Download PDF

Siemens | Using Machine Learning to Get Machines to Mimic Intuition

The ability to learn is a precondition for autonomy. With this in mind, Siemens researchers are developing knowledge networks based on deep learning-related simulated neurons and connections. Such networks can be used to generalize information by identifying associations between extraordinarily complex realms, such as the publicly accessible Internet and a company’s internal information systems. Far-reaching and generic, this technology appears to hold the potential of mimicking what humans call intuition.

  • Siemens
    Siemens is the largest engineering company in Europe. With their positioning along the electrification value chain, Siemens has the knowhow that extends from power generation to power transmission, power distribution and smart grid to the efficient application of electrical energy.Featured Subsidiaries/ Business Units:- Digital Factory- Siemens Technology to Business (TTB)
  • Transportation
  • Maintenance
  • Autonomous Robots
    Autonomous robots are intelligent machines capable of performing tasks in the world independently of either direct human control or fixed programming. Examples range from autonomous drones, to industrial production robots, to your robotic vacuum cleaner. They combine expertise from the fields of Artificial Intelligence, robotics, and information science.The autonomous robot must have the ability to perceive its environment, analyze situational data in order to make decisions based on what it perceives, and then modify its actions based on these decisions. For example, the scope of autonomy could include starting, stopping, maneuvering around obstacles, communicating to obstacles, and using appendages to manipulate obstacles. There are few autonomous robots in operation today. Even most sophisticated, dynamic robots such as those used in an automotive factory perform according to static programming. And most autonomous robots are only semi-autonomous and will likely remain so even as more fundamental autonomy becomes technically feasible. For example, the Roomba vacuum cleaner does not move according to a pre-programmed route and can modify its route dynamically as its environment changes. However, it has a very limited degree of freedom that is determined by its programming.
    Building Energy Management
    Building energy management systems (BEMS) provide real-time remote monitoring and integrated control of a wide range of connected systems, allowing modes of operation, energy use, and environmental conditions to be monitored and modified based on hours of operation, occupancy, or other variables to optimise efficiency and comfort. Building energy management systems can also trigger alarms, in some cases predicting problems and informing maintenance programmes. They maintain records of historical performance to enable benchmarking of performance against other buildings or across time and may help automate report writing. BEMS are often integrated with building automation and control (BAC) systems, which have a broaded scope of operations.
  • This specific case study analyzes the ramifications of neural networks on the renewable energy industry, specifically companies involved with wind turbines.  

  • Biological systems that learn include everything from roundworms with approximately 300 nerve cells to adult elephants, whose brains contain 200 billion neurons. But regardless of whether you’re dealing with a fruit fly, a cockroach, a chimpanzee or a dolphin, the neurons of all of these creatures process and transmit information. Moreover, they do so for the same reasons: All organisms need to be able to discern and interpret their surroundings and then react appropriately in order to avoid danger and ensure their survival, as well as their ability to reproduce. They also need to be able to recall stimuli that signal risk or reward. In other words, learning is the key to survival in the natural environment.  Through creating a computational neural network that mimics the neurons in a human brain, Siemens' engineers are getting closer to creating virtual intuition.

  • Cutting Edge (technology has been on the market for < 2 years)
  • Impact #1
    [Efficiency Improvement - Production]
    Researchers at Corporate Technology (CT) are studying how machine learning techniques could be used to enable wind turbines to automatically adjust to changing wind and weather conditions, thus boosting their electricity output. The basis for self-optimizing wind turbines is the ability to derive wind characteristics from the turbines’ own operating data.  Up until now, this type of data has been used exclusively for remote monitoring and diagnosis; however, this same data can also be used to help improve the electricity output of wind turbines.”
    Impact #2
    Impact #3
© 2021 IoT ONE