Stream-Based Asset Operation Control for Multi-Node Deviation Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial control systems in multi-node networks often operate inefficiently due to lack of real-time consideration of inter-component interactions and holistic process conditions, leading to suboptimal performance and prolonged inefficiencies between data collection and adjustment.
Innovation Solution
A stream-based processing system that receives and analyzes data from multiple nodes in real-time or near real-time, using models and analysis engines to identify deviations and adjust operations accordingly, incorporating fuzzy logic and decision tables to recommend or execute corrective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If control systems operate independently at each component level with preset thresholds, then device complexity is reduced and ease of operation is improved, but manufacturing precision and reliability deteriorate due to lack of holistic process consideration
Solution Approach 1:
The system implements feedback by continuously monitoring process data from multiple components and using this information to dynamically adjust control decisions. The control system receives real-time data from various assets, analyzes it against learned normal operating patterns, and provides feedback control actions that consider the holistic state of the industrial process rather than isolated component states.
Solution Approach 2:
The patent introduces an intermediary control system that acts as a mediator between individual component control systems and the overall industrial process. This intermediary layer aggregates data from multiple assets, applies machine learning models to determine normal operating conditions, and coordinates control actions across components to achieve reliable holistic process control.
2Productivity
If real-time stream-based data analysis is implemented across multiple assets, then productivity and reliability are improved through holistic control, but device complexity increases due to integrated multi-node analysis
Solution Approach 1:
The system segments the complex task of holistic process control into manageable components: data collection from individual assets, stream-based real-time analysis, machine learning model execution, and control decision generation. Each segment handles a specific aspect of the control process, allowing the system to manage complexity through modular organization while achieving integrated control outcomes.
Solution Approach 2:
The patent replaces traditional mechanical control systems with fixed preset thresholds and manual coordination with an intelligent software-based control system. This substitution uses stream-based data processing and machine learning algorithms to automatically coordinate multiple assets, reducing the need for complex physical control mechanisms and manual intervention while improving productivity.
3Ease of manufacture
If traditional control systems use preset threshold values for each component, then ease of manufacture and device complexity are reduced, but measurement precision and responsiveness to process conditions deteriorate
Solution Approach 1:
The system dynamically changes control parameters by replacing fixed preset thresholds with adaptive thresholds derived from machine learning analysis of normal operating conditions. The control system continuously learns from process data and adjusts the parameters used for control decisions, enabling precise detection of abnormal conditions while maintaining ease of manufacture through software-based adaptation rather than hardware reconfiguration.
4Reliability
If holistic real-time control of multi-node networks is implemented, then reliability and productivity improve through coordinated asset operation, but device complexity and difficulty of detecting and measuring process states increase
Solution Approach 1:
The patent creates a virtual model or copy of the industrial process that mirrors the physical system's normal operating conditions. This digital model is trained on historical and real-time data to learn expected behavior patterns. By comparing actual process states against this virtual copy, the system can detect deviations and measure process health without requiring complex direct measurement of every physical parameter.
Data Source
AI summary
A system includes a first asset and a second asset disposed at an industrial system, and a server communicatively coupled to the first asset and the second asset. The server device is configured to receive a first set of stream-based data from the first asset and generate a model for the second asset based on the first set of stream-based data, wherein the model is configured to output an expected set of stream-based data associated with the second asset; receive a second set of stream-based data from the second asset; and send a command to the first asset or the second asset in response to the second set of stream-based data being outside of a threshold from the expected set of stream-based data.