An energy consumption optimization and regulation method for air separation systems

Abstract

This application relates to the field of gas mixture component separation technology, and proposes an energy consumption optimization and adjustment method for air separation systems. The method includes: collecting the temperature of the extraction layer, the temperature of the upper tray, and the temperature of the lower tray; calculating the temperature difference between adjacent trays, the standard deviation of temperature fluctuation, and the reference noise constant at discrete sampling times; calculating the logarithmic ratio of the upper and lower temperature differences and the confidence weight; calculating the component migration risk value and the stability weighting factor; constructing a comprehensive target value; and combining it with a preset control cycle to achieve adaptive control of the upper column reflux ratio of the air separation device, thereby optimizing energy consumption. This application can reduce the energy consumption of air separation systems.

Description

Technical Field

[0001] This application relates to the field of gas mixture component separation technology, specifically to an energy consumption optimization and adjustment method for air separation systems. Background Technology

[0002] Air separation units extract high-purity industrial gases such as oxygen, nitrogen, and argon through cryogenic distillation. Their energy consumption level directly depends on the control strategy of the reflux ratio in the distillation column. In actual industrial production, air separation units often face variable load conditions such as power grid fluctuations and adjustments to feed gas flow, which can lead to unsteady fluctuations in the operating pressure within the column. According to the principle of gas-liquid balance, changes in operating pressure will cause a common-mode drift in the temperature of all trays in the entire column; for example, an increase in pressure will lead to an increase in overall temperature.

[0003] For a two-tower process equipped with an argon extraction system, the component distribution in the upper argon fraction extraction area is crucial to product quality. To prevent nitrogen from migrating down and penetrating the argon fraction layer, causing "nitrogen blockage," the reflux ratio of the distillation column is typically controlled by monitoring the temperature or temperature difference in this area. However, single-point temperature monitoring cannot eliminate common-mode drift caused by pressure, and temperature changes due to pressure fluctuations can easily be misinterpreted as component frontal movement. Even with simple adjacent temperature difference control, transient changes in gas-liquid equilibrium during drastic pressure fluctuations can still mask the true component separation signal. When it is impossible to effectively distinguish between the "false signal" caused by pressure fluctuations and the "real risk" caused by component migration, a conservative strategy is forced, setting a high reflux ratio safety margin to cope with uncertainty. This forces the feed air compressor to operate at high load for extended periods, resulting in energy waste. Summary of the Invention

[0004] This application provides an energy consumption optimization and regulation method for air separation systems. By controlling the reflux ratio of a distillation column through temperature difference, it addresses the problem that temperature changes caused by pressure fluctuations are easily misinterpreted as component frontal movement, leading to excessive energy consumption in the air separation system. The specific technical solution adopted is as follows: One embodiment of this application provides an energy consumption optimization and regulation method for an air separation system, the method comprising the following steps: A three-point observation structure was adopted to collect the temperature of the extraction layer, the temperature of the upper tray, and the temperature of the lower tray at each discrete sampling time, and to calculate the temperature difference between adjacent trays, the standard deviation of temperature fluctuation, and the reference noise constant at the discrete sampling time. Based on the temperature difference between adjacent trays, the logarithmic ratio of the temperature difference between the upper and lower trays is calculated. The logarithmic ratio of the temperature difference between the upper and lower trays is used to characterize the probability that a nitrogen front approaches the extraction port from above. Based on the temperature difference between adjacent trays at two discrete sampling times with a preset time interval, a confidence weight is constructed. The confidence weight is used to characterize the credibility of temperature changes caused by component migration. For the same discrete sampling time, the component migration risk value is calculated based on the value of the logarithmic ratio of the upper and lower temperature differences and the confidence weight. The stability weighting factor is calculated based on the standard deviation of temperature fluctuation and the reference noise constant. The stability weighting factor is used to correct the degree of influence of component migration risk. Based on the stability weighting factor and component migration risk value at discrete sampling times, a comprehensive target value is constructed for discrete sampling times. The comprehensive target value is a quantitative assessment result of the operating economy and safety of the working condition. Combined with a preset control cycle, adaptive control of the upper tower reflux ratio of the air separator is realized, thereby achieving optimized adjustment of energy consumption.

[0005] Furthermore, the specific calculation method for the temperature difference between adjacent trays is as follows: The temperature difference between adjacent trays includes the temperature difference between upper adjacent trays and the temperature difference between lower adjacent trays; For the same discrete sampling time, the temperature difference between adjacent upper trays is the absolute value of the difference between the temperature of the upper tray and the temperature of the extraction layer, and the temperature difference between adjacent lower trays is the absolute value of the difference between the temperature of the extraction layer and the temperature of the lower tray.

[0006] Furthermore, the specific calculation method for the standard deviation of temperature fluctuation is as follows: The variance of the temperature of all sampling layers in the data buffer is denoted as the standard deviation of temperature fluctuation at the last discrete sampling time in the data buffer.

[0007] Furthermore, the specific calculation method for the reference noise constant is as follows: When the data in the data buffer completely fills the entire data buffer, the arithmetic mean of the standard deviation of temperature fluctuations at all discrete sampling times in the data buffer is recorded as the reference noise constant; conversely, the arithmetic mean of the standard deviation of temperature fluctuations at the first to the second-to-last discrete sampling times in the data buffer is recorded as the reference noise constant at the last discrete sampling time in the data buffer.

[0008] Furthermore, the specific calculation method for the logarithmic ratio of the upper and lower temperature differences is as follows: For the same discrete sampling time, the ratio of the temperature difference between the upper adjacent trays to the temperature difference between the lower adjacent trays is taken as the independent variable of the logarithmic function with the natural constant as the base, and the calculated value of the logarithmic function is denoted as the logarithmic ratio of the upper and lower temperature differences.

[0009] Furthermore, the method for constructing the confidence weights is as follows: The rate of temperature change is calculated based on the temperature difference between the upper adjacent tray and the lower adjacent tray at two discrete sampling times with a preset time interval. The rate of temperature change includes the rate of temperature change of the upper tray and the rate of temperature change of the lower tray. For the same discrete sampling time, when the signs of the upper and lower temperature difference change rates are the same, the absolute value of the difference between the upper and lower temperature difference change rates is used as the numerator, and the sum of the absolute values ​​of the upper and lower temperature difference change rates is used as the denominator. The calculated value of the fraction is recorded as the confidence weight; otherwise, the confidence weight is assigned the value 1.

[0010] Furthermore, the method for calculating the component migration risk value is as follows: When the logarithmic ratio of the upper and lower temperature differences at discrete sampling time is greater than or equal to 0, the product of the logarithmic ratio of the upper and lower temperature differences and the confidence weight is recorded as the component migration risk value; otherwise, the component migration risk value is assigned a preset high-risk penalty value.

[0011] Furthermore, the method for determining the stability weighting factor is as follows: The square of the ratio of the standard deviation of temperature fluctuation to the reference noise constant is denoted as the first square value. The sum of the product of the first square value and the preset stiffness-sensitive gain with the number 1 is denoted as the stability weighting factor.

[0012] Furthermore, the comprehensive objective value is the result of a positive correlation between the stability weighting factor and the component migration risk value.

[0013] Furthermore, the method for adaptively controlling the reflux ratio of the upper tower of the air separator by combining a preset control cycle and optimizing energy consumption includes: The discrete sampling time at which the control decision on the reflux ratio of the upper tower is made after a preset control cycle is recorded as the adjustment time. The difference between the adjustment time and the comprehensive target value of the previous adjacent adjustment time is recorded as the comprehensive target difference of the adjustment time. When the overall target difference is less than 0, the difference between the upper tower reflux ratio at the adjustment time and the preset adjustment step size is used as the value of the upper tower reflux ratio at the next adjacent discrete sampling time after the adjustment time; otherwise, the sum of the upper tower reflux ratio at the adjustment time and the preset adjustment step size is used as the value of the upper tower reflux ratio at the next adjacent discrete sampling time after the adjustment time.

[0014] The beneficial effects of this application are: To capture the sensitive component distribution changes in the argon fraction extraction area, this application constructs a three-point observation structure on the upper part of the distillation column, calculates the temperature difference between adjacent trays, weakens the common-mode effect caused by absolute pressure changes, and quantifies the micro-pulsation intensity of the tray fluid to obtain the standard deviation of temperature fluctuations. To separate the characteristic differences between operating pressure fluctuations and component migration from aliased temperature signals, based on the sound velocity propagation characteristics of pressure waves in the gas phase and the rate-limiting characteristics of mass transfer between the gas and liquid phases, the confidence level of temperature changes caused by component migration is quantified, confidence weights are obtained, and the probability of a nitrogen front approaching the extraction port from above is analyzed to obtain the logarithmic ratio of the temperature difference between the upper and lower sections. To obtain high-confidence risk assessment indicators, independent logical judgments need to be set for severe fault conditions such as "component penetration" or "temperature inversion" while eliminating pressure interference. The method prevents logical deadlock in the control system under extreme operating conditions by calculating the component migration risk value and obtaining a stability weighting factor, which is used to correct the degree of influence of component migration risk. Finally, a comprehensive evaluation system of linear energy consumption cost and nonlinear risk penalty is constructed, and a comprehensive target value is constructed at discrete sampling time. A discrete gradient search strategy that conforms to the characteristics of industrial processes is adopted, combined with a preset control cycle, to achieve adaptive control of the reflux ratio of the upper column of the air separation unit, thereby achieving optimized adjustment of energy consumption. This solves the problem that temperature changes caused by pressure fluctuations are easily misjudged as component front movement during the process of controlling the reflux ratio of the distillation column by temperature difference, resulting in excessive energy consumption of the air separation system. The method drives the air separation system to automatically approach the minimum energy consumption point at the safety boundary, thereby reducing the energy consumption of the air separation system. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of an energy consumption optimization and adjustment method for an air separation system provided in one embodiment of this application. Detailed Implementation

[0016] Please see Figure 1 The diagram illustrates a flowchart of an energy consumption optimization and adjustment method for an air separation system according to an embodiment of this application. The method includes the following steps: Step S001: Using a three-point observation structure, collect the temperature of the extraction layer, the temperature of the upper tray, and the temperature of the lower tray at each discrete sampling time, and calculate the temperature difference between adjacent trays, the standard deviation of temperature fluctuation, and the reference noise constant at each discrete sampling time.

[0017] To optimize the energy consumption of the air separation system by controlling the reflux ratio of the distillation column through temperature difference, it is necessary to establish a standardized observation data stream, eliminate common-mode interference caused by static pressure, and complete parameter self-calibration during the system cold start phase.

[0018] To capture the subtle changes in component distribution in the argon fraction extraction area, a three-point observation structure, including a baseline point and a reference point, was constructed on the upper part of the distillation column.

[0019] The three-point observation structure includes a benchmark measuring point, an upper reference measuring point, and a lower reference measuring point, arranged as follows: 1. Reference Measurement Point: A reference measurement point is set at the theoretical plate location where the argon fraction extraction port of the upper column is located. A temperature sensor is installed at the reference measurement point to collect the real-time temperature and obtain the extraction port layer temperature. .

[0020] 2. Upper reference measuring point: The first point directly above the benchmark measuring point. A reference measuring point is set at the top of the theoretical tray, and a temperature sensor is installed at the reference measuring point to collect the real-time temperature of the upper tray. ,in, This represents the first preset parameter, and in this embodiment, the value of the first preset parameter is 5.

[0021] 3. Below reference measuring point: The first point directly below the benchmark measuring point. A reference measuring point is set at the bottom of the theoretical tray, and a temperature sensor is installed at the reference measuring point to collect the real-time temperature of the lower tray. .

[0022] This embodiment will specify the sampling period. The sampling interval is set to 1 second. At each discrete sampling moment, the temperatures of the extraction layer, the upper tray, and the lower tray are collected and interacted with the field instruments in real time through the DCS distributed control system. A data buffer is set up to store the collected extraction layer temperature, upper tray temperature, and lower tray temperature in real time according to the order of collection. In this embodiment, the time window length of the data buffer is set to 60 seconds.

[0023] According to the gas-liquid equilibrium equation, fluctuations in operating pressure within a distillation column cause a common-mode drift in the temperature of all trays across the column, with the temperature moving in the same direction and at the same amplitude. This background noise often masks subtle component separation characteristics. To eliminate the influence of common-mode drift, spatial difference calculations are performed using the temperature difference between adjacent trays to obtain the temperature difference between adjacent trays, which is mainly determined by the component concentration gradient, making it insensitive to overall changes in absolute pressure. The temperature difference between adjacent trays includes the temperature difference between the upper adjacent trays and the temperature difference between the lower adjacent trays.

[0024] Specifically, for the same discrete sampling time, the temperature difference between adjacent upper trays is the absolute value of the difference between the temperature of the upper tray and the temperature of the extraction layer, and the temperature difference between adjacent lower trays is the absolute value of the difference between the temperature of the extraction layer and the temperature of the lower tray.

[0025] It is understood that the temperature difference between the upper adjacent trays and the temperature difference between the lower adjacent trays respectively characterize the degree of gas-liquid separation in the area above the extraction port and the degree of gas-liquid separation in the area below the extraction port.

[0026] Considering that temperature data collected in industrial settings often contains high-frequency measurement noise, directly performing differential operations on the temperature difference results (such as rate calculations required by subsequent modules) would severely amplify the noise, causing control signal oscillations. Therefore, noise reduction of the temperature difference between adjacent trays is necessary before feature extraction. Differential operations, such as those used in subsequent rate calculations, are part of this process.

[0027] In this embodiment, a moving average filter is selected to denoise the temperature difference between adjacent trays, and the filter window length is 5 seconds. The moving average filter denoising is a well-known technique and will not be described in detail here.

[0028] The variance of the temperature of all sampling layers in the data buffer is denoted as the standard deviation of temperature fluctuation at the last discrete sampling time in the data buffer.

[0029] The temperature fluctuation standard deviation is used to quantify the intensity of micro-pulsations in the fluid on the tray.

[0030] It should be noted that when the data in the data buffer is insufficient to fill the entire data buffer, the system is in the data accumulation phase after startup or reset. At this time, the statistical results are not reliable. In this case, the security logic of forced safe output and baseline parameter self-calibration is executed.

[0031] Specifically, the execution of forced safety output means: instead of outputting real-time calculation results, the component migration risk value and stability weighting factor calculated subsequently are forcibly locked to a high-risk state, forcing the control system to maintain a high reflux ratio and ensuring the safe operation during the data accumulation period.

[0032] Specifically, the self-calibration of the reference parameter is as follows: when the data in the data buffer is not completely filled, the reference noise constant is assigned a preset factory experience fluctuation reference value; when the data in the data buffer completely fills the entire data buffer, the arithmetic mean of the standard deviation of temperature fluctuation at all discrete sampling times in the data buffer is recorded as the reference noise constant.

[0033] It is important to understand that the reference noise constant is the "normal fluctuation reference" of the air separation system under the current load.

[0034] When the data in the data buffer completely fills the entire data buffer, the arithmetic mean of the standard deviation of temperature fluctuations at all discrete sampling times in the data buffer is recorded as the reference noise constant. At the same time, the temperature difference between adjacent trays at all discrete sampling times in the data buffer can be obtained.

[0035] Thus, the reference noise constant and the temperature difference between adjacent trays at discrete sampling times are obtained.

[0036] Step S002: Calculate the logarithmic ratio of the upper and lower temperature differences based on the temperature difference between adjacent trays. The logarithmic ratio of the upper and lower temperature differences is used to characterize the possibility that the nitrogen front approaches the extraction port from above. Construct a confidence weight based on the temperature difference between adjacent trays at two discrete sampling times with a preset time interval. The confidence weight is used to characterize the credibility of temperature changes caused by component migration.

[0037] To separate the characteristic differences between operating pressure fluctuations and component migration from aliased temperature signals, engineering statistical features are used for analysis. Specifically, confidence weights are constructed based on the sound velocity propagation characteristics of pressure waves in the gas phase and the rate-limiting characteristics of mass transfer between gas and liquid phases.

[0038] To quantify the geometric position of the nitrogen front relative to the argon fraction extraction port, and considering the typical S-shaped (Sigmoid) nonlinear characteristics of the nitrogen component concentration distribution along the column in the extraction port region, a natural logarithmic transformation is introduced to map the nonlinear concentration distribution into a position indication signal with higher linearity.

[0039] Specifically, for the same discrete sampling time, the ratio of the temperature difference between the upper adjacent trays to the temperature difference between the lower adjacent trays is taken as the independent variable of the logarithmic function with the natural constant as the base, and the calculated value of the logarithmic function is recorded as the logarithmic ratio of the upper and lower temperature differences.

[0040] In the process of calculating the ratio, in order to avoid the denominator becoming invalid or the logarithm becoming undefined when the denominator approaches zero, a preset value needs to be added to both the denominator and the numerator. In this example, the preset value is 0.01K.

[0041] When the logarithmic ratio of the temperature difference between the upper and lower sections is greater than 0, it indicates that at the same discrete sampling time, the temperature difference at the upper section is greater than that at the lower section, suggesting that a nitrogen front may be approaching the extraction port from above. However, the logarithmic ratio of the temperature difference between the upper and lower sections still includes common-mode noise caused by pressure fluctuations. For example, an increase in pressure can cause the temperature difference between the upper adjacent tray and the lower adjacent tray to change simultaneously, so further cleaning is needed in conjunction with subsequent analysis.

[0042] To capture the dynamic characteristics of temperature, the rate of change of temperature difference over time is calculated. Considering the inertia of industrial processes, too short a differential step size will introduce noise, while too long a step size will lead to lag in control. In this embodiment, the time step size for the differential calculation is set to [value missing]. One sampling period (set in this embodiment) (corresponding to 5 seconds), to balance response speed and noise immunity. This represents the second preset parameter. In this embodiment, the value of the second preset parameter is 5, which means the time step is 5 seconds.

[0043] The rate of temperature change is calculated based on the temperature difference between adjacent trays at two discrete sampling times with a preset time interval. This rate of temperature change includes the rate of temperature change at the upper tray and the rate of temperature change at the lower tray. The preset time interval is the time step, which is 5 seconds in this embodiment.

[0044] The numerator is the difference between the temperature difference of the upper adjacent tray at the discrete sampling time and the temperature difference of the upper adjacent tray at the discrete sampling time with a preset time before the discrete sampling time. The denominator is the product of the preset time and the sampling period. The value of the fraction is recorded as the rate of change of the upper temperature difference at the discrete sampling time.

[0045] The numerator is the difference between the temperature difference of the lower adjacent tray at the discrete sampling time and the temperature difference of the lower adjacent tray at the discrete sampling time with a preset time before the discrete sampling time. The denominator is the product of the preset time and the sampling period. The value of the fraction is recorded as the rate of change of the lower temperature difference at the discrete sampling time.

[0046] It is important to understand that the rate of change of temperature difference is used to characterize the "speed" and "direction" of the temperature field change at discrete sampling moments.

[0047] Based on the principles of fluid dynamics, when the operating pressure of the distillation column fluctuates, the change in gas-liquid equilibrium temperature between adjacent trays exhibits a high degree of temporal synchronicity. That is, the rate of change of temperature difference at the top and the rate of change of temperature difference at the bottom show the same direction and similar amplitude, indicating pressure interference characteristics.

[0048] Furthermore, component migration is limited by mass transfer resistance, and the changes in the upper tray must precede those in the lower tray. This manifests as a dramatic change in the rate of temperature difference change in the upper tray, while the rate of temperature difference change in the lower tray remains relatively static or lags behind, exhibiting significant asymmetry.

[0049] For the same discrete sampling time, when the upper and lower temperature difference change rates have the same sign, the changes in the upper and lower temperature difference change rates may originate from pressure fluctuations or component migration. Asymmetric calculations are performed as follows: For the same discrete sampling time, the absolute value of the difference between the upper and lower temperature difference change rates is used as the numerator, and the sum of the absolute values ​​of the upper and lower temperature difference change rates is used as the denominator. The calculated value of the fraction is recorded as the confidence weight.

[0050] In the process of calculating the ratio, in order to avoid the denominator being zero, a preset value needs to be added to the denominator. In this example, the preset value is 0.01.

[0051] When the confidence weight is closer to 0, the changes in the rate of change of the upper and lower temperature differences are more likely to be due to pressure fluctuations; when the confidence weight is closer to 1, the changes in the rate of change of the upper and lower temperature differences are more likely to be due to component migration.

[0052] For the same discrete sampling time, when the signs of the upper temperature difference change rate and the lower temperature difference change rate are opposite, since a single pressure fluctuation cannot physically cause the temperature of adjacent trays to drift in opposite directions, the change in the upper temperature difference change rate and the lower temperature difference change rate is not caused by pressure fluctuation interference, and the confidence weight is assigned the number 1.

[0053] Thus, the logarithmic ratio of the upper and lower temperature differences at discrete sampling times and the confidence weights are obtained.

[0054] Step S003: For the same discrete sampling time, calculate the component migration risk value based on the value of the logarithmic ratio of the upper and lower temperature differences and the confidence weight. Calculate the stability weighting factor based on the standard deviation of temperature fluctuations and the baseline noise constant. The stability weighting factor is used to correct the degree of influence of component migration risk.

[0055] To obtain high-confidence risk assessment indicators, it is necessary to eliminate pressure interference and set up independent logical judgment methods for severe fault conditions such as "component penetration" or "temperature inversion" to prevent the control system from experiencing logical deadlock under extreme conditions.

[0056] When the logarithmic ratio of the upper and lower temperature differences at the discrete sampling time is greater than or equal to 0, the temperature difference between the upper adjacent trays is greater than the temperature difference between the lower adjacent trays, and the nitrogen front is located above or level with the extraction port. This is a normal "edge-locking" control area. The confidence weight can be used as the weighting coefficient of the logarithmic ratio of the upper and lower temperature differences to clean the position signal. Specifically, the product of the logarithmic ratio of the upper and lower temperature differences at the discrete sampling time and the confidence weight is recorded as the component migration risk value at the discrete sampling time.

[0057] When the confidence weight is closer to 0, even if the logarithmic ratio of the temperature difference is large, the component migration risk value obtained after cleaning will approach 0, thus avoiding the system from malfunctioning in response to pressure fluctuations. When the confidence weight is closer to 1, it is more likely that a real component migration has occurred. The component migration risk value can truly reflect the degree of approximation of the components and drive the system to make adjustments.

[0058] When the logarithmic ratio of the temperature difference between the upper and lower sections at a discrete sampling time is less than 0, the temperature difference between the upper adjacent trays is less than that between the lower adjacent trays. This typically corresponds to two dangerous situations: the first is that the nitrogen front has penetrated the extraction port and moved downwards; the second is that severe flooding occurs on the trays, leading to an inverted temperature distribution. To avoid the discrete sampling time corresponding to a logarithmic ratio of the temperature difference between the upper and lower sections less than 0 being mistakenly judged as "zero risk," causing the system to reduce the reflux ratio and exacerbate the accident, when the logarithmic ratio of the temperature difference between the upper and lower sections at a discrete sampling time is less than 0, it is determined that an abnormal operating condition has occurred at the discrete sampling time, and the component migration risk value at the discrete sampling time is assigned a preset high-risk penalty value.

[0059] The high-risk penalty value is a preset constant that is much larger than the logarithmic ratio of the temperature difference between the upper and lower parts during normal operation. In this embodiment, the high-risk penalty value is set to 5.0.

[0060] It is understandable that when the logarithmic ratio of the temperature difference between the upper and lower parts at the discrete sampling time is less than 0, assigning the component migration risk value at the discrete sampling time to a preset high-risk penalty value can force the control system to perceive extremely high risk costs, thereby immediately increasing the reflux ratio and pulling the operating condition back to the normal range.

[0061] Besides component risk, the physical carrying capacity of the tray fluid is also crucial in determining the control boundary. When fluid conditions deteriorate, the system's ability to resist external disturbances decreases. In this case, risk costs must be amplified non-linearly to suppress aggressive operations. The physical carrying capacity of the tray fluid refers to its operational stability.

[0062] For the same discrete sampling time, the square of the ratio of the standard deviation of temperature fluctuation to the reference noise constant is recorded as the first square value. The sum of the product of the first square value and the preset stiffness-sensitive gain with the number 1 is recorded as the stability weighting factor.

[0063] The stiffness-sensitive gain is used to adjust the penalty for instability factors. In this embodiment, the stiffness-sensitive gain is set to 5.0.

[0064] The first squared value is used to construct nonlinear constraints. Specifically, when the standard deviation of temperature fluctuation is less than or close to the reference noise constant, the value of the stability weighting factor is greater than 1.0 and less than the sum of the preset stiffness-sensitive gain and the number 1. The value is small and does not affect the normal energy consumption optimization. When the standard deviation of temperature fluctuation is significantly greater than or close to the reference noise constant, the value of the stability weighting factor increases sharply in a parabolic manner, which can quickly amplify the risk weight.

[0065] Thus, the component migration risk value and stability weighting factor at discrete sampling times are obtained.

[0066] Step S004: Based on the stability weighting factor and component migration risk value at discrete sampling times, construct a comprehensive target value at discrete sampling times. The comprehensive target value is a quantitative assessment result of the operating economy and safety of the working condition. Combined with a preset control cycle, adaptive control of the upper tower reflux ratio of the air separator is realized, thereby achieving optimized adjustment of energy consumption.

[0067] A comprehensive evaluation system for linear performance consumption cost and nonlinear risk penalty is constructed, and a discrete gradient search strategy that conforms to the characteristics of industrial processes is adopted to drive the air separation system to automatically approach the minimum energy consumption point at the safety boundary.

[0068] To quantitatively assess the operational economy and safety of the current operating conditions, a comprehensive target value is constructed based on two parts: "operating cost item" and "risk penalty item," according to the stability weighting factor and component migration risk value at discrete sampling times.

[0069] The positive correlation between the stability weighting factor and the component migration risk value at discrete sampling time is denoted as the comprehensive target value at discrete sampling time.

[0070] It is understood that positive correlation processing is applied to the stability weighting factor and component migration risk value at discrete sampling times, ensuring that the stability weighting factor and the component migration risk value are positively correlated with the comprehensive target value. It is understood that the positive correlation in this application refers to the relationship between the independent and dependent variables, where the independent variables are the stability weighting factor and component migration risk value at discrete sampling times, and the dependent variable is the comprehensive target value at discrete sampling times. The positive correlation means that the dependent variable increases (decreases) as the independent variable increases (decreases), and can be an additive or multiplicative relationship.

[0071] Preferably, as an embodiment of this application, the formula for calculating the comprehensive target value at discrete sampling times is: in, Represents discrete sampling time The overall target value; , , These represent the preset first weighting coefficient, second weighting coefficient, and third weighting coefficient, respectively. In this embodiment, the values ​​of the first weighting coefficient, second weighting coefficient, and third weighting coefficient are 1.0, 0.5, and 3.0, respectively. Represents discrete sampling time The set value of the upper tower reflux ratio; Represents discrete sampling time Stability weighting factor; Represents discrete sampling time Component migration risk value; This represents an exponential function with the natural constant as its base.

[0072] It is understandable that the value of the upper column reflux ratio is positively correlated with the energy consumption of the raw material air compressor, representing linear cost; the upper column reflux ratio at the first discrete sampling moment is determined by those skilled in the art based on the working requirements of the air separation system; the stability weighting factor serves as the gain coefficient of the risk term, and when the fluid is unstable, even if the component migration risk value is small, the total risk cost will be significantly amplified; the preset first weighting coefficient, second weighting coefficient, and third weighting coefficient are used to balance the magnitude of economic efficiency and safety.

[0073] Air separators exhibit significant time lag and inertia. Traditional rapid adjustments can lead to system divergence. Therefore, a strict engineering setting is required for the control cycle of the upper reflux ratio in the air separator to ensure that its length covers the steady-state response time of the device to reflux ratio adjustments. Typically, the response time of an industrial air separation unit is between 10 and 20 minutes. Therefore, in this embodiment, the control cycle for the upper reflux ratio is set to 900 seconds, or 15 minutes. This means that the controller executes an upper reflux ratio control decision every 15 minutes, allowing the air separation system sufficient time to establish a new gas-liquid balance.

[0074] The discrete sampling moment at which the control decision on the upper tower reflux ratio is made after a preset control cycle is recorded as the adjustment moment. The difference between the comprehensive target value of the adjustment moment and the previous adjacent adjustment moment is recorded as the comprehensive target difference of the adjustment moment. When the comprehensive target difference is less than 0, the benefit of energy consumption reduction outweighs the cost of increased risk, the optimization direction is correct, and the difference between the upper tower reflux ratio at the adjustment moment and the preset adjustment step size is used as the value of the upper tower reflux ratio at the next adjacent discrete sampling moment. When the comprehensive target difference is greater than or equal to 0, the risk penalty increases sharply, offsetting the energy consumption benefit, and the air separation device reaches the safety boundary. The sum of the upper tower reflux ratio at the adjustment moment and the preset adjustment step size is used as the value of the upper tower reflux ratio at the next adjacent discrete sampling moment. At the same time, the system enters a cooling lock state. Specifically, the lock counter is activated, and for four consecutive preset control cycles after the adjustment moment, the air separation system forcibly maintains the set value of the upper tower reflux ratio at the next adjacent discrete sampling moment unchanged, without attempting to reduce energy consumption.

[0075] The preset adjustment step size is 0.5% of the reflux ratio at the first discrete sampling time.

[0076] The cooling lockout state ensures that the air separator can operate stably for a long time after reaching the boundary. Only after confirming that the operating conditions are completely stable will a new energy saving attempt be initiated, thus achieving robust control that balances energy consumption optimization and long-term stability.

[0077] It is worth noting that when the purity of the bottom product or the liquid level in the lower column is detected to exceed the safety threshold, the adaptive control is interrupted and the reflux ratio in the upper column is locked to the preset safety upper limit value. The adaptive adjustment is resumed after the parameters return to normal. The safety threshold is set by those skilled in the art.

[0078] This achieves adaptive control of the upper tower reflux ratio of the air separator, reducing energy consumption.

Claims

1. A method for optimizing and adjusting energy consumption in an air separation system, characterized in that, The method includes the following steps: A three-point observation structure was adopted to collect the temperature of the extraction layer, the temperature of the upper tray, and the temperature of the lower tray at each discrete sampling time, and to calculate the temperature difference between adjacent trays, the standard deviation of temperature fluctuation, and the reference noise constant at the discrete sampling time. Based on the temperature difference between adjacent trays, the logarithmic ratio of the temperature difference between the upper and lower trays is calculated. The logarithmic ratio of the temperature difference between the upper and lower trays is used to characterize the probability that a nitrogen front approaches the extraction port from above. Based on the temperature difference between adjacent trays at two discrete sampling times with a preset time interval, a confidence weight is constructed. The confidence weight is used to characterize the credibility of temperature changes caused by component migration. For the same discrete sampling time, the component migration risk value is calculated based on the value of the logarithmic ratio of the upper and lower temperature differences and the confidence weight. The stability weighting factor is calculated based on the standard deviation of temperature fluctuation and the reference noise constant. The stability weighting factor is used to correct the degree of influence of component migration risk. Based on the stability weighting factor and component migration risk value at discrete sampling times, a comprehensive target value is constructed for discrete sampling times. The comprehensive target value is a quantitative assessment result of the operating economy and safety of the working condition. Combined with a preset control cycle, adaptive control of the upper tower reflux ratio of the air separator is realized, thereby achieving optimized adjustment of energy consumption.

2. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The specific calculation method for the temperature difference between adjacent trays is as follows: The temperature difference between adjacent trays includes the temperature difference between upper adjacent trays and the temperature difference between lower adjacent trays; For the same discrete sampling time, the temperature difference between adjacent upper trays is the absolute value of the difference between the temperature of the upper tray and the temperature of the extraction layer, and the temperature difference between adjacent lower trays is the absolute value of the difference between the temperature of the extraction layer and the temperature of the lower tray.

3. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The specific calculation method for the standard deviation of temperature fluctuation is as follows: The variance of the temperature of all sampling layers in the data buffer is denoted as the standard deviation of temperature fluctuation at the last discrete sampling time in the data buffer.

4. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The specific calculation method for the reference noise constant is as follows: When the data in the data buffer completely fills the entire data buffer, the arithmetic mean of the standard deviation of temperature fluctuations at all discrete sampling times in the data buffer is recorded as the reference noise constant; conversely, the arithmetic mean of the standard deviation of temperature fluctuations at the first to the second-to-last discrete sampling times in the data buffer is recorded as the reference noise constant at the last discrete sampling time in the data buffer.

5. The energy consumption optimization and adjustment method for an air separation system according to claim 2, characterized in that, The specific calculation method for the logarithmic ratio of the upper and lower temperature differences is as follows: For the same discrete sampling time, the ratio of the temperature difference between the upper adjacent trays to the temperature difference between the lower adjacent trays is taken as the independent variable of the logarithmic function with the natural constant as the base, and the calculated value of the logarithmic function is denoted as the logarithmic ratio of the upper and lower temperature differences.

6. The energy consumption optimization and adjustment method for an air separation system according to claim 2, characterized in that, The confidence weights are constructed as follows: The rate of temperature change is calculated based on the temperature difference between the upper adjacent tray and the lower adjacent tray at two discrete sampling times with a preset time interval. The rate of temperature change includes the rate of temperature change of the upper tray and the rate of temperature change of the lower tray. For the same discrete sampling time, when the signs of the upper and lower temperature difference change rates are the same, the absolute value of the difference between the upper and lower temperature difference change rates is used as the numerator, and the sum of the absolute values ​​of the upper and lower temperature difference change rates is used as the denominator. The calculated value of the fraction is recorded as the confidence weight; otherwise, the confidence weight is assigned the value 1.

7. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The method for calculating the component migration risk value is as follows: When the logarithmic ratio of the upper and lower temperature differences at discrete sampling time is greater than or equal to 0, the product of the logarithmic ratio of the upper and lower temperature differences and the confidence weight is recorded as the component migration risk value; otherwise, the component migration risk value is assigned a preset high-risk penalty value.

8. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The method for determining the stability weighting factor is as follows: The square of the ratio of the standard deviation of temperature fluctuation to the reference noise constant is denoted as the first square value. The sum of the product of the first square value and the preset stiffness-sensitive gain with the number 1 is denoted as the stability weighting factor.

9. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The overall objective value is the result of a positive correlation between the stability weighting factor and the component migration risk value.

10. The energy consumption optimization and adjustment method for an air separation system according to claim 1, characterized in that, The method for adaptive control of the upper tower reflux ratio of the air separator, combined with a preset control cycle, to optimize energy consumption adjustment, includes the following specific methods: The discrete sampling time at which the control decision on the reflux ratio of the upper tower is made after a preset control cycle is recorded as the adjustment time. The difference between the adjustment time and the comprehensive target value of the previous adjacent adjustment time is recorded as the comprehensive target difference of the adjustment time. When the overall target difference is less than 0, the difference between the upper tower reflux ratio at the adjustment time and the preset adjustment step size is used as the value of the upper tower reflux ratio at the next adjacent discrete sampling time of the adjustment time. Conversely, the sum of the upper reflux ratio at the adjustment time and the preset adjustment step size is used as the value of the upper reflux ratio at the next adjacent discrete sampling time after the adjustment time.

Images