Intelligent start-up method and system for gas separation device

Abstract

The invention discloses an intelligent start-up method and system for a gas separation device, and belongs to the technical field of industrial process automation control. The method comprises the steps of obtaining a production demand and generating preset control data; obtaining the current operation parameters of the gas separation device, performing anomaly detection analysis on the current operation parameters and a preset monitoring threshold value, and if a potential risk is identified, generating an intervention suggestion and dynamically correcting a process control parameter; according to the corrected process control parameters and preset process control parameters, a propane tower starting instruction is generated and a propane tower is controlled to carry out material output in combination with corresponding rules in a starting expert knowledge base; and when the output material of the propane tower is monitored to be qualified, generating an ethane tower and propylene tower start-up instruction, and controlling the ethane tower and the propylene tower to output the material according to the ethane tower and propylene tower start-up instruction. According to the invention, the stability and safety in the initial working stage are improved, the scientificity and accuracy of a control strategy are improved, and the overall linkage efficiency of the device is improved; and human intervention is reduced.

Description

Technical Field

[0001] This invention relates to an intelligent start-up method and system for a gas separation unit, belonging to the field of industrial process automation control technology. Background Technology

[0002] In industrial settings such as petrochemicals and natural gas processing, gas separation units typically consist of multiple separation units operating in series or parallel, including propane, ethane, and propylene towers. Currently, during startup, key operational parameters such as the start-up and shutdown sequence of each tower, heating rate, and feed switching are usually executed by operators based on experience or through pre-defined procedures.

[0003] In the process of realizing this invention, the inventors discovered that the existing gas separation device start-up methods have at least the following problems: the system control parameters rely on preset static values ​​and cannot be dynamically corrected according to actual fluctuations during operation, which can easily cause abnormalities such as excessively rapid heating and liquid level imbalance; the start-up rhythm between each tower depends on manual judgment of material qualification and equipment stability, lacking a unified rhythm coordination mechanism, which can easily lead to increased equipment waiting time or process coupling out of control; the control strategy is difficult to self-optimize and cannot automatically adjust the strategy template based on historical operating results, which limits the intelligence level of the system.

[0004] Existing technologies lack the ability to deeply analyze real-time operating data, cannot achieve dynamic linkage correction between process parameters and monitoring thresholds, and lack an effective intelligent rhythm control mechanism between towers, relying heavily on human experience for control decisions. Summary of the Invention

[0005] To improve the safety of gas separation units during startup, this invention proposes an intelligent startup method and system for gas separation units.

[0006] The technical solution adopted by this invention to solve its technical problem is as follows: In a first aspect, an embodiment of the present invention provides a method for intelligent start-up of a gas separation device, comprising the following steps: Obtain production requirements and generate preset control data based on production requirements. The preset control data includes preset process control parameters and preset monitoring thresholds. The current operating parameters of the gas separation unit are obtained, and anomaly detection and analysis are performed based on the current operating parameters and preset monitoring thresholds. If potential risks are identified, intervention suggestions are generated and process control parameters are dynamically corrected. The gas separation unit includes a propane tower, an ethane tower and a propylene tower. Based on the revised process control parameters and the preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, a propane tower start-up command is generated, and the propane tower is controlled to produce materials according to the propane tower start-up command. Once the output material from the propane tower is detected to be qualified, an ethane tower start-up command is generated, and the ethane tower is controlled to produce material according to the ethane tower start-up command. Once the output material from the ethane tower is detected to be up to standard, a start-up command for the propylene tower is generated, and the propylene tower is controlled to produce material according to the start-up command.

[0007] As one possible implementation of this embodiment, the preset process control parameters include feed rate, feed target value, heating rate, heating target value, reflux flow target value, reflux tank level threshold, bottom level threshold, and top pressure threshold.

[0008] As one possible implementation of this embodiment, the preset monitoring thresholds include the high and low liquid level alarm threshold at the bottom of the tower, the abnormal fluctuation range of the tower pressure, the temperature change slope limit, the judgment threshold for rapid drop in the liquid level of the reflux tank, and the judgment range for fluctuation in the feed flow rate.

[0009] As one possible implementation of this embodiment, the step of acquiring the current operating parameters of the gas separation unit, performing anomaly detection analysis based on the current operating parameters and the preset monitoring threshold, and generating intervention suggestions and dynamically correcting the process control parameters if potential risks are identified includes the following steps: Real-time operating data of the propane tower, ethane tower and propylene tower are collected. The real-time operating data includes the bottom liquid level, top pressure, temperature rise, reflux tank liquid level and feed flow rate. The real-time operating data is compared with the corresponding monitoring threshold to determine whether there is an abnormal state. The abnormal state includes liquid level exceeding the limit, pressure change, temperature slope exceeding the limit, or feed fluctuation exceeding the standard. When an abnormal state is detected, corresponding intervention suggestions are generated. These intervention suggestions include adjusting the heating rate, reducing the feed rate, delaying the start-stop timing of the reflux pump, or raising the liquid level protection line. Based on the intervention recommendations, the relevant process control parameters are dynamically adjusted to ensure that the control state within the tower returns to a safe range and to guarantee the stable execution of subsequent production operations.

[0010] As one possible implementation of this embodiment, the step of generating a propane tower start-up command based on the modified process control parameters and the preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, includes the following steps: Based on the corrected process control parameters, the target operating state of the propane tower is determined, and the rule template that best matches the target operating state in the start-up expert knowledge base is called. The rule template includes the propane tower's feed start-up and shutdown sequence, heating start-up and shutdown conditions, reflux pump activation conditions, and the operation logic of each control valve. Based on the operation logic in the rule template, and combined with the modified feed rate, heating rate and reflux setting value, the propane tower start-up command is generated.

[0011] As one possible implementation of this embodiment, the preset monitoring threshold supports dynamic adjustment during the operation of the gas separation device. The dynamic adjustment includes: Based on the current operating parameters and the trends of monitored variables, and combined with the operating status of adjacent devices, the monitoring thresholds are adjusted. The operating status of adjacent devices includes device start-up and shutdown, load changes, or operation phase switching.

[0012] As one possible implementation of this embodiment, the determination of whether the produced material is qualified includes: The obtained output material composition data, tower top pressure and temperature stability indicators, and reflux tank liquid level change trend are used to determine whether the output material meets the output material quality requirements and operating stability standards. When the judgment result meets the preset qualification standard, the output material is determined to be qualified, and the start-up instruction generation and control process of the next tower is triggered.

[0013] As one possible implementation of this embodiment, the intelligent start-up method for a gas separation device further includes the following steps: The existing templates are evaluated based on historical operational data, intervention and adjustment records, and output quality information within a preset period. When the evaluation results meet the optimization threshold conditions, the optimal execution path, parameter combination, and response instructions are extracted and written into the construction expert knowledge base as new rule templates to improve subsequent matching efficiency and the adaptability of control strategies.

[0014] As one possible implementation of this embodiment, the intelligent start-up method for a gas separation device further includes the following steps: During the execution of the start-up command for the corresponding tower, the operating feedback parameters of the corresponding tower are collected in real time. The operating feedback parameters include the current feed flow rate, bottom liquid level, top pressure, heating temperature and reflux tank liquid level. The presence of execution deviations is determined based on the trend changes of operational feedback parameters. Execution deviations include control response lag, variable offset, and insufficient regulation. When an execution deviation is detected, the control strategy in the current execution process is dynamically adjusted based on the operation feedback parameters. The control strategy includes correcting the heating rate setpoint, updating the feed flow target value, and adjusting the start-stop rhythm of the reflux pump.

[0015] Secondly, an intelligent start-up system for a gas separation device provided in this embodiment of the invention includes: The control data generation module is used to acquire production requirements and generate preset control data based on these requirements. The preset control data includes preset process control parameters and preset monitoring thresholds. An anomaly detection and parameter correction module is used to acquire the current operating parameters of the gas separation unit and perform anomaly detection analysis based on the current operating parameters and preset monitoring thresholds. If potential risks are identified, intervention suggestions are generated and process control parameters are dynamically corrected. The gas separation unit includes a propane tower, an ethane tower, and a propylene tower. The propane tower control module is used to generate a propane tower start-up command based on the corrected process control parameters and preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, and control the propane tower to produce materials according to the propane tower start-up command. The ethane tower control module is used to generate an ethane tower start-up command when the output material of the propane tower is detected to be qualified, and to control the ethane tower to produce material according to the ethane tower start-up command; The propylene tower control module is used to generate a propylene tower start-up command when the output material from the ethane tower is found to be qualified, and to control the propylene tower to produce material according to the propylene tower start-up command.

[0016] The beneficial effects of the technical solutions of the embodiments of the present invention are as follows: The aforementioned technical solution, by acquiring production requirements and generating preset control data including process control parameters and monitoring thresholds, enables the gas separation unit to have pre-configuration capabilities during startup, ensuring that each control variable has a basic operational basis, thus providing sufficient reference conditions for subsequent dynamic control and anomaly judgment. By acquiring the current operating parameters of the gas separation unit and combining them with preset monitoring thresholds for anomaly detection and analysis, intervention suggestions are generated after identifying potential risks, and control parameters are dynamically corrected, enabling real-time adaptive adjustment of the operation process, thereby improving stability and safety in the initial startup phase. By combining the corrected parameters with expert knowledge base rules to generate propane tower startup instructions, knowledge-driven intelligent control instruction generation is achieved, thereby improving the scientific nature and accuracy of the control strategy. By controlling the startup of the ethane tower and propylene tower sequentially based on the qualification judgment of the output materials, rhythm coordination and automatic connection between multiple towers can be achieved, thereby reducing human intervention and improving the overall linkage efficiency of the unit.

[0017] The above-mentioned technical solution, through dynamic parameter correction and real-time anomaly detection, can adaptively adjust the control strategy according to the actual operating status of the unit, improving the stability and safety of the start-up process and reducing the anomaly rate by about 30% compared with existing technologies. Through the start-up expert knowledge base and multi-tower intelligent linkage control, it realizes knowledge-driven intelligent decision-making and rhythm coordination among multiple towers, improving the overall operating efficiency by about 25% and reducing manual intervention by about 60%. Through historical data learning and template optimization, the system has the ability to continuously improve, and the long-term operating effect is continuously improved. Through intelligent decision-making based on expert knowledge base and multi-level anomaly detection, it realizes intelligent control of the gas separation unit start-up process, improving the safety and efficiency of the start-up process. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an intelligent start-up method for a gas separation device according to an exemplary embodiment; Figure 2 This is a flowchart illustrating an anomaly detection and parameter correction method according to an exemplary embodiment; Figure 3 This is a flowchart illustrating a method for performance evaluation of an existing template according to an exemplary embodiment; Figure 4 This is a flowchart illustrating a method for dynamically adjusting the control strategy during the current execution process based on runtime feedback parameters, according to an exemplary embodiment. Figure 5 This is a schematic diagram of an intelligent start-up system for a gas separation device according to an exemplary embodiment. Detailed Implementation

[0019] To more clearly illustrate the technical features of the present invention, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following description is merely exemplary and is not intended to limit the invention.

[0020] In this application, "gas separation unit" is an abbreviation for gas fractionation unit, specifically referring to a complete set of process equipment in the petrochemical field that separates mixed hydrocarbon materials such as liquefied petroleum gas through the principle of distillation to obtain high-purity single-component products. Its core components include at least multiple distillation columns (such as propane columns, ethane columns, and propylene columns) operating in series or parallel, along with supporting feeding systems, heating / cooling systems, reflux systems, pressure control systems, and material conveying pipelines and valves. The function of this unit is to separate mixed raw materials into products such as propane, ethane, and propylene that meet specific quality requirements by controlling process parameters such as temperature and pressure, based on the differences in the volatility of each component.

[0021] like Figure 1As shown in the figure, an intelligent start-up method for a gas separation device provided by an embodiment of the present invention includes the following steps: Step S1: Obtain production requirements and generate preset control data based on production requirements.

[0022] Specifically, the process of obtaining production demand involves calling the start-up task document in the production management system or inputting the target capacity, raw material type, and production sequence requirements through the user input interface. Then, based on the demand, the corresponding tower configuration scheme and standard process package are matched. An initial set of process control parameters is generated by calling parameter templates or calculation logic. This includes the default feed rate, target temperature value, and preset reflux operation value for each tower. At the same time, a set of matching monitoring thresholds is generated to set the safe operating boundary conditions for variables such as liquid level, pressure, temperature, and flow rate. This constitutes a set of preset control data packages to support the generation of subsequent analysis and control commands.

[0023] The preset control data includes preset process control parameters and preset monitoring thresholds; in one embodiment of the present invention, the preset process control parameters include feed rate, feed target value, heating rate, heating target value, reflux flow target value, reflux tank level threshold, bottom level threshold, and top pressure threshold.

[0024] Specifically, the preset process control parameters are generated by referring to the start-up process records of the same type in the historical operating condition database. By combining the current production needs and process design specifications, the initial value of the feed rate and the final target value of the feed rate for the corresponding tower type are determined. At the same time, the range of the heating rate and the required end temperature value of the heating process are set, and the reflux strategy parameters are initialized and configured. The target reflux flow rate is set to maintain the thermal balance and separation effect in the tower. The liquid level control threshold of the reflux tank is set to constrain the reflux material storage and prevent the liquid level from interfering with downstream operations. Furthermore, safety upper and lower limits are set for the bottom liquid level to ensure that the tower operates within the effective liquid level range during the start-up process. The target value of the top pressure is set to limit the steam load and product escape state in the tower. Thus, a set of initial control parameters that meets the requirements of process safety and efficiency is constructed.

[0025] One embodiment of the present invention incorporates multiple key parameters such as feed rate, heating rate, and liquid level threshold into preset process control parameters, which can comprehensively cover the core adjustment variables in the start-up process, thereby laying a good foundation for precise control of the operating status and process stability.

[0026] In one embodiment of the present invention, the preset monitoring thresholds include the tower bottom liquid level high / low alarm threshold, the tower pressure abnormal change range, the temperature change slope limit, the reflux tank liquid level rapid drop judgment threshold, and the feed flow rate fluctuation judgment range.

[0027] Specifically, the preset monitoring thresholds are generated based on the results of multi-round simulation data analysis and expert experience extraction. The high and low liquid level alarm thresholds at the bottom of the tower are set to the liquid level values ​​within a certain tolerance range outside the effective operating liquid level corresponding to the current tower type. This is used to prevent the risk of pump cavitation or flooding caused by liquid level fluctuations. The abnormal pressure variation range at the top of the tower is set to the maximum allowable pressure fluctuation ratio range of the equipment. This is used to identify fault trends such as pressure control failure or abnormal reflux. The temperature change slope limit is used to detect whether the heating process is too fast during the heating stage, leading to separation failure or abnormal energy consumption. The rapid drop in liquid level in the reflux tank is used to capture the violent fluctuations in liquid level caused by flow imbalance or sudden increase in discharge in the reflux path. The feed flow fluctuation judgment range is used to identify the stability problem of raw material supply at the feed end. When the above thresholds are triggered or approach the trigger boundary, they can be regarded as potential risks and need to enter the abnormal detection and analysis process.

[0028] One embodiment of the present invention incorporates indicators such as liquid level alarm threshold, pressure fluctuation range, and temperature change slope into the monitoring threshold setting, which can improve the system's sensitivity to abnormal fluctuations in key variables, thereby enabling timely early warning and intervention control of abnormal operating conditions.

[0029] In one embodiment of the present invention, the preset monitoring threshold supports dynamic adjustment during the operation of the gas separation device, and the dynamic adjustment includes: Based on the current operating parameters and the trends of monitored variables, and combined with the operating status of adjacent devices, the monitoring thresholds are adjusted. The operating status of adjacent devices includes device start-up and shutdown, load changes, or operation phase switching.

[0030] Specifically, the "adjacent equipment" refers to peripheral devices that have a material flow, heat energy, or control path coupling relationship with the current target tower (such as a propane tower, ethane tower, or propylene tower). Changes in their operating status can directly or indirectly affect the material supply, heat exchange efficiency, or pressure stability of the target tower. These include, but are not limited to, upstream raw material pretreatment towers, downstream product storage tanks, and shared heat exchange units. First, based on the current operating parameter sequence, fluctuation trend indicators are calculated by comparing it with its historical stable range. For example, the average temperature rise slope, liquid level change rate, and pressure change frequency over three consecutive cycles are considered. Then, combined with the status information of adjacent equipment, such as whether it is in a start-up / shutdown transition, load jump, or process switching stage, a state association mapping model is used to determine whether the current operating condition requires adjustment of the monitoring threshold range. When a high-load fluctuation condition is identified, the error tolerance boundary for liquid level and pressure is automatically widened to avoid frequent false alarms. When a low-load or stagnant condition is identified, the abnormal trigger thresholds for temperature rise and flow are tightened to improve sensitivity. The correction results are applied to the real-time monitoring logic for subsequent abnormal detection and judgment processes, thereby dynamically adapting to changes in the device's operating environment. For example, when the load on the feed pretreatment tower is increased (such as during high-frequency regeneration), the feed flow rate entering the current tower will fluctuate frequently. In this case, the "feed flow rate fluctuation judgment threshold" of the propane tower can be relaxed from ±3% to ±5% to tolerate the input instability during this stage. Another example is when the load on the condenser in the shared heat exchange system decreases due to the temporary shutdown of the downstream propylene tower, resulting in reduced condensation efficiency and increased fluctuations in the tower top temperature. Accordingly, the system can relax the "temperature change slope limit" from 0.3℃ / min to 0.5℃ / min to avoid misjudging condensation abnormalities. Yet another example is when the ethane product tank undergoes a switching operation, causing a temporary increase in back pressure. The system can dynamically adjust the "upper limit of tower top pressure" threshold of the ethane tower from 0.22MPa to 0.25MPa, and restore the original threshold after the tank switching is completed.

[0031] One embodiment of the present invention modifies the monitoring threshold based on the current operating parameters and the trend of the monitored variables, combined with the status of adjacent devices. This allows for dynamic adaptation to the operating characteristics under different working conditions, thereby avoiding false alarms or misadjustments caused by unreasonable static threshold settings and improving the sensitivity and accuracy of system anomaly detection.

[0032] Step S2: Obtain the current operating parameters of the gas separation unit, and perform anomaly detection analysis based on the current operating parameters and preset monitoring thresholds. If potential risks are identified, generate intervention suggestions and dynamically correct process control parameters. The gas separation unit includes a propane tower, an ethane tower, and a propylene tower.

[0033] like Figure 2 As shown, the process of acquiring the current operating parameters of the gas separation unit and performing anomaly detection analysis based on the current operating parameters and the preset monitoring threshold, generating intervention suggestions and dynamically correcting the process control parameters if potential risks are identified, includes the following steps: Step S21: Collect real-time operating data of the propane tower, ethane tower and propylene tower. The real-time operating data includes the bottom liquid level, top pressure, heating temperature, reflux tank liquid level and feed flow rate.

[0034] Specifically, the process of collecting real-time operational data is based on the sampling period and signal source identifier preset by the acquisition command. It calls the latest data values ​​uploaded by level gauges, pressure transmitters, thermocouples, and flow meters distributed in each tower. It sequentially reads the bottom liquid level and top pressure data of the tower, extracts the temperature rise curves in multiple temperature zones, synchronizes the measurement values ​​of the reflux tank level probe, and captures the instantaneous flow rate or short-cycle average value of the feed flow meter through the communication drive interface. All parameter values ​​are structured and stitched together through the data buffer channel to form a complete operational snapshot, which is stored in the status buffer area for subsequent comparison and analysis. The entire acquisition process must ensure that the sampling time of all measuring points is consistent within a controllable range, such as an error of no more than 200 milliseconds, to avoid data correlation distortion.

[0035] Step S22: Compare the real-time operating data with the corresponding monitoring threshold to determine whether there is an abnormal state. The abnormal state includes liquid level exceeding the limit, pressure change, temperature slope exceeding the limit, or feed fluctuation exceeding the standard. Specifically, when performing anomaly judgment, firstly, the corresponding monitoring threshold range or dynamic floating window value is loaded for each operating parameter according to the set monitoring point name. Then, the deviation value between the current parameter and its set range is calculated and an out-of-bounds judgment is made. At the same time, the temperature rise value is processed by equal interval difference to obtain the short-term temperature rise slope and to determine whether it exceeds the slope limit setting range. For example, if the temperature rise slope is greater than 3℃ / min, it is identified as a temperature control anomaly. Liquid level exceeding the limit is based on the absolute value comparison of the set upper and lower limit ranges. Pressure change judgment uses a short-term difference change rate exceeding 10% as the threshold. Feed fluctuation is identified based on the variance exceeding the warning level within three consecutive sampling cycles. When any of the above anomaly types is triggered, it is determined that the current state is abnormal and the intervention suggestion process is triggered.

[0036] Step S23: When an abnormal state is determined, a corresponding intervention suggestion is generated. The intervention suggestion includes adjusting the heating rate, reducing the feed rate, delaying the start and stop of the reflux pump, or raising the liquid level protection line. Specifically, the process of generating intervention suggestions calls the matching intervention strategy library entries according to the anomaly type and performs matching analysis on the current state. When the heating slope is abnormal, it is recommended to reduce the target heating rate proportionally, for example, from 3℃ / min to 2℃ / min. When a violent fluctuation in feed is detected and the reflux level drops synchronously, it is recommended to reduce the feed rate to 85% of the current value. If the liquid level drops rapidly and triggers the critical point of exceeding the limit, and the current reflux pump is in operation, it is recommended to delay the triggering cycle of the next start / stop signal of the reflux pump to prevent liquid level oscillation. If the bottom liquid level is in a suspicious area, the liquid level protection line threshold is adjusted to expand the control tolerance range to prevent false triggering of alarm logic. After all suggestions are generated, they are output as the target parameter value pairs to be corrected through parameter mapping logic.

[0037] Step S24: Based on the intervention recommendations, dynamically correct the relevant process control parameters to ensure that the control state inside the tower returns to a safe range and to ensure the stable execution of subsequent production operations.

[0038] Specifically, during dynamic correction, the corrected values ​​of each parameter in the intervention suggestion are compared and updated with the current process control parameters. Control items within the adjustment range directly override the original set values. For example, the target heating rate is corrected from 3℃ / min to 2.5℃ / min, or the target feed rate is adjusted from 100m³ / h to 90m³ / h. To enhance the intelligence and adaptability of parameter adjustments during dynamic correction, fuzzy logic reasoning rules are introduced as the decision engine for parameter correction. Combining the comprehensive state of multiple real-time operating variables, key control parameters such as the heating rate, target feed flow rate, and reflux pump start / stop logic are flexibly adjusted. The fuzzy logic reasoning mechanism mainly includes four execution stages: variable fuzzification, rule matching, fuzzy reasoning, and defuzzification. First, the collected real-time operating parameters, such as tower top pressure, tower bottom liquid level, heating rate, and reflux tank liquid level, are categorized into fuzzy language sets such as "too high," "too low," "moderate," and "rapid rise," and quantitative modeling is performed using triangular membership functions. Then, the fuzzified input states are matched with preset rule templates. These rules are preset by on-site process experts based on control experience. For example: IF heating rate is "too slow" AND tower bottom liquid level is "too low" AND tower top pressure is "slightly high," THEN feed rate "reduced slightly," heating rate "increased slightly," reflux pump "started early." After matching the rules, the weighted summation value of multiple action results is calculated using a fuzzy inference engine, and defuzzification is completed using methods such as the centroid method. The fuzzy suggestions are then converted into specific correction values, such as increasing the target heating rate from 1.0℃ / min to 1.15℃ / min, decreasing the feed rate from 55% to 52%, and adjusting the reflux pump start-up time from 90 seconds to 30 seconds. If a suggestion involves strategic parameters such as start / stop rhythm or logical boundaries, the strategy management interface is called to synchronously update the decision conditions in the relevant control logic module. At the same time, the parameter versions before and after the correction are recorded and marked to form an adjustment trajectory backup for subsequent safety analysis and backtracking control. After the correction is completed, the current status is automatically marked as "adjustment completed" and enters the stable state monitoring phase to observe whether the correction result is effective and to continuously judge whether there are still new anomalies.

[0039] One embodiment of the present invention collects real-time operating data of propane tower, ethane tower and propylene tower, and compares and judges them with monitoring thresholds. It can detect abnormal trends such as liquid level exceeding the limit and pressure change in real time, thereby triggering control strategy correction in advance. By generating an intervention plan containing adjustment suggestions and dynamically correcting the corresponding process parameters, it can realize the slow release of abnormalities and trend reversal in the process, thereby significantly enhancing the responsiveness of the control system and the level of start-up safety assurance.

[0040] In one embodiment of the present invention, the anomaly detection analysis further includes multivariate coupling analysis, specifically including: Based on historical stable operating condition data, extract the baseline of coupling relationship between multiple key operating parameters; During real-time operation, the actual coupling relationship of multiple key operating parameters is continuously monitored; When the deviation between the actual coupling relationship and the coupling relationship baseline exceeds a preset threshold, it is determined to be an abnormal coupling state; Generate intervention recommendations for abnormal coupling states, including adjusting multiple associated control parameters to restore normal coupling.

[0041] Specifically, the implementation of the multivariate coupling analysis includes: screening variable pairs or groups with stable coupling relationships, including top pressure and reflux tank level, feed flow rate and bottom level, and heating rate and top temperature change; for each set of coupled variables, constructing a baseline for their coordinated changes based on principal component analysis to form a baseline function or multidimensional relationship surface; continuously comparing the deviation between the current observation and the baseline in real-time operation using a rolling window; if the deviation exceeds the set coordinated error threshold, triggering a coupling anomaly flag and determining the current state as a nonlinear fluctuation anomaly.

[0042] To improve the accuracy and foresight of abnormal state identification, in addition to comparing each real-time operating parameter with its corresponding single-variable monitoring threshold, a multivariate coupling analysis model can be constructed and applied to identify the dynamic correlation between multiple key variables. If the coupling relationship is found to be broken, it is determined that the gas separation unit is in a potentially abnormal operating condition. Even if individual variables themselves have not exceeded the limit, early warning and intervention strategies can be generated.

[0043] The construction process of the multivariate coupling analysis model includes the following steps: First, extract continuously sampled real-time parameter data from multiple typical batches during normal operation of the device, and perform collaborative statistical analysis to screen out variable pairs or groups with stable coupling relationships, such as top pressure and reflux tank level, feed flow rate and bottom level, heating rate and top temperature change, etc.; Second, for each group of coupled variables, construct a collaborative change baseline based on principal component analysis to form a baseline function or multidimensional relationship surface, and continuously compare the deviation between the current observation value and the baseline in a rolling window manner during real-time operation; Finally, if the deviation exceeds the set collaborative error threshold, a coupling anomaly flag can be triggered, and the current state can be determined as a nonlinear fluctuation anomaly. For example, during normal operation, there is a clear reverse coupling relationship between top pressure and reflux tank level, that is, when the top pressure rises, the reflux tank level should rise synchronously due to the increased condensation load; conversely, when the reflux tank level drops too quickly, the top pressure generally drops accordingly. When constructing the model, the fitted regression relationship between the two can be extracted based on 100 historical stable operating conditions as L = -8.3·P + 12.5 (where L is the reflux tank level and P is the tower top pressure, both standardized values), and a range of ±10% deviation from this relationship curve is set as the normal coupling zone. When a continuous rise in tower top pressure is observed during actual operation, but the reflux tank level does not rise accordingly, or even falls beyond this threshold range, the coupling relationship is determined to be broken, triggering an abnormal state marker. Even if the tower top pressure and reflux level do not exceed their respective individual thresholds at this time, early identification of anomalies can still be achieved.

[0044] One embodiment of the present invention employs multivariate coupling analysis technology, which can identify system anomalies at an early stage when a single parameter does not exceed the limit, thereby achieving the technical effect of early warning and intervention.

[0045] In one embodiment of the present invention, the anomaly detection analysis further includes trend prediction analysis, specifically including: Trend prediction of key operating parameters based on time series analysis algorithms; When the prediction results indicate that the parameter will exceed the safety threshold within a preset time, an early warning signal is generated in advance. Preventive intervention recommendations are generated based on early warning signals, and adjustments are made before parameters actually exceed limits.

[0046] One embodiment of the present invention uses time series analysis for trend prediction, which can realize the transformation from passive response to proactive prevention, thereby achieving the technical effect of preventing problems before they occur.

[0047] Step S3: Based on the corrected process control parameters and the preset process control parameters, and combined with the corresponding rules in the start-up expert knowledge base, generate a propane tower start-up command, and control the propane tower to produce materials according to the propane tower start-up command.

[0048] In one embodiment of the present invention, the step of generating a propane tower start-up command based on the modified process control parameters and the preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, includes the following steps: Step S31: Determine the target operating state of the propane tower based on the corrected process control parameters, and call the rule template in the start-up expert knowledge base that best matches the target operating state. The rule template includes the feed start-up and shutdown sequence of the propane tower, the heating start-up and shutdown conditions, the reflux pump activation conditions, and the operation logic of each control valve. Specifically, the process of determining the target operating state involves analyzing the corrected feed rate, heating rate, and initial reflux parameters, combined with the current stage of the unit's startup, such as the early stage of heating, the middle stage of reflux establishment, or the early stage of feed transition. The required operating mode label is extracted and used as a key retrieval condition. Multiple startup strategy templates stored in the startup expert knowledge base are searched. In each template, the set parameter range is extracted and similarity matching calculation is performed with the current state. The rule template that is closest to the current control target is selected as the basis for subsequent execution. This rule template must meet the load response capability, control logic adaptability, and fault tolerance capability under the current operating state.

[0049] Specifically, the rule template structure includes multiple process control sections. The feed start-stop sequence describes the flow ratio adjustment path from preheating completion to cold feed switching and then to full-flow feed. The heating start-stop conditions define the start and end points of the heating section control based on temperature measurement points or temperature rise slope. The reflux pump activation conditions set the trigger threshold for switching the reflux mode based on the reflux tank level or the temperature change trend of the top condenser section. The operation logic of each control valve includes the opening and closing sequence, control curve and holding state of the shut-off valve, switching valve and regulating valve. For example, the bottom feed valve needs to be opened after the pressure stabilizes and linked with the heating electric heater for power ramp-up control. The entire template is organized into an execution graph according to the process logic sequence to support the generation and calling of subsequent instructions.

[0050] Step S32: Based on the operation logic in the rule template, and combined with the corrected feed rate, heating rate and reflux setting value, generate the propane tower start-up command.

[0051] Specifically, the start-up command generation process loads the corresponding control node parameters item by item according to the operation sequence of each stage in the template. At each control point, the currently corrected parameter value is loaded to dynamically overwrite or fine-tune the preset value in the rule template to improve the execution matching degree, forming multiple execution commands including feed start command, heating start command, reflux start command and pressure control command. Each command includes target variable value, control duration, trigger criteria and linkage control conditions. For example, when generating the feed start command, the current target feed rate is set as the control value and the precondition is set as "bottom liquid level ≥ 80% of the set value and top pressure stable for 3 minutes". Finally, all control actions are integrated into a unified start-up task list and sorted with execution time axis marks to realize segmented execution control and closed-loop management of the propane tower start-up process.

[0052] One embodiment of the present invention determines the target operating state based on the modified process control parameters and calls the matching expert knowledge base rule template, which enables the linkage decision-making of dynamic parameters and historical experience, thereby improving the adaptability and accuracy of the control strategy. By extracting operation rules such as feeding start-stop sequence and heating logic from the template and generating specific start-up instructions in combination with the current modified parameters, the operation process can be standardized and human judgment errors can be reduced, thereby improving the safety and consistency of instruction execution.

[0053] In one embodiment of the present invention, the construction expert knowledge base includes the following components: The rule template library stores control strategy templates for different operating conditions. Each template includes the target operating state, applicable raw material characteristics, key control parameter combinations, execution phase operation sequence, control valve logic and safety response conditions. Historical case database stores operational data, intervention and adjustment records, and output quality information from the historical construction process; A parameter optimization library that stores the optimal parameter combinations obtained from historical data mining. The inference engine matches the most suitable rule template based on the current running state and generates control instructions.

[0054] The initial rules for the expert knowledge base were entered during the system deployment phase. Specifically, this involved organizing process engineers with extensive field experience to analyze the correspondence between various typical operating conditions, equipment states, and control strategies in the gas separation unit's start-up process. Combining standard operating procedures, historical operating condition records, and safety control specifications, representative process control paths were extracted and converted into structured rule templates before being imported into the knowledge base. Each initial template includes at least the following information fields: target operating state, applicable raw material characteristics, key control parameter combinations, execution phase operation sequence, control valve logic, and safety response conditions. This ensures that the system has a foundation of directly callable standard strategies in the initial stage of deployment.

[0055] In one embodiment of the present invention, the construction of the construction expert knowledge base includes: An initial rule template was constructed through process specification analysis and expert experience extraction. Based on historical construction start data, typical control strategies are extracted using data mining algorithms; The extracted control strategies are transformed into structured rule templates and stored in the knowledge base; After each work process is completed, the rule template is optimized and updated based on the execution results.

[0056] The construction of the expert knowledge base integrates two approaches: manual rule-making and data mining. On the one hand, by surveying the operational experience of on-site personnel and standard process procedures, control strategies for typical scenarios are extracted and organized into structured "condition-response" rule templates. For example, when the target operating state is "rapid start-up" and the propane content in the raw material components is higher than 85%, it is recommended to adopt the feeding sequence A→B→C, with a maximum heating rate of 1.5℃ / min, while defining the activation logic of the reflux pump and the valve control rhythm. On the other hand, by performing cluster analysis and rule extraction on a large number of historical start-up batches of operating parameter sequences, intervention and adjustment records, and final output quality, data-driven strategy templates are generated using decision tree algorithms and introduced into the knowledge base to further improve the template coverage and adaptability.

[0057] Furthermore, the knowledge base has the ability to continuously evolve. That is, after the start-up process is completed, the execution path and output results are periodically evaluated. When the optimization threshold conditions are met, the optimal control parameters and response instructions in the path are automatically extracted and written into the knowledge base to form a new template after manual confirmation. It can also learn from successful intervention operations in real time during operation to generate candidate rules for subsequent recommendation and invocation, thereby realizing adaptive learning and long-term optimization of the start-up control strategy.

[0058] One embodiment of the present invention employs a knowledge base construction technique that combines process specification analysis, expert experience extraction, and data mining. This technique can construct a comprehensive and highly adaptable expert knowledge base, thereby improving the scientific nature of control decisions.

[0059] Step S4: Once the output material of the propane tower is found to be qualified, an ethane tower start-up command is generated, and the ethane tower is controlled to produce material according to the ethane tower start-up command.

[0060] Specifically, the criteria for determining the quality of the propane tower output include, but are not limited to, product sampling analysis, online component detection data, and the stability of the device operating parameters. When key product indicators are found to meet the switching criteria, and the reflux tank level fluctuation is within a controllable range, and the tower top pressure and temperature are stable within the set deviation range, the start-up logic module is triggered to call the ethane tower control parameter template and generate an ethane tower start-up command based on the current control status. This command drives the ethane tower to start the corresponding feeding and heating operations and initialize the reflux settings, ensuring that the conditions inside the tower gradually transition to the target operating range and begin a new stage of material separation process.

[0061] Step S5: Once the output material from the ethane tower is found to be qualified, a start-up command for the propylene tower is generated, and the propylene tower is controlled to produce material according to the start-up command.

[0062] Specifically, the process of determining whether the ethane tower output material is qualified also combines online quality monitoring values, reflux level stability, and fluctuations in tower top pressure and temperature. When all key process variables are maintained within the set control range and the product quality meets the switching requirements, an automatic start-up control command for the propylene tower is generated. This command triggers the opening of the propylene tower feed line, adjustment of the heating power setting, and activation of the corresponding reflux strategy. During the start-up process, the operating status is continuously tracked to dynamically optimize and adjust the process control parameters, ensuring that the propylene tower can smoothly connect to the overall start-up process based on the stable output of the previous tower, and achieving seamless transition control under multi-tower linkage.

[0063] In one embodiment of the present invention, the determination of whether the produced material is qualified includes: 1) Obtain the output material composition data, the stability index of tower top pressure and temperature, and the trend of reflux tank liquid level change to determine whether the output material meets the output material quality requirements and operating stability standards. Specifically, the process of determining whether the output material is qualified involves collecting online analysis data of the output material to obtain the content of the main components and the ratio of impurities. At the same time, the coefficient of variation of the continuous sampling values ​​of the pressure and temperature at the top of the tower is calculated to determine the degree of fluctuation. Then, the trend is determined by combining the rate of change and stability threshold of the liquid level in the reflux tank over a continuous time period. When the component data shows that the content of the main component meets the preset purity requirements and the impurity content is at the lower limit of the allowable range, and the pressure and temperature fluctuations at the top of the tower do not exceed the preset range, such as ±0.5℃ and ±1kPa, and the liquid level in the reflux tank does not show a sharp drop or irregular fluctuation, it can be comprehensively judged that the current output material meets the stable production standards in terms of quality and operating status, and the current output material of the tower is identified as "meeting the switchable conditions".

[0064] The stability index can be calculated based on the degree of data fluctuation within a continuous time window, primarily using the standard deviation method. Taking the tower top pressure as an example, during the unit's output stage, pressure data sample sequences {x1, x2, ..., x} over the past 10 minutes are collected on a rolling basis. n}, calculate its standard deviation, When σ is less than the set stability standard value (e.g., 0.05 MPa), the pressure fluctuation is considered to be within a stable range; if it exceeds this standard, the output stage is considered to have fluctuation risk and cannot be considered as qualified output. Similarly, temperature fluctuations can be set with a standard deviation limit (e.g., 0.2℃) for normalization evaluation, facilitating stability comparisons of variables with different absolute values. Furthermore, the trend of the reflux tank level within a fixed window (e.g., 5 minutes) can be used to assess whether the system exhibits continuous level decline or abrupt changes, serving as an auxiliary criterion for operational stability. Only when pressure fluctuations, temperature fluctuations, and level trends all meet the preset stability index thresholds, and the output material composition data is within the quality acceptable range, does the system confirm the output of this stage as "qualified output" and trigger the subsequent tower start-up command generation and control process execution.

[0065] 2) When the judgment result meets the preset qualification standard, the output material is determined to be qualified, and the start-up instruction generation and control process of the next tower is triggered.

[0066] Specifically, to enhance the system's comprehensive ability to assess the quality and operational status of output materials, the system does not rely solely on whether any single variable exceeds a threshold when determining compliance. Instead, it combines the results of multiple indicators and processes them through logical fusion rules to achieve a comprehensive assessment of whether the final output material is qualified. Within each production cycle, the system generates corresponding status labels or quantitative scores for the four types of indicators mentioned above. For example: whether the component data is within the acceptable concentration range (qualified as 1, deviation as 0); whether the standard deviation of the column top pressure fluctuation is less than 0.05 MPa (qualified as 1, otherwise 0); whether the column top temperature change is within ±0.3℃ / min (qualified as 1, otherwise 0); and whether the reflux tank level trend is continuous and stable or maintains a slow rise (qualified as 1, abnormal as 0). Based on this, the system can use a logical "AND" operation as the simplest form of judgment criterion. That is, when all four indicators are "1", the output material is considered qualified; when any indicator is "0", the generation of the next column start-up instruction is temporarily suspended, and the system enters an abnormal verification state. Once the output material is deemed qualified, the start-up chain rhythm management logic is immediately initiated. The end signal output from the previous tower and the evaluation result serve as the trigger source to activate the start-up judgment module of the next tower. The preset control parameters of the next tower and the previously generated template structure are loaded. Based on the current unit status, such as heat exchanger temperature or remaining raw material, the optimal delay time is calculated or the zero-wait switching process is directly executed. Subsequently, the instruction generation process is triggered, and stage action instructions such as feeding, heating, and reflux are output to the scheduling platform. At the same time, the status of the previous tower is marked as "qualified and completed", ensuring the stability of the process rhythm and the maximization of output continuity under the multi-tower series connection.

[0067] Based on specific quantitative indicators, such as component purity, operational stability, and equipment status, the process for determining the quality of the output material is as follows: Obtain the component data of the output material. When propane purity ≥ 95%, ethane purity ≥ 90%, and propylene purity ≥ 99.5%, the components are considered qualified. Calculate the standard deviation of the tower top pressure over 10 consecutive minutes. When the standard deviation ≤ 0.05 MPa, the pressure is considered stable. Calculate the fluctuation range of the tower top temperature over 10 consecutive minutes. When the fluctuation range ≤ ±1.5℃, the temperature is considered stable. Monitor the trend of the reflux tank level change. When the level change rate ≤ ±2% / min and there are no drastic fluctuations, the level is considered stable. When the conditions of component qualification, pressure stability, temperature stability, and level stability are simultaneously met, the output material is determined to be qualified. Based on multiple quantitative indicators, the compliance status of each indicator is automatically calculated. When all indicators simultaneously meet the requirements, the output material is determined to be qualified.

[0068] One embodiment of the present invention combines material composition analysis data, equipment operation stability indicators and reflux tank liquid level change trends to determine whether the produced material meets quality requirements and stability standards, and constructs an objective and reliable qualification judgment mechanism. This ensures that the next tower starts only on the premise that the output of the previous tower is sufficiently stable, thereby improving the rationality of multi-tower rhythm control and the product consistency of the overall process.

[0069] In one embodiment of the present invention, such as Figure 3 As shown, the intelligent start-up method for a gas separation device further includes the following steps: Step S5: Within a preset period, the existing template is evaluated based on the operation data, intervention and adjustment records and output quality information from the historical start-up process.

[0070] Specifically, the performance evaluation process involves retrieving key operating parameter sequences, manually or automatically generated intervention records, and corresponding batch material output quality indicators from multiple historical start-up batches. It then performs a structured comparative analysis of the execution paths and results under the same rule template. Multi-dimensional scoring is achieved using quantitative indicators such as heating time, steady-state establishment time, qualified product ratio, and the number and magnitude of interventions. A comprehensive performance scoring model is constructed by combining the efficiency of achieving the start-up rhythm. For example, a template exhibiting a high intervention frequency and low qualified rate in three start-up batches will be marked as a low-priority template. Conversely, a template that can quickly establish stable output with minimal interventions in multiple executions will be rated as an efficient template and marked as a candidate rule for updating.

[0071] Step S6: When the evaluation result meets the optimization threshold condition, extract the optimal execution path, parameter combination and response instruction, and write the optimal execution path, parameter combination and response instruction as a new rule template into the construction expert knowledge base to improve the efficiency of subsequent matching and the adaptability of control strategies.

[0072] Specifically, when the evaluation score reaches the preset optimization threshold, the sequence of execution path nodes, the target control parameter values ​​effective at each stage, and the corresponding control action execution conditions are immediately extracted from the corresponding start-up process record. These are combined into a set of closed-structure control strategy templates. Each sub-step in the template is bound to a clear judgment threshold, execution trigger condition, and feedback response mechanism. For example, the template records that when the tower top temperature reaches 90°C during the heating stage, the reflux pump should be activated and the temperature should be increased at a rate of 0.8°C / min to 110°C, and the reflux should be kept stable for no less than 10 minutes. The template also includes the final output quality label and time axis mark. After being written into the knowledge base, it can be used as the first priority in the next round of matching, thereby realizing rapid invocation and adaptive control optimization based on historical best practices in future execution.

[0073] One embodiment of the present invention quantifies the differences in the effects of different control templates in actual operation by performing performance evaluation on historical operating data and intervention adjustment records within a preset period, thereby identifying the optimal strategy combination; by updating the optimal execution path and parameter results to a new knowledge template, the dynamic evolution of the expert knowledge base can be realized, thereby enhancing the system's learning ability and long-term self-optimization ability for complex working conditions.

[0074] In one embodiment of the present invention, such as Figure 4 As shown, the intelligent start-up method for a gas separation device further includes the following steps: Step S7: During the execution of the start-up command of the corresponding tower, the operating feedback parameters of the corresponding tower are collected in real time. The operating feedback parameters include the current feed flow rate, bottom liquid level, top pressure, heating temperature and reflux tank liquid level. Specifically, the process of collecting operational feedback parameters begins after the command execution starts. The feedback monitoring thread is activated, and real-time data tags in the control platform are called at a sampling frequency of seconds. The current feed flow rate, bottom liquid level, top pressure reading, temperature curve data at multiple height points of the tower, and floating-point value of the reflux tank liquid level are extracted. The status response record package is formed by synchronizing the timestamp with the execution target in the start-up command. After collection, the record package enters the feedback buffer for rolling trend analysis. At the same time, a short-term cache window is set to save the parameter change sequence of the most recent 10-30 seconds in order to judge the continuity of the control response and the actual execution effect.

[0075] Step S8: Determine whether there is an execution deviation based on the trend changes of the running feedback parameters. The execution deviation includes control response lag, variable offset and insufficient adjustment. Specifically, the process of judging execution deviation is to compare the changing trend of the collected feedback parameters with the execution target trajectory or expected response window and calculate the fitting residual. If the actual temperature rise rate is continuously lower than the target temperature rise slope, it is judged as a lag in temperature rise response. If the bottom liquid level deviates from the target liquid level range for a long time and shows directional drift, it is judged as a liquid level variable offset. If the top pressure does not reach the expected fluctuation range or the convergence period is significantly prolonged after executing the reflux adjustment command, it is considered that the reflux adjustment is insufficient. By analyzing the joint trend indicators of multiple variables and historical matching templates, the deviation type is classified and the current execution status is marked as "abnormal execution" to enter the dynamic strategy correction process.

[0076] Step S9: When it is determined that there is an execution deviation, the control strategy in the current execution process is dynamically adjusted based on the operation feedback parameters. The control strategy includes correcting the heating rate set value, updating the feed flow target value, and adjusting the start-stop rhythm of the reflux pump.

[0077] Specifically, the dynamic adjustment strategy process calls the parameter correction module based on the deviation identification results, and locally fine-tunes the control target values ​​corresponding to the lagging or offset variables in the feedback parameters. For example, the heating rate target is lowered from the original value of 1.0℃ / min to 0.8℃ / min to avoid excessive thermal inertia causing fluctuations. The feed flow rate target value is appropriately reduced by 5% according to the current liquid level drop trend, and the trigger time of the full feed node is delayed. If insufficient reflux regulation is identified, the reflux pump opening time window is extended or the lower limit of the liquid level at the start and stop point of the next cycle is raised. This achieves real-time correction based on feedback closed loop to restore control effect and execution stability. At the same time, the parameter change path before and after the strategy adjustment is recorded for subsequent analysis, learning and template evaluation.

[0078] One embodiment of the present invention continuously collects operational feedback parameters and analyzes their trends during the execution of the start-up command, thereby identifying execution deviations such as control response lag or abnormal variable fluctuations, and thus achieving real-time adjustment during the process; by dynamically correcting the heating rate, feed target value and reflux strategy based on feedback parameters, the control strategy's ability to track the actual operating state can be improved, thereby enhancing the stability and closed-loop adaptive level of the control system during the execution period.

[0079] like Figure 5 As shown in the figure, an intelligent start-up system for a gas separation unit provided in this embodiment of the invention includes: The control data generation module is used to acquire production requirements and generate preset control data based on these requirements. The preset control data includes preset process control parameters and preset monitoring thresholds. An anomaly detection and parameter correction module is used to acquire the current operating parameters of the gas separation unit and perform anomaly detection analysis based on the current operating parameters and preset monitoring thresholds. If potential risks are identified, intervention suggestions are generated and process control parameters are dynamically corrected. The gas separation unit includes a propane tower, an ethane tower, and a propylene tower. The propane tower control module is used to generate a propane tower start-up command based on the corrected process control parameters and preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, and control the propane tower to produce materials according to the propane tower start-up command. The ethane tower control module is used to generate an ethane tower start-up command when the output material of the propane tower is detected to be qualified, and to control the ethane tower to produce material according to the ethane tower start-up command; The propylene tower control module is used to generate a propylene tower start-up command when the output material from the ethane tower is found to be qualified, and to control the propylene tower to produce material according to the propylene tower start-up command.

[0080] In one embodiment of the present invention, the anomaly detection and parameter correction module includes: The parameter acquisition submodule is used to collect real-time operating data of the propane tower, ethane tower and propylene tower. The real-time operating data includes the bottom liquid level, top pressure, heating temperature, reflux tank liquid level and feed flow rate. The anomaly detection submodule is used to compare the real-time operating data with the corresponding monitoring thresholds to determine whether there is an abnormal state. The abnormal state includes liquid level exceeding the limit, pressure change, temperature slope exceeding the limit, or feed fluctuation exceeding the standard. The intervention suggestion submodule is used to generate corresponding intervention suggestions when an abnormal state is detected. The intervention suggestions include adjusting the heating rate, reducing the feed rate, delaying the start and stop timing of the reflux pump, or raising the liquid level protection line. The control parameter correction submodule is used to dynamically correct the relevant process control parameters according to the intervention suggestions, so as to ensure that the control state in the tower returns to a safe range and ensures the stable execution of subsequent production operations.

[0081] In one embodiment of the present invention, the propane tower control module includes: The operating status judgment submodule is used to determine the target operating status of the propane tower based on the corrected process control parameters and call the rule template that best matches the target operating status from the start-up expert knowledge base. The template structure recognition submodule is used to identify the rule templates, including the propane tower feed start-up and shutdown sequence, heating start-up and shutdown conditions, reflux pump activation conditions, and the operation logic of each control valve. The instruction generation submodule is used to generate propane tower start-up instructions based on the operation logic in the rule template, combined with the modified feed rate, heating rate and reflux setpoint.

[0082] In one embodiment of the present invention, the intelligent start-up system for a gas separation device further includes: The template evaluation module is used to evaluate the performance of existing templates within a preset period based on historical operation data, intervention and adjustment records, and output quality information during the construction process. The template update module is used to extract the optimal execution path, parameter combination, and response instruction when the evaluation result meets the optimization threshold condition. The optimal execution path, parameter combination, and response instruction are then written into the construction expert knowledge base as new rule templates to improve the efficiency of subsequent matching and the adaptability of control strategies.

[0083] In one embodiment of the present invention, the intelligent start-up system for a gas separation device further includes: The threshold correction module is used to correct the monitoring threshold based on the current operating parameters and the trend of the monitored variables, combined with the operating status of adjacent devices. The operating status of adjacent devices includes device start-up and shutdown, load changes, or operation phase switching.

[0084] In one embodiment of the present invention, the intelligent start-up system for a gas separation device further includes: The operation feedback acquisition module is used to collect the operation feedback parameters of the corresponding tower in real time during the execution of the start-up command of the corresponding tower. The operation feedback parameters include the current feed flow rate, bottom liquid level, top pressure, heating temperature and reflux tank liquid level. The execution deviation judgment module is used to determine whether there is an execution deviation based on the trend changes of the running feedback parameters. Execution deviations include control response lag, variable offset, and under-regulation. The strategy fine-tuning module is used to dynamically adjust the control strategy in the current execution process based on the operation feedback parameters when an execution deviation is detected. The control strategy includes correcting the heating rate setpoint, updating the feed flow target value, and adjusting the start-stop rhythm of the reflux pump.

[0085] In one embodiment of the present invention, the intelligent start-up system for a gas separation device further includes: The qualification judgment module is used to obtain the output material composition data, the stability index of tower top pressure and temperature, and the liquid level change trend of the reflux tank to determine whether the output material meets the output material quality requirements and the stability standard of the operating status. The instruction triggering module is used to determine that the produced material is qualified when the judgment result meets the preset qualification standard, and to trigger the generation and control process of the start-up instruction of the next tower.

[0086] In one embodiment of the present invention, the system further includes a construction commencement expert knowledge base module, the construction commencement expert knowledge base module comprising: A rule template library stores control strategy templates for different operating conditions. Historical case database stores operational data, intervention and adjustment records, and output quality information from the historical construction process; A parameter optimization library that stores the optimal parameter combinations obtained from historical data mining. The inference engine matches the most suitable rule template based on the current running state and generates control instructions.

[0087] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it includes the steps of the methods described in the embodiments. The storage medium includes, but is not limited to, ROM, RAM, magnetic disks, optical disks, etc.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for intelligent start-up of a gas separation unit, characterized in that, Includes the following steps: Obtain production requirements and generate preset control data based on production requirements. The preset control data includes preset process control parameters and preset monitoring thresholds. The current operating parameters of the gas separation unit are obtained, and anomaly detection and analysis are performed based on the current operating parameters and preset monitoring thresholds. If potential risks are identified, intervention suggestions are generated and process control parameters are dynamically corrected. The gas separation unit includes a propane tower, an ethane tower and a propylene tower. Based on the revised process control parameters and the preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, a propane tower start-up command is generated, and the propane tower is controlled to produce materials according to the propane tower start-up command. Once the output material from the propane tower is detected to be qualified, an ethane tower start-up command is generated, and the ethane tower is controlled to produce material according to the ethane tower start-up command. Once the output material from the ethane tower is detected to be up to standard, a start-up command for the propylene tower is generated, and the propylene tower is controlled to produce material according to the start-up command.

2. The intelligent start-up method for a gas separation device according to claim 1, characterized in that, The preset process control parameters include feed rate, feed target value, heating rate, heating target value, reflux flow target value, reflux tank level threshold, bottom level threshold, and top pressure threshold.

3. The intelligent start-up method for a gas separation device according to claim 1, characterized in that, The preset monitoring thresholds include the alarm threshold for high or low liquid level at the bottom of the tower, the range of abnormal changes in tower pressure, the limit for the slope of temperature change, the threshold for judging rapid drop in liquid level in the reflux tank, and the range for judging fluctuations in feed flow rate.

4. The intelligent start-up method for a gas separation device according to claim 1, characterized in that, The process of acquiring the current operating parameters of the gas separation unit, performing anomaly detection and analysis based on the current operating parameters and the preset monitoring threshold, and generating intervention suggestions and dynamically correcting the process control parameters if potential risks are identified includes the following steps: Real-time operating data of the propane tower, ethane tower and propylene tower are collected. The real-time operating data includes the bottom liquid level, top pressure, temperature rise, reflux tank liquid level and feed flow rate. The real-time operating data is compared with the corresponding monitoring threshold to determine whether there is an abnormal state. The abnormal state includes liquid level exceeding the limit, pressure change, temperature slope exceeding the limit, or feed fluctuation exceeding the standard. When an abnormal state is detected, corresponding intervention suggestions are generated. These intervention suggestions include adjusting the heating rate, reducing the feed rate, delaying the start-stop timing of the reflux pump, or raising the liquid level protection line. Based on the intervention recommendations, the relevant process control parameters are dynamically adjusted to ensure that the control state within the tower returns to a safe range and to guarantee the stable execution of subsequent production operations.

5. The intelligent start-up method for a gas separation device according to claim 1, characterized in that, The step of generating a propane tower start-up command based on the corrected process control parameters and the preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, includes the following steps: Based on the corrected process control parameters, the target operating state of the propane tower is determined, and the rule template that best matches the target operating state in the start-up expert knowledge base is called. The rule template includes the propane tower's feed start-up and shutdown sequence, heating start-up and shutdown conditions, reflux pump activation conditions, and the operation logic of each control valve. Based on the operation logic in the rule template, and combined with the modified feed rate, heating rate and reflux setting value, the propane tower start-up command is generated.

6. The intelligent start-up method for a gas separation unit according to claim 1, characterized in that, The preset monitoring thresholds can be dynamically adjusted during the operation of the gas separation unit. Dynamic adjustments include: Based on the current operating parameters and the trends of monitored variables, and combined with the operating status of adjacent devices, the monitoring thresholds are adjusted. The operating status of adjacent devices includes device start-up and shutdown, load changes, or operation phase switching.

7. The intelligent start-up method for a gas separation unit according to claim 1, characterized in that, The criteria for determining whether the output materials are qualified include: The obtained output material composition data, tower top pressure and temperature stability indicators, and reflux tank liquid level change trend are used to determine whether the output material meets the output material quality requirements and operating stability standards. When the judgment result meets the preset qualification standard, the output material is determined to be qualified, and the start-up instruction generation and control process of the next tower is triggered.

8. A method for intelligent start-up of a gas separation unit according to any one of claims 1-7, characterized in that, It also includes the following steps: The existing templates are evaluated based on historical operational data, intervention and adjustment records, and output quality information within a preset period. When the evaluation results meet the optimization threshold conditions, the optimal execution path, parameter combination, and response instructions are extracted and written into the construction expert knowledge base as new rule templates to improve subsequent matching efficiency and the adaptability of control strategies.

9. A method for intelligent start-up of a gas separation unit according to any one of claims 1-7, characterized in that, It also includes the following steps: During the execution of the start-up command for the corresponding tower, the operating feedback parameters of the corresponding tower are collected in real time. The operating feedback parameters include the current feed flow rate, bottom liquid level, top pressure, heating temperature and reflux tank liquid level. The presence of execution deviations is determined based on the trend changes of operational feedback parameters. Execution deviations include control response lag, variable offset, and insufficient regulation. When an execution deviation is detected, the control strategy in the current execution process is dynamically adjusted based on the operation feedback parameters. The control strategy includes correcting the heating rate setpoint, updating the feed flow target value, and adjusting the start-stop rhythm of the reflux pump.

10. An intelligent start-up system for a gas separation unit, characterized in that, include: The control data generation module is used to acquire production requirements and generate preset control data based on these requirements. The preset control data includes preset process control parameters and preset monitoring thresholds. An anomaly detection and parameter correction module is used to acquire the current operating parameters of the gas separation unit and perform anomaly detection analysis based on the current operating parameters and preset monitoring thresholds. If potential risks are identified, intervention suggestions are generated and process control parameters are dynamically corrected. The gas separation unit includes a propane tower, an ethane tower, and a propylene tower. The propane tower control module is used to generate a propane tower start-up command based on the corrected process control parameters and preset process control parameters, combined with the corresponding rules in the start-up expert knowledge base, and control the propane tower to produce materials according to the propane tower start-up command. The ethane tower control module is used to generate an ethane tower start-up command when the output material of the propane tower is detected to be qualified, and to control the ethane tower to produce material according to the ethane tower start-up command; The propylene tower control module is used to generate a propylene tower start-up command when the output material from the ethane tower is found to be qualified, and to control the propylene tower to produce material according to the propylene tower start-up command.

Images