A method of In-situ stress field's multi-factor regression analysis, based on visual BP neural network and on ActiveX technology, is presented with an application example. The relevant theories of neural network are adopted and permit to calculate the three dimensional stress fields. The method owns the advantages of strong capability in computing of Matlab, friendly VB interface and easy user-machine conversation.
Neural network theory has been used as an effective method in the in-situ stress analysis[1]. But it will inevitably involve massive problems related to the numerical calculation, such as routine matrix calculation, model orthogonalization, least-square processing and etc. In this case, many users will choose the existing simulation software Matlab. Matlab has great advantages in some common complicated matrix calculations and simulation calculation, while it is less able in compiling the graphical user interface (GUI). Although VB is relatively weaker in matrix calculation and numerical analysis, it is flexible and convenient in the programming. Also it is one of the most widely used visual programming tools. In order to take advantages of matlab in calculation and those of VB in GUI, neural network toolbox is adopted as one of its ActiveX components in the multifactor regression analysis method of three dimensional stress field. It is capable of visualizing the BP artificial neural network model in VB.
ActiveX automation is one of ActiveX's protocols, which permits the control of one application program or component to another. ActiveX component supplied by Microsoft Corporation is a new protocol applied to module integration. Meanwhile, it is an extension of the VB toolbox as well as some executable codes written according to the ActiveX criterion, such as a document of.EXE,.DLL or.OCX.
