In a process control field, systematic training of the plant operator is executed using a training simulator of the plant. However, an adviser who guides the trainee is necessary, and the system cannot be called a completely independent one. Therefore, we developed the intelligent training system for the training of start-up operation of the distillation tower plant. It is possible to study about the plant operation without special knowledge about the simulator or the plant, and it provides a graphical interface also. In the intelligent training system, it is indispensable to provide the functions to offer timely optimum guidance information according to the situation. In this paper, we show the method to construct an intelligent training system that is used to realize such functions. Focus is on the control for switching inference to organically connect many modularized functions, and on the state transition model to represent the knowledge in a similar manner the human experts would.
This paper presents a new magnetic field sensor named zero magnetic field type after its detection principle. The detection principle is as follows: When the magnetic field level in the core used as a sensor is shifted by an unknown field to be detected, a controlled field is so superimposed in the opposite direction to the unknown field that the original zero magnetic field state is restored. Then the unknown field can easily be detected from the controlled field. Consequently, the detection characteristics are not affected with the geometrical dimension, magnetization curve and demagnetization effect of the core used as a sensor, and moreover, the temperature variation of the core, the winding resistance and the inductance do not also directly degrade the detection performances. This paper certifies that the proposed sensor certainly has such excellent properties. In addition, the followings are clarified: (i) The upper limit of the detectable range does not exist in principle; (ii) The detection accuracy is 0.02% to the full scale 20 kA/m and realizes the order of 4×10-4%/°C for the temperature variation of -76°C to 300°C; (iii) The magnetization property required to the core used as a sensor is that it has as sharply a rectangular B-H loop as possible; (iv) As a result, the design of the sensor is greatly simplified.
Fluctuation in the polarization state of the output light from a single-mode fiber degrades the coupling efficiency to an optical integrated circuit and the S/N ratio in heterodyne system or optical measurement systems. This paper describes the analysis and experimental results of the lightwave polarization control scheme which has the features such as low insertion loss, low control power and simple configuration. The scheme consists of three phase shifters with two moving λ/4 plates and a fixed λ/4 plate, and a feedback control system. We analyze the polarization state transformation and derive the condition required for the optimal control. Experimentally, we constructed an endless rotative polarization controller using step motors which has excellent performance and durability, and in order to confirm this fact, stabilized the polarization state of the light output from single-mode optical fiber in the snow.
For the viewpoint of the Probabilistic Safety Assessment, quantitative analysis of dependent failure, particularly CCF: Common Cause Failure, is one of the most important and difficult problems to deal practicaly. Parametric Model method, a kind of the methods for analysis of dependent failure, include several kinds of models, such as the Beta Factor model, the BFR model and etc.. Since the group of the Parametric Models have not been enough studied systematically yet, they are not neccessarily applied accurately for the analysis of CCF. Therefore, it is neccessary to systematize the group of the Parametric Models for applying the models effectively and successfully to the analysis. In this paper, clear classification of the above Parametric Models was obtained and their characteristics were made clear. As a result, it is easier to obtain more useful procedure for accurate applications of the Parametric Models to the analysis of CCF with actual data.