Chemical product quality multi-dimensional detection and process dynamic modeling system and method

Abstract

The invention provides a chemical product quality multi-dimensional detection and process dynamic modeling system and method, and belongs to the technical field of chemical production quality detection. Comprising the steps of collecting multi-dimensional parameter data of chemical production equipment, adding a timestamp, storing the data to obtain an original multi-dimensional data set, extracting multi-sensor data, and performing data fusion through a weighted average algorithm or a neural network algorithm to generate a comprehensive quality evaluation result; taking the comprehensive quality evaluation result and historical window data as input, constructing and training a dynamic quality model, and outputting a predicted value; and comparing the predicted value with a product quality target value, and determining whether the operation parameters of the production process need to be adjusted or not. According to the method, the problems of insufficient multi-dimensional detection, lack of real-time monitoring and modeling method lag in traditional chemical product quality control are solved, and multi-dimensional data fusion, real-time updating of a dynamic model and closed-loop optimization control of a production process are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical production quality detection, and in particular to a system and method for multi-dimensional detection and process dynamic modeling of chemical product quality. Background Art

[0002] Chemical product quality control is crucial for ensuring production safety, economic efficiency, and product competitiveness. With the increasing complexity and sophistication of chemical production processes, quality control has shifted from single-point detection to dynamic monitoring throughout the entire process. Traditional quality control systems rely on single-point sensors to monitor independent parameters such as temperature and pressure, and implement production adjustments based on manual experience or simple threshold judgments. These systems are ill-suited to the multivariable, nonlinear, and dynamic nature of modern chemical production.

[0003] However, traditional methods usually test a single quality indicator, ignoring the synergistic effects of multiple parameters such as pressure, flow, and material composition, making it difficult to achieve a systematic evaluation of product quality. At the same time, existing systems generally use periodic offline detection or single-point real-time data acquisition, lacking the ability to integrate and dynamically analyze multi-source data in real time. When an abnormality occurs in the production process (such as micro-leakage in the pipeline or attenuation of catalyst activity), the signal change of a single sensor may be overwhelmed by noise, resulting in a delayed abnormal response, making it difficult to detect production anomalies in a timely manner, affecting safety and qualification rate. In addition, the existing static historical data modeling cannot adapt to dynamic changes in production, and the lack of a feedback mechanism leads to low prediction accuracy.

[0004] Therefore, a multi-dimensional detection and dynamic modeling system and method are urgently needed to improve the efficiency and quality of chemical production. Summary of the Invention

[0005] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a system and method for multi-dimensional detection and process dynamic modeling of chemical product quality, which solves the problems of insufficient multi-dimensional detection, lack of real-time monitoring and lagging modeling methods in traditional chemical product quality control, and realizes multi-dimensional data fusion, real-time updating of dynamic models and closed-loop optimization control of the production process.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for multi-dimensional detection and process dynamic modeling of chemical product quality includes the following steps:

[0008] Data collection: Through the sensor network deployed on chemical production equipment, multi-dimensional parameter data of chemical production equipment is collected in real time, and stored after adding timestamps to obtain the original multi-dimensional data set;

[0009] Data fusion processing: extracting multi-sensor data from the stored original multi-dimensional data set, performing data fusion through a weighted average algorithm or a neural network algorithm, and generating a comprehensive quality assessment result;

[0010] Dynamic modeling: taking the comprehensive quality assessment results and historical window data as input, constructing and training a dynamic quality model, and outputting a predicted value;

[0011] Real-time optimization control: The predicted value of the dynamic quality model is compared with the product quality target value. If it exceeds the preset threshold, the production parameter adjustment amount is calculated, the production equipment adjustment parameters are controlled, and the actual parameters are fed back to the dynamic quality model to achieve closed-loop optimization of the production process.

[0012] Preferably, in the data collection step, the sensor network includes a temperature sensor, a pressure sensor, a flow meter, and a chemical composition analyzer. The multidimensional parameter data includes temperature, pressure, flow, and material composition. The collected multidimensional parameter data is pre-processed by denoising and unified formatting, and stored in a local cache queue in chronological order. A timestamp t is added to each data sample to obtain the original multidimensional data set:

[0013] {x1(t),x2(t),...,x n (t)};

[0014] Among them, x n (t) represents the data collected by the nth sensor at time t.

[0015] Preferably, in the step of data fusion processing, if the weighted average algorithm is adopted, specifically:

[0016] According to the predetermined weights of each sensor data, the following conditions are met:

[0017]

[0018] The comprehensive quality index is calculated by the formula:

[0019]

[0020] Among them, X(t) is the comprehensive quality index at time t, ω i is the weight of the i-th sensor data, x i (t) is the data collected by the i-th sensor at time t.

[0021] Preferably, in the step of data fusion processing, if the neural network fusion algorithm is adopted, specifically:

[0022] The original multidimensional data set is input into the trained neural network model. First, the j-th neuron in the hidden layer calculates the input using the formula:

[0023]

[0024] Among them, net j is the input value of the jth neuron in the hidden layer, ω ij is the connection weight from the i-th neuron in the input layer to the j-th neuron in the hidden layer, x i (t) is the data collected by the i-th sensor at time t, b j is the bias of the jth neuron in the hidden layer;

[0025] It is then processed by the ReLU activation function to obtain the output value:

[0026] y j =max(0,net j );

[0027] Among them, y j is the output value of the jth neuron in the hidden layer after being processed by the ReLU activation function;

[0028] Finally, the output value is passed to the output layer and calculated through linear combination or Softmax function to finally generate the quality evaluation value.

[0029] Preferably, in the step of dynamic modeling, the process of constructing and training a dynamic quality model includes:

[0030] The dynamic quality model is constructed by combining the LSTM neural network with the attention mechanism;

[0031] The weighted average algorithm X(t) or the neural network algorithm The data in the historical window data [tk, t] is used as input, transmitted to the dynamic quality model, and the output is the predicted value; where k is the window size, representing the selected historical data time period.

[0032] Preferably, in the step of dynamic modeling, the process of constructing and training a dynamic quality model further includes:

[0033] The error between the predicted value and the actual value is calculated using the loss function, and the formula is:

[0034]

[0035] Among them, E is the error value, s is the number of samples for a single training, is the predicted value, y l is the actual value.

[0036] Preferably, based on the determined error value, the gradient of the error value with respect to the parameters of the dynamic quality model is determined by a back propagation algorithm, and the parameters of the dynamic quality model are repeatedly updated according to a weight update formula, thereby achieving optimization and updating of the dynamic quality model; the weight update formula is:

[0037]

[0038] Among them, α is the learning rate.

[0039] Preferably, in the step of real-time optimization control, comparing the predicted value output by the dynamic quality model with a preset product quality target value includes:

[0040] like Where δ is the preset threshold, then the current production process is considered to be in a normal state and no adjustment of production parameters is made; otherwise, it is determined that there is a deviation in the production process and the production parameters are adjusted.

[0041] Preferably, if the preset threshold is exceeded, the production parameter adjustment amount is calculated, the production equipment is controlled to adjust the parameters, and the actual parameters are fed back to the dynamic quality model, specifically including:

[0042] First, the feedback control module calculates the production parameter adjustment amount according to the deviation size through the formula:

[0043]

[0044] Secondly, combined with the actual value of the current production parameter, the target value of the production parameter is calculated. The formula is:

[0045] p tar =p cur +Δp;

[0046] Among them, Δp is the production parameter adjustment amount, f is the function for calculating the production parameter adjustment amount, p cur is the actual value of the current production parameter, p tar is the target value of production parameters;

[0047] Finally, the feedback control module sends control instructions to the production equipment through the industrial bus to adjust the operating parameters of the production equipment, and at the same time feeds back the actual operating parameters to the dynamic quality model to complete the closed-loop control process.

[0048] The present invention also provides a chemical product quality multi-dimensional detection and process dynamic modeling system for executing the above-mentioned chemical product quality multi-dimensional detection and process dynamic modeling method, comprising:

[0049] The data acquisition module consists of a sensor network consisting of a temperature sensor module, a pressure sensor module, a flow meter module, and a chemical composition analyzer module. It is used to collect multi-dimensional parameter data in the chemical production process in real time and transmit it to the data fusion processing module;

[0050] The data fusion processing module is used to receive multi-source data, perform integrated analysis using data fusion algorithms, generate comprehensive quality assessment results, and transmit them to the dynamic modeling module and data analysis and monitoring platform;

[0051] The dynamic modeling module builds a dynamic quality model based on machine learning algorithms, receives comprehensive quality assessment results and historical data for model training and updating, and outputs prediction results to the feedback control module and data analysis and monitoring platform;

[0052] The feedback control module receives production parameter adjustment suggestions from the dynamic modeling module, generates control instructions and transmits them to production equipment to achieve real-time optimization of the production process;

[0053] The data analysis and monitoring platform includes a cloud data processing submodule and a user interface submodule. The cloud data processing submodule is used for in-depth data analysis and generation of monitoring charts and trend data. The user interface submodule is used for users to view indicators, receive alarms and generate reports to achieve human-computer interaction.

[0054] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0055] (1) The present invention collects multi-dimensional production data in real time by constructing a sensor network including temperature sensors, pressure sensors, flow meters and chemical composition analyzers; and fuses multi-source data through a weighted average algorithm or a neural network algorithm to generate a comprehensive quality assessment result. Compared with the single indicator detection in the traditional method, the present invention can comprehensively reflect the synergistic influence of multiple parameters on product quality and realize a comprehensive assessment of the quality of chemical products.

[0056] (2) The dynamic quality model constructed based on the neural network of the present invention can adapt to dynamic changes in the production process such as raw material fluctuations and equipment aging by receiving multi-dimensional fusion data and historical window data in real time and continuously updating model parameters, thereby significantly improving the model's adaptability to complex production environments and the accuracy of quality prediction.

[0057] (3) The present invention compares the dynamic model prediction results with the quality target values ​​in real time through the feedback control module, automatically calculates the production parameter adjustment amount and drives the equipment to perform the adjustment, forming a complete closed loop of "data collection-fusion analysis-modeling prediction-optimization control", realizing the transformation from manual experience adjustment to data-driven intelligent control, and greatly improving the stability of the production process and the abnormal response capability. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 This is a flow chart of a method for multi-dimensional detection and process dynamic modeling of chemical product quality according to the present invention;

[0060] Figure 2 This is a module schematic diagram of a chemical product quality multi-dimensional detection and process dynamic modeling system of the present invention. DETAILED DESCRIPTION

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0062] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0063] Example 1

[0064] like Figure 1 As shown, the present invention provides a method for multi-dimensional detection of chemical product quality and process dynamic modeling, comprising the following steps:

[0065] Step 100, data collection: using a sensor network deployed on the chemical production equipment, multi-dimensional parameter data of the chemical production equipment is collected in real time, and stored after adding a timestamp to obtain an original multi-dimensional data set;

[0066] Step 200, data fusion processing: extracting multi-sensor data from the stored original multi-dimensional data set, performing data fusion through a weighted average algorithm or a neural network algorithm, and generating a comprehensive quality assessment result;

[0067] Step 300, dynamic modeling: taking the comprehensive quality assessment results and historical window data as input, constructing and training a dynamic quality model, and outputting a predicted value;

[0068] Step 400, real-time optimization control: The predicted value of the dynamic quality model is compared with the product quality target value. If it exceeds the preset threshold, the production parameter adjustment amount is calculated, the production equipment adjustment parameters are controlled, and the actual parameters are fed back to the dynamic quality model to achieve closed-loop optimization of the production process.

[0069] In step 100, the sensor network includes a temperature sensor, a pressure sensor, a flow meter, and a chemical composition analyzer. The temperature sensor is used to collect real-time product reaction temperature; the pressure sensor is used to monitor the internal pressure of the production equipment; the flow meter is used to measure material flow during the production process; and the chemical composition analyzer is used to detect material composition during the production process, such as reactant concentration or product purity. The multidimensional parameter data includes temperature, pressure, flow, and material composition. The collected multidimensional parameter data is preprocessed by denoising and uniformly formatting, then stored in a local cache queue in chronological order. Each data sample is timestamped t to obtain the original multidimensional dataset:

[0070] {x1(t),x2(t),...,x n (t)};

[0071] Among them, x n (t) represents the data collected by the nth sensor at time t.

[0072] In addition, for denoising multi-dimensional parameter data, the 3σ rule can be used to eliminate jump values, and median filtering can be used to remove high-frequency noise; formatting is unified by converting the analog / digital quantities of different sensors into a standard format, including parameter type, timestamp, value and unit.

[0073] In step 200, if the weighted average algorithm is used, specifically:

[0074] The weights of each sensor data determined in advance by the entropy weight method or expert assignment method must meet the following requirements:

[0075]

[0076] The comprehensive quality index is calculated by the formula:

[0077]

[0078] Among them, X(t) is the comprehensive quality index at time t, ω i is the weight of the i-th sensor data, x i (t) is the data collected by the i-th sensor at time t. For nonlinear quantities, they need to be normalized before participating in the calculation.

[0079] Furthermore, if the neural network fusion algorithm is adopted, specifically:

[0080] The number of neurons in the input layer of the neural network model is consistent with the number of sensors. The original multidimensional data set is input into the trained neural network model. First, the j-th neuron in the hidden layer calculates the input using the formula:

[0081]

[0082] Among them, net j is the input value of the jth neuron in the hidden layer, ω ij is the connection weight from the i-th neuron in the input layer to the j-th neuron in the hidden layer, x i (t) is the data collected by the i-th sensor at time t, b j is the bias of the jth neuron in the hidden layer;

[0083] Then the nonlinear transformation is performed through the ReLU activation function to obtain the output value:

[0084] y j =max(0,net j );

[0085] Among them, y j is the output value of the jth neuron in the hidden layer after being processed by the ReLU activation function;

[0086] Finally, the output value is passed to the output layer and calculated through linear combination or Softmax function to finally generate the quality evaluation value. Reflects the comprehensive quality status of multi-dimensional data.

[0087] Furthermore, the process of building and training a dynamic quality model includes:

[0088] The dynamic quality model is constructed by combining an LSTM neural network with an attention mechanism. The input layer is used to receive the comprehensive quality assessment results at the current moment. The LSTM layer contains 256 memory units to capture the relationship between time series data. The attention mechanism layer is used to calculate the attention weight of the input data. The output layer is used to output the predicted value of the quality indicator for one or more time steps in the future.

[0089] The weighted average algorithm X(t) or the neural network algorithm The data in the historical window data [tk, t] is used as input and transmitted to the dynamic quality model, and the output is the predicted value; where k is the window size, which represents the selected historical data time period. For example, k = 100 means that the data of the first 100 time steps are selected;

[0090] Furthermore, the process of building and training a dynamic quality model also includes:

[0091] The error between the predicted value and the actual value is calculated using the loss function, and the formula is:

[0092]

[0093] Among them, E is the error value, s is the number of samples for a single training, is the predicted value, y l is the actual value.

[0094] Based on the determined error value, the gradient of the error value with respect to the parameters of the dynamic quality model is determined by a back propagation algorithm, and the parameters of the dynamic quality model are repeatedly updated according to a weight update formula, thereby achieving optimization and updating of the dynamic quality model; the weight update formula is:

[0095]

[0096] Among them, α is the learning rate, and the update frequency is not less than 10 times per second to ensure that the model adapts to changes in the production process in real time.

[0097] In step 400, the predicted value output by the dynamic quality model is compared with a preset product quality target value, including:

[0098] like Where δ is the preset threshold, then the current production process is considered to be in a normal state and no adjustment of production parameters is made; otherwise, it is determined that there is a deviation in the production process and the production parameters are adjusted.

[0099] Furthermore, if the preset threshold is exceeded, the production parameter adjustment amount is calculated, the production equipment is controlled to adjust the parameters, and the actual parameters are fed back to the dynamic quality model, specifically including:

[0100] First, the feedback control module calculates the production parameter adjustment amount according to the deviation size through the formula:

[0101]

[0102] Secondly, combined with the actual value of the current production parameter, the target value of the production parameter is calculated. The formula is:

[0103] p tar =p cur +Δp;

[0104] Among them, Δp is the production parameter adjustment amount, f is the function for calculating the production parameter adjustment amount, p cur is the actual value of the current production parameter, p tar is the target value of production parameters;

[0105] Finally, the feedback control module sends control instructions to the production equipment through the industrial bus to adjust the operating parameters of the production equipment, and at the same time feeds back the actual operating parameters to the dynamic quality model to complete the closed-loop control process.

[0106] Example 2

[0107] like Figure 2 As shown, the present invention also provides a chemical product quality multi-dimensional detection and process dynamic modeling system, which is used to execute the chemical product quality multi-dimensional detection and process dynamic modeling method described in Example 1, including:

[0108] The data acquisition module consists of a sensor network consisting of a temperature sensor module, a pressure sensor module, a flow meter module, and a chemical composition analyzer module. It is used to collect multi-dimensional parameter data in the chemical production process in real time and transmit it to the data fusion processing module;

[0109] The data fusion processing module is used to receive multi-source data, perform integrated analysis using data fusion algorithms, generate comprehensive quality assessment results, and transmit them to the dynamic modeling module and data analysis and monitoring platform;

[0110] The dynamic modeling module builds a dynamic quality model based on machine learning algorithms, receives comprehensive quality assessment results and historical data for model training and updating, and outputs prediction results to the feedback control module and data analysis and monitoring platform;

[0111] The feedback control module receives production parameter adjustment suggestions from the dynamic modeling module, generates control instructions and transmits them to production equipment to achieve real-time optimization of the production process;

[0112] The data analysis and monitoring platform includes a cloud data processing submodule and a user interface submodule. The cloud data processing submodule is used for in-depth data analysis and generation of monitoring charts and trend data. The user interface submodule is used for users to view indicators, receive alarms and generate reports to achieve human-computer interaction.

[0113] Therefore, the above-mentioned multi-dimensional detection and process dynamic modeling system and method for chemical product quality is adopted to solve the problems of insufficient multi-dimensional detection, lack of real-time monitoring and lagging modeling methods in traditional chemical product quality control, and realize multi-dimensional data fusion, real-time updating of dynamic models and closed-loop optimization control of production processes.

[0114] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0115] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for multi-dimensional detection and process dynamic modeling of chemical product quality, characterized in that: The following steps are involved: Data collection: Through the sensor network deployed on chemical production equipment, multi-dimensional parameter data of chemical production equipment is collected in real time, and stored after adding timestamps to obtain the original multi-dimensional data set; Data fusion processing: extracting multi-sensor data from the stored original multi-dimensional data set, performing data fusion through a weighted average algorithm or a neural network algorithm, and generating a comprehensive quality assessment result; Dynamic modeling: taking the comprehensive quality assessment results and historical window data as input, constructing and training a dynamic quality model, and outputting a predicted value; Real-time optimization control: The predicted value of the dynamic quality model is compared with the product quality target value. If it exceeds the preset threshold, the production parameter adjustment amount is calculated, the production equipment adjustment parameters are controlled, and the actual parameters are fed back to the dynamic quality model to achieve closed-loop optimization of the production process.

2. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 1, characterized in that: In the data collection step, the sensor network includes a temperature sensor, a pressure sensor, a flow meter, and a chemical composition analyzer. The multi-dimensional parameter data includes temperature, pressure, flow, and material composition. The collected multi-dimensional parameter data is pre-processed by denoising and formatting, and stored in a local cache queue in chronological order. A timestamp t is added to each data sample to obtain the original multi-dimensional data set: {x1(t),x2(t),...,x n (t)}; Among them, x n (t) represents the data collected by the nth sensor at time t.

3. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 1, characterized in that: In the step of data fusion processing, if the weighted average algorithm is adopted, specifically: According to the predetermined weights of each sensor data, the following conditions are met: The comprehensive quality index is calculated by the formula: Among them, X(t) is the comprehensive quality index at time t, ω i is the weight of the i-th sensor data, x i (t) is the data collected by the i-th sensor at time t.

4. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 3, characterized in that: In the step of data fusion processing, if the neural network fusion algorithm is adopted, specifically: The original multidimensional data set is input into the trained neural network model. First, the j-th neuron in the hidden layer calculates the input using the formula: Among them, net j is the input value of the jth neuron in the hidden layer, ω ij is the connection weight from the i-th neuron in the input layer to the j-th neuron in the hidden layer, x i (t) is the data collected by the i-th sensor at time t, b j is the bias of the jth neuron in the hidden layer; It is then processed by the ReLU activation function to obtain the output value: y j =max(0,net j ); Among them, y j is the output value of the jth neuron in the hidden layer after being processed by the ReLU activation function; Finally, the output value is passed to the output layer, and the quality evaluation value is finally generated through linear combination or Softmax function calculation.

5. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 4, characterized in that: In the step of dynamic modeling, the process of building and training a dynamic quality model includes: The dynamic quality model is constructed by combining the LSTM neural network with the attention mechanism; The weighted average algorithm X(t) or the neural network algorithm The data in the historical window data [tk, t] is used as input, transmitted to the dynamic quality model, and the output is the predicted value; where k is the window size, representing the selected historical data time period.

6. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 5, characterized in that: In the step of dynamic modeling, the process of building and training a dynamic quality model further includes: The error between the predicted value and the actual value is calculated using the loss function, and the formula is: Among them, E is the error value, s is the number of samples for a single training, is the predicted value, y l is the actual value.

7. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 6, characterized in that: Based on the determined error value, the gradient of the error value with respect to the parameters of the dynamic quality model is determined by a back propagation algorithm, and the parameters of the dynamic quality model are repeatedly updated according to a weight update formula, thereby achieving optimization and updating of the dynamic quality model; the weight update formula is: Among them, α is the learning rate.

8. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 7, characterized in that: In the step of real-time optimization control, the predicted value output by the dynamic quality model is compared with a preset product quality target value, including: like Where δ is the preset threshold, then the current production process is considered to be in a normal state and no adjustment of production parameters is made; otherwise, it is determined that there is a deviation in the production process and the production parameters are adjusted.

9. A chemical product quality multi-dimensional detection and process dynamic modeling method according to claim 8, characterized in that: If the preset threshold is exceeded, the production parameter adjustment amount is calculated, the production equipment adjustment parameters are controlled, and the actual parameters are fed back to the dynamic quality model, including: First, the feedback control module calculates the production parameter adjustment amount according to the deviation size through the formula: Secondly, combined with the actual value of the current production parameter, the target value of the production parameter is calculated. The formula is: p tar =p cur +Δp; Among them, Δp is the production parameter adjustment amount, f is the function for calculating the production parameter adjustment amount, p cur is the actual value of the current production parameter, p tar is the target value of production parameters; Finally, the feedback control module sends control instructions to the production equipment through the industrial bus to adjust the operating parameters of the production equipment, and at the same time feeds back the actual operating parameters to the dynamic quality model to complete the closed-loop control process.

10. A chemical product quality multi-dimensional detection and process dynamic modeling system, characterized by: The method for performing multi-dimensional detection and process dynamic modeling of chemical product quality according to any one of claims 1 to 9 comprises: The data acquisition module consists of a sensor network consisting of a temperature sensor module, a pressure sensor module, a flow meter module, and a chemical composition analyzer module. It is used to collect multi-dimensional parameter data in the chemical production process in real time and transmit it to the data fusion processing module; The data fusion processing module is used to receive multi-source data, perform integrated analysis using data fusion algorithms, generate comprehensive quality assessment results, and transmit them to the dynamic modeling module and data analysis and monitoring platform; The dynamic modeling module builds a dynamic quality model based on machine learning algorithms, receives comprehensive quality assessment results and historical data for model training and updating, and outputs prediction results to the feedback control module and data analysis and monitoring platform; The feedback control module receives production parameter adjustment suggestions from the dynamic modeling module, generates control instructions and transmits them to production equipment to achieve real-time optimization of the production process; The data analysis and monitoring platform includes a cloud data processing submodule and a user interface submodule. The cloud data processing submodule is used for in-depth data analysis and generation of monitoring charts and trend data. The user interface submodule is used for users to view indicators, receive alarms and generate reports to achieve human-computer interaction.

Images