Does Darwin’s theory of evolution mean anything to you? Arima has drawn inspiration from it to design and develop a new way to optimize production scheduling … by simulating the process of natural selection.
From biology to artificial intelligence
A population of individuals will evolve from one generation to the next. When two individuals breed, their genes are intermixed and repackaged into new chromosomes, with some genes changing through mutation. If the offspring survive and reproduce, this will be because their genetic characteristics make them well adapted to their environment. The cycle is repeated, and their surviving offspring will be even better adapted.
So in biology a genetic algorithm is an iterative process that improves the quality of an initial range of solutions through recombination, mutation, and selection to form, with each new generation, a better range of solutions.
Syncrun applies this principle to its model of the production process. The “population” is all of the possible solutions for production optimization. The “gene” is the way a resource is assigned to an operation according to a specific production schedule.
This evolutionary artificial intelligence technique has an advantage over conventional optimization approaches: the solution comes with a comprehensive view of the relevant costs.
Components of the genetic model
An initial population of chromosomes/solutions is generated in line with basic scheduling rules. It then goes through a double process of selection: selection for reproduction and selection for survival—in other words, a chromosome/solution is selected if it survives into the next generation.
The algorithm breaks up the chromosomes and creates new ones by recombining the fragments. There are also mutations where two groups of genes swap their positions on the same chromosome. May the best team win!
In practice, Syncrun’s genetic algorithm will aim for a production goal (produce on time, and at the lowest cost) by simulating hundreds of thousands of possible ways to reach that goal. Finally, it comes up with the most efficient solution.
With this pathbreaking approach, we can optimize a manufacturer’s complete planning horizon and cut all of the production costs for inventory maintenance, for setup time, for production time, and for use of available resources.
Contact us to learn more about Syncrun’s genetic algorithm and about the tangible benefits it will very quickly generate!