How to Study Probabilistic Methods?
Many engineers are familiar with solving governing equations using analytical or numerical methods. Most of engineering programs (including Excel) have been developed to render a deterministic (single) output from numerous input parameters.
Needless to say, the model input parameters will impact on the deterministic output value significantly from the governing equation, presented as a functional relationship. How can you check each significance of your input parameters in getting the reasonable output value? It does not matter whether governing equation uses theoretical approach or empirical approach, but it matters to know which input parameters predominantly impact on your model output determination.
Without this model sensitivity analysis or back-calculation analysis in other words, you cannot assure accuracy of model output especially if you don’t understand the governing system of equations quite well. The disadvantage of sensitivity analysis is narrow scope of investigation which possibly yields a local optimum value, governed by few input parameters not considering entire relationship of every single input parameter.
In lieu of using the sensitivity analysis, probability analysis can give you an idea how to estimate sensitivity of each input parameter to model output value by additional consideration of variance beside the mean or maximum likelihood value. I know many commercial software using FEA or FDA now adopt a probabilistic or stochastic model to show the data distribution of output values from assigning data distributions of input parameters.
Simply using a function of black box software cannot entitle calling you as an expert on the engineering problem. Now is right time to develop your own Excel tools for practical model examination and background knowledge on probabilistic approach which will be increasingly necessitated in future AI society. My probability online course will start in April.?https://lnkd.in/gfUUYxm
Geotechnical Engineer
1 年Thanks for sharing