Operator Tool for Non-Linear Process Parameter Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current process monitoring and diagnosis tools in industrial settings face challenges in identifying non-linear relationships between process parameters, leading to inefficient problem identification and control actions, particularly due to the limitations of linear methods like PLS and PCA, which can provide misleading results and require significant processing capacity.
Innovation Solution
A method that utilizes non-real-time processing to build statistical models describing non-linear causal interrelations between variables, organizing them hierarchically and storing them for real-time analysis, allowing operators to select key variables and dynamically organize models to identify the most impactful explanatory variables, thereby facilitating effective control actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If linear methods like PLS and PCA are used for process monitoring, then the implementation is simple and processing capacity requirements are reduced, but the identification of non-linear relationships between process parameters becomes inaccurate and misleading
Solution Approach 1:
The patent transforms the statistical modeling approach by changing from linear methods (PLS, PCA) to non-linear methods (neural networks, support vector machines, genetic algorithms). This parameter change in the mathematical model enables accurate capture of non-linear relationships between process parameters while maintaining computational feasibility through the use of modern algorithms that can handle non-linearity without requiring excessive processing capacity.
Solution Approach 2:
The patent substitutes traditional mechanical statistical methods (linear regression-based PLS and PCA) with computational intelligence methods (neural networks, support vector machines). This substitution replaces the limited linear modeling capability with more sophisticated computational models that can naturally represent non-linear relationships, thereby improving measurement precision without proportionally increasing device complexity.
2Measurement precision
If sophisticated non-linear statistical models are implemented in real-time, then the identification of process parameters becomes more accurate, but the processing capacity requirements and computational time increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-training sophisticated non-linear models (neural networks, support vector machines) offline using historical process data. During real-time operation, only inference is performed on the pre-trained models, which requires significantly less computational power than full training. This separates the computationally intensive model development phase from the real-time monitoring phase, enabling accurate non-linear relationship identification without excessive processing capacity requirements during actual process control.
Solution Approach 2:
The patent segments the statistical modeling process into distinct phases: offline model training and development, and online model inference and application. This segmentation allows complex non-linear models to be built and optimized using abundant computational resources during the offline phase, while the online phase requires minimal processing power for real-time parameter identification, thereby resolving the contradiction between model sophistication and real-time processing capacity.
3Reliability
If comprehensive process data is analyzed in real-time, then the monitoring accuracy and problem identification capability improve, but the processing time and computational load increase
Solution Approach 1:
The patent performs preliminary data processing and model training offline, preparing pre-computed statistical models and parameter relationships before real-time monitoring begins. During real-time operation, the system only needs to evaluate pre-trained models against current process data, which dramatically reduces processing time while maintaining comprehensive analysis of process parameters. This enables high reliability monitoring without excessive computational delay.
Solution Approach 2:
The patent implements dynamic adaptability by using neural networks and support vector machines that can learn and adapt to changing process conditions. These dynamic models can adjust to new operating regimes and identify emerging problems without requiring complete re-analysis of all historical data, thereby maintaining high monitoring accuracy while reducing processing time as the system adapts to familiar patterns over time.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An operator tool is provided to assist an operator of a process or machine in decision making. The tool is initiated by the operator selecting a key variable, "Variable of Interest", at the user interface (20). The variable of interest usually is an output variable of interest (such as quality, cost, etc.). At least most significant variables related to the selected variable of interest are automatically determined by a statistical method and shown at the user interface (22). The operator can adjust the related variables shown on the interface (24). The impact of one variable to another is demonstrated by a prediction method and shown in a numerical and/or graphical form. After being satisfied with the result of the analysis, the operator decides on which is the preferred parameter change in order to overcome the problem in question or achieve the desired improvement, and then implements the corresponding change on the real system (28).