An appropriate analysis of pore pressure models under irregular loading effects is provided. the model is based on liquefaction test with the constant amplitude loading and pore pressure model, based on the accumulated principle of residual pore pressure, using the differential method: superposition method and optimization method. Using constant amplitude loading liquefied test, and artificial modulation irregular load liquefied test and earthquake wave input dynamic triaxial liquefied test on model, the system has been tested, the result displayed: (1) the model can better simulate the results of constant amplitude loading liquefied test; (2) the model can better describe the trend on artificial modulation irregular load liquefied test results (3) the model can describe pore-pressure growth law under earthquake wave, and also can better analysis on differential pore-pressure under different type of earthquake wave (4) the model has less parameters, can analysis of pore-pressure in real time, easy to use.

As to the fact that the common model of repair does not suit for the immemorial exponential unit, it first proposes the general repair model based on the Repair Degree. This model can describe the repair effect of the exponential unit exactly. Then it studies the classical estimation method of the parameters for the Repair Degree as well as the Failure Rate in the condition of general repair, that’s Moment Estimation and Maximum Likelihood Estimation. On this foundation, it compares the two methods by large amount of simulative data. Further, it figures out the estimation value of the Failure Rate on the assumption of ‘As Good As new after repair’. There exists apparent difference from the exact value. So it shows that the assumption of ‘As Good As new after repair’ is not appropriate.

The traditional model for average support probability often assumes that component life is exponentially distributed. But the different types of components in electromechanical integration system may follow different non-exponential distributions. In order to solve the problem, we formulate the random event of support failure as the event of system shutdown with service parts in short supply. We show system shutdown can be formulated as a Markov process and we develop a practical method for solving the problem of how converting non-exponential distribution to exponential. In the base of above, we develop an average support probability model for series system with arbitrary-life-distribution units. By simulating the operation of a non-exponential distribution system in Extendsim software, we compare the result with our model and prove the correctness of it.

As psychology, education measure theory and with the combination of computer technology of continuous research and development, based on Computer Adaptive Test (CAT) become a new type of test form. Based on the analysis and state the topic of CAT theory and Maximum Information Selection Strategy (MISS), the method of maximum information selection strategy in the key issues in an adaptive testing system solutions is discussed. The effectiveness of the solution method which this topic given is proved by experiment.