Application Research Of Networked Sensor And Multi-sensor Data Fusion Technology

With the rapid development of computer technology and network communication technology, the concept of instruments in the 21st century will be an open system concept, and networked sensors have become the trend of sensor technology. However, in the Measurement and Control System fields there are tens and thousands of sensors/actuators, the interface standards of which are different from each other, so the complexity and the cost of the Measurement and Control network are increased and the ability of data transmission and processing of the sensors is limited. In order to solve this problem a method for constructing networked sensors system based on embedded system is given in this paper. Data of the mine environment parameters was processed by the technology of Multi-sensor Data Fusion, then the safety condition of the environment was estimated and the anticipated results were obtained.The control core of this system is the MCU P89C669 from Philips. According to the function requirements of the networked sensors, control nodes based on the combination of 1-wire bus and CAN bus were designed in the thought of modularization. The nodes have the functions of data acquisition, data transmission and Human-Computer Interaction, involving the interfaces of 1-wire bus, RS232, RS485 and CAN bus etc., so multi-construction of network can be implemented easily. In the design of the hardware of this system the resources of the MCU was used fully, and the circuit is simple and has a good stability. The environment in which the networked sensors are used was considered and a simple, effective communication protocol was designed according to relevant standards.The characters of the detection of the mine environment were analyzed, and the self-adaptive weighed algorithm and Neural Networks were studied. The self-adaptive weighed algorithm is simple and intuitionist and the Neural Networks algorithm has the features of fault tolerance, self-study and self-organization. A two-step Data Fusion method based on self-adaptive weighed algorithm and BP Neural Networks was proposed and the stimulation work was done. The results prove that this data fusion method is effective, feasible and can supply more comprehensive and more accurate mine environment information.