A preliminary inquiry on the process of work such as carrying or setting up large-sized equipments for constructing new electric power plants is indispensable for assuring efficient works and safety operations. We have been developing a computer aided beforehand-study system which assists unexperts who has not much experience about constructions of electric power plants. In this paper we propose an automatic search algorithm on suitable routes for carrying in electric power facilities; it works as function of the system mentioned above. In the algorithm the automatic route search algorithm for autonomous moving robots and manipulators is improved to fit the construction of electric facilities; the new algorithm can find not only one route but also some candidate routes and take care of the workers' space around the moving equipment. The proposed algorithm has been applied for examining the process of transformer installation work in a substation.
This paper describes the cartographic information processing system using verbal representation. The verbal representation is a new method for representing the multi-dimensional surfaces, which human beings use for representing of topography. This method has not only advantages of representing relationship between mountains and valleys, but allows one to form a concept of shapes through a little bit of information because the relationship is represented by labels (or words). In this paper, the words used for the verbal representation are defined in 3D space. Next, the outline of the cartographic information processing system is described. The system has two subsystems, which are shape representing subsystem and shape recognizing subsystem of topography. Each subsystem constructed some units. Finally, the recognizing process of the topography at the shape recognizing subsystem is shown.
Recently, robust absolute stability problems are argued for Lur'e systems which contain transfer functions with real parametric uncertainties. In dealing with this kind of systems, their linear parts with parameter perturbations are represented by interval plants. The Popov's theorem is known as a way of solving the classical absolute stability problem when nonlinear characteristics of Lur'e systems is time-invariant. It is recently shown that this theorem can be extended to the case where the linear plant is replaced by interval plants. The absolute stability of such systems is checked by the geometrical relation between a Popov line and the Popov loci of several extreme plants in the complex plane. However, it is restricted to the case that the lower bound of the sector equals to zero. In this paper, an extension is made so that nonlinear part has a general sector. This can be done by making the transformation for original systems and applying the Popov's theorem to the transformed systems. We show that the absolute stability of original systems can be guaranteed if we have only to check the Popov's condition for the systems transformed from several extreme plants.
In parallel processing, it is more efficient to handle a job by decomposing it into finely-grained tasks than trying to manipulate the larger ones. However, if the tasks are finely grained, the overhead related to the synchronization of the processors and the shared memory accesses correspondently increases. We propose an approach for reducing this overhead by constructing a parallel scheduler which considers the preparation time during plural searchings. In this scheduler, (1) the processes share two upper bounds derived by each searching, (2) the lower bound function is built taking into account the preparation time, and (3) the branching in the search tree is made by a heuristic node selection considering the node level. An application of the proposed scheduler based on a multi-DSP system to the computation of the inverse dynamics for robot arms shows that the scheduling time can reduced to be more than twice shorter than that of the conventional scheduler, for the case of four processors.
This paper presents a method for drawing power system networks using fractal technique. It is quite common to represent power systems as visual network graph. The drawing of such graph, however, was performed manually by human-hand. Automatic drawing makes possible to represent power system flexibly. The mathematical problem seems to be similar to “How to draw planar graph”. However, unlike to the planar graph problem which treats eassentially non-separable, power systems contain a few loops. Fractal graph is utilized in this paper. Our problem is defined as follows: (1) All vertex must be on the grid. All branches must be composed of horizontal or vertical line segments. The vertex that has more than 5 degree can be divided into sub-vertices. (2) Geometrical relationship between two vertices must be preserved as long as possible (In this paper, “Geometrical” means “Directional”). (3) Branches should not cross each other as long as possible. The method is applied to some practical power system networks. Although the method has some week points for grid systems, the test results show the feasibility of the proposed method.
Acceleration anomalies in paralell branch and bound algorithms may play a key role in solving NP-complete problems. As an analytical tool to deal with these phenomena, we propose a probability distribution model. From the computational results based on this model, we verify that the acceleration anomalies are accentuated by the appropriate handling of the various node selections, and if the applied lower bound function is of poor precision, this leads to a highly effective acceleration. We apply these ideas to the task scheduling problem for parallel processing and confirm that the better solutions can be obtained with plural search strategies.
Deadbeat control of the discrete-time systems has the specific feature that the error is settled to zero in the finite control step. But in the practical system, there exist input saturation on the control signal, and undesirable effect such as reset wind-up may be occured. In this paper, a new design method of deadbeat control system is proposed for the plants with input saturation, and apply it to a speed control of the current type inverter induction motor. The deadbeat control signal which considers saturation limitations can be easily obtained by matrix computation. Then, the deadbeat controller not only for the series integral action but also state feedback compensation can be design by using the sequence of the control signal. Illustrative numerical example are presented, and it is found from experimental results that the method proposed here seems to be useful for the practical system.
Analitical results of operation of magnetostrictive torque sensor were shown. A highly linear, little hysteresis and quick response torque sensor was constituted using a two-core multivibrator bridge circuit. Torque sensitivity was improved by adding two diodes between baseemitter electrodes of the two switching transistors in order to discharge the commutation capacitors. An optimal value of the magnetostrictive layer inductance was found theoretically and experimentally to realize a maximum sensitivity of torque sensing.