This paper mainly introduced the basic theory of Hidden Markov Model (HMM) and Support Vector Machines (SVM). HMM has strong capability of handling dynamic process of time series and the timing pattern classification, particularly for the analysis of non-stationary, poor reproducibility signals. It has good ability to learn and re-learn and high adaptability. SVM has strong generalization ability of small samples, which is suitable for handling classification problems, to a greater extent, reflecting the differences between categories. Based on the advantages and disadvantages between the two models, this paper presented a hybrid model of HMM-SVM. Experiments showed that the HMM-SVM model was more effective and more accurate than the two single separate models. The paper also explored the application of its database system development, which could help the managers to get and handle the data quickly and effectively.

This thesis mainly elaborated the PID neural network feed-forward algo-rithm and back propagation algorithm and the structure form of its controller, then make use of MATLAB to simulate the liquid level adjusting system, analysis its control perform-ance and choose appropriate neural network parameters, and compared with the traditional PID control effect, analyzes the advantages of PID neural network. Through the comparison with the conventional PID control, PID neural network is superior to the traditional PID. The traditional PID control tuning parameters has a large number of thumb rules for reference, but the setting out of the parameters is not necessarily good. And sometimes we have to modify the parameters if we wound the better control effect. PID neural network is set up as long as the learning step in accordance with the PID rule set. this paper has show that Liquid Level Control System based on Computer Nerve Network has good control effect of rapid and effective.

The ability to maintain events (i.e. interactions between/among objects) in working memory is crucial for our everyday cognition, yet the format of this representation is poorly understood. The current ERP study was designed to answer two questions: How is maintaining events (e.g., the tiger hit the lion) neurally different from maintaining item coordinations (e.g., the tiger and the lion)? That is, how is the event relation (present in events but not coordinations) represented? And how is the agent, or initiator of the event encoded differently from the patient, or receiver of the event during maintenance? We used a novel picture-sentence match-across-delay approach in which the working memory representation was "pinged" during the delay, in two ERP experiments with Chinese and English materials. First, we found that maintenance of events elicited a long-lasting late sustained difference in posterior-occipital electrodes relative to non-events. This effect resembled the negative slow wave reported in previous studies of working memory, suggesting that the maintenance of events in working memory may impose a higher cost compared to coordinations. Second, in order to elicit a hallmark for agent vs. patient representation in working memory, we pinged agent or patient characters during the delay. Although planned comparisons did not reveal significant differences in the ERPs elicited by the agent pings vs. patient pings, we found that the ping appeared to dampen the ongoing sustained difference, suggesting a shift from sustained activity to activity silent mechanisms. These results represent one of the uses of ERPs to elucidates the format of neural representation for events in working memory.

No abstract