Smoke Simulation Within An Aircraft Cargo Compartments Based On MIGA Optimization Method

Due to the relative tightness and isolation of the aircraft cargo compartment,it is easy to spread into a fire and cause great losses when flames appear.Therefore,airworthiness regulations require the installation of smoke detection alarm in cargo compartments and other inaccessible compartments and give visual warnings at specified times.At present,during the airworthiness certification process of smoke detection system in the aircraft cargo compartment,the smoke generator is often used to perform the flight tests instead of real smoke sources.However,there are differences in the smoke diffusion process between real and simulated smoke,and few studies have been able to determine the quantitative impact of the differences on the detection.Therefore,modeling and experimental research on the regularity and differences of detection of real and simulated smoke are conducted to quantitatively evaluate the boundary conditions of smoke detectors for equivalent detection.Objective to provide engineering reference for airworthiness certification.The main research work includes the following:Based on the fire dynamics simulator software,three-dimensional numerical models of the cargo compartment and real smoke and simulated smoke sources are established.In order to conveniently describe the rule of smoke diffusion,three-dimensional numerical models of large eddy simulation,true and simulated smoke sources,particle dispersion,and smoke release are established using computational fluid dynamics methods with a low reynolds number unsteady and incompressible Navier-stokes governing equation.The numerical method is established after the grid independence verification and the verification of the same working condition test results.The comparison of the test results with the simulation results shows the feasibility of building numerical model.Analyze the differences in smoke diffusion and detection between simulated smoke and real smoke.Based on the control variables,multiple groups of simulation and experiment schemes for changing the smoke generator's smoking program are designed,and the differences between simulated smoke and real smoke detectionunder different smoking modes are sought.It is found that under the same smoke volume and the same release time,the difference in light transmission between real and simulated smoke is large,so the detection equivalent can not be achieved.When the real and simulated smoke release rates are the same,the early simulated smoke is easier to detect than the real smoke,and the later simulated smoke dissipates faster;when the smoke generator emits smoke at intervals,the light transmission of the real and simulated smoke can be closer,and the results are affected by the number of time intervals and the smoke flow.A multi-island genetic algorithm is used to optimize the equivalent smoke generator boundary conditions for true and simulated smoke detection.Aiming at reverse solving the problem of smoke generator's conditions equivalent to the real and simulated smoke detection,an optimization method is used to calculate the solution.Based on the analysis of the difference between real and simulated smoke,the minimum value of the sum of the squared residuals of true and simulated smoke transmission is constructed as the optimization objective function.And determine the optimization variables and their constraints.In order to simplify the process of repeatedly calling the FDS calculation,an approximate model of a three-dimensional numerical model is constructed using a neural network,instead of optimizing the nonlinear N-S equation in the loop.In order to avoid the calculation falling into a local optimum,a global exploration Multi-Island Genetic Algorithm is selected for optimization calculation.As a result,the optimal combination of smoke flow and time intervals at the detection location are obtained.Finally,the optimization results are obtained and verified by experiments,which show the reliability of the optimization method for solving the optimal smoking program of the smoke generator.