جستجوی مقالات مرتبط با کلیدواژه « هوش مصنوعی » در نشریات گروه « مکانیک »

In the upcoming research, using deep neural networks, it is used to predict the maximum yield of circular metal sheets under uniform dynamic load. The neural network presented in this research was designed in the Python programming language and using the libraries available in it, including Tensorflow. The network is based on the regression problem and is of sequential type and includes 10 hidden layers which are the activation function in neurons of Leaky RELU type. The network optimizer algorithm was set to Adam and the objective function of the MSE problem and the number of network iterations was set to 700 times. The data set used in this article consists of 581 samples obtained from 16 series of experiments during the last forty years, which were standardized by the Scikit_learn library. The metal sheets are of 4 types: steel, aluminum, copper and titanium, and there is no separation between different metals.  The number of training data in the model was determined to be 443 equals to 75% of the data set. Also, the number of experimental and evaluator data was selected as 88 numbers equivalent to 15% and 50 numbers equivalent to 10% of the entire data set. Each sample has 8 features as neural network inputs and one label as output. The presented intelligent model among the 88 test data that was completely randomly selected from the data set was able to classify 76% of the data, approximately equivalent to 67 numbers, within the error range of less than 10% and 88% of the data, or in other words, equivalent to approximately 78 numbers within the error range. Predict less than 20%. The amount of the root mean square error index decreased 102 times compared to the analytical and traditional predictive relationships available in the research records. Also, the coefficient of determination criterion, which is an important indicator for evaluating the performance of neural networks based on regression problems, includes the value of 0.96.

In this study, a comprehensive analysis of driving behavior with an emphasis on fuel consumption and driver categorization is presented. Data from 80 drivers were collected using a custom-designed datalogger connected to the vehicle’s On-Board Diagnostics (OBD) port. Critical features related to driving patterns were extracted through a correlation matrix and concepts in the field of powertrains. Key variables such as acceleration and deceleration were identified and derived. Regression models were applied to predict fuel consumption based on this driving feature. Through this analysis, the most influential factors affecting fuel efficiency were highlighted. Additionally, unsupervised machine learning techniques were employed to cluster drivers into distinct groups based on their driving styles. A comparative study of various algorithms was conducted to evaluate the efficacy of different clustering methods. Valuable insights for automotive manufacturers, policymakers, and drivers are offered by the results, emphasizing the role of driving behavior in fuel efficiency and the potential for tailored driver assistance systems.

The purpose of this research is to design and build a puppet robot with the capability of deep conversational learning with a deep reasoning approach. Due to the mentioned advances in artificial intelligence and deep learning, it is possible to provide conditions where the puppeteer's dependence on the handler is reduced and communicate with the audience intelligently. This robot, by recognizing the Farsi speech of the audience, determines a suitable answer to the question and broadcasts it in Persian language. The importance of this issue is in creating a suitable dialogue mechanism. This mechanism is a deep learning algorithm that by recognizing the question posed by the user, provides a set of possibilities from the categories included in the robot's dataset, and considering the highest probability, the desired category in which the user's question is placed is determined, and among the answers For that category of questions, an answer is chosen randomly. In addition, the puppet robot dialogue mechanism has a few simple conditional sections that can provide appropriate answers to repetitive or obscure questions. In the results of various trainings, by changing the parameters in the deep learning model of this robot with a 64-class dataset, it was found that the use of crowded layers with many neurons works better than a few layers of them,

The purpose of condition monitoring is to monitor the conditions of an asset in order to predict its failure. The first step in implementing condition monitoring is to establish a health indicator. To construct the health indicator, operational data such as vibrational data should be collected from the asset during its operation, followed by extracting meaningful features from the data.This study introduces an artificial intelligence model (convolutional autoencoder) for feature extraction from the vibration data, which only requires healthy status data for training. For this purpose, the vibration data of the asset is gathered online during operation in a healthy state to establish the healthy dataset. Then, the deep learning model is trained by the healthy dataset. Finally, after the failure stage is detected, the health indicator is established by measuring the differences in the vibrations of the healthy and the failure conditions. The performance of the proposed model is evaluated by vibrational data of a gearbox. The health indicator exhibits a monotonically increasing degradation trend and has good performance in terms of detecting incipient faults.

Artificial intelligence refers to algorithms that can have correct and logical reactions as well as behaviors similar to intelligent human behaviors; including understanding complex situations, simulating human thought processes and reasoning methods, successfully responding to them, learning and the ability to acquire knowledge and reason to solve problems. Artificial intelligence is a kind of knowledge of knowing and designing intelligent agents and its scope of application is global. Artificial intelligence is a vast field of intersection of many knowledges as well as a connecting link of old and new sciences and techniques. Artificial intelligence is undoubtedly a new phenomenon in the field of science and technology, which has had a great impact on the field of naval military equipment. Artificial intelligence, like any other technology, has advantages and disadvantages, and as a result, many opportunities and threats. In this article, after stating the introduction and presenting the problem plan and its description, the opportunities of using artificial intelligence and data mining to improve the performance of electronic warfare systems in the maritime domain are discussed. Then, the advantage of using data mining with artificial intelligence compared to other methods to identify cyber-electronic threats in the maritime domain is discussed. Finally, the identification phases of cyber-electronic threats in the maritime domain are introduced by combining artificial intelligence and data mining. In the end, after the solutions and proposals are proposed, discussions, summaries and conclusions are presented.

In a holistic design; Creating a common language between different engineering, taking in to account the designer/customer's interests and preferences, removing potential shortcomings of common design methodologies and taking full advantage of the benefits of MDO approaches are the most important factors for achieving logical and realistic results. In this paper, a novel Comprehensive Preference-based Design approach is presented which attempts to achieve subjective attributes that are defined in the concept of maximization of designer/customer's satisfaction in addition to objective goals which are formulated in the form of minimization of a performance criterion in a two-phase structure using two nested optimizers. In the first phase of CPD, using the concept of satisfaction, the subjective preferences of the designer/customer are defined in terms of fuzzy relationships and operators in the form of demands and desires. Whereas the results of this phase are inaccurate, in the second phase, it is attempted to define a performance criterion and in order to achieve an optimal operational plan, attitude parameters and the compromises needed to meet the designer/customer's preferences are implemented. The methodology is utilized to design of a UAV with flight duration of 24 hours, 45 m/s of cruise speed at least, payload of 200 kg and less than 1 km take-off distance. To evaluate the results of CPD, the design process using MDO in AAO framework is also performed. Comparison of the results shows that despite the higher mass of UAV designed by CPD, overall satisfaction is higher and preferences have been satisfied.

Today human or person detection in various environments is important for smart robots and autonomous vehicle in Artificial Intelligence and robotics. In this paper a new method for human detection using two Laser Range Finder (LRF Sensor) that implemented on servicing at home robot explained. This proposed approach is extracting a curve state from laser scanner data and fusion of Legs detection and Torso detection. For visualization, used Rviz (ROS Visualizer) in ROS (Robotic Operating System) middleware. The results have been shown accuracy and efficiency of proposed algorithm.

In industrial rotatory machines, different forces in rotor bearings are generated due to various impaired mechanical sources, namely bearing misalignment and nonhomogeneous mass distribution (unbalance). By precisely analyzing and diagnosing the produced patterns of bearing forces, one can determine the unbalance parameters such as quantities of masses, their distance from the rotational axis, and characteristics of corresponding parallel planes. Consequently, it will be possible to formulate pragmatic protocols according to which the maintenance engineers of rotatory systems will pinpoint properties of problematic imbalance masses and then straightforwardly balance them. In the procedure of conducting this research, several exemplary imbalance masses are deployed on a rotatory mechanical shaft and the equations of motion and forces in perfectly aligned rigid bearings are extracted. Then, by applying a neural network-oriented system the patterns of bearing forces are recognized and the characteristics of the nominal masses including magnitudes, distances from the rotational axis, angles as well as the unbalance type are determined. The accuracy of predicting 8 variables of balancing masses was 41% and after eliminating the redundant overlaps from principal components, the accuracy of predicted 5 variables of balancing masses significantly increased to 95%. Also, by implementing another comprehensive neural network system, it was shown that by exerting two separate balancing masses, the applicability of this method in balancing any faulty systems with dynamic unbalance is possible.

In recent years, with the advent of the Fourth Industrial Revolution concepts and the development of artificial intelligence technologies, new approaches such as the digital twin have been introduced. In a digital twin, a virtual counterpart of the physical system during its whole life is created, with abilities such as analyzing, evaluating, optimizing, and predicting. The first step in creating a digital twin model is to construct a (multi) digital health indicator that describes different aspects of the physical component state during the whole life of the component. In this research, a new method for constructing health indicators based on vibration measurement and a deep learning model has been introduced. For this purpose, the Continuous Wavelet Transform was used to convert the raw vibration signals into two-dimension images; Then, the deep learning model was used to extract features from the images and the health indicator is constructed based on the differences of the images in normal and failure stages. In this article, various Autoencoder architectures are discussed, and it is demonstrated that the Convolutional Autoencoder has better performance in terms of detecting incipient faults. The performance of the proposed model is evaluated by the vibration data of the bearing, and the constructed health indicator exhibited a monotonically increasing degradation trend and had good performance in terms of detecting incipient faults.

n GEO communicational satellites, thrusts are being used for many missions such as station keeping, longitude change maneuver and actuators desaturation. These types of actuators need fuel and its estimation requires many complicated calculations. In this paper, a novel method based on availed data for previous satellites is proposed for estimation of satellite fuel mass and it does not need mathematical modeling of satellite dynamics. Two methods, least square method and artificial intelligence, are used to this aim and two mathematical model are proposed for satellite fuel mass estimation. By applying the models to several previous satellites, it is shown that the models have lower than 5% average error. The proposed method in this paper is quick and accurate and can be utilized for GEO satellites feasibility study and conceptual design

The integrated navigation contains Inertial Navigation System (INS),Global Positioning System (GPS) is an important methodin moving vehicles such as ground, sea,aerospace vehicles navigation. This study contains the introduction of INS, theintroduction of GPS, the integration method of above systems,important issues in this field. For these purposes, the mostimportant researches of Iranian,overseas scientists are studied. Besides of issues such as using a Kalman Filter (KF), ExtendedKalman Filter (EKF), Particle Filter (PF),other methods based on KF,PF, the new approaches are also surveyed. The studyresults show that recently using artificial intelligence method in integration, the problem outage or block of GPS signal in integrationand low-cost inertial navigation sensors have been the focus of attentions.

In this paper, a new approach for orbital position prediction of satellites, is introduced. As traditional methods are based on keplerian equations of motion, orbital disturbances are uasualy neglected for simplicity. This paper, suggests artificial intelligent time series peridiction methods for orbital position prediction of satellites. The advantage of this method is based on usage of actual data, so all disturbances are taken into account. For this reason use of TLE as the most reachable actual data is considered. Compariosion of output of this method with actual data, proofs the accuracy of proposed method.

In this paper a new method based on temporal difference on reinforcement learning algorithm is proposed to active noise control of periodic acoustic signal. This method does not need dynamic information and therefore it is fully robust with respect to the dynamic change. Very low computational burden and low needed memory are some advantages of this method. In the first step، with well definition of the state variables، actions، and reward signal، a reinforcement learning problem is formed and then it is solved by Q-learning technique. In the next step the problem is simplified by help of frequency domain modeling information and it is solved by temporal difference method. Finally، a multi-level approach is presented which it can increase the precision without increasing of the memory size. It is shown by simulation that this method works well and the improvement of the multi-level temporal difference method with respect to the Q-learning method is addressed.