Support Vector Machine Training With Process Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches for process control, such as direct measurement and computer models, face challenges in effectively measuring and predicting process conditions and product properties, especially when these measurements are difficult, time-consuming, or expensive, leading to inefficiencies in manufacturing and quality control.
Innovation Solution
The development of a system and method for training a support vector machine (SVM) with process constraints, including physical attributes and operational constraints, to predict and manage process conditions and product properties, using a primal or dual formulation and optimizers like SMO or NLP, allowing for real-time optimization and decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct measurement methods are used to measure process conditions and product properties, then measurement accuracy is improved, but measurement time and cost increase
Solution Approach 1:
The patent creates a virtual copy of the physical process through a computational model (support vector machine). This digital twin replicates the behavior of the actual manufacturing process, allowing measurements and predictions to be obtained from the virtual model rather than direct physical measurement, thereby reducing measurement time while maintaining accuracy
Solution Approach 2:
The patent replaces direct physical measurement systems with a computational prediction system. Instead of using physical sensors and measurement devices to directly measure process conditions and product properties, the system uses a trained support vector machine model to predict these values based on input process parameters, eliminating the need for time-consuming direct measurements
2Productivity
If conventional computer models are used for process control, then prediction capability is improved, but model accuracy for complex non-linear processes deteriorates
Solution Approach 1:
The patent transforms the modeling approach by changing the mathematical parameters and structure of the model. Instead of using traditional linear or simple statistical models, the patent employs a support vector machine with kernel functions that can capture complex non-linear relationships. The model uses optimized parameters including kernel type, regularization parameter C, and gamma parameter to achieve high accuracy in predicting process conditions and product properties
Solution Approach 2:
The patent implements a feedback mechanism where the support vector machine model is trained using historical process data and continuously refined. The model learns from past measurements and predictions, adjusting its internal parameters to improve accuracy over time. This feedback loop allows the model to adapt to changing process conditions and maintain high prediction accuracy for complex non-linear processes
3Reliability
If more comprehensive process constraints are included in the model, then model reliability is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-processing and normalizing the input data before feeding it to the support vector machine model. The system prepares the data in advance by scaling features, handling missing values, and encoding categorical variables, which simplifies the computational burden during actual prediction and improves model reliability without adding complexity during runtime operations
Data Source
AI summary
System and method for training a support vector machine (SVM) with process constraints. A model (primal or dual formulation) implemented with an SVM and representing a plant or process with one or more known attributes is provided. One or more process constraints that correspond to the one or more known attributes are specified, and the model trained subject to the one or more process constraints. The model includes one or more inputs and one or more outputs, as well as one or more gains, each a respective partial derivative of an output with respect to a respective input. The process constraints may include any of: one or more gain constraints, each corresponding to a respective gain; one or more Nth order gain constraints; one or more input constraints; and/or one or more output constraints. The trained model may then be used to control or manage the plant or process.