Automation Engineering Learning Framework for Sparse Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The use of AI in the engineering phase of industrial automation is underdeveloped due to issues such as scarce proprietary engineering data, short duration of the engineering phase, and difficulty in capturing human intent and knowledge.
Innovation Solution
The implementation of machine learning and artificial intelligence (AI) methods, including neural networks, code classification, semantic code search, and hardware recommendations, to automate engineering tasks such as code classification, hardware configuration, and code generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If AI methods are applied to automate engineering tasks, then productivity and efficiency are improved, but the scarcity of proprietary engineering data and difficulty in capturing human intent remain challenging constraints
Solution Approach 1:
The patent introduces neural networks and machine learning models as intermediary systems that learn from available engineering data and sensor information to capture human intent and domain knowledge. These intermediaries bridge the gap between limited proprietary data and the need for intelligent automation, enabling the system to infer and replicate expert decision-making patterns without requiring exhaustive explicit knowledge representation
Solution Approach 2:
The system implements feedback loops where the AI models continuously learn from engineering data and runtime sensor information. By incorporating feedback mechanisms, the system refines its understanding of human intent and engineering principles over time, improving its ability to automate tasks while preserving the nuanced knowledge that would otherwise be lost
2Reliability
If the engineering phase duration is extended to capture more human knowledge, then the quality of AI training data improves, but the overall project timeline increases
Solution Approach 1:
The patent applies preliminary action by pre-training neural networks on available engineering data and historical information before the actual engineering tasks begin. This allows the AI system to develop initial capabilities and knowledge representations in advance, reducing the need for extended engineering phases while maintaining data quality for training purposes
Solution Approach 2:
The system uses multi-functional approaches where the same AI infrastructure serves multiple purposes: training on engineering data, processing runtime sensor information, and performing various engineering tasks. This universality allows the system to maximize the value extracted from limited engineering phase time while maintaining high data quality through repeated use across different functions
3Manufacturing precision
If more comprehensive hardware configuration and code generation are performed manually, then the precision and adaptability improve, but the development complexity and time consumption increase
Solution Approach 1:
The patent implements self-service mechanisms where the AI system automatically generates hardware configurations and code based on learned patterns from training data. The system serves itself by using its own trained models to produce engineering artifacts, reducing manual intervention while maintaining precision through the accuracy of the underlying neural networks and learning algorithms
Data Source
AI summary
Applications of artificial intelligence (AI) in industrial automation have focused mainly on the runtime phase due to the availability of large volumes of data from sensors. Methods, systems, and apparatus that can use machine learning or artificial intelligence (AI) to complete automation engineering tasks are described herein.