Research Of Wireless Sensor Network Performance Testing And Intelligent Fault Diagnosis Technology

Wireless Sensor Networks(WSN) is a key technique which is great potential and influence in the world. The WSN integrates wireless communications technique, sensor technique, micro-electromechanical system technique and distributed information processing technique, will combine the logical information world and the physical world together. It changes the way of human interacts with nature, has a high application value in the civil and military and many other areas.Because of the special application background of wireless sensor networks and its own characteristics and other factors, making the maintenance of WSN is very difficult even unmaintainable. This requires the status of WSN nodes achieve the highest possible network performance optimization before deploying the node to the application environment, then the nodes need to be tested, analyze the reliability of the network, through the analysis of the results of the test or assessment to modify certain parameters or network nodes to achieve optimal performance. Therefore, the study of the work of testing and performance of wireless sensor network nodes and network analysis has important implications for a wide range of scientific applications of wireless sensor networks. In addition, as a distributed computing platform, WSN node easily affected by various disturbances to failure. When a node or a region critical node failure that will lead to the nodes within a certain range around it all does not work, it will result in connectivity and network coverage was cut loopholes. This will greatly reduce the reliability of the sensor nodes, weaken WSN intended function. In order to avoid failure by node level rise to network level, we must effectively detect node failures, and to diagnose the cause of failure prediction, then given rehabilitation program in order to effectively ensure reliable operation of the network. So, in a timely manner on the network to make the diagnosis of various abnormal conditions, find a reasonable tolerant control scheme, to improve the reliability and robustness of WSN has important significance.In order to prolong the lifetime of WSN effectively, we can make full use of the resource, carry out the timely, accurate fault diagnose on WSN. It can ensure the reliability of information transmission, and can carry out the effective routing plan and node management by the superior computer or the centre node, long-distance maintain WSN and the nodes.This paper focuses on the performance testing and intelligent fault diagnosis technology of wireless sensor networks in depth research, on the basis of the relevant research results at domestic and abroad, we presents several new and effective solutions around performance testing and intelligent fault diagnosis which two important technologies ensure high reliability for WSN. The main work of this paper as follows:In WSN performance testing and reliability analysis, this paper presents FIPES, a new fault injection and reliability evaluation method for wireless sensor networks. Firstly, we introduce the basic structure of FIPES; then from the injection fault, failure analysis and reliability assessment describes this WSN performance test methods; finally we had done many tests for FIPES. Experiments show that, FIPES can be effectively injected to WSN various failures and assess its performance, and it has a high value in terms of reliability testing WSN, and injected a new method for WSN testing and reliability verification study.In WSN intelligent fault diagnosis, this paper, we analyze WSN fault hierarchy and characteristics of fault-level of WSN, get the strategies of fault detection and fault diagnosis. And effectively make use of Rough Sets Theory advantage in the removed redundant attribute, fully integrated to improve the RBF neural network in parallel computing and overcome noise interference function, we proposes an algorithm that integrates the Rough Sets Theory and improved RBF algorithm (RSRBF), to resolve WSN node online fault diagnosis problem which is limited energy and significant uncertainty at the same time. The specific method is that:at first, we make use of the Rough Sets theory to reduction fault diagnosis decision table, and then use the data reduced to training neural networks, finally, use the improved RBF neural network to diagnose fault of the WSN nodes. In the reduction algorithm based on Rough Sets Theory, we proposed an improved attribute reduction algorithm, improve the computational efficiency in this paper.This paper proposed the RSRBF integration fault diagnosis algorithm has reveals the WSN node fault characteristic information intrinsic redundancy, can solve the WSN node online fault diagnosis problem fast accurately. In the situation that we obtain the incomplete information or the partial information contains errors, can also give the WSN node a reasonable fault diagnosis. Compared with diagnosis rules based on traditional RBF neural network, when the reliability of fault feature data is lower, the more advantages the algorithm accuracy in the diagnosis this paper mentioned. Through many simulation experiments, it shows that the algorithm proposed in this paper has a high rate of diagnostic accuracy, low communication cost and low power consumption; it improves the robustness of fault diagnosis, and enhances practicality of WSN which has limited energy. These research results are important for improving our research level and development of wireless sensor networks and will promote large-scale wireless sensor network applications.