Fixed exposure will lead to underexposure or overexposure of collected flame radiation images using CCD, which has a great influence on the temperature measuring accuracy. A temperature measurement method was proposed by image fusion with multi-exposure, which can eliminate the influence of insufficient underexposure and overexposure. The approach was first to acquire a group of flame radiation images during different exposures. Then a partial region with good exposure effect in each radiation image was obtained by segmentation, with which the complete flame image can be spliced together. An experimental system was built to calibrate the temperature measurement parameters by two-color pyrometry through a blackbody furnace. The relation between exposure time and monochromatic gray level, as well as the relation between the temperature and temperature measurement coefficient were obtained. A candle flame was selected as the measuring object and the complete and accurate flame temperature distribution was acquired following our proposed method. The experimental results show that, compared with the temperature measurement using a single exposure time, our method can effectively avoid the measurement error caused by underexposure and overexposure, and improve the measurement accuracy.

光场相机可以解决辐射测温多相机系统光路复杂、同步触发难等问题, 在辐射成像三维温度重建时有其独特优势. LSQR是求解基于大型稀疏矩阵最小二乘问题的经典算法, 该算法用于重建三维温度场时对温度初值依赖较大, 在信噪比较低的情况下重建精度不理想. 本文提出阻尼LSQR-LMBC重建算法, 通过在LSQR方法中添加阻尼正则化项, 提高火焰三维温度场重建的抗噪性能, 并结合LMBC算法, 实现吸收系数和三维温度场同时求解. 在数值模拟部分, 随着信噪比逐渐降低, 阻尼LSQR的重建效果比LSQR更加稳定, 在信噪比达到13.86 dB时, 重建精度大约提高30%. 阻尼LSQR-LMBC的平均重建误差为6.63%. 用丁烷火焰进行了实验, 重建的丁烷火焰三维温度场分布符合辐射火焰燃烧的特征, 和热电偶的测温数据对结果进行对比, 相对误差在6.8%左右.

A traditional successive cancellation (SC) decoding algorithm produces error propagation in the decoding process. In order to improve the SC decoding performance, it is important to solve the error propagation. In this paper, we propose a new algorithm combining reinforcement learning and SC flip (SCF) decoding of polar codes, which is called a Q-learning-assisted SCF (QLSCF) decoding algorithm. The proposed QLSCF decoding algorithm uses reinforcement learning technology to select candidate bits for the SC flipping decoding. We establish a reinforcement learning model for selecting candidate bits, and the agent selects candidate bits to decode the information sequence. In our scheme, the decoding delay caused by the metric ordering can be removed during the decoding process. Simulation results demonstrate that the decoding delay of the proposed algorithm is reduced compared with the SCF decoding algorithm, based on critical set without loss of performance.

Turbo codes have been widely used in wireless communication systems due to their good error correction performance. Under time division long term evolution (TD-LTE) of the 3rd generation partnership project (3GPP) wireless communication standard, a Log maximum a posteriori (Log-MAP) decoding algorithm with high complexity is usually approximated as a lookup-table Log-MAP (LUT-Log-MAP) algorithm and Max-Log-MAP algorithm, but these two algorithms have high complexity and high bit error rate, respectively. In this paper, we propose a normalized Log-MAP (Nor-Log-MAP) decoding algorithm in which the function max* is approximated by using a fixed normalized factor multiplied by the max function. Combining a Nor-Log-MAP algorithm with a LUT-Log-MAP algorithm creates a new kind of LUT-Nor-Log-MAP algorithm. Compared with the LUT-Log-MAP algorithm, the decoding performance of the LUT-Nor-Log-MAP algorithm is close to that of the LUT-Log-MAP algorithm. Based on the decoding method of the Nor-Log-MAP algorithm, we also put forward a normalization functional unit (NFU) for a soft-input soft-output (SISO) decoder computing unit. The simulation results show that the LUT-Nor-Log-MAP algorithm can save about 2.1% of logic resources compared with the LUT-Log-MAP algorithm. Compared with the Max-Log-MAP algorithm, the LUT-Nor-Log-MAP algorithm shows a gain of 0.25~0.5 dB in decoding performance. Using the Cyclone IV platform, the designed Turbo decoder can achieve a throughput of 36 Mbit/s under a maximum clock frequency of 44 MHz.

In the small-angle forward scattering technique, an inversion method of particle size distribution (PSD) using artificial fish swarm algorithm (AFSA) is proposed, which can acquire optimal characteristic parameters of the PSD. Simulations are performed to verify the effectiveness of AFSA, in which spheroidal particles of a unimodal distribution that conforms to Johnson’s S B {\mathrm{S}_{\mathrm{B}}} function is inversed at different levels of gaussian white noise. The comparison of the smooth objective function and non-smooth objective function of AFSA is conducted and discussed. The measurement system of the small-angle forward scattering based on Mie scattering theory is constructed and experiments are performed to verify the practicability of AFSA. A CCD sensor is adopted to receive the scattered light instead of the conventional photodetector. Pinhole is first proposed to replace standard particles and calibrate the measurement system, which can eliminate the influence of particle concentration. After the calibration experiments, the PSD of standardized polystyrene microspheres are measured and modified. Both the simulation and experiment results indicate that the PSD can be successfully inverted by AFSA with high reliability and stability in certain conditions, and the system calibration can further improve the inversion accuracy.

A Weibull-distribution adaptive-parameters cuckoo search (WACS) algorithm is proposed, which can converge quickly and prevent falling into local optimal values, and thus improve the global search performance of a cuckoo search (CS) algorithm. In simulations, particle size inversions were performed using the proposed algorithm for unimodal and bimodal particle systems obeying Johnson’s SB, Rosin–Rammler, and normal distribution, and the results were compared to the original CS algorithm, Weibull-distribution CS algorithm, and adaptive-parameters CS algorithm. Among them, the WACS algorithm has the best accuracy. The relative root mean squared error (RRMSE) was three to four orders of magnitude lower than the CS algorithm. The noise immunity of the algorithm was verified by comparing the particle size inversion error. Random noise [1%, 10%] was added to the scattered light energy of the target function, in 1% noise increments. The WACS algorithm prevailed, and the advantage became more obvious as the noise increased. A small-angle forward scattering experimental platform was built, and ferric tetroxide particles were selected as the measured particles. Experimental measurements were carried out on a unimodal particle system (50 µm) and bimodal particle system (50 and 100 µm), while the WACS algorithm was used on particle size distribution inversion. Compared to the CS algorithm, the RRMSE of the WACS algorithm was approximately 51% lower on unimodal and 66% lower on bimodal particle population inversions.