A limonene rectification temperature control method based on an adaptive fuzzy neural network

Abstract

The present application relates to the field of device control, more particularly, the present application relates to a kind of based on adaptive fuzzy neural network's limonene rectification temperature control method, method includes: first, according to real-time sensor data calculation comprehensive performance index. Then, the dynamic characteristics of the device are identified in real time online, obtain the process dynamic parameters such as inertia parameter and sensitivity parameter changing with time. Then, based on dynamic parameter and the estimation of index noise level, the core parameters of extremum search control algorithm are adaptively set. Finally, the adaptive set extremum search control algorithm is executed to optimize to calculate initial preheating temperature set value, and combined with super relaxation control logic is modified, under the premise of ensuring that temperature does not cross the preset safety limit, output the final preheating temperature set value. The present application not only realizes the adaptive energy saving and consumption reduction of dehydration process, but also guarantees the absolute safety of production process through the hard constraint mechanism of output end.

Description

Technical Field

[0001] This invention relates to the field of equipment control. More specifically, this invention relates to a method for temperature control in limonene distillation based on an adaptive fuzzy neural network. Background Technology

[0002] The limonene distillation process has specific requirements for the precision of the bottom heating temperature control. Conventional distillation temperature control systems typically employ temperature control models based on proportional-integral-differential algorithms or conventional neural networks. This type of control logic primarily monitors the preset target process temperature and the actual feed temperature inside the distillation vessel, adjusting the opening of the reboiler steam regulating valve based on the temperature deviation between the two. Under initial equipment operation conditions and with no scaling at the heat exchange interface, the heat injected from the steam side can be transferred to the feed liquid inside the vessel with a stable heat transfer coefficient. This control method can meet the basic heating requirements of the process.

[0003] However, in actual industrial production, as the continuous operating time of distillation equipment increases, high-boiling-point organic impurities adhere to the reboiler tube bundle walls, forming a coke layer. This coke layer increases heat transfer resistance, creating a physical barrier between steam and feed liquid, causing the actual heat transfer coefficient of the reboiler to gradually decrease. Under this condition of impaired heat transfer, the heat energy released at the steam end cannot be fully absorbed by the feed liquid in the vessel, thus slowing down the rate of temperature rise of the feed liquid. This results in the actual feed liquid temperature collected by the conventional control system remaining below the target process temperature for an extended period.

[0004] Because conventional controllers lack quantitative assessment methods for changes in the underlying heat transfer state, relying solely on surface temperature differences for feedback adjustment, they easily misinterpret slow liquid temperature rise as insufficient overall system heating power. Based on this judgment, the controller continuously outputs commands to increase the steam valve opening. This control action causes heat energy that fails to penetrate the heat exchange interface to be trapped and accumulate within the reboiler's metal matrix and coke layer. As localized heat accumulates, the local physical temperature of the reboiler gradually rises, which can easily lead to high-temperature isomerization degradation of limonene in contact with this area, thereby reducing the purity of the target product and posing safety hazards to equipment operation. Summary of the Invention

[0005] To address the problem in existing technologies where scale buildup in the reboiler of a distillation vessel hinders heat transfer, conventional controllers relying solely on temperature difference regulation are prone to heat accumulation and high-temperature degradation of limonene. This invention proposes a limonene distillation temperature control method based on an adaptive fuzzy neural network, comprising the following steps: The absolute pressure of the steam pipeline is obtained by a pressure transmitter installed in the steam pipeline network, the valve opening feedback percentage is obtained by a valve positioner, and the actual temperature of the liquid in the distillation kettle is obtained by a temperature sensor. The system energy injection rate is calculated using the valve opening feedback percentage, the absolute pressure of the steam pipeline, and the equivalent enthalpy conversion coefficient. The sensible heat response rate is calculated using the time difference of the actual temperature of the liquid in the distillation vessel and the equivalent comprehensive heat capacity. The energy loss allocation value is determined based on the difference between the two. The bottom heat accumulation is obtained by subtracting the normal heat loss rate from the energy loss ratio and performing time-series accumulation. The bottom heat accumulation is then calculated using the negative exponential decay mapping relationship to determine the adaptive thermal resistance suppression operator used for pressure regulation deviation control. The theoretical valve opening percentage output by the built-in adaptive fuzzy neural network is limited and corrected based on the adaptive thermal resistance suppression operator to determine the target valve opening percentage. The target valve opening percentage is then converted into an analog electrical signal by a programmable logic controller and sent to the electric valve positioner of the steam regulating valve to drive the physical valve stem of the steam regulating valve to produce displacement.

[0006] This invention incorporates sensing data such as absolute steam pressure, valve opening, and feed liquid temperature to calculate the system energy injection rate on the steam side and the sensible heat response rate on the feed liquid side. Based on this, the change in thermal resistance is quantified as an energy loss value, which is further converted into the heat accumulation degree at the bottom of the vessel. An adaptive thermal resistance suppression operator is generated based on this, thereby reducing the amplitude of the theoretical output of a conventional neural network. This limits excessive steam input when heat transfer is impeded, avoiding the risks of localized heat accumulation and material degradation.

[0007] Preferably, the step of obtaining the absolute pressure of the steam pipeline through a pressure transmitter installed in the steam pipeline network, obtaining the valve opening feedback percentage through a valve positioner, and obtaining the actual temperature of the liquid in the distillation vessel through a temperature sensor includes: The communication interface of the central processing unit is used to synchronously retrieve the absolute pressure data of the steam pipeline transmitted by the pressure transmitter, the valve opening feedback percentage data transmitted by the valve positioner, and the electrical signals collected by the temperature sensor. The electrical signal is preprocessed using an anti-aliasing filtering algorithm and converted into the actual temperature of the liquid in the distillation vessel.

[0008] Preferably, the system energy injection rate is calculated using the valve opening feedback percentage, the absolute pressure of the steam pipeline, and the equivalent enthalpy conversion coefficient; the sensible heat response rate is calculated using the time difference between the actual temperature of the liquid in the distillation vessel and the equivalent comprehensive heat capacity; and the energy loss allocation value is determined based on the difference between the two, specifically satisfying the following relationship:

[0009]

[0010]

[0011] In the formula, Representing the System energy injection rate at each sampling time; Representing the The percentage of valve opening feedback at each sampling time; Representing the Absolute pressure of the steam pipeline at each sampling time; Represents the equivalent enthalpy conversion factor; Representing the Sensible thermal response rate at each sampling time; and Representing the first The and the first The actual temperature of the liquid in the distillation vessel at each sampling time; Represents a fixed sampling step size; Represents the equivalent overall heat capacity; Representing the Energy loss allocation at each sampling time.

[0012] This invention quantifies the instantaneous physical heat power pushed into the steam side heat exchange interface by combining the valve opening feedback percentage with the absolute pressure of the steam pipeline and converting it with the equivalent enthalpy conversion coefficient; at the same time, by using the time difference of the actual temperature of the liquid in the distillation vessel and the equivalent comprehensive heat capacity, the effective heat power actually absorbed and converted into the internal energy of the material at the liquid end per unit time is accurately calculated. Furthermore, by subtracting the system energy injection rate from the sensible heat response rate, the stagnant heat power that could not penetrate the interface due to the fouling of the reboiler tube wall was extracted, thus transforming the physical thermal resistance, which could not be directly measured, into a quantifiable energy loss value.

[0013] Preferably, the method for obtaining the equivalent comprehensive heat capacity includes: Extract data from multiple consecutive production batches of distillation vessels while they are in a state confirmed by testing to be free of scaling. In a single batch, record the start time of the stable injection of constant power and the corresponding actual temperature of the liquid in the distillation vessel, and record the end time when the liquid reaches the preset high temperature without boiling and the corresponding actual temperature of the liquid in the distillation vessel. At the same time, record the total time consumed in the start and end stages. Using constant power, total duration, and temperature difference at the start and end times, the heat capacity of a single batch is calculated based on a division relationship, and the equivalent comprehensive heat capacity is obtained by taking the arithmetic mean of the heat capacity values ​​of multiple batches.

[0014] This invention uses baseline data of a scale-free distillation vessel for calibration, eliminating the deviations caused by scaling or phase changes in materials during the later stages of operation in heat calculation, and providing accurate engineering parameters for the calculation of sensible heat response rate.

[0015] Preferably, the step of subtracting the normal heat loss rate from the energy loss ratio and then summing the results over time to obtain the bottom heat accumulation degree specifically satisfies the following relationship:

[0016] In the formula, Representing the Thermal congestion at the bottom of the vessel at each sampling time; Representing the Thermal congestion at the bottom of the vessel at each sampling time; Representing the Energy loss allocation at each sampling time; Represents the normal heat loss rate; Represents a fixed sampling step size; This represents the heat retention ratio coefficient.

[0017] This invention combines the normal heat loss law of the equipment with the heat retention ratio coefficient, and transforms the instantaneously calculated energy loss value into a cumulative parameter in the time dimension, which objectively reflects the dynamic accumulation of heat retained inside the reboiler.

[0018] Preferably, the calculation of the heat accumulation at the bottom of the vessel using the negative exponential decay mapping relationship determines the adaptive thermal resistance suppression operator used for the pressure adjustment deviation control, specifically satisfying the following relationship:

[0019] In the formula, Representing the An adaptive thermal resistance suppression operator for each sampling time; Representing the Thermal congestion at the bottom of the vessel at each sampling time; Represents the safe thermal capacity limit threshold; This represents a zero-valued constant.

[0020] The adaptive thermal resistance suppression operator constructed in this invention can maintain the adaptive thermal resistance suppression operator near the reference value when the accumulated heat accumulation at the bottom of the vessel is at a low level, without affecting the normal control of the neural network; when the heat accumulation gradually approaches the safe tolerance limit, the adaptive thermal resistance suppression operator exhibits a smooth nonlinear decay, thereby avoiding the violent opening and closing action of the steam valve caused by using a hard threshold cutoff, and reducing hardware damage.

[0021] Preferably, the method for obtaining the safe heat capacity limit threshold includes: The total physical mass of the reboiler tube bundle and the metal matrix of the distillation vessel bottom, as well as the corresponding nominal specific heat capacity of the metal, are obtained in advance. Based on the pre-set safe temperature rise limit of the heat exchange interface, the theoretical maximum heat capacity is calculated by multiplying the total physical mass, the nominal specific heat capacity of the metal, and the safe temperature rise limit. The theoretical maximum heat capacity is subtracted from the preset engineering safety margin to determine the safe heat capacity limit threshold.

[0022] Preferably, the method for obtaining the heat retention ratio includes: In the calibration batch of distillation kettles with known scaling conditions, the energy loss ratio at each sampling moment during the heating stage is subtracted from the normal heat loss rate and then accumulated over time to record the total net energy deficit accumulated during the heating stage. Forcefully close the steam regulating valve, record the temperature difference between the moment of closure and the point when the liquid temperature reaches its peak due to residual heat release, multiply the temperature difference by the equivalent comprehensive heat capacity, and calculate the total energy of sensible heat temperature rise. The ratio obtained by dividing the total energy of sensible heat temperature rise by the total energy of net deficit is determined as the heat retention ratio coefficient.

[0023] Preferably, the theoretical valve opening percentage output by the built-in adaptive fuzzy neural network is limited and corrected based on the adaptive thermal resistance suppression operator to determine the target valve opening percentage, specifically satisfying the following relationship:

[0024] In the formula, Representing the The target valve opening percentage at each sampling time; Representing the An adaptive thermal resistance suppression operator for each sampling time; Representing the The theoretical valve opening percentage at each sampling time; This represents the minimum percentage of valve opening in the engineering system.

[0025] Preferably, the method for obtaining the minimum engineering valve opening percentage includes: The steam regulating valve is manually closed in stages until the steam condensate is critically cut off, as observed by the flow meter or sight glass. The valve opening percentage at the moment before the critical cutoff is taken as the minimum engineering valve opening percentage.

[0026] The present invention has the following beneficial effects: While retaining the advantages of conventional adaptive fuzzy neural network models, a low-level accounting and intervention mechanism based on thermodynamic energy conservation was established. By independently calculating and comparing the energy difference between supply and demand, the bottom heat accumulation index, characterizing the degree of internal physical heat transfer obstruction, was accurately extracted. This index was then transformed into an adaptive thermal resistance suppression operator to weight and limit the theoretical output of the conventional adaptive fuzzy neural network model. This control architecture solves the problem of blindly overfeeding steam caused by traditional controllers relying solely on temperature difference feedback under complex operating conditions. It enables the system to maintain normal temperature tracking performance while also proactively sensing and adaptively suppressing the accumulation of implicit heat loads in the heat exchanger. This method cuts off the path of excessive heat energy to heat-sensitive materials at the control end, significantly improving the process safety and product purity of limonene distillation while ensuring the smooth operation of the steam network and actuator valves. Attached Figure Description

[0027] Figure 1 This is a flowchart of the steps of a method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network provided in an embodiment of the present invention; Figure 2 This is a dynamic response diagram of distillation temperature control provided in an embodiment of the present invention; Figure 3 This is a comparison diagram of the actuator opening instructions provided in the embodiments of the present invention. Detailed Implementation

[0028] Please see Figure 1 The diagram illustrates a flowchart of a method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network, as provided in Example 1. The method includes the following steps: S1: Obtain the absolute pressure of the steam pipeline through a pressure transmitter installed in the steam pipeline network, obtain the valve opening feedback percentage through a valve positioner, and obtain the actual temperature of the liquid in the distillation vessel through a temperature sensor.

[0029] Preferably, as an example, the absolute pressure of the steam pipeline is obtained through a pressure transmitter installed in the steam pipeline network, the valve opening feedback percentage is obtained through a valve positioner, and the actual temperature of the liquid in the distillation vessel is obtained through a temperature sensor, including: First, the central processing unit's communication interface synchronously retrieves the absolute pressure data of the steam pipeline transmitted by the pressure transmitter, the valve opening percentage data transmitted by the valve positioner, and the electrical signal collected by the temperature sensor.

[0030] Next, the electrical signal is preprocessed using an anti-aliasing filtering algorithm to convert it into the actual temperature of the liquid in the distillation vessel.

[0031] Understandably, when the limonene distillation vessel enters the initial stage of high-power heating, the steam valve has just opened. Since the steam releases heat through condensation within the reboiler tube bundle and still needs to undergo heat conduction through the tube walls, the actual temperature of the liquid inside the vessel, as detected by the temperature sensor, has not yet shown a significant rise. Under these conditions, if only the liquid temperature is relied upon for judgment, the controller may misinterpret the heating as insufficient due to a large temperature difference. However, by simultaneously introducing the absolute pressure of the steam pipeline and the valve opening percentage feedback, both the liquid temperature in the limonene distillation vessel and the control intensity at the control end can be simultaneously sensed.

[0032] S2: Calculate the system energy injection rate using the valve opening feedback percentage, the absolute pressure of the steam pipeline, and the equivalent enthalpy conversion coefficient. Calculate the sensible heat response rate using the time difference between the actual temperature of the liquid in the distillation vessel and the equivalent comprehensive heat capacity. Determine the energy loss allocation value based on the difference between the two.

[0033] It should be noted that the heat transfer during limonene distillation is not an ideal instantaneous conversion process. Under actual operating conditions, the heat released by the steam must penetrate the physical barrier of the reboiler tube wall to be transferred to the feed liquid, and the formation of scale further increases the heat transfer resistance of this barrier. If adjustments are made directly without specific analysis, the controller cannot accurately distinguish whether the heat energy is effectively transferred to the feed liquid or is trapped in the heat exchange path due to scale blockage. Therefore, the core purpose of this step is to decompose and quantify the energy flow direction in the heat exchange path, and to convert the physical thermal resistance, which is difficult to observe directly, into a quantifiable power difference by independently calculating the input power on the steam side and the actual absorbed power on the feed liquid side.

[0034] Preferably, as an example, the system energy injection rate is calculated using the valve opening feedback percentage, the absolute pressure of the steam pipeline, and the equivalent enthalpy conversion coefficient; the sensible heat response rate is calculated using the time difference between the actual temperature of the feed liquid in the distillation vessel and the equivalent comprehensive heat capacity; and the energy loss allocation value is determined based on the difference between the two, including:

[0035]

[0036]

[0037] In the formula, Represents the sampling time index; Representing the The system energy injection rate at each sampling time, measured in watts; Representing the The valve opening feedback percentage at each sampling time is the effective opening equivalent obtained by linearizing the inherent flow characteristic curve of the regulating valve at the factory, and it is dimensionless. Representing the The absolute pressure of the steam pipeline at each sampling time, in Pascals; The equivalent enthalpy conversion factor is expressed in watts per square kPa. Representing the The sensible thermal response rate at each sampling time, in watts; and Representing the The sampling time and the first The actual temperature of the liquid in the distillation vessel at each sampling time, in degrees Celsius; This represents the equivalent total heat capacity, measured in joules per degree Celsius. This represents a fixed sampling step size, measured in seconds. For example... Take 0.5 seconds. Representing the The energy loss value at each sampling time is expressed in watts.

[0038] The theoretical basis for constructing the above relationship is as follows: In fluid mechanics, the valve throttling flow rate relationship is... Multiply both sides of the valve throttling flow rate equation by the steam enthalpy value. ,get , Indicates mass flow rate. Indicates the valve flow coefficient. Indicates the unit conversion factor. This indicates the pressure difference before and after the valve. Indicates fluid density, This represents the flow characteristic function corresponding to the valve opening degree, where the valve throttling flow rate is multiplied by the steam enthalpy. Equal to thermal power, therefore Corresponding to the above system energy injection rate relationship In the distillation process, the pressure downstream of the steam regulating valve is mainly maintained by the condensate system, and the changes are relatively small, thus the pressure difference is small. relative to upstream absolute pressure Approximate, therefore This can be equated to the above system energy injection rate calculation formula. Valve opening degree and flow characteristics Typically, valve characteristics are nonlinear. To simplify calculations, engineering practices often linearize these characteristics, meaning the original opening signal is converted to linear values ​​by adjusting the characteristic curve built into the valve's positioner or controller. This is converted into an effective opening equivalent that is linearly related to the flow rate, and therefore Replace the valve throttling flow rate relationship effect, and Although it varies with pressure, the change is relatively slow, and in engineering, it can be calibrated under a typical working condition to obtain an approximate constant. This can be equated to the above system energy injection rate calculation formula. .

[0039] The existing thermodynamic sensible heat relationship is: Differentiating both sides of the relation yields , Indicates the amount of heat absorbed. Indicates quality, This represents the specific heat capacity at constant pressure. This represents the change in temperature. It can be seen that the derived existing thermodynamic sensible heat relationship is consistent with the aforementioned sensible heat response rate relationship, where... Corresponding to the above sensible heat response rate relationship , Corresponding to the above sensible heat response rate relationship .

[0040] Understandable, The term reflects the instantaneous kinetic energy pushed from the steam side to the heat exchange interface, which is converted via the equivalent enthalpy coefficient. The correction accurately reflects the power intensity actively pushed from the energy source to the interface; correspondingly, the temperature difference value... With sampling step size The ratio reflects the instantaneous evolution rate of the thermal state at the feed liquid end, and is further analyzed by the equivalent comprehensive heat capacity. The proportional adjustment precisely characterizes the effective power that is instantaneously captured and converted into the internal energy of the material at the load end. In the scenario where organic scale forms on the reboiler wall during the later stages of limonene distillation, the scale layer creates a high thermal resistance physical barrier, causing severe blockage of energy penetration at the interface. Even if the valve maintains a high opening, resulting in a higher energy injection rate... The value is relatively large, but the temperature rise at the liquid end is limited by the physical shielding effect of the scale layer, resulting in a low calculated sensible thermal response rate. The value is relatively small. In this case, the energy loss value determined by the difference between the two is... It will increase rapidly.

[0041] The above embodiments involve equivalent enthalpy conversion coefficients. and equivalent comprehensive heat capacity The method for obtaining the above parameters will be explained below.

[0042] The equivalent enthalpy conversion coefficient The method for obtaining the equivalent enthalpy conversion factor includes: during the cold commissioning phase of the limonene distillation kettle, introducing standard constant steam with a known absolute pressure into the reboiler and recording the standard absolute pressure value; measuring the mass of steam condensate generated per unit time, querying the corresponding latent heat of vaporization through the mass standard steam enthalpy table for thermodynamic conversion, multiplying the mass of condensate generated per unit time by the latent heat of vaporization to calculate the reference thermal power under this condition; finally, dividing the reference thermal power by the square root of the standard absolute pressure value and taking the average of the calculation results for multiple consecutive test cycles to obtain the equivalent enthalpy conversion factor.

[0043] The equivalent comprehensive heat capacity Methods for obtaining [the information] include: First, data from ten consecutive production batches of the distillation vessel were extracted after physical cleaning and endoscopic inspection confirmed the absence of organic scale. In each batch, the constant power value The moment when stable injection begins is recorded as the start time, and the actual temperature of the liquid in the distillation vessel at that moment is also recorded. To eliminate the interference of material boiling and vaporization on energy calculation, the time when the liquid reaches a preset high temperature but has not yet boiled is selected as the termination time, and the actual temperature of the liquid in the distillation vessel at that time is recorded. Simultaneously, record the total time consumed from the start time to the end time. .

[0044] Then, based on the equation Solve for the first The heat capacity values ​​of each batch are calculated, and the arithmetic mean of the heat capacity values ​​of ten batches is taken to obtain the equivalent comprehensive heat capacity.

[0045] S3: The bottom heat accumulation is obtained by subtracting the normal heat loss rate from the energy loss ratio and performing time-series accumulation. The bottom heat accumulation is then calculated using the negative exponential decay mapping relationship to determine the adaptive thermal resistance suppression operator used for pressure regulation deviation control.

[0046] It should be noted that the heat energy injected but not absorbed by the feed liquid does not disappear into thin air, but accumulates and remains in the metal matrix and scale layer of the heat exchanger. This accumulated heat energy will impose a hidden heat load on the system. If heating continues unchecked, the heat will accumulate and accumulate, which can easily lead to high-temperature degradation of limonene. Therefore, the core objective of this step is to construct a risk monitoring index based on the energy imbalance state, and based on this, to establish a forward-looking adaptive suppression and control mechanism, thereby providing a foundation for avoiding the aforementioned hidden dangers in the future.

[0047] Preferably, as an example, the bottom heat accumulation is obtained by subtracting the normal heat loss rate from the energy loss ratio and performing time-series accumulation. The bottom heat accumulation is then calculated using a negative exponential decay mapping relationship to determine the adaptive thermal resistance suppression operator used for pressure regulation deviation control, including:

[0048]

[0049] In the formula, Representing the The thermal congestion at the bottom of the vessel at each sampling time, measured in joules; Representing the The heat accumulation at the bottom of the vessel at each sampling time, in joules, wherein the heat accumulation at the bottom of the vessel at the initial time is 0; This represents the normal heat loss rate, and its unit is watts. The heat retention ratio coefficient is a dimensionless unit that reflects the proportion of heat energy that is actually retained in the scale layer on the reboiler tube wall and forms deposits due to the mismatch between supply and demand. Representing the The adaptive thermal resistance suppression operator at each sampling time is in dimensionless units. Represents the safe heat capacity limit threshold, measured in joules. It reflects the maximum heat accumulation at the bottom of the distillation vessel that the reboiler tube bundle and the metal matrix at the bottom of the distillation vessel can accommodate without causing limonene to deteriorate. To prevent zero small constants, This represents a fixed sampling step size, measured in seconds. Representing the The energy loss value at each sampling time is expressed in watts.

[0050] It is understandable that the energy loss ratio is... This reflects the absolute power deviation between supply and demand, and this difference is further reduced by the normal heat loss rate. The resulting subtraction combination term It accurately characterizes the net heat energy flow rate forcibly retained by the physical structure of the heat exchanger per unit time, i.e., the net stored power; and this net stored power is related to the sampling step size. and the heat retention ratio coefficient The product of these terms transforms the instantaneous power gap into the actual physical energy increment that remains within the sampling period.

[0051] If the reboiler is severely fouled, the feed liquid temperature rises slowly. If the control system opens the steam valves without restriction, the energy injection rate will be much greater than the sum of the feed liquid's sensible heat response rate and heat loss. In this case, the cumulative term Q in the formula... kIt will increase rapidly over time and gradually approach the safety threshold M. Under the negative exponential mapping relationship, the suppression operator λ k The weight of the control output will decrease rapidly from 1 to 0, thus automatically reducing the weight of the control output. This process demonstrates the algorithm's ability to trigger safety constraints in advance before the risk of thermal penetration occurs, avoiding the risk of high-temperature isomerization of limonene due to concentrated heat release.

[0052] The above examples involve normal heat loss rate. Safe thermal capacity limit threshold and heat retention ratio coefficient The method for obtaining the above parameters will be explained below.

[0053] Among them, the normal heat loss rate Methods for obtaining [the information] include: When the distillation vessel is unloaded and the target process temperature is maintained constant, the sensible heat rise of the liquid inside the vessel no longer occurs. The physical heat energy represented by the system energy injection rate calculated by the central processing unit is entirely used to offset the natural heat dissipation of the distillation vessel and its connecting pipelines to the external environment. The central processing unit records the system energy injection rate for multiple consecutive sampling periods during this constant temperature maintenance phase, and the arithmetic mean of the recorded data is taken to obtain the normal state heat loss rate.

[0054] The safe heat capacity limit threshold Methods for obtaining [the information] include: First, the total physical mass of the reboiler tube bundle and the metal matrix of the distillation vessel bottom, as well as the corresponding nominal specific heat capacity parameters of the metal, are obtained in advance and stored in the memory of the central processing unit; the critical temperature for the thermosensitive isomerization degradation of limonene is consulted, and the upper limit of the safe temperature rise of the heat exchange interface is set accordingly; the central processing unit calls the above parameters, multiplies the total physical mass, the nominal specific heat capacity of the metal, and the upper limit of the safe temperature rise, and calculates the theoretical maximum heat capacity. Next, the central processing unit subtracts a preset engineering safety margin from the theoretical maximum heat capacity to determine the safe heat capacity limit threshold.

[0055] The method for obtaining the heat retention ratio includes: In the calibration batch of distillation kettles with known scaling conditions, firstly, the central processing unit is used to subtract the normal heat loss rate from the energy loss value at each sampling moment during the heating stage, and the difference is accumulated over time to record the total net energy deficit accumulated during the heating stage. Subsequently, the programmable logic controller sends a signal to forcibly close the steam regulating valve on the steam pipeline, and the central processing unit records the actual temperature of the liquid in the distillation vessel at the moment of closure. Due to thermal inertia and the release of residual heat from the scale layer in the reboiler, the actual temperature of the liquid in the distillation vessel monitored by the temperature sensor will continue to rise for a period of time. The central processing unit records the actual temperature of the liquid in the distillation vessel when it reaches its peak during this period. Next, the central processing unit is used to calculate the temperature difference between the actual temperature of the liquid in the distillation vessel at the peak value and the actual temperature of the liquid in the distillation vessel at the moment of closing. The temperature difference is then multiplied by the equivalent comprehensive heat capacity to calculate the total energy of sensible heat temperature rise generated by the liquid continuing to absorb residual heat after the valve is closed. Finally, the central processing unit divides the total energy of the sensible heat temperature rise by the total energy of the net gap, and the resulting ratio is determined as the heat retention ratio coefficient.

[0056] S4: Based on the adaptive thermal resistance suppression operator, the theoretical valve opening percentage output by the built-in adaptive fuzzy neural network is limited and corrected to determine the target valve opening percentage. The target valve opening percentage is then converted into an analog electrical signal by the programmable logic controller and sent to the electric valve positioner of the steam regulating valve to drive the physical valve stem of the steam regulating valve to generate displacement.

[0057] It should be noted that in this embodiment, a built-in adaptive fuzzy neural network is used as the basic control center. Its default logic is to output an aggressive valve opening command when there is a large negative deviation between the target process temperature and the actual temperature of the feed liquid in the distillation vessel. However, when heat transfer is hindered due to scaling in the reboiler, this aggressive command based on the apparent temperature difference will seriously conflict with the actual physical tolerance of the equipment. Therefore, the core purpose of this step is to establish a reverse constraint mechanism, using the adaptive thermal resistance suppression operator determined in the previous step to forcibly intervene in the control weights of the conventional neural network, thereby effectively suppressing the heat accumulation phenomenon caused by scaling in the reboiler.

[0058] Preferably, as an example, the theoretical valve opening percentage output by the built-in adaptive fuzzy neural network is limited and corrected based on an adaptive thermal resistance suppression operator to determine the target valve opening percentage. Then, a programmable logic controller (PLC) is used to convert the target valve opening percentage into an analog electrical signal, which is sent to the electric valve positioner of the steam regulating valve to drive the physical valve stem of the steam regulating valve to produce displacement. This includes: First, the built-in adaptive fuzzy neural network is used to output the theoretical valve opening percentage at the current sampling time. The built-in adaptive fuzzy neural network is the basic control model of this embodiment. It adopts the existing adaptive neural fuzzy inference system model (ANFIS), takes the deviation and the rate of change of the deviation between the target process temperature and the actual temperature of the current feed liquid in the distillation vessel as input, and takes the theoretical valve opening percentage as output.

[0059] Next, the theoretical valve opening percentage is corrected using an adaptive thermal resistance suppression operator to obtain the target valve opening percentage, specifically satisfying the following relationship:

[0060] In the formula, The index representing the current sampling time; Representing the The percentage of the target valve opening at each sampling time is in dimensionless units. Representing the The adaptive thermal resistance suppression operator at each sampling time is in dimensionless units; Representing the The theoretical valve opening percentage at each sampling time, in dimensionless units; It represents the minimum percentage of valve opening in engineering projects, and its unit is dimensionless.

[0061] It is understandable that when severe scaling occurs in the reboiler of the limonene distillation reactor, heat transfer is hindered, resulting in a lower actual temperature of the feed liquid inside the distillation reactor. This leads to a significant temperature difference between the target process temperature and the actual temperature of the feed liquid inside the distillation reactor. The adaptive fuzzy neural network... The valve will be fully opened due to a large temperature difference; however, the operator may be affected by the physical risks accumulated in the preceding steps. It has significantly decreased and is approaching its peak. This leads to A smaller value reduces the control influence of the valve opening output by the adaptive fuzzy neural network, thus... Item The control effect of valve opening is improved, thereby tightly clamping the target opening to the minimum engineered valve opening that can maintain steam micro-circulation. Nearby, it effectively prevents the risk of heat accumulation caused by heat transfer obstruction.

[0062] Finally, temperature control is performed based on the target valve opening percentage. The specific process is as follows: The programmable logic controller (PLC) uses its analog-to-digital converter to convert the target valve opening percentage, representing a digital ratio, into a corresponding standard industrial analog electrical signal, which is then sent to an electric valve positioner installed on the process pipeline. Upon receiving the analog electrical signal, the positioner drives the physical valve stem of the steam regulating valve to overcome spring and fluid resistance, generating actual mechanical displacement.

[0063] The above embodiments involve the minimum percentage of valve opening in engineering applications. The method for obtaining this parameter will be explained below.

[0064] The minimum engineering valve opening percentage Methods for obtaining [the information] include: The steam regulating valve is manually closed in stages until the steam condensate is critically cut off, as observed by the flow meter or sight glass. The valve opening percentage corresponding to the previous moment is taken as the minimum engineering valve opening percentage.

[0065] To demonstrate the effectiveness of the solution, relevant experiments were conducted. Below are the images obtained from the experiments: Figure 2 The graph shows the dynamic response of distillation temperature control. The dotted line represents the target process temperature, the dashed line represents the actual feed temperature using conventional methods, and the solid line represents the actual feed temperature using the present invention. As can be seen from the graph, in the first stage of undisturbed steady-state, both the solid and dashed lines stably follow the dotted line reference. In the second stage, where heat transfer is hindered, the dashed line trajectory, influenced by the traditional feedback adjustment mechanism, continuously increases the steam injection rate to compensate for the temperature deviation, resulting in only a slight decrease in the actual feed temperature. The solid line trajectory, under the intervention of the adaptive thermal resistance suppression operator, actively weakens the heating output, exhibiting a clear and stable downward trend, thus proactively yielding to abnormal conditions. In the third stage, where heat penetration occurs, the dashed line trajectory experiences severe temperature overshoot due to the instantaneous release of a large amount of accumulated heat energy. Conversely, the solid line trajectory, due to the effective limitation of excessive heat accumulation at the bottom layer by the previous suppression mechanism, does not exhibit significant temperature fluctuations in this stage and can quickly and smoothly recover and follow the dotted line reference. This demonstrates that the present invention has good anti-overshoot capability and system robustness under extreme conditions such as heat transfer resistance.

[0066] Figure 3The graph shows a comparison of actuator opening commands. The dashed line marked with an "X" represents the theoretical valve opening percentage output using conventional methods, while the solid line marked with a circle represents the target valve opening percentage output using the method of this invention. The graph shows that during the deteriorating operating conditions of Phase Two, the dashed line marked with an "X" continuously increases the output command due to persistent temperature deviation, rapidly climbing and remaining at full load and full opening for an extended period. When Phase Three arrives and temperature overshoot occurs, this dashed line experiences a sudden, significant step-down drop. This frequent and drastic switching of the limit stroke exacerbates the wear of the valve's mechanical packing. Conversely, the solid line marked with a circle, after identifying heat accumulation characteristics in Phase Two, is forced to limit its amplitude by the adaptive operator, and is smoothly constrained to near the lower minimum engineering valve opening. When the heat is smoothly released in Phase Three, this solid line smoothly returns to the normal dynamic fine-tuning state. This demonstrates that the present invention effectively reduces ineffective mechanical wear of the actuator, achieves physical limit protection for the terminal equipment, and eliminates the potential for system failure caused by violent mechanical component movements.

[0067] This concludes the embodiment.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network, characterized in that, include: The absolute pressure of the steam pipeline is obtained by a pressure transmitter installed in the steam pipeline network, the valve opening feedback percentage is obtained by a valve positioner, and the actual temperature of the liquid in the distillation kettle is obtained by a temperature sensor. The system energy injection rate is calculated using the valve opening feedback percentage, the absolute pressure of the steam pipeline, and the equivalent enthalpy conversion coefficient. The sensible heat response rate is calculated using the time difference of the actual temperature of the liquid in the distillation vessel and the equivalent comprehensive heat capacity. The energy loss allocation value is determined based on the difference between the two. The bottom heat accumulation is obtained by subtracting the normal heat loss rate from the energy loss ratio and performing time-series accumulation. The bottom heat accumulation is then calculated using the negative exponential decay mapping relationship to determine the adaptive thermal resistance suppression operator used for pressure regulation deviation control. The theoretical valve opening percentage output by the built-in adaptive fuzzy neural network is limited and corrected based on the adaptive thermal resistance suppression operator to determine the target valve opening percentage. The target valve opening percentage is then converted into an analog electrical signal by a programmable logic controller and sent to the electric valve positioner of the steam regulating valve to drive the physical valve stem of the steam regulating valve to produce displacement.

2. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 1, characterized in that, The process of obtaining the absolute pressure of the steam pipeline through a pressure transmitter installed in the steam pipeline network, obtaining the valve opening feedback percentage through a valve positioner, and obtaining the actual temperature of the liquid in the distillation kettle through a temperature sensor includes: The communication interface of the central processing unit is used to synchronously retrieve the absolute pressure data of the steam pipeline transmitted by the pressure transmitter, the valve opening feedback percentage data transmitted by the valve positioner, and the electrical signals collected by the temperature sensor. The electrical signal is preprocessed using an anti-aliasing filtering algorithm and converted into the actual temperature of the liquid in the distillation vessel.

3. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 1, characterized in that, The system energy injection rate is calculated using the valve opening feedback percentage, the absolute pressure of the steam pipeline, and the equivalent enthalpy conversion coefficient. The sensible heat response rate is calculated using the time difference between the actual temperature of the liquid in the distillation vessel and the equivalent comprehensive heat capacity. Based on the difference between the two, the energy loss allocation value is determined, specifically satisfying the following relationship: In the formula, Representing the System energy injection rate at each sampling time; Representing the The percentage of valve opening feedback at each sampling time; Representing the Absolute pressure of the steam pipeline at each sampling time; Represents the equivalent enthalpy conversion factor; Representing the Sensible thermal response rate at each sampling time; and Representing the first The and the first The actual temperature of the liquid in the distillation vessel at each sampling time; Represents a fixed sampling step size; Represents the equivalent overall heat capacity; Representing the Energy loss allocation at each sampling time.

4. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 3, characterized in that, The method for obtaining the equivalent comprehensive heat capacity includes: Extract data from multiple consecutive production batches of distillation vessels while they are in a state confirmed by testing to be free of scaling. In a single batch, record the start time of the stable injection of constant power and the corresponding actual temperature of the liquid in the distillation vessel, and record the end time when the liquid reaches the preset high temperature without boiling and the corresponding actual temperature of the liquid in the distillation vessel. At the same time, record the total time consumed in the start and end stages. Using constant power, total duration, and temperature difference at the start and end times, the heat capacity of a single batch is calculated based on a division relationship, and the equivalent comprehensive heat capacity is obtained by taking the arithmetic mean of the heat capacity values ​​of multiple batches.

5. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 1, characterized in that, The heat accumulation at the bottom of the vessel is obtained by subtracting the normal heat loss rate from the energy loss ratio and then summing the results over time, specifically satisfying the following relationship: In the formula, Representing the Thermal congestion at the bottom of the vessel at each sampling time; Representing the Thermal congestion at the bottom of the vessel at each sampling time; Representing the Energy loss allocation at each sampling time; Represents the normal heat loss rate; Represents a fixed sampling step size; This represents the heat retention ratio coefficient.

6. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 1, characterized in that, The calculation of the thermal accumulation at the bottom of the vessel using the negative exponential decay mapping relationship determines the adaptive thermal resistance suppression operator used for pressure regulation deviation control, specifically satisfying the following relationship: In the formula, Representing the An adaptive thermal resistance suppression operator for each sampling time; Representing the Thermal congestion at the bottom of the vessel at each sampling time; Represents the safe thermal capacity limit threshold; This represents a zero-valued constant.

7. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 6, characterized in that, The method for obtaining the safe thermal capacity limit threshold includes: The total physical mass of the reboiler tube bundle and the metal matrix of the distillation vessel bottom, as well as the corresponding nominal specific heat capacity of the metal, are obtained in advance. Based on the pre-set safe temperature rise limit of the heat exchange interface, the theoretical maximum heat capacity is calculated by multiplying the total physical mass, the nominal specific heat capacity of the metal, and the safe temperature rise limit. The theoretical maximum heat capacity is subtracted from the preset engineering safety margin to determine the safe heat capacity limit threshold.

8. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 5, characterized in that, The method for obtaining the heat retention ratio includes: In the calibration batch of distillation kettles with known scaling conditions, the energy loss ratio at each sampling moment during the heating stage is subtracted from the normal heat loss rate and then accumulated over time to record the total net energy deficit accumulated during the heating stage. Forcefully close the steam regulating valve, record the temperature difference between the moment of closure and the point when the liquid temperature reaches its peak due to residual heat release, multiply the temperature difference by the equivalent comprehensive heat capacity, and calculate the total energy of sensible heat temperature rise. The ratio obtained by dividing the total energy of sensible heat temperature rise by the total energy of net deficit is determined as the heat retention ratio coefficient.

9. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 1, characterized in that, The theoretical valve opening percentage output by the built-in adaptive fuzzy neural network is limited and corrected based on the adaptive thermal resistance suppression operator to determine the target valve opening percentage, specifically satisfying the following relationship: In the formula, Representing the The target valve opening percentage at each sampling time; Representing the An adaptive thermal resistance suppression operator for each sampling time; Representing the The theoretical valve opening percentage at each sampling time; This represents the minimum percentage of valve opening in the engineering system.

10. The method for controlling the temperature of limonene distillation based on an adaptive fuzzy neural network according to claim 9, characterized in that, The method for obtaining the minimum valve opening percentage includes: The steam regulating valve is manually closed in stages until the steam condensate is critically cut off, as observed by the flow meter or sight glass. The valve opening percentage at the moment before the critical cutoff is taken as the minimum engineering valve opening percentage.

Images