Performance Analysis And Evaluation Of Multi - Sensor Delay Control System

As an important module of large complicated timing equipment, multi-sensor delay control system(MDCS) takes charge of estimating a key delay parameter. How to make analysis and evaluation for MDCS system is the focus and difficulty for designers and customers. In this dissertation, some concerned theoretical knowledge, such as bayesian small sample theory, convex model theory, data fusion theory, optimization theory, evidence theory, decision analysis theory are applied, in-depth research of performance analysis and evaluation of MDCS system has been conducted, the main contributions are as follows.1) In order to make effective accuracy evaluation on MDCS system under small sample condition, two bayes assessment models based on credibility of prior information are proposed, the physical implications of which are also expounded. The comparative result of point estimation error among the proposed models and two conventional models is given by analytical deduction. Good performance of the proposed models is demonstrated by simulation results.2) The convex model is applied to express the uncertainty of the sensor’s estimation in interference environment, then robustness index is proposed to denote the inhibition ability of the decision algorithm on abnormal estimation. The geometrical implication of the proposed robustness index is expounded. The analytical resolution method of the robustness index of the common decision algorithms is deduced, and general simulation resolution method is also proposed.3) Quantitative improvement method of the three existed decision algorithms is proposed. Multi-object optimization problem is put forward on the improvement of difference-parallel algorithm, then pareto resolution under different weight combination is obtained by genetic algorithm, thus rational resolution is given on increment-setting optimization. In order to enhance the robustness of the decision algorithm, two adaptive data fusion algorithm is proposed.4) Four key indexes, i.e. accuracy, robustness, effective sensor number and computational complexity are applied to make comprehensive evaluation on decision algorithm of MDCS system. The similarity between different weight vectors is proposed, based on which, weight assignment optimization method is proposed based on minimum deviation principle is proposed, and the calculation platform of comprehensive evaluation is constructed. 5) In order to make quantitative reliability analysis on MDCS system when only a few statistical data are available, evidence theory and convex model theory are combined to make non-probability quantitative analysis on MDCS system. The basic probability assignment is used to express the uncertainty of the basic event’s probability, then evidence theory and structure function are applied to transfer the uncertainty, and the probability envelop of the top event is obtained, hence the reliability quantitative analysis of the complicated system is implemented.As mentioned above, this dissertation has conducted systematical research on MDCS system in depth, obtained some positive research achievements, and provided elementary resolution on the analysis and evaluation of MDCS system.