论文部分内容阅读
烧结矿质量是烧结厂生产控制的核心指标,现场一般根据烧结矿检化验的结果对过程工艺参数进行调整,但这种调整方法存在严重的滞后性。本文基于烧结矿质量预测模型与多元统计理论,建立了烧结过程性能监控系统。预测模型以烧结矿Fe O含量与转鼓强度为指标,采用BP神经网络构建,多元统计分析采用质量控制图的方法辨识质量异常的数据点,随后采用故障贡献图的方法分析其主因。所开发的监控系统在现场的运行情况表明:在原料稳定的情况下,烧结矿质量预测值与实际检测值相差5%以内的比例超过95%;质量控制图与故障贡献图能较好地发现质量异常点并分析引起异常的主要因素。系统的使用有助于减少烧结矿质量异常的情况,实现合理稳定的烧结生产。
The sinter quality is the core index of sinter plant production control. The site generally adjusts the process parameters according to the result of sinter ore inspection, but there is a serious lag in this adjustment method. Based on the sinter quality prediction model and multivariate statistical theory, this paper established a sintering process performance monitoring system. The prediction model is based on the index of FeO content and drum strength in sinter, and is constructed by BP neural network. Multivariate statistical analysis uses the method of quality control chart to identify the data points of abnormal quality, and then uses the method of fault contribution diagram to analyze the main cause. The operation condition of the monitoring system developed in the field shows that the proportion of the predicted value of sinter quality within 5% of the actual measured value is more than 95% under the condition of stable raw materials; the quality control chart and the fault contribution diagram can be well found Abnormal mass points and analyze the main factors that cause abnormalities. The use of the system helps to reduce the abnormal quality of sinter and achieve reasonable and stable sintering production.