DEVELOPMENT OF OE-BASED BROWN-FORSYTHE TEST ALGORITHM FOR CONTROL VALVE STICTION DETECTION
One of the valve nonlinearities is the existence of control valve stiction. Stiction is a condition where the valve stem resists movement and does not give the required response to the output signal from the controller. Control valve stiction has adverse effects to the control loop performance of a...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Year Project Report / IMRAD |
| Language: | en en |
| Published: |
Universiti Malaysia Sarawak, (UNIMAS)
2018
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/34356/1/DEVELOPMENT%20OF%20OE-BASED%20BROWN-FORSY24pgs.pdf http://ir.unimas.my/id/eprint/34356/4/DEVELOPMENT%20OF%20OE-BASED%20BROWN-FORSYft.pdf http://ir.unimas.my/id/eprint/34356/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | One of the valve nonlinearities is the existence of control valve stiction. Stiction is a condition where the valve stem resists movement and does not give the required response to the output signal from the controller. Control valve stiction has adverse effects to the control loop performance of a process as it introduces variability in the process parameters. This can lead to deterioration in product quality and economic loss. As a result, this project puts an emphasis on the stiction detection methods in the research areas of process control. Therefore, the main objective of the project is to develop an OE-based Brown-Forsythe test algorithm to effectively detect the presence of control valve stiction.In this study, the proposed OE model is developed using System Identification in MATLAB, where it is used to simulate the process output (PV). Residual distribution is generated from the difference between actual and simulated PV. Brown-Forsythe test statistics, H(R) are calculated using Kruskal-Wallis (non-parametric ANOVA) test and hypothesis testing is performed. Stiction is then declared if the values exceed the threshold value, X2, where the null hypothesis is rejected at 5% significance level.In order to investigate the effectiveness of the proposed method, two case studies are considered whereby step change and PRBS input signals are introduced for each case study, respectively. Case Study 1 studies three strengths of stiction, which are no stiction (Base Case 1), weak stiction (Case 1.1) and strong stiction (Case 1.2), while Case Study 2 investigates the presence of several sources of process nonlinearities in control loops, which include well-tuned controller (Base Case 2), tight-tuned controller (Case 2.1), presence of external disturbances (Case 2.2) and presence of stiction (Case 2.3).As a result, the proposed method is able to successfully detect and distinguish presence of stiction for both types of inputs at 95% confidence level. A sensitivity analysis is also conducted for process gain, K and time constant, τ model parameters, whereby the method is considered satisfactorily robust as it is shown to be insensitive to ±10% of changes in the model parameters. The method is also found to be applicable to successfully detect stiction within industrial control loops. Lastly, it is compared with other published stiction detection methods, where it performs as efficient and even better than other methods. |
|---|