| Due to high speed development of economy and society, our country has been one of the largest countries in elevator’s quantity demanded, usage amount and holdings. As a means of transport that is so closely related to lives and property of the masses, elevator’s safety performance is critical.Elevator safety evaluation is to do a comprehensive examination on the security situation of elevator operates, to find these hidden safety troubles, to prevent the occurrence of elevators’ safety accident. elevator safety evaluation is an important component of elevator safety management. Manned, without-machine-room, traction elevator that was used and evaluated most frequently was selected by this thesis as research object, first of all, through analyzing its structure and safety requirements, identified its hazards and analysed the types of elevator security incident and causes. Finally, Discrete Hopfied neural network was selected as the evaluating method through comparing and analyzing of different kinds of safety evaluation methods and consideration of the characteristics of evaluation objects, object of the evaluation and application fields. Through a lot of reading of related literatures and on the basis of consulting experts’ opinions and the suggestions, An evaluation index system was established that is applicable to the method above.As a kind of comprehensive connection type of neural network, Hopfield neural network opened up a new research approach for the development of artificial neural network. It uses the structure features and learning method of different type of stratum neural networks to simulate the memory mechanism of biological neural network, satisfactory results always could be obtained through this process. In the process of elevator safety evaluation, the main problem was we didn’t have enough experts that owned sufficient knowledge and experience to evaluate every elevator, so some elevator evaluation results were not compelling enough. Here we took advantage of the associative memory function of discrete Hopfield neural network to stimulate the evaluation model of experts.Firstly, The evaluation results that made by the knowledgeable and experienced experts were transformed by this thesis into a form that could be identified by the neural network, then input them into the neural network for training, to recall the model of expert evaluation. After training the network successfully, we used this network to evaluate the safety status of elevator and tested the results of evaluation.Applying the method of discrete Hopfield neural network in elevator safety evaluation has good application prospects and popularized value. This method doesn’t need to do a large number of tedious programming. Just need to learn the method of training function calling, greatly improving the efficiency of the safety evaluation. Through taking advantage of discrete Hopfield neural network’s features of associative memorying, self-organizing and learning to stimulate the model of experts evaluating process, making the safety evaluating process more reasonable and reliable, the evaluation results more accurate. Using the discrete Hopfield neural network to evaluate the safety of elevator has good evaluation effect and important practical value. |
