A New Sensitivity Test Method Based On The Langlie Method And The Neyer_D Method

Sensitivity tests are often used to estimate the parameters associated with latent continuous variables that cannot be measured.For example,it can be used to estimate the parameter of threshold in Military Industries,Biopharmaceutical Industry, Materials Engineering and so on.Nowadays,the Bruceton Method,the Langlie method and the Neyer_D optimal method are the most widely used sensitivity test methods.The Langlie method can obtain an overlap stable and quickly, but it finds follow-up test points blindly which leads to a waste of sample size and less efficient estimate of parameters;The Neyer_D method consists of two parts:an initial test and a D-optimal point selection procedure. The initial test is used to obtain an overlap, and the D-optimal procedure searchs follow-up stimulus levers by maximizing the Fisher information. Unlike the Langlie method, the Neyer_D method has a clear goal when searching follow-up point, but its performance is affected byσguess obviously, including the speed of obtaining overlap and precision of parameters estimation in the Neyer_D method.In order to keep the advantage of the Langlie method and the Neyer_D method while remove their defects, we propose a new sensitivity test method named Langlie_D optimal method.The new method conducts the Langlie test until an overlap comes up, then continues the test using the strategy of the Neyer_D test. Obviously the proposed test inherits the advantage of both the Langlie and the Neyer_D tests. From simulation results, in the perspective of parameter estimation accuracy, the new method performs better than the Langlie method consistently, equivalently to the best situation of the Neyer_D method. Finally, we summarize the new method's detailed test procedure.