This IDEA project developed an automated general-purpose tester with artificial intelligence capabilities that can be adapted to the testing of a variety of transit equipment electronic units.
The artificial intelligence software incorporated in the automated, programmable general-purpose test equipment consists primarily of neural networks that have the capability of being “trained” to recognize certain failures from specific waveform patterns as depicted in Figure 1. The programmable card-based instruments are under the control of a personal computer (PC) with a graphical user interface (GUI). Several graphically based, off-the-shelf software systems from National Instruments greatly simplified the encoding of the needed control and display software. Diagnosis is accomplished quickly, and often quite accurately, without the need for timeconsuming probing and circuit analysis procedures.
The project has demonstrated the productivity gains possible in the transit environment with PC-controlled programmable test equipment that employs flexible software architecture and a graphically based programming language. With such a system, even technical personnel not proficient in computer programming can configure the equipment. The automated generalpurpose tester incorporates software of programmable artificial intelligence tools, such as neural networks and inference generators, to assist in diagnosing circuit failures. Automated, programmable general-purpose test equipment greatly enhances testing efficiency while reducing overall test equipment costs.
The Bay Area Rapid Transit (BART) used the product to test and repair other transit equipment and to make appropriate modifications for troubleshooting electronic operational devices. This product has not experienced much application for transit equipment maintenance. NTIS # PB99-113201
