Temperature control method of reaction kettle for purifying polypropylene

Abstract

The invention relates to the technical field of reaction kettle temperature control, in particular to a temperature control method of a reaction kettle for polypropylene purification. The method comprises the following steps: acquiring state data of a reaction kettle in a polypropylene purification process; obtaining a degassing initial moment based on the change of each purification temperature data; obtaining a reference historical purification of the current purification based on the similar condition of the state data; according to the change of the material viscosity data in a specified time period after the degassing initial moment of the reference historical purification and the material viscosity data at the current moment of the current purification, obtaining predicted material viscosity data, and in combination with the change lag condition of the material viscosity data and the pressure data in the reference historical purification, obtaining predicted pressure data; and based on the predicted material viscosity data and the predicted pressure data, obtaining predicted temperature data to control the current temperature of the purification reaction kettle. By accurately acquiring the predicted temperature data, the temperature of the reaction kettle can be controlled in advance, and the hysteresis and inaccuracy of temperature control are effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of reactor temperature control technology, and specifically to a method for controlling the temperature of a reactor used for polypropylene purification. Background Technology

[0002] Polypropylene, as an important thermoplastic polymer, is widely used in packaging, automotive, fiber, and chemical industries due to its excellent mechanical properties, chemical stability, and processability. The properties of polypropylene products, such as mechanical strength, transparency, and heat resistance, largely depend on their purity. Therefore, efficient purification of polypropylene in the post-processing stage of production to remove residual monomers, solvents, and other low-molecular-weight impurities is crucial.

[0003] Currently, reaction vessels are widely used in industry as the core equipment for polypropylene purification. The polypropylene purification process includes stages such as melt preheating, vacuum degassing, impurity evaporation, condensation separation, and cooling solidification. Among these, the vacuum degassing and impurity evaporation stages are the key steps determining the purification effect. The polypropylene material needs to be heated to 160-170℃ (close to its melting point range) and a negative pressure environment maintained inside the vessel. Through the synergistic effect of heating and depressurization, low-boiling-point impurities in the material are rapidly volatilized, achieving separation of impurities from the polypropylene matrix. Temperature and pressure are the core parameters controlling the polypropylene purification process, and the two are closely coupled. Temperature directly determines the melting state of the material and the impurity evaporation rate, while pressure affects the removal efficiency by changing the boiling point of impurities. Fluctuations in either parameter can lead to incomplete purification or material degradation, ultimately affecting product quality.

[0004] Currently, most existing reactor temperature control methods are based on traditional adjustment strategies, relying solely on temperature detection values ​​for control without fully considering the dynamic coupling relationship between temperature, pressure, and material viscosity. In actual production, the impurity content and viscosity of raw materials vary between different purification batches, and the impurity evaporation rate dynamically changes with the reaction process. These factors disrupt the thermal balance and pressure stability within the reactor, leading to significant adjustment lag and overshoot problems in traditional control systems. This makes it impossible to predict changes in operating conditions in advance, resulting in inaccurate temperature control and ultimately causing a decrease in polypropylene purification efficiency and poor batch-to-batch quality consistency. Summary of the Invention

[0005] To address the technical problem of lag and inaccuracy in reactor temperature control, which leads to unstable polypropylene purification, the present invention aims to provide a reactor temperature control method for polypropylene purification. The specific technical solution adopted is as follows: This invention provides a method for temperature control of a reaction vessel for polypropylene purification, the method comprising the following steps: Real-time acquisition of state data in the reactor during the polypropylene purification process; the state data includes temperature data, pressure data, and material viscosity data; Based on the changes in temperature data for each purification, the initial degassing time for each purification is obtained; based on the similarity of the state data at the initial degassing time of the current purification and the historical purification, as well as the similarity of the initial material weight, the reference historical purification for the current purification is obtained. Based on the changes in material viscosity data during a specified time period after the initial degassing moment of each historical purification, and the material viscosity data at the current moment of the current purification, the predicted material viscosity data at future moments of the current purification is obtained. Based on the predicted material viscosity data, and referring to the lag of changes in material viscosity and pressure data in historical purification, the predicted pressure data for the future time of the current purification is obtained. Based on predicted material viscosity data and predicted pressure data, a temperature prediction model is used to obtain predicted temperature data for the future time of the current purification process. Based on predicted temperature data, the temperature of the reactor in the current purification process is controlled.

[0006] Furthermore, the method for obtaining the initial time of degassing is as follows: For any purification at any moment, the difference between the standard deviations of the temperature data of the purification in the first preset time period and the second preset time period at that moment is obtained as the degassing reference value at that moment; wherein, the first preset time period and the second preset time period have the same duration, the end time of the first preset time period is that moment, and the start time of the second preset time period is that moment. Obtain the degassing reference value at each moment of the purification process, and take the moment corresponding to the largest degassing reference value as the initial degassing moment of the purification process.

[0007] Furthermore, the method for obtaining the reference history purification is as follows: For any historical purification and any state data, the result of normalizing the difference between the current purification and the state data at the initial degassing moment of the historical purification is taken as the first difference of the state data. The result of normalizing the difference between the initial material weight data of the current purification and the historical purification is taken as the initial weight difference. The sum of the first difference and initial weight difference of all state data is negatively correlated and normalized to determine the similarity between the current purification and the previous historical purification. When the similarity is greater than the preset similarity threshold, the corresponding historical purification is used as the reference historical purification for the current purification.

[0008] Furthermore, the method for obtaining the predicted material viscosity data is as follows: Based on the unit change of material viscosity data within a specified time period after the initial degassing moment of each historical purification, obtain the reference viscosity unit change value at the current moment of the current purification; where the current moment must be after the initial degassing moment of the current purification, and the time between the initial moment of the specified time period and the current moment is equal to the time between the corresponding initial degassing moment. The time period consisting of a specified duration starting from the current moment is taken as the first prediction time period for the current purification. All moments within the first prediction time period are considered as the first future moments; For any first future moment, the time interval between that first future moment and the current moment is taken as the first duration; The product of the reference viscosity unit change value and the first time duration is used as the viscosity reference change value; The sum of the material viscosity data at the current purification moment and the viscosity reference change value is used as the predicted material viscosity data at the first future moment.

[0009] Furthermore, the method for obtaining the reference viscosity unit change value is as follows: For any reference historical purification, the difference in material viscosity data at each moment within the specified time period of the reference historical purification and at the previous adjacent moment shall be regarded as the second difference. The average of the ratios of the duration between all second differences and their corresponding times is taken as the unit viscosity change rate for the specified time period of the reference historical purification. The average of the unit viscosity change rate over a specified time period for all previous historical purifications is used as the reference viscosity unit change value at the current moment of the current purification.

[0010] Furthermore, the method for obtaining the predicted pressure data is as follows: Based on the lag between changes in material viscosity data and pressure data in historical purification processes, the lag time of pressure data relative to material viscosity data is obtained. The moment that is located after the current moment and whose time interval is the change lag time is taken as the initial second future moment; Starting from the initial second future time, the time period consisting of a specified duration is taken as the second prediction time period for the current purification; all times within the second prediction time period are taken as second future times; wherein, one second future time corresponds to one first future time, and the interval between each second future time and its corresponding first future time is the change lag time. Each first future moment is taken as the first historical reference moment in each reference history purification process; wherein, the time between each first future moment and its corresponding first historical reference moment and the corresponding degassing initial moment is the same. Each second future moment is taken as the second historical reference moment in each reference history purification process; wherein, the time between each second future moment and its corresponding second historical reference moment and the corresponding degassing initial moment is the same. The material viscosity data and vacuum valve position setting value at each first historical reference time in the reference historical purification, as well as the initial material weight data of the reference historical purification, are used as independent variables; the pressure data at the second historical reference time corresponding to each first historical reference time in the reference historical purification are used as dependent variables; a set of stable reference regression coefficients are obtained by performing multiple linear regression analysis using the least squares method. ;in, This is the basic pressure compensation item, which represents the initial pressure data of the reactor when there is no material and no evacuation. The viscosity coefficient represents the weight of the influence of material viscosity data on the predicted pressure data. The valve position control coefficient represents the weight of the influence of the vacuum valve position on the predicted pressure data. The material weight coefficient represents the weight of the initial material weight data on the predicted pressure data. Then, the trained regression equation is determined: In the formula, For the first Predicted pressure data at the corresponding time point; The duration of the change lag; This represents the material viscosity data at time t. B represents the vacuum valve position setting at time t; B represents the initial material weight data. The predicted material viscosity data and vacuum valve position setting value at each first future time point, as well as the initial material weight data of the current purification, are input into the trained regression equation to obtain the predicted pressure data at the corresponding second future time point.

[0011] Furthermore, the method for obtaining the change lag time is as follows: For any historical purification, the viscosity data of the material from that historical purification is arranged in chronological order to obtain a viscosity data sequence. The pressure data from this historical purification process were arranged in chronological order to obtain a pressure data sequence. The cross-correlation function between the viscosity data sequence and the pressure data sequence is calculated, and the time shift that makes the cross-correlation function reach its maximum value is used as the reference lag time corresponding to this reference history purification. The average of the reference lag times corresponding to all historical purifications is used as the lag time of the pressure data relative to the material viscosity data.

[0012] Furthermore, the method for obtaining the predicted temperature data is as follows: The predicted material viscosity data and predicted pressure data for the predicted reference time period are sequentially input into the trained temperature prediction model, and the predicted temperature data for each future moment within the reference time period are output.

[0013] Furthermore, the method for controlling the temperature of the reactor during the current purification process is as follows: When the predicted temperature data indicates that the future temperature will exceed the preset temperature limit, the heating power of the reactor should be reduced in advance. When predicted temperature data indicates that the future temperature will be lower than the preset lower limit, the heating power of the reactor should be increased in advance.

[0014] Furthermore, the temperature prediction model is a multilayer feedforward neural network with ReLU activation function, trained by minimizing the mean square error between the predicted temperature and the actual temperature.

[0015] The present invention has the following beneficial effects: This invention first obtains the initial degassing time for each purification process based on the changes in temperature data, accurately determining the start time of the vacuum degassing and impurity evaporation stages. This effectively enhances the accuracy of comparisons between different subsequent purification processes and prepares for subsequent temperature prediction, improving its reliability. To further improve the accuracy of current purification temperature prediction, it obtains a reference historical purification process based on the similarity of state data and initial material weight between the current and historical purification processes at their initial degassing times. This provides a reliable reference for the current purification process, effectively improving the accuracy and reliability of all subsequent prediction steps. Furthermore, based on the changes in material viscosity data within a specified time period after the initial degassing time of each reference historical purification process, and the material viscosity data at the current moment of the current purification process, it obtains predicted material viscosity data for future moments of the current purification process. This accurately reflects the changes in the impurity content within the material, enabling advance prediction of material state changes caused by impurity removal, laying the foundation for subsequent control processes, and fundamentally addressing the issue. This approach changes the passive response model. By leveraging predicted material viscosity data and referencing historical changes in viscosity and pressure during purification, it obtains predicted pressure data for future moments in the current purification process. This effectively avoids the problem of drastic pressure data fluctuations due to direct control, making direct modeling and prediction difficult. This improves the stability and reliability of predicted pressure data, facilitating accurate subsequent temperature prediction. Furthermore, based on predicted material viscosity and pressure data, a temperature prediction model is used to obtain predicted temperature data for future moments in the current purification process. This allows for accurate prediction of future temperature changes, enabling proactive adjustments to heating power before actual temperature deviations occur. This effectively avoids inaccurate temperature control and response lag, significantly reducing temperature overshoot or undershoot, and greatly improving the accuracy of temperature control and process stability. Finally, based on the predicted temperature data, the temperature of the reactor during the current purification process is accurately controlled, ensuring stable operation of the polypropylene purification process under optimal process parameters. This significantly improves the purity, consistency, and energy efficiency of the final product. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic flowchart illustrating a method for controlling the temperature of a reaction vessel for polypropylene purification, provided in one embodiment of the present invention; Figure 2This is a structural diagram of a temperature control system for a polypropylene purification reactor provided in one embodiment of the present invention; Figure 3 This is a schematic diagram of a computer device provided according to an embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a reaction vessel temperature control method for polypropylene purification proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following describes in detail, with reference to the accompanying drawings, a specific scheme for a reaction vessel temperature control method for polypropylene purification provided by the present invention.

[0021] Example 1: This invention proposes a method for temperature control of a reaction vessel used in polypropylene purification. Please refer to [link / reference]. Figure 1 The diagram illustrates a schematic flow chart of a method for controlling the temperature of a reaction vessel for polypropylene purification according to an embodiment of the present invention. The method includes the following steps: Step S1: Real-time acquisition of state data in the polypropylene purification reactor; the state data includes temperature data, pressure data, and material viscosity data.

[0022] Specifically, the known polypropylene purification process includes stages such as melt preheating, vacuum degassing, impurity evaporation, condensation separation, and cooling solidification. Among these, the vacuum degassing and impurity evaporation stages are the core and key of the entire purification process. This embodiment mainly focuses on analyzing the vacuum degassing and impurity evaporation stages to ensure that the temperature and pressure inside the reactor are precisely and dynamically controlled in a coordinated manner, thereby effectively improving the purity of polypropylene.

[0023] The principle of the vacuum degassing and impurity evaporation stage is as follows: After initial melting and preheating, the polypropylene material is transported to the reactor and gradually heated to a temperature close to its melting point (typically between 160°C and 170°C). Simultaneously, a vacuum (negative pressure) is applied and maintained inside the reactor through a vacuum pump and pressure regulating valve system. Under these conditions, the boiling points of residual solvents, low-molecular-weight hydrocarbons, and unreacted monomers in the material decrease significantly due to the reduced system pressure, allowing them to be efficiently volatilized and removed at a relatively low temperature, thus purifying the polypropylene material. In the vacuum degassing and impurity evaporation stage, temperature control directly determines the melting uniformity of the material and the impurity evaporation rate, while pressure control dominates the boiling characteristics and removal efficiency of the impurities. These two aspects are closely coupled and jointly determine the efficiency of the purification process and key quality indicators such as the purity, mechanical properties, and transparency of the final polypropylene product.

[0024] To ensure stable process operation, the reactor is equipped with a complete set of supporting systems and monitoring devices. The heating system (heating jacket or electric heating mantle) provides a stable heat source, ensuring precise temperature control. The vacuum pump and pressure regulating valve system work together to maintain and dynamically adjust the negative pressure within the reactor, adapting to the impurity evaporation requirements. The mechanical stirrer operates continuously to ensure uniform mixing of the molten polypropylene, avoiding process fluctuations caused by localized overheating or bubble retention. Simultaneously, the reactor is equipped with multi-dimensional monitoring equipment to acquire real-time status data of the polypropylene purification process. This status data includes temperature, pressure, and material viscosity data. Specifically, a thermocouple temperature sensor monitors the temperature in real time, a pressure transmitter tracks the pressure, and a torque sensor on the stirring motor directly reflects the torque data as material viscosity. It should be noted that temperature, pressure, and viscosity data are collected synchronously. In this embodiment, the time interval between two consecutive data acquisitions is set to 1 minute. Implementers can set the time interval between two consecutive data acquisitions according to actual conditions; it is not limited here. In addition, the initial material weight data for each polypropylene purification process is acquired.

[0025] By collecting the aforementioned temperature, pressure, material viscosity, and initial material weight data in real time, accurate and comprehensive data support can be provided for the dynamic temperature control of the entire polypropylene purification process, ensuring the scientific nature and timeliness of the control strategy.

[0026] Step S2: Based on the changes in temperature data for each purification, obtain the initial degassing time for each purification; based on the similarity of the state data at the initial degassing time between the current purification and historical purification, as well as the similarity of the initial material weight, obtain the reference historical purification for the current purification.

[0027] Specifically, it is known that in the polypropylene purification process, the reactor is in a dynamic and unstable state in the initial stage. From the time the material enters the reactor until it is heated to the target temperature range, its temperature, pressure, and state change drastically. Furthermore, the initial conditions (such as ambient temperature and initial material temperature) may differ between different purification processes, leading to different times when a stable state is reached. Considering that the temperature data inside the reactor remains stable during the vacuum degassing and impurity evaporation stages of polypropylene purification, this embodiment obtains the initial degassing moment (i.e., the moment when the temperature begins to stabilize) for each purification process based on the changes in temperature data for each purification step. Essentially, this represents the start of the vacuum degassing and impurity evaporation stage.

[0028] To accurately predict the temperature data of the current purification process and ensure stable and efficient polypropylene purification, this embodiment obtains a reference historical purification based on the similarity of the initial degassing data and the initial material weight between the current purification and historical purification processes. This allows for the accurate selection of the most similar historical purification process, enabling personalized data generated under the closest conditions to be used for prediction. This effectively improves the relevance and accuracy of subsequent current purification predictions.

[0029] Preferably, in one feasible embodiment of this method, the method for obtaining the initial degassing time is as follows: For any moment in any purification process, the difference between the standard deviations of the temperature data of that purification process within a first preset time period and a second preset time period is obtained, and this difference is used as the degassing reference value for that moment; the larger the degassing reference value, the more likely that moment is the initial degassing time for that purification process. The first preset time period and the second preset time period have the same duration, the end time of the first preset time period is this moment, and the beginning time of the second preset time period is this moment; in this embodiment, the duration of both the first and second preset time periods is set to 10 minutes. The implementer can set the duration of the first and second preset time periods according to actual conditions, and this is not limited here. Then, the degassing reference value for each moment in that purification process is obtained, and the moment corresponding to the largest degassing reference value is used as the initial degassing time for that purification process.

[0030] This completes the initial degassing moment for each purification process.

[0031] Preferably, in one feasible embodiment of this invention, the method for obtaining historical purification data is as follows: For any historical purification and any state data, the absolute value of the difference between the initial degassing time of the current purification and the initial degassing time of the historical purification is normalized and used as the first difference of the state data; the absolute value of the difference between the initial material weight data of the current purification and the historical purification is normalized and used as the initial weight difference; in this embodiment, the above absolute values ​​of difference are normalized using a normalization function. The smaller the first difference and the initial weight difference are, the more similar the reactions of the historical purification and the current purification are. Furthermore, in this embodiment, the first difference and the initial weight difference of all state data are added together and negatively correlated and normalized, and the result is used as the similarity between the current purification and the historical purification; in this embodiment, the absolute value of the difference is normalized using a normalization function. The sum of the first difference and the initial weight difference is negatively correlated and normalized, where x represents the sum of the first difference and the initial weight difference for all state data; norm is the normalization function. The greater the similarity, the more similar the historical purification is to the current purification. Therefore, this embodiment sets a preset similarity threshold of 0.6. Implementers can set the size of the preset similarity threshold according to actual conditions, which is not limited here. When the similarity is greater than the preset similarity threshold, the corresponding historical purification is used as the reference historical purification for the current purification.

[0032] Thus, the reference historical purification data for the current purification process has been accurately selected. It should be noted that this embodiment sets the number of historical purification data to 600. Implementers can set the number of historical purification data according to their actual needs; this is not limited here.

[0033] Step S3: Based on the changes in material viscosity data during a specified time period after the initial degassing moment of each historical purification, and the material viscosity data at the current moment of the current purification, obtain the predicted material viscosity data for the future moments of the current purification.

[0034] Specifically, during the degassing process in the reactor, temperature and pressure data are directly controlled in real time. This control itself introduces frequent and short-cycle corrections, resulting in significant short-term fluctuations and strong interference traces in historical temperature and pressure data. Therefore, directly relying on these historical data, which are heavily influenced by external control, to establish a continuous prediction model for temperature or pressure data during the purification process yields low reliability and highly distorted prediction results. Unlike temperature and pressure data, changes in material viscosity data are primarily driven by changes in its internal state, i.e., impurity content. This is a relatively continuous and gradual physicochemical process determined by the inherent laws of the reaction itself. Therefore, material viscosity data has higher predictability and stability. Furthermore, inferences based on the changing trends of material viscosity data can more stably and accurately reflect the evolution of the material state inside the reactor.

[0035] It is known that the residual solvents and unreacted monomers contained in the material have shorter molecular chains and typically exhibit lower viscosity than the bulk polypropylene. As the degassing process proceeds, these low-molecular-weight impurities are continuously extracted and discharged, leading to a relative increase in the proportion of high-molecular-weight components in the material system. Macroscopically, this manifests as a gradual increase in the viscosity of the entire molten material as the degassing process progresses. It should be noted that the change in material viscosity is not uniform and linear. Due to differences in the content and composition of impurities at different degassing stages, the rate of change in material viscosity varies, thus affecting the heat transfer efficiency and gas escape behavior within the reactor, directly driving subsequent temperature and pressure fluctuations. Therefore, the change in material viscosity is essentially an indirect representation of the dynamic changes in the impurity content within the reactor.

[0036] In order to accurately predict the predicted material viscosity data at future moments of the current purification process, and to accurately predict the evolution of the current purified material state, this embodiment obtains the predicted material viscosity data at future moments of the current purification process based on the changes in material viscosity data within a specified time period after the initial degassing moment of each historical purification, as well as the material viscosity data at the current moment of the current purification process. This lays the foundation for subsequent feedforward control of temperature and pressure.

[0037] Preferably, in one feasible manner of this embodiment, the method for obtaining the predicted material viscosity data is as follows: based on the unit change of material viscosity data within a specified time period after the initial degassing moment of each reference historical purification, the reference viscosity unit change value at the current moment of the current purification is obtained, which accurately reflects the trend of the unit change of material viscosity data after the current moment of the current purification; wherein, the current moment must be located after the initial degassing moment of the current purification, and the current moment is the real-time state of the vacuum degassing and impurity evaporation stage of the current purification, and the time between the initial moment of all specified time periods and the corresponding initial degassing moment is equal to the time between the current moment and the corresponding initial degassing moment. Essentially, each reference historical purification is aligned with the initial degassing moment of the current purification, and then the moment with the same time between the current moment and the initial degassing moment is determined in each reference historical purification, which is the initial moment of the specified time period in each reference historical purification; The method for obtaining the reference viscosity unit change value is as follows: For any reference historical purification, the difference between the material viscosity data at each moment within the specified time period of the reference historical purification and the previous adjacent moment is taken as the second difference; it should be noted that the initial moment within the specified time period of the reference historical purification is not analyzed because it does not have a previous adjacent moment within the specified time period; the average of the ratios of the duration between all second differences and their corresponding moments is taken as the unit viscosity change rate of the specified time period of the reference historical purification; then the average of the unit viscosity change rates of all specified time periods of the reference historical purification is taken as the reference viscosity unit change value at the current moment of the current purification. It should be noted that the duration of the specified time period needs to be set in conjunction with process requirements, data characteristics, and control objectives. This avoids situations where the specified time period is too short to identify trends in advance, or too long, leading to poor data timeliness and prediction failure. Therefore, the duration of the specified time period should be greater than or equal to the response time required for the reactor temperature control system to execute a control action (such as adjusting heating power) and for a significant temperature change to be observed within the reactor. This ensures that the prediction can cover system delays, making the feedforward control meaningful. The duration of the specified time period should also be based on the typical duration of continuous changes in material viscosity data during the vacuum degassing and impurity evaporation stages. For example, the duration of the specified time period can be set to 10% to 30% of the total duration of the vacuum degassing and impurity evaporation stages. The goal is to capture meaningful trend changes without introducing excessive errors due to overly long-term predictions. The duration of the specified time period should be an integer multiple of the system control cycle. For example, if the control system calculates and outputs a control command every 5 seconds, the duration of the specified time period can be set to 30 seconds or 60 seconds, ensuring synchronization between prediction and control. Furthermore, the duration of the specified time period should be based on the correlation coefficient between the predicted sequence of material viscosity data and the corresponding historical sequence in the reference historical purification being higher than a preset threshold. In this embodiment, the preset threshold is set to 0.7. Implementers can set the size of the preset threshold according to actual conditions, which is not limited here. The method for obtaining the correlation coefficient is a well-known technique, such as the Pearson correlation coefficient and the Kendall rank correlation coefficient, and will not be elaborated further. In one specific embodiment, by analyzing historical data on the vacuum degassing and impurity evaporation stages, the typical duration of this stage was determined to be 20 minutes. Considering the response delay of approximately 1-2 minutes due to system thermal inertia, to ensure the effectiveness of the prediction while also considering computational efficiency, the duration of the specified time period was set to 3 to 5 minutes. This approach proactively covers the system's response delay while remaining within a short period where the material viscosity data change trend can be reasonably predicted. Implementers can set the duration of the specified time period according to actual circumstances; no limitation is imposed here. The time period defined by the specified duration, starting from the current moment, is taken as the first predicted time period for the current purification. All moments within the first predicted time period are taken as the first future moments. For any first future moment, the time period between the first future moment and the current moment is taken as the first duration. The product of the reference viscosity unit change value and the first duration is taken as the viscosity reference change value. Then, the sum of the material viscosity data at the current moment of the current purification and the viscosity reference change value is taken as the predicted material viscosity data at the first future moment.

[0038] At this point, the predicted material viscosity data for each first future moment in the current purification process is obtained.

[0039] Step S4: Based on the predicted material viscosity data and the lag of changes in material viscosity and pressure data in historical purification, obtain the predicted pressure data for the future time of the current purification process.

[0040] Specifically, as the viscosity of the material gradually increases due to impurity removal, the resistance to gas diffusion and overflow in the melt increases, leading to a slowdown in the internal gas release rate. This change causes a lag-induced increase in pressure within the reactor, and the pressure change directly affects the boiling point and heat transfer efficiency of the material, thus feeding back into the system's thermal balance and ultimately causing a temperature change. Therefore, pressure data can serve as an intermediate bridge connecting material viscosity data and temperature data. This allows for the prediction of pressure data based on material viscosity data. In this embodiment, based on predicted material viscosity data and referencing the lag in changes in material viscosity and pressure data during historical purification processes, predicted pressure data for future moments in the current purification process is obtained. This provides a crucial and reliable intermediate variable for subsequent temperature feedforward control, indirectly enabling reliable inference of temperature data changes.

[0041] Preferably, in one feasible way of this embodiment, the method for obtaining the predicted pressure data is as follows: First, based on the lag between changes in material viscosity data and changes in pressure data in the reference historical purification, the lag time of the pressure data relative to the changes in material viscosity data is obtained, so as to accurately determine the time corresponding to the pressure data predicted based on the material viscosity data, thereby improving the stability and accuracy of the predicted pressure data. The method for obtaining the lag time is as follows: For any historical reference purification, the viscosity data of the material from that purification is arranged in chronological order to obtain a viscosity data sequence; the pressure data from that purification is also arranged in chronological order to obtain a pressure data sequence; the cross-correlation function between the viscosity data sequence and the pressure data sequence is calculated, and the time shift that makes the cross-correlation function reach its maximum value is taken as the reference lag time corresponding to that historical reference purification; finally, the mean of the reference lag times corresponding to all historical reference purifications is rounded down to obtain the lag time of the pressure data relative to the material viscosity data. The cross-correlation function is well-known and will not be elaborated further. The moment that is after the current moment and whose time interval is equal to the lag time is taken as the initial second future moment. Where s represents the current time, This indicates the lag time of the change; the time period consisting of a specified time interval starting from the initial second future time is used as the second prediction time period for the current refinement. Where S is the duration of the specified time period; all moments within the second prediction time period are taken as second future moments; wherein, one second future moment corresponds to one first future moment, and the interval between each second future moment and its corresponding first future moment is [duration missing]. Each first future moment corresponding to a specific moment in each historical purification process is designated as a first historical reference moment. The time interval between each first future moment and its corresponding first historical reference moment and the initial degassing moment is the same. Similarly, each second future moment corresponding to a specific moment in each historical purification process is designated as a second historical reference moment. The time interval between each second future moment and its corresponding second historical reference moment and the initial degassing moment is the same. Therefore, it can be inferred that each first historical reference moment corresponds to a second historical reference moment, and the time interval between the first historical reference moment and the corresponding second historical reference moment is also the lag time. ; The material viscosity data and vacuum valve position setting value at each first historical reference time in the reference historical purification, as well as the initial material weight data of the reference historical purification, are used as independent variables; the pressure data at the second historical reference time corresponding to each first historical reference time in the reference historical purification are used as dependent variables; a set of stable reference regression coefficients are obtained by performing multiple linear regression analysis using the least squares method. ;in, This is the basic pressure compensation item, which represents the initial pressure data of the reactor when there is no material and no evacuation. The viscosity coefficient represents the weight of the influence of material viscosity data on the predicted pressure data. The valve position control coefficient represents the weight of the influence of the vacuum valve position on the predicted pressure data. The material weight coefficient represents the weight of the initial material weight data on the predicted pressure data. Then, the trained regression equation is determined: In the formula, For the first Predicted pressure data at the corresponding time point; The duration of the change lag; This represents the material viscosity data at time t. B represents the vacuum valve position setting at time t; B represents the initial material weight data. Finally, the predicted material viscosity data and vacuum valve position setting value for each first future time step, along with the initial material weight data for current purification, are input into the trained regression equation to obtain the predicted pressure data for the corresponding second future time step. The vacuum valve position setting value for each first future time step is pre-set and can be directly accessed.

[0042] It should be noted that for the predicted pressure data at each moment within the time period consisting of the current moment and the initial second future moment, the corresponding regression equation can be obtained by using the actual material viscosity data obtained at the corresponding moment before the current purification moment, through the above-mentioned method of obtaining the regression equation. Thus, the predicted pressure data at each moment within the time period consisting of the current moment and the initial second future moment can be obtained. The duration between each moment within the time period consisting of the current moment and the initial second future moment and its corresponding moment before the current moment is the change lag duration.

[0043] At this point, we have obtained the predicted pressure data for the current purification process at future moments.

[0044] Step S5: Based on the predicted material viscosity data and predicted pressure data, obtain the predicted temperature data for the future time of the current purification process using a temperature prediction model.

[0045] Specifically, to accurately obtain predicted temperature data for future moments and enable timely and accurate temperature control within the reactor, this embodiment uses a temperature prediction model based on predicted material viscosity and pressure data to obtain predicted temperature data for future moments during the current purification process. Specifically, the predicted material viscosity and pressure data for the predicted reference time period are sequentially input into a trained temperature prediction model, which outputs predicted temperature data for each future moment within the reference time period. This clearly reflects the future evolution trend of the temperature data in the reactor without intervention. This embodiment sets the reference time period to 5 minutes. Implementers can set the reference time period length according to actual conditions; it is not limited here, but the initial moment of the reference time period must be the current moment.

[0046] The temperature prediction model is a multi-layer feedforward neural network with ReLU activation, consisting of an input layer, hidden layers, and an output layer. The input layer contains two nodes, receiving standardized predicted material viscosity data and predicted pressure data, respectively. The hidden layers can be one or more, each containing several neurons and employing the ReLU activation function to introduce a non-linear transformation, enabling the model to learn and express the complex coupling relationship between material viscosity, pressure, and temperature data. The output layer is a single neuron using a linear activation function, directly outputting the predicted temperature data, suitable for regression prediction tasks. The loss function of the temperature prediction model is the mean squared error, minimizing the mean squared error between the temperature data predicted by the neural network and the actual temperature data recorded in the historical database. The multi-layer feedforward neural network with ReLU activation, the linear activation function, and the mean squared error are all well-known and will not be elaborated further.

[0047] During the training of the temperature prediction model, a large amount of aligned material viscosity-pressure-temperature data collected from historical purification processes was used as the training set. All weight parameters and bias terms in the network were optimized using the backpropagation algorithm until the model converged, i.e., the loss function decreased to a stable range without overfitting. The backpropagation algorithm is well-known and will not be described in detail here.

[0048] Step S6: Based on the predicted temperature data, control the temperature of the reactor during the current purification process.

[0049] Specifically, based on the obtained predicted temperature data, potential temperature control problems in the reactor can be identified in advance. Then, based on the predicted temperature data, the temperature of the reactor in the current purification process can be controlled, effectively improving the quality of polypropylene purification.

[0050] Specifically, when the predicted temperature data indicates that the future temperature will exceed the preset upper temperature limit, it means that the intensity of the exothermic reaction in the reactor may exceed the heat dissipation or suppression capacity of the current control system. In this case, it is necessary to reduce the heating power of the reactor in advance to preemptively suppress the temperature rise. When the predicted temperature data indicates that the future temperature will be lower than the preset lower temperature limit, it may indicate that the heating system efficiency is decreasing. In this case, it is necessary to increase the heating power of the reactor in advance to actively make up for the expected heat gap. The preset upper temperature limit and the preset lower temperature limit in this embodiment are set by professionals according to the actual situation and are not limited here.

[0051] By controlling the temperature of the reactor during the current purification process using predicted temperature data, a leap from post-mortem correction to pre-mortem prevention is effectively achieved. This allows for control intervention before measurable temperature deviations actually occur, effectively avoiding quality problems such as overheating degradation or incomplete condensation of the product. At the same time, it significantly improves the stability, accuracy, and product consistency of temperature control throughout the polypropylene purification process.

[0052] In summary, this embodiment acquires the state data of the polypropylene purification reactor; obtains the initial degassing time based on the changes in temperature data for each purification process; obtains a reference historical purification data based on similarities in the state data; obtains predicted material viscosity data based on the changes in material viscosity data within a specified time period after the initial degassing time in the reference historical purification, and the material viscosity data at the current moment in the current purification process; and obtains predicted pressure data by combining the lag in changes in material viscosity and pressure data in the reference historical purification. Based on the predicted material viscosity and pressure data, predicted temperature data is obtained to control the temperature of the current purification reactor. This invention, by accurately acquiring predicted temperature data, facilitates early temperature control of the reactor, effectively reducing the lag and inaccuracy of temperature control.

[0053] Example 2: This invention also proposes a temperature control system for a reaction vessel used in polypropylene purification; please refer to [link / reference]. Figure 2 The diagram shows a structural diagram of a temperature control system for a polypropylene purification reactor provided in an embodiment of the present invention. The system includes: a data acquisition module 10, a reference historical purification acquisition module 20, a predicted material viscosity data acquisition module 30, a predicted pressure data acquisition module 40, a predicted temperature data acquisition module 50, and a temperature control module 60.

[0054] The data acquisition module 10 is used to acquire the status data of the reactor in the polypropylene purification process in real time; the status data includes temperature data, pressure data and material viscosity data. The historical purification acquisition module 20 is used to obtain the initial degassing time of each purification based on the change of temperature data for each purification; and to obtain the reference historical purification for the current purification based on the similarity of the state data at the initial degassing time of the current purification and the similarity of the initial material weight. The predicted material viscosity data acquisition module 30 is used to obtain the predicted material viscosity data for future times of the current purification based on the changes in material viscosity data within a specified time period after the initial degassing time of each reference historical purification, and the material viscosity data at the current time of the current purification. The predicted pressure data acquisition module 40 is used to acquire the predicted pressure data for the future moment of the current purification based on the predicted material viscosity data and the lag of changes in material viscosity data and pressure data in historical purification. The predicted temperature data acquisition module 50 is used to acquire the predicted temperature data for the future time of the current purification process based on the predicted material viscosity data and predicted pressure data, through a temperature prediction model. Temperature control module 60 is used to control the temperature of the reactor during the current purification process based on predicted temperature data.

[0055] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer equipment can be divided into different functional modules to complete all or part of the functions described above. In addition, the polypropylene purification reactor temperature control system and the polypropylene purification reactor temperature control method embodiment provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiment, which will not be repeated here.

[0056] Example 3: This invention also proposes a temperature control device for a reaction vessel used in polypropylene purification. The device includes a memory and a processor. The memory stores executable program code, and the processor calls and executes the executable program code to perform a temperature control method for a reaction vessel used in polypropylene purification provided in the embodiments of this application. Specifically, the device may be a chip, component, or module. The chip may include a connected processor and memory; the memory stores instructions, and when the processor calls and executes the instructions, the chip can perform the temperature control method for a reaction vessel used in polypropylene purification provided in the above embodiments.

[0057] Furthermore, this application also protects a computer device; please refer to [link to relevant documentation]. Figure 3 The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the computer device can perform any of the aforementioned methods for controlling the temperature of a reaction vessel for polypropylene purification.

[0058] Example 4: This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the polypropylene purification reactor temperature control method provided in the above embodiment.

[0059] Example 5: This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to implement the polypropylene purification reactor temperature control method provided in the above embodiment.

[0060] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0061] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0062] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for controlling the temperature of a reactor for purifying polypropylene, characterized by, The method comprises the following steps: Real-time acquisition of state data in a polypropylene purification process reactor; the state data comprises temperature data, pressure data and material viscosity data; Based on the change of the temperature data of each purification, the initial degassing time of each purification is obtained; based on the similarity of the state data at the initial degassing time of the current purification and the historical purification, and the similarity of the initial material weight, the reference historical purification of the current purification is obtained; Based on the change of the material viscosity data of each reference historical purification within a specified time period after the initial degassing time, and the material viscosity data at the current time of the current purification, the predicted material viscosity data at the future time of the current purification is obtained; Based on the predicted material viscosity data, and the lag of the change of the material viscosity data and the change of the pressure data in the reference historical purification, the predicted pressure data at the future time of the current purification is obtained; Based on the predicted material viscosity data and the predicted pressure data, the predicted temperature data at the future time of the current purification is obtained through a temperature prediction model; The temperature of the reactor in the current purification process is controlled based on the predicted temperature data.

2. The method for controlling the temperature of a reactor for purifying polypropylene according to claim 1, wherein The method for obtaining the initial degassing time comprises the following steps: For any time of any purification, the difference between the standard deviations of the temperature data of the purification in a first preset time period and a second preset time period is obtained as the degassing reference value of the time; wherein the lengths of the first preset time period and the second preset time period are the same, the end time of the first preset time period is the time, and the start time of the second preset time period is the time; The initial degassing time of the purification is obtained by taking the time corresponding to the maximum degassing reference value as the initial degassing time of the purification.

3. The method of claim 1, wherein the temperature of the reactor is controlled by a temperature controller. The method for obtaining the reference historical purification comprises the following steps: For any historical purification and any state data, the difference between the state data of the current purification and the initial degassing time of the historical purification is normalized as the first difference of the state data; The difference between the initial material weight data of the current purification and the historical purification is normalized as the initial weight difference; The first difference and the initial weight difference of all kinds of state data are added and negatively correlated and normalized as the similarity degree of the current purification and the historical purification; When the similarity degree is greater than a preset similarity degree threshold, the corresponding historical purification is the reference historical purification of the current purification.

4. The method for controlling the temperature of a reactor for purifying polypropylene according to claim 1, wherein The method for obtaining the predicted material viscosity data comprises the following steps: According to the unit change of the material viscosity data of each reference historical purification within a specified time period after the initial degassing time, the reference viscosity unit change value at the current time of the current purification is obtained; wherein the current time is after the initial degassing time of the current purification, and the lengths of the initial time of the specified time period and the current time from the corresponding initial degassing time are equal; The time period formed by the length of the specified time period with the current time as the starting point is taken as the first prediction time period of the current purification; The time of the first prediction time period is taken as the first future time; For any first future time, the length of the first future time from the current time is taken as the first length. The product of the reference viscosity unit change value and the first time length is taken as a viscosity reference change value; The addition result of the current purified material viscosity data at the current time and the viscosity reference change value is taken as the predicted material viscosity data at the first future time.

5. The method of claim 4, wherein the temperature of the reactor is controlled by the temperature of the cooling water. The reference viscosity unit change value is obtained by: For each reference historical purification, the difference between the material viscosity data at each time and the previous adjacent time within the specified time period of the reference historical purification is taken as a second difference; The average of the ratio of all second differences to the time length between the corresponding times is taken as the unit viscosity change rate of the specified time period of the reference historical purification; The average of the unit viscosity change rates of all reference historical purifications is taken as the reference viscosity unit change value at the current time of the current purification.

6. The method of claim 4, wherein the temperature of the reactor is controlled by a temperature controller. The predicted pressure data is obtained by: According to the lag of the change of the material viscosity data and the pressure data in the reference historical purification, the change lag time length of the pressure data relative to the material viscosity data is obtained; The time after the current time and at a distance of the change lag time length is taken as an initial second future time; The time period formed by the time length of the specified time period starting from the initial second future time is taken as the second prediction time period of the current purification; the times within the second prediction time period are all taken as second future times; wherein, one second future time corresponds to one first future time, and the interval time length between each second future time and the corresponding first future time is the change lag time length; The time corresponding to each first future time in each reference historical purification is taken as a first historical reference time; wherein, the time length between each first future time and the corresponding first historical reference time is the same as the time length from the corresponding degassing initial time; The time corresponding to each second future time in each reference historical purification is taken as a second historical reference time; wherein, the time length between each second future time and the corresponding second historical reference time is the same as the time length from the corresponding degassing initial time; The material viscosity data and the vacuum valve position setting value at each first historical reference time in the reference historical purification, and the initial material weight data of the reference historical purification are taken as independent variables; the pressure data at the second historical reference time corresponding to each first historical reference time in the reference historical purification is taken as dependent variable; a set of stable reference regression coefficients is obtained by using the least square method for multiple linear regression analysis: ; wherein, is a basic pressure compensation term, which is the initial pressure data of the reaction kettle without material and without air extraction; is a material viscosity coefficient, which represents the influence weight of the material viscosity data on the predicted pressure data; is a valve position control coefficient, which represents the influence weight of the vacuum valve position on the predicted pressure data; is a material weight coefficient, which represents the influence weight of the initial material weight data on the predicted pressure data; Further determine the trained regression equation: ; in the formula, is the viscosity data of the material at the tth moment; is the predicted pressure data at the corresponding moment; is the change lag length; is the viscosity data of the material at the tth moment; is the vacuum valve position set value at the tth moment; B is the initial material weight data; The predicted material viscosity data at each first future time and the vacuum valve position setting value, and the initial material weight data of the current purification are input into the trained regression equation to obtain the predicted pressure data at the corresponding second future time.

7. The method of claim 6, wherein the temperature of the reactor is controlled by a temperature controller. The change lag time length is obtained by: For each reference historical purification, the material viscosity data of the reference historical purification is arranged in time sequence to obtain a viscosity data sequence; The pressure data of the reference historical purification is arranged in time sequence to obtain a pressure data sequence; The cross-correlation function of the viscosity data sequence and the pressure data sequence is calculated, and the time shift amount that makes the cross-correlation function reach the maximum value is taken as the reference lag time length corresponding to the reference historical purification; The average of the reference lag time lengths corresponding to all reference historical purifications is taken as the change lag time length of the pressure data relative to the material viscosity data.

8. The method of claim 1, wherein the temperature of the reactor is controlled by a temperature controller. The predicted temperature data is obtained by: The predicted material viscosity data and the predicted pressure data in the reference time period are sequentially input into the trained temperature prediction model, and predicted temperature data at each future time in the reference time period is output.

9. The method of claim 1, wherein the temperature of the reactor is controlled by a temperature controller.

9. The method of claim 1, wherein the temperature of the reactor is controlled by a temperature controller. The method for controlling the temperature of the reaction kettle in the current purification process is: When the predicted temperature data indicates that the future temperature will exceed the preset upper temperature limit, the heating power of the reaction kettle is reduced in advance; When the predicted temperature data indicates that the future temperature will be lower than the preset lower temperature limit, the heating power of the reaction kettle is increased in advance.

10. The method of claim 8, wherein the temperature of the reactor is controlled by a temperature controller. The temperature prediction model is a multi-layer feedforward neural network with a ReLU activation function, which is trained by minimizing the mean square error between the predicted temperature and the actual temperature.

Images