ADS-79D-HDBK Downloaded from https://www.everyspec.com simulation-type test. If the verification proves that the requirements of the system elements are met, verification is complete. At the end of verification, a ―Prognostics Verification Document‖ needs to be generated that captures the verification tests and process. K.5.2.2 Mechanical Health-based Prognostics Validation Process. Upon successfully verifying the prognostics methodology, the production unit is generated. The production unit is essential for validating the methodology. This section explains the process by which the Mechanical Health-based Prognostics methodology should be validated (See Figure K-9).
The entry criterion for the Mechanical Health-based Prognostics Validation Process is that the prognostics verification process must have been completed. The starting assumption here is that upon successful completion of verification, the system elements will now be production software. Validation can only be performed on production software. To continue with the validation process, the next step is to establish validation requirements and metrics. At this point, a decision should be made regarding which of the following three processes should be used: Data Validation, In-Service Validation or Seeded/Natural Fault Test Validation. Data Validation:
To go down this process path, samples of fault cases for the target component should be available. The available fault cases can be used to determine whether the prognostics methodology is able to accurately diagnose the RUL of the component. However, if such a dataset is not available, one of the other validation processes needs to be followed. The data validation is successfully complete if the candidate system can accurately (based on criteria established in steps before) predict the RUL of the component. A document about the validation process will have to be generated to close out the process. In-Service Validation:
In order to perform in-service validation, the candidate algorithm and software has to be installed on the target platform. Aircraft operations will continue normally. The idea behind this validation process is that the system will acquire prognostics validation when it successfully predicts the target component‘s RUL through normal aircraft operation. In this validation method, the health algorithm shall define when a part is to be removed from the aircraft; it should be removed even if the prognostics algorithm does not say it should. When the component is removed, check that the algorithm‘s RUL estimate is correct based on tear down analysis.
There is no time limit for in-service validation. The prognostics algorithm is considered validated when the system has successfully determined the RUL of an in-service component. A document about the validation process will have to be generated to close out the process. 208