Journal of Quality Engineering and Management

Abstract Determining the Factors Influencing the Prediction of Helicopter Rotor Failures

Purpose: The purpose of this paper is to investigate and identify the variables that influence the occurrence of helicopter accidents caused by different types of rotor failures. These crucial factors include flight conditions, maintenance conditions, and helicopter configuration. With this approach, accidents can be investigated more effectively and flight safety can be significantly improved.

Methodology: By analyzing 135 rotor faults accident from a comprehensive dataset containing 5652 helicopter-related accidents, eight classes of rotor faults were identified. Based on expert surveys and a review of studies in the field of helicopter accidents, nine features were proposed as crucial factors to such accidents. The significance of these factors was assessed using five feature selection methods. The input features included maximum takeoff weight, flight hours since the last inspection, type of last inspection, engine power, flight hours, altitude, wind speed, wind direction, and flight phase. Five well-known feature selection techniques—Correlation Matrix, Extreme Gradient Boosting (XGBoost), Mutual Information, Deep Learning, and Neural Network—were employed to identify the most essential factors.

Findings: "Maximum weight", "helicopter engine power", "flight phase" and "flight hours" were identified as variables with the highest degree of importance in predicting faults class of helicopter rotor, which also have a strong and acceptable justification in flight mechanics.

Originality/Value: The distinction of the present study from similar works lies in the inclusion of a broader range of variables, such as flight conditions and helicopter configuration, in contrast to previous studies that considered only a limited set of variables. By prioritizing these variables, the findings pave the way for proactive measures to prevent rotor faults, aiming to enhance prediction accuracy, reliability, and flight safety.

Abstract In this study, the usual kriging method as a linear statistical estimator and two intelligent methods of artificial neural network ANN and adaptive neural fuzzy inference system ANFIS were evaluated in predicting the amount of electric conductivity and nitrate in groundwater. In order to conduct studies, nitrate concentration in 40 wells in Lanjanat plain of Isfahan was measured by spectrophotometer and electrical conductivity. The input data of the artificial neural model, including the length and width of the geographies, the nitrate concentration, and the electrical conductivity value were determined as the output of the model. In order to investigate the performance and efficiency of artificial intelligence models in predicting qualitative information, qualitative information of 50% of the wells was used for calibration and 50% of the wells were used for validating the models. Finally, the output of the models was compared with the value measured in the observation wells based on the mutual error evaluation criteria. The results showed that the ANFIS model performed better than the other two interpolation models in predicting the value of electrical conductivity and nitrate, respectively, with the root mean square error (RMSE) and (mg/l) of 5.362, with the mean bias error (MBE) 2.365 with a correlation coefficient (R) of 0.767. Also, the ANN model had far better results than the usual kriging method. Based on this, ANFIS model is proposed for spatial prediction of electrical conductivity and nitrate in the study area.

Abstract In this research, a hybrid approach based on fuzzy logic and artificial neural network is presented to predict the failure of machines in order to increase productivity. The subject of this research is one of the factories of the automobile industry named Diaco Ide Aria, which operates in the field of automobile parts production. Preventive maintenance requires correct prediction of breakdowns and accidents, equipment and machines so that productivity can be increased by timely and correct maintenance of machines as well as fixing defects and breakdowns. To model the multi-layer perceptron fuzzy-neural network (MLP), first, 100 failures and stops were collected in a period of 15 months and then entered into MATLAB software. The obtained results show that the implementation of fuzzy-neural network and the prediction of machine failure time has reduced the duration and cost of repairs. Therefore, the working time and accessibility of the machines increased and ultimately increased the productivity by 57%, also, the accuracy of the developed neural-fuzzy model was estimated at 94%.

Abstract  The issue studied in this paper, considering redundancy allocation and buffer allocation problems simultaneously in a series-parallel production system. The purpose of this research is to improve the availability, total system costs and buffer capacity through determining the optimal buffers size between work stations, selecting high reliability machines and assigning them to work stations, and developing a proper maintenance and repair plan. In this paper, the preventive and emergency repairs to machines are allowed and their cost is considered in the cost function. Furthermore, it is assumed that machine failure rates are random and follows a distribution function such as Weibul. Given these assumptions, it is very difficult to obtain the availability and cost functions via mathematical relations, explicitly. Thus, a hybrid simulation, design of experiments, and neural network approach are applied to estimate the availability and cost functions. In order to analyze the proposed model, a numerical example was used
 and based on the proposed methodology, was analyzed and evaluated. The model related to the problem was coded and solved by the NSGA-II algorithm and the Pareto set of answers was obtained. The results of the research indicated the validity of the proposed methodology for the problem under study. 

Abstract The main purpose of this study is to identify the causes of vibration and detectable defects of bearings through sonometry using a multilayer neural network. Neural network is an intelligent method and due to its main properties, ie its high ability to estimate nonlinear functions and adaptive learning, it has been used to troubleshoot mechanical vibrations of machines, ie bearing acoustics and their frequency analysis. To collect the data, a type of healthy ball bearing cone bearing and a similar bearing with defective bullets were used and tested in desktop drills and radial-based drills in 5 different rounds. In this study, according to the network with 10 hidden layers, the signal frequency is considered as the input of the multilayer neural network and finally the bearing defects and its probable cause are determined and corrective measures are proposed.

Abstract In most cases, the alert received from a control chart does not indicate the actual time of the process change due to the delay between the actual change time and the time of receiving the alert from the control chart. As a result, it is necessary to examine the real time of change, which is referred to as the "point of change". By reviewing the literature on identifying real-time process changes, it can be concluded that most research in this field focuses on univariate processes and little research is devoted to multivariate processes. In addition, most research in the field of estimating change time in multivariate processes has focused on changes in the mean process vector, and only one research has been done on the covariance matrix. In this paper, a model based on artificial neural network is proposed to estimate the point of change in the covariance matrix of multivariate normal processes. The method presented in phase 2 is control diagrams and the type of change that occurred in the variance of qualitative characteristics is assumed to be the type of step changes. The performance of the proposed method in estimating the change point is evaluated based on two criteria of experimental distribution of estimates as well as the mean and standard deviation of the change point estimator for different step shifts in the variance of process variables in a simulation study. Finally, in order to further explain the proposed method, a numerical example is provided. The results show the proper performance of the proposed method in estimating the change point in the covariance matrix of multivariate normal processes.