Projects / Programmes
On-line System Identification for Model-Based Prognostics and Health Management
| Code |
Science |
Field |
Subfield |
| 2.06.03 |
Engineering sciences and technologies |
Systems and cybernetics |
Methods and tools for design and implementation of control systems |
| Code |
Science |
Field |
| T125 |
Technological sciences |
Automation, robotics, control engineering |
| Code |
Science |
Field |
| 2.02 |
Engineering and Technology |
Electrical engineering, Electronic engineering, Information engineering |
on-line system identification, parameter estimation in dynamical systems, signal processing, prognostics, estimation of remaining useful life
Researchers (1)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
28479 |
PhD Matej Gašperin |
Systems and cybernetics |
Head |
2013 - 2015 |
60 |
Organisations (1)
| no. |
Code |
Research organisation |
City |
Registration number |
No. of publicationsNo. of publications |
| 1. |
0106 |
Jožef Stefan Institute |
Ljubljana |
5051606000 |
90,742 |
Abstract
Mechanical systems deteriorate in time. These deteriorations cause mechanical faults that eventually lead to interruptions in performance, expensive maintenance and reduced lifetime of the entire system. Despite the vast amount of research devoted to condition monitoring in the last decades, the on-line evaluation of component health or the ability to predict incipient failure remains difficult. It is believed that a shift in traditional health management tools, such as trend analysis, to techniques from different disciplines is required to achieve the promise of condition-based maintenance and prognostics.
Control theory provides a powerful set of modelling tools to study the interactions and behaviour of a system when subjected to conditions or inputs. In this scope a state-space representation of a system can be used to model machinery components. In such a model, the input variables could be for example torque and RPM into a gearbox, and the output variables are heat and vibration. Additionally, any number of derived outputs can be calculated that are indicators of the systems component conditions (a Condition Indicator (CI)). These inputs and output can be used to represent the state of the system.
The goal of the proposed project is to develop new algorithms for data-driven prediction of RUL with on-line estimation of the state-space model. This is achieved by combining the state estimation algorithm with Maximum-Likelihood parameter estimation in the form of the Expectation-Maximization (EM) algorithm. The important additional feature of the proposed framework is the ability to automatically adapt to the eventual changes in the process (change of operating conditions or other incipient events).
The project comprises the three main goals, which can be summarized as follows:
To develop algorithms for model-based prognostics with on-line estimation of nonlinear dynamical models.
To validate the system identification framework for model-based prognostics under non-stationary operating conditions.
To validate the developed tools on real-world applications and disseminate the results.
The proposed project is the continuation of the research carried out in the scope of the PhD thesis, which has already received attention in both scientific community and industry. Additionally, the Department of Systems and Control has rich experience in development of advanced diagnostics algorithms and electronic components. The department has strong connections with industrial partners, which is a result of successful collaboration on numerous past projects. This environment therefore provides a solid base for successful implementation of the project goals.
Significance for science
Nowadays, the requirement for automated procedures for monitoring of equipment is becoming more important in several industrial branches. The beginnings of the field can be traced back to military industry and energy sectors. With the maturing of the research field, the application areas are being broadened. However, the transfer of the technology from highly specialized application areas to wide industrial applications poses additional requirements, which have to be adequately addressed. In the scope of the research project Online System Identification for Model-Based Prognostics and Health Management (SysID-PHM) we have addressed one of these requirements, which is the need for autonomous identification of mathematical models for prognostics. With this, we can highly contribute to the decrease in time needed to apply such algorithms to a wide array of processes. The novelty of our solution is that we adopt a mathematical model in the state-space, where the parameters of the model are automatically inferred from available operation data. With this we are, up to our knowledge, the first group to effectively incorporate model identification and prognostics in one algorithm. From the industrial viewpoint, the implementation of the predictive maintenance gives opportunity for significant decrease of the direct as well as indirect maintenance costs in every company. New monitoring technologies allow radically new ways of maintenance i.e. predictive maintenance and moving the maintenance from the company (outsourcing) to the companies specialized for that purpose. Both will in the end decrease the costs and increase the availability of the equipment and hence the efficiency of the overall production process.
Significance for the country
For a relatively small and export oriented economy, such as Slovenian, the niche high-technology fields, such as the one covered in the The SysID-PHM project have a great potential to contribute to either formation of new start-up companies, or increase the competitiveness of the existing companies active in the area. In the duration of the project, we have established a working relationship with companies INEA d.o.o. and COSYLAB d.d., both of which are interested in the specific results of the project. Furthermore, apart from anticipated benefits for the economy, the results contribute to the international recognition of the Department of Systems and Control and entire Jožef Stefan Institute. Following several impactful publications, we have been able to raise the reputation of the research group from the targeted community and increase the potential for future collaborations.
Most important scientific results
Annual report
2013,
2014,
final report,
complete report on dLib.si
Most important socioeconomically and culturally relevant results
Annual report
2013,
2014,
final report,
complete report on dLib.si
