Available Technologies

    Innovation Sphere

Communicate

Details

Project TitleTimes Series Method for Industrial Modeling
Track Code2017-116
Short Description
Abstract
 
TagsData analytics, latent variable methods, principal component analysis, projection to latent structures, canonical correlation analysis, time series, industrial data modeling, fault detection, diagnosis
 
Posted DateOct 9, 2017 12:36 PM

Market Opportunity

Industrial manufacturing and operation facilities require large scale data acquisition systems that utilize hundreds or even thousands of sensors. Process operation data are massive and highly dimensional due to the complexity of the processes. Fault diagnosis methods have been intensively studied and applied in many industrial processes. However, these methods exploit relations among the data that are linear and static. Dynamics or time correlations among the data are useful for prediction and interpretation. It is therefore necessary to develop dynamic versions of the fault diagnosis methods for efficient and effective modeling.

USC Solution

USC researchers have developed a novel method and system for industrial, multivariate data modeling, fault detection, and diagnosis. This technology extracts desirable features and variables from data and applies the newly acquired knowledge to support effective decision-making at an industrial plant. Through the use of data analytics these processes are streamlined and provide reliable, up-to-date information reflecting changes in operation.

Value Proposition

  • Provides up-to-date information reflecting changes in operation

  • Effective fault detection, diagnosis, and decision-making

  • Real-time data

  • Streamlined operations

Applications

  • Process operations and control

  • Industrial manufacturing

  • Data analytics

  • Fault diagnosis

Stage of Development

  • Experimentally validated

  • Available for exclusive and non-exclusive license

Intellectual Property

Status:

Provisional patent filed

62/473,971


Key Publications:

Data distillation, analytics, and machine learning

Contact Information

Nikolaus Traitler, Licensing Associate

USC Stevens Center for Innovation

(213) 821-3550

traitler@usc.edu

Files

File Name Description
NCD 2017-116 - Times Series Method for Industrial Modeling.pdf None Download