非均质物料链式组合称重定量算法优化与试验

非均质物料质量差异较大且不可分割,组合称重定量过程中组合对象不确定,存在组合称重定量精度与组合速度的矛盾。该研究针对链式组合称重定量系统,提出以定量精度及组合效率为目标,对组合样本数和抽样数进行优化分析,达到保证组合称重定量精度下,减少数据计算量以提高组合定量速度的目的。研究表明,在相同允许组合误差下,增大组合样本数可提高组合成功概率,但组合计算量随组合样本数增加而呈指数增加。通过对服从正态分布N(100,102)的质量数据进行10 000轮组合计算发现,当组合定量目标质量为500 g,允许组合误差为0.1 g时,组合计算时间较短的组合样本数为14。并对优化组合样本数和抽样数的组合算法进行了链式组合称重定量试验验证。试验结果表明,在物料质量标准差≤30 g,允许定量组合误差为0.1 g时,优化后的组合算法与优化前遍历组合算法在定量组合成功概率总体上保持在95%左右,且优化后的算法组合计算时间减少了40%。研究结果可为非均质物料链式组合称重定量系统的研制提供参考。

Optimization and experiments of non-uniform objects quantitative combination algorithm based on chain transmission

Effective mass of non-uniform objects varies differently and indivisibly, thereby making the combined object uncertain during the combined weighing and quantification process. Thus, there is a great contradiction between the quantitative accuracy of combined weighing and combined speed. In this study, a chain-drive combination weighing and quantitative system was proposed, where the precision-efficiency was treated as the combined operation target. A combined error of influence parameters was analyzed to evaluate the accuracy of combined weighing quantification. Besides, the number of combined samples and sampling were optimized to reduce the combined calculation time with a high combined quantitative speed. The results indicated that the number of combined samples increased the probability of combination success. However, the amount of combined calculation increased exponentially with the increase of the number of combined samples. Thus, the number of combined samples needed to be optimized for the tradeoff between the combined error and calculation time. A normal distribution was followed after 10 000 rounds of combined calculation on the quality data, where the mean value was equal to 100, and the variance was 102. It was found that the number of combined samples was 14 with the shorter calculation time when the target mass of combined quantification was 500 g and the allowable combined error was ±0.1g. Moreover, the sampling numbers needed to be screened for the high requirements of combined error, due to the characteristics of non-uniform objects. In addition, a simulation experiment was designed to explore the influence of the total sampling number on the combined error and calculation time. The test results showed that the improved and previous combination maintained the success probability of quantitative combination at about 95% when the standard deviation of weight distribution was less than 30 g and the combined quantitative error was less than 0.1-1.0 g. The calculation time of the optimized combination was reduced by 40%, compared with the conventional one. The findings can provide a sound reference for the potential development of a chain combined weighing and quantitative system for non-uniform objects.