The Research And Application Of The Intelligent Control Of Flotation Column

Flotation column with its highly selective separation develop rapidly.At present,the flotation column is mostly separated by flotation workers according to their experience,there are some problems that adjustment is not timely,quality fluctuation,high labor intensity of workers.So the realization of automatic control for the flotation column is important to play the advantage of equipment of flotation column,stabilize and improve product quality in coal flotation index number.At present,the domestic research on automation of flotation column is mostly in a single variable input/output stability control stage,the effect of some variables of flotation process was determined by using PID controller to control,however,the flotation process is multi variable,large time delay,nonlinear,so single input/single output control method can not achieve the optimization of the control of the flotation process.This paper,based on sorting needle of flotation column process characteristics,is proposed a new ideas of flotation column intelligent control which use artificial neural network technology to build an multi input/multi output variable control structure.Stability control is a prerequisite to ensure normal operation of flotation column,also it is the basis of optimal control,the design of column flotation stability control system,mainly including:rinse water automatic control,automatic control of flotation feed concentration,foam layer thickness(liquid level)automatic control,automatic pressure control based on circulating pulp,chemical dry the amount of slime to add automatic control,stability control to ensure the stable operation of flotation column can also create the conditions for optimal control.The research indicated that,the adjustment variables mainly has two:foam layer thickness(level)and reagent regime,the two work together to ensure the flotation effect is best,the other variables are in constant value control,adjust the frequency is low,so the paper based on flotation column stability control,mainly control the foam layer thickness and reagent system optimization.The foam layer thickness is controlled by fuzzy control method,the flotation feed variation and the feeding quantity change rate as input variables of fuzzy language,the foam layer thickness set values for the fuzzy output variable,with reasonable fuzzy rules,automatically adjust the online implementation of foam layer thickness set value.Flotation reagent adding is controlled with the conbination dry coal quantity of medicament add method with the expert system to realize online automatic adjustment of the coal quality fluctuation.From the control theory,the foam layer thickness automatic control and automatic control of adding medicament are two independent control of process and structure,so learn from the theory,technology and method of variable structure control and intelligent control of the flotation column,variable structure intelligent control strategy is proposed,the basic way of thinking is the organization and coordination of foam layer thickness controller and agent add controller,based on the switching real-time condition two controllers are implemented,in order to maximize the realization of optimal operation of flotation column.This paper designed a variable structure algorithm of multi input/multi output based on artificial neural network,concrete ideas:first,a pattern recognition system based on BP neural network,to obtain floating concentration,floating flow,pressure of circulating pulp,froth flotation column layer thickness of four sensor variables through the sensor,while the introduction of foaming agent and trap collecting agent dosage,as artificial neural network with 6 inputs,output layer is defined as 1,its output is defined respectively as "0" and "1",automatic control two controller structures representing the foam layer thickness automatic control and reagent addition,through the training of artificial neural network for data acquisition,training results:when the numbers of hidden layer,hidden layer structure of 13*1,the number of function training is the least,convergence speed is the fastest,and mean square error(MSE)for a minimum is 7.19e-06,the ability to identify system model of the strongest,fastest speed of recognition,the system stability is better.Finally,the 6*13*1*1 BP neural network structure is designed.Finally,an industrial test was conducted in the field,the test results show that:the system is running well,flotation column of each variable is stable and controllable,and the ash content in coal flotation product quality is stable,flotation concentrate yield increased,the volume of reagent consumption is significantly reduced,and good effect is achieved.