Multiple-model Adaptive Estimation Approach Research Articles

Cooperative Guidance Law for Target Pair to Lure Two Pursuers into Collision

A novel cooperative defensive guidance law is presented for a two-on-two engagement. Instead of classical strategies in which evasive maneuvers are performed or additional agents (e.g., defender missiles) are deployed, the target pair lures the pursuing missiles into collision. The optimal cooperative strategy is solved using state-dependent Riccati equation method. Linearized kinematics and arbitrary-order linear adversaries’ dynamics are assumed in the guidance law derivation. Imperfect information is assumed on the relative states and the guidance laws employed by the missile pair. Guidance strategies that the pursuing missiles may employ are assumed to belong to a finite set of linear guidance laws. In addition to the proposed cooperative defensive strategy, a decentralized estimation scheme based on the multiple-model adaptive estimation approach is also presented. Guidance law and estimation performance are demonstrated using nonlinear simulations. Simulation results show the viability of the proposed guidance law and highlight the sensitivity of the guidance law to range measurement accuracy.

Selective-Reinitialization Multiple-Model Adaptive Estimation for Fault Detection and Diagnosis

The existing multiple-model adaptive estimation approach is able to detect faults quickly. However, there are three main problems when it is used for fault detection and diagnosis: false alarms, requirement of designing additional models to identify the faults, and slow response to detect the removal of the faults. In this paper, a novel selective-reinitialization multiple-model adaptive estimation approach is proposed. This approach introduces a state augmentation strategy that can identify the faults without designing additional models, as well as reduce false alarms. The major contribution of this approach is that three selective-reinitialization algorithms are proposed that can improve the performance of the multiple-model adaptive estimation significantly. The selective-reinitialization multiple-model adaptive estimation approach eliminates false alarms and is quick to detect the removal of the faults. The performance of the proposed approach is compared with the multiple-model adaptive estimation and the interacting multiple model with an example of the fault diagnosis of the inertial measurement unit and air data sensors for a Cessna Citation II aircraft. The simulation results suggest that the selective-reinitialization multiple-model adaptive estimation outperforms the multiple-model adaptive estimation and interacting multiple model in effectiveness and efficiency.

Space Object Shape Characterization and Tracking Using Light Curve and Angles Data

This paper presents a new method, based on a multiple-model adaptive estimation approach, to determine the most probable shape of a resident space object among a number of candidate shape models while simultaneously recovering the observed resident space object’s inertial orientation and trajectory. Multiple-model adaptive estimation uses a parallel bank of filters, each operating under a different hypothesis to determine an estimate of the physical system under consideration. In this work, the shape model of the resident space object constitutes the hypothesis. Estimates of the likelihood of each hypothesis, given the available measurements, are provided from the multiple-model adaptive estimation approach. The multiple-model adaptive estimation state estimates are determined using a weighted average of the individual filter estimates, whereas the shape estimate is selected as the shape model with the highest likelihood. Each filter employs the unscented estimation approach, reducing passively collected electro-optical data to infer the unknown state vector composed of the resident space object’s inertial-to-body orientation, position, and respective temporal rates. Each hypothesized shape model results in a different observed optical cross-sectional area. The effects of solar radiation pressure may be recovered from accurate angles data alone, if the collected measurements span a sufficiently long period of time, so as to make the nonconservative mismodeling effects noticeable. However, for relatively short arcs of data, this effect is weak, and thus the temporal brightness of the resident space object can be used in conjunction with the angles data to exploit the fused sensitivity to both the resident space object shape model and associated trajectory. Initial simulation results show that the resident space object model and states can be recovered accurately with the proposed approach.

Observer Kalman filter identification and multiple-model adaptive estimation technique for classifying animal behaviour using wireless sensor networks

The identification of a mathematical model capable of describing the behaviour of animals given input such as feed has great potential for behavioural control purposes. Such models will allow to make predictions which are fundamental to any closed loop control such as control of the feeding. This paper investigates the problem of mathematically modelling animal behaviour. An observer Kalman filter identification method was successfully applied to input–output data and two models representing the hypotheses that animals are actively feeding and the hypotheses that animals are inactive were identified. The input and output of each of the identified models were feed dry matter offer and the pitch angle of the neck, respectively. The pitch angle of the neck of the animal was successfully measured and aggregated by a ZigBee-based wireless sensor network. Two fourth-order models describing the dynamics of an animal in the active and inactive behaviour modes showed good performance in terms of prediction error, cross-correlation between the residual and the output as well as cross-correlation between the residual and the input with 99% confidence interval. A multiple-model adaptive estimation approach was applied to determine the likelihood of each of the two models being the correct model for a specific input of dry matter feed. The average classification success rate was 87.2% for the whole experiment.