Hoisting machinery as a material handling equipment, widely used in the national economy departments, in the national “safe, efficient, green and harmonious” under the application requirements, to improve the intrinsically safe hoisting machinery, a complex system, in this paper, the affecting the safe operation of the hoisting machinery hazards, summary and analysis based on the intrinsic safety theory and correlation analysis method, on the nature of the hoisting machinery safety assessment model is established. The theory of information entropy and fuzzy mathematics, the safety evaluation method of hoisting machinery based on neural network is studied. Through summarizing the hazard factor of hoisting machinery, lifting machinery design, manufacture, installation, alteration, use, and management and so on, this paper analyzes advantages and disadvantages of commonly used safety assessment or prediction method, based on the “human–environment” of safety evaluation of ideas, will influence of lifting machinery into the ontology equipment hazards, organizational security hazards, essence of safety culture and emergency fault handling of hazards. In the paper, two neural networks are used to predict the failure rate, and the accuracy of the two methods is compared. Firstly, BP neural network is optimized by genetic algorithm for prediction. BP neural network optimized by genetic algorithm is the most widely used neural network for prediction. Secondly, Elman neural network is used for prediction. Two neural networks are used to predict the failure rate, study the structural weight of neural network, obtain the prediction result graph and prediction error graph of neural network, and analyze the results, so as to judge the availability of using neural network method to predict the failure rate.

起重机械作为一种物料搬运设备，在国民经济部门广泛使用，在国家“安全，高效，绿色，和谐”的应用要求下，完善了本安型起重机械的复杂系统。对影响起重机械安全运行的危害，基于本质安全理论和相关分析方法进行归纳与分析，建立起起重机械安全性评估模型。研究了信息熵理论和模糊数学理论，研究了基于神经网络的起重机械安全性评估方法。通过总结起重机械的危害因素，起重机械的设计，制造，安装，改造，使用和管理等，本文分析了常用的安全评估或预测方法的优缺点，基于“人与环境”安全评估的思想，将举升机械对本体设备的危害，组织安全的危害，安全文化的本质和突发事件的影响错误处理危害。在本文中，使用两个神经网络来预测故障率，并比较了两种方法的准确性。首先，利用遗传算法对BP神经网络进行优化，进行预测。通过遗传算法优化的BP神经网络是预测中使用最广泛的神经网络。其次，将Elman神经网络用于预测。使用两个神经网络来预测故障率，研究神经网络的结构权重，