The control system of a smart pig porridge feeder was designed based on MCU whose kernel was the ARM CortexM3. The control system could well solve the problems of pig porridge feeder which include both the weight proportion between water and dry feeding, and the dry feeding weight delivered for every day. The ways of dynamical adjustment for the weight proportion of water and dry feed were shown as follows. Firstly, the least square support vector regression (LS-SVR) model between the delivered speed of the dry feeding and the motor’s rotary speed was founded. Then referred to the LS-SVR model, the abnormal realtime sampling weight value during continuous delivering feed was removed by using some data process methods, such as the threshold value judgment and the Grubbs criterion. At last, the value of feeding at certain times was predicted via the line regression model attained by least squares fitting the rest of sampling data. Because the water flow could be truly detected by a flow sensor, the weight proportion of water and dry feed was well adjusted. Using recursive weight compensation combined the LS-SVR model and static measurement means, the delivered dry feeding weight for one day was accurately controlled by the control system. After being mixed with stirred fully by the feeder using the control system, the water and the dry feed were turned to porridge whose nutrients should be better absorbed by pigs. The weight proportion of water and dry feed error was below 4%. The weight error of feeding for a pig per day was less than 1g.