BP网络在位场解释中的应用

介绍了ＢＰ网络的特点，阐述了它在位场资料解释中应用的方法、特点和可行性。通过模型计算证实了该方法在位场异常识别中的效果，正确率达９７％。此外，通过修改网络单元的激发函数，增强网络的稳定性和内插功能，使ＢＰ网络在多层密度界面反演中得到满意的结果，模型和实例计算证明了该方法的有效性。

THE APPLICATION OF BP NETWORKS TO THE INTERPRETATION OF POTENTIAL FIELD DATA

In the paper,the features of BP network are intruduced. The methods,features and possibility of the application of BP network to the interpretation of potentialfield data are also disctissed.The effectiveness of the method has been proved by modeltest which use BP network to recognize potential anomaly.97 percent of the anomalieshave been correctly distinguised. In addition,a satisfying result is acqtuired by theapplication of BP network to multi-layer densitV interface inversion. Bv means ofadjusting the cell activation ftinction,the interpolating capability and the steadiness of thenetwork are increased.Model and field data are practicfed to test the efficiencv of themethod.