Building an Intelligent Testing System for Robotic Systems Based on Sigma-Pi Neural Networks

The paper considers the construction of an intelligent testing system for robotic systems based on sigma-pi neural networks. When robots are used on industrial production lines, the task of testing industrial robots for performance arises. There are two ways to solve this problem: routine checks of robots or constant observation of the operator at the robotic line. In this work, an intelligent system based on sigma-pi neural networks is being built, which will be able to solve a similar problem using readings from sensors located at different nodes of the robot. A neural network, pre-trained according to the algorithm that is given in the work, can continuously monitor the state of the robotic complex and make a decision to stop the line in case of suspicion of a breakdown. As an example of the operation of a sigma-pi neural network in this work, an example is given based on 5 input data, that is, data from 5 sensors, normalized according to the principle "there is a signal" or "there is no signal".