Sustainable Organizational Performance, Cyber-Physical Production Networks, and Deep Learning-assisted Smart Process Planning in Industry 4.0-based Manufacturing Systems.

• Author: Kovacova, Maria

1. Introduction

   As a result of the shortage of instantaneous assessment and precise prediction of manufacturing performance (Ciobanu, 2021; Kliestik et al, 2021; Lyons and Lazaroiu, 2020), the production customization requirements are frequently released after inconsistencies occur, and decisions are typically made derived from historical data (Lazaroiu et al., 2021; Popescu et al., 2017), resulting in discontinuity or performance decrease. (Wang et al., 2021) Sensing and computing technologies optimize the performance and time management of integrated assembly systems. (Cohen et al, 2019)

2. Conceptual Framework and Literature Review

   Internet of Things-based decision support systems can configure a cloudedge networking across a smart factory environment, with smart resources through distributed monitoring capacity, and cloud center and edge resources can interconnect fluidly towards production performance prediction by extracting real-time manufacturing data. (Wang et al, 2021) Because of industrial big data and intricate process nonlinearity, the balance between modeling rigor and computation inconvenience should be harmonized. (Qian et al., 2021) By articulating the smart factory, the organizational manufacturing competence can be optimized. (Jo, 2021) Sustainable Industry 4.0 wireless networks facilitate improved, streamlined, and heterogeneous throughput of large-scale production assembly lines (Andronie et al, 2021a, b; Hamilton, 2021; Lazaroiu et al, 2020; Paskaleva and Stoykova, 2021; Valaskova et al, 2021), enabling the manufacturing of items on the assembly system and straightening out material stream to a virtual assembly line by use of control algorithms. (Cohen et al, 2019) Networking across the organizational structure and throughout the value chain furthers the networking of physical and virtual undertakings (Du?manescu et al, 2016; Konhäusner et al, 2021; Mircica, 2020; Valle, 2020; Vatamanescu et al, 2021): interoperability supports the implementation of smooth production, through links between manufacturing systems and transfer of knowledge and skills. (Cugno et al, 2021)

3. Methodology and Empirical Analysis

   Using and replicating data from Capgemini, the Economist Intelligence Unit, McKinsey, Management Events, and World Economic Forum, we performed analyses and made estimates regarding how data-driven supervision, predictive analytics, and optimization systems integrate product traceability, manufacturing maintenance, and process performance in smart manufacturing. Descriptive statistics of compiled data from the completed surveys were calculated when appropriate.

4. Study Design, Survey Methods, and Materials

   This survey employs statistical weighting procedures to clarify deviations in the survey sample from known population features, which is instrumental in correcting for differential survey participation and random variation in samples. Throughout the research process, the total survey quality approach, designed to minimize error at each stage as thus the validity of survey research would be diminished, was followed. At each step in the survey research process, best practices and quality controls were followed to minimize the impact of additional sources of error as regards specification, frame, nonresponse, measurement, and processing. The interviews were conducted online and data were weighted by five variables (age, race/ethnicity, gender, education, and geographic region) using the Census Bureau's American Community Survey to reflect reliably and accurately the demographic composition of the United States. Independent t-tests for continuous variables or chisquare tests for categorical variables were employed. Results are estimates and commonly are dissimilar within a narrow range around the actual value. If a participant began a survey without completing it, that was withdrawal of consent and the data was not used. To prevent missing data, all fields in the survey were required. Any survey which did not reach greater than 50% completion was removed from subsequent analysis to ensure quality. The data was weighted in a multistep process that accounts for multiple stages of sampling and nonresponse that occur at different points in the survey process. Test data was populated and analyzed in SPSS to ensure the logic and randomizations were working as intended before launching the survey. To ensure high-quality data, data quality checks were performed to identify any respondents showing clear patterns of satisficing (e.g., checking for high rates of leaving questions blank). Sampling errors and test of statistical significance take into account the effect of weighting. Question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of opinion polls. The sample weighting was accomplished using an iterative proportional fitting process that simultaneously balanced the distributions of all variables. Stratified sampling methods were used and weights were trimmed not to exceed 3. Average margins of error, at the 95% confidence level, are +/-2%. The design effect for the survey was 1.3. For tabulation purposes, percentage points are rounded to the nearest whole number. The precision of the online polls was measured using a Bayesian credibility interval. Confirmatory factor analysis was employed to test for the reliability and validity of measurement instruments. Addressing a significant knowledge gap in the literature, the research has complied with stringent methodology, reporting, and data analysis requirements.

5. Statistical Analysis

   To ensure reliability and accuracy of data, participants undergo a rigorous verification process and incoming data goes through a sequence of steps and multiple quality checks. Descriptive and inferential statistics provide a summary of the responses and comparisons among subgroups. An informed e-consent was obtained from individual participants. Study participants were informed clearly about their freedom to opt out of the study at any point of time without providing justification for doing so. All data were interrogated by employing graphical and numeric exploratory data analysis methods. Descriptive analyses (mean and standard deviations for continuous variables and counts and percentages for categorical variables) were used. Descriptive statistical analysis and multivariate inferential tests were undertaken for the survey responses and for the purpose of variable reduction in regression modeling. Multivariate analyses, and not univariate associations with outcomes, are more likely to factor out confounding covariates and more precisely determine the relative significance of individual variables. An Internetbased survey software program was utilized for the delivery and collection of responses. Panel research represents a swift method for gathering data recurrently, drawing a sample from a pre-recruited set of respondents. Only participants with non-missing and non-duplicated responses were included in the analyses. Individuals who completed the survey in a too short period of time, thus answering rapidly with little thought, were removed from the analytical sample. Behavioral datasets have been collected, entered into a spreadsheet, and cutting-edge...