Virtual Sensors

Not all values for the description of traffic, which are necessary or beneficial for traffic management and control, can be measured directly. For example the capacities and reserves of a traffic line cannot be measured. Virtual or indirect sensors are software functions, which calculate or estimate values that cannot be ascertained from other available sensor data and information.

For example, only data from induction loops measuring the flow rate are available. With this information the flow reserve of a traffic line is not given directly, or what the probability is of even minor disruptions causing congestion. These readings also depend on the traffic load itself in combination with the setting of the traffic-lights. A virtual sensor can be used to obtain the values of the flow reserves.

“Virtual sensors” can be used

  • for the evaluation of traffic values that cannot be measured or are only barely measurable (such as traffic load dependent capacities or type and amount of flow reserves),
  • for the uniform formulation of traffic values from different sensors and data sources for traffic control (e.g. in the form of sensor fusion technologies),
  • for the compensation of faulty sensor signals and information.

It is essentially always about the appropriate combination and conversion of different sensor values into information that can be better applied for control and/or in traffic information systems. Implementation is easiest by means of machine learning approaches.

ANDATA supplies comprehensive tools for the quick and efficient development and calibration of virtual sensors as well as ready-made software modules for the quick and simple implementation of virtual sensors in arbitrary platforms and environments.

Features and Advantages

  • Indirect measurement of values which cannot be measured directly or which are difficult to read (e.g. traffic dependent capacities and reserves).
  • Better use of existing sensors, retrieving more information from fewer sensors and data.
  • Sensor fusion for the uniform combination of different information and conversion into standardized values.
  • Usage of measurements restricted in time.
  • Compensation of defective sensors.
  • Integration into arbitrary platforms and environments.