Research On Collision Security Level Detected Method Based On Array Sensing Technology

In general,most of injury of human body in the process of human-machine interaction are caused by the impact of mechanical equipment on human body which exceeds a certain human tolerance.Therefore,the classification of safety level for the possible collision in the process of human-machine interaction will be conducive to the control of mechanical equipment and provide reliable and effective basis for the coming emergency rescue,which has a positive and necessary role in the follow-up operation.This research topic combines array pressure sensor and probabilistic neural network to detect the security level of mechanical equipment collision.It has carried out detailed exploration and research on the existing problems such as response time,spatial perception and accuracy.Especially in the core problem of response time,it is solved by improving the three parts of sensor network,data transmission and algorithm recognition.Its core work includes three aspects which are hardware system design,communication rule protocol and software system design.The hardware system includes array pressure sensor,processing circuit and upper computer.The communication rule protocol includes the communication interface design and data transmission rule design.The software system includes the lower computer acquisition program and upper computer processing program.In order to solve the problem of the result deviation which is caused by the different tolerance of different parts of human body in the research process,it uses method which identifies the hardness of object to solve the problem.Then,it clarifies the classification of the safety level through the simplified classification standard of injury.In the simulation experiment,the proportional human dummy model is used to obtain collision data with small mechanical arm,and three different collision safety level detection algorithm models are constructed from the obtained data that collision safety level detection algorithm model based on improved genetic algorithm optimized BP neural network,convolutional neural network and probabilistic neural network.To compare and analyze the accuracy and recognition time of the classification test results of the three algorithm models,it is found that the average accuracy of BP neural network model based on improved genetic algorithm is average and it's not ideal in identifying time.The performance of the algorithm model based on convolution neural network in the detection of collision safety level is better than that of the above in terms of accuracy and recognition time.The performance of collision security level detection algorithm model based on probabilistic neural network in accuracy and recognition time is the best of the three algorithm models.It is also easy to train the test model,and the convergence speed of the algorithm is faster when training the network.As the number of computing units in each layer is relatively determined and easy to implement,it is an excellent collision safety level detection algorithm in a comprehensive view.