Abstract:Laser welding is a widely used but complex industrial process. In this work, we propose the use of an integrated machine intelligence architecture to help address the significant control difficulties that prevent laser welding from seeing its full potential in process engineering and production. This architecture combines three contemporary machine learning techniques to allow a laser welding controller to learn and improve in a self-directed manner. As a first contribution of this work, we show how a deep, auto-encoding neural network is capable of extracting salient, low-dimensional features from real high-dimensional laser welding data. As a second contribution and novel integration step, these features are then used as input to a temporal-difference learning algorithm (in this case a general-value-function learner) to acquire important real-time information about the process of laser welding; temporally extended predictions are used in combination with deep learning to directly map sensor data to the final quality of a welding seam. As a third contribution and final part of our proposed architecture, we suggest that deep learning features and general-value-function predictions can be beneficially combined with actor-critic reinforcement learning to learn context-appropriate control policies to govern welding power in real time. Preliminary control results are demonstrated using multiple runs with a laser welding simulator. The proposed intelligent laser-welding architecture combines representation, prediction, and control learning: three of the main hallmarks of an intelligent system. As such, we suggest that an integration approach like the one described in this work has the capacity to improve laser welding performance without ongoing and time-intensive human assistance. Our architecture therefore promises to address several key requirements of modern industry. To our knowledge, this architecture is the first demonstrated combination of deep learning and general value functions. It also represents the first use of deep learning for laser welding specifically and production engineering in general. We believe that it would be straightforward to adapt our architecture for use in other industrial and production engineering settings. (C) 2015 Elsevier Ltd. All rights reserved.

Experimental phase control of a 100 laser beam array with quasi-reinforcement learning of a neural network in an error reduction loop

High environmentally adaptable phase control with reinforcement learning for coherent beam combination

Deep Learning of Coherent Laser Arrays in Angular Domain for Orbital Angular Momentum Beams Customization

Machine learning assisted two-dimensional beam-steering for integrated optical phased arrays

Cascaded Internal Phase Control of All-Fiber Coherent Fiber Laser Array

The self-learning AI controller for adaptive power beaming with fiber-array laser transmitter system

Single-step phase identification and phase locking for coherent beam combination using deep learning

Stabilization of the 81-channel coherent beam combination using machine learning

System design for coherent combined massive fiber laser array based on cascaded internal phase control

Directly Emitting a High-Power Phase-Locked Laser Array by an Internal Phase Sensing System

Coherent control using adaptive learning algorithms

Mode visualisation and control of complex lasers using neural networks

Control of Nonaffine Nonlinear Discrete-Time Systems Using Reinforcement-Learning-Based Linearly Parameterized Neural Networks

A phase-error prediction method for coherent beam combining via convolutional neural network

Model-free reinforcement learning with noisy actions for automated experimental control in optics

Deep learning-based phase control method for coherent beam combining and its application in generating orbital angular momentum beams

First Steps Towards an Intelligent Laser Welding Architecture Using Deep Neural Networks and Reinforcement Learning

Stabilizing Coherently Combined Beam Power Using a Robust Learning Algorithm

TempoRL: laser pulse temporal shape optimization with Deep Reinforcement Learning

Intelligent Laser Welding Through Representation, Prediction, and Control Learning: an Architecture with Deep Neural Networks and Reinforcement Learning

Deep reinforcement learning for optical systems: A case study of mode-locked lasers