VISUALIZATION-BASED EXPERT SYSTEM FOR EDUCATING CIVIL
ENGINEERING STUDENTS IN EARTHWORK ACTIVITIES AND EQUIPMENT

ABSTRACT

The use of expert systems which accumulate human experience can lead to the improvement of construction engineering and management. This study involves the development of an expert system program for determining the most suitable method and equipment to perform earthwork activities such as excavating and moving soil. To decide which method and equipment are the most suitable depends on many factors such as site conditions, hauling distance, and material types. Experientially based knowledge is vital to making such decisions. In this study, an expert system program is developed as a training tool to enhance the skill of foremen or young engineers who have little or no experience about equipment selection. Moreover, as a consulting tool, this program is developed in order to help experienced civil engineers to justify their selection. Further, with the visualization components that are integrated into the system, this knowledge is highly accessible to learners.

INTRODUCTION

Background and problems

The construction of the Maesung earth dam in Thailand (capacity 12 million cubic meters) is a project that requires the use of various types of equipment. This project consists of two main earth dams, a concrete spillway, and a steel pipe outlet for an irrigation system. The main structure of each dam is compacted cohesive soil with a mixed-aggregate drain system inside. Removal of loose soil layers and soft rock at the base of the dams was required before construction. During construction, many earth moving activities were delayed due to inappropriate use of equipment. In particular, this project dealt with various types of materials, terrain, and activities making it quite difficult for young engineers to select the optimal equipment. Although no accident occurred during the project implementation, the delay due to inappropriate equipment selection caused time and cost overruns.

Our experience in this project shows an urgent need for young engineers to be thoroughly trained to make appropriate equipment decisions. One approach to fill this need is to exploit knowledge on construction methods and equipment which is abundant but not readily accessible in the context of regular training sessions, books, or manuals. Such knowledge can be gained only from individuals who have years of on-site experience. However, much of this knowledge has been neglected or lost with the passing of individuals. Typically, if this specialized knowledge is retained, it is retained by only a select few.

Inexperienced engineers can use quantitative analysis as a tool to decide which type of equipment performs most economically. By determining cost per unit of material, which equals the cost per hour of equipment divided by its production per hour, they can easily decide which equipment performs the most economically. However, this analysis may result in impractical choices because it is sometimes impossible to exactly predict the equipment's production, which varies from site to site. Therefore, engineers usually make their selection based on their experience.

Expert system program

One approach to solve such problems is to develop a knowledge base that can be used as a training tool for improving the decision making skills of inexperienced engineers. To develop the knowledge base, experts' experience was gathered and implemented before it was entered into the knowledge base. By using the expert system shell, LEVEL5 OBJECT, the program is expected to be a user-friendly expert system.

Visualization component

In addition to the expert system, a visualization component helps inexperienced engineers to grasp complicated concepts and visualize a construction method without visiting construction sites, which might be dangerous for inexperienced engineers. Animation of various types of equipment was created by using the 3D Studio package, one of the most widely used animation tools, and then integrated into the knowledge base.

EQUIPMENT SELECTION

Introduction

Earthmoving equipment selection, one of the most important aspects in construction projects, has become more sophisticated. The selection depends on many considerations, such as material properties, excavation area and depth, and hauling distance. For example, experience shows that shovels can be used efficiently for excavating above ground level soils while backhoes are commonly used to excavate materials below the ground level. Scrapers and dozers are suitable for moving materials in a large area where shallow excavations are performed. Scrapers, on the other hand, can be used more easily than dozers for transporting and spreading materials; their use is more effective for a larger area of excavation.

Earthmoving equipment

There are various types of equipment for earthmoving activities and each piece of equipment has different capabilities. For example, dozers and scrapers are suitable for the task of moving broadly dispersed materials with a shallow depth, while using mass earthmoving equipment such as excavators to do this would not be recommended. Scrapers seem to be more popular than dozers in this task because scrapers can transport and spread material more easily than dozers. Loaders and shovels are suitable for moving material which is above machine level, like a stockpile. Wheel loaders are the most suitable equipment for stockpiling material because of their high production.

Track-mounted excavators are presently more popular than any other equipment because they are suitable for various kinds of tasks such as moving material which is above or below machine level or in a deep trench. Furthermore, their arm and bucket can be modified for various tasks. For instance, their arms can be enlarged for excavating deep trenches or shortened for excavating very hard materials. However, very deep trench clamshells are more suitable than excavators. The disadvantages of clamshells are their lower production and less precise position compared to excavators, thus they are not widely used. Draglines, like excavators, are a very versatile piece of equipment which can dig from above or below machine level. Draglines are suitable for stockpiling materials at concrete batching plants and digging underwater materials. However, because they drag materials by using cables, they are not suitable for precise position work.

EXPERT SYSTEM

Introduction

One of the most important recent developments in the rapidly advancing field of computer technology has been intelligent computer systems called 'Artificial Intelligence'(AI). AI is a computer system that imitates four human abilities: thinking, seeing, hearing, and speaking. An expert system is one type of AI that imitates a human's thinking and understanding by using a reasoning process to solve problems. To develop an expert system, knowledge engineers create a knowledge base that contains knowledge and expertise of the experts.

Knowledge base

There are four processes involved in creating a knowledge base. The first process is knowledge acquisition. This is the procedure for transferring the knowledge or expertise from the experts to the knowledge base. In this process, the knowledge engineers are the key individuals to interview the experts in order to collect the right information. The second process is knowledge representation. After being recorded or noted, this knowledge will be transformed into a standard format which helps knowledge engineers to more easily group them together or create the decision tree diagrams which represent the structure of the knowledge base. This process should be done only after all knowledge has been acquired. The third process is to develop the user interface. This process is an important part of the knowledge base in that it helps users to easily use and understand the program. Moreover, it is not only developed for the users but also the knowledge engineers in order to create or modify the program efficiently. The last process is knowledge validation and evaluation. This process validates and confirms all knowledge before it is used by the users.

Knowledge Acquisition

The knowledge acquisition process is considered to be a significant process of expert systems in order to transfer the knowledge or expertise from the experts to the knowledge base. In this process, the knowledge engineers are the key individuals to collect the knowledge and expertise for creating the reliable and practical expert systems. The quality of knowledge often determines the success of an expert system [Parsaye 1988]. To gather knowledge, there are three basic methods: observation, literature review, and interview. Of these three methods, interviewing seems to provide the most direct access to compile human knowledge.

To interview experts, knowledge engineers use various communication techniques such as taking notes, using tapes, video tapes, or questionnaires. By using audio or video tapes, knowledge engineers can accurately capture the information offered by experts; however, some experts may feel uncomfortable being taped. Therefore, in some cases, using a questionnaire or taking notes can diminish experts' anxiety.

Develop a questionnaire

In this study, the questionnaire contains a number of questions which were carefully developed after the authors made an extensive review of related literature. There are two alternatives for developing the questionnaire. The first alternative is to list each question in multiple choice form with full explanation so that experts can quickly check and indicate the choice to be considered. Based on the first alternative, however, the experts may hesitate to indicate the choice because the hypothesis created by the authors may be different from the experts' opinion. The second alternative is to ask the experts to create the hypotheses and then list the choices or the factors which are considered to be significant. This alternative, which was selected for this study, can free the experts' thinking and can obtain more information and comments.

There are three main questions in this questionnaire. The first question asks the experts to indicate the most common types and model of earthmoving equipment. This question provides many types of earthmoving equipment such as excavators, shovels, loaders, scrapers, and backhoe-loaders and allows the experts to check off the appropriate ones. The purpose of this question is to generate the scope of this study by considering only the commonly used equipment. Since it is known that there are various kinds of earthmoving equipment, limiting only the commonly used equipment makes this study more practical for actual construction.

The second question requests the experts to describe the suitable conditions and limitations for using equipment which is specified in the answer to the first question. The purpose of this question is not only to provide the advantages and disadvantages of such equipment but to help the authors as the knowledge engineers to discover some significant factors for further consideration.

The last question is the most important question among three because the results from this question will be considered to generate the rules in the knowledge representation process. First, it asks the experts to indicate the factors which are significant in selecting the types of earthmoving equipment. Second, for each factor that the experts indicated, they are asked to list the subfactors that affect the choice of equipment. Finally, for each subfactor that is specified, the experts are requested to indicate their preference on a scale of 5 to 1, where 5 is the most commonly used equipment and 1 is seldom used equipment. The results from this question indicate which types of earthmoving equipment are suitable for each subfactor and which one is the most preferable equipment among those types of equipment.

The answers to these three questions provide the set of raw data which specifies the detailed information for each type of equipment or each subfactor; however, in reality, a combination of many subfactors is considered in order to select the suitable types of earthmoving equipment. Thus, this raw data will be implemented in the process of knowledge representation as discussed below.

Knowledge Representation

Knowledge representation is an expert system process that organizes acquired knowledge and represents it in a form that is easy to understand and ready for implementation. One of the most popular knowledge representation schemes is the rule-based system, especially the production rule system (IF-THEN), which represents knowledge in a format that is close to natural language. For example, regarding the experts' expertise, scrapers are commonly used for large areas with a hauling distance between 33 meters and 2.5 kilometers. A rule can be stated that if the excavation size is large and hauling distance is between 33 meters and 2.5 kilometers, then scrapers are the most commonly used equipment. In this study, the authors only use the production rule system to represent the knowledge and the experts' expertise.

In the expert system program, the production rules are selected for processing by search control strategies of inference engine which is already provided in expert system shells. In general, there are two types of search control strategies that the inference engine uses in searching for rules: forward chaining and backward chaining. Forward chaining is the general term used to describe a group of search strategies which start with known facts and try to infer the conclusions implied by those facts [Pederson 1989]. In other words, conclusions from the forward chaining are the result of all available facts. The forward chaining may continue searching rules for new facts until no new facts can be found.

In contrast, backward chaining starts with known conclusions and tries to search the relevant conditions or facts which can support when the rule is true. Like forward chaining, backward chaining continues searching rules until no new conclusions can be found. Usually, both search control strategies can generate the same results; however, which one should be used is based on the type of application. In general, forward chaining is applied for recommendation problems and backward chaining is more applicable for diagnosis problems. In this study, types of earthmoving equipment are recommended for particular conditions of construction tasks; therefore, using forward chaining is more suitable than using backward chaining.

Develop the production rule

As an example in Table 1, excavated material, size of excavation, excavation depth, and hauling distance are considered as the factors in earthmoving activities. Suppose that these factors have the following characteristics: 1) muddy clay, 2) large area, 3) shallow excavation, and 4) hauling distance < 33 meters. The preference values of each type of equipment for each factor are accumulated into a total preference value and then converted into a percentage. The most suitable type of equipment is that which has the highest percentage of preference. Note that a zero preference value of any factor makes the total preference also zero.

Table 1 The % preference for a combination of four significant factors. up

Regarding the information from Table 1, a production rule can be generated as the following statement.

IF Material is muddy clay
and Size of excavation is large area
and Excavation depth is shallow
and Hauling distance < 33 meters.

THEN "Excavator(track) and crawler-loader are the most suitable equipment for moving muddy clay in large area, shallow excavation and hauling distance < 33 meters with 70% preference."

User interface

The user interface is an important component in the expert system since it serves as the link between the system and users. Users can learn by themselves and become acquainted with the program through its user interface. Moreover, the interface with the visualization part, developed from a different package, enhances the capability of the expert system program.

The interface allows users to indicate the type of material, terrain, and other significant factors via input screen, as can be seen in Figure 1, and then present all outputs, i.e. the conclusions, via the conclusion screen. Figure 2 shows an example of a conclusion screen for a particular condition of site and material. In addition, the explanation of all technical words is provided to enrich the comprehension of the users.

VISUALIZATION COMPONENT

Providing the visualization component can not only help users to grasp complex knowledge but also to visualize the method of construction. An animation of various types of equipment is created by using the 3D Studio package, which runs on a personal computer, and then is integrated into the knowledge base after it is converted to ".avi" files. To develop an animation, five main functions of the program have to be considered: 2D Shaper, 3D Lofter, 3D Editor, Keyframer, and Materials Editor.

2D Shaper

The 2D Shaper function allows program designers to create two dimensional objects which can become three dimensional objects by adding the thickness. Although most three dimensional objects can be created by using 3D Editor function, complex objects such as texts or free hand objects can be easily built by using a 2D Shaper function.

Figure 2 An example of a conclusion screen for a particular condition of site and material. up

3D Lofter

The main objective of the 3D Lofter function is to convert two dimensional objects built by using 2D Shaper function into three dimensional objects. By adding the thickness, they become three dimensional objects and are adopted in the 3D Editor function later. Furthermore, this function allows users to adjust the path of two dimensional objects so that the shape of objects can be more complicated.

3D Editor

The 3D Editor function is the most important part of 3D Studio because most objects are usually built and adjusted in this part, including creating surfaces, lights, and cameras. There are four main viewpoints which help users create the models: top, front, left or right, and user or camera view. Moreover, users can also adjust the shape of three dimensional objects using Boolean operations. This operation, which allows users to merge or subtract the objects from each other, results in a new object which is hard to create through the use of the Object Create operation. The other important function of 3D Editor is to render pictures, the non-movement picture. Users can create an image file such as ".tga" files or ".bmp" files by using this operation.

Keyframer

The principle of keyframer is to create the animation by integrating movements of objects, lights and cameras in each frame. In order to create the smooth movement of objects, lights or cameras, the Path operation can be used to adjust each position in each frame. Users can roughly check their movements by using the Preview operation, which can save time for rendering. Furthermore, the three dimensional objects can link or relate together using the Hierarchy operation, which allows users to indicate the pivot point between two objects. Animation is created in ".flc" files and will be converted to ".avi" files latter in order to integrate it with the knowledge base.

Materials Editor

The Material Editor function is an important part for adjusting the texture and color of surfaces of objects. Most materials which are available in 3D Studio are not always suitable for assigning to objects, so users have to adjust that material by following these procedures:

1) get the material from the library.
2) adjust the color or the texture of that material.
3) put the adjusted material into the library by creating a different file name.

After these steps, users can assign the new material to particular objects. The use of this function results in the flexibility and variety of 3D Studio for various kinds of objects.

CONCLUSION

Based on experts' knowledge, the authors developed a visualization-based expert system for making decisions regarding the use of equipment. The system is expected to become a self-contained interactive educational and training tool that will facilitate learning the avoidance of mistakes and inappropriate use of construction equipment. In addition, its visualization component adds clarity and interactive capability. Furthermore, in the future, the system will become a knowledge bank housing the collective wisdom of many experts; hence, we expect the system will furnish reliable solutions to a variety of construction methods and equipment selection problems.

REFERENCES

3D Studio.,[1994]. 3D Studio Release 3 Manual. Autodesk, Inc., California.

Nichols, Jr. Herbert L., [1989]. Moving The Earth. Prentice-Hall Inc., Englewood Cliffs, NJ.

Nunnally, S. W., [1987]. Construction Methods and Management. Prentice-Hall, Englewood Cliffs, NJ.

Parsaye, Kamran., [1988]. Expert Systems for Experts. Prentice-Hall Inc., Englewood Cliffs, NJ.

Pederson, Ken., [1989]. Expert Systems Programming. John Wiley & Sons, Massachusetts.

Wentworth, James A., [1990]. Road Transport Research: a guide, Organization for economic cooperation and development.

Back to Table of Contents