Predictive Variables, Artificial Neural Networks and Discrete Event Simulation in Manufacturing

Rebecca Santos

Rebecca Santos ProModel Tech Support Engineer

During my Master’s program at BYU, I worked with a team to complete my thesis.  The work turned into a published article by the International Journal of Modelling and Simulation!  Since I work for ProModel, it’s only appropriate that I share it with my fellow simulation enthusiasts.  Please let me know what you think by commenting below.

Discrete event simulation (DES) is a powerful tool that can help users make better decisions. Over the years tools such as ProModel and Process Simulator have been developed to simplify the application of simulation, decreasing the learning curve and increasing its use. One of the advantages of simulation is that it is able to create lots of data with valuable information. However, the data analysis process can be challenging and relevant information may not be fully analyzed.

Observing the opportunity to more fully learn from the data, we decided to use data mining algorithms to help in the data analysis process, and more than that, to guide the modeler to the variables that most impact the outcome of the system being modeled.

The data mining algorithm picked was Artificial Neural Networks (ANNs) which has been good at learning from the data and making accurate predictions, according to the scientific literature. We applied ANNs to the data generated by the simulation model and we were able to create ANN models that could predict the simulation model results. The ANN models created were then interpreted and through the interpretation it was possible to rank variables according to their impact on the output results. This makes it possible for decision makers to know how to prioritize and where to place their investments.

Here is the abstract from the article:

This research used a discrete event simulation to create data on a shipment receiving process instead of using historical records on the process. The simulation was used to create records with different inputs and operating conditions and the resulting overall elapsed time for the overall process. The resulting records were used to create a set of predictive artificial neural network models that predicted elapsed time based on the process characteristics. Then, the connection weight approach was used to determine the relative importance of the input variables. The connection weight approach was applied in three different steps:

(1) On all input variables to identify predictive and non-predictive inputs

(2) On all predictive inputs and

(3) After removal of a dominating predictive input.

This produced a clearer picture of the relative importance of input variables on the outcome variable than applying the connection weight approach once.

You are welcome to access the published article at the journal’s website or the full original paper at the ProModel website.

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s