Gear Estimation Using Clustering for Test Bench Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for testing internal combustion engines on test benches struggle to accurately simulate real driving conditions, particularly in estimating gear shifts and engine torque, which are crucial for realistic emission and consumption behavior analysis.
Innovation Solution
A method using a clustering algorithm to identify linear relationships between vehicle speed and engine speed from recorded test drive data, allowing for the estimation of gear shifts and other target values without direct measurement, and correcting for outliers and measurement noise to generate a chronological sequence of gears for a realistic test run.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If gear shift information is not directly measured during test drives, then measurement equipment and system complexity are reduced, but the accuracy of test run simulation deteriorates
Solution Approach 1:
The patent replaces direct mechanical measurement of gear shifts with a computational approach using clustering algorithms. The system processes measured data (vehicle speed, engine speed, accelerator pedal position) through algorithmic analysis to identify gear shift events, substituting physical sensors with information processing methods.
Solution Approach 2:
The patent introduces intermediate calculated parameters (vehicle speed, engine speed ratios, accelerator pedal position) that serve as mediators to indirectly determine gear shift events. These intermediate values bridge the gap between available measurements and the target gear shift information.
2Productivity
If gear shifts are estimated using simple algorithms, then processing time and computational resources are reduced, but the accuracy of test run specification deteriorates
Solution Approach 1:
The patent employs feedback mechanisms where the clustering algorithm iteratively refines gear shift detection by comparing detected events with expected patterns based on vehicle dynamics. The system continuously adjusts detection thresholds and parameters based on feedback from the measured data characteristics.
Solution Approach 2:
The patent implements dynamic adaptation of detection parameters during data processing. The clustering algorithm adjusts its sensitivity and thresholds based on the actual characteristics of the measured data, allowing optimal performance across different driving conditions and vehicle types.
3Reliability
If real driving conditions are simulated on test benches, then emission and consumption behavior accuracy are improved, but the complexity of test run specification increases
Solution Approach 1:
The patent enables the test run specification system to automatically generate its own input data by processing measured test drive data through clustering algorithms. The system serves itself by converting raw measurement data into structured test run specifications without requiring manual intervention or complex external configuration.
Solution Approach 2:
The patent transforms raw measurement parameters (vehicle speed, engine speed, accelerator pedal position) into derived parameters (gear shift events, test run specification) through systematic parameter changes and transformations. This automated parameter transformation simplifies the overall testing system while maintaining accuracy.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
The aim of the invention is to estimate a gear of a transmission for a test run in a simple manner from other available measured values (MW) of a test drive. This aim is achieved, according to the invention, in that a temporal sequence of vehicle velocity (v) and of an engine speed (N) is used as measured values (MW). From data points (DP) of associated vehicle velocities (v) and engine speeds (N), a number of ranges (Bn) having a linear relationship between the vehicle velocity (v) and the engine speed (N) is identified by means of a clustering algorithm. The clustering algorithm assigns the data points (DP) to the number of ranges (Bn) and calculates a cluster center (CZn) for each range (Bn), each cluster center being interpreted as a gear (Gn). The gear (Gn) associated with the cluster center (CZn) of a range (Bn) is associated with the data points (DP) of said range (Bn) in order to obtain a temporal sequence of gears (Gn), and the determined temporal sequence of the gears (Gn) is used as a set point value (SW) of the test run or is used to determine another set point value (SW) of the test run.