0

0

2659 days ago,
655 views

PowerPoint PPT Presentation
Outline of Experiments Lecture I

Review of Error Analysis Theory & Experimentation in Engineering Some Considerations in Planning Experiments Review of Statistical equations and hypothesis Begin Statistical Design of examinations ("DOE" or "DOX") Topics Today

Part 1: Review of Error Analysis Uncertainty or "irregular blunder" is natural in all estimations Statistical premise Unavoidable-look to gauge and consider Can minimize with better instruments, estimation methods, and so forth

Review of Error Analysis Systematic blunders (or "strategy mistakes") will be mix-ups in suspicions, strategies and so on. that prompt to non-irregular predisposition Careful exploratory arranging and execution can minimize Difficult to portray; can just take a gander at proof sometime later, investigate procedure to discover source and dispense with

Graphical Description of Random and Systematic Error

Why do we have to gauge vulnerability and incorporate into expressed test values? Likelihood of being off-base will impact handle and additionally budgetary choices Cost/advantage of tolerating result as "actuality"? What might be the impact downstream as the vulnerability proliferates through the procedure? At the point when looking at two values and figuring out whether they are diverse Overlap of instability? What is the likelihood that the distinction is huge ?

Stating Results +/ - Uncertainty Rule for Stating Uncertainties Experimental vulnerabilities ought to quite often be adjusted to one huge figure. Lead for Stating Answers The last huge figure in any expressed answer ought to for the most part be of a similar request of extent (in a similar decimal position) as the vulnerability. Express Uncertainty as mistake bars and certainty interim for graphical information and bend fits (relapses) individually

Determining Propagated Error: Non-factual Method Compute from aggregate differential

Propagated blunder OR Can do affectability examination in spreadsheet of other programming program Compute conceivable instability in ascertained result in view of shifting estimations of contributions as per the vulnerability of every info Example: Use "Solver" enhancement device in Excel to discover most extreme and least estimations of registered esteem in a cell by changing the estimation of every information cell Set imperative that the information values lie in the scope of instability of that esteem

mean standard deviation of every estimation standard deviation of the mean of the estimations Confidence interims on dependant variable Confidence interims on relapse parameters Or Can Use rehash estimations to gauge instability in an outcome utilizing likelihood and insights for arbitrary blunders :

Statistical Formulas from part 4 of Taylor

Relationship of standard deviation to certainty interims

Confidence interims on non-straight relapse coefficients Can be perplexing use programming however comprehend hypothesis of how figured for direct case

Error bars that speak to instability in the dependant variable

How estimations at a given x,y would be appropriated for various estimations

Determining Slope and Intercept In Linear Regression

Confidence interims (SD) on incline B and Intercept A

Regression Output in Excel Simple ANOVA-we will take a gander at more mind boggling cases in DOE Slope and catch Confidence limits (+/ - ) om slant & capture

Confidence Intervals in TableCurve

Confidence Intervals in TableCurve

Regression in Polymath

Statistical Process Control Very Widely Used for quality control and in conjunction with DOE for process change Control Charts give measurable confirmation That a procedure is carrying on ordinarily or if something incorrectly Serve as information yield (dependant variable )from process in outlined measurable investigations

Variation from expected conduct in control diagrams like relapse and point measurements C ontrol L imit is the mean of an all around carried on process yield (i.e. item) U pper and bring down C ontrol L imits speak as far as possible on mean of "all around carried on" process ouptut Expect arbitrary deviations frame mean simply like in relapse

Part 2: Theory and Experimentation in Engineering

Theory and Experimentation in Engineering Two principal ways to deal with critical thinking issues in the revelation of information: Theoretical (physical/numerical displaying) Experimental estimation ( Most regularly a mix is utilized )

Example of mix of hypothesis and experimentation to get semi-exact relationship

Theoretical Models Simplifying suspicions required General results Less offices normally required Can begin concentrate instantly Experimental approach Study "this present reality"- no disentangling presumptions required Results particular to device concentrated High exactness estimations require complex instruments Extensive lab offices perhaps required Time delays from building device, troubleshooting Features of option techniques

Functional Types of Engineering Experiments Determine material properties Determine segment or framework execution lists Evaluate/enhance hypothetical models Product/handle change by testing Exploratory experimentation Acceptance testing Teaching/learning through experimentation

Some imperative classes of Experiments Estimation of parameter mean esteem Estimate of parameter changeability Comparison of mean qualities Comparison of fluctuation Modeling the reliance of dependant Variable on a few quantitative or potentially subjective factors

Practical Experimental Planning Experimental outline: Consider objectives Consider what information can be gathered. Trouble of acquiring information What information is most critical What estimations can be overlooked Type of information: Categorical? Quantitative? Test to ensure that estimations/mechanical assembly are solid Collect information precisely and archive completely in ink utilizing bound journals. Make duplicates and keep independently

Preview of Uses for DOE Lab tests for research Industrial process tests

Four building issue classes to which DOE is connected in assembling 1. Examination 2. Screening/portrayal 3. Displaying 4. Streamlining

Comparison Compares to check whether an adjustment in a solitary "element" (variable) has brought about a procedure change (preferably a change)

Screening/Characterization Used when you need to see the impact of an entire scope of components to know which one(s) are generally imperative. Two factorial tests generally utilized

Modeling Used when you need to have the capacity to develop a scientific model that will foresee the impact on a procedure of controlling a factors or different factors

Optimization When you need to decide the ideal settings for all elements to give an ideal procedure reaction.

SPONSORS

No comments found.

SPONSORS

SPONSORS