Temporal And Spatial Alignment And Performance Assessmenta For Multi-Platform Multi-Sensor And Multi-Source Information Fusion System

The multi-platform multi-sensor multi-source information fusion has widely application to the field of national defence. But, the issues of particular concern in multi-platform information fusion, in addition to those involved in single-platform fusion, are: (1) spatial and temporal alignment, (2) combination expulsion of computation, (3) jam-up and saturation in communication, (4) performance assessment for multi-platform multi-sensor and multi-source information fusion system. This dissertation focuses on the temporal and spatial alignment problem of sensors and the performance assessment for multi-platform multi-sensor and multi-source information fusion system. This dissertation has provided some effective methods for solving the two problems and the contributions are summarized as follows:1.Background, meaning, status, and exiting problem for development of multi-platform multi-sensor and multi-source information fusion are summarized systematically. The research idea of temporal and spatial alignment problem of sensors and the performance assessment is proposed for multi-platform multi-sensor and multi-source information fusion system.2. The conception of the platform alignment and system alignment is presented for multi-platform multi-sensor and multi-source information fusion with systematic temporal and spatial alignment based on the different spatial distributions between single-platform and multi-platforms. To do as this, we can deal with different problem with different alignment approaches, and it is propitious to solve the problem.3. Based on the characteristic of the sensor distribution inside a platform, a sensor registration model for single-platform alignment is given. With the Taylor series, we can get the first-order approximation of the registration model. Then, the estimates of the system biases can be obtained by applying Kalman filtering techniques to the approximate model. The simulation results indicate that the satisfied resolution can be obtained by this method to the problem of sensors registration, of which the distance between the sensors is small, that is, sensors inside the same platform.4. Two registration methods for system alignment are presented. The 3-D Exact Maximum Likelihood Registration Algorithm (EML) incorporates the effects of measurement noise. The registration estimates are obtained by the maximum likelihood function of the sensors measurement. The simulation results indicate that the estimates of registration errors have better consistency and stability. The Maximum Likelihood Registration algorithm based on Earth-Centered Earth-Fixed (ECEF) coordinate system considers the geometry of the global, and eliminates errors introduced by the stereographic projection. The simulation result based on thisregistration algorithm shows that the satisfied angular biases can be preliminarily obtained.5. A Registration Algorithm for Sensor Alignment Based on Stochastic Fuzzy Neural Network (SNFF) is presented. In stochastic fuzzy neural network based registration system, the network can be taught to correct the measurement so as to trace the measurement coming from another sensor (reference sensor). The main advantage of the solution is that the same network can study the different types of systemic errors by training. The simulation result indicates that this method is applicable to all kinds of biases and errors. But it needs much time than others; and as long as the environmental conditions are changed, the network must be trained again.6. A multi-platform multi-sensor time aligning algorithm ?multi-rate estimate algorithm based on Maximum Entropy Inference Engine is presented. The merit of this algorithm is that higher rate estimated signals can be obtained with different lower rate signals. It indirectly reduces the communication rate between sensors. The simulation result indicates that the satisfied temporal alignment result can be obtained.7. The complexity problem of the multi-platform multi-sensor and multi-s...