ABSTRACT
In this study, Artificial Neural Network and Neuro – Fuzzy models were developed using data extracted from a residential two – storey reinforced concrete framed structure construction schedule and project execution documents. The evaluation of project performance indicators in earned value analysis from 0 – 100% progress at 5% increment with a total of seventeen tasks were carried out using Microsoft Project software and data obtained from the computation were utilized for model development. Pearson Correlation results obtained for the model variables indicated stronger positive relationship between the response factors Earned Value (EV) and Performance Indicators namely; Planned Progress, Actual Time (AT), Earned Schedule (ES), Actual Cost (AC) and Cost Variance (CV) while negative linear relationships were observed to exist for the Schedule Performance Indicator (SPI) and Schedule Variance (SV) factors. Using input – output and curve fitting (nftool) function in MATLAB, a 6 – 10 – 1 two – layer feed – forwards network with Tansig Activation Function (AF) for the hidden neurons and linear Activation Function (AF) output neurons was generated with Levenberg – Marquardt (Trainlm) training algorithm. Similarly, with the aid of ANFIS toolbox in MATLAB software, the training, testing and validation of the ANFIS model were carried out using hybrid optimization learning algorithm at 100 epochs and Gaussian Membership Function (gaussmf). Loss function and statistical parameters; Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and R-values were taken as the performance evaluation criteria of the developed smart intelligent models. The generated statistical results indicate no significant difference between model results and experimental values with MAE, RMSE, R2 of 1.9815, 2.256 and 99.9% respectively for ANFIS model and MAE, RMSE, R2 of 2.146, 2.4095 and 99.998% respectively for the ANN model. The model performance shows adaptive and robust behavior to deal with complex relationships between the model variables to produce accurate target response.
TABLE OF CONTENTS
Title Page i
Declaration ii
Certification iii
Dedication iv
Acknowledgement v
Table of Contents vi
List of Tables ix
List of Figures x
Abstract xi
CHAPTER 1: INTRODUCTION
1.1 Statement of Problem 4
1.2 Aim and Objectives of Study 5
1.3 Significance of Study 5
1.4 Scope of Study 6
CHAPTER 2: LITERATURE REVIEW
2.1 Construction Schedule 8
2.1.1 Importance of scheduling in construction projects 10
2.2 Project Scheduling Process 11
2.2.1 Plan schedule management 12
2.2.2 Define the project – activities 12
2.2.3 Determine dependencies 13
2.2.4 Sequence activities 13
2.2.5 Estimate resources 13
2.2.6 Estimate durations 14
2.2.7 Develop the project – schedule 14
2.2.8 Monitor and control 15
2.3 Project Scheduling tools and Techniques 15
2.3.1 Tool lists 15
2.3.2 Calendar 15
2.3.3 Gantt charts 16
2.3.4 Guidelines for generating a better project – schedule 16
2.4 The Critical – Path Method 16
2.4.1 Mathematical formulation of CPM scheduling 18
2.4.2 Critical path – scheduling algorithms 18
2.5 Other Scheduling Methods 21
2.5.1 Line – of – balance (LOB) scheduling technique 21
2.5.2 Resource – oriented scheduling 22
2.5.3 Q Scheduling 22
2.6 Earned Value Management (EVM) 22
2.6.1 Calculating earned value 24
2.6.2 The EVM indicators 24
2.7 Fundamentals of Earned Value Management 26
2.7.1 Organization and scope of project 26
2.7.2 Planning, scheduling and budgeting 27
2.7.3 Accounting for actual costs 28
2.7.4 Analyzing and reporting on project performance 28
2.7.5 Revisions and data maintenance 28
2.8 Artificial Intelligence Application in Construction Management 31
2.8.1 Artificial neutral network and fuzzy – inference – systems 32
CHAPTER 3: METHODOLOGY
3.1 Experimental Design 36
3.2 Steps to Critical Path Calculation 37
3.2.1 Forward scroll algorithm 37
3.2.2 Backward scroll algorithm 37
3.3 Model Performance Evaluation 41
CHAPTER 4: RESULTS AND DISCUSSION
4.1 Data sets for Model Development 46
4.1.1 Pearson correlation 49
4.2 Artificial Neural Network (ANN) Model Development 51
4.2.1 Training state of the ANN 54
4.2.2 Validation performance of the ANN 56
4.2.3 Error histogram of the ANN 58
4.2.4 Regression plot of the ANN 60
2.4.5 Selection of optimized ANN model 62
4.3 Neuro – Fuzzy Model Development 64
4.3.1 Testing and Training ANFIS 67
4.3.2 Graphical plots of the membership function 69
4.3.3 Selection of optimized ANFIS model 72
4.3.4 ANFIS – model variables graphical expression 74
4.4 Model Validation 76
4.5 Sensitivity Analysis 81
CHAPTER 5: SUMMARY, CONCLUSION AND RECOMMENDATION
5.1 Conclusion 83
5.2 Recommendation 84
REFERENCES 85
APPENDICES 91
LIST OF TABLES
4.1: Start and End Times of the Activities 43
4.2: Performance Indicators Computation Results 44
4.3: Pearson’s Correlations for Model Parameters 50
4.4: Artificial Neural Network Processing Parameter Settings 52
4.5: ANN Architectures’ comparison to derive an optimized model during training and testing 63
4.6: ANFIS Network Parameter 65
4.7: ANFIS Architectures’ comparison to derive the optimized model during training and testing 73
4.8: Actual and Model predicted results 77
4.9: Performance Evaluation of ANFIS Model 78
LIST OF FIGURES
2.1: Project Scheduling Process 12
2.2: Earned Value Analysis 30
3.1: Precedence Relations and Durations for a Nine Activity Project 39
3.2: Gantt chart 40
4.1: Interpretation of Value for Indicators of Project Performance 45
4.2 a and b: Distribution histogram chart for input variables (AT and ES) 46
4.2 c and d: Distribution histogram chart for input variables (SV and Planned Progress) 47
4.2 e and f: Distribution histogram chart for input variables (SPI and Planned AC) 47
4.2 g and h: Distribution histogram chart for input and variables (CPI and CV) 48
4.2 i: Distribution histogram chart for output variables (EV) 48
4.3: ANN Architecture 53
4.4: ANN Training State 55
4.5: Validation performance of the ANN 57
4.6: ANN Error Histogram 59
4.7: ANN Training, Testing and Validation Regression Plot 61
4.8: ANFIS Model Variables and Architecture 66
4.9: ANFIS Model Training and Error Plot 68
4.10: Plot of Testing Datasets 68
4.11 ANFIS Membership Function Plots 70
4.12: 3D-Surface Plots of ANFIS – Model Variables 75
4.13: Goodness of Fit Plot for ANFIS model 79
4.14: Goodness of Fit Plot for ANN model 80
4.15: ANN model Sensitivity analysis results 82
4.16: ANFIS model Sensitivity analysis results 82
CHAPTER 1
INTRODUCTION
Artificial intelligence (AI) is a combined term used to describe a machine that has the ability to imitate and copy human ability to solve problems, recognize patterns, cognitive capabilities and learning. Machine learning is an aspect of AI capable of statistical manipulations to offer computer-systems its ability to "learn" from data, with no explicit programming (Das et al., 2011). A machine becomes better at understanding and providing insights on exposure to more data. Artificial intelligence is a science on the research and involvement of the techniques of human intelligence activities. It is unarguably a far reaching cross – frontier subject, after more than fifty years advancement. The field’s goal is to device means to perform similarly and carryout certain intelligent function of human brain, such that people can develop technology products and establish relevant theories.
More often than not, this evolutionary technology is adopted in many area of specialization such as knowledge based system, intelligent database system, expert-system, and intelligent robot system (Flood, 2008). In Civil Engineering field, vast problem areas especially, in design, construction management, and decision making for certain programs affected by many uncertainties which could be solved not only by using Mathematics, Physics, and Mechanics calculations but also using knowledge acquired from the practitioners (Virle and Mhaske, 2013). Such knowledge and practice are illogical, not complete and without precision, as a matter of fact, cannot be solved with conventional procedures. However, artificial intelligence possesses its own advantages including the ability to handle complex challenges to expert levels by imitating the experts experience and knowledge. In all, artificial intelligence is vast and has broad applicability potentials in civil engineering practice. The capability of machines to perform tasks intelligently is known and called Artificial Intelligence. Artificial Intelligence in today’s modern world has become very popular. It entails simulating natural intelligence abilities in machines and programming such machines to learn and carbon-copy the activities of human beings. The machines are formulated to be able to learn experientially and deliver tasks like humans. The continuous growth and expansion of technologies for example AI, makes impactful changes on quality and easiness of life.
It is obvious that every major industry has a role for Artificial intelligence and for obvious good reasons: It is effective for solving extremely hard and challenging problems that reduces productivity and accuracy. It is utilized in construction to monitor situations and progress at job site to identify unsafe behavior, for instance (Zhang and Haghighat, 2010). Earned Value Management simply called EVM is a standard process used to locate project variances through comparison of actual work executed versus work in plan. It is able to adequately integrate measurements of the project management constraints triangle; scope, time, and costs. In a given integrated system, EVM provides effective forecasting of performance aberrations in project, which is a very essential contribution for project management. Project managers use the method to monitor cost and schedule and apply it for project forecasting. The method provides quantitative data for project decision making and also provides technique that measures project performance against the project baseline (Iranmanesh and Zarezadeh, 2008).
EVM method makes it possible for the manager of a project to compute the extent of work actually carried out on a project by doing more than just reviewing cost and schedule reports. EVM makes it possible that a method is advanced and projects are measured by the progress made. By that, the project manager will be able to use the measured progress to forecast the actual cost and completion-date of a project, taking into cognizance trend-analysis or usability of the project’s ‘burn-rate’. Such a model depends on a key measure which is the project’s earned value (Ma and Yang, 2012). Projects are subjected to numerous factors that interfere with their efficient completion. This is the reason that makes it veritable to closely check progress made in work in real time, and analyze systematically a given abnormally in work-schedules and project costs as against the previously planned values, in order to take preventive steps to minimize the negative impact of perplexing components (Senouci and Al-Derham, 2008).
Artificial Intelligence in construction is potentially able to help stakeholders realize value throughout project lifecycles, including: Design, bidding, and financing; procurement and construction; operations and asset management; and, business model transformation. Construction participants and researchers develop smart-technologies that replicate human intelligence and knowledge to reduce the dependence level of expert in construction planning and schedule control. An excellent project-schedule is a critical step toward finishing a project within budget and timely (Aziz et al., 2014; Eber, 2019). Design schedules for construction project are expected to offer a clearer and descriptive summary of all project timelines, milestones and deadlines. It should be routinely updated to track and examine progress of several activity-phases needed for project completion. A construction project schedule makes for a comprehensive explanation and prototype of the contractor’s plans on how to construct the project and makes visible the work scope. That work scope is a sequential representation of the schedules’ work activities, and the duration required. One critical point to be abreast of is that a project-schedule is singlehandedly the management document capable of forecasting the time a project will be rounded up. Scheduling entails listing of specific task and activities, achievements indicating a start and finish date as planned (Zheng and Liu, 2010). Scheduling in construction cannot be overemphasized because of how crucial it is to the success of a project. Proper schedules has the potential to ensure a timely project completion and in line with the budget. It does not only highlight the speed of the work but includes how/when a tasks is performed. More so, scheduling outlines method and procedure for materials delivery. And conclusively, it gives room for seasonal readjustments to allow alterations and uncertainties to be accommodated (Jacob and Kane, 2004).
1.1 STATEMENT OF PROBLEM
Construction projects are inherently complex, with several moving-parts. A common school of thought is that attention should be paid to execution, in practice, it is overt that success of a project depends more on appropriate up-front planning. Here is an area where Artificial Intelligence makes significant difference. New product development requires planning and effective arrangement of project activities to ensure their controlling, and finally to meet the project requirements. Planning and effective scheduling of project activities utilizes the predictors of required time for each activity, their dependencies and amount of resources needed. The estimators’ quality is a function of the new product type deployed for forecasting. A good performance in terms of prediction accuracy is rather difficult to obtain for products of technological breakthroughs because they are required to create their market. In turn, for modified products being new to the company but not to the market, there is the possibility to use data from previous product development projects. Planning is essentially forecasting future events and speed. With the advent of artificial intelligence, this process of planning can further benefit the significant amounts of historical productivity and as built robust-performance data that was too difficult and tedious to mine through to help influence the forecast being put together. Planning, and specifically scheduling, tends to be highly labor-intensive processes. Scheduling tools are recently deployed to help in the process of planning by providing informed details regarding what durations, sequence and work cost should be.
1.2 AIM AND OBJECTIVES OF STUDY
This study is aimed at application of Artificial Intelligence to construction scheduling for better prediction of project duration and cost minimization in Nigeria construction industry.
The specific objectives are to:
• Develop a robust optimization tool to monitor and track project progress.
• Enhance the efficiency of reallocation, rescheduling and forecasting process with decision support system.
• Use Earned Value – Management (EVM) model for better forecasting of progress and performance.
• Deployment of Artificial Intelligence methods for constructing better forecasts to estimate the final period of a project
1.3 SIGNIFICANCE OF STUDY
Numerous simple scheduling practices are implemented in construction. However, with more complex and larger projects these days, there is need for more practical and formal scheduling procedures. Having a handy, easy to use, and explicit construction-schedule indicates that the suppliers exactly know what to deliver at required timeframe. Moreover, subcontractors and critical project team players will learn proactively when to book their skilled laborers and artisans. A good project schedule is updated continuously, detailed accurately, wherein communication about such a project with constraints involved is analyzed. Corporation of project team players is an important factor which supports successful completion of projects. Scheduling is essential to the completion and success of a construction project, it empowers project-managers to match labor, materials, equipment, and all other resources associated with activities and construction tasks over time. Properly planned project schedule guarantees the effective completion of the project by analyzing how and when all required tasks are carried out, plus the sequences and materials delivery methods.
1.4 SCOPE OF STUDY
The coverage of this research study is limited to the application of soft-computing techniques in the investigation of construction schedule management through earned-value-analysis to evaluate factors propagating cost overruns and innovating better methods that help achieve success in the project goal. Three states in Nigeria were selected as the study area namely; Enugu, Abia and Rivers. A better and more concise explanation is that effective project-management is all that is needed to get a project done on time and strictly as budgeted to deliver the needed scope and quality. A project manager has must make a tradeoff between cost, time and scope while ensuring the required quality. The study involves; gathering of data from selected large construction sites, also using well-structured questionnaires, and modeling using fuzzy-logic, neural-network and Neuro – Fuzzy soft-computing techniques in MATLAB software and prediction performance validation using Multiple regression statistical models and analysis of variance (ANOVA).
Click “DOWNLOAD NOW” below to get the complete Projects
FOR QUICK HELP CHAT WITH US NOW!
+(234) 0814 780 1594
Login To Comment