Optimization Of Magnesite Process For Magnesium Carbonate Preparation Based On BP Neural Network

Magnesium carbonate trihydrate（Mg CO₃·3H₂O）whiskers and anhydrous magnesium carbonate（Mg CO₃）crystals,as new magnesium chemical materials,have a wide range of applications in plastics,medicine,fire retardant and metallurgy due to their excellent mechanical properties and flame retardant performance.In this paper,a method to optimize the preparation process of magnesium carbonate was investigated using magnesite as raw material.The suitable process conditions and mechanism of magnesium hydroxide prepared by acid leaching method were analyzed.An orthogonal test was designed for magnesium carbonate trihydrate and anhydrous magnesium carbonate,and the process parameters were selected.BP neural network was established for the preparation of magnesium carbonate trihydrate and anhydrous magnesium carbonate to predict the product size under different combinations of process parameters,and the combinations of process parameters for the maximum aspect ratio magnesium carbonate trihydrate and the minimum particle size magnesium carbonate anhydrous were obtained.The study on the preparation of magnesium hydroxide by acid leaching of lightly burned magnesium powder with citric acid,the results showed that the highest leaching rate was80.60%when the concentration of citric acid was 7%,the solid-liquid ratio was 1:40,the acid leaching temperature was 70℃and the acid leaching time was 60min;the magnesium citrate solution was used as the magnesium source,Na OH as the precipitant,Mg Cl₂as the additive,polyethylene glycol and sodium dodecylbenzene sulfonate as the compound dispersant,the reaction temperature was 70℃and the reaction time was 60min,the particle size of the amorphous magnesium hydroxide was the smallest,which was 845nm.Optimization of magnesium carbonate trihydrate process based on BP neural network.The results showed that the optimized experimental conditions by orthogonal test were 80min reaction time,50℃reaction temperature,800r/min stirring rate,p H 7±0.1,Mg Cl₂dosage0.5g/L,and the maximum aspect ratio of magnesium carbonate trihydrate was 19.95.The BP neural network was established by combining the orthogonal test data to predict the length-to-diameter ratio under different values.The average error of the validation group was only 7.5%,and the network performance was good.Under the optimized experimental conditions,the magnesium carbonate trihydrate crystals were in the form of one-dimensional rods,and the aspect ratio reached 24.57,compared with 19.95 before optimization,which was improved by 23.2%.Optimization study of anhydrous magnesium carbonate process based on BP neural network.Using magnesium carbonate trihydrate as the magnesium source,anhydrous magnesium carbonate with particle size of 7.03μm can be obtained by using orthogonal test with optimized conditions of Mg Cl₂dosage of 10g/L,reaction time of 11h,reaction temperature of 180℃and solid-liquid ratio of 1:30.BP neural network was established to predict the particle size under different process taking values.It was verified that the average error was about 3%and the network had good prediction performance.The rhombus-shaped anhydrous magnesium carbonate was prepared under the optimized process parameters with a particle size of 5.09μm,which was reduced by 27.6%compared with the orthogonal test result of 7.03μm.The orthogonal test was designed to optimize the process with bicarbonate water as raw material,the amount of Na HCO₃was 14g/L,the reaction time was 16h,the reaction temperature was 180℃,the concentration of bicarbonate water was 0.3mol/L,and the particle size of anhydrous magnesium carbonate reached the minimum value of 13.9μm.BP neural network was established,and the average error of the network validation group was 3.3%,and the network performance was good,which can be used for comprehensive optimization of process parameters.The network performance is good and can be used to optimize the process parameters.After optimization,the crystal morphology was rhombic block,and the particle size reached 12.06μm,which was reduced by 13.2%compared with the orthogonal test result of 13.9μm.