The Key Technology Research On Intelligent Airbag Algorithm

Statistics show that, accident casualty can be reduced by60%with the use ofairbag and belt. Airbag control algorithm is crucial in airbag control system, but mostof the algorithms are designed for HIII50th passengers which cannot protect otherpassengers perfectly. This group adopts the method of data fusion, considered thecrew types, studied and tested the airbag ignition control algorithm.Firstly, according to the basic model and algorithm of data fusion, the ignitioncontrol algorithm of data fusion strategy is proposed and the collision strengthclassification algorithm and occupant head displacement prediction algorithm isdesigned. Algorithm of collision strength classification regards the belt, the gradientof acceleration and the acceleration time as the classification index. Crew headdisplacement prediction algorithm considered the influence of Angle collision to thecrew. Two-way acceleration is used as the crew head prediction of input parameters.Used neural network which based on genetic algorithm to forecast the displacement ofoccupant head, and system controlled the airbag ignition based on the displacement ofoccupant head. Considered the disadvant age of occupant head displacement predictionmethod under the condition of severe collision, speed variation method is used forignition in severe collision, which can avoid the delay of ignition caused by using thehead displacement prediction algorithm.Secondly, validation of the airbag controller development needs a lot ofexperiments, so developed the airbag control algorithm validation test apparatusbased on simulating impact. The device is based on collision principle, usingpendulum to impact block, aluminum cylinder energy absorption, to simulate thecollision acceleration. Practicability is verified for the test device, and thedevelopmental airbag controller is initialed verified with the device, which make abase for the further validation of airbag ignition.Finally, the control algorithm is verified by sled testes. Four groups of test havebeen done, which proved that the algorithm can accurately realize the classification ofcollision strength and ignition control when passenger without belt.