Sensor board with intelligence

In view of Germany's climate targets and the associated energy transition, the heat pump has its own special charm. However, as environmentally friendly as heat pumps are, they are not completely trouble-free. In addition to signs of wear on the electric motor that drives the compressor, faults in the heat pump can occur, for example, in the form of too low or too high temperature differences in the evaporator and condenser. For example, the temperature of the coolant could still be too high when it leaves the condenser, so that when it enters the evaporator the difference to the energy source is too small. As a result, the coolant can absorb less energy from the environment - the heat pump then works inefficiently and produces too little power.

In addition to a high-pressure and low-pressure fault, other faults can occur in the refrigerant circuit that can cause the heat pump to fail, for example defective evaporator fans, defective defrosting, dirty registers, closed shut-off valves and blocked filter dryers.

Prevent with predictive maintenance

Almost all potential weak points in the heat pump can be monitored using AI. Critical conditions can be detected and rectified at an early stage before they occur with the help of intelligent sensor technology. Maintenance of the heating system can therefore be planned conveniently and efficiently without unplanned downtime. On the technology side, the use of intelligent sensors ("embedded AI system components"), where the AI is housed directly on the sensor board and the heat pump can be monitored without an internet or cloud connection, is a good option. Embedded AI has the advantage that it processes large amounts of data - up to several terabytes a day - which is not possible with conventional cloud or server solutions, as such large amounts of data are almost impossible to transfer.

Embedded AI means the fusion of sensor and AI on one board, which is referred to as an embedded AI system component. This means that the sensor data is evaluated on site and in much greater depth and in real time. Instead of the data, only the evaluation results are passed on. This means that such solutions are also particularly secure in terms of data protection. The main areas of application are predictive maintenance, user interaction and functional innovations. Due to the limited resources of embedded AI - low memory and energy requirements - the service life is extended and energy generation is optimized. The energy requirement can even be covered in a heat pump by means of energy harvesting.

Areas of application for embedded AI in production

Embedded AI also enables precise quality control during the manufacturing process. Cameras and sensors such as lidars, radars or those for spectrography and ultrasound can monitor products in real time and detect defects or deviations from standards. Defective products can be identified at an early stage and ejected from the production process. Embedded AI can also demonstrate its advantages in autonomous robots, the monitoring and control of energy consumption and safety in the workplace.