The exponential stability and trajectory bounds of the motions of equilibria of an associative neural network under structural variations while learning a new pattern are investigated. The filtering ability of the associative neural network whose input is contaminated by noise is analysed. The obtained results are expected to instruction for further studies of the learning associative neural networks.

This paper analyzes transient characteristics of high temperature CMOS inverter and gate circuits, and gives computational formulas of their rise time, fall time and delay time. It may be concluded that the transient characteristics of high temperature CMOS inverter and gate circuits deteriorate due to reduction of carrier mobility and threshold voltages of MOST's and increase of leakage currents of MOST's orain terminal pn junction. The calculation results can explain experimental phenomena.

By using singular value decomposition, this paper elaborates the quantitative relation among oriented energies of the measurable voltage-change matrix, fault-information matrix and departure-disturbance matrix in departure-fault circuits, describes separation of the maximal oriented energy subspace, and proposes a maximal closeness criterion. Some novel algorithms, under the name of Generalized Inverse Projection (GIP), are presented to locate k-faults. The location accuracy of the GIP is so high as that of the excellent Nonlinear Least Square estimation (NLS), but testing time required by the GIP is only one over several hundreds of that by the NLS.

The fast and efficient method for measuring high-order Volterra kernels along axis is given. This is a sequel of the authors former papers (1990). Using the conclusions of those papers mentioned above, and combining the DC component with the input signals, the high-order kernels of nonlinear system H₂(jω,0),H₃(jω,0,0), H₃(jω₁,jω₂,0) can be, measured with many points.

It is the key to suppress ground clutter for airborne early warning (AEW) radars. In this paper, a temporal-spatial nonadaptive joint filtering processing approach is proposed to suppress clutter for AEW radars. The computation load of this approach is small, and the performance for detecting slowly moving targets is better than that of the conventional cascaded nonadaptive processing.

Some basic properties of the time-sidelobe of arbitrary sequence s auto-correlation function have been studied and several theorems of the time-sidelobe of sequence s autocorrelation function are proved also in the paper. Some practical exam-ples are given using the theorems. The first example is to search low sidelobe codes, and the second example use sidelobe s self-complementary property to cancel sidelobes of 63 bit PN code. Finally, it is point out that the method for cancelling sidelobe needs only add operation.

It is an ideal selection to use a-Si:H film, which is cheap and with small lag, as the photoconductive layer of liquid crystal light valve. By using the method of impedance matching, an a-Si:H liquid crystal light valve is designed and fabricated, and the clear image on a large screen projective system is obtained. The resolution of the valve is 201p/mm. The sensitometry measurement shows that the projection light level varies linearly with the writing light level when the latter is below 100μW/cm². The measurement of transient response at the writing light level of 100μW/cm² indicates that the rise time is 60ms and the decay time is 80ms.

The effect of metallization thickness on a leaky wave antenna based on a metal-strip-loaded dielectric insert waveguide is investigated. The intrinsic equation is set up by. the extended version of the network analytical method of electromagnetic field. Numerical results show that the effect of metallization thickness is smaller on the direction of main beam and very noticeable on the beamwidth and aperture efficiency.

The field-enhanced carrier generation of deep level centers in semiconductor space charge region has been studied. This paper points out that both Cou-lombic emission and non-Coulombic emission must be considered into the carrier generation rates of deep level centers. On this basis, a formula of generation rate has been given. The analysis of the computing results shows tkat previous model, in which only Coulombic emission was considered, is rather simple, and tke tkeory proposed in this paper can more satisfactorilyt explain the experimental results.

The linear relationship between Doppler centroid and range walki, in spaceborne SAR data is analysed,and a new method to estimate Doppler centroir is proposed, which is called amplitude correlation method. Compared with Li s(1985) clutterlock method that is widely used now, the new method has much less compv tation burden and is able to give higher estimation accuracy for a quasi-homogeneous or a non-homogeneous scene. This is clearly verified by the experimental results of estimating Doppler centroid for S...

Based on the high frequency approximation theory, the complex ray expansion of plane waves are derived. The results obtained may be regarded as the basis of the numerical expansion of plane wave, which has been used successfully in some problems.

The problem of a linear dipole antenna mounted on a dielectric slab is investigated theoretically and experimentally. A set of integral equations for the current vector along the dipole and the field in the slab is derived by the concept of polarized current. The equations are solved numerically by moment methods. Finally, the computational results are given and compared with experimental ones. The influence of the dielectric upon the dipole is revealed.

Data are sampled manually using a electromagnetic induction type underground metal pipe detectror. After that, an empirical formula for a metal pipe s second field distribution function is obtained by fitting the data. On the other band, the depth of underground metal pipe can be detected by use of the detector. Then, the pipe s diameter can be calculated by use of the half width of the distribution func-tion s curve and the depth. The error of the result is ±10% of the pipe s diameter.

Some new experimental results of optically controlled dielectric resonators (DRO) are presented. A very stable X-band DRO was found to be optically tunable up to 17.5MHz with modulation rate of 1.17MHz/mW with red light illumination. And an even higher modulation rate of 2.24MHz/mW with illumination of violet light was obtained. Instead of a drop in optically controlled DRO output power, a little rise of output power was achieved.

By analysing the difficulty of previous flip-flops with a high radix,this paper proposes a logic design scheme with two presetting inputs. The circuit of a quaternary CMOS flip-flop is designed by using the transmission function theory. The result shows that its structure is simpler and its processing speed is higher than that of two binary flip-flops which store the equal information.

A derivation of the Z-domain charge coupled model is given for a general switched-capacitor branch. A new loop charge method for the Z-domain analysis of multiphase switched-capacitor networks is presented.

This paper describes some new active bandpass filters using only a single second-generation current cenveyer (CCII), which are superior to early realizations for using fewer components. By choosing proper values of the circuit elements, they can be also used to simulate RCD series resonators.

Developments of the microcomputer applications to the microwave radiometer system is reviewed. A microcomputer controlled and demodulated 8mm total power radiometer with periodic calibration is realized. Its opreation principle, analysis method, technique superiority and test results are also described.