Industrial Control Tags for ML-Based Variable Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial manufacturing environments face challenges in extracting enterprise-level insights due to the vast amount of data generated quickly, and existing technologies struggle to effectively integrate machine learning models into industrial automation environments to optimize target variables.
Innovation Solution
A system comprising a design application and a processing application that selects program tags representing target variables, identifies correlated independent variables, and modifies control programs to optimize target variables based on determined correlations, facilitating the integration of machine learning models into industrial automation environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning models are integrated into industrial automation environments, then the ability to optimize target variables is improved, but the complexity of the programming system increases
Solution Approach 1:
The patent introduces a machine learning model as an intermediary component that sits between the existing control system and the optimization goal. This model receives data from control tags and outputs optimized control actions, allowing the system to gain advanced optimization capabilities without fundamentally restructuring the entire programming system. The model acts as a mediator that translates between traditional control data and optimized control decisions.
Solution Approach 2:
The patent segments the optimization function into a separate machine learning model component rather than embedding it throughout the entire control program. This allows the complex optimization logic to be isolated in a dedicated model that can be trained and updated independently, while the rest of the control system remains unchanged. The segmentation enables incremental integration and reduces overall system complexity.
2Adaptability or versatility
If machine learning algorithms are used to recognize patterns and improve through training, then the ability to learn from examples is improved, but the difficulty of integration into industrial programming systems increases
Solution Approach 1:
The patent uses copying by creating a virtual representation of the industrial process within the machine learning model. The model learns from historical data that copies past process behaviors and outcomes, allowing it to recognize patterns without requiring direct modification of the physical system or complex integration protocols. This copying approach simplifies integration by working with data replicas rather than the actual control infrastructure.
Solution Approach 2:
The machine learning model serves as an intermediary layer that handles the complexity of pattern recognition and learning, shielding the industrial programming system from these complexities. The model translates learned patterns into standard control commands that the existing system can execute, making the learning capability accessible without requiring the host system to understand or manage the learning process directly.
3Quantity of substance
If a large number of program tags are available to represent industrial assets and devices, then the coverage of industrial variables is improved, but the difficulty of targeting specific variables increases
Solution Approach 1:
The patent implements feedback through the machine learning model that continuously analyzes data from multiple program tags and provides optimized control actions. The model receives feedback from process outcomes and adjusts its predictions accordingly, enabling it to identify which specific tags are most relevant for optimization goals. This feedback mechanism automatically filters through the large number of available tags to identify the critical ones.
Solution Approach 2:
The patent replaces manual variable targeting (a mechanical/search-based approach) with a data-driven machine learning approach. Instead of programmers manually searching through numerous tags to find relevant variables, the ML model automatically identifies and weights the importance of different tags based on their relationship to optimization goals, substituting computational intelligence for manual search and selection processes.
Data Source
AI summary
Various embodiments of the present technology generally relate to industrial automation environments. More specifically, embodiments include systems and methods to optimize a target variable in an industrial automation environment. In some examples, a design application generates a control program configured and selects a program tag that represents a target variable in an industrial process. A processing application identifies a set of available program tags that represent independent variables in the industrial process and determines correlations between ones of the independent variables and the target variable. The processing application selects available program tags that represent independent variables correlated with the target variable and generates a recommendation that indicates the selected available program tags. The design application modifies the control program using the selected available program tags to optimize the target variable. The design application transfers the control program for implementation by the programmable logic controller.