Centralized Fusion Method of Single Out-of-Sequence Measurements for Arbitrary Delay Steps in Nonlinear Systems

Abstract

The invention discloses a single out-of-sequence measurement centralized fusion method for any delay steps of a nonlinear system, which adopts the generalized Gaussian smoothing method to solve the smooth probability density function of any nonlinear system, and uses the particle filter algorithm to realize the update. Whether the measurement is a sequential measurement, if so, use the particle filter algorithm to obtain the sequential posterior state, otherwise perform out-of-order measurement fusion, that is, first use the generalized Gaussian smoothing method to solve the smoothed probability density function at the time when the out-of-order measurement is generated, and then jointly smooth the probability The density function and the partial likelihood function are used to solve the asynchronous likelihood function, and finally the sequential posterior state of the previous moment is updated using the asynchronous likelihood function to obtain the fusion posterior state, and the fusion posterior state represented by the particle sample is approximated as a Gaussian parameter The storage effectively solves the problem of difficult fusion of out-of-sequence measurement data caused by data processing and communication delays in asynchronous sensor networks.

Description

technical field [0001] The invention belongs to the technical field of multi-sensor data fusion, and in particular relates to a single out-of-sequence measurement centralized fusion method for arbitrary delay steps of a nonlinear system. Background technique [0002] Using multiple sensors to work together can improve system reliability and robustness, expand time and space coverage, and improve target detection performance. With the rapid development of information processing and network communication technologies, multi-sensor data fusion technology has received extensive attention . Among them, in the centralized data fusion method, each local sensor sends local measurements to the fusion center for processing, there is no information loss, and the optimal global estimation value can be obtained, so in the area monitoring, target tracking, target detection and positioning It has been widely used in many fields. [0003] In the centralized fusion system, due to the diffe...

AI Technical Summary

Problems solved by technology

[0003] In the centralized fusion system, due to the different data processing time and communication transmission time of each sensor, there is a random delay in the local measurement of each sensor arriving at the fusion center, resulting in out-of-sequence measurements from the same target, that is, out-of-sequence measurement problems

A common solution is to ignore out-of-sequence measurements, but this method usually degrades tracking performance, especially when out-of-sequence measurements occur frequently or come from local sensors with higher accuracy, directly ignoring out-of-sequence measurements will seriously affect Fusion accuracy

The literature "An optimal multi-delay out-of-order measurement processing algorithm, Systems Engineering and Electronic Technology, 2010, 31(11): 2592-2596" is based on the idea of ​​optimal linear fixed-point smoothing, directly using out-of-order measurement to update the current time State estimation effectively solves the problem that multi-step delayed out-of-sequence measurements are difficult to update and fuse, but this method is based on the Kalman filter algorithm and is only suitable for dynamic systems with weak linear or nonlinear strength; the literature "Out-of-sequencemeasurement processing for particle filter: exact bayesian solution, IEEETrans.Aersp.Electron.Syst., vol.48, no.4, pp.2818-2831, 2012" proposed a particle filter algorithm based on deterministic Bayesian method (A -PF algorithm), this method is suitable for general nonlinear systems, and when the number of particles is large enough, this method is approximately optimal and has high tracking accuracy. Large, storage requirements are too high, only suitable for low-dimensional or non-real-time scenarios

Images