Earticle

In allusion to the low correctness and efficiency of fault diagnosis for the complex industrial system, rough set theory, particle swarm optimization and back propagation (BP) neural network are introduced to propose a hybrid intelligent fault diagnosis(RPBPNN) method in this paper. In the proposed RPBPNN method, rough set theory as a new mathematical tool is used to process inexact and uncertain knowledge in order to obtain the minimum fault characteristic set for simplifying the structure and improving learning efficiency of BPNN. The particle swarm optimization (PSO) algorithm with the global optimization ability is directly used to train the weights of BP neural network in order to establish the optimized BP neural network model. Then the minimum fault characteristic set is used to train the optimized BP neural network model in order to obtain the optimal BP neural network model for realizing the fault diagnosis. Finally, the proposed RPBPNN method is applied to an actual application case for verifying the effectiveness. The experimental results show that PSO algorithm can search for the optimal values of BPNN parameters and the proposed RPBPNN method can accurately eliminate false and improve the diagnostic accuracy. So the proposed RPBPNN method takes on better generalization performance and prediction accuracy in the real industrial application system.

At the present learning through online video is very popular. But there is no way to de-termine whether student is actually watching video or not. In this paper, an algorithm for real time eye state classification using simple web-cam is presented. Here one application is developed in which eyes of the person seated in-front of camera are detected using classifier. Four different eye positions: looking straight, looking left, looking upward and looking right are classified with the help of K-means clustering of the features of detected eyes. Here looking downward is not considered because it seems closed eyes and when closed eyes are detected the video will automatically pause. This approach is also used to detect constant gaze towards screen to prevent Computer Vision Syndrome. Another application of eye detection and eye state classification is to detect driver fatigue during driving. The experimental results prove the effectiveness of the presented meth-ods.Given that two eyes are detected in a face; the system classifies the eye-states with an accuracy of 95%.

A novel neural network algorithm optimized by particle swarm optimization (PSO) for function approximation is proposed in this paper. The prior information extracted from the upper and lower bound of the approximated function is coupled into PSO. Since the prior information narrows the search space and guides the movement direction of the particles, the convergence rate and the approximation accuracy are improved. Experimental results demonstrate that the new algorithm is more effective than traditional methods.

Deadlocks are serious runtime bugs and are difficult to expose, reproduce and diagnose. Once suffering from them, programs may be afflicted with increasing response time, decreasing throughputs, or even crashes. We present Mocklinter, a dynamic deadlock detection tool to capture a deadlock as soon as it happens and spit out enough information to support source-level debugging. Mocklinter tracks the synchronization state of the target program by dynamically constructing and maintaining a lock allocation graph. Mocklinter uses this graph to decide whether a deadlock is confronted or not. Mocklinter handles all types of pthread mutexes and can detect any number of deadlocks at a time. Each deadlock captured by Mocklinter can involve any number of threads. We implemented Mocklinter in Linux-3.2.0 and evaluated it with ten applications, including Dining-Philosophers, Sshfs, SQLite, OpenLDAP, MySQL and so on, whose sizes varies from 0.1K to 1021.0K in terms of LOC. The results demonstrate effectiveness against real or artificial deadlock bugs, while incurring modest performance overhead and scaling to more than one thousand of threads.

The design of interactive systems is a process of transformation from problem domain to implementation domain. A mobile interactive system design differs from traditional software design in the perspective providing to conceptual modeling. With a diversity of user requirements and devices, it more concentrates on knowledge about the context of use rather than static information of problem domain. This paper proposes a multi-dimension description method which can describe those concepts related to special user requirements in the special context from four aspects: the set of static conceptual elements, process set, constraint set and interactive set. Then based on the conceptual model, a mobile interactive system generation framework is proposed to accurately map those concepts into requirement analysis and the implementation process; under the guidance of design pattern, a usable and useful interactive system can be achieved. Finally, we give a prototype of campus navigation system in which we successful used the generation framework.

A method for image compression, adaptive difference compensation vector quantization using dynamic image block adjustment is proposed in this paper. Before image coding, this proposed method analyzes the similarity values between the encoding image sub-block and its 8-neighbor sub-blocks, then determines the encoding rule according to the similarity value and the preset threshold. If the similarity value is approximate to the threshold, we use the same codeword to encode the neighbor block and the encoding image block; otherwise, the neighbor block is separately encode. When encoding, the difference between each image block and its matching codeword is first computed to obtain the difference image, and then the sign bits of the pixel difference is imposed the running length coding on and attached after the codeword index. When decoding, this proposed method restores the compressed image according to the codeword index, performs the running length encoding to deal with decode the attached information, uses the window of 33 size to tackle the adaptive difference compensation and derive the final decoding image. The experiment results reveal that this proposed method can improve the encoding speed and image restoration performance against the normal vector quantization.

Medical care requires intensively Image segmentation which a significant tool used to extract origin of interest from the background. Different segmentation techniques are deployed in medical images leading to an essential developments in both diagnosis and detection process. Such development has managed to assist specialists and doctors specified in the medical care system to diagnose the patients accurately. This study will propose a more developed methodology via incorporating cluster representatives and second derivative filter technique on Carcinoma image. The conventional segmentation algorithm is widely deployed in medical care image, in spite of its advantages; being able to produce a complete division of the image, but it contains major drawbacks represented by the over-segmentations and sensitivity to false edges. This study will expose these major problems of the conventional algorithm and fixing these problems via deploying Cluster Representatives; that uses a set of pixels as representatives that represent the clusters and producing advanced results. The other process deployed is a second derivative filter applied to the segmented image resulting of the demolition of unwanted region from the image. Nevertheless, results of this study has proved that the number of partitions in medical care images are much fewer when comparing partitions produced by deploying the traditional algorithm deployed in medical care images. Such result will assist experts to easily diagnose health problems in a more accurate measure.

The development process of the software system is influenced by multi factors, thus the risk control of software projects is of vital importance. Targeting at the features including multi-factor, multilayer, fuzziness and uncertainty during the process of risk assessment of software projects, this paper analyzed the risk measurement of software projects and proposed a multilevel risk measurement algorithm of software projects based on grey variable weight clustering model. This algorithm firstly analyzed the influencing factors of the risk control of software projects, establishing a multilevel risk measurement system for software projects; next, establishing the whitenization weight functions of different measurement indexes and different measurement grades of software projects risks respectively on the basis of the grey system theory, and under the condition of considering the degree of importance of the indexes of different levels, the corresponding risk measurement algorithm of software projects based on grey variable weight clustering model was established; then the paper presented the detailed execution steps of this algorithm and discussed the detailed process to achieve the algorithm. Finally, the algorithm was verified through the case study.