基于改进粒子群算法优化的染色木材颜色检测算法研究

Research on Color Detection Algorithm for Stained Wood Based on Improved Particle Swarm Optimization Algorithm

为提高染色木材颜色的检测精度和速度,对樟子松木材单板进行染色,选取染色单板的光谱反射率作为输入,以极限学习机模型为基础构建预测模型,对染色单板的色度参数L^(*)、a^(*)、b^(*)进行预测,运用粒子群算法对ELM权值和阈值进行寻优,并引入非线性惯性权重和新的位置与速度更新策略改进粒子群算法,以消除其易陷入局部最优的缺点。此外,以L^(*)、a^(*)、b^(*)平均绝对误差为评价指标,与基础ELM模型及其他模型作对比,发现优化后的模型平均绝对误差为0.16,测色效果相较于基础ELM的0.68、麻雀算法优化的ELM的0.37等具有明显优势,这对于提高木材染色生产效率具有重要意义。

This study focused on enhancing the accuracy and efficiency of color detection in stained wood.Specifically,the research investigated the dyeing process of Chinese pine wood veneers.The spectral reflectance of the stained veneers was extracted and used as input data.The Extreme Learning Machine(ELM)model was employed as the predictive model to estimate the L^(*)、a^(*)、b^(*)color parameters of the stained veneers.The ELM model was optimized using the Particle Swarm Optimization(PSO)algorithm,which addressed the issue of local optima by introducing non-linear inertia weights and a novel position-velocity update strategy.Evaluation of the models was based on the L^(*)、a^(*)、b^(*)average absolute error metric.The results indicated that the optimized model achieved an average absolute error of 0.16,outperforming the baseline ELM model(0.68)and the ELM model optimized with the sparrow algorithm(0.37).This research contributed to improving the efficiency of wood staining production and holds implications for the field of color detection in wood materials.