Showing papers on "Fault detection and isolation published in 1968"

Detection of Changes in the Characteristics of a Gauss-Markov Process

Abstract: By an extension to the theory of sequential detection with dependent measurements, it is possible to develop a sequential probability ratio test (SPRT) to detect changes in regime in a Gauss-Markov process rather than detecting which of the two regimes exists. It is shown how a posterior form of this extended SPRT may be simplified to reduce computational complexity. The simplified SPRT's are in fact modifications of the original SPRT detecting the regime and not the change. The tests are applied to the problem of fault detection in a gyro navigational system; the results of a detailed computer simulation are given.

Fault detection in a linear cascade of identical machines

Abstract: Many machines can be decomposed into a linear cascade of identical sequential machines. We consider the problem of implementing fault detection experiments for such machines. Cascade equivalence and information lossless properties of cascade machines are studied. Sufficient conditions for efficient testing of these machines are derived. Procedures are presented for testing cascades of arbitrary length N with either 2 or 3 simple present experiments. The total length of the fault detection sequence obtained by employing the methods presented is shown to be significantly less than if the classical testing procedures were taken.

Periodic motor vehicle inspection and predictive analytical modeling

Abstract: PERIODIC MOTOR VEHICLE INSPECTION (PMVI), IS AMENABLE TO EVALUATION USING PREDICTIVE ANALYTICAL TECHNIQUES. USING A MODEL RELATING VEHICLE COMPONENT FAILURE AND INSPECTION EFFECTIVENESS PARAMETERS, THE AUTHORS HYPOTHESIZE THAT AN ANNUAL MOTOR VEHICLE INSPECTION SHOULD HAVE LITTLE EFFECT ON THE DEFECT RATES OF THOSE COMPONENTS WHICH FAIL FREQUENTLY BUT ARE EASILY DETECTED AND RATHER READILY OR WILLINGLY REPAIRED BY OWNERS. ON THE OTHER HAND, PERIODIC INSPECTION SHOULD BE QUITE EFFECTIVE WHEN APPLIED TO COMPONENTS FOR WHICH FAULT DETECTION IS LOW. RESULTS OF THIS INVESTIGATION SUGGEST THAT FOR CERTAIN COMPONENTS, DEFECTS MAY BE BETTER CONTROLLED BY INCREASING THE FAULT DETECTION RATE OF THE OWNERS. AUDIBLE SIGNALS OF DEFECTIVE WORN BRAKES, VISIBLE INDICATORS OF WORN TIRE TREADS, SIGNALS ON THE DRIVER'S CONTROL PANEL TO INDICATE FAILURES IN THE VEHICLE'S LIGHTING SYSTEM ARE PRESENTLY AVAILABLE TECHNIQUES THAT FULFILL THIS FUNCTION. /AUTHOR/

Solid-State, High-Speed Phase-Comparison Relay Based on Zero Crossings: II - Design

Abstract: A relay scheme has been developed based on the instantaneous phase current zero crossings. Transient and steady- state unbalances will produce shifts of the phase currents, therefore measurement of the deviation of successive zero crossings from 60 degrees leads to six possible fault indications per cycle. Comparison of the deviation magnitude at either end of a relay zone can be used to locate the fault. This scheme requires no additional delay due to analog filtering. It is shown that all types of faults with zero or finite impedance give at least two and a maximum of six fault indications per cycle.

Dynamic fault diagnosis techniques.

Abstract: : The work was directed toward the advancement of a recently developed technique for diagnosing faults in conventional electronic circuits. Results of a field data collection effort are presented and discussed. An automatic tester, FRIEND, is described which implements the tests prescribed by an IBM 7094 computer program. Also included is a discussion on the extension of the technique to the system level. The technique for fault isolation is based on the analysis and measurement of various network functions of the circuit under test. The required test points are limited to the normally available input-output terminals, thereby reducing the complexity of required test procedures and equipment. Modifications and improvements to a FORTRAN-IV computer program, which was developed under a previous contract, are described. The computer program is capable of automatic generation of test specifications and fault diagnosis tables for conventional linear circuits of practical size. An updated computer program listing has been delivered as a separate item from this report due to its size, approximately 14,000 FORTRAN statements. The detailed theory of operation, wiring diagrams, calibration procedures, etc., for FRIEND have also been submitted as a separate document. In addition, the results of an experiment on a demonstration circuit used with FRIEND are included in this report, along with the results of some special studies. (Author)

Data systems and data analysis methods to support onboard in-flight operations.

Abstract: In aircraft or spacecraft operations, it is desirable for safety and economic reasons to monitor continuously the condition of critical vehicle systems such as engines, controls, nagivation, etc Currently, onboard systems to accomplish in-flight monitoring and checkout are gaining acceptance, and are being used more widely in aircraft and spacecraft As the capability and capacity of onboard hardware for in-flight monitoring improves, more elaborate analytical methods may be employed for data processing As an example, this paper suggests how some postflight data analysis methods for launch vehicle evaluation may be used within an onboard in-flight checkout system A concept is proposed for an integrated data processing system using modifications to existing techniques, to permit near-real-time fault detection, isolation, diagnosis, and prediction The emphasis is on sequential processing on only out-of-tolerance data The approach is to reduce continually the amount of data to be processed, on the basis of whether additional information can be gained from the next processing step Recommendations are given for further work necessary to develop and implement the concept to permit onboard in-flight monitoring of critical systems

Transformer coupled multi vibrator pulse generator for fault detection

Abstract: An AC pulse generator for use in line fault detection is constructed as a manually portable transistorized multivibrator for producing a distinctive signal having a pip, which is applied to the circuit under test. Doing so produces unbalanced currents which produce a net magnetic field along the current path, which can then be detected.

Hardware fault detection

Abstract: One of the most pressing problems currently confronting the design automation specialist is that of the automatic generation of component failure detection and diagnostic test sequences. This problem is made more difficult due to the new LSI circuits where we may have hundreds of components on a chip with only a few accessible input/output terminals.

Synthesis requirements for fault detection.

Abstract: Requirements for analysis and synthesis of diagnostic tests for detecting faults in combinational logic circuit