arXiv subjects

Taku Nonomura

Publications and source records attributed to Taku Nonomura.

24 records · Page 2Linked to original sources

arXiv2021

Randomized Subspace Newton Convex Method Applied to Data-Driven Sensor Selection Problem

The randomized subspace Newton convex methods for the sensor selection problem are proposed. The randomized subspace Newton algorithm is straightforwardly applied to the convex formulation, and the customized method in which the part of the update variables are selected to be the present best sensor candidates is also considered. In the converged solution, almost the same results are obtained by original and randomized-subspace-Newton convex methods. As expected, the randomized-subspace-Newton methods require more computational steps while they reduce the total amount of the computational time because the computational time for one step is significantly reduced by the cubic of the ratio of numbers of randomly updating variables to all the variables. The customized method shows superior performance to the straightforward implementation in terms of the quality of sensors and the computational time.

eess.SY↗

arXiv2020

Effect of Oxygen Mole Fraction on Static Properties of Pressure-Sensitive Paint

The effects of oxygen mole fraction on the static properties of pressure-sensitive paint (PSP) were investigated. Sample coupon tests using a calibration chamber were conducted for polymer-based PSP (PHFIPM-PSP), polymer/ceramic PSP (PC-PSP), and anodized-aluminium PSP (AA-PSP). The oxygen mole fraction was set to be between 0.1-100% and the ambient pressure was set to be between 0.5-140 kPa. The localized Stern-Volmer coefficient $B_{\rm local}$ once increases and then decreases as the oxygen mole fraction increases. The value of $B_{\rm local}$ depends on both ambient pressure and oxygen mole fraction, but the effect of this parameter can be characterized as a function of the partial pressure of oxygen. The value of $B_{\rm local}$ of AA-PSP and PHFIPM-PSP, which are low-pressure type and relatively low-pressure type PSP, have a peak at the relatively low partial pressure of oxygen, and $B_{\rm local}$ of PC-PSP, which are atmospheric pressure type PSP, has a peak at the relatively high partial pressure of oxygen. The peak of the intensity change with respect to pressure fluctuation proportional to the ambient pressure $S_{\mathcal{PR}}$ appears at the lower partial pressure of oxygen than that of $B_{\rm local}$. This is because the intensity of PSP becomes quite low at the high partial pressure of oxygen even if $B_{\rm local}$ is higher. Hence, an optimal partial oxygen mole fraction exists depending on the type of PSP and ambient pressure range of the experiment, and its optimal value can be found based on the partial pressure of oxygen.

physics.chem-ph↗

arXiv2020

Data-driven sparse sensor placement based on A-optimal design of experiment with ADMM

The present study proposes a sensor selection method based on the proximal splitting algorithm and the A-optimal design of experiment using the alternating direction method of multipliers (ADMM) algorithm. The performance of the proposed method was evaluated with a random sensor problem and compared with the previously proposed methods such as the greedy method and the convex relaxation. The performance of the proposed method is better than an existing method in terms of the A-optimality criterion. In addition, the proposed method requires longer computational time than the greedy method but it is quite shorter than the convex relaxation in large-scale problems. The proposed method was applied to the data-driven sparse-sensor-selection problem. A data set adopted is the NOAA OISST V2 mean sea surface temperature set. At the number of sensors larger than that of the latent state variables, the proposed method showed similar and better performances compared with previously proposed methods in terms of the A-optimality criterion and reconstruction error.

eess.SY↗

arXiv2020

Data-driven Vector-measurement-sensor Selection based on Greedy Algorithm

A vector-measurement-sensor problem for the least squares estimation is considered, by extending a previous novel approach in this paper. An extension of the vector-measurement-sensor selection of the greedy algorithm is proposed and is applied to particle-image-velocimetry data to reconstruct the full state based on the information given by sparse vector-measurement sensors.

eess.SP↗

arXiv2018

Dynamic mode decomposition using a Kalman filter for parameter estimation

A novel dynamic mode decomposition (DMD) method based on a Kalman filter is proposed. This paper explains the fast algorithm of the proposed Kalman filter DMD (KFDMD) in combination with truncated proper orthogonal decomposition for many-degree-of-freedom problems. Numerical experiments reveal that KFDMD can estimate eigenmodes more precisely compared with standard DMD or total least-squares DMD (tlsDMD) methods for the severe noise condition if the nature of the observation noise is known, though tlsDMD works better than KFDMD in the low and medium noise level. Moreover, KFDMD can track the eigenmodes precisely even when the system matrix varies with time similar to online DMD, and this extension is naturally conducted owing to the characteristics of the Kalman filter. In summary, the KFDMD is a promising tool with strong antinoise characteristics for analyzing sequential datasets.

physics.flu-dyn↗

arXiv2018

Extended-Kalman-filter-based dynamic mode decomposition for simultaneous system identification and denoising

A new dynamic mode decomposition (DMD) method is introduced for simultaneous online system identification and denoising in conjunction with the adoption of an extended Kalman filter algorithm\color{black}. The present paper explains the extended-Kalman-filter-based DMD (EKFDMD) algorithm and illustrates that EKFDMD requires significant numerical resources for many-degree-of-freedom (many-DoF) problems and that the combination with truncated proper orthogonal decomposition (trPOD) helps us to apply the EKFDMD algorithm to many-DoF problems. The numerical experiments of the present study illustrate that EKFDMD can estimate eigenvalues from a noisy dataset with a few DoFs better than or as well as the existing algorithms, whereas EKFDMD can also denoise the original dataset online. In particular, EKFDMD performs better than existing algorithms for the case in which system noise is present. The EKFDMD with trPOD can be successfully applied to many-DoF problems, including a fluid-problem example, and the results reveal the superior performance of system identification and denoising. Note that these superior results are obtained despite being an online procedure.

physics.flu-dyn↗