The Intelligent Fire Control Algorithm Of Automotive Airbag Based On Fuzzy Logic And Neutral Network

Airbag for automotive plays a very important role on vehicle passive safety and occupant protection, and the fire control algorithm is the key of the whole airbag control system, it contribute most for the performance of the occupant protection of the airbag.Firstly, this paper introduces the basic theory of airbag control and its conventional fire algorithm whose advantages and disadvantages are analyzed. On the basis of this discussion, with the modern intelligent control algorithm, this paper advances an airbag intelligent fire control algorithm which base on the fuzzy logic inference and artificial neutral network. According to the"127mm-30ms"rule which is commonly applied, a hybrid fuzzy neutral network model is established. It includes a pure neutral network for predicting the occupant head displacement and a fuzzy neutral network for estimate if it fulfills the airbag's fire condition. These two sub networks'outputs are both considered to decide if the airbag will be fired finally. The neutral network adopts the acceleration's variety for its input, and predicts the occupant's head displacement 30 milliseconds later. This method ensures the optimal fire time. Fuzzy neutral network adopts both acceleration's variety and speed's variety for its input, partition range of the value of acceleration's variety fuzzy, then decide if the airbag must be fired by checking that if the acceleration's variety and its correspond speed's variety are large enough meanwhile. Also, the algorithm recognizes the crash better when training the fuzzy rules by neutral network. Using the two sub networks to estimate the fire condition and confirm the fire time respectively, the intelligent algorithm in this paper gets good results in airbag control. The algorithm model is established with the fuzzy toolbox and neutral network toolbox in MATLAB 7.1. Training and validating of the algorithm network model is completed by acceleration-head displacement curves that gain from simulating with MADYMO and driver-side frontal crash model. The acceleration curves are simulated by computer based on sled crash. Results show that the fire algorithm this paper advances is effective and veracious, and such a study will be of significance in the airbag intelligent fire control field.