Efficient product development with the aid of artificial intelligence (AI)
As part of product development a multitude of calculations and experiments are required, which subsequently have to be validated by prototypes. The production of complex and cost-intensive tools is not rarity.
Also, transient CFD calculations often have to be excuted to evaluate the product functionality, whose computing times are difficult to determine in advance and thus making the cost and time expenditure for finding a solution incalculable.
How can an improved basis for decision-making with regard to expense and benefits in the development process be created and variants reduced?
How can a multitude of existing data and experiences from the past be used for future decisions so that only the most auspicious solution approaches have to be pursued?
We have dedicated ourselves to the question of how statements about other questions can be generated from a large number of data already available in the company from experiments and/or CFD analyes using artificial intelligence (AI) in order to make product development more efficient.
Artificial neural networks (ANN) can provide us the responses and increase the efficiency of finding solutions.
As part of the Central Innovation Programme for small and medium-sized enterprises (SMEs), we have created possibility to predict tendencies for different command variables from a large number of available CFD models and CFD results with the help of artificial neural networks (ANN), such as the prediction of computing times from CFD simulations or the cooling capacity for a heat exchanger.
Furthermore, methods were developed with which the performance of the results can be evaluated qualitatively and the risk of accuracy can be derived.