Method for predicting mechanical property of propellant based on BP (Back Propagation) artificial neural network

Abstract

The invention discloses a method for predicting the mechanical property of a propellant based on a BP (Back Propagation) artificial neural network. The method comprises the following specific steps ofstep I, performing normalization preprocessing on collected propellant related data to obtain a training sample set; step II, constructing a BP artificial neural network model by designing of an input layer, a hidden layer and an output layer, and selecting a transfer function, a training function and a learning function; step III, performing iterative training on the BP neural network by use ofthe training sample set and establishing a prediction model; and step IV, inputting a variable into an optimal prediction network model for prediction to obtain the mechanical property of the propellant under conditions of different bonding agents, different cure parameters and different test temperatures. In the method, the proper BP neural network model is established according to existing dataconditions, the mechanical property of the composite solid propellant in any formula under the conditions of different bonding agent contents, cure parameters and tests can be predicted to achieve thepurposes of improving efficiency and reducing production cost.

Description

technical field The invention relates to the field of composite solid propellants, in particular to a propellant mechanical performance prediction method based on BP artificial neural network. Background technique Composite propellant is a kind of propellant composed of polymer binder, solid powder oxidizer, powder metal fuel and other additional components. It is usually divided into polysulfide rubber composite propellant, polyurethane propellant according to the type of polymer binder. Composite propellants, hydroxyl-terminated polybutadiene composite propellants, and carboxyl-terminated polybutadiene composite propellants, etc. When a variety of process additives are added to the composite propellant, there is a complex nonlinear relationship between the various additives on the mechanical properties of the propellant, but this relationship lacks a mathematical model. The research on the influence of mechanical properties has been limited to slowly debugging through exp...

AI Technical Summary

Problems solved by technology

When a variety of process additives are added to the composite propellant, there is a complex nonlinear relationship between the various additives on the mechanical properties of the propellant, but this relationship lacks a mathematical model. The research on the influence of mechanical properties has been limited to slowly debugging through experiments. This method not only requires a large number of propellant samples, but also has a long development cycle and high cost, making it difficult to adapt to new weapon development platforms.

Since the artificial neural network has very strong fault tolerance and adaptability, especially with a high degree of nonlinear ability, the emergence of BP network that can approximate any nonlinear function provides us with a new method to solve these traditional problems. , has been widely used in solving nonlinear problems. At present, there is no research on the application of BP network to the mechanical properties of propellants.

Images