ML Predictive Models for Tire Performance

Estimating vehicle wheel traction and stability in real-time using machine learning to improve vehicle control in varying road conditions. The method involves training regression and classification models offline using input features like wheel dynamics, acceleration, velocity, etc. to predict wheel traction limit and stability status. These models are then used onboard the vehicle to calculate predicted wheel characteristics based on sensor inputs. This allows estimating wheel performance in real-time without direct measurements, aiding control systems like traction control and ABS by providing estimated limits.

Source

A tire wear prediction method using machine learning to accurately predict tire wear and improve driving safety by finding excessive wear or aging conditions in time. The method involves collecting tire wear data under varying road conditions, vehicle loads, and tire types. This data is used to train a tire wear prediction model that can analyze road factors like quality and curves, as well as vehicle factors like load, to comprehensively predict tire wear. The model can also optimize itself incrementally as new data is added.

Source

Modeling a tire slip rate prediction model using artificial intelligence to accurately estimate tire slip rate in real-time without relying on vehicle speed. The model is trained using tire test data to label vertical acceleration measurements with slip rate for each revolution. This labeled acceleration data is then used to train an AI model for predicting tire slip rate from vertical acceleration. The model can be used in autonomous vehicles to improve stability, braking, and acceleration performance by providing more accurate tire slip rate estimates.

Source

A method for evaluating tire performance that uses machine learning to predict ride comfort and handling based on measured tire properties. The method involves acquiring tire measurements, inputting them into a predictive model, and deriving an output that allows the driver to evaluate ride comfort and handling. This output corresponds to the driver's sensory evaluation of a vehicle equipped with that tire compared to a reference tire. The model is trained using measured tire properties and sensory evaluations.

Source

Improving tire pressure monitoring accuracy in vehicles using machine learning. Instead of installing pressure sensors in each tire, the method determines tire pressure indirectly based on wheel speed and temperature. A tire pressure prediction model trained using historical tire data is used. The model takes wheel speed and temperature as inputs and outputs predicted tire pressure. The model can be updated over time as actual tire pressure readings are received. This allows improving model accuracy based on the specific vehicle's tires.

Source

Modeling an end-to-end tire performance margin identification model using vehicle data to accurately estimate tire condition without direct sensor measurements. The method involves obtaining tire performance margins from vehicle data, segmenting the labeled data, and training an AI model on the segmented data to predict tire performance margins from new vehicle data.

Source

Optimizing heavy vehicle motion management and reducing tire wear by using dynamic tire models to predict tire behavior and wear rates based on vehicle conditions. The models estimate tire parameters like wear, stiffness, rolling resistance, etc. given input like tire data, vehicle state, and environmental factors. By iteratively updating tire models as conditions change, the vehicle control can be optimized to minimize tire wear for specific maneuvers and loads. This involves coordinating motion support devices like brakes and steering based on the tire models. The models also predict stopping distance to optimize braking.

Source

Detecting tire lateral force using deep learning to enable tire uniformity quality control without relying on expensive equipment. The method involves training a stacked denoising self-encoder model with attention using tire production process parameters. The trained model can then predict tire lateral force from the process parameters, allowing in-line tire uniformity detection using data that is already being collected during production. This provides real-time tire quality monitoring at lower cost compared to traditional equipment-based lateral force measurement.

Source

Method for managing tire safety of vehicles using IoT to predict tire wear and optimize driving routes to mitigate tire failure risks. The method involves collecting tire service data, analyzing wear rates on different road surfaces, building a tire wear prediction model, determining optimal driving routes based on tire wear, and generating real-time tire parameter alerts. It leverages IoT, machine learning, and blockchain to provide proactive tire management for reducing accidents caused by tire failure.

Source

A method, device, and computer program for predicting tire life of electric vehicles using big data analysis. The method involves collecting driving, road, and weather data from electric vehicles and maps, processing the data to extract factors affecting tire wear, grouping vehicles by wear levels, and building a tire life prediction model using the grouped wear data. The model can estimate tire life based on current driving conditions and provide warnings to vehicle owners.

Source

Tire dynamic characteristic prediction method that can reduce the number of teacher data necessary for creating a prediction model. The method includes inputting the design factors of a plurality of tires having at least one different tire design factor into a computer; a step of inputting the static characteristics into the computer respectively; a step of inputting the dynamic characteristics of the plurality of tires into the computer; and a step of creating a prediction model capable of outputting the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted, using the dynamic characteristics as training data; a step of inputting the design factors and static characteristics of the tire to be predicted; and the computer inputting the design factors and static characteristics of the tire to be predicted into the prediction model, and outputting the dynamic characteristics.

Source

Digital tire health estimation for predicting tire wear, fatigue, and damage using machine learning models. The method involves generating tire health models based on input variables like tire type, load, pressure, speed, temperature, etc. These models are iteratively refined over time using historical data. When a tire's input values are measured, the appropriate model is selected and tire health variables like tread depth, carcass health, and ply crack growth are estimated. An output signal indicates the tire's overall health. The method allows predicting tire wear, fatigue, and damage for proactive maintenance and tire management.

Source

Estimating tire wear more accurately by extracting features from tire sensor data using machine learning models. The system involves installing a sensor on the tire to measure physical quantities related to deformation. A learning-based model extracts features from the sensor data and estimates the tire wear index. This improves wear estimation accuracy compared to directly using the raw sensor data. The extracted features are more representative of tire condition compared to raw sensor values.

Source

System for accurately estimating tire wear using machine learning. It involves calculating tire loads using a separate load calculation model, then feeding those loads into a wear calculation model. The wear estimation is compared to actual tire wear measurements to update and refine the wear calculation model over time. This iterative learning process improves the accuracy of estimating tire wear based on load data.

Source

Accurately calculating tire wear rate using vehicle data, tire measurements, and machine learning. The method involves obtaining tire, vehicle, and telematics data, and calculating tire wear rate using a self-tuning mathematical model. The model continuously improves by training on tire wear data from multiple vehicles. This allows more accurate tire wear estimation compared to static models.

Source

Estimating tire wear using sensors in the tire and vehicle to improve accuracy and adaptability compared to methods like visual inspection or tread depth gauges. It leverages radial and longitudinal acceleration, pressure, and contact duration from tire sensors to learn tire wear relationships using neural networks. The tire module collects sensor data and transmits it wirelessly to the central vehicle module. The central module estimates tire wear using the transmitted data. The neural network model takes tire pressure, ground contact features, and cycle as input to estimate wear.

Source

Method and system for designing tires that balances performance and handling. The method involves using machine learning to train a tire model based on tire parameters and performance. By determining the target tire performance requirements, the closest matching parameters are found. Then, the tire parameters are further optimized based on vehicle handling stability. This allows designing tires tailored to specific applications while maintaining good overall performance.

Source

A tire physical information estimation system that accurately estimates tire forces and other parameters using a deep learning model. The system has a convolutional neural network (CNN) with input layers, feature extraction layers, and output layers. The CNN takes tire motion data as input, extracts features, and outputs estimated tire forces and other physical parameters. The CNN uses convolution operations and normalization to process the tire motion data and extract relevant features. This allows accurate estimation of directional tire forces using a learning-based approach.

Source

A method for improving the accuracy of tire condition estimation systems by extracting changes in tire characteristics over the life of the tire. The method involves capturing tire data like pressure, temperature, and tread depth, storing it in a historical log, applying time-series decomposition to isolate underlying trends, and using a learned model to predict tire condition based on those trends. This accounts for bias and variance in tire characteristics that change over time.

Source

Method for predicting the wear of tires on real vehicles using both real vehicle data and indoor tire testing data. The method involves receiving tire, vehicle, wear condition, and initial mileage/wear information. It analyzes this data using a random forest machine learning algorithm to predict the final tire wear and life based on the initial values. This allows faster wear prediction compared to actual vehicle testing. The random forest algorithm learns wear trends from the indoor and real data, then applies that knowledge to predict final wear based on initial values.

Source