Sign Up
ENعربي
All articles

Project Management

Artificial Intelligence in Project Management in the Metal Industry

  • Introduction

Companies are under immense pressure to reduce costs, improve quality, and shorten delivery times. The metalworking industry, a critical producer of intermediate goods, is particularly affected by these demands. One of the most persistent challenges in this sector is meeting delivery deadlines, which can be compromised by a multitude of factors such as supply chain disruptions, project management inefficiencies, technical issues, design alterations requested by clients, workforce-related problems, workplace safety concerns, and inaccuracies in determining activity times.

To overcome these challenges, various approaches and methodologies have been adopted. Agile management tools like Scrum, Prince2 Agile, Lean Project Management, Agile Project Management, and DSDM (Dynamic Systems Development Method) have been implemented to enhance the flexibility and adaptability of production processes. Classic analytical solutions, such as the Gantt/Waterfall Method, PMBOK Guide, Spiral Model, Prince2, Critical Path Method (CPM), and PERT method, are also applied to provide a clearer and more structured view of projects.

Artificial Intelligence (AI) has emerged as a promising tool in project management, offering potential solutions to predict and mitigate delays, improve operational efficiency, and optimize resource allocation. This literature review aims to explore the applications of AI in project management, with a specific focus on its potential within the metalworking industry. The structure of this review will begin with an overview of AI methods, followed by an examination of AI applications in project management and a discussion of the current state and future potential of AI in metalworking project management.

  • Overview of AI Methods

Artificial intelligence has demonstrated great potential, and its increasing adoption by companies signals a significant shift in how industries operate. With the exponential growth of data, AI applications are expanding significantly. Scientific studies suggest that AI can greatly improve project control and production management.

The primary goal of AI is to create systems capable of simulating human intelligence, enabling them to learn, reason, and make decisions. Artificial Neural Networks (ANNs), inspired by the human brain, are a critical AI technique. ANNs consist of interconnected layers of artificial neurons that process and transmit information. These neurons receive input signals, process them using mathematical functions, and emit output signals to other neurons. The network learns by adjusting its weights and internal parameters to minimize the error between inputs and desired outputs. Once trained, the model can make predictions or classifications of new data. ANNs are used in various fields, including speech and image recognition, natural language processing, time series forecasting, recommendation systems, games, robot control, and project management.

AI methods can be classified based on their capability, functionality, and purpose. One common classification divides AI into two main groups: Narrow or Weak AI and General or Strong AI.

  • Narrow or Weak AI: This type of AI is designed to focus on and execute a specific, delimited task. It lacks the ability to apply knowledge gained from that task to other domains, formulate plans, comprehend abstract concepts, or undergo subjective experiences. Instead, it reacts to a finite set of predefined inputs in a manner that simulates intelligence. Machine Learning (ML) and Deep Learning (DL) are two branches of Narrow AI that rely on algorithms and data to solve complex problems.
  • Machine Learning (ML): This involves teaching a machine to learn from historical data without explicit programming. ML techniques are classified into Supervised Learning and Unsupervised Learning. Supervised learning uses labeled training data to develop predictive models, while unsupervised learning analyzes unlabeled datasets to identify patterns and structures without specific guidance. Supervised learning techniques include Linear Regression, Logistic Regression, Support Vector Machines (SVM), Decision Trees, Artificial Neural Networks, Ensemble Algorithms (Random Forest and Gradient Boosting Machines), and Gaussian Naive Bayes. Unsupervised learning techniques include K-Means, Principal Component Analysis (PCA), and Variational Autoencoder (VAE).
  • Deep Learning (DL): This is a method of machine learning that uses artificial neural networks with multiple layers to extract patterns and features from data. DL techniques are applied in Natural Language Processing (NLP) and Computer Vision. NLP enables machines to understand, interpret, and generate human language effectively. Computer vision allows computer systems to interpret and understand visual information, enabling them to "see" and process visual data. NLP techniques include Tokenization, Part-of-Speech Tagging, Named Entity Recognition (NER), Sequence-to-Sequence Models, Transformers (BERT), Sentiment Analysis, Word Embeddings (Word2Vec e GloVe), Long Term Memory Networks (LSTM), and Short-Term Memory Networks (GRU). Computer Vision techniques include Convolutional Neural Networks (CNN), Region-based CNN (R-CNN), Recurrent Neural Networks, Transfer Learning, and Generative Adversarial Networks (GANs).
  • General or Strong AI: This is a theoretical form of AI that aims to replicate human functions, such as reasoning, planning, and problem-solving. It can perform any intellectual task that a human can do, including cognitive abilities like reasoning, learning, planning, creativity, and emotional understanding. General or Strong AI would function analogously to a human being, capable of reacting to stimuli and performing a wide range of tasks. Techniques in this category include Reinforcement Learning, Evolutionary Algorithms, and Hybrid Approaches. Reinforcement Learning involves autonomous agents learning to make decisions by interacting with dynamic environments and receiving rewards or penalties based on their actions. Evolutionary Algorithms replicate biological processes to optimize solutions, employing methods such as selection, mutation, and recombination. Hybrid Approaches combine Reinforcement Learning and Evolutionary Algorithms to create more robust and adaptable systems.

III. AI Applications in Project Management

The literature review methodology involved analyzing scientific articles from the Scopus database, using keywords such as "Project management," "AI," "Planning," "Schedule," and "Forecast". The search was conducted using three groups of keywords: "Project management AND AI AND Forecast," "Project management AND AI AND Schedule," and "Project management AND AI AND Planning". The articles were filtered to include those published between 2019 and 2023.

The initial search yielded 198 articles, which were then narrowed down to 107 unique articles. After a meticulous analysis, 48 articles were identified that presented the main AI methodologies used successfully in project management. However, some articles were excluded because they were duplicates, bibliographic reviews, or provided qualitative information without relevant data. Filters were applied to include only articles in English, published in journals, and with full access to the content. The final selection comprised 23 articles.

The analysis of these articles revealed the following main AI methodologies used in project management:

  • Machine Learning (ML): ML methodologies are the most prevalent in project management. Among ML techniques, Artificial Neural Networks (ANNs) are the most widely used. ANNs are capable of learning complex patterns through supervised data. Ensemble Algorithms (Random Forest and Gradient Boosting Machines) are also reliable alternatives. Support Vector Machines (SVMs) are used in medical research and the electrical sector, and Logistic Regression is applied in the field of medicine.
  • Deep Learning (DL): Within Deep Learning, Convolutional Neural Networks (CNN) are the most popular, applied in projects across different domains, such as the agricultural sector, construction sector, and medical research. Generative Adversarial Networks (GANs) are employed in the operations and construction sectors.

The review found no scientific articles that mentioned Reinforcement Learning AI, Evolutionary Algorithms, or Hybrid Approaches in project management with positive results.

  • AI in Metalworking Project Management

A significant finding of this literature review is the scarcity of studies specifically focused on AI applications in the metalworking industry. Despite the advancements and applications of AI in project management in other sectors, its potential in metalworking remains largely unexplored.

The metalworking industry could greatly benefit from AI, particularly in optimizing processes, enhancing quality and efficiency, and potentially revolutionizing production. AI can be applied to improve the accuracy of time estimates, optimize resource allocation, predict and mitigate delays, and enhance overall project effectiveness.

One promising direction for future research is the development of an AI tool for multi-project planning in metalworking, with a focus on using Artificial Neural Networks (ANNs). Such a tool could predict production times more accurately and contribute to optimizing industrial processes, reducing costs, increasing operational efficiency, and decreasing project delivery times. This tool could quantify project times based on the parameterization of a set of parameters that characterize the system at a precise moment.

  • Conclusion

This literature review highlights the challenges and gaps in research on the use of AI in project management. Methodological screening revealed that many articles were either qualitative research, unrelated to the study's scope, or duplicates. The analysis of AI methodologies showed a preference for Machine Learning (ML), particularly Artificial Neural Networks (ANNs).

A notable gap is the lack of studies on AI applications in the metalworking industry, which suggests a promising area for future research. The Transformer architecture, while primarily used in Natural Language Processing (NLP) with models like BERT and GPT, also holds potential for other areas, including project management. Future research could explore the use of Transformer architectures in project management to improve various processes.

Overall, AI has significant potential to improve production times, reduce costs, and increase efficiency in the metalworking industry. The development of AI-driven tools, particularly those using ANNs, could lead to more accurate project planning and execution in this sector. Further research and implementation of AI in metalworking are needed to fully realize these benefits and drive innovation in the industry.

FAQ on AI in Project Management, Specifically in the Metalworking Industry

  • Why is there a need to improve project management in industries like metalworking?

Intense competition pushes companies to reduce costs, improve quality, and shorten delivery times. The metalworking industry faces challenges in meeting delivery deadlines due to supply chain problems, project management difficulties (lack of coordination, communication issues), technical issues (equipment defects, lack of skills), client-driven design changes, workforce problems (lack of training, absenteeism), safety concerns, and inaccurate time estimation for project activities. Inaccurate activity time estimation is a particularly relevant cause of delays in the metalworking industry.

  • What traditional approaches have been used to address project management challenges and time estimation inaccuracies?

Traditional approaches include agile management tools like Scrum, Prince2 Agile, Lean Project Management, Agile Project Management and DSDM (Dynamic Systems Development Method). Classic analytical solutions are also used, such as the Gantt/Waterfall Method, PMBOK Guide, Spiral Model, Prince2, and the Critical Path Method (CPM) and/or the PERT method, to provide a clearer and structured view of projects. Algorithms like Monte Carlo, Genetic Algorithms, Linear and Non-Linear Programming Algorithms and Swarm Algorithms are used to optimize resource management.

  • How is Artificial Intelligence (AI) being applied to project management, and what specific AI techniques are most commonly used?

AI is emerging as a promising tool for project management, particularly in predicting and mitigating delays, improving operational efficiency, and optimizing resource allocation. Machine Learning (ML) is the most preferred AI technique, with a strong emphasis on Artificial Neural Networks (ANNs). Deep Learning (DL) is also utilized, though less frequently than ML.

  • What are the two main categories of AI, and how do they differ in their application to project management?

AI is broadly divided into Narrow or Weak AI and General or Strong AI.

  • Narrow or Weak AI focuses on specific, delimited tasks with limited ability to apply knowledge to other areas. Machine Learning (ML) and Deep Learning (DL) fall under this category.
  • General or Strong AI aims to replicate human intelligence, including reasoning, planning, and problem-solving. While theoretically powerful, this study found no evidence of General or Strong AI techniques (like Reinforcement Learning, Evolutionary Algorithms, or Hybrid Approaches) being successfully applied in project management.
  • Can you describe some specific Machine Learning (ML) and Deep Learning (DL) techniques and their applications in project management?
  • Machine Learning (ML) Techniques:
  • Artificial Neural Networks (ANNs): Used for learning complex patterns and are the most used technique for projects in the construction industry.
  • Ensemble Algorithms (Random Forest, Gradient Boosting Machines): Reliable alternatives for various projects, including construction.
  • Support Vector Machines (SVMs): Applied to classification and regression, particularly in medical research and the electrical sector.
  • Logistic Regression: Used for binary classification problems.
  • Deep Learning (DL) Techniques:
  • Convolutional Neural Networks (CNNs): Popular within Deep Learning, with applications across agriculture, construction, and medical research.
  • Generative Adversarial Networks (GANs): Used for generating realistic synthetic data, applicable in operations and construction.
  • Long Short-Term Memory (LSTM) networks: Used for time-series prediction.
  • What challenges are there in applying AI to project management, particularly in the metalworking industry?

Research indicates a scarcity of studies on AI applications in project management, especially within the metalworking industry. A significant portion of the reviewed articles were either qualitative research, unrelated to the study's scope, or duplicates. This reveals a gap in the literature and suggests an unexplored field for implementing AI technologies in metalworking. No articles reviewed mentioned the application of Reinforcement Learning, Evolutionary Algorithms, and Hybrid Approaches with positive results in project management.

  • What specific area of project management shows the most promise for AI application?

Based on the reviewed literature, AI, and specifically machine learning techniques like ANNs, shows promise in improving the accuracy of time estimation in project management. Addressing this is critical to mitigating delays and improving overall project efficiency.

  • What future research directions are suggested by this study, especially regarding the metalworking industry?

The study suggests the development of an AI tool using Artificial Neural Networks (ANNs) to support multi-project planning and improve the accuracy of production time determination in the metalworking context. This could lead to optimized industrial processes, cost reduction, increased operational efficiency, and decreased project delivery times. Further, the Transformer architecture has potential beyond natural language processing in project management applications.

(Silva et al., 2024)

Reference:

Silva, J., Ávila, P., Matias, J., Faria, L., Bastos, J., Ferreira, L., & Castro, H. (2024). Bibliographic review of AI applied to project management and its analysis in the context of the metalworking industry. IFAC-PapersOnLine, 58(27), 177–187. https://doi.org/10.1016/j.procir.2024.10.073