Manufacturing Systems Get Smarter

A number of intelligent manufacturing systems were under the spotlight at the recent Innovative Applications of Artificial Intelligence (IAAI '95) show, held last month in Montreal. While the technology embedded in these systems is state-of-the-art, an even more impressive feature is the substantial return on investment that manufacturers have realized from AI-based applications. Generally, most companies report payback on the initial development investment in less than a year. Even more significantly, many report up to an order of magnitude improvement in productivity.

Following are capsule descriptions of some noteworthy systems presented at IAAI '95.

Automobile manufacturer General Motors' Metal Fabrication Division and Manufacturing Center (Warren, Mich.) has developed GMGear, a generative gear design system. This intelligent system generates multiple possible design solutions for a gear set, then evaluates and recommends the best design to meet performance requirements, geometric constraints and manufacturing considerations.

The system explores the solution space for multiple key parameters (more than 1,000 combinations for minimally constrained designs) simultaneously. It makes tradeoffs to balance performance while meeting constraints. The system provides a "glass box" view of the design methodology, design decisions and reasoning, value descriptions and sources, and descriptions of terms and equations.

The system speeds up the gear design process up to 95% from weeks to a day or less, with better quality designs; captures knowledge from across several GM divisions; creates better understanding of the gear design process; and promotes commonality of terminology and design practices throughout GM.
The Product Cost Estimator (PCE), developed by computer giant IBM (San Jose, Calif.), is an expert system for generating plans for new products. Before developing and marketing a new product, it is crucial to accurately estimate life cycle costs and revenues to determine anticipated profitability of the product. The rapid evolution of computer hardware means the fast pace of new product introductions, and a quick, detail business planning cycle.

To this end, IBM has developed and deployed the PCE, a shell for producing business plans and service cost estimates for new products. It enables the timely exploration of a variety of alternative strategies. PCE, deployed in the first half of 1994, has been used in several parts of the business.

Its payoff includes substantial time savings and improved accuracy. Where producing a business plan typically takes days to months at IBM, using PCE the company was able to do business planning with fewer people in approximately one-tenth the elapsed time, with the same accuracy as previous planning methods.
Pratt & Whitney Government Engine and Space Propulsion (West Palm Beach, Fla.), a manufacturer of gas turbine engines, has developed AGETS (Automated Ground Engine Test Set), a jet engine diagnostic system designed to isolate failures using an overall system troubleshooting approach. AGETS is used by the U.S. Air Force and Air National Guard to test, trim and diagnose problems with Pratt & Whitney jet engines. Sixty-six AGETS units exist worldwide, and have been in operation since 1985.

The benefits of AGETS are difficult to quantify given the low frequency of trouble calls and a decision to limit diagnosis capability to only high-level system modules. Efforts are underway, however, to relax the operational requirements, which will make payoff calculations in terms of financial and personnel savings easier to track.

Click here to return to Table of Contents for the Intelligent Manufacturing September issue.