Temperature Identification Of Submersible Motor Based On Neural Network

Submersible motor units generally work in deep underground oil wells and the operating environment is very harsh.In order to ensure their efficient and stable operation,relevant parameters such as the temperature of submersible motors under the ground must be monitored all the time in order to master the operating status of submersible motors,prevent them and avoid the occurrence of accidents with submersible motors.Due to the unique working environment of the submersible motor,the conventional temperature sensor is difficult to be directly applied at high temperatures.Special temperature measuring devices are highly cost and difficult to implement.In order to accurately monitor the temperature of the submersible motor,the neural network based temperature monitoring method for submersible motor is proposed in this paper.The method can use the parameters easily measured on the ground to push down the temperature of the submersible motor,and doesn’t require the accurate mathematical model,which is suitable for complex nonlinear applications.According to the steady state model of the asynchronous motor and close relationship between the stator resistance and the motor temperature,it can be introduced that the temperature of the submersible motor can be reflected by measuring the stator current and the stator voltage.The sample data of stator current,stator voltage and submersible motor temperature are obtained through experiments.The BP network and RBF network are respectively used to train the data through fitting in MATLAB,and comparing the generalization of trained BP network and RBF network.The result shows that the RBF network has a faster response speed,higher identification accuracy,and stronger generalization ability than the BP network in the submersible motor temperature identification system.The RBF neural network algorithm is used to identify the temperature of the submersible motor.In order to further reduce the error,this paper uses the hybrid algorithm to train the number of nodes in the hidden layer,the center value,width and weight of the radial basis function in the training process of the RBF neural network to obtain the RBF neural network identification.The simulation experiment of MATLAB shows that the RBF neural network based on hybrid algorithm has better identification accuracy and generalization ability.The trained model can accurately and quickly identify the temperature of the submersible motor.Therefore,the method proposed in this paper for the identification and monitoring of submersible motor temperature based on neural network has certain practical value and application prospect.