An Expert CAD Flow for Incremental Functional Diagnosis of Complex Electronic Boards

Abstract

Functional diagnosis for complex systems can be a very time-consuming and expensive task, trying to identify the source of an observed misbehavior. We propose an automatic incremental diagnostic methodology and CAD flow, based on data mining. It is a model-based approach that incrementally determines the tests to be executed to isolate the faulty component, aiming at minimizing the total number of executed tests, without compromising 100% diagnostic accuracy. The data mining engine allows for shorter test sequences with respect to other reasoningbased solutions (e.g., Bayesian belief networks), not requiring complex pre- and post-conditions management. Experimental results on a large set of synthetic examples and on three industrial boards substantiate the quality of the proposed approach

Recent Advances in Intelligent Algorithms for Fault Detection and Diagnosis

“…To simulate the failures of the instrumentation used, such as transmitters and valves, simulated deviations are introduced in their parameters as additive disturbances of step type and an amplitude of 10% in the instrument parameters, such as failures in the transmitter span (FST), Transmitter Span Dead Zone Failure (ZMST), Transmitter Zero Calibration Failure (FCCT), Transmitter Zero Dead Zone Failure (FZMCT), Valve Span Failure (FSV), and failure in the valve span dead zone (FZMSV) [ 7 , 25 , 32 , 33 ]. After simulating the deviations in the parameters and with the specified amplitudes, the outputs of each of the previously found models are obtained, measuring the absolute value of the error and thus obtaining the failure code to determine if there is any abnormality in the process [ 34 , 35 , 36 ]. If the failures occur over some variations in values in which they have been modeled, there are offline models, and they will will only be used if the main models fail to detect the origin of the existing failure, according to the failure code returned by the system’s detection [ 7 , 21 , 37 , 38 , 39 , 40 , 41 ].…”

A Fully Automated and Configurable Cost-Aware Framework for Adaptive Functional Diagnosis

“…The process required the diagnosis engineer's experience. To overcome such limitations, an engine based on data mining has been proposed in [10], and one on statistical data in [12] while a comparative analysis of different ML-based engines is presented in [11].…”

“…It drives the actual diagnosis process by identifying a test execution order and by determining when to stop the process because the faulty component(s) can be identified while also determining components that are surely fault-free. Details about the two engines upon which the runtime analyzer relies can be found in [10] and [12].…”

A configurable board-level adaptive incremental diagnosis technique based on decision trees

“…The approach, based on Bayesian Belief Networks, incrementally executes (groups of) tests and, based on their outcomes, constituting a partial syndrome, adapts the execution order of the remaining tests and interrupts the process as soon as the faulty component can be identified. Subsequently, Data Mining has been adopted as an alternative engine to support the methodology ( [7]), offering interesting improvements. A comparative analysis of different machine learning-based engines has preliminarily shown that decision trees constitute a good candidate ( [8]).…”