An intelligent fault detection system for a heat pump installation based on a geothermal heat exchanger

Abstract

The heat pump with geothermal exchanger is one of the best methods to heat up a building. The heat exchanger is an element with high probability of failure due to the fact that it is an outside construction and also due to its size. In the present study, a novel intelligent system was designed to detect faults on this type of heating equipment. The novel approach has been successfully empirically tested under a real dataset obtained during measurements of one year. It was based on classification techniques with the aim of detecting failures in real time. Then, the model was validated and verified over the building; it obtained good results in all the operating conditions ranges.

Holistic management of a smart city thermal energy plant with sewage heat pumps, solar heating, and grey water recycling

“…Casteleiro-Roca et al designed a novel intelligent system to detect faults in BHE systems (Casteleiro-Roca et al, 2016). Based on a classification method, the system can detect when the BHE in a heat pump system fails in real-time.…”

Development of efficient, flexible and affordable heat pumps for supporting heat and power decarbonisation in the UK and beyond: Review and perspectives

“…Field trials of heat pumps in the UK in the early 2010s showed some 40-50% lower COP than the brand labeled values [71], which were partially due to wrongly sized and incorrectly for set up systems for the properties, which led to operation in sub-optimal modes [72]. Additionally, the heat pumps may be damaged by extreme weather events (such as flood) or accidental activities (such as like crushing, perforations) [73]. Active maintenance and performance monitoring are needed to ensure the high energy efficiency over the systems' lifetime.…”

A fault detection system based on unsupervised techniques for industrial control loops

“…The well‐known support vector machine algorithm employed in many different applications (Schölkopf et al, ; Calvo‐Rolle et al, ; Casteleiro‐Roca et al, ; Jove et al, ) is frequently used to solve the one‐class classification problem as well as support vector data description (Tax, ). To solve anomalies issues in different parts of industrial plants, the use of virtual sensors or missing data imputation techniques is very common (Casteleiro‐Roca et al, ; Fernández‐Serantes et al, ; Jove et al, ; Casteleiro‐Roca et al, ; Gonzalez‐Cava et al, ).…”