Modular Analytics Engine for Real-Time Industrial 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 are not adaptable to various applications, limiting their computational efficiency and flexibility.
Innovation Solution
A modular analytics engine system that uses agnostic, data-driven tools for modeling, optimization, classification, and control, allowing for flexible deployment and adaptation to different applications without prior knowledge of the system or process, utilizing a set of base functions to build models from variables of interest and enabling modular operations in series or parallel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a model predictive control (MPC) system uses a detailed process model to optimize performance, then controller performance is improved, but computational efficiency deteriorates
Solution Approach 1:
The system segments the modeling and control architecture into multiple levels: a detailed process model runs offline to generate training data, while a simplified neural network model executes online for real-time control decisions. This segmentation allows the detailed model to inform the simplified model without burdening real-time computation.
Solution Approach 2:
The system performs preliminary modeling work offline by training a neural network model using data from a detailed process model. This preliminary action creates a pre-computed, simplified model that can be rapidly executed during real-time control without requiring the computational resources of the original detailed model.
2Measurement precision
If a control system uses application-specific models tailored to particular machines, then model accuracy is improved, but adaptability to different applications deteriorates
Solution Approach 1:
The system employs universal neural network modeling techniques that can be applied across different industrial applications. The same neural network architecture and training methodology serve multiple purposes: process modeling, fault detection, performance optimization, and control, making the approach adaptable to various machines and processes while maintaining accuracy through data-driven customization to each specific application.
3Reliability
If sophisticated monitoring and control approaches are developed for specific applications, then performance optimization is improved, but ease of deployment to other applications deteriorates
Solution Approach 1:
The neural network model serves multiple functions automatically without requiring separate systems for each task. The same model structure performs process modeling, fault detection, performance optimization, and control operations, enabling the system to adapt to different applications through data-driven learning rather than requiring manual reconfiguration or redeployment of application-specific models.
Data Source
AI summary
A modular modeling engine is provided for industrial automation applications. The module may be instantiated upon demand, such as upon receipt of annotated data for a system or process being monitored and/or controlled. The model is agnostic insomuch as little or no prior knowledge is required of the system or process. Variables, functions, and their combinations are selected and the model is refined automatically. A data structure is received for instantiation of the model, and following modeling, a similar data structure is produced. The module may be used together with other modules for caning out complex automation processing at the same or multiple levels in an automation setting.