The Gold of the 21st Century – Digital Vehicle Data

Public transport rail vehicles and buses generate many different types of data. These include 

• sensor data and vehicle control parameters
• location and on-time data from AVMS and passenger information system, and 
• data from driver assistance and monitoring systems.

Most digital data treasures still lie unused in the vehicle, hidden in the memory chips of countless sensors and system components. And even after being collected and stored in databases, the important information is often embedded in a confusing mass of redundant data.

Turning raw data into information that can be evaluated usually requires prefiltering as a first step. This filtering process depends heavily on the respective application and can evolve over time in response to changing requirements.

Various tools and methods are used to extract the desired information from the prefiltered raw data (e.g., time series analysis, data mining, business analytics), which is then further processed in downstream back-office systems. 

A standardized system architecture and interfaces is required to ensure that raw data is collected and processed in a compatible manner across manufacturers and users. 

In addition to the widely-used CAN (Controller Area Network)-bus, there can also be any number of additional fieldbuses, for example Profibus, WorldFIP and Bitbus. 

Here, the trend is moving toward replacing these with Industrial Ethernet and IP-based communication, except for time and safety-critical control tasks.

In 2002, seven European commercial vehicle manufacturers launched the FMS (Fleet Management Interface) standard, which makes specific vehicle data accessible to third party suppliers. However, only a small percentage of all telematics data present in the vehicle is available via FMS. 

Reading additional data from the internal fieldbuses is largely prevented by the vehicle manufacturers. This is understandable since this data may make it possible to draw conclusions about the internal functioning and quality of the individual vehicle components. Here, a balance must be struck between the legitimate interests of vehicle manufacturers and that of users.

For rail vehicles, the challenges are even greater. Although standardized fieldbuses exist, e.g., MVB (Multifunction Vehicle Bus) or WTB (Wire Train Bus), the messages and error codes transmitted are manufacturer and device specific.

The FMS standard is geared toward vehicles with internal combustion engines. In the age of electromobility, the vehicle data required is changing, for example, to enable charge management for electric buses.

In addition to the well-known SOC (State-of-Charge), there are many other important parameters, for example

• SoH (State of Health) 
• charge throughput, and
• battery cell voltage and temperature.

These values are of crucial for intelligent charge management, which optimizes charging times while also considering battery life. 

The nonprofit association ITxPT, which PSI Transcom recently joined, has set itself the goal of promoting the standardization of IT applications for public transport. ITxPT (Information Technology for Public Transport) consistently pursues a data-focused approach, independent of the vehicle hardware.

The TiGR protocol (Telediagnostic for Intelligent Garage in Real-time) has laid a foundation for standardization in the area of vehicle and diagnostic data. But there is still a need to extend this, for example to include electromobility.

To make the information contained in the raw data usable, PSI Transcom GmbH is developing the VDC (Vehicle Date Center) as a structured data platform for vehicle data.

The first step is

• to acquire the existing raw data from the vehicle, 
• to filter it, and  
• to store it in a well-defined form.

The interface of the VDC for data acquisition is either 

• a telematics box installed in the vehicle, which "listens" to the data traffic on the fieldbuses or directly acquires sensor data, or 
• a connection to the IT backbone of the system supplieror vehicle manufacturer, which has already transferred this data to its background system using its own OEM solutions. 

VDC supports both variants.

A filter which extracts the desired raw data in the required granularity from the available data stream is usually necessary for data acquisition. Ideally, this filter can be configured online or remotely (over-the-air) to enable a flexible reaction to different data requirements at runtime.

The filtered raw data is first stored in the "big data area" of the VDC. Various tools can be used to invoke different data analysis methods, for example qualitative labeling and KPI analysis.

Thanks to their many years of experience with AI-based software, tools like the Deep Qualicision Tools from PSI are an excellent solution.

The processed data (smart data) is either forwarded to system overviews (dashboards) or to downstream systems (e.g. depot and workshop management) for further processing.

VDC is able to merge different data sources and evaluate them together. When storing raw data, the data owner can assign usage and access rights.

Not only is data protection and security ensured, but cross-user analyses are also made possible.

For example, multiple transport companies may operate the same type of vehicle under similar conditions. They grant each other permission to jointly use the collected data. Now, analyses can be performed based on a cross-user data set, thereby improving the prediction of future events, e.g., for predictive maintenance.

As part of a research project, the interfaces and architecture of the VDC will be standardized and developed in close cooperation with vehicle manufacturers, potential users and the ITxPT association.

The focus will be placed on 

• the ITxPT-compliant integration of various data sources, 
• data usability across users, 
• the use of various plug-ins (AI toolbox) for data analysis, and 
• the development of data validation tools.

The Vehicle Data Center thus becomes the central tool for mining valuable gold nuggets from raw data.