AI and Machine Learning for Mechanical and Electrical Engineering ...
Unknown
First edition
About this book
Practical and informative, AI and Machine Learning for Mechanical and Electrical Engineering examines how artificial intelligence (AI) is changing the status quo in mechanical engineering, electrical systems, and management. Real-world examples and case studies demonstrate the application of AI in such diverse settings as industry and policymaking. This book illustrates how AI is playing a crucial role in enhancing productivity and innovation in various industries. It discusses transition methods and the ethical implications of using AI in mechanical engineering. Chapter highlights include the following:
- Developing a smart algorithm to integrate fault detection and classification
- Algorithms to investigate different testing scenarios for various anomalies in electric motors
- Data fusion to detect and assess electromechanical damage
- Neural networks for rolling bearing fault diagnosis
- Evolutionary algorithms to optimize deep learning models for water industry forecasts
- AI-based anomaly detection and root-cause analysis
An overarching theme is the transition from traditional mechanical, electrical, and management systems to AI-enabled smart systems. The book helps readers make sense of the challenges of integrating smart systems. It equips engineers with theoretical understanding as well as insight based on hands-on expertise. It shows how to better link and automate systems and improve productivity. This book not only shows how to implement smart solutions now but also shows the way to a more intelligent, productive, and interconnected future.
Questions & Answers from this book
Questions and answers are connected to the referenced book and its available source material.
Chapter 1: Development of a Smart Algorithm to Integrate Fault Detection and Classification of End-to-End Monitoring of Autonomous Transfer Vehicles
Chapter 2: Data Science and ML Algorithms to Investigate Different Testing Scenarios for Various Anomalies in Driven Electric Motor
Chapter 3: A Data Fusion Technique to Detect and Assess Electromechanical Damage
Chapter 4: AI: Classifications and Protection of the Smart Grid Systems
Chapter 5: An Artificial Intelligence-Based Solar Radiation Prophesy Model for Green Energy Utilisation in the Energy Management System
What is the inverse distance weighting method and how is it used to fill in missing weather data in the solar radiation prediction model?
Inverse distance weighting is an interpolation method used to estimate a missing meteorological value at a target station by combining observed values from nearby stations, with each observation weighted by the inverse of its distance from the target. In the solar radiation prediction model, it is applied to incomplete weather time series data to fill gaps, restoring a complete dataset that can be used to train and test the AI prediction model. The inverse coefficient n in the distance weighting is commonly set to 2.
According to the chapter, which AI model (ANN, SVM, or RF) performed best in both training and testing phases for solar PV forecasting, and what were its RMSE values for training and testing?
According to the chapter, ANN performed best in both the training and testing phases for solar PV forecasting, with RMSE values of 1.4568 for training and 1.9874 for testing.
Chapter 6: Two-Channel Convolutional Neural Networks for Rolling Bearing Fault Diagnosis in Unbalanced Datasets
How does TC-CNN maintain high fault diagnosis performance even when the ratio of healthy to faulty samples reaches 50:1, and what happens to other methods under such imbalance?
TC-CNN keeps high fault diagnosis performance at a 50:1 healthy-to-fault ratio by combining FFT-based frequency spectra and GST-based time-frequency images in two parallel CNN branches. Fusing these complementary feature sets enriches fault-related information, making the minority fault samples easier to identify even when very few are available. Other methods degrade sharply under this imbalance, producing low precision and F1 scores because they cannot learn enough from the small number of faulty training samples.
Why does TC-CNN outperform CWT+2D-CNN in fault diagnosis, especially under high imbalance ratios?
TC-CNN outperforms CWT+2D-CNN under high imbalance because it fuses FFT-based frequency features and GST-based time-frequency features from two channels. This complementary, multi-dimensional feature set lets TC-CNN extract enough fault information from very few fault samples, so its precision and F1 stay nearly stable even as the imbalance ratio rises from 6:1 to 51:1, whereas CWT+2D-CNN's performance drops sharply as fault sample counts shrink.
Chapter 7: The Implementation of Artificial Intelligence for Auto Gearbox Failure Detection
Chapter 8: Evolutionary Algorithms to Optimise Deep Learning Model for Water Industry Forecasts
What are the limitations of mechanistic models in wastewater treatment, and how do data-driven models like deep learning address these limitations?
Mechanistic models in wastewater treatment are based on mass conservation and biochemical rate equations, which approximate aggregated biological groups. They are useful for design and interpretation but cannot capture all behaviors, especially in complex, data-rich systems or when governing mechanisms are unknown. Data-driven models like deep learning address these limits by learning directly from large sensor datasets, and architectures such as Bi-LSTM can capture both short- and long-term process dynamics, improving effluent quality prediction and nutrient removal forecasting.
How does the chapter's GA approach handle the optimization of hyperparameters and architecture for the Bi-LSTM model?
The GA encodes the Bi-LSTM's hyperparameters and architecture as floating-point chromosomes and uses mean squared error (MSE) on the training data as the fitness function. Each generation applies selection, crossover, mutation, and elitism to evolve batch size, sampling interval, learning rate, per-layer neuron counts, hidden layer count, optimizer, and activation function. Optimizers and activations are float-coded and mapped to discrete options, and later hidden layers can be dropped when their encoded value is negative. Population-size tests showed that about 50 individuals is best for avoiding premature convergence.
Chapter 10: Artificial Intelligence and Internet of Things-Based Intelligent Scheduling for Load Distribution in Power Grids
According to the experimental results, by what percentage does the proposed source–grid–load–storage collaborative scheduling reduce the average load adjustment compared to no coordinated scheduling and compared to the existing mechanism that only considers energy storage?
The proposed source–grid–load–storage collaborative scheduling reduces the average load adjustment by 50% compared to no coordinated scheduling and by 40% compared to the existing mechanism that only considers energy storage.
What are the two main methods for peak shaving and valley filling in power grids, and how does UPIoT enable them to work together?
The two main methods are user-side policy measures such as time-of-use pricing that encourage voluntary load shifting, and coordinated dispatch of flexible loads and energy storage devices that discharge during peaks and charge during valleys. Under UPIoT, data barriers between the grid and users are broken, so real-time user consumption intentions can be fed into the distribution node and combined with controllable loads and storage in a single coordinated dispatch, allowing both methods to work together more effectively.
Chapter 11: Coordinated Response Strategies: Swarm Robotics for Crisis Management
What are the core capabilities of swarm intelligence for crisis management as listed in the chapter?
The chapter lists ten core capabilities: search and rescue; surveillance and listening; communication and networking; logistics and delivering supplies; response to natural disasters; epidemic control; resource allocation and deployment; routing and logistics; collecting data; and real-time reporting.
How does the Glowworm Swarm Optimization (GSO) algorithm balance global discovery and local exploitation?
GSO balances global discovery and local exploitation by combining random movement with attraction. Random movement lets glowworms explore the broader search space, while brighter glowworms attract others, causing them to group around promising solutions and fine-tune locally.
Chapter 12: Smart Farming and Human Bioinformatics Systems Based on IoT and Sensor Devices
What are the four layers of the basic IoT architecture as described in the chapter?
The chapter describes the basic IoT architecture as consisting of four layers: the sensing layer, network layer, processing layer, and application layer. The sensing layer gathers data from the environment using sensors and actuators. The network layer provides communication and connectivity among IoT components. The processing layer analyzes and interprets data, and the application layer interacts with end users.
What are the five major components of IoT architecture as listed in the chapter?
The chapter lists five major components of IoT architecture: device, gateway, cloud, analytics, and UI (user interface).
How do soil sensors and nutrient sensors contribute to precision crop management?
Soil sensors and nutrient sensors gather field data on moisture, temperature, nutrient content, and pH. Farmers use that data to adjust irrigation and fertilization schedules, avoiding overwatering or nutrient imbalances. Nutrient sensors track nitrogen forms and potassium, helping farmers optimize fertilizer use for healthier crops and better yields.
Chapter 13: Machine Learning Techniques Applied in Predictive Maintenance: A Review
What are the five consecutive stages of the intelligent PdM planning model introduced by Mustufa Haider Abidi et al., and what was the purpose of using SVM and RNN in that model?
The five consecutive stages are data cleansing, data normalization, optimal feature selection, precision network decision-making, and prediction. In the model, SVM was used to identify a network that could handle prediction values with different ranges, while RNN was employed to provide predictions throughout a variety of ranges, with the J-SLnO algorithm optimizing the RNN's weights.
How does the PdM method described in the text enable dynamic decision rules for maintenance management, and what is the role of classification modules with different time horizons?
The PdM method enables dynamic decision rules by allowing maintenance actions to be adjusted based on real-time, data-driven information rather than fixed schedules. Classification modules are trained with different prediction time horizons, each producing distinct tradeoffs between the frequency of unplanned stoppages and unused equipment lifetime. The operating cost-based decision system then uses those results to select maintenance actions that minimize expected costs.
Chapter 14: Optimization of Parameters During Tribological Investigations on Azadirachta indica-Based Bio-Composites
Chapter 15: ANFIS Modelling Study on Surface Water Analysis
How does the ANFIS model combine ANN and fuzzy logic, and what water quality parameters were used in the study mentioned?
ANFIS combines an artificial neural network's learning structure with fuzzy logic rules, using a fuzzy if-then inference system embedded in an ANN architecture to handle nonlinear and uncertain water-quality data. In the Kanchipuram temple pool study from 2017 to 2020, the parameters modelled were total dissolved solids, pH, chloride, sulphate, dissolved oxygen, the heavy metal iron, and total coliform count.
What are the five layers in the ANFIS model and what is the function of each layer?
ANFIS has five layers: fuzzification, output of the product of membership values, normalization of firing strengths, Takagi-Sugeno rule computation, and final summation. Layer 1 converts each input into linguistic membership grades using membership functions. Layer 2 multiplies the incoming membership degrees to create firing strengths. Layer 3 normalizes those strengths. Layer 4 computes each rule's weighted consequent using Takagi-Sugeno if-then rules. Layer 5 sums all weighted rule outputs to give the final ANFIS output.
Chapter 16: WSN-Based Optimal Crude Oil Storage Health Monitoring Framework
What are the two main problems faced by probabilistic decision-making frameworks for real systems in WSN health monitoring, and how does the proposed PSS framework address them?
The two main problems are a lack of sufficient detection data, particularly sensor failure data, and the complexity of the monitored system, which makes sensor knowledge fuzzy, incomplete, and difficult to use. The proposed PSS framework addresses these by combining quantitative detection data with qualitative sensor knowledge in a probabilistic rule base, allowing it to handle ambiguity, uncertainty, and incomplete information and to work effectively despite limited failure data and complex system operations.
In the WSN-based crude oil storage health monitoring framework, what are the reference values for the health status (HS) attribute at the points M, TM, N, TG, and G?
For the HS attribute, the reference values are: M = 0.3696, TM = 0.4, N = 0.42, TG = 0.44, and G = 0.45.