To resolve the problems of short working life and narrow operating voltage range of the spark gap with one-offin gas filling structure, a spark gap with repeatable gas filling structure was designed. The spark gap was composed of three electrodes and two gaps as well as gas filling devices. Electrodes structures adopted the form of field aberration three-electrode gas spark switch forming two discharge gaps between three electrodes of anode, cathode and triggering electrode. Gas filling devices were composed of fast connector, gas pipe, barometer and air valve, by which the gaps could be repeatedly charged and deflated from a hole at the center of anode. Working life is lengthened by exchanging gas, and operating voltage range is widened by changing gas pressure in a certain range. The spark gap was manufactured and applied successfully in a TEA CO2 laser with pulse energy 170J. The results prove the designs feasible.

To resolve the problem of data transmitting in multi-sensor network possessing of a lot of sensors and processing sensors signals on the same time base and transmitting the results remotely, a three classes’ bus topology structure based on intelligent node was designed. The system comprised scene devices, transmission cable, and console. The scene devices included several gateways, more nodes and sensor groups. The console communicated remotely with gateways via Ethernet, and gateway communicated with nodes via CAN bus, each node managed a group of sensors via RS-485 bus. The designs of intelligent node were presented in detail including hardware design, transport protocols, as well as program flow. Above conceptual design was applied successfully in a certain multi-sensor network system principle prototype consisting of a console, a gateway and three nodes possessing of ten sensors. The results prove the conceptual design feasible.

To detect week signal of underwater magnetic fields, we designed a negentropy maximization blind source separation algorithm based on the stochastic gradient descent algorithm by bring in a penalty factor. The simulation study demonstrates that the new algorithm is effective to separate the mixed-signal with high precision, to decrease the iterative calculation and to enhance the convergence rate.

To resolve the problem of acquiring multi sensors signals synchronously in a distributed system, a project of distributed synchronous data acquisition based network time server was designed. The system comprised multi distributed sensor units, a centralized control unit and D&C (display and control) center. Each sensor unit was equipped with a data acquisition module. All data from sensor units were concentrated and sent to the D&C center, and the D&C center implemented power supply and management of sensor units via the centralized control unit. To synchronize the data acquisition modules, a network time server was employed in the D&C center. It received standard time information from GPS and outputted the time signal with DCF77-encoding. Each data acquisition module received and decoded DCF77 time signal, obtaining absolute time and synchronizing its time base. Above project was applied successfully in a system comprising 20 distributed sensor units. The results prove the designs feasible.