Abstract: A new neural network parameter identification method is proposed by using the upper and lower bounds of the parameters of an electro-hydraulic load simulator. First, a mathematical model of the electro-hydraulic load simulator is built; then searching is performed for the best parameter values between the upper and lower bounds of the parameters using the neural network method. The goal is for the output under the inputs of the system to approximate the output of experiments. Finally, the efficiency of the identification parameters is tested by means of another group of experimental data. The result shows that the identification parameters can approximate the real system well. This identification method can be used in the parameter identification of complex dynamic systems.

Key words: electro-hydraulic load simulator, neural networks, parameter identification, parameter bound value, dynamic model identification