Date of Award
Chemical and Biomedical Engineering
Filters and filtration, Kalman filtering, Estimation theory
In reality many processes are nonlinear and in order to have a knowledge about the true process conditions, it is important to make decisions based on the state of the system. Process measurements such as pressure, temperature, and pH, are available at time instances and this information is necessary in order to obtain the state of the system. Filtering is a state estimation technique by which the estimate is obtained at a time instant, given the process measurements at their respective time instances. Several lters have been developed so far for the estimation of the states of the system. Kalman lters are the optimal lter algorithms used for linear state and measurement models. Approximations are made to this algorithm in order to apply to non-linear systems. Particle lter (PF) is one such approximation made to the Kalman ltering technique. It involves drawing a set of samples or particles from the state of the system. It works on the principle of importance sampling, where, the samples are derived from a probability density function which is similar to the state model. The particles are resampled according to their weights in order to determine the estimate. Taking into account the diculties in particle ltering technique, a nested particles lter (NPF) was developed. NPF works in such a way that there is a set of particles under each sample of the original particle lter, and from these nest of samples the transition prior is updated using an extended Kalman particle lter (EKPF). The idea of nested particle lter was developed from the unscented particles lter (UPF), which uses the concept of local linearization to develop the importance density. Better importance densities are formulated in this case through which better posteriori are obtained. It is important to note that the update of the NPF can be done with any suboptimal nonlinear lter available. This thesis work is based on developing the NPF with a direct sampling particle lter (DSPF) update. Some modications are made to the unscented particle
Srinivasan, Swathi, "State Estimation Based on Nested Particle Filters" (2013). ETD Archive. 763.