Damage probability calculation device and damage probability calculation method
A damage probability calculation device which calculates a damage probability in an evaluation target region of a gas turbine computes strength distributions including a mean, a standard deviation and a certainty factor of the strength of a material used for the evaluation target region with respect to stresses on the basis of strength test data of the material. In addition, the damage probability calculation device computes stress distributions including the mean, the standard deviation and the certainty factor of stresses input to the evaluation target region on the basis of operation data of the gas turbine acquired through a sensor. Further, the damage probability calculation device computes damage probability distributions indicating distributions of damage probabilities of the evaluation target region on the basis of the strength distributions and the stress distributions.
1. A damage probability calculation device for calculating a damage probability in an evaluation target region of a machine, comprising:
a strength distribution computation unit configured to compute strength distributions, including a mean, a standard deviation and a certainty factor, of the strength of a material used for the evaluation target region, based on test data of the strength of the material with respect to stresses;
a stress distribution computation unit configured to compute stress distributions of stresses input to the evaluation target region based on operation data of the machine acquired through a sensor, the stress distributions including the mean, the standard deviation and the certainty factor of the stresses input to the evaluation target region; and
a damage probability distribution computation unit configured to compute damage probability distributions indicating distributions of damage probabilities of the evaluation target region based on the strength distributions and the stress distributions,
wherein the strength distribution calculation unit and the stress distribution calculation unit are configured to obtain the respective certainty factors according to the Bayes estimation,
and wherein the strength distribution computation unit is configured to update the strength distributions, which are prior distributions, into posterior distributions whenever the number of pieces of data of the test data increases.
2. The damage probability calculation device according to claim 1 , wherein the stress distribution computation unit is configured to compute the stress distributions based on the operation data and manufacture data of the machine.
3. The damage probability calculation device according to claim 1 , wherein the stress distribution computation unit is configured to compute boundary conditions with respect to an analysis model of the evaluation target region from the operation data and compute the stress distributions based on a response curved surface indicating a response of the stresses to the boundary conditions.
4. A damage probability calculation method for predicting a damage probability in an evaluation target region of a machine, comprising:
computing strength distributions, including the mean, the standard deviation and a certainty factor, of the strength of a material used for the evaluation target region, based on test data of the strength of the material;
computing stress distributions, including the mean, the standard deviation and a certainty factor, of stresses input to the evaluation target region, based on operation data of the machine acquired through a sensor;
computing damage probability distributions indicating distributions of damage probabilities of the evaluation target region based on the strength distributions and the stress distributions; and
updating the strength distributions, which are prior distributions, into posterior distributions whenever the number of pieces of data of the test data increases,
wherein in the computing of the strength distributions and the computing of the stress distributions, the respective certainty factors are computed according to the Bayes estimation.