The Emulsifier Fault Early Warning On GA-BP Neural Network Theory

Emulsion explosive is a kind of industrial explosive widely used in agriculture,conservancy,transportation,mining and other fields,which is a powerful and cheap energy and plays an important role in the economic construction and social improvement.However,while bringing economic benefits,some faults of the emulsion explosive production line also occur.Due to the particularity of the explosives industry,these faults will often bring disastrous consequences.The emulsifier is the device which is prone to malfunction in the whole production,the current safety measure for emulsifier is to set the threshold of the important parameters,and an alarm generates when the parameters exceed.Meanwhile,the equipment is regularly maintained.In fact,these methods cannot find the potential faults in time to prevent the possible risks.The traditional planned maintenance and breakdown maintenance can not meet the requirements of the safe operation of the emulsifier,and the fault alarm of emulsifier needs to transform to predictive maintenance and on-demand service.The development of prediction technique provides the possibility for above requirements.Through applying prediction technique to the safe operation of the emulsifier,this paper proposes a new method of fault alarm of emulsifier based on genetic algorithm to optimize the neural network.This method establishes multivariable model to predict the operation status of the emulsifier.Researches done in this paper are as follows:(1)On the basis of acquiring the production process of emulsion explosive,the common faults of emulsion explosive production line are listed.Through the analysis from the angle of process,the faults of the emulsion explosive usually occurs in the emulsion link.Then the fault factors of emulsifier are analyzed by using the way of fault tree,on this basic,the need to establish dynamic and multivariate fault early warning model for fault early warning of emulsifier is proposed.(2)Three common prediction methods for current fault early warning are introduced,through combining the comparison of the advantages and disadvantages of three methods with the actual situation of emulsifier,BP neural network is selected to establish the fault early warning model of emulsifier,and the genetic algorithm is used to optimize the BP neural network for the disadvantages of BP neural network,which improve the accuracy of fault early warning model.(3)The process of establishing the fault early warning model is described,and a method for judging whether the emulsifier has potential faults or not based on the error between the actual and predicted values is introduced.Finally,the validity of fault early warning model is analyzed by an example.(4)On the basis of the existing equipment and safety monitoring of the emulsion explosive production line,the preliminary design of the fault early warning system of the emulsion is accomplished.