confidence interval of the prediction results online.

7. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 1, characterized in that, In the data analysis unit, the generation process of the dynamic optimization control strategy includes: Using the rolling prediction results and the real-time state dataset as input, a multi-objective optimization function is constructed; The multi-objective optimization function takes minimizing the overall system operating cost as the primary objective and maximizing the system efficiency as the secondary objective, and introduces the life loss cost, which is positively correlated with the equipment operating intensity, as a penalty term. The comprehensive operating cost includes real-time electricity costs calculated based on time-of-use pricing, as well as equivalent fatigue loss costs incurred due to equipment start-up, shutdown, and load regulation.

8. The oxygen recovery and energy storage system for an air separation unit under peak-valley electricity conditions according to claim 1, characterized in that, The execution control unit performs coordinated adjustments according to the dynamic optimization control strategy, including: During peak electricity price periods, the load of the nitrogen compressor is dynamically reduced to below the economic operating point or partially shut down. At the same time, the energy storage capacity of the solid bed accumulator is increased, and the liquid oxygen stored in the liquid oxygen vacuum tank is used preferentially. After being pressurized by the liquid oxygen pump and reheated and vaporized by the vaporization heat exchanger, the liquid oxygen is supplied to the oxygen product to transfer electricity consumption. During periods of low electricity prices, the load of the nitrogen compressor and the nitrogen expander is increased to the high-efficiency operating range, the liquefaction rate of the oxygen liquefaction system is increased, the excess oxygen of the air separation unit is converted into liquid oxygen and stored in the liquid oxygen vacuum tank, and the solid bed accumulator is controlled to release cold, thereby reducing the liquefaction energy consumption of the auxiliary system. Based on the predicted load and electricity price trends, a distributed model predictive control algorithm is used to perform feedforward compensation on the opening of the liquid nitrogen throttling valve and the liquid oxygen throttling valve, and to perform feedback PID control on the speed of the nitrogen circulation fan.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the oxygen recovery and energy storage system of the air separation unit under peak-valley electricity conditions as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the oxygen recovery and energy storage system of the air separation unit under peak-valley electricity conditions as described in any one of claims 1 to 8.

Images