Knowledge Extraction and Application

Summary

The Knowledge Extraction and Application (KEA) project will contribute to standards and test methods that normalize models, methods, and technologies for connecting shop floor information to operations decision making.  This project will improve the utilization of available information by synthesizing and contextualizing information from traditionally incompatible sources ("linking" data together). The result is increased manufacturing system efficiency through improved operations management capabilities leveraging linked data.

Industrial analytics and decision-support models must be supported by the most up-to-date information available about part quality, equipment reliability, process performance, and other information. Dissimilar information systems collect, handle, and provide data in highly different ways – often to the point of direct incomparability or incompatibility between information sources. Even within each information regime, the myriad of data types requires specialized models to synthesize or extract pertinent knowledge that is relevant to the high-level decision planning process.

One major component of this information regime is human generated data. The prevailing paradigm underlying smart manufacturing technologies often stops short of explicit integration of human knowledge and human-machine interactions in operational analysis. Unfortunately, this means human-generated knowledge is captured in text-based documents that are often riddled with inconsistencies, relatively few-in-number, and filled with domain-specific jargon. The problems with these documents leads to difficulties in consistently using them when making operational decisions on the manufacturing floor as they are difficult to analyze in their current form.

NIST is uniquely situated to tackle this problem facing manufacturers by having the expertise and motivation to look across the multiple areas of the production process -- linking disparate information from multiple products and services that are not inherently designed to interact efficiently. Successful decision-making in this context relies on the integral parts of capturing information, contextualizing that information into knowledge, then interpreting that knowledge to develop actionable plans. Each piece of this process is necessary to ensure optimal operations management.

Description

Objective - To develop and deploy advances in standards, measurement science, and software tools using actionable, computable, domain knowledge and data in operations and logistics to improve the reliability, quality, and efficiency of advanced manufacturing systems.

What is the technical idea?

The technical idea is to develop methodologies and architectures to collect and transform streaming data and tacit knowledge stored in documents, such as maintenance work orders, into actionable intelligence. Where suitable industrial-grade solutions exist, the strategy is to select, implement, and test technologies supporting the required functionality. Where they do not, the project will develop and test methodologies delivered as best practice guides and reference implementations.

To address the broader operations management context, the project will explore and test suitable solutions to address the functional requirements of data collection, performance analysis, and tracking (ISO/IEC-62264) and the cross-cutting concerns of shop floor connectivity, distributed data management, and knowledge extraction and industrial analytics (IIRA).

Within a knowledge extraction and industrial analytics focus, there is a need for methodologies supporting industrial analytics lifecycle management, including algorithm selection and tuning, verification and validation, deployment and integration, and configuration management.

Both the broader and narrower focus areas enable higher-level operations management functions such as  intelligent, flexible, and resilient (operational) control, or scheduling and execution management. The target use cases address maintenance and production scheduling, where predicting resource failures through advanced analytics and mitigating resource downtime and production delays through intelligent and resilient scheduling.

The goal of this project is to enable manufacturers to collect and synthesize data collected and contextualized from the shop-floor and information technology systems and transform and analyze these data and documents themselves with aid from methodologies realized in open source toolkits and guidelines developed at NIST.

What is the research plan?

There is a need for standard models, methods, and recommended practices for efficiently using information available in modern manufacturing facilities to enable intelligent operations decision-making. The goal of this project is to explore methods for processing information (extracting knowledge) collected from the areas of part quality, process efficiency, and equipment reliability and methods for enabling critical operations decision-support capabilities. The two primary use cases for KEA will focus on scheduling (operational control) and prognostics (reliability).

Major Milestones / Thrust Areas:

• Scalable Data Pipeline for Advanced Manufacturing Systems
• Synthesizing Heterogeneous Data Sources
• Industrial Analytics Lifecycle Management
• Annotation Tool Development
• Maintenance Operation Improvement
• Technical Language Processing of Maintenance Documents