Method, device and equipment for controlling temperature of reaction kettle and storage medium

Abstract

The application provides a kind of reaction kettle temperature control method, device, equipment and storage medium, it is related to reaction kettle control technical field.The method comprises: obtaining real-time operating parameter input mechanism model, obtain mechanism temperature prediction value;Wherein, mechanism model is based on the heat balance equation of reaction kettle construction;Mechanism temperature prediction value is solved heat balance equation using four-order Runge-Kutta method to obtain;According to mechanism temperature prediction value and sensor measured temperature, calculate residual value, and the residual value is standardized;Obtain the standardized residual value in current time and before setting time window, construct standardized residual value sequence, and the standardized residual value sequence is input TCN model, and the residual correction value is output;Through dynamic weight adjustment mechanism fusion mechanism temperature prediction value and residual correction value, calculate mixed temperature prediction value, and determine DCS control scheme according to mixed temperature prediction value.The application can improve the adaptability of temperature prediction to complex dynamic process, improve prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of reactor control technology, and in particular to a method, apparatus, equipment and storage medium for controlling the temperature of a reactor, especially a method and apparatus for controlling a reactor used in the synthesis of sodium carboxymethyl cellulose. Background Technology

[0002] Sodium carboxymethyl cellulose is an anionic, linear, water-soluble cellulose ether. It is a derivative obtained by chemically modifying natural cellulose with chloroacetic acid. Its aqueous solution has the functions of thickening, film forming, binding, water retention, colloidal protection, emulsification, and suspension. It can be used as a flocculant, chelating agent, emulsifier, thickener, water retainer, and film forming material, and is widely used in food, medicine, electronics, pesticides, leather, plastics, printing and other fields.

[0003] Sodium carboxymethyl cellulose (CMC) can be prepared using methods such as the aqueous method, the low-concentration solvent method, and the slurry method (also known as the multi-concentration solvent method). The slurry method can produce not only high-purity CMC but also CMC with a high degree of substitution and uniform substitution. However, the slurry method for producing high-purity CMC is time-consuming, and factors such as raw materials and environmental conditions can negatively impact the economic efficiency of the enterprise. Traditional control methods rely on linear assumptions, which are ill-suited to the nonlinear exothermic processes and sudden changes in latent heat of vaporization during CMC production due to the heterogeneity of raw materials and the large variations in environmental parameters, easily leading to overshoot or response lag. Summary of the Invention

[0004] This invention provides a method, apparatus, equipment, and storage medium for controlling the temperature of a reaction vessel, in order to solve the problem that the control scheme cannot adapt to the dynamic processes such as nonlinear exothermic and sudden changes in latent heat of vaporization in the production of sodium carboxymethyl cellulose, which easily leads to overshoot or response lag.

[0005] In a first aspect, embodiments of the present invention provide a method for controlling the temperature of a reaction vessel, used in a reaction vessel for the synthesis of sodium carboxymethyl cellulose, the method comprising: Real-time operating parameters are input into the mechanism model to obtain the predicted mechanism temperature; wherein, the mechanism model is constructed based on the heat balance equation of the reactor; the predicted mechanism temperature is obtained by solving the heat balance equation using the fourth-order Runge-Kutta (RK) method; The residual value is calculated based on the predicted temperature value and the actual temperature measured by the sensor, and the residual value is then standardized. Obtain the standardized residual values ​​within the current time and the previously set time window, construct a standardized residual value sequence, and input the standardized residual value sequence into a temporal convolutional network (TCN) to output the residual correction value; The predicted temperature value is fused with the residual correction value through a dynamic weight adjustment mechanism to calculate the mixed temperature prediction value, and the control scheme of the distributed control system (DCS) is determined based on the mixed temperature prediction value.

[0006] In one possible implementation, the step of fusing the mechanistic temperature prediction value and the residual correction value through a dynamic weight adjustment mechanism to calculate the mixed temperature prediction value includes: The number of times the residual value exceeds a set residual threshold within a set time period is cumulatively counted. The weights of the predicted mechanism temperature and the corrected residual value are adjusted based on the cumulative count; wherein the sum of the weights of the predicted mechanism temperature and the corrected residual value is 1.

[0007] In one possible implementation, the formula for calculating the predicted mixed temperature is:

[0008] in, The predicted value for the mixing temperature; The predicted temperature value of the mechanism derived from the mechanism model; This is the residual correction value; The weight of the predicted temperature value for the mechanism.

[0009] In one possible implementation, the initial weights of the predicted mechanism temperature and the corrected residual value are different for different reaction stages in the reactor. The reaction stages include etherification stage I, etherification stage II, and boiling stage; the weighted sum of the predicted mechanism temperature and the corrected residual value is 1; the weight of the predicted mechanism temperature corresponding to etherification stage I is greater than the weight of the corrected residual value.

[0010] Optionally, the weight of the predicted temperature values ​​for the etherification stage I mechanism. =0.6; Weight of the predicted temperature value for the etherification stage II mechanism =0.5; the weight range of the predicted temperature value for the boiling stage mechanism is 0.4-0.6.

[0011] In one possible implementation, adjusting the weights of the predicted mechanism temperature and the corrected residual value based on the accumulated count includes: The weights of the predicted mechanism temperature are updated according to the following formula:

[0012] in, The weights for the updated predicted temperature values ​​based on the mechanism are as follows: The weights for updating the temperature prediction values ​​based on the previously mentioned mechanism; This refers to the number of times the residual value exceeds a set residual threshold within a given time period.

[0013] In one possible implementation, the heat balance equation includes the heat input from steam heating, heat loss, heat release from chemical reactions, and latent heat of vaporization at the boiling point; the real-time operating parameters include steam flow rate, reactor temperature, ambient temperature, stirring frequency, and reactor pressure. The heat input for steam heating is calculated based on steam flow rate, steam enthalpy, and condensate enthalpy; the heat loss is calculated based on the temperature inside the reactor and the ambient temperature; and the latent heat of vaporization at the boiling point stage is calculated based on the pressure inside the reactor and the evaporation rate.

[0014] The steam enthalpy and condensate enthalpy are obtained by referring to a saturated steam table. The steam flow rate is measured by a flow meter, the temperature inside the reactor and the ambient temperature are measured by a resistance temperature detector (RTD) sensor, and the pressure inside the reactor is measured by a pressure transmitter.

[0015] In one possible implementation, the heat balance equation is:

[0016] in, Specific heat capacity of the material (KJ / kg); The mass of the material inside the vessel (kg); Input heat (kJ / s) for steam heating; Heat loss (kJ / s); For chemical reactions that are exothermic (endothermic reactions are negative values, kJ / s); The latent heat of vaporization at the boiling point stage (kJ / s).

[0017] Steam heating input heat The formula is:

[0018] in, Steam flow rate (kg / s); This is the steam enthalpy (kJ / kg, refer to the saturated steam table); This is the enthalpy value of condensate (kJ / kg).

[0019] Heat loss The formula is:

[0020] in, The overall heat transfer coefficient of the vessel body (kW / (m²·℃)); The surface area of ​​the vessel is (m²). The temperature inside the reactor; The ambient temperature.

[0021] Heat of chemical reaction The formula is:

[0022] in, The reaction rate constant is 1 / s; The activation energy of the reaction is (kJ / mol). The gas constant is 8.314 J / (mol·K). The molar enthalpy change of the reaction (kJ / mol); The activation energy of the reaction is (kJ / mol). The reaction volume is (m³). Latent heat of vaporization at boiling point The formula is:

[0023] in, The latent heat of vaporization (kJ / kg) is obtained by looking up a table based on the pressure inside the vessel. This table is derived from experimental data obtained by measuring the latent heat of vaporization of the solvent at characteristic pressure points during the reaction process of the mixed solvent in an offline experiment. The evaporation rate is (kg / s, estimated using the material balance equation).

[0024] In one possible implementation, determining the DCS control scheme based on the mixed temperature prediction value includes: The predicted mixed temperature is input into the pressure-temperature coupled latent heat calculation model to obtain the pressure value; The steam valve opening adjustment value and the stirring frequency adjustment value are determined based on the pressure value.

[0025] In one possible implementation, after determining the DCS control scheme based on the mixed temperature prediction value, the method further includes: The steam valve opening adjustment value and the stirring frequency adjustment value are sent to the actuator, which then performs the control operations of the steam valve and the stirring motor.

[0026] Secondly, embodiments of the present invention provide a device for controlling the temperature of a reaction vessel, used in a reaction vessel for the synthesis of sodium carboxymethyl cellulose, the device comprising: The mechanism temperature prediction calculation module is used to obtain real-time operating parameters input into the mechanism model to obtain the mechanism temperature prediction value; wherein, the mechanism model is constructed based on the heat balance equation of the reactor; the mechanism temperature prediction value is obtained by solving the heat balance equation using the fourth-order Runge-Kutta method; The residual processing module is used to calculate the residual value based on the predicted temperature value of the mechanism and the actual temperature measured by the sensor, and to standardize the residual value. The residual correction value calculation module is used to obtain the standardized residual values ​​within the current time and the previously set time window, construct a standardized residual value sequence, input the standardized residual value sequence into the TCN model, and output the residual correction value. The mixed temperature prediction calculation module is used to fuse the mechanism temperature prediction value and the residual correction value through a dynamic weight adjustment mechanism to calculate the mixed temperature prediction value, and determine the DCS control scheme based on the mixed temperature prediction value.

[0027] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect or any possible implementation thereof.

[0028] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect or any possible implementation thereof.

[0029] In this embodiment of the invention, a dynamic weight adjustment mechanism is used to fuse the mechanistic temperature prediction value and the residual correction value to calculate the mixed temperature prediction value. This mechanism adjusts the weights of the two by accumulating the number of times the residual value is greater than a set residual threshold within a set time period. This allows the mixed temperature prediction value to rely more on the physical laws of the mechanistic model when the mechanistic model is reliable, and to rely more on the residual correction value to compensate for unmodeled factors when the mechanistic model has large errors. This improves the adaptability of temperature prediction to complex dynamic processes and increases prediction accuracy. Attached Figure Description

[0030] Figure 1 This is an application scenario diagram of the reaction vessel temperature control method provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the implementation of a reactor temperature control method according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a model integration architecture provided in an embodiment of the present invention; Figure 4 This is a control closed-loop flowchart provided in a specific embodiment of the present invention; Figure 5 A schematic diagram of a reaction vessel temperature control device according to an embodiment of the present invention is shown; Figure 6 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0031] Existing methods fail to balance the heating rate with the overheating safety threshold, leading to long production cycles or high risks. This invention aims to propose a solution combining mechanistic models and deep learning to address difficulties in daily production. The mechanistic model (fourth-order Runge-Kutta method for solving the heat balance equation) provides the physical framework, while the TCN residual model (hereinafter referred to as the TCN model) dynamically compensates for unmodeled errors (stirring power fluctuations, raw material moisture content, etc.). An adaptive weighting mechanism improves control accuracy and heating rate (e.g., initial...). =0.6, adjust when the error exceeds ±0.5℃, control accuracy is controlled within ±0.3℃, and heating rate is improved.

[0032] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] Figure 1 This diagram illustrates an application scenario of a reactor temperature control method provided in an embodiment of the present invention. Figure 1 As shown, the control terminal acquires the operating data of the reactor, generates a control scheme for the reactor based on the operating data using a combination of mechanism model and deep learning, and sends the control scheme to the DCS system, which then controls the specific parameters of the reactor.

[0034] Figure 2 This diagram illustrates an application scenario of a reactor temperature control method provided in an embodiment of the present invention. Figure 2 As shown, it includes the following steps: S201, real-time operating parameters are input into the mechanism model to obtain the predicted mechanism temperature; the mechanism model is constructed based on the heat balance equation of the reactor; the predicted mechanism temperature is obtained by solving the heat balance equation using the fourth-order Runge-Kutta method.

[0035] The execution entities of the various embodiments of this application are as follows: Figure 1 The control terminal shown can be a server, processor, microprocessor, or other device with data processing capabilities. In actual implementation, the specific implementation method of the execution entity can be selected according to actual needs. This embodiment does not impose any particular restrictions on this, as long as the device has data processing capabilities. For ease of understanding, the control terminal is used as the execution entity in the following embodiments.

[0036] In the specific implementation process, a mechanistic model of the reactor heating process was established. Based on the energy conservation equation, a heat balance equation for the reactor was established, considering various energy inputs and outputs in the system, as well as the influence of chemical reactions and phase transitions on energy. The heat balance equation of the reactor was constructed by integrating the heat input from steam heating, heat loss, heat release from chemical reactions, and latent heat of vaporization at the boiling point. The predicted value of the mechanistic temperature was obtained by numerically solving the heat balance equation.

[0037] S202, calculate the residual value based on the predicted temperature of the mechanism and the actual temperature measured by the sensor, and then standardize the residual value.

[0038] The residual value represents the dynamic error not covered by the mechanistic model, including unmodeled disturbances (such as fluctuations in stirring power and changes in raw material moisture content), and its formula is:

[0039] in, The actual temperature measured by the sensor; This is the predicted temperature value for the mechanism.

[0040] The residual values ​​are standardized to eliminate the influence of dimensions, making the residuals comparable under different operating conditions. The formula for standardizing the residual values ​​is:

[0041] in, The historical residual mean (unit: °C) is calculated using data from previously set batches. The standard deviation of historical residuals (unit: °C) reflects the range of residual fluctuations.

[0042] Optionally, the batch size can be set to 100 times.

[0043] S203: Obtain the standardized residual values ​​within the current time and the previously set time window, construct a standardized residual value sequence, input the standardized residual value sequence into the TCN model, and output the residual correction value.

[0044] Compared to purely data-driven models, residual models typically require fewer computational resources because mechanistic models provide an approximate solution, which deep learning models simply tune to adapt to specific error patterns. This approach enhances the model's flexibility and robustness in different environments, offering an effective technical route for handling complex real-world problems.

[0045] Set the standardized residual value within the time window as follows:

[0046] The time window length is set to ;in, To set the first time window The standardized residual values.

[0047] Optionally, the time window can be set to 5-30 minutes. Alternatively, the time window can be set to 5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes, or 30 minutes. Taking a 10-minute time window as an example, the formula for the standardized residual sequence of the 10 minutes preceding the current time is: .

[0048] S204 uses a dynamic weight adjustment mechanism to integrate the predicted temperature value and the residual correction value, calculates the mixed temperature prediction value, and determines the DCS control scheme based on the mixed temperature prediction value.

[0049] Integrating numerical mechanism models with deep learning in terms of structure or process is an important method to leverage the advantages of both models. Specifically, the TCN model, as a deep learning model, incorporates the data flow of the numerical mechanism model for feature enhancement. Feature enhancement is achieved through bidirectional interaction between physical feature injection and the data flow, specifically including: First, the key physical parameters output by the mechanism model (the predicted mechanism temperature calculated based on the heat input of steam heating, heat loss, exothermic chemical reaction, and latent heat of vaporization at the boiling point) are fused with real-time sensor data to form the TCN model input.

[0050] Secondly, the feature weights are dynamically switched according to the reaction process, and the predicted temperature value and residual correction value are integrated through a dynamic weight adjustment mechanism to dynamically adjust the input weights of the TCN model.

[0051] Finally, during the actual implementation process, adversarial examples are generated every 30 minutes using the residual data of the mechanistic model to enhance the generalization ability of the TCN model in reverse.

[0052] Based on the aforementioned method for controlling the reactor temperature, in the preparation of sodium carboxymethyl cellulose, cellulose raw material is added to an organic solvent to obtain a reaction system. An alkalizing agent and an etherifying agent are then added sequentially to the reaction system to obtain carboxymethyl cellulose. The liquid-to-solid ratio between the solvent and the cellulose raw material in the reaction system determines the reaction rate and heat. When the numerical mechanism model cannot capture all complex nonlinear relationships, residual modeling uses deep learning (TCN model) to focus on compensating for the steady-state error of the mechanism model (±1.8℃→±0.5℃). Standardized time-series residuals are input, and a limited compensation value (±2℃) is output, achieving anti-overfitting with 60% data requirement and adapting to scenarios with liquid-to-solid ratio fluctuations (1:2-1:5) and moisture content changes (±5%).

[0053] In this embodiment, a dynamic weight adjustment mechanism is used to fuse the mechanistic temperature prediction value and the residual correction value to calculate the mixed temperature prediction value. This mechanism adjusts the weights of the two by accumulating the number of times the residual value is greater than a set residual threshold within a set time period. This allows the mixed temperature prediction value to rely more on the physical laws when the mechanistic model is reliable, and to rely more on the residual correction value to compensate for unmodeled factors when the mechanistic model has large errors. This improves the adaptability of temperature prediction to complex dynamic processes and increases prediction accuracy.

[0054] In one possible implementation, a dynamic weighting mechanism is used to fuse the mechanism temperature prediction and the residual correction value to calculate the mixed temperature prediction, including: The number of times the residual value exceeds a set residual threshold within a set time period is cumulatively counted. The weights of the mechanism temperature prediction and residual correction values ​​are adjusted based on the cumulative count; where the weights of the mechanism temperature prediction and residual correction values ​​are summed to 1.

[0055] In this embodiment, the calculation formula for the mixed temperature prediction value clarifies the fusion ratio between the mechanism temperature prediction value and the residual correction value. This preserves the physical interpretability of the mechanism model based on the heat balance equation, while also compensating for the errors not covered by the mechanism model through the residual correction value. This allows the mixed temperature prediction value to combine the advantages of both and more accurately reflect the actual temperature trend of the reactor in a nonlinear and time-varying reaction environment.

[0056] In one possible implementation, the formula for calculating the mixed temperature prediction is as follows:

[0057] in, This is a predicted value for the mixing temperature; The predicted mechanism temperature is derived from the mechanism model. This is the residual correction value; The weights for the predicted mechanism temperature values.

[0058] In one possible implementation, the initial weights of the predicted mechanism temperature and the corrected residual value are different for different reaction stages in the reactor. The reaction stages include etherification stage I, etherification stage II, and boiling stage; the weights of the predicted mechanism temperature and the corrected residual value are summed to 1; the weight of the predicted mechanism temperature corresponding to etherification stage I is greater than the weight of the corrected residual value.

[0059] Optionally, the weight of the predicted temperature values ​​for the etherification stage I mechanism. =0.6; Weight of the predicted temperature value for the etherification stage II mechanism =0.5; the weight range for the predicted temperature of the boiling stage mechanism is 0.4-0.6, i.e. .

[0060] Among them, the etherification stage I reaction is relatively stable and the mechanism model is more reliable, therefore a weight is set. =0.6 The mechanism model is the main factor in reducing unnecessary corrections; the exothermic nonlinearity of the etherification II stage reaction is enhanced, and the mechanism model and residual correction value are equally important; the temperature inside the vessel is close to the boiling point during the boiling stage, the latent heat of vaporization changes abruptly (the thermal effect fluctuates violently), and the pressure inside the vessel and the evaporation rate are easily affected by the reflux of condensate and the fluctuation of steam pressure. The mechanism model has a large estimation error for the phase change process, but a certain mechanism weight still needs to be retained to avoid the residual correction deviating too much from the physical law. Therefore, the weight of the predicted mechanism temperature value is dynamically adjusted based on the pressure to adapt to the reaction characteristics inside the vessel.

[0061] For example, when the operating conditions are stable, such as pressure fluctuations within the vessel ≤ ±0.01MPa and a stable evaporation rate, ω=0.6 is used. The mechanistic model still dominates, ensuring that the predicted value conforms to the coupling law of temperature and pressure (the temperature inside the vessel approaches the boiling point during the boiling stage, and the pressure inside the vessel will suddenly increase), avoiding frequent temperature fluctuations caused by residual correction. When the operating conditions are significantly disturbed, such as pressure fluctuations ≥ ±0.02MPa and unstable condensate reflux, ω=0.4 is used to strengthen the compensation effect of the residual correction value, specifically offsetting the mechanistic error caused by the sudden change in latent heat of vaporization, and reducing the error of the predicted mixing temperature.

[0062] In this embodiment, different initial weights are set for the predicted mechanism temperature and the residual correction value for different reaction stages in the reactor to adapt to the differences in reaction characteristics at each stage. By allocating the initial weights, adjustment space can be reserved in advance for the residual correction, thereby improving the pertinence and accuracy of temperature prediction at each stage.

[0063] In one possible implementation, the weights of the predicted mechanism temperature and the residual correction value are adjusted based on the cumulative count, including: The weights of the predicted mechanism temperature values ​​are updated according to the following formula:

[0064] in, To update the weights of the predicted mechanism temperature, The weights for updating the predicted mechanism temperature values; This refers to the number of times the residual value exceeds a set residual threshold within a given time period.

[0065] Optionally, The value of n is 3, 4, or 5. When n takes the value of 3, 4, or 5, a dynamic balance can be achieved between prediction accuracy and system stability. The value of n can be adjusted according to the specific application scenario.

[0066] Among these parameters, when n is 3, the response speed to residual anomalies is the fastest, quickly triggering weight decay mechanisms (such as ω adjustment), making it suitable for scenarios with frequent operating condition fluctuations (such as the start-up and shutdown phases of industrial equipment). When n is 4, it balances response speed and anti-interference capability, capturing both continuous residual offsets and filtering out brief abnormal fluctuations, making it suitable for most systems operating in steady or quasi-steady state (such as continuous production processes). When n is 5, the judgment of residual anomalies is more stringent, requiring 5 consecutive residual exceedances before triggering correction, effectively avoiding noise interference, but the response speed is relatively slow, making it suitable for scenarios with stable operating conditions and high requirements for the stability of prediction results (such as temperature control systems for precision instruments).

[0067] In the specific implementation process, the initial weight of the predicted mechanism temperature corresponding to the etherification stage I is 0.6, and the calculation formula for the initial predicted mixing temperature is as follows: .

[0068] During the reaction process, the initial weights are updated based on the number of times the residual value exceeds a set residual threshold. Optionally, the residual threshold is set to 0.5. -1.5 Optionally, the residual threshold can be set to 0.5. 1 Or 1.5 When the residual threshold is set to 1 At that time, If the error exceeds the tolerance, the corresponding count result is incremented by 1.

[0069] In this embodiment, the weights of the predicted mechanistic temperature values ​​are updated based on the number of times the residual value exceeds a set residual threshold within a set time period. This ensures that when the mechanistic model consistently exhibits large errors, its weight dynamically decreases as the number of errors increases, while the weight of the residual correction value increases accordingly. This mechanism can respond in real-time to changes in the reliability of the mechanistic model, preventing the prediction bias from expanding due to the accumulation of errors in the mechanistic model. It also enhances the hybrid model's resistance to unmodeled disturbances and improves the system's robustness.

[0070] In one possible implementation, the heat balance equation includes the heat input from steam heating, heat loss, heat release from chemical reactions, and latent heat of vaporization at the boiling point; real-time operating parameters include steam flow rate, reactor temperature, ambient temperature, stirring frequency, and reactor pressure. The heat input for steam heating is calculated based on steam flow rate, steam enthalpy, and condensate enthalpy; the heat loss is calculated based on the temperature inside the reactor and the ambient temperature; and the latent heat of vaporization at the boiling point stage is calculated based on the pressure inside the reactor and the evaporation rate.

[0071] The steam enthalpy and condensate enthalpy were obtained by referring to the saturated steam table. The steam flow rate was measured by a flow meter, the temperature inside the reactor and the ambient temperature were measured by a resistance temperature detector (RTD) sensor, and the pressure inside the reactor was measured by a pressure transmitter.

[0072] In one possible implementation, the heat balance equation is:

[0073] in, Specific heat capacity of the material (KJ / kg); The mass of the material inside the vessel (kg); Input heat (kJ / s) for steam heating; Heat loss (kJ / s); For chemical reactions that are exothermic (endothermic reactions are negative values, kJ / s); The latent heat of vaporization at the boiling point stage (kJ / s).

[0074] Steam heating input heat The formula is:

[0075] in, Steam flow rate (kg / s); This is the steam enthalpy (kJ / kg, refer to the saturated steam table); This is the enthalpy value of condensate (kJ / kg).

[0076] Heat loss The formula is:

[0077] in, The overall heat transfer coefficient of the vessel body (kW / (m²·℃)); The surface area of ​​the vessel is (m²). The temperature inside the reactor; The ambient temperature is used. The overall heat transfer coefficient of the reactor body is calculated during unloaded or water-tested operation by applying a known heating power and measuring the steady-state temperature difference between the inside and outside of the reactor. The calculation formula is:

[0078] in, The heating power is known.

[0079] Heat of chemical reaction The formula is:

[0080] in, The reaction rate constant is 1 / s; The activation energy of the reaction is (kJ / mol). The gas constant is 8.314 J / (mol·K). The molar enthalpy change of the reaction (kJ / mol); The activation energy of the reaction is (kJ / mol). The reaction volume is (m³). Latent heat of vaporization at boiling point The formula is:

[0081] in, The latent heat of vaporization (kJ / kg) is obtained by looking up a table based on the pressure inside the vessel. This table is derived from experimental data obtained by measuring the latent heat of vaporization of the solvent at characteristic pressure points during the reaction process of the mixed solvent in an offline experiment. The evaporation rate is (kg / s, estimated using the material balance equation).

[0082] The process of obtaining the predicted mechanism temperature through numerically solving the heat balance equation using the mechanism model is as follows: 1. Discretization of differential equations: The continuous heat balance equation is converted into a discrete time step form, as shown in the formula below.

[0083] in, for Predicted temperature values ​​for the mechanism at any given time; for Predicted temperature values ​​for the mechanism at any given time; For the mass of the material inside the vessel; The specific heat capacity of the material; For time step.

[0084] 2. Numerical integration method: The fourth-order Runge-Kutta method (RK4) is used to iteratively calculate the differential equation, and the final output is a temperature sequence.

[0085] During the iteration process, the average slope over the entire time step is obtained by weighted averaging four local slope estimations, and then the temperature value is updated. For the first... At the nth time step, solve for the nth time step. Temperature at each time step.

[0086] In this embodiment, the heat balance equation encompasses the heat input from steam heating, heat loss, exothermic chemical reactions, and latent heat of vaporization at the boiling point. Real-time operating parameters include key parameters such as steam flow rate, reactor temperature, and ambient temperature. The calculation of each heat term is based on reliable data such as measured data (e.g., steam flow rate, temperature, and pressure) or saturated steam meters. This allows the mechanistic model to more comprehensively and accurately reflect the thermal dynamics of the reactor, providing a solid physical basis for predicting mechanistic temperatures and reducing prediction errors caused by incomplete modeling of the mechanistic model itself.

[0087] In one possible implementation, the DCS control scheme is determined based on the mixed temperature prediction values, including: The predicted mixed temperature is input into the pressure-temperature coupled latent heat calculation model to obtain the pressure value; The steam valve opening adjustment value and the stirring frequency adjustment value are determined based on the pressure value.

[0088] The formula for calculating the pressure value is as follows:

[0089] in, The saturated vapor pressure, determined based on the predicted mixing temperature, is calculated using a pre-fitted offline saturated vapor pressure-temperature relationship curve. This is the dynamic pressure compensation term caused by the latent heat of vaporization. = Based on real-time monitoring of evaporation rate and latent heat of vaporization The latent heat of vaporization was calculated and obtained by looking up a table using the pressure inside the vessel. This is the pressure compensation coefficient.

[0090] In this embodiment, when determining the DCS control scheme based on the predicted mixing temperature, the pressure value is obtained by inputting it into a pressure-temperature coupled latent heat calculation model. Based on this, the steam valve opening adjustment value and the stirring frequency adjustment value are determined. This can associate the temperature prediction with key parameters such as the pressure inside the reactor and the latent heat of vaporization, avoiding the "false steady state" problem caused by phase change. This makes the control scheme more in line with the actual thermal effect of the reaction, improves the control accuracy of the DCS system for complex phase change processes, and adapts to nonlinear temperature control scenarios.

[0091] In one possible implementation, after determining the DCS control scheme based on the mixed temperature prediction values, the following is also included: The steam valve opening adjustment value and the stirring frequency adjustment value are sent to the actuator, which then performs the control operations of the steam valve and the stirring motor.

[0092] like Figure 3 This is a schematic diagram of a model integration architecture provided in an embodiment of the present invention, combined with... Figure 3The control-end data acquisition layer acquires real-time data input from the sensors and inputs the data into the mechanistic model and the TCN residual model respectively. (Specifically, the measured temperature from the sensors is input into the TCN residual model, and the parameters required for calculation in the heat balance equation are input into the mechanistic model.) The calculation results from the mechanistic model and the TCN residual model are simultaneously input into the hybrid model to calculate the predicted hybrid temperature value. Based on the temperature prediction value, the DCS (including actuators and frequency converters, etc.) control scheme is determined, and control signals are output for the actuators and frequency converters, etc.

[0093] The actuators include steam valve actuators, stirring motor actuators, etc.

[0094] In this embodiment, after determining the DCS control scheme, the steam valve opening adjustment value and the stirring frequency adjustment value are sent to the actuator to perform the control operation, forming a complete closed loop from temperature prediction to control command execution. This process ensures that the control scheme generated based on the mixed temperature prediction value can be promptly transformed into actual adjustment actions, quickly responding to changes in reactor temperature, maintaining reactor temperature stability, and ensuring the continuity and stability of the sodium carboxymethyl cellulose production process.

[0095] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0096] Figure 4 This is a control closed-loop flowchart provided in a specific embodiment of the present invention, as follows: Figure 4 As shown, the loading process parameters and the degree of substitution of sodium carboxymethyl cellulose are... Safety threshold The model is initialized, including setting the mechanistic model parameters. The TCN model is loaded with historical weights. Real-time operating data of the reactor (such as internal temperature and steam pressure) is then collected. Based on the collected real-time data, the predicted mechanistic temperature and residual correction values ​​are calculated for both the mechanistic model and the TCN model. Subsequently, the predicted mechanistic temperature and residual correction values ​​are input into the formula for calculating the mixed temperature prediction, and the weights in the formula are adjusted in real time to calculate the predicted mixed temperature. Afterward, the steam valve opening and stirring frequency are adjusted according to the predicted mixed temperature. The actuators are then controlled according to the adjustment scheme to achieve control of the steam valve and stirring motor. Figure 4 The specific parameters mentioned (such as DS=1.0±0.05; Tmax=120℃; Cp=2.1kJ / (kg℃); T_in-vessel: 85℃ (10 seconds / cycle), etc.) are for illustrative purposes only. After adjusting the reactor's operating status based on the predicted mixing temperature, closed-loop feedback control is required to monitor the reactor's operating status. Specifically, residual values ​​are checked for exceeding tolerances. When the number of consecutive deviations exceeds a threshold, it indicates large temperature fluctuations within the reactor, necessitating adaptive adjustments to the TCN weights. Otherwise, actuator control data and reactor data are recorded, and the TCN weights are periodically updated based on online data through self-learning to improve reactor control accuracy.

[0097] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0098] Figure 5 A schematic diagram of a reactor temperature control device according to an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below: like Figure 5 As shown, the reactor temperature control device 5 includes: The mechanism temperature prediction calculation module 501 is used to obtain real-time operating parameters and input them into the mechanism model to obtain the mechanism temperature prediction value. The mechanism model is constructed based on the heat balance equation of the reactor. The mechanism temperature prediction value is obtained by solving the heat balance equation using the fourth-order Runge-Kutta method. The residual processing module 502 is used to calculate the residual value based on the predicted temperature of the mechanism and the actual temperature measured by the sensor, and to standardize the residual value. The residual correction value calculation module 503 is used to obtain the standardized residual values ​​within the current time and the previously set time window, construct a standardized residual value sequence, input the standardized residual value sequence into the TCN model, and output the residual correction value. The mixed temperature prediction calculation module 504 is used to calculate the mixed temperature prediction value by integrating the mechanism temperature prediction value and the residual correction value through a dynamic weight adjustment mechanism, and to determine the DCS control scheme based on the mixed temperature prediction value.

[0099] In one possible implementation, the hybrid temperature prediction calculation module 504 is specifically used for: The number of times the residual value exceeds a set residual threshold within a set time period is cumulatively counted. The weights of the mechanism temperature prediction and residual correction values ​​are adjusted based on the cumulative count; where the weights of the mechanism temperature prediction and residual correction values ​​are summed to 1.

[0100] In one possible implementation, the hybrid temperature prediction calculation module 504 is specifically used to update the weights of the mechanism temperature prediction values ​​according to the following formula:

[0101] in, To update the weights of the predicted mechanism temperature, The weights for updating the predicted mechanism temperature values; This refers to the number of times the residual value exceeds a set residual threshold within a given time period.

[0102] In one possible implementation, the hybrid temperature prediction calculation module 504 is specifically used for: The predicted mixed temperature is input into the pressure-temperature coupled latent heat calculation model to obtain the pressure value; The steam valve opening adjustment value and the stirring frequency adjustment value are determined based on the pressure value.

[0103] In one possible implementation, a sending module is also included, which, after determining the DCS control scheme based on the predicted mixing temperature, sends the steam valve opening adjustment value and the stirring frequency adjustment value to the actuator, so that the actuator can perform the control operations of the steam valve and the stirring motor.

[0104] In this embodiment, a dynamic weight adjustment mechanism is used to fuse the mechanistic temperature prediction value and the residual correction value to calculate the mixed temperature prediction value. This mechanism adjusts the weights of the two by accumulating the number of times the residual value is greater than a set residual threshold within a set time period. This allows the mixed temperature prediction value to rely more on the physical laws when the mechanistic model is reliable, and to rely more on the residual correction value to compensate for unmodeled factors when the mechanistic model has large errors. This improves the adaptability of temperature prediction to complex dynamic processes and increases prediction accuracy.

[0105] Figure 6 This is a schematic diagram of an electronic device provided according to an embodiment of the present invention. For example... Figure 6 As shown, the electronic device 6 of this embodiment includes a processor 60 and a memory 61. The memory 61 stores a computer program 62. When the processor 60 executes the computer program 62, it implements the steps in the various method embodiments described above. Alternatively, when the processor 60 executes the computer program 62, it implements the functions of each module / unit in the various device embodiments described above.

[0106] For example, computer program 62 may be divided into one or more modules / units, which are stored in memory 61 and executed by processor 60 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 62 in electronic device 6.

[0107] Electronic device 6 may include, but is not limited to, processor 60 and memory 61. Those skilled in the art will understand that... Figure 6This is merely an example of electronic device 6 and does not constitute a limitation on electronic device 6. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 6 may also include input / output devices, network access devices, buses, etc.

[0108] The processor 60 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0109] The memory 61 can be an internal storage unit of the electronic device 6, such as a hard disk or RAM. The memory 61 can also be an external storage device of the electronic device 6, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 61 can include both internal and external storage units of the electronic device 6. The memory 61 is used to store the computer program 62 and other programs and data required by the electronic device 6. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0110] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0111] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0112] This invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0113] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0114] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0115] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for controlling the temperature of a reaction vessel, used in a reaction vessel for the synthesis of sodium carboxymethyl cellulose, characterized in that, The control method includes: Real-time operating parameters are input into the mechanism model to obtain the predicted mechanism temperature; wherein, the mechanism model is constructed based on the heat balance equation of the reactor; the predicted mechanism temperature is obtained by solving the heat balance equation using the fourth-order Runge-Kutta method; The residual value is calculated based on the predicted temperature value and the actual temperature measured by the sensor, and the residual value is then standardized. Obtain the standardized residual values ​​within the current time and the previously set time window, construct a standardized residual value sequence, and input the standardized residual value sequence into the temporal convolutional network (TCN) model to output the residual correction value; The predicted temperature value is fused with the residual correction value through a dynamic weight adjustment mechanism to calculate the mixed temperature prediction value, and the DCS control scheme is determined based on the mixed temperature prediction value.

2. The method for controlling the temperature of the reactor according to claim 1, characterized in that, The step of fusing the predicted mechanistic temperature value and the corrected residual value through a dynamic weight adjustment mechanism to calculate the mixed temperature prediction value includes: The number of times the residual value exceeds a set residual threshold within a set time period is cumulatively counted. The weights of the predicted mechanism temperature and the corrected residual value are adjusted based on the cumulative count; wherein the sum of the weights of the predicted mechanism temperature and the corrected residual value is 1.

3. The method for controlling the temperature of the reactor according to claim 2, characterized in that, The formula for calculating the predicted mixing temperature is as follows: in, The predicted value for the mixing temperature; The predicted temperature value of the mechanism derived from the mechanism model; This is the residual correction value; The weight of the predicted temperature value for the mechanism.

4. The method for controlling the temperature of the reactor according to claim 2 or 3, characterized in that, The initial weights of the predicted mechanism temperature and the corrected residual value are different for different reaction stages in the reactor. The reaction stages include etherification stage I, etherification stage II, and boiling stage; the weighted sum of the predicted mechanism temperature and the corrected residual value is 1; the weight of the predicted mechanism temperature corresponding to etherification stage I is greater than the weight of the corrected residual value.

5. The method for controlling the temperature of the reactor according to claim 1, characterized in that, The heat balance equation includes the heat input from steam heating, heat loss, heat release from chemical reactions, and latent heat of vaporization at the boiling point; the real-time operating parameters include steam flow rate, reactor temperature, ambient temperature, stirring frequency, and reactor pressure. The heat input for steam heating is calculated based on steam flow rate, steam enthalpy, and condensate enthalpy; the heat loss is calculated based on the temperature inside the reactor and the ambient temperature; and the latent heat of vaporization at the boiling point stage is calculated based on the pressure inside the reactor and the evaporation rate.

6. The method for controlling the temperature of the reactor according to claim 1, characterized in that, The step of determining the DCS control scheme based on the predicted mixed temperature includes: The predicted mixed temperature is input into the pressure-temperature coupled latent heat calculation model to obtain the pressure value; The steam valve opening adjustment value and the stirring frequency adjustment value are determined based on the pressure value.

7. The method for controlling the temperature of the reactor according to claim 6, characterized in that, After determining the DCS control scheme based on the predicted mixed temperature value, the method further includes: The steam valve opening adjustment value and the stirring frequency adjustment value are sent to the actuator, which then performs the control operations of the steam valve and the stirring motor.

8. A device for controlling the temperature of a reaction vessel, used in a reaction vessel for the synthesis of sodium carboxymethyl cellulose, characterized in that, The control device includes: The mechanism temperature prediction calculation module is used to obtain real-time operating parameters input into the mechanism model to obtain the mechanism temperature prediction value; wherein, the mechanism model is constructed based on the heat balance equation of the reactor; the mechanism temperature prediction value is obtained by solving the heat balance equation using the fourth-order Runge-Kutta method; The residual processing module is used to calculate the residual value based on the predicted temperature value of the mechanism and the actual temperature measured by the sensor, and to standardize the residual value. The residual correction value calculation module is used to obtain the standardized residual values ​​within the current time and the previously set time window, construct a standardized residual value sequence, input the standardized residual value sequence into the TCN model, and output the residual correction value. The mixed temperature prediction calculation module is used to fuse the mechanism temperature prediction value and the residual correction value through a dynamic weight adjustment mechanism to calculate the mixed temperature prediction value, and determine the DCS control scheme based on the mixed temperature prediction value.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images