Visualization system and method for the whole production process of MVR scraper crystallizer

Abstract

The present invention relates to the field of industrial automation and intelligent control technology, and in particular to a visualization system and method for the entire production process of an MVR scraper crystallizer. The content includes: collecting production data in real time, analyzing and reconstructing the production data after cleaning and filtration to obtain reconstructed production data; establishing a dynamic tensor model of the production process to predict the system state tensor; defining and optimizing the objective function, determining the optimal control input tensor, and optimizing the production parameters, while displaying the reconstructed production data, the predicted system state tensor, and the adjustment of the actual production parameters in real time in a graphical manner to achieve intelligent control. It solves the problems of being unable to effectively process multi-dimensional complex data, and being difficult to achieve real-time data monitoring and processing; being unable to accurately predict state changes in the production process, resulting in a lag in the adjustment of the control strategy; being difficult to achieve global optimization and dynamic adjustment, and being difficult to achieve intelligent control and optimization of the entire process.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation and intelligent control technology, and in particular to a visualization system and method for the entire production process of an MVR scraper crystallizer. Background Art

[0002] MVR (Mechanical Vapor Recompression) scraper crystallizer is a key equipment widely used in chemical, pharmaceutical, food and other industries. It is used to process high-viscosity and high-concentration solutions and prepare high-purity solid products through evaporation and crystallization processes. With the continuous development of industrial production, people have higher and higher requirements for product quality and production efficiency. Traditional control and monitoring methods are difficult to meet the needs of modern production. Therefore, there is an urgent need for an efficient and intelligent system to improve the operating performance and production effect of MVR scraper crystallizer.

[0003] With the rapid development of sensor technology and data processing technology, intelligent production has gradually become a trend in industrial development. Through advanced sensor technology, various production data in the production process can be collected in real time. Through efficient data processing and analysis algorithms, valuable information can be extracted from massive production data, providing a scientific basis for the optimization and control of the production process. At the same time, based on data-driven intelligent optimization algorithms, dynamic adjustment and optimization of production parameters can be achieved to ensure that the production process runs in the optimal state, thereby improving product quality and production efficiency.

[0004] However, the above technology has the following technical problems: it is unable to effectively process the multi-dimensional and complex data generated in the production process, resulting in insufficient production data analysis and inaccurate feature extraction; it is difficult to achieve real-time data monitoring and processing, affecting the response speed of the control system and the control accuracy of the production process; it is unable to accurately predict state changes in the production process, resulting in delayed adjustment of the control strategy; it is difficult to achieve global optimization and dynamic adjustment, the control effect is not ideal, and it is difficult to achieve intelligent control and optimization of the entire process. Summary of the invention

[0005] The present invention provides a visualization system and method for the entire production process of an MVR scraper crystallizer, so as to solve the problems that the multidimensional complex data generated in the production process cannot be effectively processed, resulting in insufficient production data analysis and inaccurate feature extraction; it is difficult to achieve real-time data monitoring and processing, which affects the response speed of the control system and the control accuracy of the production process; it is impossible to accurately predict state changes in the production process, resulting in delayed adjustment of the control strategy; it is difficult to achieve global optimization and dynamic adjustment, the control effect is not ideal, and it is difficult to achieve intelligent control and optimization of the entire process.

[0006] The visualization system and method of the whole production process of the MVR scraper crystallizer of the present invention specifically include the following technical solutions:

[0007] The visualization method of the whole process of MVR scraper crystallizer production includes the following steps:

[0008] S1. Real-time collection of production data, cleaning and filtering of the production data; analysis and reconstruction of the cleaned and filtered production data through a multi-dimensional tensor iterative transformation analysis algorithm to obtain reconstructed production data, and display the reconstructed production data in real time in a graphical manner;

[0009] S2. Based on the reconstructed production data and control input tensor, a dynamic tensor model of the production process is established through an adaptive dynamic tensor control algorithm to predict the system state tensor and visualize the predicted system state tensor; the optimal control input tensor is determined by defining and optimizing the objective function;

[0010] S3. Apply the optimal control input tensor to the actual production parameters, monitor and feedback data in real time, optimize the production parameters through the adaptive multi-nested variational optimization algorithm, and display the adjustment status of the actual production parameters and feedback data in real time in a graphical manner, ultimately achieving intelligent control.

[0011] Preferably, the S1 specifically includes:

[0012] A multidimensional tensor iterative transformation analysis algorithm is designed. The cleaned and filtered production data is preliminarily processed through multidimensional tensor transformation. The cleaned and filtered production data is mapped to the tensor space, and components of different frequencies are separated to obtain the production data after multidimensional tensor transformation.

[0013] Preferably, the S1 specifically includes:

[0014] Iterative analysis is performed on the production data after multi-dimensional tensor transformation, and high-frequency and low-frequency features in the production data after multi-dimensional tensor transformation are extracted through high-order partial derivatives and gradient iteration to obtain feature data.

[0015] Preferably, the S1 specifically includes:

[0016] Reconstruction is performed based on the extracted feature data to obtain reconstructed production data.

[0017] Preferably, the S2 specifically includes:

[0018] Based on the reconstructed production data, combined with the control input tensor and the noise tensor, a dynamic tensor model of the production process is constructed; through the dynamic tensor model of the production process, the system state tensor is predicted.

[0019] Preferably, the S3 specifically includes:

[0020] The feedback data is compared with the target value, the feedback error is calculated, and the production parameters are adjusted and controlled through an adaptive multi-nested variational optimization algorithm to iteratively adjust the production process.

[0021] Preferably, the S3 specifically includes:

[0022] During the implementation of the adaptive multi-nested variational optimization algorithm, collaborative optimization is performed to obtain the global optimal production parameters and ultimately achieve intelligent control.

[0023] The visualization system of the entire production process of MVR scraper crystallizer includes the following parts:

[0024] Data acquisition module, data processing module, real-time monitoring module, control optimization module, parameter adjustment module, and visual data display module;

[0025] The data acquisition module collects production data in the production process in real time through sensors; and transmits the production data to the data processing module in real time;

[0026] The data processing module cleans and filters the production data, and analyzes the cleaned and filtered production data through a multi-dimensional tensor iterative transformation analysis algorithm; performs preliminary processing on the cleaned and filtered production data through multi-dimensional tensor transformation to obtain the production data after multi-dimensional tensor transformation; performs iterative analysis on the production data after multi-dimensional tensor transformation, extracts feature data, and reconstructs based on the feature data to obtain the reconstructed production data; transmits the reconstructed production data to the real-time monitoring module and the visual data display module;

[0027] The real-time monitoring module monitors the production process in real time, establishes a dynamic tensor model of the production process based on the reconstructed production data, and predicts the system state tensor; transmits the predicted system state tensor to the control optimization module and the visual data display module;

[0028] The control optimization module defines the objective function based on the predicted system state tensor, optimizes the control input tensor, and obtains the optimal control input tensor by optimizing the objective function; the optimal control input tensor is transmitted to the parameter adjustment module;

[0029] The parameter adjustment module applies the optimal control input tensor to the actual production parameter adjustment, monitors the status of the production process in real time, and obtains feedback data. Based on the feedback data, the production parameters are optimized through an adaptive multi-nested variational optimization algorithm to obtain the adjusted production parameters; the production process is iteratively adjusted to obtain the global optimal production parameters; and the adjusted production parameters are transmitted to the visual data display module;

[0030] The visual data display module displays the reconstructed production data output by the data processing module, the predicted system state tensor output by the real-time monitoring module, and the adjusted production parameters output by the parameter adjustment module in real time in a graphical manner.

[0031] The beneficial effects of the technical solution of the present invention are:

[0032] 1. The present invention adopts a multidimensional tensor iterative transformation analysis algorithm, and effectively processes and analyzes the production data collected during the production process through multidimensional tensor transformation and complex iterative steps; through multidimensional tensor transformation, the production data can be separated into components of different frequencies, and multidimensional features can be extracted to achieve efficient processing and analysis of the production data; the iterative analysis and reconstruction steps further ensure the accuracy and speed of feature extraction, and finally obtain high-quality cleaned and filtered production data;

[0033] 2. Using the adaptive dynamic tensor control algorithm, a dynamic tensor model of the production process is established, which can monitor the production status of the MVR scraper crystallizer in real time; the dynamic tensor model of the production process not only describes the change of the control system state tensor over time, but also considers the impact of the reconstructed production data and control input tensor on the system state tensor; through the dynamic tensor model of the production process, the system state tensor can be accurately predicted, providing a scientific basis for the adjustment of the control strategy;

[0034] 3. Through the objective function defined by the control optimization module, the control system can evaluate the effect of the control strategy in multiple time steps and determine the optimal control input tensor by solving the optimization problem; the optimal control input tensor contains the optimal values ​​of different control variables, such as temperature, pressure, flow, etc.; the parameter adjustment module applies the optimal control input tensor to the actual production parameter adjustment to ensure that the production process runs in the optimal state;

[0035] 4. The present invention introduces an adaptive multiple nested variational optimization algorithm. Through the adaptive multiple nested variational process and collaborative optimization, the production parameters are gradually optimized to ensure the convergence and optimization effect of the production parameters; the collaborative optimization ensures the coordination between multiple production parameters to achieve the global optimum; through continuous monitoring and feedback, the production process is continuously iterated and optimized, and ultimately intelligent optimal control is achieved;

[0036] 5. The control system obtains feedback data by real-time monitoring of the status of the production process, and uses the feedback data to evaluate the actual production effect; the next round of production parameters are optimized through the adaptive multiple nested variational optimization algorithm, ensuring the dynamic adjustment and optimization of the production process; the continuous monitoring and feedback mechanism ensures the efficiency and reliability of the control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1It is a visualization system structure diagram of the whole production process of the MVR scraper crystallizer according to the present invention;

[0038] Figure 2 It is a flow chart of the visualization method of the whole production process of the MVR scraper crystallizer described in the present invention. DETAILED DESCRIPTION

[0039] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0041] The specific scheme of the visualization system and method for the whole production process of the MVR scraper crystallizer provided by the present invention is described in detail below with reference to the accompanying drawings.

[0042] See attached Figure 1 , which shows a visualization system structure diagram of the entire production process of an MVR scraper crystallizer provided by an embodiment of the present invention, and the system includes the following parts:

[0043] Data acquisition module, data processing module, real-time monitoring module, control optimization module, parameter adjustment module, and visual data display module;

[0044] The data acquisition module collects production data such as temperature, pressure, flow, crystallization speed, etc. in real time during the production process through sensors installed at key parts of the MVR scraper crystallizer; and transmits the production data to the data processing module in real time through wireless transmission technology;

[0045] The data processing module cleans and filters the collected data, and analyzes the cleaned and filtered production data through a multidimensional tensor iterative transformation analysis algorithm; performs preliminary processing on the cleaned and filtered production data through multidimensional tensor transformation to separate components of different frequencies and obtain production data after multidimensional tensor transformation; performs iterative analysis on the production data after multidimensional tensor transformation, extracts feature data, and reconstructs based on the feature data to obtain reconstructed production data; transmits the reconstructed production data to the real-time monitoring module and the visual data display module;

[0046] The real-time monitoring module monitors the production process in real time, establishes a dynamic tensor model of the production process based on the reconstructed production data, describes the time-varying characteristics of the control system, and predicts the system state tensor to facilitate timely adjustment of the control strategy; the predicted system state tensor is transmitted to the control optimization module and the visual data display module;

[0047] The control optimization module defines the objective function based on the predicted system state tensor, optimizes the control input tensor, determines the best control input tensor by optimizing the objective function, and obtains the optimal control input tensor; transmits the optimal control input tensor to the parameter adjustment module;

[0048] The parameter adjustment module applies the optimal control input tensor to the actual production parameter adjustment, monitors the status of the production process in real time, and obtains feedback data. Based on the feedback data, the production parameters are optimized through an adaptive multi-nested variational optimization algorithm to obtain the adjusted production parameters; the production process is iteratively adjusted to obtain the global optimal production parameters; and the adjusted production parameters are transmitted to the visual data display module;

[0049] The visual data display module displays the reconstructed production data output by the data processing module, the predicted system state tensor output by the real-time monitoring module, and the adjusted production parameters output by the parameter adjustment module in real time in a graphical manner.

[0050] See attached Figure 2 , which shows a flow chart of a visualization method for the entire production process of an MVR scraper crystallizer provided by an embodiment of the present invention, the method comprising the following steps:

[0051] S1. Real-time collection of production data, cleaning and filtering of the production data; analysis and reconstruction of the cleaned and filtered production data through a multi-dimensional tensor iterative transformation analysis algorithm to obtain reconstructed production data, and display the reconstructed production data in real time in a graphical manner;

[0052] The data acquisition module is used to uniformly manage the scattered production process data. The data acquisition module uses sensors (such as temperature sensors, pressure sensors, flow sensors, etc.) installed at key positions of the MVR scraper crystallizer to collect production data such as temperature, pressure, flow, crystallization speed, etc. in real time during the production process. The collected production data is transmitted to the data processing module in real time through wireless transmission technology to ensure the real-time and accuracy of the production data.

[0053] The data processing module cleans and filters the collected production data, designs a multidimensional tensor iterative transformation analysis algorithm to analyze the cleaned and filtered production data, and ensures the efficiency and accuracy of production data processing by introducing multidimensional tensor transformation and complex iterative steps; the cleaned and filtered production data is preliminarily processed through multidimensional tensor transformation to separate components of different frequencies. The specific calculation formula of multidimensional tensor transformation is:

[0054] ,

[0055] in, Represents the production data after multi-dimensional tensor transformation; is the weight, used to adjust each dimension , , , The interaction between them and their impact on the final result (i.e., the data after multidimensional tensor transformation) ) , , , The four dimensions of the input cleaned and filtered production data represent different physical quantities, such as temperature, pressure, flow rate, and time; is with dimension The relevant frequency coefficient, expressed in Oscillation frequency in dimension; is with dimension The relevant frequency coefficient, expressed in Oscillation frequency in dimension; is with dimension The relevant attenuation coefficient, expressed in The decay rate in dimension; is with dimension The relevant frequency coefficient, expressed in Oscillation frequency in dimension; are integer variables used for indexing, ranging from 1 to , 1 to , 1 to , 1 to , represents the number of expanded items in each dimension of the production data after cleaning and filtering; It is the total number of items in each dimension of the cleaned and filtered production data. The cleaned and filtered production data is mapped to the tensor space, and different data features are extracted by setting different weights and frequency parameters, ensuring the multi-dimensional separation of the production data.

[0056] Among them, in the visualization method of the whole process of MVR scraper crystallizer production, the weight Directly affects the extraction effect of production data features in different dimensions;

[0057] Weight The specific value range of:

[0058] The value range of is set to [0,1], indicating that the influence of the weight on different combinations of dimensions and frequencies varies from no influence (0) to maximum influence (1);

[0059] In practical applications, the values ​​of weights will be fine-tuned according to the specific application scenarios. For example, the optimal weight combination can be determined using empirical data or through optimization algorithms (such as gradient descent, genetic algorithms, etc.).

[0060] The specific steps for weight setting are as follows:

[0061] Initial setting: Initial setting is available is the default value, which means that each dimension and frequency combination has an equal influence on the result of multidimensional tensor transformation;

[0062] Gradual adjustment: According to the production data characteristics and goals in actual applications, the weights of specific dimensions or frequency combinations are gradually adjusted by increasing or decreasing them. ;

[0063] Automatic parameter adjustment: Use algorithms (such as random search, Bayesian optimization, etc.) to automatically adjust weights.

[0064] Weight value examples:

[0065] Assuming I = J = K = L = 3, the weight matrix It can be set as follows:

[0066] : represents the high influence of temperature at low frequency;

[0067] : represents the medium influence of pressure at medium frequency;

[0068] : represents the low influence of traffic at high frequency;

[0069] The weights of other combinations can be adjusted according to specific circumstances.

[0070] By iteratively analyzing the production data after multi-dimensional tensor transformation, characteristic data is extracted. The iterative analysis formula is:

[0071] ,

[0072] ,

[0073] in, Indicates The feature data extracted by the iteration; Indicates The feature data extracted by the iteration; is the starting feature data; and is the iteration parameter; is the gradient, ; Through high-order partial derivatives and gradient iteration, high-frequency and low-frequency features are extracted from the production data after multi-dimensional tensor transformation. Control the contribution of high-order partial derivatives and affect the accuracy of feature extraction; Control the contribution of the gradient term and affect the speed of feature extraction;

[0074] Among them, the iteration parameter and The process of obtaining and quantifying is as follows:

[0075] and Initial value: Set the initial value based on expert experience or through simulation experiments. For different data types and application scenarios, you can test the data to observe the impact of the initial value of the iteration parameter. A common practice is to set and The initial value of is set to a positive number less than 1, usually between 0.01 and 0.1.

[0076] The optimal combination of iteration parameters is selected through Bayesian optimization model prediction. In the process of iterative parameter tuning, cross-validation (such as K-fold cross-validation) is usually used to evaluate the performance of each set of iteration parameters. Through multiple training and validation, the average performance index of each set of iteration parameters is calculated to select the best iteration parameters.

[0077] The feature data extracted by iterative analysis is reconstructed, and the specific calculation formula is:

[0078] ,

[0079] in, is the reconstructed production data; , , , is the extracted feature component; is the normalization coefficient. The extracted feature data are fused and reconstructed to obtain the reconstructed production data. Through the multi-level reconstruction process, the high quality and high reliability of the reconstructed production data are ensured.

[0080] The visualization data display module displays the reconstructed production data in real time in a graphical manner, using multi-dimensional graphical representations such as heat maps, line graphs, pie charts and three-dimensional graphs to dynamically display the changing trends and states of key production parameters during the production process. Among them, the heat map represents the distribution of production data collected by each sensor, and the color depth intuitively reflects the high and low data values; the line graph is used to dynamically display the changing trends of temperature, pressure, flow, etc. over time; the three-dimensional graph displays the production data after multi-dimensional tensor transformation in a three-dimensional space.

[0081] S2. Based on the reconstructed production data and control input tensor, a dynamic tensor model of the production process is established through an adaptive dynamic tensor control algorithm to predict the system state tensor and visualize the predicted system state tensor; the optimal control input tensor is determined by defining and optimizing the objective function;

[0082] The reconstructed production data It is passed as input to the real-time monitoring module, which monitors the production process in real time and establishes a dynamic tensor model of the production process through an adaptive dynamic tensor control algorithm. The specific formula is:

[0083] ,

[0084] in, is the system state tensor, which indicates that the control system is The state tensor at the moment; Indicates that the control system is The state tensor at the moment; is the control input tensor, indicating the control strategy in The role of moments; , , They are all time-varying tensor matrices, which are used to describe the relationship between the system state tensor and time, the impact of the reconstructed production data on the system state tensor, and the impact of the control input tensor on the system state tensor; is a noise tensor, which represents the random disturbance in the control system. By constructing a dynamic tensor model of the production process to describe the changes in the system state tensor in the production process and the time-varying characteristics of the control system, the dynamics and accuracy of the dynamic tensor model of the production process are ensured;

[0085] Among them, the time-varying tensor matrix , , The determination of usually involves multiple steps such as system modeling, data analysis, and optimization. The following are common methods for obtaining and determining the time-varying tensor matrix:

[0086] Physical laws and mechanism modeling: First, establish a basic physical model based on the physical laws of the production process, process characteristics and equipment parameters. For example, in the MVR scraper crystallizer system, The matrix can represent the autoregressive properties of the control system state, The matrix can represent the reconstructed production data The impact on control systems, The matrix represents the effect of the control input tensor on the control system. The initial form of the time-varying tensor matrix is ​​derived through physical laws (such as conservation of mass, conservation of energy, etc.);

[0087] Time-varying considerations: Due to the dynamic characteristics of the production process, the time-varying tensor matrix may not be static. In order to capture the time-varying characteristics of the control system, the physical model needs to be extended to consider the influence of the control system parameters changing over time, which may include the changing laws of production process parameters such as temperature, pressure, and flow.

[0088] The dynamic tensor model of the production process is used to predict the system state tensor. The specific formula is:

[0089] ,

[0090] in, is the predicted system state tensor; is the prediction step length, which indicates the number of time steps for prediction; It is the index variable of the prediction step length, which is used for accumulation in the prediction process. The future system state tensor is estimated through the dynamic tensor model of the production process, so as to adjust the control strategy in time; by setting an appropriate prediction step length , achieving accurate prediction of future system state tensors.

[0091] The visual data display module uses visualization tools to display the predicted system state tensor using color coding or dynamic curves. Different colors represent different areas of system status over time, making it easy to quickly identify abnormal trends.

[0092] The control optimization module defines the objective function and optimizes the control input tensor. The specific calculation formula is:

[0093] ,

[0094] in, is the objective function; express The reconstructed production data at each moment represents an ideal description of the production process in a multidimensional space and serves as the reference trajectory tensor of the control system, i.e., the target value; is the total number of time steps considered in the entire optimization process, and the objective function will be The sum is performed over time steps to evaluate the effect of the control strategy and optimize the control input tensor; is a weight matrix used to adjust the influence of the system state tensor error in the objective function; is the weight matrix, which is used to adjust the influence of the control input tensor in the objective function; Represents the transpose of the matrix. The optimal control input tensor is determined by optimizing the objective function. The weight parameters in the objective function and , can be adjusted through experimental data to ensure the optimality of the control strategy.

[0095] By optimizing the objective function, the optimal control input tensor is solved. The specific formula is:

[0096] ,

[0097] in, The optimal control input tensor is obtained by solving the optimization problem to obtain the best control strategy, ensuring that the control system operates in the optimal state. The accuracy and efficiency of the control input tensor are guaranteed through iterative optimization in the solution process. The method for solving the optimization problem can be adopted by existing technology.

[0098] S3. Apply the optimal control input tensor to the actual production parameters, monitor and feedback data in real time, optimize the production parameters through the adaptive multi-nested variational optimization algorithm, and display the adjustment status of the actual production parameters and feedback data in real time in a graphical manner, ultimately achieving intelligent control.

[0099] The parameter adjustment module inputs the optimal control tensor Applied to actual production parameter adjustment, optimal control input tensor It contains the optimal values ​​of different control variables, such as temperature, pressure, flow and other production parameters, expressed as: , Indicates The control variables are The optimal value at the moment; is the number of control variables. The optimal control input tensor is applied to each specific controller of the MVR scraper crystallizer to adjust the corresponding production parameters, such as temperature, pressure, flow rate, crystallization speed, etc. Based on the applied optimal control input tensor, the state of the production process is monitored in real time and feedback data is obtained; the feedback data is used to evaluate the actual production effect and serve as the input for the next round of optimization.

[0100] Compare the feedback data with the target value, calculate the feedback error, and adjust the production parameters through the adaptive multi-nested variational optimization algorithm to carry out the next round of optimization. The specific calculation formula is:

[0101] ,

[0102] ,

[0103] in, is the adjusted production parameter; is the production parameter before adjustment; is the learning rate; is the loss function; is the variational parameter; is the number of variational parameters; yes The feedback data at the moment indicates that the control system is The actual output production parameters at all times; is the regularization parameter; It is The production parameters are gradually optimized by using an adaptive multiple nested variational optimization algorithm, ensuring the gradual convergence and optimization effect of the production parameters.

[0104] In the process of implementing the adaptive multi-nested variational optimization algorithm, collaborative optimization is performed to ensure the global optimal solution. The specific formula is:

[0105] ,

[0106] in, To optimize the rate of collaboration; It is Through collaborative optimization, the coordination among multiple production parameters is ensured to achieve global optimum, thus ensuring the globality and consistency of overall parameter optimization.

[0107] The adjusted variational parameters Applied to the control system, it can realize dynamic adjustment and optimization of the production parameters of MVR scraper crystallizer. Through continuous monitoring and feedback, combined with adaptive multi-nested variational optimization algorithm, it continuously iterates and optimizes the production process, and finally realizes intelligent optimal control.

[0108] The visual data display module monitors and displays the changes and adjustments of actual production parameters in real time through a visual interface. Through the data dashboard, operators can view the status of key production parameters in real time and monitor the feedback data of production effects through dynamic charts.

[0109] In summary, the visualization system and method for the entire production process of MVR scraper crystallizer have been completed.

[0110] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0112] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. Visualization method of the whole production process of MVR scraper crystallizer, characterized in that: The following steps are involved: S1. Real-time collection of production data, cleaning and filtering of the production data; analysis and reconstruction of the cleaned and filtered production data through a multi-dimensional tensor iterative transformation analysis algorithm to obtain reconstructed production data, and display the reconstructed production data in real time in a graphical manner; The multidimensional tensor iterative transformation analysis algorithm specifically includes: The cleaned and filtered production data is preliminarily processed through multidimensional tensor transformation, the cleaned and filtered production data is mapped to the tensor space, and components of different frequencies are separated to obtain the production data after multidimensional tensor transformation; the specific calculation formula of multidimensional tensor transformation is: ， in, Represents the production data after multi-dimensional tensor transformation; is the weight, used to adjust each dimension , , , The interaction between them and the data after multidimensional tensor transformation The impact of , , , The four dimensions of the input cleansed and filtered production data; Is with dimension The relevant frequency coefficients; Is with dimension The relevant frequency coefficients; Is with dimension The associated attenuation coefficient; Is with dimension The relevant frequency coefficients; is an integer variable used for indexing; is the total number of items in each dimension of the production data after cleaning and filtering; Iteratively analyze the production data after multi-dimensional tensor transformation to extract feature data. The iterative analysis formula is: ， ， in, Indicates The feature data extracted by the iteration; Indicates The feature data extracted by the iteration; is the starting feature data; and is the iteration parameter, Used to control the contribution of higher-order partial derivatives, Used to control the contribution of the gradient term; is the gradient, ; Reconstructing the characteristic data extracted through iterative analysis to obtain reconstructed production data; S2. Based on the reconstructed production data and control input tensor, the dynamic tensor model of the production process is established through the adaptive dynamic tensor control algorithm, the system state tensor is predicted, and the predicted system state tensor is visualized; the specific formula of the adaptive dynamic tensor control algorithm is: ， in, is the system state tensor, which indicates that the control system is The state tensor at the moment; Indicates that the control system is The state tensor at the moment; is the control input tensor; , , They are all time-varying tensor matrices, which are used to describe the relationship between the system state tensor and time, the impact of the reconstructed production data on the system state tensor, and the impact of the control input tensor on the system state tensor; is the noise tensor; Represents the reconstructed production data; Determine the optimal control input tensor by defining and optimizing the objective function; S3, apply the optimal control input tensor to the actual production parameters, monitor and feedback data in real time, compare the feedback data with the target value, calculate the feedback error, and optimize the production parameters through the adaptive multi-nested variational optimization algorithm. The specific calculation formula is: ， ， in, is the adjusted production parameter; is the production parameter before adjustment; is the learning rate; is the loss function; is the variational parameter; is the number of variational parameters; yes Feedback data at all times; express The production data reconstructed at each moment represents the target value; represents feedback error; is the regularization parameter; It is Variational parameters; The actual production parameter adjustment and feedback data are displayed in real time in a graphical manner, ultimately achieving intelligent control.

2. The visualization method of the whole production process of MVR scraper crystallizer according to claim 1 is characterized in that: In said S1, it specifically includes: Iterative analysis is performed on the production data after multi-dimensional tensor transformation, and high-frequency and low-frequency features in the production data after multi-dimensional tensor transformation are extracted through high-order partial derivatives and gradient iteration to obtain feature data.

3. The visualization method of the whole production process of MVR scraper crystallizer according to claim 2 is characterized in that: In the S1, it also includes: Reconstruction is performed based on the extracted feature data to obtain reconstructed production data.

4. The visualization method of the whole production process of MVR scraper crystallizer according to claim 1 is characterized in that: In said S3, it specifically includes: During the implementation of the adaptive multi-nested variational optimization algorithm, collaborative optimization is performed to obtain the global optimal production parameters and ultimately achieve intelligent control.

5. A visualization system for the entire production process of an MVR scraper crystallizer, applied to the visualization method for the entire production process of an MVR scraper crystallizer as claimed in claim 1, characterized in that: Includes the following sections: Data acquisition module, data processing module, real-time monitoring module, control optimization module, parameter adjustment module, and visual data display module; The data acquisition module collects production data in the production process in real time through sensors; and transmits the production data to the data processing module in real time; The data processing module cleans and filters the production data, and analyzes the cleaned and filtered production data through a multi-dimensional tensor iterative transformation analysis algorithm; Performing preliminary processing on the cleaned and filtered production data through multi-dimensional tensor transformation to obtain production data after multi-dimensional tensor transformation; Iteratively analyzing the production data after multi-dimensional tensor transformation, extracting feature data, and reconstructing based on the feature data to obtain reconstructed production data; Transmit the reconstructed production data to the real-time monitoring module and the visual data display module; The real-time monitoring module monitors the production process in real time, establishes a dynamic tensor model of the production process based on the reconstructed production data, and predicts the system state tensor; transmits the predicted system state tensor to the control optimization module and the visual data display module; The control optimization module defines the objective function based on the predicted system state tensor, optimizes the control input tensor, and obtains the optimal control input tensor by optimizing the objective function; the optimal control input tensor is transmitted to the parameter adjustment module; The parameter adjustment module applies the optimal control input tensor to the actual production parameter adjustment, monitors the status of the production process in real time, and obtains feedback data. Based on the feedback data, the production parameters are optimized through an adaptive multi-nested variational optimization algorithm to obtain the adjusted production parameters. Iteratively adjust the production process to obtain the global optimal production parameters; transmit the adjusted production parameters to the visual data display module; The visual data display module displays the reconstructed production data output by the data processing module, the predicted system state tensor output by the real-time monitoring module, and the adjusted production parameters output by the parameter adjustment module in real time in a graphical manner.

Images