Dan Friedlander, former Head of MBT/IAI Components Engineering has published an ongoing articles series dealing with COTS in general and use of COTS in space applications in particular. Following is the link to an index of the already published articles.

This paper deals with possible approaches to optimization of maintenance strategy of complex systems. It is based on the experience with Aircraft systems and their elements, but the methodology can be applied to any industry. It is well known that Preventive Maintenance applied instead of (or in addition to) Corrective Maintenance may increase Reliability yet influencing expenses: cost of maintenance and downtime losses. Optimization of Preventive Maintenance Plan can lead to considerable savings and improve profitability of equipment users. Two main approaches based on available data for system elements are used for the optimization of the Preventive Maintenance: 1) Traditional statistical approach, based on the reliability characteristics of the population of items; 2) PHM approach, based on measureable parameters of the individual items. The paper considers both approaches, their advantages and disadvantages, and proposes improvements to them.

The present work describes an advanced text categorization procedure developed and successfully used in aerospace industry, especially for safety assessment, analysis and improvement. The purpose is the computerized analysis and interpretation of human reported free-text aviation safety records, in order to automatically “read”, discover and treat anomalies occurred in the field. The methodology and algorithms were verified on actual, significant and appropriate ASRS (Aviation Safety Reporting System) data base (http://asrs.arc.nasa.gov/index.html) as well as other similar data bases containing millions of unprocessed safety and reliability reports. One of the most important applications and goals of the research is to assign new incoming safety event reports to one or more of several predefined categories on the basis of their textual content. Optimal categorization functions can be constructed from labeled training examples (i.e., after human expertise) by means of supervised learning algorithm and cross-validation. Numerous methods for text categorization have been previously developed such as Neural Networks, Naive Bayes, AdaBoost, Linear Discriminant Analysis, Logistic Regression, Support Vector Machines (SVM), etc. SVM has become a popular learning algorithm, used in particular for large, high-dimensional classification problems; it has been shown to give most accurate classification results in a variety of applications. However, the direct application of these methods to Aerospace Anomaly Discovery is restricted for the following reasons: a) fully automatic procedure can support only middle values of Recall and Precision (50-75 %); b) lack of stability of the reports statistical parameters - i.e. the frequency of words in a report has been changing on a "year to year" basis.

Failure Reporting, Analysis and Corrective Action Systems (FRACAS) are widely used in today's practice in almost every industry. While the principles of FRACAS development and operation are well known and defined in various standards (military, industrial, commercial), the resulting picture is not always satisfactory. It is not sufficient to see FRACAS as a "technical" data collection and data processing system. It is important to see and understand the function of FRACAS as a management system. In the framework of FRACAS, multiple management functions, related to events, people and processes, are implemented.