Learning-based estimation of individual absorption profiles from a single room impulse response with known positions of source, sensor and surfaces

In situ estimation of the individual absorption profiles of a room remains a challenging problem in building acoustics. This work is aimed at studying the feasibility of this estimation in a shoebox room of fixed and known geometry, using a room impulse response measured from a source and sensor at fixed and known positions. This problem is tackled using supervised learning. Three neural network architectures are compared. Simulated training and validation sets featuring various types of perturbations (surface diffusion, geometrical errors and additive white Gaussian noise) are generated. An extensive empirical simulated study is carried out to determine the influence of these perturbations on the performances of learned models, and to determine which components of the room impulse response are most useful for absorption coefficients prediction. Trained models are shown to yield errors significantly smaller than those of a naive mean estimator on every simulated datasets, including those featuring realistic perturbation levels. Our study outlines the benefit of using convolutional neural network layers, especially when geometrical errors exist. It also reveals that early acoustic echoes are the most salient feature of room impulse responses for absorption coefficient prediction under a fixed geometry.

Finite element modeling of residual stress and geometrical error formations in selective laser melting of metals

Recent works, both numerical and experimental, on residual stress and geometrical errors in selective laser melting-produced parts highlighted the preponderance of these phenomena. However, their mechanisms of appearance are not yet fully explained. An in-house finite element model was developed and implemented to reproduce their formations. The consistence of the model with existing simulation results and with respect to experimental observations was checked. Simulations were then performed using a computational design of experiments to better comprehend the underlying phenomena and the influence of the laser speed and power. Relationships between process parameters and residual stress, plastic strain, and geometrical errors formations have been put into evidence which can support optimization procedures at design stage.

Robust Design of Aero-Engine Assembly Using Taguchi Method Based on Jacobian-Torsor Model

Aero-engine assembly is the core tache in the whole process of aero-engine manufacturing. Assembly variations are unavoidable due to parts’ geometrical errors. Statistical variation analysis is an effective method for robust design that can quantitatively predict product quality in the original design stage. However, traditional methods focus on the modeling of plane dimension chain and extremum analysis, which is difficult to comprehensively consider the rich geometrical errors and their relationship to each other; meanwhile, the precision prediction is too conservative to reduce the parts’ rework frequency and adjusting difficulty; in addition, traditional methods overemphasize the promotion of parts’ machining precision, and ignore the means of overall stack optimization. To overcome these problems, firstly, Jacobian-Torsor (J-T) model is used to build the variation propagation, which is well suited to a complex assembly that contains large numbers of joints and geometric tolerances; secondly, combining with Monte Carlo simulation and J-T statistical contribution solution, the percentage contribution of each part could be solved; Finally, Taguchi multi-objective optimization method is adopted for robust design of the whole system. A case study on an entire aero-engine assembly is presented to illustrate the proposed method and the results show that this new method can effectively evaluate the assembly performance and determine the optimal assembly plan, which has strong practical guiding significance.