This paper addresses synthesis of strictly positive real H2 controllers. An upper bound on the H2 norm of the closed-loop transfer function is minimized under constraint of strict positive realness on controllers. We propose a multi-stage design procedure to construct this type of controllers. First, we parametrize both Lyapunov solutions and controllers. Second, we show that if we partially fix them alternatively, we can improve the H2 performance iteratively while allowing for distinct Lyapunov solutions. A numerical example illustrates the applicability of the proposed method.
We have proposed a practical method to solve a hierarchical logistics optimization problem under uncertain demands based on the meta-heuristic method termed Hybrid Tabu Search. For this purpose, we have formulated the problem by employing the idea from flexibility analysis. There, first, we classified the decision variables into two kinds depending on their generic natures, e.g., hard one like facility location not easy to change after decided once and soft one like distribution route possible to change somewhat according to the demand fluctuation. Then we have developed a method to derive a flexible logistic network design that can cope certainly with the worst case only by changing the distribution routes while keeping the performance at higher level in the nominal case. Moreover, we have given a procedure to carry out parametric computations for trade-off analysis helpful for practical decision making. Through numerical experiments, we confirmed effectiveness of the proposed method.
In a human-machine system, an interface displays an abstracted state of the machine. The abstracted state may differ from that anticipated by a user. Such a phenomenon is called an automation surprise and may cause a serious human error. In this paper, we assume that the user operates the machine according to a manual which is modeled by LTL (Linear-Time Temporal Logic). We propose a formal method for detection of three automation surprises and livelocks due to the manual. First, we transform the LTL representation of the manual into a transition system. Next, we propose a composite model derived from the transition system and a machine model, which describes concurrent behavior of the machine and the user. Finally, solving a reachability problem and searching strongly-connected components for the composite model, both the automation surprises and live lock can be detected formally.
Independent component analysis (ICA) is an unsupervised technique for signal processing, and is useful for projection pursuit as well. This paper proposes an enhanced technique of ICA, which extracts independent components that are useful for revealing mutual relationship between observations and some external criteria. Fast ICA algorithm is performed after the preprocessing by regression-principal component analysis that extracts latent variables closely related to external criteria from observations. Numerical experiments including knowledge discovery from POS transaction data reveal the characteristic feature of the proposed method.