Development and Application of Diagnosis Logic of Ion Exchange Filter Depletion

Abstract

<div class="section abstract"><div class="htmlview paragraph">With the widespread adoption of fuel cell electric vehicles, electrical insulation resistance is required for driver safety. However, there are two ways in which resistance decreases: the first is electrical shorts because of failure of high-voltage components, and the second is increased conductivity of fuel cell coolant because of depletion of ion exchange filter. In the conventional solution, since these two decreases could not be distinguished due to noise in the resistance value, a vehicle alerted customers without determining the cause and severity when the resistance value falls below a certain threshold.</div><div class="htmlview paragraph">As a corrective maintenance, when an alert occurs, the vehicle is forced to be immediately delivered to the service center. However, in most cases where the alert came on, the cause was low-risk ion filter depletion. This resulted in customers complaining that they were startled and considering the alert to be non-threatening. As a result, the provider recommended customers to replace the ion exchange filter earlier than the recommended interval which results in an increase in maintenance cost.</div><div class="htmlview paragraph">Therefore, by developing diagnosis the depletion of the ion filter separately as a preventive maintenance and showing simple notice to customer in cluster, the alert can inform customers of only truly dangerous situations, increasing trust in the warning and maximizing the usage of ion filters. To address the limitations of conventional approaches, this paper presents the world's first technique for diagnosis of depletion of ion exchange filters based on pattern recognition of insulation resistance data with high accuracy and noise avoidance. Using driving data from fuel cell electric vehicles, such as the NEXO, this paper demonstrates that the proposed diagnosis logic can detect ion filter depletion with an accuracy of 99.3% (3,093/3,115) for Type I errors and 85.9% (269/313) for Type II errors.</div><div class="htmlview paragraph">By adopting this proposed diagnosis, it is possible to extend the usage period of ion exchange filters and enhance customer convenience, maintenance cost and environmental sustainability. Since eco-friendly cars have many chemical parts (batteries, fuel cell stacks, etc.) that gradually deteriorate, such as ion filters, research on these diagnosis technologies is expected to continue to expand. The diagnosis logic would be updated to Hyundai-Kia Motors' 2nd-generation fuel cell electric vehicles (including NEXO), and for vehicles after the 3rd generation, the diagnosis logic would be basically applied.</div></div>