巻号一覧
9 巻 (1973)
- 6 号 p. 625-
- 5 号 p. 499-
- 4 号 p. 373-
- 3 号 p. 251-
- 2 号 p. 123-
- 1 号 p. 1-
9 巻, 6 号
選択された号の論文の17件中1～17を表示しています
- |<
- <
- 1
- >
- >|
-
輻射伝熱による連続加熱炉の特異制御椹木 義一, 池田 三郎1973 年 9 巻 6 号 p. 625-632
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThis paper deals with a singular control problem which appears in the optimal control of flow processes described by first-order partial differential equations. By applying Sirazetdinov's maximum principle to flow processes, optimal control and various conditions to be satisfied for the singular control are derived.
When system performance is evaluated at the outlet of a flow process, the following results are obtained.
i) The optimal control is attained by only one switching from a bang-bang operation to a singular one.
ii) The singular control is determined by the stational heat or mass balance at the outlet of the flow process.
By the use of quasilinearization technique, a numerical method is developed to calculate the singular control. To illustrate the present results, a continuous furnace under a radiative heat-transfer is analyzed.抄録全体を表示PDF形式でダウンロード (912K) -
宮下 照夫, 佐藤 力1973 年 9 巻 6 号 p. 633-639
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThis paper investigates 1/ν subharmonic oscillations in the following nonautonomous system:
x+x=εf(x.x)+Ecosνt. (0<ε<<1, ν=2, 3, …)
The analysis is made essentially by an averaging method which includes a transformation from a nonautonomous to an autonomous equation.
Here non-numerical analysis by computer is applied to the above transformation in order to show the transformation in a table. When the nonlinear function f(x, x) is limited to the form xpxq, the transformation is expressed in an algorithm, and non-numerical tables of the transformation are made for p, q=0, …, 7 and ν=2, 3, 4.抄録全体を表示PDF形式でダウンロード (742K) -
松山 泰男, 白井 克彦, 秋月 影雄1973 年 9 巻 6 号 p. 640-646
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThe state of a neuron is described by the neuron potential (NP), which often has the nature of a stochastic process.
In this paper, the stochastic information processes of the spontaneous type neuron and the forced type neuron are discussed through the density of the first passage time when the NP reaches the threshold, and its subsidiary statistical quantity, the transition probability density function. These quantities can be obtained from Kolmogorov's equation and the law of probability conservation. However, the analytical solution is obtainable only in very simple cases. Hence, we took the position of obtaining these quantities by means of numerical analysis and investigating some rather complex properties of the neurons. Accordingly, we could discuss the varying threshold problem and the case when input pulse trains carry the PFM information.
And in the case of the periodic PFM, the output spike interval density becomes the important quantity because it is measured in practice. Therefore we clarified its relationship to the first passage time density and then presented the method of approximation.抄録全体を表示PDF形式でダウンロード (793K) -
木村 英紀1973 年 9 巻 6 号 p. 647-651
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーIf the state transition process proceeds so slowly that the system can be regarded as staying in the static equilibrium state at all times during the process, it is called a quasi-static process by physicists. The quasi-static process can be used to control a kind of system in which system dynamics may be a source of unwelcome results.
In this paper we consider the mathematical foundation of the quasi-static process from the viewpoint of system theory. The main result is the derivation of a sufficient condition for the possibility of the quasi-static state transition. We also discuss the application of the quasi-static process to the problem of controllability for nonlinear systems.抄録全体を表示PDF形式でダウンロード (570K) -
小宮 勤一1973 年 9 巻 6 号 p. 652-657
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーA pulsed wire anemometer which consists of a pulsed wire and two sensor wires was developed for the measurement of a highly turbulent flow by detecting the flying time of a heated tracer. This paper describes the effects of a high pass filter and of the viscous wake of the pulsed wire on the signal which is detected by the sensor wires.Theoretical analysis shows that the effect of the high pass filter is remarkable at the tail part, but is very small on the front part of the signal. This means the high pass filter has little effect on the measurement of the flying time. To measure the response curve precisely, an Analogue-to-Digital converter and a core memory system is used. The experimental response curve is compared with the theoretical one, and close agreement between the thoertical and the experimental signals is obtained. The effects of the viscous wake on the signal are examined by the same measurement system. The results show that the viscous wake of the pulsed wire effects the flying time especially for low Péclet number flow measurement, and the optimum condition for the probe design is obtained.抄録全体を表示PDF形式でダウンロード (678K) -
坂和 愛幸1973 年 9 巻 6 号 p. 658-664
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThis paper is concerned with questions regarding observability and state determination on the basis of the measurement data from a finite number of sensors. The questions are studied for distributed-parameter systems described by linear partial differential equations of parabolic type. First, observability is defined and the necessary and sufficient condition for the observability is given. It has become clear that for the observability of the system, at least the same number of sensors is needed as the multiplicity of a differential operator of the partial differential equation.
When we determine the state of the system at t=T on the basis of the measurement data for 0≤t≤T, it is required that the state to be determined should depend continuously on the measurement data. We obtain the sufficient condition for that requirement. Locations of sensors for ensuring the observability and the continuous dependence of the state on the measurement data are discussed for several examples.抄録全体を表示PDF形式でダウンロード (822K) -
西村 卓也, 長田 朗1973 年 9 巻 6 号 p. 665-671
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーIn this paper we consider a method of constructing observers for linear time-invariant, discrete-time systems. The derived results are as follows.
It is assumed that a system S to be observed is completely observable. Then it is shown that an observer for S always can be constructed. And its constructing method is described precisely. The order of the observer is n-m, where n is the order of the state of S and m is that of the output. The number of steps necessary to make the output of the observer coincide with the state of S from the beginning of observation is ν0-1 where ν0 is the observability index of S. This observer is a minimum-time and low-order observing system.
The behavior of this observer is also considered when it is used to observe the linear functions of the state of S.抄録全体を表示PDF形式でダウンロード (789K) -
北嶋 龍雄, 示村 悦二郎1973 年 9 巻 6 号 p. 672-677
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThis paper is concerned with improvement of the convergence rate of the gradient method. If the Hessian of a quadratic performance index has the maximum spectra M and the minimum spectram (M≥m≥0), the convergence rate of a gradient methods for this problem generally depends on the ratio M/m. In this paper, a variable transformation is introduced to make the ratio M/m smaller. Based on this transformation technique, the modified gradient method is developed. An example is given to show the effectiveness of the proposed method.抄録全体を表示PDF形式でダウンロード (613K) -
荻野 修作, 伊藤 一壽1973 年 9 巻 6 号 p. 678-687
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーIt is well known that limit cycles or periodic solutions existing in a single-loop feedback system with one nonlinear element are analyzed by the describing function method.
Johnson used the perturbation method in order to calculate the correction terms of the approximate periodic solution given by the describing function method.
However Johnson's theory is essentially applied only to a nonlinear element whose describing function is real, as Gibson stated in his book. So the procedure for the case in which the describing function is complex is developed in this paper. It is an extension of Johnson's results.
The procedure by the perturbation method in this paper is similar to Johnson's. However the basic differential equation system is different from Truxal's or Gibson's description and is not obscure.The calculation procedure for a third-order system having both saturation and hysteresis is given in detail, and numerical examples are given also.
In the numerical examples. the exact values of both amplitude and angular frequency of the periodic solution are obtained by numerical analysis, and the relative errors of the approximate values of both amplitude and angular frequency obtained by the describing function method or by the perturbation method are evaluated. Thus it is concluded that the application of the perturbation method is also effective in our case.抄録全体を表示PDF形式でダウンロード (1120K) -
前川 禎男, 雛元 孝夫1973 年 9 巻 6 号 p. 688-695
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThis paper describes a method for identification of a class of nonlinear systems that are represented by a cascade connection of zero-memory nonlinear and linear elements. The characteristics of the zero-memory nonlinear element are approximated by the polynomial equation, and the dynamics of the linear element are represented by the impulse response.
This method is as follows: the same input signal is applied both to the unknown system and to its mathematical model. Then the parameters of the model are adjusted so as to bring about the same output at all times through minimization of the performance index, which is the squared difference of the outputs of both systems. And when the unknown system is sufficiently approximated by the model, the dynamics of the unknown system are recognized from the model.
The model of the unknown system is composed such that the output is expressed as a linear combination of the vector function generated by powers of the input signal and the parameter vector introduced from the system parameters. The iterative estimation of the parameters is based upon a new improved gradient method. That is, the coefficients of the increments of the estimated parameters are considered as functions of time which are adjusted at each step to the optimum value for quick convergence, while they are constant in the ordinary steepest gradient method.
The convergence condition of the estimated parameters and the optimum condition for quick convergence in this method are studied theoretically.
This method guarantees for the estimated parameters to converge from any initial state and does not need a complex procedure for the execution of the parameter estimation algorithm.
Two examples are simulated by a digital computer and these results support the theory described in this paper.抄録全体を表示PDF形式でダウンロード (892K) -
入江 寿一, 石崎 長光1973 年 9 巻 6 号 p. 696-703
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーOperation of thyristor choppers at high frequencies reduces device ratings, filter size and response time in the control system. However, most thyristor choppers require a long time for preparation of the commutation i.e. the charging time of a commutating capacitor etc., and the voltage range at high frequencies is small. To reduce the preparation time is one of the most important points of the high frequency thyristor chopper.
This paper describes a new thyristor chopper circuit which has a commutating circuit separated from the load-main-thyristor circuit and requires no preparation time of the load-main-thyristor circuit for the commutation. Moreover, as a result of separating the commutating circuit, the chopper can recover automatically from a commutation failure, and can operate with a very light load. Therefore, it is suitable for the control device.
An experimental circuit of 100V, 10A has a voltage range of 0.1-1 at 3.5kHz and has a commutation loss of 12-27W at the same frequency.抄録全体を表示PDF形式でダウンロード (1224K) -
古荘 純次, 嘉納 秀明, 増淵 正美1973 年 9 巻 6 号 p. 704-712
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThe problem of designing optimal observers of order n-m and n for an nth order plant with m outputs is considered. The cost functional for the regulator is the usual integral quadratic form. In this paper, the case of the known means and covariances of the initial conditions, and thee case of the uniform distribution of the initial conditions are discussed by using the degree of optimality which quantitatively measures the total performance of the control system.
The necessary condition is derived for the optimal observer of order n-m. This condition is much simpler than the one Newmann derived when the means and covariances of the initial conditions are known. Some examples show that this necessary condition is useful.
In the latter half of this paper, the observer of order n is considered. It is shown that the cost can be decreased by making the real part of some eigenvalues of the observer approach minus infinity while keeping the eigenvalues in a certain mutual relation. For this case, Bongiorno and Youla showed that the cost would be increasingly large when the real part of all the eigenvalues of the observer approach minus infinity.抄録全体を表示PDF形式でダウンロード (995K) -
馬場 〓一1973 年 9 巻 6 号 p. 713-718
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーIt is well known as the optimal control problem with discontinuity that whenever the system trajectories hit the boundary either of the state variable constraint sets or of the sub-regions on the state space, the equations describing the system change.
In this paper, a computational algorithm is given which solves this problem numerically for linear sampled-data systems. Also the theoretical background is presented to show that this algorithm is a modification of the optimization method of Vector Performances. This algorithm has two phases, referred to as Algorithms I and II. The sample instants when the system trajectories hit the boundary of the state variable constraint sets are determined by Algorithm I. Then the given objective function is optimized subject to the input and the state variable constraints by Algorithm II with the aid of the fixed sample instants using Algorithm I.
The characteristics of this algorithm are as follows. This problem is of the decomposed type, but Varaiya's decomposition algorithm can not be used here. However the singular control problem arising in problems of this type can be dealt with by iterative computation. The convergence of this algorithm is exactly the same as that of Vector Performances Optimization Algorithm.抄録全体を表示PDF形式でダウンロード (865K) -
山浦 弘夫1973 年 9 巻 6 号 p. 719-728
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーA particular class of optimal control problems in which the Hamiltonian function is linear in the control variable may have the singular solution to be optimum as a part of the whole process. This happens when the switching function vanishes for a limited while. The singular solution has two attributes, maximum and minimum performance indices, and they must be classified to find out the real optimal process.
This paper proposes the so-called selection rule for this purpose. The selection rule can be applied to third-order terminal control problems in which singular trajectories are decided unconditionally in the state space. To specify the geometrical property, a vector method is introduced into the calculus of variation. By this method the sufficient condition for the optimum is shown to be related to the sign of a vector product component, and the selection rule is stated simply using the directions of two vectors in a plane. The transition law for the optimum trajectory is derived from the discontinuity of the second derivative of the switching function. This law is able to indicate the optimum transition from bang-bang to singular trajectory and vice versa among candidates, and thus eliminates the necessity of comparison of values of the performance index along each trajectory.
Three examples are worked out in detail to illustrate the application of the proposed graphic method.抄録全体を表示PDF形式でダウンロード (1032K) -
舘 〓1973 年 9 巻 6 号 p. 729-738
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーA method is proposed for separating a desired periodic signal, i.e., for detecting its direction and estimating its Fourier components, from other periodic signals scattered in a field contaminated with Gaussian random noise.
The auto-bispectrum of a signal composed of a periodic signal plus Gaussian random noise preserves the information about both relative amplitudes and relative phases of the components of the periodic signal, while it cancels out the over-all phase shift caused by the time delay of the signal.
The cross-bispectrum of the two waveforms observed at two different points, on the other hand, preserves information about the inter-regional phase shift of each component.
These characteristics are clarified and used as the basis, of the method. The direction of the desired signal is measured by means of the auto- and cross-bispectra of the observed waveforms. The Fourier components of the desired signal are estimated by using the auto-bispectrum, and the desired signal can be reconstructed by means of Fourier synthesis.
An example of reconstruction is shown, and the estimation errors of the program used are discussed. The estimated value varies because of finite averaging, and the distribution of the deviation from the true value can be regarded as almost normal. The relative error diminishes almost linearly as the signal/noise ratio increases when the signal/noise ratio is larger than 1.0.抄録全体を表示PDF形式でダウンロード (1161K) -
室津 義定, 大場 史憲, 小沢 功1973 年 9 巻 6 号 p. 739-744
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーThis paper presents a new method for finding the optimum solution for a linear programming problem with uncertainty in the coefficients.
Total cost is defined by adding penalty cost to activity cost when the constraints are violated, and a stochastic programming problem is set up to minimize the expected total cost. It is shown that the problem is reduced to a convex programming. The algorithm for finding the optimum solution is also presented, using the gradient method.抄録全体を表示PDF形式でダウンロード (709K) -
中川 正雄1973 年 9 巻 6 号 p. 745-746
発行日: 1973/12/30
公開日: 2009/03/27
ジャーナル フリーPDF形式でダウンロード (303K)
- |<
- <
- 1
- >
- >|