Bosch wants to better use the batteries. With a networking of the monitoring systems and self-learning algorithms, the actual, individual battery state is to be considered stronger. Ideally, a performance increase is accompanied by a deflection of life.
The mostly used lithium batteries in electric cars hold around 500 to 1000 charging cycles. According to bosch, the manufacturers normally become a mileage between 100.000 to 160.000 kilometers ared. At an average use, the batteries thus hold around eight to ten years – slightly more long than a vehicle generation. Since he is the most expensive component on an electric car, the battery management in the vehicle is already monitoring the operating conditions down to the cell level.
In the development of battery monitoring, the engineers had to hold up to typical parameters so far and comply with potentially extensive values to potentially harmful values, especially at cargo and drainage. In the case of programming, as well as standardized amptions were previously the aging of the cells.
The main aging factors are in addition to the already mentioned pure number of charging cycles already high stroms, as they occur during fast shop and in all driving situations, in which speed is accelerated or strongly recuperated. Coarse influence take high or low temperatures – at the beginning of the charge or start it can be the aux temperature, in operation then heat due to charge and / or unloading.
So it can already have a significant effect if only 95 percent is charged with cold or brick battery. No for the user in the range lane-bearing difference, but in the long term a life-prolonged maaking. This can make a good vehicle battery management, but only within its system boundaries.
Step over system limits
Bosch now goes one step further and captures "over the air" all for a particular battery relevant parameters such as charging speed, temperature, age of memory in real time. Since this information is so for a rough amount of identical batteries, bosch can use machine learning methods. Algorithms trained on the basis of the collected data, for example, the remaining life and performance of a particular battery can be predicted as before. That’s – says bosch – so far "not yet accurate" and therefore a "novum".
The findings from the data of the entire captured vehicle fleet should identify even more stress factors for vehicle batteries by means of learning software for the vehicle batteries in real time. Batteries are supposed to pick up their capacity better and still hold long. The new service calls bosch "battery in the cloud".