In this paper, we propose a model set identification method for Hammerstein systems. This method gives a local model set near an equilibrium point by evaulating the l∞, gain of the model error for the given input level. The upper bound of l∞ gain can be obtained by a natural extension of an existing worst-case l1 identification method for linear systems, which is the main motivation to adopt l∞ gain as the uncertainty measure in this method. Also, this method gives less conservative model sets with more experimental data by using the noise set which consists of hard-bounded noises but takes account of a low correlation property of noise signals, simultaneously.
This paper discusses dissipative control for discrete-time systems and provides a design method for a controller to ensure specified stability margins (e.g. gain and phase margins). The paper extends an existing continuous-time dissipative control into the discrete-time case, with emphasis on quadratic dissipativeness of the loop-transfer function. Such dissipativeness ensures that the feedback system has a robust stability against some class of simultaneous perturbations on gain and phase inserted into the feedback loop. The class is characterized according to the metric Q of quadratic dissipativeness. It is shown that the Q can be suitably selected from specified gain and phase margins and if the loop-transfer function is quadratic dissipative with the Q then specified stability margins are ensured.
Generation of a stationary environmental map is one of the important tasks for computer vision. Under the assumption of known motion of a camera, environmental maps of a real scene can be successfully generated by monitoring azimuth changes in an image. Several researchers have used this property for robot navigation. However, it is difficult to observe the exact motion parameters of the camera because of encoder measurement error of the robot. Therefore, observation errors in the generated environmental map accumulate in long movements of the robot. To generate a large environmental map, it is desirable not to assume known camera motion. In this paper, under the assumption of unknown motions of the camera, we propose a method to generate a stationary environmental map and estimate the egomotion by using an omnidirectional image sensor.
The installation of various equipments and devices in close proximity necessitate the establishment of a design method capable of securing safety in a systematic manner. Regarding the safety index worked out for the sake of safety evaluation, including the reliability of safety devices, and the attention of operators/maintenance personnel, it is necessary to prepare the TAS (Tree Analysis for Safety) at each FA system, and though it is suitable to a system safety evaluation, it is inappropriate as a system safety functional design. In this paper, I show that the degree of surplus indicating the degree of accomplishment of the safety standard at each logical product can be defined. Using this degree of surplus, I propose the method seeking a combination of safety devices which can cope with the multiplexing of safety devices. Thus, I show that the safety functional design for FA systems can be constructed using the safety index
Since a fuzzy classifier with ellipsoidal regions is based on the Mahalanobis distance, the generalization ability is degraded when the associated covariance matrices become singular. In this paper, we discuss two methods for improving the generalization ability : 1) during the symmetric Cholesky factorization of the covariance matrix we replace the input of the root with a prescribed positive value when it is smaller than the prescribed value, and 2) we tune the slopes of the membership functions so that the margins are maximized. We demonstrate the validity of our methods by computer simulations.
In conventional support vector machines (SVMs), an n-class problem is converted into n two-class problems. For the i-th two-class problem we determine the optimal decision function which separates class i from the remaining classes. In classification, a sample is classified into class i only when the value of the i-th decision function is positive. In this architecture, the sample is unclassifiable if the values of more than two decision functions are positive or all the values are negative. In this paper, to overcome this problem, we propose fuzzy support vector machines (FSVMs). Using the decision functions obtained by training the SVM, for each class, we define a truncated polyhedral pyramidal membership function. Since, for the data in the classifiable regions. the classification results are the same for the two methods, the generalization ability of the FSVM is the same with or better than that of the SVM. We evaluate our method for three benchmark data sets and demonstrate the superiority of the FSVM over the SVM.
This paper aims at developing a decision model of how much bonds should the government issue, how much should an annual interest rate and principal rate guarateed be.Firstly, financial risk on the government caused by a big earthquake is described and earthquake bond is defined.Then, by using nonlinear programming a decision model is formulated.In this model the government tries to minimize the financial risk described by the variance on payment subject to the constraint on the expectation on payment by the government and subject to the constraint that the earthquake bond is more attractive for the investor than the other investment plan.Some numerical examples are included for the case of Hanshin-Awaji earthquake.