Companies use Industrial Intelligence to gain significant advantages over their competitors. By cleverly combining various AI methods, they can
Green Steelmaking Webinar 2023 - 30 Min.
Vision 2022 - 16 Min.
Machine Learning Conference 2021 - 41 Min.
The goal is the intelligent use of sensor and diagnostic data from public transport vehicles for the economic use of vehicles and the further improvement of the CO2 balance. In this research project, a system architecture is being developed and validated by field tests at two application partners. Data from different vehicle types and components from various manufacturers will be collected, enriched with external data and analyzed and refined by applying AI-based tools.
The goal of the research project is to explore innovative and powerful self-learning tools for analysis, prediction and decision support on the state of gas grids. With the help of Industrial Artificial Intelligence, anomalies that are triggered by natural disasters, geopolitical tensions, terrorism, organized crime and large-scale emergencies can be detected more quickly and to quickly identify appropriate measures more efficiently.
Together with a network of partners, PSI is developing an eIoT platform (Energy Internet of Things platform) in the research project DISEGO (Critical Components for Distributed and Secure Grid Operation). Smart grid services will be functionally raised to a new level on this interoperable platform by AI and machine learning methods and extended by aspects of explainability of AI decisions.
Within the research project, the foundations are being created and implemented for the use of electric mobile heavy-duty transport machines to maintain process reliability in production and ensure electrical supply security, as well as coupling to renewable energy sources. PSI supports the project with a software platform for monitoring and controlling production resources as well as with AI and cloud-based software products.
In the ENSURE research project, PSI is working with partners in phase 3 to develop an AI-based optimization for maintenance management and asset management strategies in order to bring more intelligence, data exchange and transparency to the grids. This leads to resource conservation through more efficient use of operating resources and improved integration of renewable energies into the electricity grids, thus making a decisive contribution to the energy transition.
Machine learning and artificial neural networks are the basic technologies for forecasting loads and infeed by distributed energy resources. The results provide important input data for look-ahead calculations to determine the expected network state. Hence future challenges in network management can be safely and efficiently managed today.
The SASO system makes forward-looking recommendations to eliminate current and expected malfunctions in the transport or distribution grid, which are evaluated on the basis of fuzzy logic and provide the basis for optimal decisions. SASO can thus be expanded into a self-learning system, a kind of power grid autopilot.
Advanced industrial engineering processes can be used, for example, to predict and ensure the gas volumes required to fulfil gas supply contracts. Optimization processes in the operation of compressors and plants lead to cost savings and a reduction in CO2 emissions.
Anomaly detection systematically monitors the power grid using artificial intelligence and machine learning specifically configured for security applications. It is currently used to detect anomalies on individual infeeds and transformers, both individually and in system context.
PSIgasguide identifies the best operational modes for reliable gas grid operations using boundary values and decision criteria. The software first utilizes future gas transport volumes and determines potential modes. A ranking is then calculated to assist the operator in selecting the optimal mode. The ranking is determined by evaluating different criteria based on a multi-criteria decision making technique.
Balancing open positions via the market until shortly before the start of delivery and thus saving expensive balancing energy costs is becoming increasingly important in the energy trading process. PSI uses the AI-based Smart Day Trader with KPI-driven multi-criteria optimization for intuitive and understandable determination of structured remaining positions and flexible contract management for modeling.
The Intelligent Grid Operator – PSIngo – offers a transparency and automation platform for the distribution network. As companion of network providers in the process of a comprehensively digital transformation PSIngo supports among others the entire scalable expansion path for high-automated network operation powered by Qualicision AI-based Rollout Planning at KPI-oriented and comprehensible network design.
In the process of value-adding process data analysis by Deep Qualicision, data relations for predictive control and optimization of business processes are machine-learned from historical data using Qualitative Labeling.
For optimized production control of processes in the tire industry, artificial intelligence methods are used to learn the test and validation data required for automatic image recognition. The accuracy of the classification is over 99 percent. Manual interventions in the production process are reduced and quality is constantly assured.
Extended fuzzy logic optimizes the key performance indicator-supported production processes with dynamic goal requirements and a certain and uncertain information situation. The advantages of optimization are, for example, cost savings through sequencing with flexible use of resources, the balancing of goal conflicts and the ability to react in real time.
Extended fuzzy logic is Industry 4.0-compatible and open for flexible control of processes far from rigid physical structures whether in areas like field/workforce management, predictive maintenance or agile production e.g. Automated Guided Vehicles for modular manufacturing with PSI swarm production.
Optimize complex processes in steel or aluminum mills despite interdependent production steps, reduce energy consumption and thus carbon emissions, and stay competitive with changing market demands. Planning, scheduling, and logistics require AI-based production software that supports industrial customers on their way to decarbonization for efficient, resource-saving and sustainable production.
AI-based business process optimization uses qualitative labeling to evaluate machine data on a multi-criteria basis and recommends maintenance actions in advance. This increases machine availability, improves planning reliability, and reduces total operating costs.
The Online Heat Scheduler optimizes production in hybrid steel plants using Qualicision AI. It combines real-time control and KPI-driven optimization to save costs, maximize energy, and decrease CO2 emissions.
PSI uses Qualicision AI to enhance its software solutions' optimization and decision-making processes, such as the integrated planning offered by the Casthouse Scheduler, resulting in improved team communication and KPI analysis.
Simultaneously ensuring high plant availability and minimized maintenance costs is a daily challenge in predictive asset management, in addition to scaling from one machine to plant networks. PSI uses the AI-based Qualicision technology to provide holistic and understandable maintenance recommendations and to plan specific operations and their monitoring.
KPI-based production optimization combines planning and real-time control of chained workflows across multiple resources. In production planning, both simple and complex work processes with different data layers such as order data, deadlines and capacities are projected onto the schedules.
PSI uses Qualicision AI to enhance decision-making and optimization processes in their software solutions. The Predictive Quality Tool leverages Deep Qualicision AI and Machine Learning to identify the root cause of defects and suggest solutions to minimize risk.
Extended fuzzy logic optimizes storage and transport of goods. Conflicts of goals between different key performance indicators are recognized and balanced. Delivery costs are reduced and packaging logistics are simplified.
The use of artificial neural networks optimizes automatic baggage detection based on high-resolution images from several surveillance cameras. High quality real-time detection and baggage tracking throughout the baggage handling process is guaranteed. For example, damage is automatically detected and logged.
Various combinatorial processes support supply chain optimization, e.g. in the optimal selection of locations for logistics centers. Several criteria are taken into account and the most cost-effective solution is determined as a result. The processes are scalable and reusable.
Extended fuzzy logic optimizes semi-automatically or fully automatically all processes in the depot of transport companies. This includes route optimization as well as vehicle positioning and tracking. The flexible configuration of all disposition criteria ensures improved vehicle maintenance while maintaining the required vehicle quality.
Extended fuzzy logic supports infrastructure operators in managing traffic flows and ensures optimum operation for all road users. Conflicts of objectives are balanced according to predefined key performance indicators. Maintenance procedures can be planned and smoothly integrated into the management of the transport networks.
AI real-time optimization combines the multi-criteria optimization requirements of line operation on the road from the perspective of passenger transport as well as from the economic perspective of the depot. In particular, the real-time aspect of optimization, which dynamically adapts itself to the current situation by analyzing the conflict of goals, constitutes the advantages of extended fuzzy logic as an AI method.
In industrial use, whether in production or in energy supply, significantly higher demands are placed on AI applications:
For popular applications such as speech recognition on smartphones, a recognition accuracy of 95% is sufficient. In industrial processes, a deviation of a few percentage points can paralyze the entire production.
Production must continue reliably even with incomplete data. This is where fuzzy logic comes in, providing the best decision support even under difficult conditions.
AI is not a stand-alone application. It must be securely integrated into existing processes and interact with other services via interfaces and protocols.
Artificial Intelligence also has its limits. Only in combination with conventional optimization methods does AI fully play out its advantages.
Even in flow manufacturing, production processes can be optimized using advanced fuzzy logic. Cost savings of 8 -10% are achievable through sequencing with flexible resource deployment, balancing conflicting goals, and real-time responsiveness.
A dynamic, self-organizing Industry 4.0 swarm production based on manufacturing islands and intelligent workpieces can realize additional productivity gains of 20% with appropriate optimization.
In the metal industry, PSI software enables energy savings of around 10% and accompanies the structural change to low-CO2 production.
Significant sustainability effects are achieved in optimized charging management for electric vehicles and in the management of power grids through precise feed-in forecasts based on AI and the determination of optimal switching measures.
Compared with conventional grid expansion, investment in intelligent hardware and software saves around 40% of the costs that would otherwise be required for new transformers, cables and earthworks.
In logistics, the use of Artificial Intelligence results in an increase in efficiency of more than 10%, e.g., through the use of AI optimization in the distribution center.
Using AI methods for traffic control in cities can reduce CO2 levels by up to 15%.