計測自動制御学会論文集
Online ISSN : 1883-8189
Print ISSN : 0453-4654
ISSN-L : 0453-4654
53 巻, 10 号
選択された号の論文の5件中1~5を表示しています
論文
  • 石原 新士, 山北 昌毅
    2017 年 53 巻 10 号 p. 527-538
    発行日: 2017年
    公開日: 2017/10/13
    ジャーナル フリー
    This paper addresses state estimation problems for parametric uncertain nonlinear systems. We developed Robust Nonlinear Kalman filters (RNKF) by using two linearization methods to consider the influence of parameter uncertainties of covariance matrixes. The first method is Taylor expansion which is used in Extended Kalman Filter and another is equivalent linearization. The RNKF is more accurate than conventional NKF. However the RNKF has some disadvantages: (1) when there is no parameter uncertainty, estimation accuracy of the RNKF may be inferior to that of the NKF and (2) the estimated values of the RNKF can have some offsets by the influence of parameter uncertainties. So, we developed an adaptive RNKF by introducing an adaptive scheme into RNKF to automatically tune the influence of parameter uncertainties. We also developed Approximated Minimum Variance Unbiased Filter (AMVUF) by solving constrained optimization problem to reduce the influence of parameter uncertainties. Furthermore, we develop a new simultaneous states and parameters estimator for nonlinear systems based on the AMVUF. We confirm the validity of the proposed methods by Monte Carlo simulations.
  • 豊田 充, 申 鉄龍
    2017 年 53 巻 10 号 p. 539-546
    発行日: 2017年
    公開日: 2017/10/13
    ジャーナル フリー
    The optimization problem of stochastic logical systems is studied in this paper. To deal with a system without knowledge of the objective function, a Bayesian optimization framework is extended with the learning algorithm called Gaussian process. Firstly, the regret bound, which represents the difference between the true optimal value and the achieved objective function value, is evaluated with exploiting the statistic features of Gaussian process. A numerical example is illustrated for the purpose of validation on the optimization algorithm afterward.
  • 吉田 智哉, 小林 一行, 鈴木 郁, 山下 浩二, 渡邊 嘉二郎
    2017 年 53 巻 10 号 p. 547-556
    発行日: 2017年
    公開日: 2017/10/13
    ジャーナル フリー
    This paper describes a new type of level switch which detects the existence of light, hot particulates such as ash in a tank. Here we employed an acoustic tube method. A port of the tube was driven by a piezo-sounder at less than its resonance frequency and the length L of the tube was selected to have the same resonance frequency, i.e., L=n×(half of the wave length). If the other port is covered by particulates, the acoustic impedance at the resonance frequency increases, which yields increases of the electric impedance of the piezo device. The sound velocity increases in proportion to the temperature, which may deteriorate the optimal measurement condition, i.e., the length of tube is not optimal at the altered temperature. However theoretical studies showed that covering a detected port by particulates mostly influenced the acoustic viscosity which is less influenced by the temperature change. As a result of theoretical studies, we showed that the proposed method can work in a certain temperature range. We verified the theoretical results using experiments which showed that the proposed method worked in the temperature range from 5°C to 80°C with a piezo sounder with 760 resonance frequency and an acoustic tube with half the wave length. Moreover, it detected particles with low bulk density such as those with 13kg/m3. We also conducted an experiment to judge the existence of gravel heated to 450°C in at a room temperature of 23°C which was successful.
  • 吉田 達哉, 辻内 伸好, 伊藤 彰人, 鞍谷 文保, 安藤 博昭, 青柳 多慶夫, 村山 栄治, 荒井 邦晴
    2017 年 53 巻 10 号 p. 557-563
    発行日: 2017年
    公開日: 2017/10/13
    ジャーナル フリー
    Response to a regulation of exhaust gas, saving energy and automation are the most important assignment in the construction machinery industry. In late years, automated construction systems including ICT machinery are researched and partially in practical use. The systems require simulation techniques in early stage of the design. A highly reliable model enables large cost reduction if nonlinearity and uncertainty of the machinery is reproduced with more precision. This paper refers to high-accuracy parameter identification for a hydraulic servo model used in controller developments. The parameters are identified from a measurement data recorded with a commercial machine. In the model, input is rate of lever operation and output is operational velocity of hydraulic cylinder. A plant model consists of a dead band, dead time and quadric transfer function, and parameters of the plant model is identified by using differential evolution (DE). The parameter identification by DE shows superior accuracy and convergence than one by genetic algorithm. The result shows that a useful plant model for development is identified by DE.
ショート・ペーパー
feedback
Top