The magnetic field of a recording head is measured by projection electron-beam tomography with a resolution of about ten nanometers, and the magnetic field closer to the sample surface than the measurement plane is estimated by numerical calculation. The magnetic field at 20 nm calculated from the field measured at 50 nm adequately agrees with the directly measured field at 20 nm. The combination of projection electron-beam tomography and this calculation method make it possible to determine the magnetic field close to a head (air-bearing) surface.
This paper presents some experimental validation results of an already-proposed switching control method for simultaneous achievement of collision avoidance and tracking control for a vehicle in a non-cooperative situation. To validate the method, an experimental control system is made, in which the vehicle is a toy model car possible to remotely control via infrared ray and a camera is used to measure the vehicle's state. After presenting the constructed control system, the effectiveness of the method is investigated with the results obtained from the several control experiments.
One of the important classes of MIMO processes is that of serial processes. We propose a combined control technique of PID control and disturbance observer for the temperature control of a deionized (DI) water heater that has the structure of a serial process. The results of both simulation and actual equipment experiment verify that the proposed method enables us not only to use fewer temperature sensors than the conventional system but also to solve the offset problem that is typical for the serial processes.
This paper introduces an improved quasi-ARX neural network and discusses its application to adaptive control of nonlinear systems. A switching mechanism is employed to improve the performance of the quasi-ARX neural network prediction model which has linear and nonlinear parts. An adaptive controller for a nonlinear system is established based on the proposed prediction model and some stability analysis of the control system is shown. Simulations are given to show the effectiveness of the proposed method both on stability and accuracy.
This paper provides a design method of fixed-structure controllers satisfying multiple H∞ norm specifications by using the covariance matrix adaptation evolution strategy (CMA-ES). The CMA-ES is a kind of stochastic optimization such as particle swarm optimization (PSO), and has been shown to have a good performance for nonconvex optimization problems. However, there are few control applications of the CMA-ES, and therefore, its superiority is not clear in control problems. The effectiveness of the proposed method is demonstrated through numerical examples in comparison with the PSO-based method that has recently been proposed as a good approach.
This paper is concerned with causal linear periodically time-varying (LPTV) scaling for stability analysis. The effectiveness of such a scaling technique was first shown for the robust stability analysis of linear sampled-data systems, and it was then extended to a more general technique called noncausal LPTV scaling. Even though noncausal LPTV scaling has been shown to be more effective than causal LPTV scaling in such a context, this paper aims at showing that causal LPTV scaling, unlike noncausal LPTV scaling, can be applied even to sampled-data systems with static sector nonlinearities. Numerical examples are also studied to compare the effectiveness with another method based on the conventional multiplier technique.
This paper presents an emulation of fuzzy logic control schemes for an autonomous parallel parking system in a backward maneuver. There are four infrared sensors sending the distance data to a microcontroller for generating an obstacle-free parking path. Two of them mounted on the front and rear wheels on the parking side are used as the inputs to the fuzzy rules to calculate a proper steering angle while backing. The other two attached to the front and rear ends serve for avoiding collision with other cars along the parking space. At the end of parking processes, the vehicle will be in line with other parked cars and positioned in the middle of the free space. Fuzzy rules are designed based upon a wall following process. Performance of the infrared sensors is improved using Kalman filtering. The design method needs extra information from ultrasonic sensors. Starting from modeling the ultrasonic sensor in 1-D state space forms, one makes use of the infrared sensor as a measurement to update the predicted values. Experimental results demonstrate the effectiveness of sensor improvement.
This paper is concerned with equivalence of nonlinear systems from a viewpoint of geometric congruence of system orbits in the state space. The notion of orbital equivalence was originally exploited by Sampei and Furuta in the context of the time-scale transformation approach. They gave a fundamental characterization of orbital equivalence in the form of similarity conditions in the system vector-fields; however, it was not satisfactory in the sense that there remains a significant gap between the necessary condition and the sufficient one, mainly due to its treatment of unavoidable singularity of the time-scale functions. In this paper, we intend to cast a new light on this problem with the aid of a viewpoint of “almost” that neglects the inconsistency on measure zero subsets. By introducing the notions of almost orbital equivalence and almost similarity of the vector-fields, we provide a sole necessary and sufficient conditions connecting them. Furthermore, we discuss the combination of the orbital equivalence analysis and a stability issue based on a recent development of the density function approach for stability analysis. The discussion is also illustrated by numerical examples.
A distributed sensor system is highly desirable for detecting, locating, and monitoring fine cracks at unknown locations in advanced ceramics. This paper presents a distributed high-resolution fiber optic sensor based on the Brillouin scattering principle, and its application in ceramic crack detection for the first time. The existence of cracks, together with their locations, is identified by measuring the strain distribution on a sensing fiber bonded to the ceramic surface. By employing the innovative coherent probe-pump interaction technique, the Brillouin sensor developed in this study achieves a high spatial resolution (100 mm) and measurement accuracy. Capable of detecting and locating fine cracks less than 40 µm, the efficacy of the distributed Brillouin fiber optic sensor is demonstrated through experiments.
This paper tackles two design problems, i.e. stabilizing and H∞ control problems, using Gain-Scheduled (GS) state-feedback controllers for Linear Parameter-Varying (LPV) systems under the condition that the scheduling parameters are inexactly measured. The LPV systems are supposed to have polynomially parameter-dependent state-space matrices and the measured scheduling parameters are supposed to have a priori defined bounded uncertainties. Using parameter-independent Lyapunov functions and parameter-dependent scaling matrices related to the uncertainties in the measured scheduling parameters, we give formulations for designing parametrically affine GS stabilizing and H∞ state-feedback controllers in terms of parametrically affine Linear Matrix Inequalities (LMIs). Our proposed methods encompass the design methods of robust state-feedback controllers as a special case. A simple numerical example for an H∞ control problem is included to illustrate our results.
To maintain stable operation of semiconductor fabrication lines, statistical process control (SPC) methods are recognized to be effective. In semiconductor fabrication lines, there exist a huge number of process state signals to be monitored, and these signals contain both normally and non-normally distributed data, which make the application of SPC difficult. The authors have already proposed an SPC method “ESCM” which can solve these problems. However, this method has a problem that it is highly sensitive to outliers contained in the reference data. In this paper, we propose a new outlier removal method to solve this problem. This method uses the “effective standard deviation” we used in the ESCM to deal with a wide range of non-normal distributions. Basic performances of this method are demonstrated by comparing abilities to detect outliers correctly, and to mistakenly detect non-outlier data with conventional methods. In the demonstration, we use various kinds of artificially generated data and real data observed in process tools in a semiconductor fabrication line. Usefulness of this method is demonstrated by studying the effects of this method when used with ESCM.
Robots must be presently taught by human workers to execute given manufacturing tasks. The current problem is that the task of teaching robots is rather time-consuming, especially within the robotic assembly domain. This problem is caused by insufficient accumulation of human expertise that should be reused in this domain. Therefore, a knowledge-intensive method for acquiring human expertise is proposed in this paper. Our method is able to acquire human expertise in the robotic assembly domain by observing robot-teaching demonstrations of human experts. What distinguishes our method from others is that there are two modes of learning: 1. learning from an example directly given by human workers, and 2. learning expertise on error recovery by observing revisions made by human workers in handling execution errors that occur in reusing previously acquired knowledge. The acquired human expertise is required to be represented in a way that satisfies two requirements. The first one is operability so that the representation is easy to transform into robot programs (commands & parameters). The second one is understandability so that the representation is easy for human workers to understand the robot program. A specific robot assembly example is given to illustrate the proposed method.