Evaluation of the suitability of different chiller performance models for on-line training applied to automated fault detection and diagnosis (RP-1139)

Chemical Compounds

• Multilayer 95%
• Black 47%
• Error 40%
• Time 38%
• Length 32%
• Weight 28%

Engineering & Materials Science

• Fault detection 100%
• Failure analysis 92%
• Multilayer neural networks 11%
• Polynomials 8%
• Statistical Models 4%
• Vapors 4%
• Uncertainty 3%
• Neural networks 3%

Earth & Environmental Sciences

• detection 63%
• evaluation 51%
• parameter 8%
• prediction 5%
• artificial neural network 4%
• sliding 4%
• compression 3%
• engineering 3%
• test 1%
• analysis 1%