Research And Development Of Software For Compensation Of Caustic Soda Concentration In Alumina Evaporation Process Based On Industrial Big Data

The evaporation process is one of the key processes in alumina production process,and its role is to evaporate the excess water in spent liquor through multi-effect falling film evaporator and multi-stage flash evaporator.The operating index of the evaporation process is the caustic soda concentration of discharge.The goal of the operation control of the evaporation process is to control the concentration of caustic soda at the discharge port within the target range specified by the process.The current caustic soda concentration of alumina production enterprises adopts offline assay method.The offline assay value is seriously lagging,and it is difficult to adjust the flow setting value of the control circuit in time.For this reason,the current development trend is to use a measuring instrument based on the principle of refracting and temperature online fitting of caustic soda concentration,but this instrument requires research and development personnel to frequently calibrate the instrument according to the production process,poor measurement accuracy,and it is difficult to apply to the evaporation process.Therefore,it is of great significance to study the accuracy compensation of caustic soda concentration measurement.In view of the above problems,this paper based on the actual project of an alumina plant in Shanxi Province,carried out the research on the accuracy compensation algorithm and software of caustic soda concentration measurement in the evaporation process.The main work of this paper is as follows:(1)In view of the fluctuation of raw ore composition(aluminum silicon ratio,iron oxide,calcium oxide,etc.)and the dynamic change of the production process,which lead to the low accuracy of caustic soda concentration measurement,a caustic soda concentration measurement accuracy compensation algorithm based on long short-term memory network(LSTM)is proposed.The depth learning model of caustic soda concentration measurement accuracy compensation is established by taking the instrument values of alkali refractivity,temperature and caustic soda concentration as the model input and the error between the instrument value of caustic soda concentration and the test value as the model output.The sum of the output value of the compensation model and the instrument value of caustic soda concentration is the prediction value of the precision compensation algorithm of caustic soda concentration measurement.Using the big data of the actual industrial process to train the parameters of the caustic soda concentration measurement precision compensation model,and finally determine the two-layer long short-term memory network to meet the requirements of the caustic soda concentration measurement precision.(2)With the proposed algorithm,an on-site measurement accuracy compensation software for caustic soda concentration is researched and developed.The hardware platform of the software consists of distributed control system(DCS),virtual machines in cloud servers,process historical database(PHD),computers of engineer station and operator station.The software consists of system communication module,data acquisition and storage module,data preprocessing module,precision compensation algorithm module and human-computer interaction module.The human-computer interaction module has the functions of data monitoring,index monitoring and model precision monitoring.The software uses python language to develop the functional modules,and foxdraw to develop the precision compensation interface,which realizes the online real-time monitoring of caustic soda concentration.(3)The experimental verification is carried out with the actual industrial process data.The verification results show that the developed software for measuring accuracy compensation of caustic soda concentration runs stably,and the measurement accuracy after compensation is 43.3%higher than that of the original instrument for measuring caustic soda concentration,which shows that the developed software for accuracy compensation algorithm has practical application value for the production of alumina evaporation process.