Undetected emergency in a car could lead to accidents and fatalities.Benefits
Save lives and damages. Increases customer satisfaction. Increases brand awareness.
Emergencies can be detected beforehand and can be alerted to near hospitals, ambulances, driver's family members, etc. This use case focuses on the person's facial emotions to detect the same medical emergency.
Demand planning Forecast
Order unfulfillment. Surplus or lack of stock at customer's premises. Customer dissatisfaction.Benefits
Improve supply chain efficiency (production, labour, warehousing, shipping…). Adequate cash flows. Cost-saving opportunities.
There could be a spike in the demand for a product or a service due to some external factors. An ML model tracks these factors by means of old data and avoids surplus or deficit during these spikes.
Knowledge discovery on R&D reports, tech. specifications and regulations
Having a human go through the massive amount of documents manually is costly and faulty.Benefits
Machine precision not only delivers an instantly result, but also can detect hidden common patterns in millions of documents that humans would never find.
After defining the right ontology, NLP models will scrap documents to find the wished information. e.g., Query is "Injector temperature for polipropilene parts". The model scraps millions of documents and outputs all information similar to the query.
Chatbot and conversational AI for customer complains
Communication between quality employees or with other clients and suppliers (B2B and B2C) is sometimes slow and unprecise.Benefits
Avoid calls and emails. Improve customer and supplier satisfaction. Reduce misunderstandings and therefore errors. Reduce costs.
Chatbots and conversational AI can be connected to well-knitted quality knowledge database to automatically communicate issues among stakeholders.
Prescriptively solve quality issues
Sometimes similar or events the exact same problems happened in the past. Quality managers might forget and treat the issues from zero.Benefits
Avoid reprocessing and existing quality issues. Propose solutions to similar problems from the past. Both lead to cost reduction.
With old quality issues and their solutions, it is possible to train a model. For future quality issues, the model will output past solutions with a high level of coincidence (accuracy).
Text mining in quality issues
Quality texts are normally summarized into one single word for tracking reasons. This leads to information loss.Benefits
Hidden relationships between quality texts can be discovered which leads to better quality and decision making.
With NLP and ML techniques and a proper ontology, similarities in text can be detected and therefore hidden knowledge extracted (e.g., 70 of the last 300 complaints about price, pointed to a subscription issue).
Automation in CAE
Repetitive adjustment of finite elements in similar parts projects after project.Benefits
Reduction of finite elements adjustment by +80% therefore reducing costs and engineers fatigue.
For those parts that present similarities project after project (e.g., fatigue test on a door handle ), an ML model can be created that, with dimensional and finite elemnts data from old parts, can automatically propose the correct finite element distribution for a new part with the current dimensional restrictions as input.
Automatic validation planing
Validation plans constantly change due to external circumstances (parts did not arrive on time, a test failed before expected leading to machine availability, machine breakdowns, urgencies of all sorts).Benefits
Automatic generation of validation plans in real time considering all external events and avoiding human errors, therefore reducing costs.
With input data such as parts and machines availability, customer deadlines, and test requirements (e.g., length), an ML model is created that outputs a validation planing in real-time.
Energy dynamic pricing
"Overpricing and underpricing.
Devices not used efficiently."
Reasonable pricing for both supply and demand and better dynamic use of assets. This leads to customer satisfaction and better revenues.
"Electricity can be priced differently depending on many variables such as location, type of car, cost of electricity at a certain time in the day, type of customer, etc., An AI system can fairly and dynamically adequate the price adapted to each circumstance. Additionally, device operations can be controlled based on pricing signals.
Technology demonstration of real-time pricing service impacts upon sending signals to connected devices to control their operation based on pricing signals and/or grid conditions"
Prediction of optimal charging station location/usage
It is not clear what are the success factors of an EV charger. Therefore, the proper location of new charging stations is unknown. Additionally, EV users can not schedule their charging due to the lack of ML.Benefits
Among others, prediction of hourly charger utilization leads to the Identification of the best location for new chargers, charging scheduling for EV users, and better city planning.
Based on charging station data utilization such as time of the day, electricity purchased, the price paid, postcode of charger, type of charger, etc., an ML model can be trained to predict how chargers will be next used.
Advanced Volt / Var optimization functionalities
Transmission congestion in the grid. Volt Var optimization is not leveraged.Benefits
Offering ancillary (voltage and reactive power control) services for a higher benefit. Reducing congestion in the grid.
"By means of machine learning techniques, Volt Var can be optimized based on real-time circumstances such as electricity needs, DER distribution, smart meters, weather, etc. Demonstrate enhanced algorithms to leverage Volt Var Optimization (VVO) for grid management services. Potential use cases include distribution capacitors to reduce transmission congestion and leveraging distributed smart inverters (further improve grid stability)"
DERMS / ADMS advanced functionality
DERs not aggregated into one unique system for advanced functionalities.Benefits
Increase DER coordination, optimized dispatch, VVO funcitonalities, direct resource aggregation, load cycling.
"Based on DER utilisation historical data, an ML model can be created to leverage DERMS thus facilitating enhanced visibility and control over DERs and other grid assets. Leverage DERMS to facilitate enhanced visibility/control over DERs & other grid assets; incorporate additional technologies into DERMS and increase DER coordination through aggregation for optimized dispatch, including DR & EV Integration, VVO functionality, direct resource aggregation, and load cycling. The project may also explore dispatch DERs for restoration switching use cases leveraging the estimated time of restoration forecasts"
EV BtM aggregation for demand management
Aggregation of batteries for electricity storage and G2V is still not fully implemented.Benefits
Loss in electricity surplus is minimized. Economical benefits for V2G. Sustainability increased.
Batteries connected to the grid are aggregated so that, in moments of need, electricity is injected back into the grid, and in moments of abundance, electricty is stored in the batteries.
Assessment of quality indicators in processes
Statistical tools can be limited when it comes to extracting knowledge out of real-time data.Benefits
Automatic adjustments of parameters to improve quality in real-time which reduces quality costs.
Applying machine learning algorithms (anomaly detection) in real-time data flows (quality indicators such as ratio machine parameters/scraped parts) allows the reaction to reduce quality costs in real-time (such as scrapped parts).
Quality text classification and processing
Manually classifying texts is time-consuming. Normally issues are classified once a week which might lead to delays in urgent matters.Benefits
Automatically classification of quality issues as they happen. Freeing employees form repetitive tasks. Ensuring fast reaction to urgent matters.
With NLP and ML techniques, future quality issues can be automatically classified based on past classified quality issues.
"Breakdown of a system which shuts down the production for a costly period of time."Benefits
"Reduction of costs due to Elimination of preventive and corrective maintenance,
increase in equipment life expectancy."
Through anomaly detection, variables from equipment are measured, unusual patterns detected, and the maintenance operator is alerted. E.g., rotos, haters, electric devices, pumps.
Reduction in TAKT time
Tool changeover over time. Excess of scrap. Unefficient production. Unsustainability.Benefits
Reduction in TAKT time. Fewer risks for employees. Sustainable goals.
Using changeover historical data, it is possible to improve the efficiency of equipment changeover. Using scrap data and additional sources such as machine parameter, operator, weather, etc., it is possible to reduce scrap by discovering patterns in the data.
Calculates optimal settings for robots and machines
Humans are limited when it comes to computing the best combination of machine settings for a specific case.Benefits
ML find the most optimal set of settings which leads to efficient usage of the workforce, reduction of scrap, and increase in energetic efficiency.
Using the data from past settings, ML can determine the best set of parameters to be used for a new part.
Factory/production line energy efficiency
Cost of resources are increasing because of global demand. The need to efficiently use these resources increases.Benefits
Avoid energy failures. Consumes enegry considering hourly price and factory needs. These lead to cost reduction and the achievement of sustainability goals.
Artificial intelligence monitors collect information, controls, evaluates and manages energy consumption in factories. It controls energy usage and reduces it during peak hours, identifies and signals problems, and detects equipment failures before they occur.
Voice assistant to access customer digital service
Customers might have problems with the products and services and sometimes it takes time to get in touch with the supplier/provider.Benefits
Quicker and more effective response to customer problems which leads to customer satisfaction and also information gathering.
With NLP and computer audition techniques, it is possible to connect customer questions with the right answer and output human-like audio.
Assigning and answering new driver questions correctly
Car features (of all sorts) are not entirely adapted to the unique user's profile.Benefits
User experience is radically increased leading to customer satisfaction and brand awareness.
Car users activate thousands of features while being in the car. By means of AI, these activities can be tracked and a profile created so that the car features adapt to the user for a unique experience.
Predicting vehicle breakdown and alerting drivers in advance
Some problems are reported once it is too late (corrective actions).Benefits
Major accidents and hazardous incidents can be avoided. Avoidance of corrective costs.
With anomaly detection, suspicious events are identified through the sensors installed in each system and communicated to the driver/service supplier.
Transport and ride-sharing demand forecast
Shared cars are not available due to a spike in demand. Too many cars are in a location where they are not needed.Benefits
Better distribution of cars within the city depending on circumstances and needs. Higher customer satisfaction.
There are spikes in need of vehicles in determined locations depending on several variables like weather, time, ocurring events, etc. A model is trained which predicts the demand for cars considering all aforementioned variables.
Increase fleet performance by identifying driving patterns
Not accurately locating fleet vehicles on the map. Vehicle performance not entirely adapated to the unique user's profile.Benefits
Reduction of unnecessary relocation and imporvements in performance (e.g., petrol consumption) therefore reducing costs.
An AI system gathers information from fleet usage and needs, recommending which user should use which vehicle, and adapting the car features to each user every time.
Predictive maintenance of vehicle fleet
Cars break down leading to corrective maintenance and customer churn (costs).Benefits
Reduction of preventive and corrective maintenance. Increase in customer satisfaction and brand awareness. Increase in revenues.
Sensors located throughout the car allow machine learning (anomaly detection) to detect deviation in the car components ensuring preventive maintenance.
Dynamic routing based on traffic flow and unforeseen events
Heavier traffic and higher amount of accidents. Deliveries not on time. Delays in emergencies, leading to fatalities.Benefits
Lesser traffic and therefore reduction in human risks. Better route planning which leads to customer satisfaction and better transport demand/capacity forecast.
Having a live update in the traffic flow could result, by means of ML, in predicting the best rerouting of vehicles. This can also help making space for the emergency vehicles such as fire brigets, ambulance, police vehicles.
Dynamic pricing to best determine a price for each ride
Overpricing and underpricing.Benefits
Reasonable pricing for both supply and demand leading to customer satisfaction and better revenues.
A car ride could be priced differently depending on many variables such as location, type of car, price of electricity, type of customer, etc., An AI system can fairly and dynamically adequate the price adapted to each circumstance.
Knowledge discovery on R&D reports and regulations
Having a human go through massive amount of documents manually is costly and faulty.Benefits
Machine precision not only delivers and instantly result, but also can detect hidden common patterns in millions of documents that humans would never find.
After defining the right ontology, NLP models will scrap documents to find the wished information. e.g., Query is "screw with diameter bigger than 2mm". The model scraps millions of documents and outputs all information similar to the query.
Optimizing product design using customer feedback
Customer not satisfied with the product design. Complaints not reflected in new designs. At times 'updating design concepts' process become stagnant.Benefits
Information is not lost. Feedback reflected in new design. Customer satisfaction.
By means of NLP, customer feedback is gathered. All desgin-related topics are clustered and visualized by the development team, which can act accordingly after.
Automation in CAD generation
Repetitive design from similar parts project after project.Benefits
Reduction of design parts by +80% therefore reducing costs and engineers fatigue.
For those parts that present similaritries project after project (e.g., part fixtures for testing or production), an ML model can be created that, with dimensional data from old parts, can automatically generate a new part with the current dimensional restrictions as input.
Similarities and anomalies detection on complex engineering data
Impractical timesloted descriptive identification of anomalies.
Corrective approach and therefore waste of resources.
Real-time improvements and therefore reduction in costs.
Better decision making.
ML permits to detect data anomalies (outliers) in real-time and with high precision in comparison to manual or statistical anomaly detection.