| id |
ndltd-OhioLink-oai-etd.ohiolink.edu-osu1155661130
|
| record_format |
oai_dc
|
| spelling |
ndltd-OhioLink-oai-etd.ohiolink.edu-osu11556611302021-08-03T05:51:19Z Stochastic control of animal eiets: optimal sampling schedule and diet optimization Cobanov, Branislav We hypothesized that the application of stochastic process control and stochastic optimization would result in significant improvements to diet optimization. The first part addresses the problem of monitoring the composition of forages as they are removed from their storage facilities. This forage variation is of three types: analytical, white-noise, and large threshold change (economically most significant). A TQC model was developed based on renewal reward process theory, accounting for the costs while the process is in-control (sampling costs, white-noise costs, false alarms costs) and out-of-control (sampling costs, lost production costs, assignable cause investigation costs, process repair costs). The model uses 13 input variables and optimizes for 3 sampling design parameters: number of samples, sampling frequency, and X-bar control limits. Results showed that the current practice of taking one sample on a monthly basis and to intervene when the composition mean > 2 standard deviations is far from optimal. 3 input variables accounted for 87.6% of the total sensitivity index for TQC. These were: the size of the herd, the milk production loss when forage composition changes, and the mean time the process is in-control. Three assumptions were made and tested for robustness during the derivation of the TQC: (1) no outliers, (2) a symmetric distribution, and (3) an abrupt change when the process goes out-of-control. Model proved to be robust: the TQC change was <1% when outliers were present, 1.4% when log-normal distribution was used, and the TQC was slightly to significantly lower in the presence of gradual change. Current diet formulation models assume perfect certainty in a chemical composition of feeds used. However, feeds do vary, and their nutritional compositions are always somewhat uncertain. Thus, an accurate model should factor nutrient variance when optimized. Monte Carlo techniques, using a pseudo-random number generator for a multivariate normal distribution, were used to numerically solve for the estimates of total diet means and variances of nutrients that are themselves nonlinear functions of random multivariate normal nutrients. The economic benefits of stochastic programming appear substantial: $0.17•cow-1•d-1 in increased milk revenues and $0.35•cow-1•d-1 in lower feed cost, combining into $0.52•cow-1•d-1 in additional income-over-feed-costs. 2006-09-14 English text The Ohio State University / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=osu1155661130 http://rave.ohiolink.edu/etdc/view?acc_num=osu1155661130 unrestricted This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws.
|
| collection |
NDLTD
|
| language |
English
|
| sources |
NDLTD
|
| author |
Cobanov, Branislav
|
| spellingShingle |
Cobanov, Branislav
Stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| author_facet |
Cobanov, Branislav
|
| author_sort |
Cobanov, Branislav
|
| title |
Stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| title_short |
Stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| title_full |
Stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| title_fullStr |
Stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| title_full_unstemmed |
Stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| title_sort |
stochastic control of animal eiets: optimal sampling schedule and diet optimization
|
| publisher |
The Ohio State University / OhioLINK
|
| publishDate |
2006
|
| url |
http://rave.ohiolink.edu/etdc/view?acc_num=osu1155661130
|
| work_keys_str_mv |
AT cobanovbranislav stochasticcontrolofanimaleietsoptimalsamplingscheduleanddietoptimization
|
| _version_ |
1719426643571769344
|
