Safe Torque Estimation Through Neural Network

Sammanfattning: Mass electrication of the vehicles is spreading everywhere and new solutionsand applications related to the automotive eld have to be studied. In particular,a big problem concerning automotive industries is safety, which imposesrequirements that must be fullled. Considering the control system of a motor,requirements are promptly translated into limitations for the hardware and software,not allowing the use of important measurements for the estimation of theelectromagnetic torque produced by the motor. For this reason, it is necessaryto investigate the possibility of a new kind of torque estimator.The purpose of this work is to evaluate whether it is possible to design anew type of safe torque estimator based on a neural network, and implementin the software of the inverter ACH6530 produced by Inmotion. As the namesuggests, the neural network can use as inputs only measurements consideredsafe according to ISO standards and must return an estimation which is able tofulll all the safety requirements of level ASIL C dened by the document ISO26262.Therefore, after a long trial and error process based on dierent choicesrelated to network structures and parameters, a neural model capable of givingsatisfactory results has been designed. Implementation in the system has beencarried out after evaluating on a board which neural network structures thesystem could bear. Finally, the neural network estimator has been tested on theactual motor, giving positive results and showing that this type of applicationis possible and its accuracy is comparable to the current safe torque estimationimplemented in the system.