The Studies About Grinding Steel Ball Loss Based On Artificial Neural Network Model

Grinding process is an important working procedure in the field of metallurgy,construction,cement,materials and other industrial production,especially in the mineral processing sector,it is a very important part in the process of mineral processing.Ball mill is the main equipment of the grinding operations,the efficiency of its work has a direct impact on the selection of the indicators of the product.At the same time,the weight and the ratio of mill ball load is reasonable or not is of grinding ore operations are important factors,steel ball of the reasonable loading is mainly dependent on the research in the field of mill ball loss.Therefore,the study of grinding loss model has a great significance.In foreign countries,there have been many developed countries,especially those in the mineral resources of the same rich countries,in the field of mineral processing industry,has achieved a large and medium-sized concentrator processing automation control of the whole process.In China,although in practice the mineral processing field,researchers working on the study summed up the ball in a lot of achievements and experience,and study the movement of grinding media,mill power consumption,and other aspects of the mathematical model grinding mill made in some progress,but the study of ball mill loss problem is still not resolved to draw definitive theoretical guidance policy.In order to improve the production efficiency and reduce the energy consumption and the steel consumption of the grinding process,it is necessary to maintain the stability of the ratio of steel balls in the ball mill.Therefore,in order to make grinding balls to keep the ball within the balance loss,while providing a better basis for people to study the movement of a ball mill media,the need for a reasonable and workable ball loss mathematical model.However,due to the complexity of the ball movement of the ball mill,and the movement of the ball trajectory is also random,so in other words,there is no mathematical model can accurately describe the motion of the steel ball.The artificial neural network has a great advantage in the application of this system.Practice has proved that the artificial neural network and computer technology will become the mainstream and trend in this field,and also bring us good economic benefits and social benefits.In this paper,based on the summary of the research on the mill ball losses at home and abroad,taking into account the working characteristics of the grinding operation,a mathematical model of the relationship between grain size and steel ball losses is established.The paper used of BP neural network and based on genetic algorithm optimization algorithm in two ways of loss of steel ball and filling rate of the prediction model.The experimental results show that the,the BP artificial neural network modeling has strong fitting,more accurate prediction of ore in the process of steel ball wear rule of grinding.In general,we need to spend a lot of time and energy to produce the scene to collect the data after the model is established.This paper aimed to establish a better model,or to optimize the neural network algorithm,so that we can reduce the amount of data,while achieving good simulation results.This paper proposed the establishment of mathematical model within the ore particle size and ball mill grinding process.We have known that grinding system have some characteristics like interference factors,and strong time-varying,like that we start directly from the product that ore particle size after grinding to establish the relational model.In this paper,two mathematical models of mill grinding loss based on BP neural network are presented,one is the mathematical model that the relation of ratio of the ball mill and product size,The second of relationship mathematical models is that the quality of the ball mill and ore particle size.The first model established the BP neural network model that the product of each fraction ratio as input and the proportion of each ball diameter of steel ball in the mill value as output.The second BP neural network model was that it is selected the grinding quality of product before and after each graded as input parameters,and each quality of steel ball of the ball diameter in the mill as the output of the neural network.Finally,the simulation results has been analyzed,The first model's quality of error was about 150g;the second model to meet 0?k <0.05 the total sample of test samples has been 80%,indicating that the experimental model had a certain stability in the laboratory mill and hasbeen achieved better result.Then our work is adding more model neural network algorithms and Contacting the mine production,provide theoretical guidance for information collection and process automation mill balls loss.