Scheduling and duty schedule optimization

In duty schedule optimization, personnel resources must be allocated to different tasks. An example would be the planning of the technical field service or the technical maintenance service. This use case should also take into account that tasks often arise at short notice. For example, a Quantum Boltzmann Machine can be used to train a neural network with data from older duty rosters and their short-term changes.

Municipal registers AI

Automated linking of citizen information from so-called specialized processes that have been separated until today. The event of a change of address at the residents’ registration office should automatically lead to the offer of a new residents’ parking permit and the re-registration of waste disposal, without the systems communicating with each other via standardized interfaces. Thus a kind of recommendation system is created, which in turn can be solved as an optimization problem on quantum computers.

Water anomaly detection in public buildings

Detection of pipe bursts and other faulty operating conditions via noises on the main water supply line. The early detection of pipe bursts, for example in sports halls and schools, is of great importance for administrations and insurance companies because of the potentially high damage amounts. A Quantum Support Vector Machine, for example, can be used here as an application in the area of predictive maintenance.

Network Operation

In order to ensure a smooth operation of the core or access network, the load on all nodes and edges must be minimized. At the planning level, it is important to find a cost-minimizing expansion of the capacity matrix (i.e. minimum network expansion costs) for a given development of the network throughput in order to operate the future volumes in the network with minimum “maximum utilization”. For many related graph problems, formulations exist as optimization problems that can be solved or their solutions improved based on QAOA or Quantum Annealing.

Industrial Production Lines

In a (sheet metal) production facility, it must be decided upon receipt of an order how parts are to be distributed on raw material sheets and when which of these raw material sheets are to be processed on which machine. Of course, the restrictions of the problems must be observed, e.g. a machine can only process one sheet at a time, the number of machines is given and an order must be completed within the scheduled time. This use case can be treated as a decision making problem using Quantum-enhanced Reinforcement Learning or as a linear equation system using HHL.

Customer Behavior Prediction

Customer Churn Prediction” is the ability to recognize the probability of a contract change from the behavior of the customer. Here, model-driven AI approaches can be used, which learn a behavior model from customers and thus act similar to methods of predictive maintenance. Training an AI on concrete countermeasures (better offers, more direct contact, etc.) is promising here and could be supported by Quantum Reinforcement Learning, for example.