Neural Network Misfire Detection in Dynamic Skip Fire Engines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting misfires in dynamic skip fire (DSF) controlled internal combustion engines are inadequate due to the difficulty in distinguishing misfires from skipped cylinders, as both result in similar angular acceleration profiles.
Innovation Solution
The use of machine learning, specifically neural networks, to model expected exhaust manifold pressure for fired and skipped cylinders, allowing for the comparison of measured pressure to determine if a misfire has occurred by defining distribution ranges for successful firings and skips based on empirical data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional misfire detection methods are used in DSF engines, then the detection system is simple, but misfire detection accuracy deteriorates due to inability to distinguish misfires from skipped cylinders
Solution Approach 1:
The patent introduces an intermediary classification system that mediates between the raw sensor data and misfire detection. A classification module is added to interpret cylinder events, distinguishing between skipped cylinders and misfires by analyzing patterns in angular acceleration and other sensor signals. This intermediary layer resolves the contradiction by enabling accurate misfire detection without directly modifying the fundamental detection methodology.
Solution Approach 2:
The patent applies dynamics by making the detection system adaptive to the changing operational states of DSF engines. The classification module dynamically adjusts its analysis based on real-time engine conditions, firing patterns, and sensor data characteristics. This dynamic approach allows the system to maintain high detection accuracy across varying engine loads and skip fire patterns without requiring a completely static, overly complex system design.
2Difficulty of detecting and measuring
If angular acceleration profiles are used for misfire detection, then the measurement method is simple, but the ability to distinguish misfires from skipped cylinders deteriorates
Solution Approach 1:
The patent merges multiple detection approaches by combining angular acceleration analysis with additional sensor signals and classification logic. Instead of relying solely on angular acceleration profiles, the system integrates data from multiple sources including crankshaft position sensors, oxygen sensors, and other available inputs. This merging of detection methods maintains relative simplicity while significantly improving the ability to distinguish misfires from skipped cylinders through pattern recognition.
Solution Approach 2:
The patent changes the parameters used for detection by moving beyond simple angular acceleration magnitude to include temporal patterns, rate of change, and comparative analysis across multiple cylinders and engine cycles. The classification module analyzes multiple parameters simultaneously, changing from a single-parameter detection approach to a multi-parameter analysis that improves misfire identification accuracy while keeping the overall method accessible and implementable.
Data Source
AI summary
Using machine learning for cylinder misfire detection in a dynamic firing level modulation controlled internal combustion engine is described. In a classification embodiment, cylinder misfires are differentiated from intentional skips based on a measured exhaust manifold pressure. In a regressive model embodiment, the measured exhaust manifold pressure is compared to a predicted exhaust manifold pressure generated by neural network in response to one or more inputs indicative of the operation of the vehicle. Based on the comparison, a prediction is made if a misfire has occurred or not. In yet other alternative embodiment, angular crank acceleration is used as well for misfire detection.