Spatio-Temporal Broadband Sequential State Filtering For Dynamical System Consisted Of Spiral Line Array, Channel And Target

The ocean environment is a limited waveguide environment-channel. The top boundary is sea surface and the below boundary is sea seabed. The sound field is a three-dimensional field represented by bearing, range and depth three-dimensional cylindrical coordinates. In ocean waveguide, to really implementing passive and active target detection-effectively receive and light, three-dimensional source localization should be done. Sensor array, which is used to receive and transmit signals, is often installed in the ship and hull platform dome, resulting array size is limited.Ocean channel multipath propagation and scattering is with space complexity and time complexity characteristics. Complexity refers to broadband. Temporal complex waveform is broadband waveform, and spatial complex wave-front is broadband wave-front. Time broadband corresponds to multi-frequency, and space broadband corresponds to multiple wave numbers. In addition to the effect of ocean channel, in passive sonar, noise radiated by target is temporal broadband complex waveforms, and if point target is extended to distribution targets, spatial broadband complex wave-front will also happen. In active sonar, transmitting signals are generally temporal broadband complex waveforms, and thus the reflection waveforms are also temporal broadband complex waveforms. Likewise, if the target is extended to distribution targets, echo will generate spatial broadband complex wave-front. To make full use of space and time structure information in sound field, large space-bandwidth product array and large time-bandwidth product signal should be used to do space and time data acquisition.In this paper, space broadband double ellipsoid spiral line array (DSLA) is used to do three-dimensional sampling in the waveguide. DSLA is composed of two spiraling along ellipsoid curve array (SLA), the start of which, stagger a certain interval in height, and stagger nearly 180 degree in bearing, and clockwise spiral downward. This result that every height of the ellipsoid will have array elements in eastern and western hemispheres.In this paper, DSLA three-dimensional sampling is used for three-dimensional source localization. Three-dimensional localization here, refers to three-dimensional joint localization, is not combination of bearing estimation using horizontal array and range-depth estimation using vertical array. If readers are just interested in one-dimensional or two-dimensional position, estimation can be obtained from results of three-dimensional localization by marginalization or taking a slice method.Platform array - channel - target constitute a dynamic system with state regular dynamic evolution, and this knowledge will improve the reliability of localization. In this paper, using DSLA, as time broadband, space broadband and space-time evolution method, do spatio-temporal broadband sequential Bayesian filtering to ensure a reliable three-dimensional localization.The main content of this article is:(a) Horizontal array/circle array bearing estimation and vertical array range-depth estimation.(b) SLA and tilted line array matched field three-dimensional localization.(c) SLA three-dimensional localization using sequential Bayesian state filtering.After theoretical analysis, computer simulation and experimental data processing, the following conclusions are derived:(a) The horizontal array can estimate bearing, with no ability to estimating range and depth, and vertical array can estimate the depth of targets, with no ability to estimate bearing. These space fuzzy characteristics lead to degradation of the detection and localization. The main reason for the weak detection and localization of horizontal array and vertical array is that they are space narrowband array.(b) Though tilted line array and pushpin array are three-dimensional array, with ability to do three-dimensional localization, they are still space narrowband array. Tilt array is a surface array, which is positioned vertically, having no space broadband characteristics. The pushpin array composed of a circle array and a vertical array, is the combination of the two space narrowband array, thus it isn't a space broadband array. In fact, horizontal line array, horizontal surface array, vertical array, tilted line array and pushpin array are all fading form of the DSLA. DSLA is large space-bandwidth product array, with better three-dimensional localization performance than space narrowband arrays.(c) In this paper, the localization framework uses SLA spatio-temporal broadband sequential Bayesian filtering, which uses knowledge about dynamic evolution of target state vector. Laboratory waveguide experiments show that, compared with the Bayesian approach which directly uses target state vector posterior probability density before as a priori of the current moment, this method has better localization performance.(d) Uses Monte Carlo sound propagation and probability density function estimation to do statistical modeling of the sound filed in uncertain environment. Simulation and laboratory waveguide experiments verify the validity of the method.