On-Demand Analytics Modules for Real-Time Control Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial monitoring and control systems face challenges in achieving real-time optimization due to the need for complex and specific models, which often require significant manual input and adaptation, limiting their flexibility and computational efficiency.
Innovation Solution
A modular analytics engine system that includes sensors, actuators, and an automation controller, capable of instantiating analytics modules for modeling, optimization, classification, and control operations based on annotated data structures, allowing for adaptive and agnostic processing that can operate independently and in various configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a model predictive control (MPC) system uses a simplified process model for computational efficiency, then real-time optimization performance is improved, but model quality and controller performance deteriorate
Solution Approach 1:
The patent segments the modeling and control system into multiple independent analytics modules (e.g., system identification module, state estimation module, optimization module) that can be selectively instantiated. This allows the MPC system to use simplified models for fast real-time control while maintaining the option to use more complex models for offline analysis or specific critical operations, thus resolving the contradiction between computational speed and model quality.
Solution Approach 2:
The patent implements dynamic model instantiation where the complexity and type of analytics modules are adjusted in real-time based on operational conditions. The system can dynamically select between simplified and complex models, or instantiate different module configurations depending on the control horizon, process state, and computational resources available, allowing optimal balance between speed and accuracy at different times.
2Reliability
If MPC systems use complex and specific models for accurate process representation, then controller performance is improved, but device complexity and manual adaptation requirements increase
Solution Approach 1:
The patent creates a universal analytics engine framework that can handle multiple types of processes and control problems through a common set of modular analytics modules. These modules are designed to be application-agnostic and can be configured for different processes through parameter settings rather than structural changes, reducing the need for complex custom modeling for each specific application while maintaining high controller performance.
Solution Approach 2:
The patent implements self-identifying analytics modules that automatically detect process characteristics, select appropriate modeling approaches, and configure themselves based on incoming data. The system identification module can automatically identify process dynamics and parameters without extensive manual tuning, and the state estimation module adapts to different process states autonomously, significantly reducing manual adaptation requirements while maintaining reliable control performance.
3Adaptability or versatility
If analytics modules are instantiated upon request in a modular fashion, then adaptability and flexibility are improved, but system complexity increases
Solution Approach 1:
The patent introduces a standardized data structure intermediary (annotated input data structure with standardized fields and schemas) that mediates between diverse analytics modules and the underlying process data. This standardized interface allows different modules to be instantiated and configured independently without increasing overall system complexity, as all modules communicate through the same well-defined data structure, enabling high adaptability while maintaining architectural simplicity.
Data Source
AI summary
An analytics engine is provided for industrial automation applications. The engine may be modular, and may be instantiated upon receipt of a data structure, such as containing annotated data from or relating to a monitored and/or controlled machine or process. The module may be data-driven so that it is instantiated only as needed, upon receipt of the input data structure. The module then carries out analysis on the data, and outputs a data structure that can be used for further analysis, or directly by other modules for modeling, classification, optimization and/or control.