Showing 3 results for Artificial Neural Network
Seyedeh Reyhaneh Shams, Ali Jahani, Mazaher Moeinaddini, Nematallah Khorasani, Saba Kalantary,
Volume 10, Issue 4 (11-2020)
Abstract
Introduction: As a metropolitan area in Iran, Tehran is exposed to damage from air pollution due to its large population and pollutants from various sources. Accordingly, research on damage induced by air pollution in this city seems necessary. The main purpose of this study was to forecast ozone in the city of Tehran. Considering the hazards of ozone (O3) gas on human health and the environment and its ascending trend over the past decades, it is also essential to study and predict its quantities in the air. Forecasting ozone in the air can be further used to prevent and control pollution by authorities.
Material and Methods: Using an analytical-applied research method, this study was to predict ozone gas in this metropolitan area via daily ozone data of air quality measurement stations, traffic variables, green space, as well as time factors such as one-day time delay. In this regard, an artificial neural network (ANN) model was employed to forecast ozone concentration using the MATLAB software.
Results: The results of the ANN model were compared with a linear regression one. Correlation coefficient and root-mean-square error (RMSE) of the ANN model were subsequently compared with R2=0.734 and RMSE=0.56 as well as R2=0.608 and RMSE=11.69 regression equations.
Conclusion: It was concluded that the error in the ANN model was smaller than that in the regression one. According to the results of the sensitivity analysis of the season parameters, the length of sunshine hours had the most significant effect on the amount of ozone gas in Tehran air.
Elahe Allahyari, Abdollah Gholami, Morteza Arab-Zozani, Hosein Ameri, Negin Nasseh,
Volume 11, Issue 3 (9-2021)
Abstract
Introduction: These days, there is a consensus that emotional intelligence plays an important role in the success of individuals in different areas of life. Persons with higher emotional intelligence had lower stress in dealing with demands and pressures in the workplace. The purpose of this study was to use artificial neural network to predict job stress and to compare the performance of this model with the multivariate regression model.
Material and Methods: In order to do that, 892 participants were selected randomly in different job categories. Then, 15 dimensions of Bar-On questionnaire, 10 job categories, age and education were considered as input variables and 7 dimensions of health and safety executive HSE were determined as output variables in models.
Results: The results revealed that an artificial neural network with hyperbolic tangent and sigmoid transfer functions respectively in hidden and output layers with 375 hidden neurons had significantly better performance than multivariate regression. So that, correlation of predicted values and job stress were only between 0.192-0.364 in regression model, but neural network had at least correlation 0.527 in all dimensions of job stress.
Conclusion: In predicting job stress using emotional intelligence, artificial neural network method was much better than multivariate regression model.
Mojtaba Zokaei, Marzieh Sadeghian, Mohsen Falahati, Azam Biabani,
Volume 13, Issue 4 (12-2023)
Abstract
Introduction: Due to the increase in the provision of electronic services to citizens in government offices, the number of computer users and the occurrence of musculoskeletal disorders have increased. Therefore, this study aimed to predict and model the complex relationships between the risk factors of musculoskeletal disorders in computer users working in government offices by an artificial neural network.
Material and Methods: The current cross-sectional study was conducted in 2020 on 342 employees of various government offices in Saveh city. First, the researcher visited the work environment to identify the problems and measure the environmental factors. Then, ergonomic risk assessment and psychosocial factors were evaluated using the Nordic questionnaire and the ROSA method. The effect of various factors in causing musculoskeletal disorders was investigated using a logistic regression test.Then the resulting data were collected and modeled by one of the neural network algorithms. Finally, artificial neural networks presented an optimal model to predict the risk of musculoskeletal disorders.
Results: The results showed that by increasing the level of social interactions, the level of demand, control, and leadership in the job, musculoskeletal disorders in men and women decrease. There was a significant relationship between the prevalence of musculoskeletal disorders and job demand, job control levels, social interaction levels, leadership levels, organizational climate levels, job satisfaction levels, and stress levels, in addition between reports of pain in the neck and shoulder and wrist/hand region. There was a significant relationship with the overall ROSA score. Also, there was a significant relationship between the report of pain or discomfort in the neck area with the phone screen risk score, wrist/hand with the keyboard-mouse risk score, and shoulder, upper back, elbow, and lower back with the chair risk score. The accuracy of the presented model for predicting musculoskeletal disorders was also about 88.5%, which indicates the acceptability of the results.
Conclusion: The results showed that several factors play a role in causing musculoskeletal disorders, which include individual, environmental, psychosocial, and workstation factors. Therefore, in the design of an ergonomic workstation, the effects of the mentioned factors should be investigated. Also, predicting the effectiveness of each of the mentioned factors using an artificial neural network showed that this type of modeling can be used to prevent musculoskeletal disorders or other multifactorial disorders.
