Fuel Injection Control Device Dynamic Learning Rate Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fuel injection control devices for internal combustion engines face challenges in accurately compensating for individual differences in fuel injection valves, especially due to aged deterioration and environmental discrepancies between testing and actual engine environments, leading to unreliable initial learned values and prolonged test operation periods.
Innovation Solution
A fuel injection control device with a controller that adjusts the learning rate of the learned value based on the accumulated number of updates, increasing the learning rate when the number is smaller than a predetermined value to quickly establish reliable compensation for individual differences, and switching to a usual learning rate when the number exceeds the threshold, while also considering engine operating regions and using fuel pressure sensors to detect fluctuations and calculate deviation tendencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the learning rate of the learned value is decreased to ensure stable updating, then the reliability of the learned value is improved, but the time required to update the learned value to an appropriate value is extended
Solution Approach 1:
The patent applies dynamics by making the learning rate adjustable rather than fixed. The control device dynamically changes the learning rate based on the accumulation of operational data: using a first (higher) learning rate initially to quickly reduce reliability issues, then switching to a second (lower) learning rate once sufficient data is accumulated. This dynamic adjustment resolves the contradiction between fast updating and stable updating.
Solution Approach 2:
The patent implements periodic action by dividing the learning process into distinct phases. The control device periodically evaluates whether a predetermined number of updates have been accumulated and switches between different learning rates at these evaluation points. This periodic switching allows the system to benefit from both high initial learning rates and subsequent stable updating.
2Productivity
If the learning rate is increased to quickly compensate for individual differences, then the productivity is improved, but the accuracy of learned value updating deteriorates when noise or disturbance occurs
Solution Approach 1:
The control device dynamically adjusts the learning rate based on the stage of the learning process. By using a higher learning rate only during the initial phase when sufficient operational data has not yet been accumulated, the system achieves fast updating without the risk of noise interference that would occur with continuously high learning rates.
Solution Approach 2:
The patent applies preliminary action by accumulating a predetermined number of updates before switching to the more stable, lower learning rate. This preliminary accumulation phase allows the system to quickly establish an initial learned value, and only after this foundation is set does it transition to the more precise, noise-resistant updating mode.
3Manufacturing precision
If the initial learned value is set through testing before assembly, then the manufacturing precision is improved, but the reliability deteriorates due to environmental differences between testing and actual engine operation
Solution Approach 1:
The patent combines preliminary testing with preliminary operational accumulation. While initial learned values are set through manufacturing testing, the system also accumulates a predetermined number of updates during actual engine operation before considering the learned value sufficiently reliable. This dual approach addresses both manufacturing precision and operational reliability.
Solution Approach 2:
The control device implements feedback by continuously monitoring the accumulation of operational updates and using this information to adjust the learning rate. This feedback mechanism ensures that the system adapts to actual operating conditions, compensating for environmental differences between testing and real-world operation.
Data Source
AI summary
An electronic control unit detects a manner of fluctuation of a fuel pressure with injection of fuel by a fuel injection valve with the use of a fuel pressure sensor, and calculates a tendency of deviation of an actual fuel injection characteristic of the fuel injection valve with respect to a basic fuel injection characteristic on the basis of a result of comparison between a detected temporal waveform and a basic temporal waveform. The tendency of deviation is reflected at a predetermined reflection rate (R) at the time of updating a learned value (Gij) for compensating for an individual difference of the fuel injection valve. The predetermined reflection rate (R) is increased when an accumulated value (D) of a vehicle travel distance is shorter than a predetermined value (Dth) as compared with when the accumulated value (D) is longer than or equal to the predetermined value (Dth).