Method and system for automatically adjusting parameters of rectifying tower based on PLC (Programmable Logic Controller)

Abstract

The invention discloses a PLC-based rectifying tower parameter automatic adjustment method and system, and belongs to the technical field of rectifying tower control. The method comprises the steps of performing clustering processing on historical regulation and control parameters to construct a classification area, constructing regulation and control samples marked as primary samples and secondary samples based on the classification area, and a sample intelligent model, determining target regulation and control parameters based on the primary samples, and transmitting the target regulation and control parameters to a PLC control system to execute regulation and control actions, and acquiring real-time operation parameters of the rectifying tower, inputting the real-time operation parameters into the sample intelligent model to output a simulation regulation and control parameter set, further judging regulation and control feedback, and triggering updating of the sample intelligent model when the regulation and control feedback is in a deviation state. According to the method, through double screening of historical regulation and control parameters and simulation pre-judgment before regulation and control, the accuracy of the regulation and control parameters is improved, and the stability, safety and robustness of the system are enhanced through closed-loop self-learning and fault-tolerant mechanisms.

Description

Technical Field

[0001] This invention belongs to the field of distillation column control technology, specifically relating to a PLC-based method and system for automatic adjustment of distillation column parameters. Background Technology

[0002] As a core separation device in modern chemical, petroleum refining, and pharmaceutical production processes, the distillation column's operational efficiency directly determines the purity and yield of the final product, as well as the energy consumption and economic benefits of the entire production process. The essence of the distillation process is to achieve precise separation of multi-component mixtures through multiple vapor-liquid phase equilibrations within the column. Therefore, ensuring that the distillation column maintains a long-term, stable, and efficient operating state under various operating conditions, and accurately and timely monitoring and controlling its key operating parameters, are crucial technical aspects for guaranteeing product quality, reducing production costs, and enhancing the core competitiveness of enterprises.

[0003] Existing technologies still have many shortcomings in achieving stable and optimized control of distillation columns. Traditional automatic control systems, whether based on simple feedback loops or relying on preset process models, lack foresight and predictive capabilities. Most of these systems adopt "post-compensation" control logic, that is, they only begin to adjust after detecting deviations in process parameters from the setpoint. This lag response is difficult to effectively cope with disturbances such as sudden changes in feed composition or fluctuations in upstream operating conditions, resulting in significant time delays in the control process and affecting the timeliness of control. The adaptive capabilities of existing technologies are generally insufficient. The distillation process is a complex, nonlinear, multivariate, and strongly coupled time-varying system. Traditional control models or parameters, once set, are often difficult to self-tune online according to the dynamic changes in actual operating conditions. When actual production conditions deviate from the design conditions, their control performance will significantly decrease, and may even cause system oscillations, destroying operational stability. Although process simulation-based optimization techniques can theoretically find the best, their effectiveness depends on the accuracy of the mechanistic model. Building and maintaining a high-fidelity model that can accurately reflect the dynamic characteristics of the actual distillation process is costly and there is model mismatch. This makes it difficult for model-based control strategies to achieve the expected results in practical applications, affecting the overall control efficiency and robustness of the distillation column.

[0004] In view of this, the present invention proposes a PLC-based method and system for automatic adjustment of distillation column parameters. Summary of the Invention

[0005] The purpose of this invention is to provide a PLC-based method for automatically adjusting the parameters of a distillation column, in order to solve the problem that in the prior art, the relevant parameters in the distillation column need to be adjusted in real time according to changes in operating conditions during operation. By reasonably controlling these parameters, the distillation process can be further optimized, and the purity and yield of the product can be improved.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] This invention provides a PLC-based method for automatically adjusting distillation column parameters, comprising:

[0008] The target control parameter is determined based on a first-level sample constructed by clustering historical control parameters.

[0009] The target control parameters are transmitted to the PLC control system of the distillation column to execute the control action;

[0010] The system acquires the real-time operating parameters of the distillation column and, based on the sample intelligent model corresponding to the real-time operating parameters, determines whether the control feedback is in an execution state or a deviation state.

[0011] When the control feedback is determined to be in a deviated state, an update to the sample intelligent model is triggered;

[0012] The steps to determine the target control parameters include:

[0013] The historical control parameters related to the operation of the distillation column were clustered to obtain multiple classification regions that include historical control parameters and corresponding control states;

[0014] A control sample is constructed based on each classification region, and the control sample is labeled as a primary sample and a secondary sample.

[0015] The average control value is calculated based on the primary sample, and the average control value is determined as the target control parameter.

[0016] In a preferred technical solution, the step of labeling the control samples as primary samples and secondary samples includes:

[0017] Arrange the control states within the classification area in chronological order, and calculate the difference between any two temporally adjacent control states to obtain the control amplitude.

[0018] Calculate the average value of all control amplitudes within the classification region to obtain the average control amplitude;

[0019] Based on the comparison results of each control amplitude with the average control amplitude, the control samples are labeled as primary samples or secondary samples.

[0020] In a preferred technical solution, the step of clustering historical control parameters includes:

[0021] The first historical control parameter in the time-ordered historical control parameters is set as the initial reference value, and the subsequent historical control parameters are set as comparison values;

[0022] Calculate the absolute value of the difference between each comparison value and the current reference value in turn to obtain the corresponding value difference;

[0023] When the difference between corresponding values ​​is greater than or equal to the warning threshold, the comparison value corresponding to that difference will be determined as the new reference value.

[0024] Based on the initial reference value and all new reference values, historical control parameters are divided into multiple classification regions.

[0025] In a preferred technical solution, the step of determining whether the control feedback is in an execution state or a deviation state includes:

[0026] The real-time operating parameters are input into the sample intelligent model, and the simulated control parameter set is output.

[0027] Based on whether the parameter values ​​in the simulated control parameter set exceed the preset rated range and whether the rate of change of the parameter values ​​is greater than the preset rated rate, the control feedback is determined to be in execution state or deviation state.

[0028] In a preferred technical solution, the steps to trigger an update to the sample intelligent model include:

[0029] When the control feedback is determined to be in a deviated state, the type of the deviated state is further determined as either a state that can continue to operate or a state that cannot operate.

[0030] When the deviation state is a continuing operation state, the real-time operation parameters of this state will be used to update the secondary samples corresponding to the classification region to which the real-time operation parameters belong, and the sample intelligent model will be updated based on the updated secondary samples.

[0031] When the deviation state is an inoperable state, the historical control parameters associated with the current inoperable state are retrieved from the historical control parameters and input into the sample intelligent model to regenerate the simulated control parameter set.

[0032] In a preferred technical solution, the historical control parameters include heating power, feed rate, and catalyst dosage.

[0033] Furthermore, the control states corresponding to the historical control parameters include reflux ratio, bottom residue amount, and condensation temperature.

[0034] This invention also provides a PLC-based automatic adjustment system for distillation column parameters, comprising:

[0035] The target parameter generation module is used to determine the target control parameter based on a first-level sample constructed by clustering historical control parameters.

[0036] The PLC control module is used to transmit the target control parameters determined by the target parameter generation module to the PLC control system of the distillation column to execute the control action.

[0037] The operation status determination module is used to acquire the real-time operating parameters of the distillation column and, based on the sample intelligent model corresponding to the real-time operating parameters, determine whether the control feedback is in an execution state or a deviation state.

[0038] The model adaptive update module is configured to trigger an update of the sample intelligent model in response to the running status determination module determining that the control feedback is in a deviated state.

[0039] In a preferred technical solution, determining the target control parameter specifically includes:

[0040] Historical control parameters are clustered to obtain multiple classification regions;

[0041] Based on the classification regions, control samples are constructed and labeled as primary and secondary samples;

[0042] The average control value is calculated based on the primary sample to determine the target control parameter.

[0043] In a preferred technical solution, determining whether the control feedback is in an execution state or a deviation state specifically includes:

[0044] Real-time operating parameters are input into the sample intelligent model to output a set of simulated control parameters;

[0045] Based on whether the parameter values ​​in the simulated control parameter set exceed the preset rated range and whether the rate of change of the parameter values ​​is greater than the preset rated rate, the control feedback is determined to be in an execution state or a deviation state.

[0046] In a preferred technical solution, triggering the update of the sample intelligent model specifically includes:

[0047] The type of deviation state is determined as either a continue-running state or a non-running state;

[0048] When the system is determined to be in a state where it can continue to operate, the corresponding secondary samples and sample intelligent models are updated based on real-time operating parameters.

[0049] When a non-operable state is determined, the historical control parameters associated with the current non-operable state are retrieved from the historical control parameters and input into the sample intelligent model to regenerate the simulated control parameter set.

[0050] Beneficial effects

[0051] This invention provides a PLC-based automatic adjustment method for distillation column parameters. It clusters historical control parameters to construct classification regions to address different operating conditions. Within each classification region, the control amplitude of the control state is calculated, dividing the control samples into primary and secondary samples. The target control parameter is then determined based on the primary sample. This dual screening of historical control parameters effectively filters out interference from transitional phases or abnormal fluctuations in the intelligent sample model, ensuring that the generated target control parameter originates from a stable operating range under specific operating conditions. This mechanism makes the automatic adjustment of distillation column parameters more targeted and precise, improving the rationality and control accuracy of the target control parameter.

[0052] Before executing control actions, this invention inputs the acquired real-time operating parameters into a sample intelligent model to output a simulated control parameter set. Based on whether the values ​​of the simulated control parameter set exceed the preset rated range and whether the rate of change is greater than the preset rated rate, it pre-determines whether the control feedback is in an execution state or a deviation state. Only when the state is determined to be execution state is the target control parameter transmitted to the PLC control system. Through the simulation prediction mechanism before control, potential risks that may lead to system instability can be identified in advance, avoiding the direct application of inappropriate control commands to the distillation column. Compared with traditional passive response control, this invention achieves forward-looking assessment and intervention of the operating state, thereby effectively avoiding the risk of equipment overload operation and enhancing the stability and safety of the distillation column control process.

[0053] When the control feedback is determined to be in a deviation state, this invention can distinguish the type of deviation state: for a state where operation can continue, the secondary samples and sample intelligent models are updated using the corresponding real-time operating parameters; for a state where operation is not possible, the relevant historical parameters are searched and called from the historical control parameters to respond. Based on the differentiated deviation state handling strategy, this invention constructs a closed-loop self-learning and fault-tolerant mechanism. On the one hand, the sample intelligent model can continuously learn from new non-critical deviation data, achieve self-iteration and optimization, and improve its adaptability and prediction accuracy; on the other hand, its fault-safe mechanism ensures that in the event of a serious anomaly, the system can fall back to a safe operating condition, ensuring the continuity of production and the robustness of the system. Attached Figure Description

[0054] Figure 1 The method flow provided by this invention Figure 1 ;

[0055] Figure 2 The method flow provided by this invention Figure 2 ;

[0056] Figure 3 This is a system module diagram provided by the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely for explaining the invention and are not intended to limit the scope of protection of the invention.

[0058] Example 1

[0059] Please see Figures 1-2 As shown, this embodiment provides a PLC-based automatic adjustment method for distillation column parameters, which is suitable for intelligent adjustment and control of multiple parameters in the distillation separation process, aiming to ensure product quality, energy efficiency and equipment safety.

[0060] This method includes the following steps:

[0061] Historical control parameters related to the long-term operation of the distillation column are obtained from the distributed control system (DCS) or the supervisory control system (SCADA). These historical control parameters may include heating power, feed rate, and catalyst dosage. The corresponding control status at the same time stamp of the historical control parameters is obtained. The control status may include, for example, reflux ratio, bottom residue amount, and condensation temperature.

[0062] Clustering of historical control parameters yields multiple independent classification regions. This clustering step aims to identify fundamental changes in operating patterns caused by factors such as raw material fluctuations, product scheme switching, or equipment maintenance. The specific steps are as follows: The first historical control parameter in the time-ordered historical control parameters is set as the initial reference value; all historical control parameters after the initial reference value are set as comparison values; the absolute value of the difference between each comparison value and the current reference value is calculated sequentially to obtain the corresponding value difference. Here, the current reference value is the initial reference value used in the first calculation; the corresponding value difference is compared with a preset warning threshold, which is set based on the working principle... Based on experience, if the difference between corresponding values ​​is greater than or equal to the warning threshold, it indicates that the operating status of the distillation column has changed significantly. At this time, the comparison value corresponding to the difference between corresponding values ​​is determined as the new reference value, and the new reference value is used as the current reference value for the next comparison. Repeat the above calculation, comparison and determination steps until all comparison values ​​are processed and one or more new reference values ​​are obtained. Based on the initial reference value and all new reference values, the historical control parameters are divided to form multiple independent classification regions. Specifically, the initial reference value and the new reference value are defined as the starting nodes of the classification regions, and the data set between two adjacent starting nodes constitutes an independent classification region.

[0063] Based on the historical control parameters and corresponding control states contained within the system, control samples are constructed, and a corresponding sample intelligent model is generated for each control sample. The purpose of this step is to transform historical operational experience into a sample intelligent model that can be used for real-time decision-making. The sample intelligent model refers to a computational model used to predict the evolution trend of state parameters of the distillation column within a short time step in the future, based on the corrected real-time operating parameters. Its specific formula is as follows:

[0064]

[0065] In the formula, This represents the set of simulation control parameters, which means the vector of predicted values ​​of key state parameters (such as temperature, pressure, and component concentration) of the distillation column by the model for one or more future time steps. This represents the activation function, which is a function that performs a nonlinear mapping on the result of a linear transformation, such as the Sigmoid function or the ReLU function, and is used to enhance the nonlinear fitting ability of the model. This represents the weight matrix, which means that the matrix has a size of 1. The matrix, whose elements are the parameters learned by the model through training, reflects the influence weights of different input parameters on different output parameters; This represents the replacement parameter vector, which is the preprocessed and corrected real-time running parameter vector used as the input features of the model. This represents the bias vector, which means that the size is... The vector provides additional adjustable parameters for the model, used to adjust the activation threshold of the activation function.

[0066] To differentiate between different types of operational experience, control samples with control amplitudes greater than or less than the average control amplitude need to be labeled as primary and secondary samples, respectively. This labeling process is achieved by quantifying the volatility of control states. The specific steps are as follows: Arrange the control states within the classification region in chronological order, calculate the difference between any two temporally adjacent control states to obtain a series of control amplitudes. These amplitudes reflect the intensity of control actions within a specific operational phase. Calculate the average of all control amplitudes within the classification region to obtain the average control amplitude, which represents the average operational intensity within that region. Based on the comparison between each control amplitude and the average control amplitude, samples are labeled. For example, if a control sample has a control amplitude greater than the average control amplitude, it can be labeled as a primary sample, corresponding to critical control actions. Conversely, if the control amplitude is less than the average control amplitude, it can be labeled as a secondary sample, corresponding to routine stable maintenance operations. This distinction allows the system to focus on historical operations that have the greatest impact on production.

[0067] After the sample labeling is completed, the average control value is calculated based on the historical control parameters contained in the primary sample. The average control value that is sent to the PLC control system to execute specific control actions is determined as the target control parameter. Since the primary sample corresponds to the key control operations in history, the average control value calculated based on the primary sample is determined as the target control parameter under the current operating condition. The target control parameter is then passed to the control execution queue, waiting to be transmitted.

[0068] Real-time operating parameters of the distillation column are acquired via PLC. To enhance the robustness of the sample intelligent model and handle potential sensor anomalies, the following preprocessing steps are performed: The difference between the real-time operating parameters and the preset benchmark value is calculated to obtain the deviation. The preset benchmark value can be a theoretically calculated value or an empirical value from long-term stable operation. Based on the deviation, replacement parameters are extracted from a preset parameter library. This library stores verified and safe parameter values ​​within different deviation ranges. The purpose of this step is to correct and smooth the real-time data, avoiding overreaction of the entire control system due to instantaneous jumps or drifts of a single sensor. The replacement parameters are input into the sample intelligent model that matches the current operating conditions to output a set of simulated control parameters. The sample intelligent model can employ a neural network regression model, which can output a set of simulated control parameters based on the input replacement parameters. This set of simulated control parameters can be used to predict the evolution trend of the distillation column's state parameters within a short time step in the future.

[0069] After obtaining the simulated control parameter set, the control feedback needs to be determined based on this parameter set to decide the next operation. The control feedback includes an execution state and a deviation state. This determination process includes a dual verification mechanism: the first verification is amplitude verification, which calculates the amount by which the parameter value in the simulated control parameter set exceeds the upper limit of its corresponding preset rated range. The preset rated range is the operating range defined by the process safety regulations. If the amount is greater than the preset offset threshold, the control feedback is directly determined to be a deviation state. If the amount is not greater than the preset offset threshold, the second verification, namely rate verification, is entered, which further calculates the rate of change of the parameter value. If the rate of change is greater than the preset rated rate, the control feedback is determined to be a deviation state. If the amount is not greater than the preset offset threshold and the rate of change is not greater than the preset rated rate, the control feedback is determined to be an execution state.

[0070] When the control feedback is determined to be in a deviation state, it is necessary to further determine the type of deviation state in order to take different levels of countermeasures. The types of deviation states include a state that can continue to operate and a state that cannot operate. The basis for determination is: whether the parameter value in the simulated control parameter set exceeds its corresponding preset rated range upper limit. If the parameter value exceeds the preset rated range upper limit, the deviation state is determined to be a state that cannot operate; if the parameter value does not exceed the preset rated range upper limit, the deviation state is determined to be a state that can continue to operate.

[0071] Based on the determined control feedback, the corresponding control action is executed. When the control feedback is in the execution state, it indicates that the system predicts that the current operating condition is stable. The determined target control parameters are transmitted to the PLC control system of the distillation column, and the PLC directly controls the actuator to execute the control action, so that the distillation column moves closer to the optimal operating point.

[0072] When the deviation state is determined to be a state that can continue to operate, the real-time operating parameters of that state will be used to update the secondary samples corresponding to the classification region to which the real-time operating parameters belong. Based on the updated secondary samples, the sample intelligent model will be updated. The significance of this is that the system can learn from controllable deviations and continuously fine-tune its sample intelligent model to enhance its adaptability to the current working conditions.

[0073] When the deviation state is determined to be an inoperable state, the historical control parameters associated with the current inoperable state are searched from the historical control parameters. The searched historical control parameters are then input into the sample intelligent model to regenerate the simulated control parameter set. This is equivalent to drawing on successful human intervention experience in the past to provide a verified solution for the current crisis, thereby guiding the system out of the inoperable state.

[0074] This method may also include a dynamic rolling sampling step: when the control parameter deviates systematically and exceeds the preset systematic deviation threshold, the difference between the simulated control parameter set in the feedback process and the actual execution result is reconstructed into a new control sample for real-time updating of the sample intelligent model.

[0075] This method is highly scalable and can be integrated into industrial automation DCS systems. It can interface with existing equipment through the OPC UA interface or MODBUS protocol to enable the algorithm optimization capabilities to be applied to actual production.

[0076] In summary, this embodiment can minimize manual intervention in distillation parameter adjustments, reduce the accident rate, and improve operational stability and energy efficiency. It is suitable for application scenarios with complex process control requirements in modern chemical, pharmaceutical, and petroleum refining industries.

[0077] Example 2

[0078] Please see Figure 3As shown, this embodiment provides a PLC-based automatic parameter adjustment system for a distillation column. This system autonomously generates optimized control parameters through in-depth analysis of historical operating data. During real-time monitoring, it adaptively updates the control model based on actual operating feedback from the distillation column, thereby achieving closed-loop, intelligent control of the distillation column's operating status. Physically, this system can be deployed on industrial control computers, servers, or embedded systems, and interacts with the PLC control system and various sensors at the distillation column site via industrial Ethernet, fieldbus, and other communication methods. Logically, the system can be divided into the following collaborative modules:

[0079] Target parameter generation module

[0080] Based on historical data, target control parameters are generated to guide the operation of the distillation column. The stored historical control parameters related to distillation column operation are clustered. In a specific execution flow, a reference-value-based iterative clustering method is used. The first parameter in the temporally ordered historical control parameter sequence is set as the initial reference value. Subsequent historical control parameters are used as comparison values, and the absolute value of the difference between each comparison value and the current reference value is calculated to obtain the corresponding value difference. When a corresponding value difference is greater than or equal to a preset warning threshold, it indicates a significant change in the operating condition of the distillation column. At this point, the module determines the comparison value corresponding to this corresponding value difference as the new reference value. By continuously iterating this process, using the initial reference value and all established new reference values ​​as base points, the entire historical control parameter sequence is divided into multiple classification regions. Each region represents a relatively stable operating condition. Historical control parameters may specifically include heating power, feed rate, and catalyst dosage. After classifying the regions, control samples are constructed based on the data within each region. Each sample contains historical control parameters and their corresponding control states at that point in time, such as reflux ratio, bottom residue amount, and condensation temperature. To differentiate data stability, control samples are further labeled as primary and secondary samples. The control states within a certain region are arranged chronologically, and the difference between any two temporally adjacent control states is calculated to obtain a series of control amplitudes. The average control amplitude is obtained by calculating the average of all control amplitudes within the region. By comparing each control amplitude with the average control amplitude, if the control amplitude corresponding to a control sample is significantly less than the average, it indicates that it is in a stable operating phase and is labeled as a primary sample; otherwise, it is labeled as a secondary sample. Based on the control samples labeled as primary samples, the average of the control parameters in these stable samples is calculated to obtain the average control value, which is then determined as the target control parameter for the current control cycle for use by subsequent modules.

[0081] PLC control module

[0082] The decision results are transformed into actual physical control. The system receives the target control parameters determined by the target parameter generation module. After receiving the parameters, it encapsulates them into instructions that conform to the target equipment communication protocol and transmits them to the PLC control system of the distillation column through the data link. After parsing the instructions, the PLC control system directly drives the corresponding actuators, such as heaters, valves, and pumps, to perform specific control actions, thereby applying the optimized parameters to the actual production process.

[0083] Running status determination module

[0084] The operating status of the distillation column after the execution of control actions is monitored and evaluated in real time. Real-time operating parameters of the distillation column are acquired through sensor interfaces. These real-time operating parameters are input into the sample intelligent model corresponding to the current operating condition, i.e., the classification region. This sample intelligent model is built based on historical first-level and second-level samples and can simulate the theoretically ideal control state based on real-time input. After the model runs, it outputs a set of simulated control parameters. Then, the parameter set is analyzed, and the current control feedback is determined according to preset rules. It checks whether the values ​​of each parameter in the simulated control parameter set exceed a preset rated range and whether the rate of change of the parameter values ​​is greater than a preset rated rate. If all parameters are within the normal range, the control feedback is determined to be in an execution state, indicating stable operation. If any parameter exceeds the limit, it is determined to be in a deviation state, indicating that the actual operation does not match the model expectation and intervention is required.

[0085] Model Adaptive Update Module

[0086] The module is activated when the operational status determination module identifies the control feedback as a deviation state. Its core task is to enable the model to learn and correct itself to adapt to changes in operating conditions. Upon receiving a deviation state trigger signal, it determines the type of deviation state, classifying it as either a continue-operating state or an inoperable state. This determination can be based on the degree of deviation and the criticality of the parameters involved. If it is determined to be a continue-operating state, meaning the deviation is still within a controllable range, the real-time operational parameters of this state will be used as new data points to update the secondary sample library corresponding to the classification region of these real-time operational parameters. Since the secondary samples represent the unstable or boundary states of the system, this update operation can enrich the model's ability to handle abnormal conditions. Based on the updated secondary samples, the sample intelligent model is updated or retrained online to improve the model's prediction accuracy when dealing with similar deviations in the future. If the system is determined to be inoperable, i.e., a serious fault or dangerous condition has occurred, a simple model update is no longer applicable. A safe rollback strategy will be implemented, which will search for historical control parameters associated with the current inoperable state from a complete historical control parameter database. For example, parameter configurations that have successfully handled similar faults in the past will be used. This set of verified historical parameters will be input into the sample intelligent model to regenerate a safe set of simulated control parameters to guide the system to return to a stable state or provide operators with emergency response guidelines.

[0087] Through the collaborative work of the aforementioned target parameter generation module, PLC control module, operating status determination module, and model adaptive update module, this embodiment constructs a complete closed-loop control system from data analysis, decision generation, instruction execution to status feedback and model self-learning. This system can effectively improve the automation level and operational stability of the distillation process, and is particularly suitable for chemical production scenarios with complex and variable operating conditions and high product quality requirements.

Claims

1. A PLC-based method for automatically adjusting distillation column parameters, characterized in that, include: The target control parameter is determined based on a first-level sample constructed by clustering historical control parameters. The target control parameters are transmitted to the PLC control system of the distillation column to execute the control action; The system acquires the real-time operating parameters of the distillation column and, based on the sample intelligent model corresponding to the real-time operating parameters, determines whether the control feedback is in an execution state or a deviation state. When the control feedback is determined to be in a deviated state, an update to the sample intelligent model is triggered; The steps to determine the target control parameters include: The historical control parameters related to the operation of the distillation column were clustered to obtain multiple classification regions that include historical control parameters and corresponding control states; A control sample is constructed based on each classification region, and the control sample is labeled as a primary sample and a secondary sample. The average control value is calculated based on the primary sample, and the average control value is determined as the target control parameter.

2. The method for automatic adjustment of distillation column parameters based on PLC according to claim 1, characterized in that, The steps for labeling control samples as primary and secondary samples include: Arrange the control states within the classification area in chronological order, and calculate the difference between any two temporally adjacent control states to obtain the control amplitude. Calculate the average value of all control amplitudes within the classification region to obtain the average control amplitude; Based on the comparison results of each control amplitude with the average control amplitude, the control samples are labeled as primary samples or secondary samples.

3. The method for automatic adjustment of distillation column parameters based on PLC according to claim 2, characterized in that, The steps for clustering historical control parameters include: The first historical control parameter in the time-ordered historical control parameters is set as the initial reference value, and the subsequent historical control parameters are set as comparison values; Calculate the absolute value of the difference between each comparison value and the current reference value in turn to obtain the corresponding value difference; When the difference between corresponding values ​​is greater than or equal to the warning threshold, the comparison value corresponding to that difference will be determined as the new reference value. Based on the initial reference value and all new reference values, historical control parameters are divided into multiple classification regions.

4. The method for automatic adjustment of distillation column parameters based on PLC according to claim 1, characterized in that, The steps to determine whether the control feedback is in an executing state or a deviating state include: The real-time operating parameters are input into the sample intelligent model, and the simulated control parameter set is output. Based on whether the parameter values ​​in the simulated control parameter set exceed the preset rated range and whether the rate of change of the parameter values ​​is greater than the preset rated rate, the control feedback is determined to be in execution state or deviation state.

5. The method for automatic adjustment of distillation column parameters based on PLC according to claim 1, characterized in that, The steps to trigger an update to the sample intelligent model include: When the control feedback is determined to be in a deviated state, the type of the deviated state is further determined as either a state that can continue to operate or a state that cannot operate. When the deviation state is a continuing operation state, the real-time operation parameters of this state will be used to update the secondary samples corresponding to the classification region to which the real-time operation parameters belong, and the sample intelligent model will be updated based on the updated secondary samples. When the deviation state is an inoperable state, the historical control parameters associated with the current inoperable state are retrieved from the historical control parameters and input into the sample intelligent model to regenerate the simulated control parameter set.

6. The method for automatic adjustment of distillation column parameters based on PLC according to claim 1, characterized in that: Historical control parameters include heating power, feed rate, and catalyst dosage; Furthermore, the control states corresponding to the historical control parameters include reflux ratio, bottom residue amount, and condensation temperature.

7. A PLC-based automatic adjustment system for distillation column parameters, characterized in that, include: The target parameter generation module is used to determine the target control parameter based on a first-level sample constructed by clustering historical control parameters. The PLC control module is used to transmit the target control parameters determined by the target parameter generation module to the PLC control system of the distillation column to execute the control action. The operation status determination module is used to acquire the real-time operating parameters of the distillation column and, based on the sample intelligent model corresponding to the real-time operating parameters, determine whether the control feedback is in an execution state or a deviation state. The model adaptive update module is configured to trigger an update of the sample intelligent model in response to the running status determination module determining that the control feedback is in a deviated state.

8. The PLC-based automatic adjustment system for distillation column parameters according to claim 7, characterized in that, Determining the target control parameters specifically includes: Historical control parameters are clustered to obtain multiple classification regions; Based on the classification regions, control samples are constructed and labeled as primary and secondary samples; The average control value is calculated based on the primary sample to determine the target control parameter.

9. The PLC-based automatic adjustment system for distillation column parameters according to claim 7, characterized in that, Determining whether the control feedback is in an execution state or a deviation state specifically includes: Real-time operating parameters are input into the sample intelligent model to output a set of simulated control parameters; Based on whether the parameter values ​​in the simulated control parameter set exceed the preset rated range and whether the rate of change of the parameter values ​​is greater than the preset rated rate, the control feedback is determined to be in an execution state or a deviation state.

10. The PLC-based automatic adjustment system for distillation column parameters according to claim 7, characterized in that, The specific steps to trigger an update to the sample intelligent model include: The type of deviation state is determined as either a continue-running state or a non-running state; When the system is determined to be in a state where it can continue to operate, the corresponding secondary samples and sample intelligent models are updated based on real-time operating parameters. When a non-operable state is determined, the historical control parameters associated with the current non-operable state are retrieved from the historical control parameters and input into the sample intelligent model to regenerate the simulated control parameter set.

Images