Presentation is loading. Please wait.

Presentation is loading. Please wait.

ALGORITHM FOR AN OPTIMAL SIZING OF DIFFERENT VARIANTS

Similar presentations


Presentation on theme: "ALGORITHM FOR AN OPTIMAL SIZING OF DIFFERENT VARIANTS"— Presentation transcript:

1 ALGORITHM FOR AN OPTIMAL SIZING OF DIFFERENT VARIANTS
IASI ALGORITHM FOR AN OPTIMAL SIZING OF DIFFERENT VARIANTS OF BLENDED WINES Erol MURAD, Edmond MAICAN, Mihaela DAVID - Politehnica University of Bucharest - 4-th ACME Conference

2 IASI Generalities The expert system – tested with different standards produced by means of blending traditional wine assortments, at research institutes in viticulture; Setting for the standard wine properties - carried out interactively by an oenologist expert, depending on the type and characteristics of the blended product; Checking the expert system in this stage – by comparison with real experiments made by partner institutions; The main objective of the verification stage – to find (among the variations determined by the expert system) a blended wine variant which approaches the initial experimental data;  2,5% - tolerance value accepted for all the properties of the blended product (the oenologist interaction with the expert sistem - not yet been completed); 4-th ACME Conference

3 CUPEXVIN – Blending Algorithm
Standard properties Computing Engine Blending components properties Validation criteria Set of solutions Sorting criteria Sort Selected recipe OENOLOGIST EXPERT OENOLOGIST EXPERT Dosing {2…3 batches} Blending Taste/evaluation Accept recipe Yes No Sorting criteria Yes STOP No 4-th ACME Conference

4 Simulated experiments to determine optimal blending solutions
IASI Simulated experiments to determine optimal blending solutions Number of random searches: 170; Valid solutions: 10; Selection criterion: minimum CO1 (optimization criterion); Check for seven types of blended products, covering white, red, sweet and dry wines. OPT2 – additional optimization criterion – current harvest weight, with coefficient K [1] 4-th ACME Conference

5 Optimal blending solutions
IASI Optimal blending solutions Blended product: wine from Dealu Mare/Valea Calugareasca, code SR-20 Blending recipe: Component 1: M 25%; Component 2: CS 25%; Component 3: FN 25%; Component 4: PN 25%; Poz. Kdoz[1] Kdoz [2] Kdoz [3] Kdoz [4] CO1 1 0.254 29.43 2 72.44 3 443.81 4 472.13 5 568.14 6 1337.6 7 1552.4 8 0.0495 1690.3 9 3501.8 10 15856 4-th ACME Conference

6 Optimal blending solutions
IASI Optimal blending solutions Blended product: wine from Dealu Mare/Valea Calugareasca, code SR-21 Blending recipe: Comp.1: BM 20%; Comp.2: M 20%; Comp.3: CS 20%; Comp.4: FN 20%; Comp.5: PN 20%; Poz. Kdoz[1] Kdoz [2] Kdoz [3] Kdoz [4] Kdoz [5] CO1 1 0.1754 112.97 2 202.94 3 0.2067 0.1707 289.74 4 489.51 5 587.72 6 856.13 7 1157.1 8 9 10 0.1058 4-th ACME Conference

7 Optimal blending solutions
IASI Optimal blending solutions Blended product: wine from Dealu Mare/Valea Calugareasca, code SR-27 Blending recipe: Component 1: BM 10%; Component 2: M 30%; Component 3: FN 50%; Component 4: PN 10%; Poz. Kdoz[1] Kdoz [2] Kdoz [3] Kdoz [4] CO1 1 53.42 2 62.67 3 0.0985 254.16 4 5 6 0.4345 7 8 9 10 4-th ACME Conference

8 Optimal blending solutions
IASI Optimal blending solutions Blended product: wine from Dealu Bujor, code: SR-34 Blending recipe: Component 1: FN 10%; Component 2: M 60%; Component 3: BN 10%; Component 4: CS 20%; Poz. Kdoz[1] Kdoz [2] Kdoz [3] Kdoz [4] CO1 1 7.05 2 9.48 3 0.2057 138.27 4 166.1 5 166.11 6 614.18 7 0.2109 8 9 10 4-th ACME Conference

9 Optimal blending solutions
IASI Optimal blending solutions Blended product: wine from Drăgăşani, code: SA-39 Blending recipe: Component 1: RI 33%; Component 2: FR 33%; Component 3: CR 34%; Poz. Kdoz[1] Kdoz [2] Kdoz [3] CO1 1 0.62 2 3 4 5 17678 6 7 8 9 10 0.45 0.55 4-th ACME Conference

10 IASI Conclusions Imposed value of the tolerances for the standard wine properties ( 2.5%) - extremely restrictive (standard wine properties - different weights); Initially - 10 acceptable blended variants were determined (the dosing coefficients were calculated); The sorted positions of variants similar to the initial blending are not in accordance with the criterion CO1; ⇒ CO1 must be modified (it should also depend on the type of blended product: white, red, dry, sweet); In case no variant closest to the standard was determined, or if there are major deviations - the search area was widened to 30 allowable variants; Results that differ may be due to the fact that, in some actual blendings, volumetric determinations were used, leading to mismatches with the results from mass dosing (which can provide better blending recipes); The expert system finds a range of blending variants, whom are used as a starting point by the oenologist expert, who can choose the most feasible options to achieve the actual blending variant. 4-th ACME Conference

11 Thank you for your attention!
IASI Thank you for your attention! 4-th ACME Conference


Download ppt "ALGORITHM FOR AN OPTIMAL SIZING OF DIFFERENT VARIANTS"

Similar presentations


Ads by Google