Search

KR-102962124-B1 - APPARATUS AND METHOD FOR PREDICTING AND CONTROLLING PROCESS PERFORMANCE BASED ON ARTIFICIAL INTELLIGENCE

KR102962124B1KR 102962124 B1KR102962124 B1KR 102962124B1KR-102962124-B1

Abstract

The objective of the present invention is to provide an AI-based process performance prediction and control device and method that enable process performance to be predicted and controlled with minimal computing resources using a tree-based AI model. To achieve the above objective, the present invention is characterized by comprising: a learning unit that trains a tree-based artificial intelligence model using training data; a data collection unit that collects process data used for monitoring the operating status during process execution; and a prediction unit that inputs the collected process data into the trained tree-based artificial intelligence model to predict process performance.

Inventors

  • 손문
  • 자히드 울라

Assignees

  • 한국과학기술연구원

Dates

Publication Date
20260511
Application Date
20230509

Claims (13)

  1. A learning unit that trains a tree-based artificial intelligence model using training data; An importance determination unit that identifies the importance of each input factor by quantifying the influence of each input factor on process performance prediction using the above-mentioned learned tree-based artificial intelligence model; A data collection unit that collects process data used for monitoring the operating status during process execution; A prediction unit that predicts process performance by inputting the collected process data into the above-mentioned learned tree-based artificial intelligence model; and A control unit that controls input parameters so that the process performance is included within a preset process performance range according to the predicted process performance above; The above importance judgment unit is, By using the learned tree-based artificial intelligence model to calculate the prediction error while removing the influence of any one of the input factors through Permutation Feature Importance analysis, the influence of each input factor on process performance prediction is quantified—wherein, the method of removing the influence of an input factor is implemented by randomly shuffling the values of the corresponding input factor to make the corresponding input factor noise, The above control unit is, Characterized by performing control starting from the input factor with the highest importance based on the importance of each input factor identified by the importance determination unit above. AI-based process performance prediction and control device.
  2. delete
  3. delete
  4. In paragraph 1, The above learning unit is, Characterized by receiving process identification information that distinguishes each process together with the training data when the process consists of multiple processes, and training the tree-based artificial intelligence model. AI-based process performance prediction and control device.
  5. In paragraph 4, The above prediction unit is, Characterized by receiving the above process data and process identification information corresponding to each process data together to predict process performance. AI-based process performance prediction and control device.
  6. delete
  7. A step of training a tree-based artificial intelligence model using training data; A step of quantifying the influence of each input factor on process performance prediction using the above-mentioned learned tree-based artificial intelligence model to identify the importance of each input factor; A step of collecting process data used for monitoring the operating status during process execution; and A step of predicting process performance by inputting the collected process data into the above-mentioned learned tree-based artificial intelligence model; and The method includes the step of controlling input factors so that the process performance is included within a preset process performance range according to the predicted process performance above; and In the step of identifying the importance of each of the above input factors By using the learned tree-based artificial intelligence model to calculate the prediction error while removing the influence of any one of the input factors among the above input factors through Permutation Feature Importance analysis, the influence of each input factor on process performance prediction is quantified—wherein, the method of removing the influence of an input factor is implemented by randomly shuffling the values of the corresponding input factor to make the corresponding input factor into noise, and In the step of controlling the above input parameters, Characterized by performing control starting from the input factor with the highest importance, based on the importance of each input factor identified in the step of identifying the importance of each input factor above. Artificial intelligence-based process performance prediction and control method.
  8. delete
  9. delete
  10. In Paragraph 7, The step of training the above tree-based artificial intelligence model is, Characterized by a step of training the tree-based artificial intelligence model by receiving process identification information that distinguishes each process together with the training data when the process consists of multiple processes. Artificial intelligence-based process performance prediction and control method.
  11. In Paragraph 10, The step of predicting the above process performance is, The step of predicting process performance by receiving the above-mentioned process data and process identification information corresponding to each process data together, characterized by Artificial intelligence-based process performance prediction and control method.
  12. delete
  13. A computer-readable recording medium having a computer program recorded thereon for performing an artificial intelligence-based process performance prediction and control method according to any one of paragraphs 7, 10, and 11.

Description

Apparatus and Method for Predicting and Controlling Process Performance Based on Artificial Intelligence The present invention relates to an artificial intelligence-based process performance prediction and control device and method, and more specifically, to an artificial intelligence-based process performance prediction and control device and method that enables process performance to be predicted and controlled with minimal computing resources by utilizing a tree-based artificial intelligence model. Chemical, physical, and biological-based oil refining processes, environmental purification processes, and water treatment processes are widely utilized in industries related to chemical engineering and environmental engineering. These processes allow for monitoring of operation status in the form of standardized (tabular) data. In other words, conventional numerical process modeling has been widely utilized due to the distinct correlation between input factors (influent properties, reaction temperature, pressurization, etc.) and output factors (product purity, production yield, etc.). However, since the types of related factors are very diverse, it is difficult to optimize the process beyond a certain level using conventional numerical modeling, making it challenging to perform optimization by considering all factors affecting process performance simultaneously. Therefore, research on process optimization using data-driven machine learning-based artificial intelligence techniques is currently being actively conducted. Machine learning-based artificial intelligence technology is known to offer significantly higher efficiency in process optimization compared to existing technologies when big data is available; however, there is a problem where trained models are overfitted to only a few types of data. In addition, there is a disadvantage that the computing resources and time required for model training increase exponentially as the amount of data increases and the model structure becomes more complex. For example, deep learning models have multiple artificial neural networks called hidden layers hidden between the input and output, and these multiple neural networks are trained through iterative calculations until they reach a stage where they can best predict the correlation between input and output factors. Therefore, deep learning models composed of multiple hidden layers consume a significant amount of computing resources. FIG. 1 is a diagram illustrating a tree-based artificial intelligence model applied to the present invention. FIG. 2 is a schematic diagram showing the configuration of an artificial intelligence-based process performance prediction and control device according to one embodiment of the present invention. FIG. 3 is a diagram illustrating an exemplary process of training a tree-based artificial intelligence model by receiving process identification information together according to the present invention. FIG. 4 is a process diagram illustrating an artificial intelligence-based process performance prediction and control method according to an embodiment of the present invention. Before describing the present invention in detail, it should be understood that the terms and words used in this specification should not be interpreted as being limited to their ordinary or dictionary meanings, and that the inventor of the present invention may appropriately define and use the concepts of various terms to best describe their invention, and furthermore, that these terms and words should be interpreted in a meaning and concept consistent with the technical spirit of the present invention. In other words, it should be understood that the terms used in this specification are used merely to describe preferred embodiments of the present invention and are not intended to specifically limit the content of the present invention, and that these terms are defined in consideration of various possibilities of the present invention. In addition, it should be noted that in this specification, singular expressions may include plural expressions unless the context clearly indicates a different meaning, and that even if they are expressed in a similarly plural form, they may include a singular meaning. Throughout this specification, where it is stated that a component "includes" another component, unless specifically stated otherwise, this may mean that it does not exclude any other component but may include any other component. Furthermore, it should be noted that in cases where it is stated that a component "exists inside or is installed in connection with" another component, this component may be installed in direct connection or contact with the other component, or it may be installed at a certain distance apart, and in the case where it is installed at a certain distance apart, there may be a third component or means for fixing or connecting the component to the other component, and a description of this third c