Font Size: a A A

Research On Fault Diagnosis Methods For Sampled-Data Systems

Posted on:2011-04-21Degree:DoctorType:Dissertation
Country:ChinaCandidate:A B QiuFull Text:PDF
GTID:1118330338495792Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
With the wide application of digital comuputer and multisensor techniques in industrial process, most morden systems and equiments are essentially appeared as sampled-data systems. Comparing with general systems, these type of systems have higher level of automation, more complex structure and thus bigger possibility of occurring faults. The research of fault diagnosis for sampled-data systems has attracted considerable attention and some results have achieved. However, due to the hybridism of sampled-data systems, the unknownness of intersampling behaviour, the diversity of sampling pattern and the uncertainity of sampling instant, the difficulties of fault diagnosis are increasing and the existing research has many deficiencies and profound problems. Therefore, with the aid of lifting technique, the theories of hybrid system and time-delay system, the thesis develops several new fault diagnosis design methods for various kinds of sampled-data systems. The main contributions are as follows:1. Under the framework of continuous lifting, the direct design approach to the diagnositic observer which has flexible structure has been establised for a class of sigle-rate sampled-data systems.2. Based on the hybrid system approach and linear matrix inequality (LMI) technique, a direct design methodology of fault detection for a class of sampled-data system with both continuous process noise and discrete measurement noise is presented. It not only resolves the harsh condition of strictly properness caused by continuous lifting, but also avoids solving Riccati equation with jumps which are difficult.3. With the aid of discrete lifting, the extended QR decomposition and elementary transformation, a new fast rate fault detection method, which is intuitive and simple, is proposed for the multirate sampled data systems corrupted by general disturbance.4. Based on the idea of sequential filtering, a fast rate fault detection scheme consisting of a fast rate steady-state residual generation and the corresponding residual evaluation, which can avoid the complex problem of causality constraint, is further developed for multirate sampled-data systems with stochastic disturbance. Meanwhile, the performance of three asynchronous multisensor fusion estimation algorithms based on sequential filtering, the left synchronization lifting and the right synchronization lifting are analyzed and compared. Some conclusions are valuable in practical engineering applications.5. Based on the output-delay approach, a robust sensor fault detection design approach is proposed for the general nonuniform sampled-data systems.6. The problem of fault estimation for sampled-data systems is firstly investigated from the time delay viewpoint. Based on the analysis of the inapplicability of the adaptive fault diagnosis observer in sampled-data system, a novel augmented fault estimation observer design method is proposed to guarantee the exponential convergence of the estimation errors. Furthermore, an extension to the case of time varying fault estimation for the uncertain sampled-data system is studied.7. The effectiveness and superiority of the proposed methods are demonstrated by numerical simulations and aircraft examples.
Keywords/Search Tags:fault diagnosis, sampled-data systems, multirate asynchronous sampling, nonuniform sampling, lifting, hybrid system approach, output delay
PDF Full Text Request
Related items