Metallurgical Acid SO ₂ Soft Measurement Of The Conversion Rate

Compared with pyrite acid and sulfur-based sulfuric acid, smelting off-gas acid have some special advantages:it recycles SO2of smelting off-gas acid in the production process, reduces the environmental pollution caused by SO2, at the same time can create good economic benefit. Metallurgical off-gases process is separated into four main parts that are clean, dry, conversion, absorption, and the conversion process is the core part. At present, there are two kinds of conversion technology,"one convert and one absorption" and "two convert and two absorption", while the "two convert and two absorption" is the world’s leading technology of acid. Conversion technology and the level of control have direct impact on the conversion rate of SO2. Therefore, SO2conversion rate and its stability has become an important indicator to measure the product quality and production of smelting off-gas acid.This paper aiming at the problem of SO2conversion rate at a smelter "two convert and two absorption" of smelting off-gas acid can’t be measured online directly, using the soft measurement technology to predict and research the conversion rate of SO2. That is base on deeply analysis of its technological process and the transformation principle, this paper selected instrumental variables of the soft-sensing of SO2conversion rate(including the entrance and exit temperature of five layers, SO2inlet concentration, SO2outlet concentration, as fan speed,13variables), processing the data that collected (according to the error processing criterion excluded data and normalization methods for data normalization), and using multiple linear regression and BP neural network to build SO2conversion rate model, and ultimately validates the soft measurement model. Verification results show that the BP neural network’s fitting and prediction result is better than multiple linear regression model, it has high prediction accuracy, and can be used for the SO2conversion ratio on-line prediction.However, in the actual production process, soft-sensing model can’t be static, when conditions changes, in order to obtain more accurate predictive value, the model need to corrected. A short-term correction method is used to the model in this paper, the method can improve the prediction accuracy of the model, and the model predicted value is more close to the measured values.In order to use the value of model in industrial production, the paper used the Siemens WinCC7.0design the interface of monitoring, at last, revised model is applied to the platform of WinCC7.0that industrial control platform, and mock implementation the fitting of SO2conversion rate online.