US 12,106,233 B2
Parameter-searching method, parameter-searching device, and program for parameter search
Yusuke Tagawa, Kyoto (JP); and Yuki Ishikawa, Kyoto (JP)
Assigned to SHIMADZU CORPORATION, Kyoto (JP)
Appl. No. 16/959,704
Filed by SHIMADZU CORPORATION, Kyoto (JP)
PCT Filed Apr. 24, 2019, PCT No. PCT/JP2019/017373
§ 371(c)(1), (2) Date Jul. 2, 2020,
PCT Pub. No. WO2019/244474, PCT Pub. Date Dec. 26, 2019.
Claims priority of application No. 2018-116364 (JP), filed on Jun. 19, 2018.
Prior Publication US 2021/0089952 A1, Mar. 25, 2021
Int. Cl. G06N 7/01 (2023.01); G06F 16/951 (2019.01); G06F 17/15 (2006.01); G06N 20/00 (2019.01)
CPC G06N 7/01 (2023.01) [G06F 16/951 (2019.01); G06F 17/15 (2013.01); G06N 20/00 (2019.01)] 16 Claims
OG exemplary drawing
 
1. A parameter-setting method to obtain a parameter value with which an observed value or an index value derived from the observed value satisfies a predetermined condition in a target chromatograph and/or spectrometer system which outputs the observed value by performing an observation under a given parameter value, the target chromatograph and/or spectrometer system being a physical or chemical analysis system configured to perform a predetermined type of physical or chemical analysis of samples, and to set the parameter value in the target chromatograph and/or spectrometer system, the parameter-setting method comprising:
a model estimation step to estimate a posterior distribution of a model function of the target chromatograph and/or spectrometer system based on target observation data observed with the target chromatograph and/or spectrometer system and reference observation data observed with a reference system related to the target chromatograph and/or spectrometer system, the model estimation step including a variation estimation step to estimate an amount of variation between the target observation data and the reference observation data within a parameter space based on the two kinds of observation data, and the model estimation step further estimates a posterior distribution of the model function after a correction corresponding to the estimated amount of variation is made;
a parameter determination step to determine a parameter value which is a condition for a subsequent observation, based on the posterior distribution of the model function acquired through the model estimation step;
a data acquisition step to set the determined parameter value in the target chromatograph and/or spectrometer system, and acquire additional target observation data from the target chromatograph and/or spectrometer system by performing the predetermined type of physical or chemical analysis of a sample while the determined parameter value is set in the target chromatograph and/or spectrometer system;
repeating a process of the model estimation step, the parameter determination step and the data acquisition step to update the determined parameter value;
setting an updated determined parameter value in the target chromatograph and/or spectrometer system as an optimal parameter value of the target chromatograph and/or spectrometer system; and
performing a measurement with the target chromatograph and/or spectrometer system with the updated determined parameter value set in the target chromatograph and/or spectrometer system.