J Intell Manuf (2008) 19:233–240 DOI 10.1007/s10845-008-0076-6 Application of artificial neural network for determination of standard time in machining Izabela Kutschenreiter-Praszkiewicz Received: 1 January 2006 / Accepted: 1 July 2007 / Published online: 18 January 2008 © Springer Science+Business Media, LLC 2008 Abstract The purpose of this article is to present the appli- cation of neural network for time per unit determination in small lot production in machining. A set of features consid- ered as input vector and time consumption in manufacturing process was presented and treated as output of the neural net. A neural network was used as a machining model. Sensitiv- ity analysis was made and proper topology of neural network was determined. Keywords Neural network · Standard time · Machining · Sensitivity analysis Introduction One of the main problems in the activity of enterprises is determination of time and cost consumption of a production order of a new or modernized product. A manager should know the time consumption of a new product processing exactly before acceptance. Analytical time consumption methods require time and information about machining parameters; therefore it’s necessary to develop a new fast method used in the first steps of offer preparation. It seems useful to use a modern method of data analysis such as a neural network in this area of enterprise activity. An engi- neer samples time consumption of machining based on his experience, as well as the output data which is also related with his experience. A disadvantage of this method is con- nected with the reappraisement of output data and overstated cost calculation, and finally lost or unprofitable contract. I. Kutschenreiter-Praszkiewicz (B ) Department of Industrial Engineering, University of Bielsko-Biala, ul. Willowa 2, Bielsko-Biala 43-300, Poland e-mail: ipraszkiewicz@ath.bielsko.pl Analysis of the existing methods of work measurement There are a number of different techniques for work measure- ment, which can be either direct, where actual observations are made by workers (time study, activity sampling) or indi- rect comprising synthetic timing, analytical estimating and predetermined motion time systems (Muhlemann et al. 1993). The most widely used of those is time study, which is a structured process of directly observing and measuring, using a timing device, human work in order to establish the time required for completion of the work by a qualified worker when working at a defined level of performance. The next method of work measurement is activity sam- pling. This method was defined by British Standard as a tech- nique in which a number of successive observations are made over a period time of one or a group of machines, processes or workers. Each observation records what is happening at that instant, and the percentage of time during which that activity or delay occurs. It is not possible to use direct methods of work measure- ment for standard time setting of a new product processing before offering acceptance. These methods are applicable after the start up of new product processing. Work content can be obtained before the task is actually carried out using indirect methods of work measurement. In the synthetic timing the appropriate elements are iden- tified and the time for the new work is established by adding the previously assessed times. Many organizations will build up books of tables of these times, entering new elements as they are discovered. In the analytical estimating method the time required for the task is built up of synthetic data where possible, but sup- plemented where such data is not available for particular ele- ments by estimations based on the best available knowledge and experience. 123