| As the structure of energy being adjusted this years, the coal-fired power plants still in an important position, though there are many new ways appeared. It will threaten it,s safty operating for it’s safety loophole in later use process coused by settlement deformation of construct for The nonuniformity of the foundation soil, underground water level changes and the interaction of inner stress. So it is important to pay attention to the problem of foundation settlement of the structure and there is no time to delay.There are many kinds of methods to forecast the foundation settlement and each method has its own advantages, disadvantages and applicable conditions. It is time consuming for the low automation in the comprehensive comparison to select the suitable method. This paper mainly adopts eight kinds of settlement prediction method that based on the measured data. They are hyperbolic method, exponential curve method,three point method, Weibull model, Gompertz model, Logistic model, Usher model and MMF model. This paper uses MATLAB graphical user interface(GUI), combined with the powerful calculation function of MATLAB, to combinating design the selection of 8kinds methods in subsidence prediction. To improve the degree of automation of the foundation settlement prediction and make it convenient to choose the settlement predicting methed, the model can finish data fitting and parameter counting in one platform. And it make it easy to get the parameters. To make it convenient to decide the group of starting point in the hyperbola and exponential curve method which make quite different final result by different starting point, this paper will use curve fitting toolbox cftool in MATLAB to find it which can make it approach to reality. In this paper, we will test the system by the measured data of a power plant.The analysis results showed that the settlement prediction model which is based on MATLAB has good application prospect for it’s advantages, such as, friendly interface,easy operation, easy modification, good real-time performance. |