An automatic monitoring and optimization system for fine chemical production processes

Abstract

The present application relates to the field of automation, in particular to a fine chemical production process automation monitoring and optimization system, comprising a time-frequency domain fusion characteristic quantity of stirring torque power time sequence fluctuation data extracted in real time through an association matrix construction module, a dynamic inversion thixotropic index combined with a deep neural network model, solving the problem that traditional methods are difficult to perceive material rheological characteristics in real time; a dynamic coupling analysis module analyzes the material viscosity change rate based on the thixotropic index, generates stirring speed adjustment instructions and jacket temperature compensation values dynamically matched with viscosity by using fuzzy PID control, overcoming the defects of uneven mixing and local overheating caused by process parameter lagging adjustment; a multi-objective collaborative optimization module locks the quality optimization weight in the viscosity mutation stage, quickly and stably adjusts the reaction conditions through the feedforward compensation algorithm, and dynamically balances the stirring power consumption and heat transfer efficiency in the steady state based on the Pareto frontier search, solving the conflict between quality and energy efficiency targets.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automation, in particular to a fine chemical production process automation monitoring and optimization system. BACKGROUND

[0002] In the pesticide production process, the rheological properties of the material directly affect the reaction mixing uniformity, effective ingredient dispersion and the quality of the final product. Traditional pesticide production relies on fixed process parameters or manual experience to adjust the stirring speed and temperature, and it is difficult to capture the dynamic viscosity changes caused by the thixotropic behavior (such as shear thinning or viscoelasticity mutation) of the material in real time. Especially when synthesizing high-viscosity pesticide emulsions or suspensions, abnormal fluctuations in stirring torque can easily cause uneven mixing, local overheating and other problems, resulting in poor batch stability or increased side reactions. In the prior art, offline viscosity detection has a lag, and a single device control strategy cannot simultaneously adapt to viscosity mutations and energy efficiency optimization requirements, resulting in energy waste or product quality fluctuations.

[0003] In addition, the batch difference of pesticide raw materials and the presence of toxic substances further exacerbate the insufficient response of traditional systems to dynamic working conditions, and there is an urgent need for a precise control scheme that can analyze thixotropic properties in real time and optimize multiple targets. SUMMARY

[0004] The purpose of the present application is to provide a fine chemical production process automation monitoring and optimization system to solve the problems raised in the background art, and the specific technical problems include how to perceive the thixotropic properties of pesticide production materials in real time and quickly analyze the viscosity change trend to solve the cooperative lag problem of stirring mixing efficiency and temperature control; and how to switch between quality priority and energy efficiency priority modes according to the dynamic behavior of viscosity to achieve balanced optimization of stability and energy consumption in the pesticide synthesis process.

[0005] To achieve the above purpose, the present application provides the following technical scheme: a fine chemical production process automation monitoring and optimization system, comprising a correlation matrix construction module, a dynamic coupling analysis module and a multi-objective collaborative optimization module, wherein:

[0006] The correlation matrix construction module uses sliding window segmentation processing to segment the stirring torque power time series fluctuation data, extracts the peak value and standard deviation of the torque fluctuation in each time window as the basic characteristic quantity in the time domain, and obtains the main frequency band energy distribution characteristic through fast Fourier transform in the frequency domain. The two are normalized and fused to form a fluctuation amplitude characteristic quantity, which comprehensively represents the rheological properties of the pesticide material;

[0007] The correlation matrix construction module constructs a nonlinear mapping model based on a deep neural network, an input layer receives real-time fluctuation amplitude characteristic quantities, a hidden layer dynamically captures the correlation mode of torque fluctuation and material thixotropic behavior through an adaptive activation function, and an output layer generates a thixotropic index.

[0008] The change rate calculation unit in the dynamic coupling analysis module adopts a time series trend analysis algorithm to dynamically track the thixotropic index, and accurately obtains the material viscosity change rate through polynomial fitting and first-order derivative calculation in a sliding time window.

[0009] The device control parameter set generation unit in the dynamic coupling analysis module generates a device control parameter set including stirring speed adjustment instructions and jacket temperature compensation values according to the preset material viscosity change rate and process parameter mapping rule, and adopts a fuzzy PID control algorithm; when the material viscosity change rate is positively increased, the gain coefficient of the stirring speed adjustment instruction is dynamically increased and the jacket temperature compensation value is increased; when the material viscosity change rate is negatively attenuated, the stirring speed is reversely adjusted and the temperature compensation amplitude is reduced, so as to realize precise control under dynamic viscosity change.

[0010] The multi-objective collaborative optimization module executes a dynamic weight distribution strategy according to the device control parameter set, wherein:

[0011] When the material viscosity change rate exceeds the preset mutation threshold, the quality optimization weight locking mechanism is activated, a step increment of the jacket temperature compensation value is generated through a feedforward compensation algorithm, the increment is determined by the square term, the linear term and the material thermal conductivity constant of the viscosity change rate, the device thermal capacity constraint is integrated and a segmented amplitude limiting strategy is adopted to suppress temperature overshoot, and the quality optimization weight is forcibly locked to ensure the stability of pesticide synthesis.

[0012] When the material viscosity change rate returns to the steady state interval, a multi-objective optimization function is constructed based on the stirring speed adjustment instruction, the stirring power consumption and the heat transfer efficiency are taken as core variables, the feasible solution set is analyzed in real time through a Pareto frontier search mechanism, the optimal weight distribution point is determined by using a gradient projection method with constraints, the speed step size is dynamically adjusted in combination with the sensitivity matrix of the frequency converter speed regulation, and a lagging attenuation factor is introduced in the jacket temperature compensation value, the distribution proportion of the quality weight and the energy efficiency weight is adjusted in real time according to the second derivative direction and the normalized rate value of the material viscosity change rate, and dynamic balance in the energy consumption optimal mode is realized.

[0013] Compared with the prior art, the present application has the following advantages:

[0014] By real-time analyzing the thixotropic properties and dynamic viscosity changes of pesticide materials, the mixing efficiency and the response speed of temperature control are significantly improved, effectively avoiding product failure caused by local overheating or uneven mixing; the adaptive learning ability of deep neural networks enhances the system's robustness to toxic components and batch differences of pesticide raw materials, ensuring control accuracy under complex conditions; the combination of fuzzy PID and multi-objective optimization algorithm prioritizes pesticide synthesis quality when viscosity changes, and reduces stirring energy consumption and heat loss in the steady state, achieving dynamic balance between quality and energy efficiency. Through feedforward compensation and Pareto optimization mechanism, energy waste and waste generation in pesticide production are reduced, providing an intelligent solution for the production of high-toxicity and high-stability pesticides. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is the overall module schematic diagram of the present application;

[0016] Figure 2 is the dynamic coupling analysis module unit schematic diagram of the present application.

[0017] In the figure: 100, correlation matrix construction module; 200, dynamic coupling analysis module; 201, change rate calculation unit; 202, device control parameter set generation unit; 300, multi-objective collaborative optimization module. DETAILED DESCRIPTION

[0018] In order to more clearly understand the purpose, technical scheme and advantages of the present application, the present application is described and explained below in conjunction with the drawings and examples.

[0019] Unless otherwise defined, technical terms or scientific terms used in the present application shall have the same meaning as those commonly understood by a person having ordinary skill in the art to which the present application belongs. The terms "one", "a", "an", "the", "these", and similar terms in the present application do not mean "only one" or "exactly one", but can mean "one or more". The terms "include", "contain", "have", and any variant thereof in the present application are intended to cover the inclusions without being exclusive; for example, a process, method, and system, product or device containing a series of steps or modules (units) are not limited to the listed steps or modules (units), but can include steps or modules (units) not listed, or can include other steps or modules (units) inherent to the process, method, product or device. The terms "connect", "connected", "coupled" and the like in the present application are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The term "multiple" in the present application means two or more. The term "and / or" describes the association relationship between the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean that A exists alone, A and B exist together, and B exists alone. Generally, the character " / " means that the objects before and after are in an "or" relationship. The terms "first", "second", "third" and the like in the present application are only used to distinguish similar objects, and do not represent a specific order of the objects.

[0020] Next, please refer to Figure 1 The present application provides a technical solution: an automatic monitoring and optimization system for fine chemical production process, comprising a correlation matrix construction module 100, a dynamic coupling analysis module 200 and a multi-objective collaborative optimization module 300.

[0021] The correlation matrix construction module 100 collects stirring torque power time series fluctuation data in the running process of the stirring device through a high-precision torque sensor, and performs sliding window segmentation processing on the stirring torque power time series fluctuation data to divide continuous time window units; in the time domain dimension, the peak value and the standard deviation of the torque fluctuation signal in each time window are calculated, which are respectively used as the basic characteristic quantities representing the torque instantaneous impact intensity and the fluctuation dispersion degree;

[0022] Simultaneously, in the frequency domain, a Fast Fourier Transform is performed on the torque fluctuation data for each time window to extract the energy distribution characteristics of the main frequency band and quantify the energy proportion of different frequency components. A normalization algorithm is then used to fuse the time-domain basic features with the frequency-domain main frequency band energy distribution characteristics in a multi-dimensional manner. After eliminating dimensional differences, a normalized torque fluctuation amplitude feature is formed, thus comprehensively characterizing the material's rheological properties in the time-frequency domain. This provides comprehensive input parameters for deep neural network models to characterize the thixotropic behavior of materials, supporting the accurate inversion of the thixotropic index. This technology effectively solves the problem of lost frequency domain information for complex rheological properties in traditional time-domain analysis, enhancing the system's ability to analyze the dynamic response of pesticide mixture rheological states.

[0023] A deep neural network model is trained based on historical production data to establish a nonlinear mapping relationship between torque fluctuation amplitude features and thixotropic index. The input layer of the deep neural network model receives real-time extracted torque fluctuation amplitude features, the hidden layer captures the dynamic correlation pattern between torque fluctuation and material thixotropic behavior through an adaptive activation function, and the output layer generates the thixotropic index at the current moment. An online incremental learning mechanism continuously updates the network weight parameters, enabling the deep neural network model to dynamically adapt to the rheological property variations caused by different raw material batches. Ultimately, this achieves the real-time analytical capability of inferring material property parameters based on equipment operation data.

[0024] The online incremental learning mechanism constructs a dynamic incremental training sample set by collecting real-time data on the temporal fluctuations of stirring torque and power during the production of new raw material batches, along with the corresponding measured values ​​of the thixotropic index. An elastic weight solidification algorithm is used to update the weights of the deep neural network model. By calculating the diagonal elements of the Fisher information matrix of historical batch data, the importance of each network weight parameter to the learned rheological properties is quantified. Regularization constraints are applied during the training of new batches to retain key weight parameters while allowing dynamic adjustment of non-key weights. Combined with an adaptive learning rate scheduler, the model update amplitude is dynamically adjusted based on the shift between the raw material batch switching signal and the feature distribution of the new samples. This allows the network to continuously integrate the nonlinear characteristics of the thixotropic behavior of new batch materials while avoiding catastrophic forgetting, achieving dynamic tracking of cross-batch rheological property variations and enhancing the model's generalization ability.

[0025] In pesticide production, the correlation matrix construction module 100 monitors torque fluctuation data in the reactor stirring process in real time through a high-precision torque sensor. Combined with time-frequency domain feature fusion technology, it accurately analyzes the rheological properties (such as thixotropic index) of the pesticide mixture and dynamically tracks viscosity changes caused by batch differences in raw materials. This provides real-time inversion support for physical property parameters for subsequent process optimization, ensuring uniform dispersion and reaction stability of active ingredients in different batches of pesticides.

[0026] Please see Figure 2, the change rate calculation unit 201 in the dynamic coupling analysis module 200 adopts a time series trend analysis algorithm to dynamically track the material viscosity change process by receiving the thixotropic index output by the correlation matrix construction module 100, wherein:

[0027] The thixotropic index is intercepted by using a sliding time window mechanism, a trend function of the thixotropic index changing with time is constructed by using a second-order polynomial fitting algorithm, and a first-order derivative thereof is calculated as the material viscosity change rate; wherein the sliding window length is dynamically adjusted according to the material batch switching signal and the rheological response delay characteristic, and when the raw material batch switching is detected, the window length is automatically shortened to improve the trend tracking sensitivity; in the polynomial fitting process, the fitting coefficients are iteratively optimized based on the least square method principle, and abnormal data points caused by torque sensor noise or process disturbance are removed through residual analysis; at the same time, a window overlap factor is introduced to smooth the fitting results of adjacent windows, and the rate calculation fluctuation caused by window boundary mutation is suppressed.

[0028] The change rate calculation unit 201 intercepts the time series data of the thixotropic index by using a sliding time window mechanism, constructs a trend function by using a second-order polynomial fitting algorithm and calculates a first-order derivative thereof, accurately analyzes the material viscosity change rate; the tracking sensitivity is improved by dynamically adjusting the window length (such as shortening the window when the raw material batch is switched), abnormal points caused by sensor noise and process disturbance are removed in combination with residual analysis, and a window overlap factor is introduced to smooth the fitting results of adjacent windows, effectively suppressing the rate calculation fluctuation. This technology solves the problems of noise sensitivity and hysteresis of the traditional difference method, and in pesticide production, it can capture viscosity mutations (such as shear thinning in the emulsification stage of herbicides) in real time, provide accurate trend basis for subsequent dynamic adjustment of equipment parameters, and ensure the stability of the reaction process.

[0029] The device control parameter set generation unit 202 in the dynamic coupling analysis module 200 generates a device control parameter set by using a fuzzy PID control algorithm according to a preset material viscosity change rate and process parameter mapping rule; when the viscosity change rate is positively increasing, the gain coefficient of the stirring speed adjustment instruction is dynamically increased according to the rate gradient to strengthen the shearing effect, and a jacket temperature compensation value is calculated based on the energy conservation equation to offset the viscous heating effect; when the viscosity change rate is negatively attenuated, the stirring speed is adjusted in the opposite direction and the temperature compensation amplitude is reduced, thereby forming a device control parameter set of the stirring speed adjustment instruction and the jacket temperature compensation value that matches the material rheological state in real time, realizing closed-loop cooperative control of the process parameters and the evolution of the material properties.

[0030] The device control parameter set generation unit 202 generates stirring speed adjustment instructions and jacket temperature compensation values that match the rheological state in real time based on the material viscosity change rate using a fuzzy PID control algorithm. When the material viscosity change rate is positively increasing, the stirring speed gain coefficient is dynamically increased to strengthen the shear mixing, and the temperature compensation value is calculated based on the energy conservation equation to offset the viscous heat generation. When the material viscosity change rate is negatively attenuated, the parameters are adjusted in the opposite direction to stabilize the system. This technology adjusts the PID parameters adaptively through a fuzzy rule base, overcomes the adjustment hysteresis of traditional linear control for non-linear rheological behavior, and realizes closed-loop collaborative control of stirring intensity and temperature compensation in pesticide synthesis (such as insecticide condensation viscosity mutation scenarios), inhibits local overheating and uneven mixing, and ensures reaction efficiency and product consistency.

[0031] For the problem of raw material viscosity mutation in pesticide synthesis process, the dynamic coupling analysis module 200 dynamically adjusts the stirring speed and jacket temperature compensation strategy by calculating the change rate of thixotropic index in real time, for example, rapidly increasing the shear force to inhibit phase separation during the emulsification stage of herbicides, while compensating for local overheating caused by viscous heat generation, realizing closed-loop collaborative control of reaction conditions and material rheological state, and ensuring the synthesis efficiency of pesticide active ingredients and product quality consistency.

[0032] The multi-objective collaborative optimization module 300 uses a state-triggered dynamic weight distribution strategy to realize process target priority switching based on the device control parameter set (including stirring speed adjustment instructions and jacket temperature compensation values) output by the device control parameter set generation unit 202, wherein:

[0033] When the material viscosity change rate exceeds the preset mutation threshold, it is determined that the rheological property is in a stage of violent fluctuation, the quality optimization weight locking mechanism is activated, the jacket temperature compensation value is used as the core control variable, the temperature compensation amplitude and response speed are increased to the preset upper limit through a feedforward compensation algorithm, a first control instruction is generated to forcibly lock the quality optimization weight (for example, set to the interval of 0.9-1.0), and the material mixing uniformity and reaction stability are preferentially maintained; wherein:

[0034] The feedforward compensation algorithm constructs a fast response mechanism based on the pre-stored material viscosity mutation mode and device thermal inertia parameters, calls a preset compensation coefficient matrix (including compensation amplitude increments and response acceleration parameters corresponding to different viscosity mutation levels) according to the gradient direction and intensity of viscosity change, and directly calculates the step increment of the jacket temperature compensation value; the increment is determined by the square term, the linear term and the material heat conduction constant of the viscosity change rate, and the response time of the temperature control loop is compressed through a dynamic acceleration factor; the algorithm synchronously integrates the device heat capacity constraint, monitors the jacket temperature overshoot risk in real time, uses a piecewise amplitude limiting strategy to suppress temperature overshoot when the compensation value is generated, and ensures the rapidity and stability of the compensation process.

[0035] When the viscosity change rate falls back to the steady state interval (±5% of the preset reference value), switch to the energy efficiency optimization mode, build a multi-objective optimization function based on the stirring speed adjustment instruction, dynamically adjust the mass and energy efficiency weight distribution ratio (0.5-0.7 interval) by real-time calculation of the Pareto frontier of the stirring power consumption (proportional to the cube of the speed) and the jacket heat transfer efficiency (linearly related to the temperature difference), generate the second control instruction to drive the frequency converter to execute the speed fine tuning, and introduce a hysteresis attenuation factor in the jacket temperature compensation value, realize the dynamic balance of the minimum stirring energy consumption and the maximum heat transfer efficiency, and finally form a closed-loop optimization control chain that adapts to the evolution of the material state; wherein:

[0036] The multi-objective optimization function takes stirring efficiency and heat transfer power as the core variables, dynamically balances the mass and energy consumption targets, and adjusts the distribution ratio of the mass weight and the energy efficiency weight in real time according to the second derivative direction of the viscosity change rate and the normalized rate value: if the viscosity change is accelerating, the mass weight is preferentially increased; if it is decelerating, the energy efficiency weight ratio is increased; during the optimization process, a Pareto boundary search mechanism is introduced, the feasible solution set of stirring power and heat transfer efficiency is analyzed in real time, the optimal weight distribution point is determined by using the gradient projection method with constraints, and the speed step is dynamically adjusted combined with the sensitivity matrix of the frequency converter speed regulation, finally realizing the adaptive balance of mass and energy consumption targets under time-varying working conditions.

[0037] In the continuous production of pesticides, the multi-objective collaborative optimization module 300 intelligently switches between mass priority and energy efficiency priority modes according to the viscosity change trend: when the insecticide polycondensation reaction is intense, the temperature compensation weight is locked to ensure smooth reaction; in the steady state stage, the stirring speed and heat transfer parameters are optimized to balance the reaction rate and energy consumption cost, for example, reducing the energy consumption of bactericide production by 15% while maintaining a product yield of 98%, forming a dynamic control chain of mass and energy efficiency dual-objective optimization.

[0038] As can be seen from the above description, the fine chemical production process automation monitoring and optimization system provided in the embodiment has the following technical effects:

[0039] Through time-frequency domain feature fusion and deep neural network online incremental learning, the high-precision real-time analysis of material flow behavior characteristics in pesticide production process is realized, the thixotropy variation caused by raw material batch difference is dynamically tracked, the uniformity of mixing and the stability of reaction are ensured; based on fuzzy PID control and viscosity change rate self-adaptive equipment parameter dynamic adjustment, through feedforward compensation and segmented limiting strategy, the reaction conditions are quickly and stably adjusted in the viscosity mutation stage, the risk of local overheating and phase separation is inhibited; combined with Pareto frontier search and multi-objective weight dynamic allocation mechanism, the stirring energy consumption and heat transfer efficiency are balanced in the steady state stage, the seamless switching of quality priority and energy efficiency priority mode is formed, and finally the closed-loop control of controllable product quality, energy optimization and process parameter collaborative response is realized under complex working conditions, which significantly improves the intelligent level and comprehensive benefit of pesticide production.

[0040] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above-mentioned embodiments, and the above-mentioned embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. An automated monitoring and optimization system for fine chemical production processes, characterized in that, It includes an association matrix construction module (100), a dynamic coupling analysis module (200), and a multi-objective collaborative optimization module (300), wherein: The correlation matrix construction module (100) extracts the fluctuation amplitude feature quantity that characterizes the rheological properties of the material based on the online collected stirring torque power time-series fluctuation data, and establishes a nonlinear mapping model between the fluctuation amplitude feature quantity and the thixotropic index. The nonlinear mapping model is used to calculate the thixotropic index in real time based on the fluctuation amplitude feature quantity. The nonlinear mapping model is a deep neural network model trained based on historical production data. Its input layer receives the fluctuation amplitude feature quantity extracted in real time, the hidden layer captures the dynamic correlation pattern between torque fluctuation and material thixotropic behavior through an adaptive activation function, and the output layer generates the thixotropic index. The thixotropic index is used to characterize the dynamic index of the material thixotropic behavior. The dynamic coupling analysis module (200) analyzes the material viscosity change trend and obtains the material viscosity change rate based on the thixotropic index; and generates a set of equipment control parameters including stirring speed adjustment commands and jacket temperature compensation values ​​based on the material viscosity change rate. The multi-objective collaborative optimization module (300) executes a dynamic weight allocation strategy based on the equipment control parameter set, wherein: When the rate of change of material viscosity exceeds the preset abrupt change threshold, a first control command is generated based on the jacket temperature compensation value to forcibly lock the quality optimization weight. When the rate of change of material viscosity returns to the steady-state range, a second control command is generated based on the stirring speed adjustment command in the same set of control parameters of the equipment to switch to the energy efficiency optimization mode, so as to dynamically balance the stirring power consumption and heat transfer efficiency.

2. The automated monitoring and optimization system for fine chemical production processes according to claim 1, characterized in that, The process of extracting the fluctuation amplitude feature includes: The time-series fluctuation data of stirring torque power is segmented by sliding window segmentation, and the peak value and standard deviation of torque fluctuation within each time window are extracted as basic feature quantities in the time domain. In the frequency domain, the energy distribution characteristics of the main frequency band are obtained through fast Fourier transform, and the basic characteristic quantity is normalized and fused with the energy distribution characteristics of the main frequency band to form the fluctuation amplitude characteristic quantity.

3. The automated monitoring and optimization system for fine chemical production processes according to claim 1, characterized in that, The deep neural network model continuously updates the network weight parameters through an online incremental learning mechanism to dynamically adapt to the variations in rheological properties caused by different batches of raw materials.

4. The automated monitoring and optimization system for fine chemical production processes according to claim 1, characterized in that, The dynamic coupling analysis module (200) includes a rate of change calculation unit (201). The rate of change calculation unit (201) obtains the rate of change of material viscosity by using a time series trend analysis algorithm to dynamically track the thixotropic index and by obtaining it through polynomial fitting and first derivative calculation within a sliding time window.

5. The automated monitoring and optimization system for fine chemical production processes according to claim 1, characterized in that, The dynamic coupling analysis module (200) includes a device control parameter set generation unit (202), and the process by which the device control parameter set generation unit (202) generates the device control parameter set includes: Based on the preset mapping rules between the material viscosity change rate and process parameters, a fuzzy PID control algorithm is used to generate a set of equipment control parameters. When the viscosity change rate increases positively, the gain coefficient of the stirring speed adjustment command is dynamically increased and the jacket temperature compensation value is calculated. When the viscosity change rate decreases negatively, the stirring speed is adjusted in the opposite direction and the temperature compensation amplitude is reduced.

6. The automated monitoring and optimization system for fine chemical production processes according to claim 1, characterized in that, When the multi-objective collaborative optimization module (300) executes the dynamic weight allocation strategy, it includes: When the rate of change of material viscosity exceeds the preset mutation threshold, the quality optimization weight locking mechanism is activated, the compensation magnitude and response speed of the jacket temperature compensation value are increased to the preset upper limit, and the first control command is generated to forcibly lock the quality optimization weight. When the rate of change of material viscosity returns to the steady-state range, a multi-objective optimization function is constructed based on the stirring speed adjustment command. The weight allocation ratio of mass and energy efficiency is dynamically adjusted to generate a second control command, and the system switches to the energy efficiency optimization mode.

7. The automated monitoring and optimization system for fine chemical production processes according to claim 6, characterized in that, The quality optimization weight locking mechanism generates a step increment of the jacket temperature compensation value through a feedforward compensation algorithm. The feedforward compensation algorithm integrates equipment thermal capacity constraints and adopts a segmented amplitude limiting strategy to suppress temperature overshoot. The step increment is determined by the square term, linear term, and material thermal conductivity constant of the viscosity change rate.

8. The automated monitoring and optimization system for fine chemical production processes according to claim 6, characterized in that, The multi-objective optimization function takes stirring power consumption and heat transfer efficiency as core variables, analyzes feasible solution sets in real time through Pareto front search mechanism, and uses constrained gradient projection method to determine the optimal weight allocation point. Combined with the sensitivity matrix of inverter speed regulation, the speed step size is dynamically adjusted.

9. The automated monitoring and optimization system for fine chemical production processes according to claim 6, characterized in that, The energy efficiency optimization mode introduces a hysteresis attenuation factor into the jacket temperature compensation value, and adjusts the allocation ratio of mass weight and energy efficiency weight in real time according to the direction of the second derivative of the material viscosity change rate and the normalized rate value.

Images