基于神经网络的产品质量预测方法

Product Quality Prediction Method Based on Neural Network

绿色可持续发展政策下,为了节能减排,节约不可再生的矿物资源以及加工所需的能源,需要提高矿石加工质量。文章主要分析在矿石加工过程中,保持其他环境变量不变,矿石质量和温度控制对矿石产品质量的影响,并进行产品质量指标的预测,以及给定指标,对温度进行设定,最后进行模型推广。对于文中的问题1,选择通过给定的生产加工数据,建立相应数学模型以研究系统温度对产品质量的影响,进而给出用系统温度预测产品质量的方法。首先建立一个系统温度之间的数学模型Ⅰ,以观测两者可能存在的潜在关系;其次建立一个系统温度与指标的模型Ⅱ,对结果进行预测,同时用神经网络预测法进行预测并将结果进行比较;最后得出指标数值。对于文中的问题2,首先在建立模型前用多元线性回归分析,再用神经网络预测,得到原数据的预测值,通过比对原数据与预测数据得到预测方法的准确性;再由所确定的预测方法得到系统所需参数,得出系统设定的温度。

Under the policy of green and sustainable development,in order to save energy and emissions,non renewable mineral resources,and energy required for processing,it is necessary to improve the quality of ore processing.The article mainly analyzes the impact of ore quality and temperature control on the quality of ore products while keeping other environmental variables constant during the ore processing process,predicts product quality indicators,sets temperature for given indicators,and finally promotes the model.For question 1 in the article,choose to establish a corresponding mathematical model based on the given production and processing data to study the impact of system temperature on product quality,and then provide a method for predicting product quality using system temperature.Firstly,establish a mathematical model Ⅰ between system temperatures to observe the potential relationship between the two.Secondly,establish a model Ⅱ between system temperature and indicators to predict the results.At the same time,use neural network prediction method to predict and compare the results.The final indicator value is obtained.For question 2 in the article,first use multiple linear regression analysis before establishing the model,and then use neural network prediction to obtain the predicted value of the original data.By comparing the original data with the predicted data,the accuracy of the prediction method is obtained.Then,the required parameters of the system are obtained from the determined prediction method,and the set temperature of the system is obtained.