Publicação
Predictive maintenance for forklifts using IOT data and survival analysis techniques
| Resumo: | Predictive Maintenance is a data-driven methodology of conducting maintenance on industrial assets. It serves a strategic function. Being data-driven, this methodology depends on the volume, frequency and quality of the data being collected from industrial assets, as well as the solution methods employed. In literature, and often in industry, Machine Learning based methods are more popularly employed to devise solution strategies for Predictive Maintenance. However, these methods require large volumes of data and are parametric methods which are constrained to the mathematical functional specification of the method used and may not yield good results in all cases. This work investigates the applicability of a non-parametric method in the field of Survival Analysis for devising a decision-support methodology applied to the predictive maintenance of forklifts by the company "KION Group". This methodology has an additional innovation in that the author has also implemented unsupervised clustering to both inform and refine the inputs to and results from the survival analysis. After validating this method with the field service personnel of KION, it was found that the insights provided by the method would make the predictive maintenance strategies more effective and enable cost optimisation while at the same time increase the uptime of the equipment for KION's customers. |
|---|---|
| Autores principais: | Godbole, Sanjyot |
| Assunto: | Predictive Maintenance Industrial-Internet-of-Things (IIoT) Survival Analysis Unsupervised clustering forklifts material-handling equipment SDG 9 - Industry, innovation and infrastructure SDG 12 - Responsible production and consumption |
| Ano: | 2023 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | Predictive Maintenance is a data-driven methodology of conducting maintenance on industrial assets. It serves a strategic function. Being data-driven, this methodology depends on the volume, frequency and quality of the data being collected from industrial assets, as well as the solution methods employed. In literature, and often in industry, Machine Learning based methods are more popularly employed to devise solution strategies for Predictive Maintenance. However, these methods require large volumes of data and are parametric methods which are constrained to the mathematical functional specification of the method used and may not yield good results in all cases. This work investigates the applicability of a non-parametric method in the field of Survival Analysis for devising a decision-support methodology applied to the predictive maintenance of forklifts by the company "KION Group". This methodology has an additional innovation in that the author has also implemented unsupervised clustering to both inform and refine the inputs to and results from the survival analysis. After validating this method with the field service personnel of KION, it was found that the insights provided by the method would make the predictive maintenance strategies more effective and enable cost optimisation while at the same time increase the uptime of the equipment for KION's customers. |
|---|
Atividades financiadas
Não foram encontrados projetos financiados.