Ye Qingwei

The duplicated of partial content detection based on PSO

It is discussed how to detect the duplicated of partial contents between two documents in this paper. There are some algorithms which can detect similarity among documents. But these algorithms cannot detect the duplicated of partial contents in documents. A new effective algorithm of the duplicated of partial contents detection in documents is put forward in this paper. The new algorithm is using PSO algorithm to search the optimized partial contents which is the most similar in two documents. For PSO algorithm, it provides the encoding of the particles. A new related coefficient of strings is defined for strings similarity. And the new evaluation function of PSO is designed based on the related coefficient function. The hybrid mutation PSO algorithm is used for searching the most similar partial contents quickly and accurately. The simulation experiments indicate that the algorithm can search the most similar partial contents in two documents effectively.

Research and application of ultrasonic signal detection based-on wavelet-time entropy

Ultrasonic signal detection is one kind of the non-destructive testing methods, and it can be non-destructive inspection of the internal defects of materials or mechanical components. One of the difficulties of ultrasonic signal detection is the detection of the echo signal in the ultrasonic signal. With the wavelet transform and Shannon entropy theory, the wavelet-time entropy algorithm is introduced in this paper. And the numerical simulations are carried out to prove the feasibility of this algorithm in abnormal signal detection. An improved wavelet-time entropy algorithm is put forward in this paper. The high-frequency coefficients of wavelet transform are replaced by the low-frequency coefficients, and the optimal width of window is worked out for actual ultrasonic signal. The results show that this improved wavelet-time entropy algorithm can effectively detect the echo signal in ultrasonic signal.

Spectral clustering algorithm based on K-nearest neighbor measure

Analysing the defect on different similarity matrix in spectral clustering, we propose a new algorithm—Spectral clustering algorithm based on K-nearest neighbor measure. The K-nearest neighbor measure focuses on using data points between the common number of nearest neighbors to measure the degree of similarity, and avoids the degree of similarity is large and unstable by contrast. The experiment results show the efficiency and performance of the algorithm. Meanwhile, the algorithm effectively resolves the problem that two data points belonging to different clusters are close. It possesses the advantage of discriminating the clusters with variable density, and also has the advantage of clustering in a sample space of any shape.

A mutation hybrid algorithm of particle swarm optimization

Many mutation functions of particle positions are discussed in this paper. It is discussed the global searching capability, the local searching capability and the traversal capability of these mutation functions. And a mutation hybrid algorithm of particle swarm optimization is brought out in this paper. In each iteration, the position of particles which is satisfied the mutation condition are mutated with many kind of mutation functions, and each mutation function is endowed a probability. The probability distribution is relied on the specific optimization problem. Some experiments results of standard PSO, single mutation PSO and multi-mutation PSO are contrasted. It indicates that if the probability distribution of mutation functions set is well formed, the multi-mutation PSO will search the global best value more quickly and effectively.

The optimized estimation of impulse response function using PSO

A new estimation algorithm of impulse response function is put forward using PSO in this paper. Takes all points of impulse response function as one particle, and searches the best particle with PSO iterations. The evaluation function of PSO iterations is designed to satisfy the convolution equation of the linear vibration system, which requires the minimum sum of errors between the best impulse response function and the response signal with the impulse signal. And the smoothness of impulse response function and the consistency of frequency curves are considered in the design of the PSO evaluation function. The multi-mutation PSO iterations are used, which can avoid falling into the local optimal result. The simulating experiments indicate that the optimized estimation algorithm of impulse response function can work more stably and accurately than the related coefficient algorithm of impulse response function.

Loosen Detection of Screw Thread Based on Time Frequency Entropy

Several kinds of time-frequency transforms of vibration signals are compared to retrieve the features of loosening detection of screw thread in this chapter. The pseudo-Wigner-Ville time-frequency distribution is selected because its mutual coherence is the least. The entropy features of the time and frequency projected curves of the pseudo-Wigner-Ville distribution image are defined in this chapter. The relation between the entropy feature of the time-projected curve and the screw thread loosening moment is descending, and the relation between the entropy feature of frequency projected curve and the screw thread loosening moment is ascending. When applied to the loosening detection experiments, these relations are used to detect the loosening of screw thread very well.

Redundancy insertion of frame-based multiple description video codec

In video stream transmission over unreliable networks, the loss of a packet may result in an entire video frame, which may consequently cause severe degradation of all subsequent frames. To solve the problem, based on prediction, presented in this paper is a new pre- and post processing method with flexible redundancy insertion between descriptions for Frame-based multiple description code. The simulation shows that the presented method demonstrates its adequate efficiency in preventing error propagation. Moreover, the system is more robust in the case of packet-loss over different networks.