Polyvinyl chloride polymerization intelligent control system

Abstract

The present application relates to the technical field of industrial automation control, in particular to a polyvinyl chloride polymerization reaction intelligent control system, the system comprises a trend identification module, a coupling analysis module, a target generation module, a deviation evaluation module and a control execution module.In the present application, the change direction of the polyvinyl chloride chain growth rate and the temperature rise amplitude is combined and analyzed, which surpasses the dependence on single parameter threshold, realizes early identification of potential risks such as reaction acceleration or retardation, introduces external meteorological changes as a key variable, quantifies the coupling influence with the internal reaction state, makes the temperature control change from passive response to prediction-based proactive intervention, and the regulation and control target constructed on this basis not only dynamically associates the current working condition and environmental disturbance, but also can be adapted according to the real-time capacity of the cooling device, and before executing specific adjustment, the expected response of the control action is deduced by referring to historical similar data, and the possible deviation between the adjustment result and the target is quantified.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, and in particular to an intelligent control system for polyvinyl chloride polymerization reaction. Background Technology

[0002] The field of industrial automation control technology mainly involves the use of automation technology to achieve precise operation and process regulation of production equipment in industrial production, including on-site signal acquisition, control algorithm design, actuator regulation and system integration.

[0003] Among them, the intelligent control system for polyvinyl chloride polymerization reaction refers to the operation and control system for the polymerization reaction section in the polyvinyl chloride production process, which monitors and dynamically adjusts key parameters such as temperature, pressure, stirring speed, feed rate and reaction time in real time.

[0004] Existing technologies mainly rely on real-time monitoring and dynamic adjustment of independent key parameters such as temperature and pressure inside the polymerization reactor. The control logic is based on threshold judgment of the absolute value of a single parameter. This method has an inherent lag when abnormalities occur in the early stages of the reaction trend. For example, the control system will only respond when the temperature inside the reactor has deviated significantly from the set value, by which time the best intervention opportunity has often been missed. At the same time, since its operating model does not consider external environmental factors, the system cannot foresee the superimposed impact on the exothermic reaction process when faced with external disturbances such as drastic changes in ambient temperature. As a result, its adjustment actions can only be used as a post-event remedy. Its control strategy is usually preset and fixed, lacking a mechanism to dynamically adjust the intensity and method of adjustment based on subtle differences in the current reaction state and prediction of future trends. This is very likely to cause control overshoot or insufficient response, thereby affecting the uniformity of the molecular weight distribution of PVC products and the overall stability of the production process. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an intelligent control system for polyvinyl chloride polymerization reaction.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a smart control system for polyvinyl chloride polymerization reaction includes:

[0007] The trend recognition module obtains the PVC chain growth rate and the corresponding temperature rise of the reaction region during a specified period in the PVC polymerization reaction, determines the direction of change of the two over a continuous period of time and uses them as combined features for cross-analysis, and assigns polymerization reaction state labels based on the analysis results.

[0008] The coupling analysis module acquires the polymerization reaction status label and meteorological data for the corresponding time period, analyzes the response relationship between temperature changes during the reaction process and external meteorological factors, and outputs temperature trend analysis results.

[0009] The target generation module, based on the polymerization reaction status label and the temperature trend analysis results, constructs a temperature control target with the PVC chain growth rate and the corresponding temperature rise in the reaction region as the main control factors.

[0010] The deviation assessment module uses the chain growth rate response record corresponding to the temperature control target in all operating data as a comparison benchmark to estimate the growth rate generated by this cooling control, compare it with the target growth rate, and output the chain growth deviation prediction result.

[0011] The control execution module, combining the temperature control target and the chain growth offset prediction result, determines the adjustment actions for cooling injection and heat exchange, and generates the temperature control result.

[0012] As a further aspect of the present invention, the polymerization reaction status label specifically includes the direction of chain growth rate change, the direction of temperature rise change, and trend combination characteristics; the temperature trend analysis results include temperature change sensitive sections, meteorological factor response intensity, and staged temperature rise probability; the temperature control target specifically includes the target chain growth rate range, the target temperature rise range, and the adjustment parameter boundary; the chain growth offset prediction results include the predicted growth rate value and the target offset degree; and the temperature control results specifically refer to the cold adjustment command, the heat exchange adjustment command, and the execution adjustment sequence.

[0013] As a further aspect of the present invention, the trend recognition module includes:

[0014] The trend extraction submodule obtains the chain growth rate and the corresponding temperature rise of the reaction region within a specified period during the polyvinyl chloride polymerization reaction, calculates the direction of change of each time period, and assigns a label to the direction of change of chain growth rate and the direction of change of temperature rise according to the positive or negative value of the difference, thereby generating trend direction labels.

[0015] The feature pairing submodule, based on the trend direction label, combines the chain growth rate change direction label and the temperature rise amplitude change direction label into a trend pairing feature, obtains the aggregation stage temperature control target state of the time period corresponding to the trend pairing feature, determines the deviation direction between the pairing feature and the temperature control state, filters the associated feature pairing, and generates a state offset trend combination.

[0016] The state allocation submodule, based on the state offset trend combination, divides the aggregation reaction state type corresponding to the combination trend and assigns corresponding labels to generate aggregation reaction state labels.

[0017] As a further aspect of the present invention, the coupling analysis module includes:

[0018] The tag reading submodule acquires the aggregation reaction status tag, collects meteorological data for the corresponding time period, extracts temperature change parameters from the meteorological data, and pairs them with the status tag in a time synchronization manner to generate a status meteorological pairing group.

[0019] The meteorological coupling submodule, based on the state meteorological pairing group, constructs feature pairs of state labels and temperature change parameters, inputs them into a Bayesian regression model, calculates the joint probability distribution of temperature change to response temperature change value under state label conditions, establishes the response relationship of meteorological change to temperature evolution in the response process, and generates temperature response probability distribution data.

[0020] The trend recognition submodule, based on the temperature response probability distribution data, determines whether the temperature value in the current polymerization process is in a probability region sensitive to air temperature changes, identifies the trend region based on the probability distribution threshold, and outputs the temperature trend analysis results.

[0021] As a further aspect of the present invention, the target generation module includes:

[0022] The parameter adaptation submodule obtains the polymerization reaction state label and the temperature trend analysis results, calls the polyvinyl chloride chain growth rate and reaction area temperature rise data for the corresponding time period, and extracts the set of control input parameters based on the state type and trend area.

[0023] The scheme selection submodule calls the preset cooling adjustment plan in the parameter library, matches the response characteristics and trend area information of the corresponding stage of the status label, judges the degree of fit between each plan and the control input parameter set, and generates a matching cooling plan group.

[0024] The target construction submodule, combined with the real-time cooling device operating capacity parameters, and according to the matching cooling plan group, limits the target range of the chain growth rate adjustment value and the temperature rise adjustment value, forming the next cycle control index combination, and outputs the temperature control target.

[0025] As a further aspect of the present invention, the deviation evaluation module includes:

[0026] The response recording submodule calls the temperature control target, references the cooling control data corresponding to the current control type and temperature control target from all operating data, extracts the chain growth rate response record from the cooling control data as a source of comparison data, and generates a chain growth rate benchmark record set.

[0027] The growth estimation submodule, based on the chain growth rate benchmark record set, calls the adjustment content of the current temperature control target as input, infers the chain growth rate change trend in the data sequence, estimates the response value generated by this cooling control, and uses it as the predicted value of the chain growth rate.

[0028] The offset assessment submodule calls the target growth rate set in the temperature control target, calculates the residual sequence distribution with the predicted chain growth rate, quantifies the degree of offset, and outputs the chain growth offset prediction result.

[0029] As a further aspect of the present invention, the control execution module includes:

[0030] The action setting submodule obtains the temperature control target and the chain growth offset prediction result, calls the control interval and the predicted offset direction, calculates the adjustment range and sorting order of cooling injection and heat exchange for each cooling control point, determines the sequence of adjustment operations and the target range, and generates the adjustment action execution sequence.

[0031] The instruction generation submodule constructs the cooling control point's injection and heat exchange instructions according to the execution sequence of the adjustment actions, encodes the execution position and adjustment range of the instructions respectively, forms a control-end identifiable format, and generates a set of injection and heat exchange operation instructions.

[0032] The control feedback submodule synchronously transmits the set of cooling and heat exchange operation instructions to the corresponding execution unit at the control end, collects and records the real-time response results of the regulation process, and outputs the temperature control results.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0034] In this invention, by combining the analysis of the changes in the growth rate of polyvinyl chloride chains and the magnitude of temperature rise, the reliance on a single parameter threshold is overcome, enabling early identification of potential risks such as accelerated or delayed reactions. Furthermore, external meteorological changes are introduced as a key variable, quantifying their coupling influence with the internal reaction state. This transforms temperature control from a passive response to a predictive, proactive intervention. The control target constructed on this basis not only dynamically correlates the current operating conditions with environmental disturbances but also adapts to the real-time capabilities of the cooling device. Before executing specific adjustments, the expected response of the control action is extrapolated by referencing historical data of similar types, predicting and quantifying the possible deviation between the adjustment result and the target. Finally, based on this deviation prediction, the execution sequence and adjustment magnitude of various adjustment methods such as cooling injection and heat exchange are precisely set, effectively avoiding over- or under-adjustment and ensuring the stable operation of the reaction process and the consistency of product quality. Attached Figure Description

[0035] Figure 1 This is a system flowchart of the present invention;

[0036] Figure 2 This is a flowchart of the trend recognition module of the present invention;

[0037] Figure 3 This is a flowchart of the coupling analysis module of the present invention;

[0038] Figure 4 This is a flowchart of the target generation module of the present invention;

[0039] Figure 5 This is a flowchart of the deviation evaluation module of the present invention;

[0040] Figure 6 This is a flowchart of the control execution module of the present invention. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] Please see Figure 1 A smart control system for polyvinyl chloride polymerization reaction includes:

[0043] The trend recognition module obtains the PVC chain growth rate and the corresponding temperature rise of the reaction region during a specified period in the PVC polymerization reaction, determines the direction of change of the two over a continuous period of time and uses them as combined features for cross-analysis, and assigns polymerization reaction state labels based on the analysis results.

[0044] The coupling analysis module acquires the polymerization reaction status label and meteorological data for the corresponding time period, analyzes the response relationship between temperature changes during the reaction process and external meteorological factors, and outputs temperature trend analysis results.

[0045] The target generation module, based on the polymerization reaction status label and temperature trend analysis results, uses the PVC chain growth rate and the corresponding temperature rise in the reaction region as the main control factors to construct a temperature control target.

[0046] The deviation assessment module uses the chain growth rate response record corresponding to the temperature control target in all operating data as a comparison benchmark to estimate the growth rate generated by this cooling control, compare it with the target growth rate, and output the chain growth deviation prediction result.

[0047] The control execution module, combining the temperature control target and the chain growth offset prediction results, determines the adjustment actions for cooling injection and heat exchange, and generates temperature control results.

[0048] The polymerization reaction status labels specifically include the direction of chain growth rate change, the direction of temperature rise change, and trend combination characteristics. The temperature trend analysis results include temperature change sensitive sections, meteorological factor response intensity, and staged temperature rise probability. The temperature control targets specifically include the target chain growth rate range, the target temperature rise range, and the adjustment parameter boundaries. The chain growth offset prediction results include the predicted growth rate value and the target offset degree. The temperature control results specifically indicate the cold adjustment command, heat exchange adjustment command, and the order of execution of adjustment.

[0049] Please see Figure 2 The trend recognition module includes:

[0050] The trend extraction submodule obtains the chain growth rate and the corresponding temperature rise of the reaction region within a specified period during the polyvinyl chloride polymerization reaction, calculates the direction of change of each time period, and assigns a label to the direction of change of chain growth rate and the direction of change of temperature rise according to the positive or negative value of the difference, thereby generating trend direction labels.

[0051] To obtain the chain growth rate and corresponding temperature rise in the reaction zone during a specified period in the polyvinyl chloride (PVC) polymerization reaction, this process begins by retrieving the continuous monitoring records of the most recent hour from an online Raman spectrometer deployed inside the polymerization reactor. This time window is then divided into six time periods at ten-minute intervals, yielding a chain growth rate sequence of [0.48, 0.52, 0.51, 0.58, 0.60, 0.59], in kilograms per molar second. Simultaneously, the temperature rise during the same time period is collected from a distributed fiber optic temperature sensor array covering the reactor jacket. After calculating the arithmetic mean of the data from three key monitoring points, the corresponding temperature rise sequence is obtained as [0.19, 0.23, 0.22, 0.29, 0.30, 0.59].

[28] The unit is Celsius. Then, the direction of change in adjacent time periods is calculated. This calculation is completed by subtracting the value of the next time period from the value of the previous time period. Then, according to the sign of the difference result, labels are assigned to the direction of change of chain growth rate and the direction of change of temperature rise. The assignment criteria are: if the difference result is positive, a "positive" label is assigned; if the difference result is negative, a "negative" label is assigned; if the difference result is zero, a "stable" label is assigned. Based on this standard, the label sequence of the direction of change of chain growth rate is determined as [positive, negative, positive, positive, negative], and the label sequence of the direction of change of temperature rise is determined as [positive, negative, positive, positive, negative]. Finally, a trend direction label composed of these two sets of direction labels is generated.

[0052] The feature pairing submodule combines the chain growth rate change direction label and the temperature rise magnitude change direction label into a trend pairing feature based on the trend direction label. It obtains the temperature control target state of the aggregation stage in the time period corresponding to the trend pairing feature, determines the deviation direction between the pairing feature and the temperature control state, filters the associated feature pairing, and generates the state offset trend combination.

[0053] Based on the trend direction labels, namely the chain growth rate change direction label sequence and the temperature rise change direction label sequence generated in the previous step, the labels at the same time position in the two sequences are combined one-to-one, thus forming a trend pairing feature for five consecutive time periods. Specifically, this feature is [(positive, positive), (negative, negative), (positive, positive), (positive, positive), (negative, negative)]. During this period, the temperature control target state of the polymerization stage corresponding to these five time periods in the production process specification is retrieved from the manufacturing execution information archive. This state sequence is [constant temperature maintenance, constant temperature maintenance, programmed temperature rise, programmed temperature rise, constant temperature maintenance]. Next, the direction of deviation between the paired features and the temperature control state is determined. This determination is based on a set of fixed logical rules. The core of this rule set is as follows: In the "constant temperature maintenance" state, the theoretically ideal paired features are (stable, stable). Any non-"stable" combination is judged as a deviation. Among them, the (positive, positive) combination is defined as "leading deviation", and the (negative, negative) combination is defined as "lagging deviation". In the "programmed temperature rise" state, the theoretically ideal paired features are (positive, positive). All other combinations are judged as deviations. Then, based on this benchmark, the associated feature pairs with deviations are screened, and finally, the state offset trend combination is generated.

[0054] The state assignment submodule, based on the combination of state offset trends, divides the aggregation reaction state types corresponding to the combination trends and assigns corresponding labels to generate aggregation reaction state labels.

[0055] Based on the state deviation trend combination, whose content is [(Time Period 1, (Positive, Positive), Leading Deviation), (Time Period 2, (Negative, Negative), Lagging Deviation), (Time Period 5, (Negative, Negative), Lagging Deviation)], the polymerization reaction state type corresponding to the combination trend is classified and assigned a corresponding label. This classification is based on a rule knowledge base, which is based on statistical clustering analysis of more than 10,000 batches of historical polymerization reaction data. For example, one rule clearly states: when the temperature control target is "constant temperature maintenance" and the deviation direction is "leading deviation", the polymerization reaction state type is classified and assigned a corresponding label. The reaction state type is defined as "accelerated reaction risk"; when the temperature control target is "constant temperature maintenance" and the deviation direction is "lagging deviation", the state type is defined as "slow reaction risk". Based on these established rules, the state deviation trend combinations selected in the previous steps are matched and assigned one by one. For the combination of time period one, it is labeled "accelerated reaction risk" and for the combination of time period two and time period five, they are both labeled "slow reaction risk". In this way, a clear aggregation reaction state label is generated for each time period in which deviation occurs.

[0056] Please see Figure 3 The coupling analysis module includes:

[0057] The tag reading submodule acquires the aggregation reaction status tag, collects meteorological data for the corresponding time period, extracts temperature change parameters from the meteorological data, and pairs them with the status tag according to the time synchronization method to generate a status meteorological pairing group.

[0058] The aggregation reaction status label generated by the state allocation submodule during the subsequent time period (T7) is obtained. Assuming the label obtained at this moment is "reaction acceleration risk", the outdoor ambient air temperature is collected during the time period completely synchronized with T7 by accessing the data interface of the meteorological monitoring station deployed at the aggregation reactor site. For example, the measurement value at the beginning of T7 (15:00) is 30.1 degrees Celsius, and the measurement value at the end of T7 (15:10) is 30.5 degrees Celsius. The temperature change parameter is extracted from the collected meteorological data. This parameter is obtained by subtracting the temperature value at the beginning from the temperature value at the end, and the result is 0.4 degrees Celsius. Then, the status label "reaction acceleration risk" and the calculated temperature change parameter 0.4 degrees Celsius are strictly matched according to the timestamp to generate a state meteorological pairing group.

[0059] The meteorological coupling submodule, based on the state-meteorological pairing group, constructs feature pairs of state labels and temperature change parameters, inputs them into the Bayesian regression model, calculates the joint probability distribution of temperature change to response temperature change value under the state label condition, establishes the response relationship of meteorological change to temperature evolution in the response process, and generates temperature response probability distribution data.

[0060] Based on the state-meteorological pairing group, namely ("Accelerated Reaction Risk", 0.4 degrees Celsius), feature pairs between state labels and temperature change parameters are constructed. This process first quantifies and normalizes the non-numerical state label "Accelerated Reaction Risk". The quantification standard is pre-set based on the statistical correlation strength between different state labels and the reaction exothermic rate in the historical database. For example, "Accelerated Reaction Risk" is quantified with an original score S equal to 1.5 because it has the highest positive correlation. Then, normalization is performed to eliminate dimensions. The normalization method is to divide the original score by a baseline value. The baseline value is set to 1.5, which is based on selecting the largest absolute value among all possible state label quantization values, thereby normalizing the state variables. The value range of is constrained to be between [-1, 1], and the current state is The calculation is 1.5 divided by 1.5, resulting in 1.0. This feature pair is then input into an optimized Bayesian regression model to calculate the joint probability distribution of the state label and temperature change on the response temperature change value. The model expression is as follows:

[0061] ;

[0062] In this expression, each term has a clear physical meaning and a basis for its definition: This represents the expected change in the reaction temperature inside the reactor for the next cycle, expressed in degrees Celsius (°C), and is the final prediction target for this module. It is the intercept of the base temperature change, with a value of 0.08, in degrees Celsius (°C). It represents the state in which the reaction is most stable (i.e., ...). And the outside temperature remains unchanged (i.e.) Under ideal baseline conditions, the natural temperature rise rate inside the reactor caused solely by the exothermic reaction itself is set based on the statistical average of the actual temperature rise rates under this ideal baseline conditions from over five thousand batches of historical data. This is the state influence coefficient, with a value of 0.55, in degrees Celsius (°C). It represents the temperature change caused by a unit normalized state variable. It is dimensionless, therefore The dimension of the coefficient is the same as that of temperature. It is used to quantify the independent contribution of internal reaction states (such as accelerated or slowed reaction) to temperature changes. The coefficient is set based on the results of historical data regression analysis. The more significant the influence of internal reaction states on temperature runaway in the past, the larger the absolute value of the coefficient. It is the normalized state variable calculated above, with a value of 1.0, dimensionless. It transforms the qualitative label "accelerated reaction risk" from paragraph 3 into a standardized numerical value for calculation in the model. It is the meteorological main effect coefficient, with a value of 1.10. It is dimensionless and represents the multiple of the change in reaction temperature inside the reactor caused by a change of one degree Celsius in the outside temperature. It reflects the heat insulation performance of the reactor and its overall sensitivity to the ambient temperature. Its setting is based on the results of regression analysis of historical data. The larger the overall heat transfer coefficient of the reactor and the worse the heat insulation performance, the larger the value of this coefficient. This is meteorological data from paragraph 4, which is the measured change in outside temperature, with a value of 0.4, in degrees Celsius (°C). This is the interaction coefficient, with a value of 0.50, dimensionless. It quantifies the coupling effect between the internal reaction state and external temperature changes. Specifically, it measures whether the impact of external temperature changes on the reactor temperature is amplified or reduced when the reaction state itself is unstable. Its setting is also based on historical data regression analysis. If historically, there have been frequent cases of deteriorating reaction states coinciding with extreme weather leading to drastic temperature changes, the absolute value of this coefficient will significantly deviate from zero. The calculation process of substituting the current value into this formula is as follows: The calculation results show that the expected value of the predicted temperature change is 1.27 degrees Celsius. The model also shows that this expected value follows a normal distribution with a mean of 1.27 degrees Celsius and a standard deviation of 0.16 degrees Celsius. This is the generated temperature response probability distribution data.

[0063] The trend recognition submodule, based on temperature response probability distribution data, determines whether the temperature value in the current polymerization process is in a probability region sensitive to air temperature changes, identifies the trend region based on the probability distribution threshold, and outputs the temperature trend analysis results.

[0064] Based on the temperature response probability distribution data, specifically a normal distribution with a mean of 1.27 degrees Celsius and a standard deviation of 0.16 degrees Celsius, this study determines whether the current temperature value during the polymerization reaction is within a probability region sensitive to air temperature changes. This determination is based on a probability distribution threshold, set at a predicted temperature change exceeding 0.80 degrees Celsius. This threshold is determined through backtesting of historical data. Statistical results show that when the predicted temperature rise exceeds 0.80 degrees Celsius, 98% of cases ultimately trigger cooling adjustments within the following 30 minutes. The currently calculated expected value of 1.27 degrees Celsius is greater than the 0.80 degree Celsius threshold. Therefore, the current polymerization reaction is identified as being within a probability region highly sensitive to air temperature changes. Based on this determination, the current reaction trend is categorized as "high-risk external disturbance-driven warming," and this temperature trend analysis result is output.

[0065] Please see Figure 4 The target generation module includes:

[0066] The parameter adaptation submodule obtains the polymerization reaction state label and temperature trend analysis results, calls the PVC chain growth rate and reaction area temperature rise data for the corresponding time period, and extracts the set of control input parameters based on the state type and trend area.

[0067] The polymerization reaction status label "Reaction Acceleration Risk" and the temperature trend analysis result "High-Risk External Disturbance-Driven Heating" are obtained. Simultaneously, the measured values ​​of the PVC chain growth rate and the temperature rise of the reaction zone, which correspond exactly to the current time period, are retrieved from the real-time database. These values ​​are 0.68 kg / mol / s and 0.39 degrees Celsius, respectively. Based on the two combined conditions of status label and trend attribution, a precise query is performed in the historical production database to retrieve all historical operating condition records that simultaneously meet these two conditions. From these records, the values ​​of all control input parameters used within 15 minutes after the occurrence and before successful control intervention are extracted, forming a set containing dozens of sets of parameter states before successful historical intervention. This set is the control input parameter set.

[0068] The scheme selection submodule calls the preset cooling adjustment plan in the parameter library, matches the response characteristics and trend area information of the corresponding stage of the status label, judges the degree of fit between each plan and the control input parameter set, and generates a matching cooling plan group.

[0069] The system calls upon preset cooling control plans from the parameter library, which stores over twenty standard operation sequences labeled "Plan-A" and "Plan-B". Then, using the status label "Accelerated Reaction Risk" and the trend area information "High-Risk External Disturbance-Dominated Temperature Rise" as matching indices, it evaluates the suitability of each preset plan. The evaluation process involves: first, calculating the arithmetic mean vector of all parameters in the control input parameter set obtained in the previous step; then, for each preset plan, calculating its theoretical control effect vector; and finally, calculating the Euclidean distance between this theoretical effect vector and the parameter set's average vector. The smaller this distance, the higher the suitability. The selection criterion is that the distance value must be less than a preset suitability threshold of 1.25. This threshold is set based on the 80th percentile of the distance between the selected scheme and the parameter set under the current operating conditions in all historical successful control cases. All plans with a calculated distance less than 1.25 are selected, generating a matching cooling plan group.

[0070] The target construction submodule, combined with the real-time cooling device operating capacity parameters, and according to the matching cooling plan group, limits the target range of the chain growth rate adjustment value and the temperature rise adjustment value to form the next cycle control index combination and outputs the temperature control target.

[0071] Combining real-time cooling device operating capacity parameters obtained from equipment monitoring services, for example, by reading ultrasonic flow meter data to determine that the current maximum output flow of the main cooling water pump is 96% of its rated value, and based on the matching cooling plan group generated in the previous step (assuming "Plan-A" is selected), target ranges are defined for the chain growth rate adjustment value and the temperature rise adjustment value. The ideal chain growth rate adjustment target set in "Plan-A" is -0.12 kg / mol / s, and the ideal temperature rise adjustment target is -0.60 degrees Celsius. Considering that the real-time operating capacity is the rated value... 96% of the ideal adjustment target value is multiplied by 0.96 to form the lower limit of the adjustment target range. That is, the lower limit of chain growth rate adjustment is -0.1152 and the lower limit of temperature rise adjustment is -0.576. At the same time, the upper limit of adjustment is set. The upper limit is set according to the maximum allowable adjustment amount specified by the process specification to prevent excessive inhibition of reaction, that is, not exceeding 110% of the ideal value. Thus, the upper limit of chain growth rate adjustment is calculated to be -0.132 and the upper limit of temperature rise adjustment is -0.66. Finally, the control index combination for the next cycle is formed, and this temperature control target is output.

[0072] Please see Figure 5 The deviation assessment module includes:

[0073] The response recording submodule calls the temperature control target, references the cooling control data corresponding to the current control type and temperature control target from all running data, extracts the chain growth rate response record from the cooling control data as a source of comparison data, and generates a chain growth rate benchmark record set.

[0074] The temperature control target is invoked, namely the chain growth rate adjustment target range [-0.132, -0.1152] and the temperature rise amplitude adjustment target range [-0.66, -0.576]. Using these as search criteria, all historical cooling control data corresponding to the current control type and the current temperature control target are referenced from all historical operation data archives. The specific screening criteria are: the minimum and maximum values ​​of the control target range in the historical records both fall within ±5% of the minimum and maximum values ​​of the current target range. From all historical cooling control events that pass this criterion, the actual chain growth rate response change data within 15 minutes after the execution of the control command is specifically extracted. These data sequences together constitute a comparative data source, and finally, a chain growth rate benchmark record set is generated.

[0075] The growth estimation submodule, based on the chain growth rate benchmark record set, calls the adjustment content of the current temperature control target as input, infers the chain growth rate change trend in the data sequence, estimates the response value generated by this cooling control, and uses it as the predicted value of the chain growth rate.

[0076] Based on a baseline record set of chain growth rate data, which includes, for example, 30 historical chain growth rate response data sequences under similar conditions, and using the current temperature control target adjustment content as the inference input, specifically the median of the current target adjustment range (i.e., a chain growth rate adjustment target of -0.1236 kg / mol / s), the process of inferring the chain growth rate trend from the data sequences is as follows: First, the arithmetic mean of the 30 historical response data sequences in the baseline record set is calculated at each time point to construct an average response curve. Then, the current input adjustment target value is compared with the average adjustment target value of these 30 historical events, and a scaling factor is calculated. For example, if the historical average adjustment target is -0.1200, the scaling factor is 1.03. Next, each data point on this average response curve is multiplied by the scaling factor 1.03 to obtain a scaled predicted response curve. Finally, the value corresponding to this predicted response curve at the end of the control cycle is taken to estimate the final response value generated by this cooling control. This value is the predicted chain growth rate value.

[0077] The offset assessment submodule calls the target growth rate set in the temperature control target, calculates the residual sequence distribution with the predicted chain growth rate, quantifies the degree of offset, and outputs the chain growth offset prediction result.

[0078] The target growth rate set in the temperature control target is called, which is the median of the interval -0.1236 kg / mol / s. The residual between this target value and the chain growth rate prediction value (assumed to be -0.1250) output by the growth estimation submodule is calculated, which is -0.0014. To more comprehensively quantify the degree of deviation, this residual calculation is extended to every corresponding time point between the entire predicted response curve and the target response curve, resulting in a complete residual sequence distribution. Subsequently, the root mean square error of the residual sequence is calculated, which is used as the final quantitative indicator of the degree of deviation. The evaluation criteria for the degree of deviation are divided into three intervals: a root mean square error less than 0.002 is defined as "low deviation", between 0.002 and 0.005 is defined as "moderate deviation", and greater than 0.005 is defined as "high deviation". The boundary values ​​of these intervals are set according to the statistical distribution of the deviation between historical predictions and actual results, corresponding to the 30th and 80th percentiles of the distribution, respectively. Based on the calculated root mean square error value and its interval, the chain growth deviation prediction result is output.

[0079] Please see Figure 6 The control execution module includes:

[0080] The action setting submodule obtains the temperature control target and chain growth offset prediction results, calls the control range and predicted offset direction, calculates the adjustment range and sorting order of cooling injection and heat exchange for each cooling control point, determines the sequence of adjustment operations and target range, and generates the adjustment action execution sequence.

[0081] Obtain the temperature control target range and chain growth offset prediction results. Assuming the offset prediction result is "negative moderate offset", this result indicates that the estimated cooling effect may be slightly stronger than expected. Therefore, when setting specific adjustment actions, the original plan needs to be fine-tuned. Call the control range and the predicted offset direction, and calculate the adjustment amplitude and sorting order of refrigerant injection and heat exchange for each cooling control point. Based on the prediction of "negative moderate offset", the adjustment strategy will prioritize the jacket heat exchange adjustment with higher adjustment accuracy and relatively smoother response, and reduce the refrigerant injection volume which has a rapid response but is difficult to control precisely. First, determine the order of adjustment operations as adjusting the jacket circulating water first. Second, according to the offset value quantified by the offset prediction result, slightly lower the total adjustment target, and split the revised total target according to the ratio of 85% allocated to heat exchange adjustment and 15% allocated to refrigerant injection adjustment. Finally, generate a detailed sequence of adjustment actions that includes the order and specific adjustment amplitude.

[0082] The instruction generation submodule constructs the cooling control point's injection and heat exchange instructions according to the sequence of adjustment actions, encodes the execution position and adjustment range of the instructions respectively, forms a control-end recognizable format, and generates a set of injection and heat exchange operation instructions.

[0083] Following the sequence of adjustment actions, injection and heat exchange commands for the corresponding cooling control points are constructed. First, the adjustment range allocated to heat exchange adjustment is converted into a specific valve opening increment command by querying the characteristic curve of the heat exchanger valve. For example, increasing the opening of the circulating water outlet valve in jacket A area by 18% generates the command "HWCV-A:SETPOS:INC:18". Then, the adjustment range allocated to injection adjustment is converted into a precise pump start-up time command by querying the physical properties of the injection refrigerant and the flow calibration data of the injection pump. For example, starting the injection pump at port B for 0.8 seconds generates the command "INJ-B:RUNTIME:SET:0.8". The execution location, operation type, and adjustment range of the command are encoded into a standardized string format that can be directly recognized and executed by the control hardware, ultimately forming the injection and heat exchange operation command set.

[0084] The control feedback submodule synchronously transmits the set of cooling and heat exchange operation instructions to the corresponding execution unit at the control end, collects and records the real-time response results of the regulation process, and outputs the temperature control results.

[0085] The set of cooling and heat exchange operation instructions is synchronously transmitted via a fieldbus network to the corresponding execution units of the field programmable logic controllers (PLCs) that control the jacket valves and cooling pumps, respectively. The execution units receive and parse the instructions, and drive the valve positioners and pump relays in the field to complete the specified operations. Within a preset observation period from instruction issuance, execution, and completion, the entire real-time response results of the entire control process are continuously collected and recorded at a frequency of not less than 1 Hz. The collected data includes the temperature, pressure, stirring power, and chain growth rate calculated in real time by a Raman spectrometer in the reactor. These high-frequency response data are bound and associated with the executed instruction content and the timestamp accurate to milliseconds, and stored in a long-term historical database. Finally, a complete and detailed data log containing instructions, process responses, and final status is output as the result of this temperature control. By implementing the temperature control results, abnormal temperature rises caused by meteorological changes, heat load fluctuations, etc., can be effectively addressed, preventing local overheating in the polymerization reaction that leads to side reactions or explosive polymerization, deviation of the chain growth rate from the target, impact on molecular weight distribution, and increased energy consumption and equipment wear due to heat exchange imbalance.

[0086] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A smart control system for polyvinyl chloride polymerization reaction, characterized in that, The system includes: The trend recognition module obtains the PVC chain growth rate and the corresponding temperature rise of the reaction region during a specified period in the PVC polymerization reaction, determines the direction of change of the two over a continuous period of time and uses them as combined features for cross-analysis, and assigns polymerization reaction state labels based on the analysis results. The coupling analysis module acquires the polymerization reaction status label and meteorological data for the corresponding time period, analyzes the response relationship between temperature changes during the reaction process and external meteorological factors, and outputs temperature trend analysis results. The target generation module, based on the polymerization reaction status label and the temperature trend analysis results, constructs a temperature control target with the PVC chain growth rate and the corresponding temperature rise in the reaction region as the main control factors. The deviation assessment module uses the chain growth rate response record corresponding to the temperature control target in all operating data as a comparison benchmark to estimate the growth rate generated by this cooling control, compare it with the target growth rate, and output the chain growth deviation prediction result. The control execution module, combining the temperature control target and the chain growth offset prediction result, determines the adjustment actions of cooling injection and heat exchange, and generates temperature control results; The trend recognition module includes: The trend extraction submodule obtains the chain growth rate and the corresponding temperature rise of the reaction region within a specified period during the polyvinyl chloride polymerization reaction, calculates the direction of change of each time period, and assigns a label to the direction of change of chain growth rate and the direction of change of temperature rise according to the positive or negative value of the difference, thereby generating trend direction labels. The feature pairing submodule, based on the trend direction label, combines the chain growth rate change direction label and the temperature rise amplitude change direction label into a trend pairing feature, obtains the aggregation stage temperature control target state of the time period corresponding to the trend pairing feature, determines the deviation direction between the pairing feature and the temperature control state, filters the associated feature pairing, and generates a state offset trend combination. The state allocation submodule, based on the state offset trend combination, divides the aggregation reaction state type corresponding to the combination trend and assigns corresponding labels to generate aggregation reaction state labels.

2. The intelligent control system for polyvinyl chloride polymerization reaction according to claim 1, characterized in that, The polymerization reaction status labels specifically include the direction of chain growth rate change, the direction of temperature rise change, and trend combination characteristics. The temperature trend analysis results include temperature change sensitive sections, meteorological factor response intensity, and staged temperature rise probability. The temperature control targets specifically include the target chain growth rate range, the target temperature rise range, and the adjustment parameter boundaries. The chain growth offset prediction results include the predicted growth rate value and the target offset degree. The temperature control results specifically refer to the cold adjustment command, the heat exchange adjustment command, and the execution adjustment sequence.

3. The intelligent control system for polyvinyl chloride polymerization reaction according to claim 1, characterized in that, The coupling analysis module includes: The tag reading submodule acquires the aggregation reaction status tag, collects meteorological data for the corresponding time period, extracts temperature change parameters from the meteorological data, and pairs them with the status tag in a time synchronization manner to generate a status meteorological pairing group. The meteorological coupling submodule, based on the state meteorological pairing group, constructs feature pairs of state labels and temperature change parameters, inputs them into a Bayesian regression model, calculates the joint probability distribution of temperature change to response temperature change value under state label conditions, establishes the response relationship of meteorological change to temperature evolution in the response process, and generates temperature response probability distribution data. The trend recognition submodule, based on the temperature response probability distribution data, determines whether the temperature value in the current polymerization process is in a probability region sensitive to air temperature changes, identifies the trend region based on the probability distribution threshold, and outputs the temperature trend analysis results.

4. The intelligent control system for polyvinyl chloride polymerization reaction according to claim 3, characterized in that, The target generation module includes: The parameter adaptation submodule obtains the polymerization reaction state label and the temperature trend analysis results, calls the polyvinyl chloride chain growth rate and reaction area temperature rise data for the corresponding time period, and extracts the set of control input parameters based on the state type and trend area. The scheme selection submodule calls the preset cooling adjustment plan in the parameter library, matches the response characteristics and trend area information of the corresponding stage of the status label, judges the degree of fit between each plan and the control input parameter set, and generates a matching cooling plan group. The target construction submodule, combined with the real-time cooling device operating capacity parameters, and according to the matching cooling plan group, limits the target range of the chain growth rate adjustment value and the temperature rise adjustment value, forming the next cycle control index combination, and outputs the temperature control target.

5. The intelligent control system for polyvinyl chloride polymerization reaction according to claim 4, characterized in that, The deviation assessment module includes: The response recording submodule calls the temperature control target, references the cooling control data corresponding to the current control type and temperature control target from all operating data, extracts the chain growth rate response record from the cooling control data as a source of comparison data, and generates a chain growth rate benchmark record set. The growth estimation submodule, based on the chain growth rate benchmark record set, calls the adjustment content of the current temperature control target as input, infers the chain growth rate change trend in the data sequence, estimates the response value generated by this cooling control, and uses it as the predicted value of the chain growth rate. The offset assessment submodule calls the target growth rate set in the temperature control target, calculates the residual sequence distribution with the predicted chain growth rate, quantifies the degree of offset, and outputs the chain growth offset prediction result.

6. The intelligent control system for polyvinyl chloride polymerization reaction according to claim 5, characterized in that, The control execution module includes: The action setting submodule obtains the temperature control target and the chain growth offset prediction result, calls the control interval and the predicted offset direction, calculates the adjustment range and sorting order of cooling injection and heat exchange for each cooling control point, determines the sequence of adjustment operations and the target range, and generates the adjustment action execution sequence. The instruction generation submodule constructs the cooling control point's injection and heat exchange instructions according to the execution sequence of the adjustment actions, encodes the execution position and adjustment range of the instructions respectively, forms a control-end identifiable format, and generates a set of injection and heat exchange operation instructions. The control feedback submodule synchronously transmits the set of cooling and heat exchange operation instructions to the corresponding execution unit at the control end, collects and records the real-time response results of the regulation process, and outputs the temperature control results.

Images