A neural network approach to the selection of feed mix in the feed industry

Due to frequent changes of feed mix, the anticipation of pellet quality becomes a cumbersome task for a mill. This paper suggests that the artificial neural network can be used to predict the production rate and percentage of dust for a particular mill. Applying the suggested method, the potentially costly test production that otherwise required is prevented. Four models based on process parameters (i.e., feeder speed, die size, die thickness and press motor amperage), the proportions of nutrient contents (i.e., protein, fat, fibre, and ash) and the type and proportions of raw materials included in each formula (i.e., corn, rice bran, and soybean) were applied. The data used in this study are the actual data collected from a feed mill in Thailand. The assumption of the first model is that the process parameters and the proportions of nutrient contents in a formula affect the production efficiency and pellet quality. The assumption of the second model is that the process parameters and the proportions of nutrient contents of each main raw material included in a formula affect the production efficiency and pellet quality. The assumption of the third model is that the process parameters and the proportions of main raw materials included in the formula affect the production efficiency and pellet quality. The assumption underlying the fourth model is that only the proportions of main raw materials included in each formula affect the production efficiency and pellet quality. The results show that the process parameters and the percent inclusion of main raw materials (Model 3) give a more accurate prediction of pelleting rate. For the prediction of the dust level, Model 1 gives the least accurate results, while the results from the other models are not significantly different. The ability to predict the production rate and dust level enhances the mill's capability to incorporate the cost of production during the feed formulation process. Hence, the mill will be able to consider both production and raw material costs simultaneously when deciding on the most economical formula for producing feed.

[1]  Louis Sanzogni,et al.  Milk Production estimates using feed forward artificial neural networks , 2001 .

[2]  A.F.B. van der Poel,et al.  Physical quality of pelleted animal feed 3. Contribution of feedstuff components , 1998 .

[3]  Jonathan Seville,et al.  Using intelligent software to predict the effects of formulation and processing parameters on roller compaction , 2008 .

[4]  Peter R. Tozer,et al.  A multi-objective programming approach to feed ration balancing and nutrient management , 2001 .

[5]  A.F.B. van der Poel,et al.  Physical quality of pelleted animal feed 1. Criteria for pellet quality , 1996 .

[6]  F. Dubeau,et al.  Reducing nitrogen excretion in pigs by modifying the traditional least-cost formulation algorithm , 2001 .

[7]  A. E. Chappell.  Linear Programming Cuts Costs in Production of Animal Feeds , 1974 .

[8]  A.F.B. van der Poel,et al.  Physical quality of pelleted animal feed. 2. contribution of processes and its conditions , 1997 .

[9]  C. Panne,et al.  Minimum-Cost Cattle Feed Under Probabilistic Protein Constraints , 1963 .

[10]  Pablo Lara,et al.  Multicriteria fractional model for feed formulation: economic, nutritional and environmental criteria , 2005 .

[11]  Graeme M. Mohr.  The Bulk Constraint and Computer Formulations of Least-Cost Feed Mixes , 1972 .

[12]  Georges Brigham,et al.  Letter to the Editor—A Classroom Example of Linear Programming , 1959 .

[13]  M. Ngadi,et al.  Predicting mechanical properties of fried chicken nuggets using image processing and neural network techniques , 2007 .

[14]  S. Prasher,et al.  Artificial neural networks to predict corn yield from Compact Airborne Spectrographic Imager data , 2005 .

[15]  Lyndon Smith,et al.  A neural network approach for solution of the inverse problem for selection of powder metallurgy materials , 2002 .

[16]  Joze Balic,et al.  Neural network based manufacturability evaluation of free form machining , 2005 .

[17]  H. C. De Kock,et al.  Multi-Mix Feedstock Problems on Microcomputers , 1987 .

[18]  Lyndon N. Smith,et al.  A neural network approach for selection of powder metallurgy materials and process parameters , 2000, Artif. Intell. Eng..

[19]  Rajesh Ransing,et al.  Application of neural computing in basic oxygen steelmaking , 2002 .

[20]  Jesús M. Zamarreño,et al.  Prediction of hourly energy consumption in buildings based on a feedback artificial neural network , 2005 .

[21]  François Dubeau,et al.  Reducing phosphorus concentration in pig diets by adding an environmental objective to the traditional feed formulation algorithm , 2007 .

[22]  José S. Torrecilla,et al.  Optimization of an artificial neural network for thermal/pressure food processing: Evaluation of training algorithms , 2007 .

[23]  R. B. Schüller,et al.  Physical and nutritional effects of pelleting of broiler chicken diets made from wheat ground to different coarsenesses by the use of roller mill and hammer mill , 2004 .

[24]  A. M. Frattini Fileti,et al.  Neural modeling helps the BOS process to achieve aimed end-point conditions in liquid steel , 2006, Eng. Appl. Artif. Intell..