Applying dynamic Bayesian Networks to process monitoring
| dc.contributor.advisor | Auret, Lidia | en_ZA |
| dc.contributor.advisor | Kroon, R. S. (Steve) | en_ZA |
| dc.contributor.author | Wakefield, Brandon Jason | en_ZA |
| dc.contributor.other | Stellenbosch University. Faculty of Engineering. Dept. of Process Engineering. | en_ZA |
| dc.date.accessioned | 2018-11-29T09:42:14Z | |
| dc.date.accessioned | 2018-12-07T07:00:13Z | |
| dc.date.available | 2018-11-29T09:42:14Z | |
| dc.date.available | 2018-12-07T07:00:13Z | |
| dc.date.issued | 2018-12 | |
| dc.description | Thesis (MEng)--Stellenbosch University, 2018. | en_ZA |
| dc.description.abstract | ENGLISH ABSTRACT: In efforts to reduce the impact of human error on the operation of chemical and mineral processing plants, reliable process monitoring solutions attempt to assist plant operators and engineers to detect and diagnose process faults before significant loss is incurred. An existing solution, the traditional multivariate statistical process monitoring (MSPM) approach, is able to reliably detect abnormal process behaviour but struggles to unambiguously identify the root cause of the abnormal behaviour. It was identified that this is caused by a lack of incorporation of existing process knowledge into the framework of the MSPM approach. It was proposed to investigate a different fault diagnosis approach which directly incorporates process knowledge into its framework. Lerner et al. (2000) and Lerner (2002) present such an approach, using probabilistic methods to infer process behaviour given a particular process model. This model is in the form of a dynamic Bayesian network (DBN), and would contain various models which each describe particular process behaviour given information about the operational status of various process components. In particular, these DBN models were able to describe normal process behaviour in addition to highly specific abnormal process behaviour caused by, for instance, a sensor fault or a blocked pipe. Using optimised methods, the authors could then use a DBN model to make predictions about process behaviour and infer, given observation of actual process behaviour, which combination of component statuses best describe that observation. Therefore, solving the fault diagnosis problem could be reduced to performing inference in a DBN using this approach. A probabilistic fault diagnosis (PD) approach based on Lerner et al. (2000) and Lerner (2002) was therefore implemented and investigated in this thesis. A survey of recent DBN-based PD approaches was also performed, and it was determined that relatively little research had been done on the topic. Furthermore, published results presenting fault diagnosis performance for DBN-based PD approaches were typically found to be useless for meaningful comparison with a traditional MSPM approach. In this regard, this thesis aimed to investigate the usefulness of the PD approach in comparison to the MSPM approach, while providing useful fault diagnosis performance metrics to facilitate comparison with other fault diagnosis approaches. The PD approach tested in this research also extended upon Lerner et al. (2000) and Lerner (2002) by including models for regulatory control systems and recycle streams based on the work by Yu and Rashid (2013). Additionally, from the same paper, the concept of abnormality likelihood index (ALI) was implemented in the PD approach. This enabled the PD approach to function more similarly to the MSPM approach, facilitating direct comparison. Generally, it was found that the PD approach could provide competitive fault detection when compared with the MSPM approach. However, this was at the cost of real-time fault detection as well as longer detection delay for incipient faults. On the other hand, it was found that the PD approach performed better at root cause analysis than the MSPM approach. In particular, the PD approach typically provided better isolation for the root cause of fault conditions. Despite some issues, similar results were observed for the PD approach when scaling up to larger processes. Nonetheless, these issues may be addressed with additional research, further improving the capabilities of the PD approach. Therefore, it was concluded that the PD approach is useful for fault diagnosis and should be investigated further in future research. | en_ZA |
| dc.description.abstract | AFRIKAANSE OPSOMMING: In pogings om die impak van menslike foute op die bedryf van chemiese en mineraalprosesseringsaanlegte te verminder, probeer betroubare prosesmonitering oplossings om aanlegoperateurs en -ingenieurs by te staan om prosesfoute op te spoor en te diagnoseer voor beduidende verliese aangegaan word. ’n Bestaande oplossing, die tradisionele meervariaat statistiese prosesmonitering (MSPM) benadering, kan abnormale prosesgedrag betroubaar opspoor maar sukkel om die grondoorsaak van die abnormale gedrag eenduidig te identifiseer. Dit is geïdentifiseer dat dit veroorsaak word deur ’n gebrek aan inkorporasie van bestaande proseskennis in die raamwerk van die MSPM-benadering. Dis voorgestel om ’n ander foutdiagnose benadering te ondersoek, wat proseskennis direk in sy raamwerk inkorporeer. Lerner et al. (2000) en Lerner (2002) lewer so ’n benadering deur probabilistiese metodes te gebruik wat prosesgedrag aflei gegewe ’n bepaalde prosesmodel. Die model is in die vorm van ’n dinamiese Bayes-netwerk (DBN) en kan verskeie modelle bevat wat elk bepaalde prosesgedrag beskryf gegewe informasie oor die operasionele status van verskeie proseskomponente. Hierdie DBN modelle kon veral normale prosesgedrag bo-op hoogs gespesifiseerde, abnormale prosesgedrag veroorsaak deur byvoorbeeld, ’n sensorfout of geblokkeerde pyp, beskryf. Deur geoptimaliseerde metodes te gebruik, kan die outeurs dan ’n DBN model gebruik om voorspellings oor prosesgedrag te maak en aflei watter kombinasie van werklike komponent-statusse die observasie die beste beskryf, gegewe observasie van werklike prosesgedrag. Die oplossing van die foutdiagnose probleem kan dus gereduseer word tot die uitvoering van inferensie in ’n DBN, deur hierdie benadering te gebruik. ’n Probabilistiese foutdiagnose (PD) gebaseer op Lerner et al. (2000) en Lerner (2002) is daarom geïmplementeer en ondersoek in hierdie tesis. ’n Opname van onlangse DBN-gebaseerde PD-benaderings is ook uitgevoer, en dis vasgestel dat relatief min navorsing oor hierdie onderwerp gedoen is. Verder, gepubliseerde resultate wat foutdiagnose uitvoering vir DBN gebaseerde PD-benaderings wys, is tipies nutteloos gevind vir die vergelyking met ’n tradisionele MSPM-benadering. In hierdie opsig het die tesis beoog om die nuttigheid van die PD-benadering in vergelyking met die MSPM-benadering te ondersoek, terwyl dit nuttige foutdiagnose werkverrigtingmetrieke verskaf om die vergelyking met ander foutdiagnose-benaderings te fasiliteer. Die PD-benadering wat in hierdie navorsing getoets is, het ook op Lerner et al. (2000) en Lerner (2002) uitgebrei deur modelle vir regulerende beheerstelsels en herwinningstrome in te sluit, gebaseer op die werk deur Yu and Rashid (2013). Daarby, uit dieselfde publikasie, is die konsep van abnormaliteit aanneemlikheidsindeks (AAI) geïmplementeer in die PD-benadering. Dit het die PD-benadering toegelaat om meer soortgelyk aan die MSPM-benadering te funksioneer, wat ’n direkte vergelyking fasiliteer. Oor die algemeen is gevind dat die PD-benadering kompeterende foutdeteksie kon verskaf, as dit met die MSPM-benadering vergelyk word. Dit was egter ten koste van intydse foutdeteksie sowel as langer deteksie vertraging vir aanvangsfoute. Aan die ander kant is dit gevind dat die PD-benadering beter met grondoorsaak analise gedoen het as die MSPM-benadering. Die PD-benadering het in besonder tipies beter isolasie vir die grondoorsaak van foutkondisies verskaf. Ten spyte van ’n paar kwessies, is soortgelyke resultate vir die PD-benadering waargeneem toe daar na groter prosesse opgeskaal is. Nietemin, hierdie kwessies kan geadresseer word met bykomende navorsing, wat die vermoëns van die PD-benadering verder sal verbeter. Die gevolgtrekking is dat die PD-benadering nuttig is vir foutdiagnose en moet verder ondersoek word in toekomstige navorsing. | en_ZA |
| dc.format.extent | 242 pages : illustrations | en_ZA |
| dc.identifier.uri | http://hdl.handle.net/10019.1/105116 | |
| dc.language.iso | en_ZA | en_ZA |
| dc.publisher | Stellenbosch : Stellenbosch University | en_ZA |
| dc.rights.holder | Stellenbosch University | en_ZA |
| dc.subject | Multivariate analysis | en_ZA |
| dc.subject | UCTD | en_ZA |
| dc.subject | Manufacturing processes -- Monitoring | en_ZA |
| dc.subject | Fault location (Engineering) | en_ZA |
| dc.subject | Bayesian field theory | en_ZA |
| dc.title | Applying dynamic Bayesian Networks to process monitoring | en_ZA |
| dc.type | Thesis | en_ZA |