Research on digital instrument recognition technology of inspection robot

In this paper, an improved Fletcher-Reeves conjugate gradient descent algorithm based on BP neural network is proposed for the instrument recognition system of inspection robot in a chemical plant. Compared with the traditional standard gradient descent algorithm, adaptive learning rate gradient descent algorithm, and Fletcher-Reeves conjugate gradient algorithm, the improved Fletcher-Reeves conjugate gradient descent algorithm has fewer iterations under the condition of optimal detection performance. The organizational structure of this paper is as follows: Firstly, a BP neural network model with double hidden layers and 10 neurons is built. Then, the four algorithms are tested on the neural network model, and the number of iterations under the optimal detection performance is compared and analyzed. Finally, the recognition experiment is carried out on the inspection robot, and the recognition accuracy of each digit is obtained by analyzing the experiment.