Artificial neural network chemistry solving for high-pressure hydrogen–air combustion
Bibliographic record
Abstract
The simulation of the time evolution of chemical species in hydrogen flames by artificial neural networks (ANN) is discussed for operating conditions representative of aeronautical combustion chambers, e.g., above the atmospheric pressure, in the presence of recirculating burnt gases and secondary air-dilution. This machine learning strategy aims at replacing the integration of ordinary differential equations for the computation of chemical source terms, which can be costly, particularly at high-pressure conditions. The training database is obtained from a population of stochastic particles undergoing turbulent micro-mixing and chemical reactions. The ANN performance is then validated a posteriori on the same micro-mixing canonical problem, replacing the chemistry integration with the ANN regression. Various operating conditions are addressed in terms of global fuel/air ratio, dilution by cooling air, and burnt gases, at 1 bar and 20 bars pressure levels. Two chemical schemes are employed: a 9-species mechanism for H 2 chemistry and a 15-species mechanism containing a reduced NOx sub-mechanism. Various neural network training methodologies from the literature are tested and compared. It is found that the most effective strategies lie in data addition, underlying the importance of the database in ANN training procedures. Furthermore, existing methodologies fail when accounting for NOx chemistry. An alternative network architecture and training method are therefore proposed and tested. The accurate monitoring of atomic mass conservation is also addressed. • The time integration by neural networks of stiff H 2 -air chemical systems is studied. • A strategy is discussed to build training databases for conditions representative of aeronautical combustion chambers. • Three different methodologies are evaluated to organize the database and the neural networks architecture. • The influence on the network accuracy of the background pressure (from 1 to 20 bars) is examined. • A methodology is proposed to predict with neural networks both the major and intermediate species, including NOx.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".