The feedback path (FBP) compensation has been a research topic in the field of active noise control (ANC) for decades and a number of offline and online FBP modeling techniques have been proposed in the literature. In...
详细信息
The feedback path (FBP) compensation has been a research topic in the field of active noise control (ANC) for decades and a number of offline and online FBP modeling techniques have been proposed in the literature. In this paper, a novel ANC syst.m equipped with an online FBP modeling (FBPM) and neutralization (FBPMN) subsyst.m is proposed. The new syst.m consists of three adaptive subsyst.ms, namely 1) ANC controller with only two weights for each frequency target, 2) second-order adaptive IIR notch filter, and 3) an FBPMN filter. It yields very promising performance for narrowband noise signals generated by rotating machines such as fans, motors, large-scale strandcutters, etc. An auxiliary white Gaussian noise is injected into the ANC syst.m to implement the FBPMN, which is scaled by a function of the residual noise to reduce its cont.ibution to the residual noise power. Extensive simulations with synthetic/real secondary paths reveal that the proposed ANC scheme ensures excellent noise reduction performance as well as practically attractive computational efficiency.
Supervisory control and data acquisition (SCADA) syst.ms that run our critical infrastructure are increasingly run with Internet-based protocols and devices for remote monitoring. The embedded nature of the components...
详细信息
Identification of certain syst.m needs search for class of functions or model forapproximation of the syst.m input-output behavior in the "best" possible way. In most situations, as in dynamic syst.ms identi...
详细信息
Identification of certain syst.m needs search for class of functions or model forapproximation of the syst.m input-output behavior in the "best" possible way. In most situations, as in dynamic syst.ms identification, pattern recognition etc., the value of the output of the mode ling syst.m is function of the previous values of outputs and inputs. In the current circumstances almost all efforts for using neural networks for identification and control are based on feed-forward networks. If the syst.m order or the upper limit of the order is known, all the necessary previous values of the inputs and outputs of the mode led syst.m can be put as input in the network. The network can learn non-memory transformations that include thesyst.m output dependence on specified previous inputs and outputs.
In this paper a new observation model is presented to improve the state estimation and prediction in a target tracking problem. Comparing with conventional approaches, the following are distinguished points of the app...
详细信息
In this paper a new observation model is presented to improve the state estimation and prediction in a target tracking problem. Comparing with conventional approaches, the following are distinguished points of the approach. First, the measurement equation is set up in the polar coordinate and even combines the derivation measurement with the usual position measurements (i.e. there are 6 sensor data now: range, azimuth, elevation angle, range rate, azimuth rate, and elevation rate). Second, the observation noise of sensor data is considered as a colored one and be set up as the model of AR(1), and by means of a pseudo measurement model, the requirement of Kalman filter will be satisfied. As a result, the accuracy of both the observation and the prediction will be increased.
暂无评论