Edible fungus residue as an efficient and low-cost precursor was used to produce Edible Fungus residue Activated Carbon (EFAC) using the zinc chloride activation method at a 1 : 2 impregnation ratio and 600 °C activation for 3 hours. The activation process does not need gases like nitrogen and is suitable for mass production. Fungal biodegradation facilitates efficient chemical activation, which might have generated abundant pores on the activated carbon sample. Using BET, X-ray diffraction (XRD), scanning electron microscopy (SEM), and FTIR characterization, reveals that EFAC exhibits a large specific surface area (1070 m2 g−1), and large pore volume (0.68 cm3 g−1), with its surface displaying a honeycomb-like structure. The EFAC adsorbs methylene blue (MB) and aniline in water, with maximum adsorptions of 662.25 and 27.10 mg g−1, respectively. Various adsorption conditions, such as the EFAC dosage, pH, contact time and initial concentration were investigated. The adsorption is characterized by the pseudo-second-order kinetic and Langmuir isotherm models, with thermodynamics studies indicating that the adsorption is endothermic and spontaneous. Furthermore, the EFAC exhibited good regeneration performance by a 90% ethanol solution. The EFAC is a low-cost and environmentally friendly adsorbent for removing organic contaminants in wastewater.
According to the poor robustness of the extreme learning machine as a classifier for human face recognition, a swarm optimization algorithm is proposed. The algorithm combines principle component analysis and independent component analysis to extract human face features, and extreme learning machine is used as a classifier. In order to improve the classification performance of extreme learning machine and achieve higher recognition accuracy and better robustness, the swarm optimization algorithm is introduced in the classification stage. The experimental results show that compared with the traditional algorithm, the human face recognition system using the improved algorithm not only improves the face recognition rata but also reduces the influence on the result of training data when the numbers of the hidden layer nodes change, and has good robustness, has a great promotional value in similar classification model.