AI in electronics manufacturing: smarter production

Despite an already high level of automation in manufacturing, the need for even more efficient production processes is growing. The use of AI opens up numerous productivity levers along the electronics production chain.

Manufacturing companies have been working for years to digitize their production. Machines are being networked and more and more data is being generated. Artificial Intelligence (AI) has the potential to extract valuable information from this treasure trove of data. Especially in the highly competitive electronics industry, they can automate processes, increase production flexibility, reduce energy costs and resource consumption, and improve quality.

Deep Learning vs. Machine Learning

AI is first and foremost an attempt to artificially imitate human decision-making. Machine Learning (ML) and Deep Learning, which is based on artificial neural networks and is a subcategory of Machine Learning, are combined under this umbrella term.

For businesses, the difference between “classic” machine learning and deep learning is particularly important. For example, the statistical methods of machine learning are not able to process unstructured data. The result: complex feature extraction (feature engineering) by experts. On the other hand, ML does not place high demands on hardware and data volume. A typical application is predictive maintenance, which can be implemented with little programming effort, using Microsoft Azure for example, and a solid database.

Deep learning does not require feature extraction. All that is required is a sufficient amount of information, such as unstructured image or sound data, for example. In addition to machine learning models such as Support Vector Machines (SVP), Deep Learning is the current industry standard for intelligent automation of visual quality control. However, this requires large amounts of data, time-consuming training intervals and powerful computers.

AI in production

In principle, AI applications can be implemented in almost all areas of a company. In practice, however, certain areas are particularly well-suited.

  • Maintenance
    Predictive maintenance has long been recognized as a key issue in the manufacturing industry. Condition monitoring, which continuously collects measurement and production data from machines and systems, is essential for its implementation. Based on this data, machine learning algorithms optimize maintenance intervals to prevent future failures.
  • Quality Management
    When companies are asked about the benefits of AI in manufacturing, quality assurance is the area that comes up most often. Both traditional machine learning models and deep learning combined with high-resolution image processing transform quality management and control into a highly automated process. For example, after training on “good images,” the AI can independently detect any deviation from the optimal condition of a part.
  • Logistics
    Artificial intelligence can identify and prevent bottlenecks in the supply chain during production. In addition, optimizing inventory levels based on trend forecasts reduces capital tie-up.
  • Digital assistance systems
    AI-based digital assistance systems increase flexibility and productivity by providing workers with the right information for each step of a process. The intelligent algorithms learn from the interaction and adapt the messages to the user's skill level. Gesture and speech recognition also make collaboration more intuitive and secure.
  • Production control
    Artificial Intelligence is an ideal complement to modern Manufacturing Execution Systems (MES) because the data base is already available in most cases. Machine learning can be used to derive forecasts as the basis for forward-looking planning and production control. For example, throughput times can be increased, disruptions or bottlenecks can be identified in advance, and appropriate action can be taken.
  • Sales / After Sales
    There is currently discussion about the possibility of using ChatGPT—a deep learning based variant of the GPT (Generative Pre-training Transformer) language model—for human-like communication in sales and after-sales. However, data protection concerns and inaccuracies due to false assumptions on the part of the AI remain.
  • Sustainability
    Artificial intelligence enhances sustainability in almost all areas of a company. Examples include material- and energy-saving manufacturing processes, predictive process control that reduces waste, optimized control of air conditioning systems, and resource-efficient logistics.

Supporting program

Live demonstrations