Compensation-based methodology for maintenance time prediction in a virtual environment
Introduction
Maintenance time is a fundamental quantitative parameter in maintainability design and has considerable influence on product availability. Thus, maintenance time prediction, which can be conducted in the experiment, design, and manufacture phases, is an important part of the product life cycle. However, maintenance time prediction cannot be performed effectively in actual operation situations, and current maintainability experimentations waste considerable time, manpower, and material resources. Predicting maintenance time at an early stage is important in maintainability design, particularly for new products.
Traditional maintenance time prediction mainly covers six typical methods, namely, the probability simulation method, function level method, sampling prediction method, regression analysis method, time-accumulated method, and comparative analysis method [1], [2], [3], [4], [5], [6], [7]. The probability simulation method predicts the maintenance time systematically according to the maintenance activity distribution. This method can obtain a comprehensive and detailed prediction result but is complicated in calculation and dependent on basic data. The function level method has simple operations that predict the maintenance time according to the existing maintenance activity table and product level table. However, corresponding data are usually outdated and need to be added and amended for new products. The sampling prediction method samples enough replaceable units and assesses the maintenance work by referring to corresponding check tables. This method predicts the maintenance time with empirical equations and obtains increasingly detailed prediction results with product design. The regression analysis method [8] analyzes the relationship between maintenance time and equipment character on the basis of necessary experiments on similar product or statistical data. This method establishes models by using regression analysis to predict the maintenance time for new or improved products. The prediction precision of this method depends on the similarity between existing and new products and the relationship between system design and maintainability design. The time-cumulated method [9] confirms the necessary time of each maintenance project, task, and activity according to historical experience or existing data and diagrams. This method predicts the maintenance time by accumulating or averaging under a given procedure. The basic standard time in this method is obtained mainly from electronic equipment that needs to be amended for other types of products. The comparative analysis method selects a component whose maintenance time is known as a basic reference and predicts the corresponding maintenance time of other units compared with the selected basic reference. This method relies on little design information, thus making this method suitable for various products. The abovementioned traditional methods mostly focus on electronic equipment and extensive application in other types of products, such as mechanical products. These methods lack basic maintenance work time or maintenance activity information. They also have different requirements in data form, high computation complexity, and low operation efficiency.
In recent years, some improved or innovative methods with the characteristics of integration, intelligentization, and visualization have been presented to provide more approaches for maintenance time prediction. First, integration focuses on the combination of existing traditional methods because a single method cannot fulfill the requirements properly to predict the maintenance time of complex products that contain both electronic equipment and mechanical equipment. Various integrations make these improved methods highly adaptable when launching maintenance time prediction for different objects. In actual implementation, proper method selection for integration depends on experience, and an inappropriate integration may lead to an unexpected result. Second, intelligentization provides a new direction in maintenance time prediction. Methods such as language simulation and artificial intelligence are adopted to predict time, thus providing effective approaches for realization. These methods include real-time simulation [10], [11], systematic simulation [12], [13], Monte Carlo simulation [14], and genetic algorithm [15], [16]. Some intelligentization methods, such as the Bayesian network [17], tree network [18], nerve network model [19], fuzzy neural network [20], hidden Markov model [21], and multilayer perception model [22], are also used in software maintainability prediction. However, these methods involve long programming times and have low degrees of visualization. Third, visualization transforms data into graphs or images that display the transformation on-screen on the basis of computer graphics and image processing technologies, thus providing an innovative approach for data processing and representation. A maintenance flow diagram is established to conduct maintainability design analysis on the basis of product structure and possible fault information [23]. The established diagram is also adopted for maintenance process simulation and maintenance time prediction. Petri net technology is used to establish an executable specification tool theory model that combines visualization and formalization for maintenance time simulation and prediction [24]. Colored petri net is used to establish a simulation model for maintenance time prediction by considering the disassembling process sequence and system fault model [25]. Other network models, such as the critical path method [26], graphical evaluation and review technique [27], and generalized reliability analysis simulation program [28], are also presented in maintenance time prediction. In the visualization domain, the continuous development of computer graphics and computer-aided design technologies, particularly virtual reality technology, has increased research interest on product simulation, interaction, and analysis. The advantages of virtual reality systems can be summarized as follows. A virtual prototype provides a visual process to protect product design against hardware limitations. Moreover, a virtual prototype can be operated and tested in a virtual environment to assess product performance, thus decreasing physical prototype requirements and corresponding development costs.
The corresponding historical data of maintenance time is also fundamental in maintainability prediction [29], and several mathematical methods have been adopted for data processing and maintenance time prediction. A maintenance prediction method that consists of graphic evaluation, likelihood parameter evaluation, and Sharpiro–Wilk normal distribution validation is proposed on the basis of the principle of statistical modeling [30]. The demonstration method of maintenance time with random weighted method is also studied [31]. Aircraft maintenance historical data and life cycle cost information are used separately for aircraft design, including maintenance time prediction from the perspective of maintainability [32], [33].
Virtual maintenance technology, as an extension of virtual reality technology in the maintainability domain, provides a creative, intuitive, and holistic way to conduct maintenance task simulation and analysis by a virtual prototype in a virtual maintenance environment [34], [35], [36]. The applications of virtual maintenance for each stage of product design has been increasing recently [37], [38], [39], [40]. Virtual maintenance can intuitively represent the entire maintenance process, including human manipulation, cooperation, and tool operation. Virtual maintenance also provides a reasonable basis for maintainability analysis, such as visibility, accessibility, operation space, and maintenance safety [41]. Therefore, flaws can be exposed during the early stage of product design to avoid the hysteresis of conventional ways.
In the virtual simulation domain, simulation time inaccuracy is universally recognized because of the following reasons. (1) Objectivity: in a virtual simulation environment, the movement principles of virtual humans or virtual products, such as motion frame-based movement, scripting programming-based movement, motion capture-based movement, and reproduced-based movements, exist objectively. Once a virtual platform is selected for simulation, all simulation processes are generated under the movement principle. The movement principle does not fully reflect the corresponding real conditions, such as the pretightening force in bolt assembly, the collision among products, and the flexible component movement. (2) Subjectivity: the simulations generated by different designers significantly vary, even for the same maintenance process. For instance, a human motion can be generated by 5 or 10 motion frames. Both two situations reflect the same human motion, but the durations are clearly different. (3) Diversity: in practical operations, diverse virtual maintenance tools are used in different domains and maintenance processes vary in different virtual maintenance environments. Therefore, the durations from virtual simulation are different. The acquired maintenance process reflects only the maintenance steps following the maintenance procedure, and the performance of the simulation process cannot cover all maintenance activities. Therefore, the simulation time in a virtual maintenance environment is always less than the actual time because of the simplified simulation. Inaccurate time in a virtual maintenance environment clearly causes negative effects to new products in maintainability analysis. A proper method is necessary for developing a reasonable compensation to the simulation time and obtaining an accurate time.
A maintenance process consists of a series of virtual human motions [42]. Consequently, analyzing the inherent relationship between actual operations and human motions is the most significant point for maintenance time prediction. The motion time mechanism was proposed to illustrate that the time required for skilled personnel to finish a certain basic action is constant in actual conditions [43]. This fundamental scientific theory has been improved into a new method called the predetermined motion time standard (PMTS). PMTS, including work factor (WF), methods time measurement (MTM), MTM-universal analyzing system, and modular arrangement of predetermined time standard (MODAPTS), classifies human motions and determines their standard times with a constant value in different ways [44], [45], [46], [47]. PMTS provides a theoretical reference for analyzing the interrelationship between actual operations and human motions. The WF method was proposed by Quick to analyze the relationship between moving time and human body and to summarize the corresponding time for each motion. MTM, which was proposed by Maynad [48], [49], divides human motions into 18 types for time analysis in the ergonomic domain. However, MTM cannot be used directly to describe the maintenance process because of complicated maintenance operations. Heyde [50], [51], [52], [46] reported that MODAPTS classifies human motions into three types, namely, moving, terminating, and assisting, to determine the cycle time in ergonomic design. The abovementioned methods focus on ergonomic analysis and design and are appropriate for measuring the time in line production. Maintenance is becomes increasingly complex and accurate with the logic process; thus, a suitable model is needed for maintenance time prediction.
Based on the aforementioned observations, this study proposes a novel methodology for maintenance time prediction in virtual maintenance environments. To achieve this goal, a time compensation model including three relevant issues is proposed as shown in Fig. 1. First, a virtual maintenance platform is selected as the maintenance process simulation tool, and the corresponding principle for simulation generation is confirmed. The principle must be systematic and regulatory. The simulation process is generated step by step based on the confirmed principle following the corresponding maintenance process schedule. All human motions are shown after the simulation. Second, based on the generated simulation principle, the time compensation principle for virtual human motions is confirmed, including posture adjustment and hand operations. Third, the interrelationship between actual operation and corresponding single virtual human motion in virtual environment is analyzed to obtain an accurate basic compensation time for each motion frame and basic ratio for incomplete operations in virtual maintenance environment. Compensation is then established for all human motions in the entire simulation process. Thereafter, the compensation time of each step is calculated, and the compensation time for the entire simulation is defined. Finally, the simulation time and compensation time are calculated with corresponding equations to obtain the predicted time. The key point of the methodology is the accuracy of the simulation and compensation model. Detailed discussions on these principles are presented in this study.
The rest of this paper is organized as follows. Section 2 shows the details of the principles for simulation generation and time compensation. Section 3 describes the entire maintenance time prediction process. Section 4 presents a case study for the implementation of the methodology. Section 5 concludes the study and presents several discussions.
Section snippets
Human motions in the maintenance process
The maintenance process should be decomposed into reasonable human motions to support the corresponding simulation and analysis. The maintenance process was decomposed into a hierarchical structure at the CAD Center of the University of Ranko [53]. At the System Realization Laboratory of the University of Georgia, an operation process was divided into grabbing, removing, releasing (work with hands), and unfastening (work) [54]. At the US Air Force Research Laboratory, seven basic actions were
Procedure for maintenance time prediction in a virtual maintenance environment
The maintenance time prediction process based on the proposed model in the virtual maintenance environment is conducted as follows.
Maintenance simulation generation
We selected DELMIA as the virtual simulation platform in this case study. A disassembly process simulation for electronic control devices is used for the maintenance time prediction (Fig. 4). The virtual maintenance scene is built by loading necessary product prototypes, including maintenance objects and tools, supporting devices, virtual humans, and other surroundings. Thereafter, single tasks for virtual humans are created by connecting corresponding human motions generated with hand
Conclusion and discussion
This study proposes a maintenance time prediction methodology that is applicable to a virtual maintenance environment. This methodology is developed with the help of the maintenance simulation process and corresponding compensation during the early stage of product design. The case study and other cases in Table 5 show the relative accuracy of the methodology in predicting maintenance time in a virtual maintenance environment. During the prediction in this case study, the basic compensation
Acknowledgements
The authors express their sincerest gratitude to the designers for their support in the practical investigation and data collection. The authors would also like to thank the National Natural Science Foundation of China (Grant No. SKVR-10-17) for financially supporting this research.
References (58)
- et al.
Real-time simulation of time-of-flight sensors
Simul. Model. Pract. Theory
(2009) - et al.
A real-time 3D motion planning and simulation scheme for nonholonomic systems
Simul. Model. Pract. Theory
(2011) - et al.
Optimization of aircraft maintenance plan based on genetic algorithm
Phys. Procedia
(2012) - et al.
An application of Bayesian network for predicting object-oriented software maintainability
Inf. Software Technol.
(2006) - et al.
Desktop virtual reality for maintenance training: an object oriented prototype system (V-REALISM)
Comput. Ind.
(2003) - et al.
Virtual hands and virtual reality multimodal platform to design safer industrial systems
Comput. Ind.
(2007) - et al.
The use of virtual prototyping and simulation in ITER maintenance device development
Fusion Eng. Des.
(2007) - et al.
Virtual manufacturing for press line monitoring and diagnostics
Int. J. Mach. Tools Manuf.
(2008) - et al.
Analysis of the ITER ECH Upper Port Launcher remote maintenance using virtual reality
Fusion Eng. Des.
(2009) - et al.
Employing virtual humans for education and training in X3D/VRML worlds
Comput. Educ.
(2007)
Development of a collaborative virtual maintenance environment with agent technology
J. Manuf. Syst.
A modeling approach for maintenance safety evaluation in virtual maintenance environment
Comput. Aided Des.
The validity of predetermined motion time systems in setting production standards for industrial tasks Original Research Article
Int. J. Ind. Ergon.
ASPEN: computer-aided assembly sequence planning and evaluation system based on predetermined time standard Original Research Article
CIRP Annals – Manuf. Technol.
Computerized predetermined motion-time systems in manufacturing industries
Comput. Ind. Eng.
An approach to assessing motor workload in assembly tasks by the use of predetermined-motion-time systems
Appl. Ergonom.
Computer application of a natural-language predetermined motion time system
Comput. Ind. Eng.
Human factors: A computerized system to measure repetitive motion stress on the lower back
Comput. Ind. Eng.
Combining ergonomics and work measurement for job analysis
Comput. Ind. Eng.
Quantification of the cognitive difficulty of mail sorting
Int. J. Ind. Ergon.
A modern maintainability prediction technique, reliability
IEEE Trans.
Evaluation of support systems utilizing maintainability prediction
IEEE Trans. Aerosp. Electron. Syst.
Maintainability prediction in department of defense development programs
IEEE Trans. Reliab.
Maintainability prediction and demonstration technique development of prediction
Techniques
A multivariate approach to maintenance prediction of wet areas
Construct. Manage. Econom.
Analysis of locomotive maintainability prediction based on time-cumulated method
Electric Locomotives Mass Transit Vehicles
Cited by (19)
Proactive and visual approach for product maintainability design
2023, Advanced Engineering InformaticsA maintenance time estimation method based on virtual simulation and improved modular arrangement of predetermined time standards
2020, International Journal of Industrial ErgonomicsCitation Excerpt :Zhou et al. (Dong et al., 2013) proposed a VM-based maintenance time measurement method, in which a maintenance task is simulated in the virtual environment and decomposed into a series of basic actions, then the maintenance time of the task is estimated based on the approximation of each elementary actions. Geng et al. (2014) proposed a compensation-based methodology for maintenance time prediction in a virtual environment, in which a compensation principle is constructed to make up the time estimation error caused by the difference between the virtual simulation and real maintenance operations. Cai et al. (2019) studies the maintenance processes of assembly and disassembly, and proposed a time prediction method based on a standard posture library and virtual maintenance.
Applications and research trends of digital human models in the manufacturing industry
2019, Virtual Reality and Intelligent HardwareA virtual maintenance-based approach for satellite assembling and troubleshooting assessment
2017, Acta AstronauticaCitation Excerpt :(G1-1) Time: The time factor means the whole duration of all operations in the given possible procedure. Virtual simulation can be used as reference for time prediction [38]. Note that the obtained time in a virtual environment does not have to be equal to the real time.