Downloaded from https://www.everyspec.com ADS-79D-HDBK estimates as well as some preliminary engineering judgment regarding failure modes. From the allocation and preliminary data, some choices can be made to focus on particular components and failure modes for CBM feasibility. Again, using data from previous similar designs and experience, some estimates can be developed which model the CBM benefits and costs (weight, power, complexity). The initial design stage can then mature those estimates through Component Advanced Design (CAD) studies prior to the completed system preliminary design. D.6.2.2 Determine the best means of measurement.
From a review of the physical and functional models of the components, engineers can match the parameters to sensor requirements for sensitivity and range. These designs occur in parallel during CAD, using models and any other means to assess the effectiveness of sensor placement and to estimate the signal strength and fault feature characteristics. D.6.2.3 Determine the design system capabilities. During CAD and subsequent design iterations, determining the system performance through modeling and potentially small scale testing can improve the CBM system design and mitigate risks of CI development in later testing phases. D.6.2.4 Identify candidate feature extraction/CI algorithms.
Candidate features can be identified through literature searches for new techniques as well as trials of previously developed work for analogous systems and fault modes (See Appendix E for examples of proven CIs for vibration based fault detection). Another approach is to use simulation and modeling. Figure D-10 shows an approach to model based development of a CI, in this case involving a crack in a transmission subcomponent. Using finite element modeling and estimated load profiles, it is possible to develop a simulation of the fault behavior that can be used as a starting point for CI development.
As in the case of data driven selection for a legacy system, it may take several iterations to develop CIs with the appropriate accuracy, detectability, separability, and identifiability D.6.2.5 Obtain data to train and evaluate the CI. The only way to ensure the CI is sufficient is to test the CI with data. In early stages of development, surrogate data from a similar component or simulated data from extensive simulation and modeling may be the only means to test the CI.
As the development matures and actual devices from vendors are placed under test (or previous test data is made available), CI testing and iterative improvement is possible if sufficient time and resources are allocated to the effort. D.6.2.6 Code the algorithms and test performance. After selecting a number of candidate algorithms for the CI, the algorithms are converted to software, typically through the use of engineering development packages such as MatLab™ in the same manner as the legacy aircraft. These programs are easily configured to read the data files obtained in Section 5.2.5 and run through the algorithm calculations.
The algorithms are subjected to the same analysis for accuracy, detectability, separability and identifiability. Once performance has been validated and verified at the system level, on aircraft testing for the full system is accomplished as discussed above in the legacy case. The validation and verification process for the new development should be able to address the key metrics of availability and impact on MTTR, with some statistically reasonable approach to factor in the limited number of aircraft and flying hours accumulated during Developmental Test or Operational Test. These methods and techniques are 91