Applied Mechanics and Materials Vol. 344

Abstract: In this paper, two kinds of time varying parameter identification methods which are called identification method of using broken line to approximate time varying parameter and identification algorithm with auto-regulation forgetting factor are studied. The two methods are used for short-term time-varying system parameter identification in simulations. According to the results of the simulations, the applicable conditions of the two kinds of identification methods are analyzed. The results of the simulations indicate the effectiveness of the two methods. Furthermore whether the two kinds of identification methods are sensitive to noise or not is studied by setting different noise levels in the simulations. Finally, experiments of a variable mass cylindrical shell are adopted to demonstrate the efficiency of the two kinds of time varying parameter identification methods.

Abstract: This paper presents the development of a virtual therapist to guide the patient through the cognitive tele-rehabilitation exercise. The intelligent assistant uses fuzzy logic to regulate the visual and audio cues based on the patients behaviors being monitored. The framework and the structure of the virtual therapist are discussed in this paper. A preliminary clinical trial indicates that the virtual therapist could complement the electronic exercises.

Abstract: Parameters of support vector regression (SVR) have a great impact on complexity, training accuracy and prediction accuracy of the model. To solve the problem that excessive pursuit of training accuracy will decrease the generalization ability of model, a method of parameter optimization based on artificial fish swarm algorithm (AFSA) is put forward. A part of predictive sample is used to calculate error, and build fitness function according to it, which set up the feedback from prediction to training and avoid overfitting. The simulation results show that the method decreases the prediction error of SVR, and improve the generalization ability.

Abstract: Multimedia has been widely used on the mobile platform. Due to its mobility and instability,the mobile terminal inevitably produces some blurred pictures (especially when shooting human faces). Hence, if these blurred and normal images are well classified and separated, it will be significant to improve the browsing efficiency. This paper focuses onresearch of two popular blur detection algorithms, DCT (Discrete Cosine Transform) and edge detection algorithm. It also offers the implementation of the blurred face image detection and classification system based on these two algorithms. At last it contrasts these two algorithms and draws a conclusion.

Abstract: A fast stereo video coding method based-on C-Means Fuzzy Hopfield neural network and AVS (Audio and Video coding Standard) is presented in this paper. In the system, an AVS encoder encodes the left-view sequence. For the right-view sequence, combined predictions from the temporal or disparity domains are exploited, with a hierarchical two-stage neural classifier for fast mode decision. Experiment results show that our method can achieve up to 30% of time savings with nearly ignored quality degradation and only 7% of bit-rate increase.

Abstract: Pattern matching algorithm is a key technology of intrusion defense system in computer network. Based on the pattern matching algorithm AC-BM, This paper proposes an structure of binary tree to express the pattern tree , the structure model of binary tree can reduce the time of forming pattern tree, and then through the using of improved BMHS algorithm of bad character, the algorithm can increase moving distance of pattern tree decrease the matching time, and promote the speed of pattern matching. The experimental data show that, through the improvement of AC-BM algorithm, it indeed can improve the matching speed, achieving good time performance.

Abstract: The Digital Speckle Correlation Method is a non-contact in-plane displacement measurement method based on machine vision. The spline, the linear or the bi-cubic interpolation algorithm is often used in the sub-pixel algorithm. The block effect or the edges blurring, however, could occur in these algorithms. Therefore, a fast interpolation algorithm based on fractal theory is proposed in this paper. The results indicate that this algorithm can overcome either the blocking effect or the edge blurring caused by the traditional interpolation algorithm. In addition, with the same sub-pixel accuracy, the shortest time is required.

Abstract: This paper studies how to use the cognitive model to model the behavioral decision-making for the tactical level simulation. The decision making process includes initial behavior planning and behavior modification. This method has the advantages of accuracy and real-time, this makes the behaviors of the simulation entity more realistic and improves the credibility of simulation.

Abstract: Collaborative engineering design based on distributed networks is an important way to increase the productivity of mechanical product design and to reduce the development cycles. This paper presents a novel method for the implementation of a collaborative engineering design. It is based on the integration of a video conferencing system and a CAD system. Issues about the general requirements and the architecture of the system are discussed. The paper also distinguishes three levels for design conflicts and provides specific solutions to them respectively.

Abstract: CGF architecture model reflects the physical behavior of operational entity Agent and the whole system. As to the CGF system, it is necessary to build not only the physical behavior model, but also intelligent decision model to reflect the subjective awareness behavior of the system. Taking offensive and defensive confrontation CGF decision-making system as the background, we analyze the relevant factors which could affect CGF decision, and build a kind of fuzzy petri net to model the CGF decision-making system. At last, a relevant inference algorithm is proposed.