Hybrid micro and nanocomposites reinforced with 1.5 weight
Early treatment through brain tumor diagnosis via medical imaging is crucial, while deep learning approaches in the area suffer the most from issues like overfitting, lack of interpretability, and computational costs. To address these issues, the lightweight multi-modal autoencoded quantum dilated attention convolutional neural network (LMAQDAC), a lightweight model that enhances tumor detection by combining in a single efficient pipeline segmentation, feature extraction, and classification, is presented. The modified Hiking optimization algorithm (MHOA) further optimizes this method in terms of training speed and stability. Interpretability is improved by Gradient-weighted Class Activation Mapping (Grad-CAM), which highlights tumor regions, thus supporting clinician trust in the AI method. Model compression techniques are employed, enabling real-time performance, drastically reducing computation and memory overhead. This allows for deployment on small edge devices, thus providing localized, fast, and secure inference. The proposed method has 99.96
The unified power quality conditioner (UPQC) makes use of a fuzzy logic sliding mode controller (FLSMC) in order to enhance the reliability and quality of the distribution power system. This is accomplished by addressing dynamic performance and power quality issues such as current disturbances, voltage sag, swell, and total harmonic distortion (THD) under nonlinear loads. When compared to traditional controllers, FLSMC-based UPQCs perform better in terms of dynamic performance and power quality. In order to reliably extract reference current and voltage signals for UPQC, the FSMPWM architecture makes use of sliding surface implementation. Similarly, control principles are derived for shunt converters and series converters. The Mamdani fuzzy rule basis for switching pulse generation was meant to be implemented at the sliding surface. Chattering is eliminated, and a fixed switching pulse is produced for shunt and series converters through the utilization of the method that has been provided. The results indicate that when a FLSMC controller is utilized, the total harmonic distortion (THD) values of the source voltage are 0.38%, and the THD values of the source current are 2.01%. Additional evidence demonstrates that the compensation utilized by this controller is successful for all parameters. As a result, the FLSMC controller is the most effective of the four options that have been offered. The FLSMC is superior to the FLC, ANFIS, and FOFLC in terms of its ability to increase dynamic performance and power quality. MATLAB/Simulink is used to actually implement the controller-based system architecture for the FOFLC and FLSMC applications. MATLAB/Simulink, Power Quality, and Fractional Order FLC, as well as Fuzzy Sliding Mode Controller, are some of the keywords that are associated with this topic.
Tomatoes face a range of diseases that can really impact both quality and quantity. Relying on visual inspections, which is what traditional diagnostic methods typically do, can be a bit of a drag. It is time-consuming and few faults may happen. To make diagnosing tomato leaf diseases a bit easier, this work proposes a deep learning-based model built on sequential models that's great for identifying multiple classes of diseases. The model is trained using a kaggele dataset containing pictures of both healthy and infected tomato leaves with spots, making it capable of detecting widespread conditions like Alternaria Solani, Phytophthora Infestans, and Leaf Mold. Several optimizers Adam, SGD, and RMSprop are used to optimize its performance and enhance the speed of the Process. In this work, examined different deep learning models and observed that CNN + Resnet18 model with adam optimizer given $\mathbf{9 8. 1 2 \%}$ accuracy.
The research examines the implementation of an IoT AI-driven Predictive Maintenance (PdM) system to be used in manufacturing premises to forecast and prevent equipment failure. This system is equipped with advanced IoT sensors that monitor critical parameters in real time, including temperature, vibration as well as pressure. All data are further analysed with edge computing and cloud computing for additional and detailed analytics and storage. Therefore, the methodology explains the system’s architecture in detail, including thorough data preprocessing and AI model elaboration. Six months of experimental validation and a couple of cases of motors and pumps substantiate the expected outcomes of the system’s functioning. Eventually, the neural network model was the most efficient, with an accuracy of 92% along with an F1-score of about 89% as compared to the remaining models, whose accuracy was barely above 80%. The major findings prove there is a possibility to forecast the failure to schedule a time for repair and hence reduce the unplanned downtime and repair costs. Thus, the motor and pump cases demonstrate that the repairs of $10,000 and $8,000 were indeed avoided, respectively. However, it is also necessary to acknowledge certain issues, including data integration and the necessity for highly qualified personnel. Thus, the study recommends possible directions for further development of AI models and expansion of potential perspectives for use across various industrial equipment. Thus, the research provides an opportunity to employ modern IoT and AI perspectives in enhancing the efficiency of maintaining equipment in a manufacturing plant.
