Orestes Llanes-Santiago

An approach to multiple fault diagnosis using fuzzy logic

The development of systems capable of diagnosing new and multiple faults in industrial systems is an active research topic. In this paper a model-based diagnostic system capable of diagnosing new and multiple faults using fuzzy logic as a fundamental tool is proposed. Also, the wavelet transform is used for isolating noise present in measurements. The proposed model was applied to the Continuously-Stirred Tank Heater model benchmark. The results demonstrate the feasibility of the proposed model, improving the robustness in the diagnostic, without loss of sensitivity to incipient or small magnitude faults.

Fault Diagnosis in Industrial Systems Using Bioinspired Cooperative Strategies

This paper explores the application of bioinspired cooperative strategies for optimization on Fault Diagnosis in industrial systems. As a first step, the Differential Evolution and Ant Colony Optimization algorithms are considered. Both algorithms have been applied to a benchmark problem, the two tanks system. The experiments have considered noisy data in order to compare the robustness of the diagnosis. The preliminary results indicate that the proposed approach, basically the combination of the two algorithms, characterizes a promising methodology for the Fault Detection and Isolation problem.

Accelerated Direct-Problem Solution: A Complementary Method for Computational Time Reduction

This paper presents a proposal designed to reduce the time required by the process to estimate the parameters of a system by accelerating the direct-problem solution as the slow phase in any estimation method. This proposal is considered a complement to existing procedures, such as the combination of different optimization methods for the purpose of reducing the number of calls to the objective function. The proposal consists of a procedure that helps study the relation between the direct-problem solution step and the time required for this solution, as well as the influence of the direct solution’s built-in error on the accuracy of the estimated parameters. Consequently, the extent in which the estimation process can be accelerated without impairing estimation accuracy can be determined. For the purpose of testing its viability, this proposal was applied to the estimation problem of the kinetic parameters of a chromatography column process, as modeled using the front-velocity method. The results from this test show that, by accelerating the direct-problem solution, the estimation time can be reduced significantly, without affecting the accuracy of the estimation.

Diagnosing Time-Dependent Incipient Faults

This chapter focuses on a formulation for fault diagnosis (FDI) using an inverse problem methodology. It has been shown that this approach allows for diagnoses with adequate balance between robustness and sensitivity. The main contribution of this chapter is the expansion of this approach to include the diagnosis of time-dependent incipient faults. The FDI inverse problem is formulated as an optimization problem that is then solved with two metaheuristics: Differential Evolution and its variation Differential Evolution with Particle Collision. The proposed methodology is tested using simulated data from the Two Tanks system, which is recognized as benchmark for control and diagnosis. The results indicate that this proposal is suitable for the aforementioned diagnosis.

An Indirect Kernel Optimization Approach to Fault Detection with KPCA

This chapter discusses a new indirect kernel optimization criterion for the adjustment of a fault detection process that is based on the dimension–reduction technique known as kernel principal component analysis. The kernel parameter optimization proposed here involves the computation of the false alarm rate and false detection rate indicators that are combined in a single indicator: the area under the ROC curve. This approach was tested on the Tennessee Eastman (TE) process, where a significant decrease in false and missing alarms was observed.

Uncertainty analysis in mass transfer parameter estimations for chromatographic separation of glucose and fructose

In this paper, a statistical approach for the analysis of the propagation of uncertainty is shown, in the estimate of the kinetic parameters of mass transference used to model a chromatographic column in Simulated Moving Bed. The modeling of the chromatography column was accomplished intervening the new approach front velocity. The analysis of how it is propagated the operational factors uncertainty involved in the process of chromatography toward the estimated parameters was carried out by the use of response surface methodology. Furthermore, chromatographic regions where factors cause bigger variation in the output and their respective patterns were determined. The analysis was applied to the separation process of glucose and fructose.