Case-Based Reasoning is a strategy which gets the most similar source case in the memory by the hints of the target case and gets the answers by the instruction of the source case.

First,three kinds of retrieving models are adopted to retrieve the set of source case which the symptom is similar to the symptom of suspected transformer,then an algorithm of synthetic estimation is introduced to achieve the optimal one of source cases,the diagnosis result of suspected transformer could be acquired by the comparison of symptoms between the optimal case and the suspected transformer at last.

After great many calculations the topological structure and training samples are screened and chosen, through the expert system based on case-based reasoning the optimal source case is retrieved out and the diagnosis conclusion is given.

And it confirms the important degree of all types of characteristic attributes. It aims to build up a retrieval based on case's key attributes. Then it makes use of the nearest neighbor strategy to implement the similar matching between the target case and source case.

This paper studies the intraband crosstalk mechanisms in wavelength division multiplexed (WDM) ring networks. In WDM ring networks, the crosstalk analysis can be simplified to the single crosstalk source case.

Theoretical analysis shows that the performance of GML algorithm is consistant with DML's in incoherent sources case, and it improves greatly in coherent source case.

Furthermore,quantitative calculation shows that compared with single source case,the SNR of reflected wave for 12-unit phased-array source increases by 19.25dB average- ly;

For the oblique light source case, the reconstructed surface is the viscosity solution to a different partial differential equation.

The surface is a viscosity solution of an Eikonal equation for the vertical light source case.

Thus this study provides empirically derived evidence that a multiple data source case study is more trustworthy than a comparable single data source case study.

This study investigates a multiple data source case study with the objective of identifying whether more findings, trustworthier findings and other findings are made using multiple data source triangulation, than had a single data source been used.

The core issue of analogical reasoning is the transfer of relational knowledge from a source case to a target problem.

In this paper, we consider the nonlinear partial differential equation which is obtained for the arsenic diffused in the silicon. In the constant source case, the analytical expression of its solution is obtained. Then obvious boundary surface presents. At the certain time the diffusion profile is the parabolic curve. The diffusion front is steeper than that in the linear case. The shiftingrate of diffusion boundary is closely related with the total quantity of the diffusion material. These characteristics...

In this paper, we consider the nonlinear partial differential equation which is obtained for the arsenic diffused in the silicon. In the constant source case, the analytical expression of its solution is obtained. Then obvious boundary surface presents. At the certain time the diffusion profile is the parabolic curve. The diffusion front is steeper than that in the linear case. The shiftingrate of diffusion boundary is closely related with the total quantity of the diffusion material. These characteristics also present in the constant surface concentration case

In point-source case, solving an integrated equation of the charge densityon 3-dimensional electrostatic interfaces is made by means of the momentmethod through the sub-domain matching method combined with the point-matching method, choosing the Legendre polynomial as the base function ofthe whole-domain and using a transformation technique from sub-domaininto whole-domain in the computing process. The comquting accuracy(relative average square-root error)for several models amount to 0.69%,The computing...

In point-source case, solving an integrated equation of the charge densityon 3-dimensional electrostatic interfaces is made by means of the momentmethod through the sub-domain matching method combined with the point-matching method, choosing the Legendre polynomial as the base function ofthe whole-domain and using a transformation technique from sub-domaininto whole-domain in the computing process. The comquting accuracy(relative average square-root error)for several models amount to 0.69%,The computing speed of this method is about 3.5 times as great as that ofthe finite element method and the finite difference method.

Real-time implementation of high resolution algorithms is difficult, especially in coherent environment' Starting from my previous work with the IMP (Incremental MultiParameter) algorithm['.s], 1 now propose a new method to speed up its convergence procedure.The original IMP algorithm estimates the source directions iteratively. With the addition of each new source, iterations begin anew, resulting in greater number of iterations.Wax et al[2] and Viberg et al[3] offer ways of determining number of sources....

Real-time implementation of high resolution algorithms is difficult, especially in coherent environment' Starting from my previous work with the IMP (Incremental MultiParameter) algorithm['.s], 1 now propose a new method to speed up its convergence procedure.The original IMP algorithm estimates the source directions iteratively. With the addition of each new source, iterations begin anew, resulting in greater number of iterations.Wax et al[2] and Viberg et al[3] offer ways of determining number of sources. Thus my new method can start on the basis of known number of sources and thus save iterations as compared with original IMP.We take a 3-source case. Fis. 1 shows that in a particular simulation trial the IMP requires 23 iterations and Fig. 2 shows that in the same trial our modified IMP requires 14 iterations. Eqs. (4) and (5) are made use of in obtaining Fig. 2.For the same 3-source case, the results of 100 simulation trials presented in Fis. 3 indicate that arbitary assignment of initial estimates does not affect convergence.Again for the same 3-source case, Fig. 4 presents the results of 100 simulation trails for both our modaled IMP and for the orisinai IMP algorithm. These 100 trials indicate that original IMP needs most probably 20 iterations and our modified IMP algorithm requires most probably only 14 iterations, or only about 70% as compared with orisinal IMP, which means a saing of 30%.For a 4-source case, our work shows a saving of about 40%. As the number of sources increases, a further increase in saving of iterations prcentage-wise may be expected.