Unsupervised isolation of abnormal process variables using sparse autoencoders

Keyphrases

• Abnormal Change 100%
• Sparse Autoencoder 100%
• Process Data 60%
• Changes in Processes 60%
• Historical Process 40%
• Autoencoder 40%
• Process Behavior 20%
• Integral Component 20%
• Fault Diagnosis 20%
• Neural Network 20%
• Process Faults 20%
• Process Operation 20%
• Without Prior Knowledge 20%
• Simple Faults 20%
• Inconclusive Result 20%
• Acquired Knowledge 20%
• Abnormal Events 20%
• Complex Fault 20%
• Sparse Networks 20%
• Sparsity Constraint 20%

Engineering

• Process Variable 100%
• Autoencoder 100%
• Process Behavior 20%
• Tasks 20%
• Fault Diagnosis 20%
• Sparsity 20%