A Two-Stream CNN Model with Adaptive Adjustment of Receptive Field Dedicated to Flame Region Detection

Convolutional neural networks (CNN) have yielded state-of-the-art performance in image segmentation. Their application in video surveillance systems can provide very useful information for extinguishing fire in time. The current studies mostly focused on CNN-based flame image classification and have achieved good accuracy. However, the research of CNN-based flame region detection is extremely scarce due to the bulky network structures and high hardware configuration requirements of the state-of-the-art CNN models. Therefore, this paper presents a two-stream convolutional neural network for flame region detection (TSCNNFlame). TSCNNFlame is a lightweight CNN architecture including a spatial stream and temporal stream for detecting flame pixels in video sequences captured by fixed cameras. The static features from the spatial stream and dynamic features from the temporal stream are fused by three convolutional layers to reduce the false positives. We replace the convolutional layer of CNN with the selective kernel (SK)-Shuffle block constructed by integrating the SK convolution into the deep convolutional layer of ShuffleNet V2. The SKnet blocks can adaptively adjust the size of one receptive field with the proportion of one region of interest (ROI) in it. The grouped convolution used in Shufflenet solves the problem in which the multi-branch structure of SKnet causes the network parameters to double with the number of branches. Therefore, the CNN network dedicated to flame region detection balances the efficiency and accuracy by the lightweight architecture, the temporal–spatial features fusion, and the advantages of the SK-Shuffle block. The experimental results, which are evaluated by multiple metrics and are analyzed from many angles, show that this method can achieve significant performance while reducing the running time.

Thermodynamic barrier to nucleation for manganese oxide nanoparticles synthesized by high-temperature gas-to-particle conversion

A complementary experimental and modeling study is reported here for nucleation of manganese oxide nanoparticles in premixed stagnation flames. The current synthesis occurs at relatively high flame temperature and low precursor loading. Thermodynamic analysis based on the postulated nucleation process, Mn(g) + O2(g) MnO(s), is carried out to quantify precursor supersaturation and potential impacts of the Kelvin effect on particle formation. Nucleation and growth are analyzed based on the computed temperature-time-oxygen history in the post-flame region. Agreement between measured and computed flame position for the base flame and precursor doped flames indicates that the manganese methylcyclopentadienyl tricarbonyl precursor does not inhibit flame chemistry for the conditions currently studied. Particle size distributions measured by mobility particle sizing and TEM images show reasonable agreement. Moreover, the measured particle size is predicted much more closely by a nucleation-limited mechanism rather than the size predicted by coagulation-limited growth.

Study of a Premixed Turbulent Counter-Flow Flame with a Large Eddy Simulation Method

The turbulent counter-flow flame (TCF) has proven to be a useful benchmark to study turbulence-chemistry interactions, however, the widely observed bulk flow fluctuations and their influence on the flame stability remain unclear. In the present work, premixed TCFs are studied numerically using a Large Eddy Simulation (LES) method. A transported probability density function (pdf) approach is adopted to simulate the sub-grid scale (sgs) turbulence-chemistry interactions. A solution to the joint sgs-pdf evolution equation for each of the relative scalars is obtained by the stochastic fields method. The chemistry is represented using a simplified chemical reaction mechanism containing 15 reaction steps and 19 species. This work compares results with two meshing strategies, with the domain inside nozzles included and excluded respectively. A conditional statistical approach is applied to filter out the large scale motions of the flame. With the use of digital turbulence, the velocity field in the flame region is well reproduced. The processes of local extinction and re-ignition are successfully captured and analysed together with the strain rate field, and local extinctions are found correlated to the turbulent structures in the reactant stream. The predicted probability of localised extinction is in good agreement with the measurements, and the influence of flame stoichiometry are also successfully reproduced. Overall, the current results serve to demonstrate the capability of the LES-pdf method in the study of the premixed opposed jet turbulent flames.

A kinetic evaluation on no2 formation in the post-flame region of pressurized oxy-combustion process

Pressurized oxy-combustion is a promising technology that can significantly reduce the energy penalty associated with first generation oxy-combustion for CO2 capture in coal-fired power plants. However, higher pressure enhances the production of strong acid gases, including NO2 and SO3, aggravating the corrosion threat during flue gas recirculation. In the flame region, high temperature NOx exists mainly as NO, while conversion from NO to NO2 happened in post-flame region. In this study, the conversion of NO ? NO2 has been kinetically evaluated under representative post-flame conditions of pressurized oxy-combustion after validating the mechanism (80 species and 464 reactions), which includes nitrogen and sulfur chemistry based on GRI-Mech 3.0. The effects of residence time, temperature, pressure, major species (O2/H2O), and minor or trace species (CO/SOx) on NO2 formation are studied. The calculation results show that when pressure is increased from 1 to 15 bar, NO2 is increased from 1 to 60 ppm, and the acid dew point increases by over 80?C. Higher pressure and temperature greatly reduce the time required to reach equilibrium, e.g., at 15 bar and 1300?C, equilibrium is reached in 1 millisecond and the NO2/NO is about 0.8%. The formation and destruction of NO2 is generally through the reversible reactions: NO+O+M=NO2+M, HO2+NO=NO2+OH, and NO+O2=NO2+O. With increasing pressure and decreasing temperature, O plays a much more important role than HO2 in the oxidation of NO. A higher water vapor content accelerates NO2 formation in all cases by providing more O and HO2 radicals. The addition of CO or SO2 also promotes the formation of NO2. Finally, NO2 formation in a Pressurized oxy-combustion furnace is compared with that in a practical atmospheric air-combustion furnace and the comparison show that NO2 formation in a Pressurized oxy-combustion furnace can be over 10 times that of an atmospheric air-combustion furnace.

Validation of Emission Spectroscopy Gas Temperature Measurements Using a Standard Flame Traceable to the International Temperature Scale of 1990 (ITS-90)

Accurate measurement of post-flame temperatures can significantly improve combustion efficiency and reduce harmful emissions, for example, during the development phase of new internal combustion engines and gas turbine combustors. Non-perturbing optical diagnostic techniques are capable of measuring temperatures in such environments but are often technically complex and validation is challenging, with correspondingly large uncertainties, often as large as 2 % to 5 % of temperature. This work aims to reduce these uncertainties by developing a portable flame temperature standard, calibrated via the Rayleigh scattering thermometry technique, traceable to ITS-90, with an uncertainty of 0.5 % of temperature (k = 1). By suitable burner selection and accurate gas flow control, a stable, square, flat flame with uniform post-flame species and temperature is realised. Following development, the standard flame is used to validate two IR emission spectroscopy systems, both measuring the line-integrated emission spectra in the post-flame region. The first utilises a Hyperspectral imaging FTIR spectrometer capable of measuring 2D species and temperature maps and the second, a high-precision single line-of-sight FTIR spectrometer. In the central post-flame region, the agreement between the Rayleigh and FTIR temperatures is within the combined measurement uncertainties and amounts to 1 % (k = 1) of temperature.

A Real-Time Fire Detection Method from Video with Multifeature Fusion

The threat to people’s lives and property posed by fires has become increasingly serious. To address the problem of a high false alarm rate in traditional fire detection, an innovative detection method based on multifeature fusion of flame is proposed. First, we combined the motion detection and color detection of the flame as the fire preprocessing stage. This method saves a lot of computation time in screening the fire candidate pixels. Second, although the flame is irregular, it has a certain similarity in the sequence of the image. According to this feature, a novel algorithm of flame centroid stabilization based on spatiotemporal relation is proposed, and we calculated the centroid of the flame region of each frame of the image and added the temporal information to obtain the spatiotemporal information of the flame centroid. Then, we extracted features including spatial variability, shape variability, and area variability of the flame to improve the accuracy of recognition. Finally, we used support vector machine for training, completed the analysis of candidate fire images, and achieved automatic fire monitoring. Experimental results showed that the proposed method could improve the accuracy and reduce the false alarm rate compared with a state-of-the-art technique. The method can be applied to real-time camera monitoring systems, such as home security, forest fire alarms, and commercial monitoring.

Combustion of a plane hydrogen microjet at subsonic and supersonic speeds

In this paper, we presented the results of experimental studies of the diffusion combustion of a plain hydrogen microjet flowing from a slit micronozzle at subsonic and supersonic speeds. For the first time, four scenarios of diffusion combustion of a plain hydrogen microjet including supersonic combustion in the presence of supersonic cells in both air and hydrogen are presented. The stabilization of the subsonic combustion of a hydrogen microjet was established to be due to the presence of a «bottleneck flame region» while the stabilization of the supersonic combustion of a microjet was found to be associated with the presence of supersonic cells. The observed hyster­esis of diffusion combustion of a plain hydrogen microjet depends on both the method of igniting the microjet (near or far from the nozzle exit) and the direction of change in the rate of its outflow (growth or reduction).