Skip to main content


Project Dates
12/01/1999 - 12/31/2001
Principal Investigators
Project Status

New York City Transit Department of Buses faced a problem where they found buses frequently breaking down during operation. In order to reduce the number of breakdowns, two approaches are being explored. First, it was felt, using the data present in their database, to develop statistical models to effectively predict the failures of critical components. This would also help in scheduling the buses for maintenance operations. In order to have an impact, using Pareto Analysis, systems that had the most number of failures were identified for the study. These systems consisted of a plethora of components, which were grouped into families. Relevant data was extracted from the central database and the data was validated before the data sets were reduced and prepared for statistical analysis. Various models were studied, but reliability analysis was preferred over the rest. A software application using Visual Basic was developed to automate the process and assist in determining the life various components in a bus.

The second approach to reducing breakdowns is through the use of smart sensor technology. It is evident that a more accurate method of predicting a bus's next failure, such as predictive maintenance, will be useful. One specific application of predictive maintenance is to install sensors on each bus that collect and analyze data in real-time (while the bus is running).? This data can be used to predict when the failure of certain components will occur. With the advent of high-speed personal computers and the boom of technological advances in the electronics industry the cost of using computers to collect data from a network of sensors is decreasing rapidly. Such technology is the focus of companies like Clever Devices, developers of the IVN II system.

The IVN II system, which stands for Intelligent Vehicle Network, uses automatic vehicle monitoring (AVM), a network of sensors connected to a computer used to collect the types of data that a predictive maintenance system needs. This study identifies what data should be collected from buses to affectivity implement a predictive maintenance program.? Several steps are taken to achieve this goal. The first step is to determine which systems, such as the air system, engine, and transmission, are critical to the operation of buses, and which systems cause the most unscheduled maintenance. After the critical systems were identified each system was studied in detail to determine the components critical to the operation of that system. Consequently,? a catalogue of sensors that can collect data to predict the failure of components was generated. After the completion of this study further work is expected be done to develop methods of checking for proper operation of buses based on data from the sensors.