Experimental validation of a neural-network-based method for fault detection in heat pump evaporators

Energy and AI, vol. 25, pp. 100845

Abstract

The heat pump industry expects a significant shortage of qualified technicians, which will affect quick replacement and service work. It is therefore sensible to reduce unnecessary maintenance work on heat pumps. However, undetected faults can have a significant impact on efficiency and therefore, operating cost. A shift from regular maintenance to demand-based strategies is therefore desirable. Fault Detection and Diagnosis has been shown to be a valuable tool for implementing demand-based maintenance by enabling the automatic detection of malfunctions. Several methods for the automatic detection of evaporator fouling have been developed. However, experimental validation of complete fault-detection frameworks in this context remains scarce. In this study, an anomaly detection of the evaporator air volume flow is performed as a proxy for evaporator fouling. It is based on an artificial neural network that learns the behavior of the fault-free system, which is used as a reference to evaluate the current state. The residual is evaluated using a threshold-based approach and Bayesian classification in order to detect fault states. Using this method, about 98% of our experimental datapoints were correctly classified, with a detection rate of up to 91.1%. This experimentally validated proof of concept highlights the potential of this approach for predictive maintenance applications.

Authors 6

1. Lasse Jöhnk corresponding Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, E.ON Energy Research Center, Institute for Energy Efficient Buildings and Indoor Climate, Mathieustrasse 10, 52074 Aachen, Germany

2. Sebastian Borges Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, E.ON Energy Research Center, Institute for Energy Efficient Buildings and Indoor Climate, Mathieustrasse 10, 52074 Aachen, Germany

3. Leon Weingarten Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, E.ON Energy Research Center, Institute for Energy Efficient Buildings and Indoor Climate, Mathieustrasse 10, 52074 Aachen, Germany

4. Tim Klebig Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, E.ON Energy Research Center, Institute for Energy Efficient Buildings and Indoor Climate, Mathieustrasse 10, 52074 Aachen, Germany

5. Christian Vering Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, E.ON Energy Research Center, Institute for Energy Efficient Buildings and Indoor Climate, Mathieustrasse 10, 52074 Aachen, Germany

6. Dirk Müller Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University, E.ON Energy Research Center, Institute for Energy Efficient Buildings and Indoor Climate, Mathieustrasse 10, 52074 Aachen, Germany