Expert system for selecting speed reduction components for a power transmission

This paper describes the development of an expert system that will allow for the concurrent consideration and analysis of the many facets of product constraints, particularly function and manufacturability, at the early stages of power transmission development. The Power Transmission Design Assistant methodology is based on the concept of directed refinement, where the product and analysis methodology become less abstract as more information is given. This allows for analysis and guidance throughout the product development process. The methodology has been demonstrated on a small-scale, semi-automatic tool. The tool was the foundation for a larger tool that incorporates expert rules to help optimize power transmission component design.
Power transmission systems are a crucial component in many types of machinery. Mechanical power transmissions are used in products such as food mixers, automobiles, and aircraft. The power transmission design problem involves deciding the input, intermediate components, and their layout for a required motion, force, or torque output. With the increased popularity of computer-based design tools, many difficult engineering problems, such as power transmission design, can be eased, which allows engineers to spend more time optimizing the product.

A partial listing of the various components involved in power transmission is shown in Figure 1. Power transmission can be achieved by either electrical, mechanical, or hydraulic means; however, most transmission systems are designed for mechanical, rotary applications. Power transmission of rotary systems typically consists of components such as shafts, bearings, gears, pulleys, sprockets, chains, belts, connectors, and fasteners. These elements can be prepackaged to meet a specific design goal, or they may be individual elements that the designer selects to meet a need. In either case, each has its own design criteria and constraints.

In the past, optimization of power transmission systems involved achieving the following interdependent functional requirements in an iterative, trial-and-error approach:

Geometry-Are there space limitations between elements, especially moving and stationary components? What is the relationship between input and output locations?

Kinematics-What is the direction of motion of the input and output transmission components? Is it at constant speed or is it transient?

Kinetics-What are the forces and torques present in the system?

Speeds-What are the required input and output speeds and forces?

Design adequacy-Once the loads and speeds are known, are the stresses, vibrations, and weights within acceptable limits?

However, after functional optimization occurs, the product is often still suboptimal in terms of cost, especially when life cycle costs are considered. The main reason for the suboptimal result is that the trade-offs between the requirements are often impossible to consider manually. It is necessary to develop a set of design tools to tackle this complex problem and integrate the many areas of the life cycle cost into a set of optimal solutions.

The iterative nature of the product development process implies that a considerable amount of time is spent repeating common but sometimes difficult tasks. If, as in power transmissions, there are a substantial number of functional constraints, these tasks become even more difficult. The process of finding optimal solutions to design problems is usually difficult and is often abandoned in favor of a quick solution that works but may not be optimal. The difficulty may simply be a result of computational ability to solve for the optimal solution or lack of a metric for deciding the optimal solution.

Engineers must be able to make correct engineering decisions early in the product development process to maintain profitability and competitiveness. The consequences of poor decision making include increased development time, increased time to market, pressure from internal groups who depend on the design decisions, and increased change order costs. Decisions or changes made later in the product development are more costly because they have greater repercussions. It is therefore advantageous to make decisions as early in the product's design as possible. The major hindrance to achieving this is the lack of quantitative knowledge on which to base these decisions.

In recent years, computer-aided engineering (CAE) has been used to relieve some of the pressures on design decisions. CAE is used to size components, evaluate products, and perform low-level design tasks. CAE has mostly been used to automate complicated and repetitive mathematical tasks; however, automation of these mathematical tasks usually requires information about product parameters that is not available early in the product development process. Currently, such design automation software falls into one of several categories: CAD-based design for manufacturability/design for assembly (DFM/DFA)1,2