Sustainable Hyperautomation in High-Tech Manufacturing Industries: A Case of Linear Electromechanical Actuators

Hyperautomation is a promising but sparingly implemented concept in intelligent manufacturing. One of the reasons for the suboptimal adoption of hyperautomation is the large gap between current theoretical frameworks and practical methodologies and tools that can be applied in a real industrial production scenario. This situation has become much more complicated in high-tech enterprises, which face a particular set of issues in terms of innovation, cost-effectiveness, and supply chain management in today’s globalized environment. This manuscript provides a new conceptual business framework and technological background for achieving sustainable hyperautomation in the manufacturing of linear electromechanical actuators (LEMA), a key component of several cyberphysical actuators. A set of digital tools and innovative concepts, such as intra-enterprise 3-level factory and definitive designs based on unified solutions, which enable mass customization and offer up to 1000 variants of the LEMAs, are introduced to achieve synergistic interaction between different business functions and provide significant cost and technological advantages. To make manufacturing more customizable, a modular design approach is used, and simultaneously, to facilitate mass production, the focus is given on roller screw transmission modules, representing approximately three-fourths of the added value of LEMA. Furthermore, the concept of synergetic forward integration is proposed and explained using an example of robotic resistance spot welding. This framework involves a closed loop of industrial mature digital tools that enables autonomous product design and manufacturing via Responsive R&D (Research and Development) and feedback-driven dynamic interactions with the market and production system. These steps allow intelligent and automatic decision making throughout the digitally connected systems within the company and out of the company through a digital networked connected intra-enterprise world inside the supply chain with minimal human intervention.

in the context of sustainable hyperautomation in the high-tech 89 industry. 90 Electromechanical actuators (EMAs) are integral compo-91 nents of CPSs used in various industries. With the grow-92 ing trend towards automation and robotics, the demand for 93 EMAs is expected to grow. However, because of stiff com-94 petition in this segment, there has been a substantial reduc-95 tion in the market price of EMAs. Recent economic crises 96 have adversely affected market dynamics. In the context of 97 decreasing demand, falling market prices, and technological 98 advances, sustainable and intelligent manufacturing of EMAs 99 is only possible by developing a reconfigurable production 100 line for mass customization in small and medium batches, 101 while maintaining technological superiority and production 102 costs comparable to mass production. 103 Automating manufacturing using integrated DTs and pre-104 dictive models allows for iterative real-time product and 105 production process optimization. However, technical com-106 plexity, innovation challenges, rigidities caused by external 107 and internal systems, and turbulent market dynamics pose 108 a substantial barrier to high-tech industries achieving suc-109 cessful digitalization and automation [20], [21]. Furthermore, 110 although digitalization has received great interest in scholarly 111 research and managerial practices, there is limited under-112 standing of the comprehensive business framework that can 113 be applied to achieve hyperautomation in high-tech manu-114 facturing firms. Indeed, even in industries where CPSs and 115 other digital tools have been implemented, to the best of 116 the authors' knowledge, automation objectives have only 117 been partially realized, and there is no prior study reporting 118 a business framework for hyperautomation in a high-tech 119 firm. 120 The term ''de-massified production'' was first used in 121 1980, and in 1993, the concept of mass customization was 122 introduced [22]. In 1995, a description of the structure of a 123 decentralized enterprise appeared [23]. Wang et al. proposed 124 a framework to bridge the gap between mass ñustomiza-125 tion and mass personalization using I4.0 technologies [24]. 126 Specifically targeting intermediate product configurations 127 that are neither generic nor standardized, Song et al. proposed 128 an uncertain decision-making model for mass personalization 129 of production within I4.0 [25]. These authors presented the 130 theoretical foundations and practical implementation of an 131 assembly system for high-tech products, using the principles 132 of standardization and redundancy. Mourtzis et al. presented 133 a web-based support platform for mass customization and 134 personalization [26]. The platform is responsive and allows 135 interaction with customers during the product design phase. 136 The proposed solution was integrated with a decentralized 137 manufacturing platform implemented using web technolo-138 gies. In a recent study, Lee et al. demonstrated the feasi-139 bility of an OrderAssistant system that generates product 140 specifications from customers' voices [27]. To maximize 141 customer satisfaction, the authors used the Kano model and 142 various optimization methods. The resulting characteristics 143 were transferred to a top-level decision support system that 144 VOLUME 10, 2022 allowed for the simultaneous design of a new product config-145 uration and a change in the production cycle. 146 Recently, significant attention has been paid to the recon-147 figuration and optimization of assembly production using a 148 late customization approach [28]. Rossit et al. presented a 149 framework for reconfiguring the assembly line sequence in 150 the final stages of production depending on the requirements 151 of the customer [29]. It was suggested that the framework 152 be implemented as an interactive online system for setting ods that can be easily combined or transformed into smart 178 distributed scheduling [30], [31]. The emergence of big data 179 and artificial intelligence has brought new insights into inno-180 vation in decision support systems (DSS) [32], [33]. The main 181 impact and key aspects of these smart systems are the product 182 lifecycle approach and the use of DTs. For example, an effec-183 tive DSS can be built using a set of digital tools to describe 184 and model a product and its production processes. This is 185 also termed the DT of production facilities and involves 186 collecting, storing, and using data at all stages of the decision-187 making cycle. The methodology for calculating metrics of the 188 complexity of the production of customized products has also 189 been recently implemented in a real enterprise that manufac-190 tures laser-processing equipment [34]. It was also shown that  Recently, Grassi et al. [35] proposed a semi-heterarchical 203 manufacturing planning and control architecture. Based on 204 this architecture, the production management model can 205 dynamically distribute assignments according to various dis-206 patch rules based on the queueing theory. The performance 207 of the model was evaluated for various production scenarios 208 using hybrid modeling systems. Modeling was carried out 209 exclusively for the shop floor, but the authors claimed that 210 the methods under consideration could also be applied to 211 dynamic dispatching at other levels of enterprise manage-212 ment. It is also worth noting that the developed planning 213 algorithms did not consider feedback from the market.

214
Going beyond of what is currently known, this paper 215 presents a comprehensive account of a sustainable hyper-216 automation approach describing the application of a novel 217 methodology for achieving hyperautomation in manufactur-218 ing of LEMAs within cyberphysical production systems. The 219 business framework, associated to the hyperautomation of the 220 LEMA production relies on responsive Research and Devel-221 opment, mass customization and use of DTs for Industry 222 4.0-compliant reconfigurable products, production processes, 223 production control and management, and added-value ser-224 vices for intra-and inter-enterprise businesses. Table 1 com-225 pares the properties of frequently reported hyperautomation 226 approaches with the framework provided in this study for 227 sustainable hyperautomation, highlinghting the differences 228 and the novel aspects introduced in this manuscript.

229
At this point, it is important to reinforce the fact that the 230 applicability and particularly the impact of the hyperautoma-231 tion approach discussed in this paper has been validated based 232 on the real experience with hyperautomation implementation 233 at Diakont premises, a prominent multinational corporation 234 that develops and manufactures a wide variety of high-tech 235 goods in different regions of the world [36].

236
Following this introductory section that includes a brief 237 literature review addressing relevant reported related works, 238 Sections 2, 3, and 4 introduce and develop the novel hyper-239 automation components. In these sections, after providing the 240 major characteristics of the business framework, the techno-241 logical background for achieving sustainable hyperautoma-242 tion in the manufacturing of LEMAs is presented. A set of 243 digital tools and new concepts, such as intra-enterprise 3-level 244 factory and definitive design-based unified solutions that 245 allow mass customization and provide over 1000 LEMA vari-246 ations, are also described. These novel techniques promote 247 synergistic interactions across many business functions while 248 providing considerable economic and technological benefits. 249 The complete framework, based on the application of the 250 DIN Specification 91345 RAMI4.0, as it has been codified 251 and effectively implemented in the Diakont industrial system, 252 is described, together with an explanation of the research 253 methodology, in Section 5. Section 6 discusses the impact 254 and mainly favorable effect of the hyperautomation method 255 on production costs and overall return on investment. Finally, 256  [51]. The following discussion is centered on a specific 285 example of hyperautomation in the manufacturing of LEMA 286 with Roller Screw Gear (RSG); however, the core concepts of 287 the hyperautomation framework presented in this manuscript 288 have also been employed by authors in the manufacturing of 289 components of CPS, such as feedback sensors, servo drives, 290 and electric motors. The market for LEMA has grown in the recent past, and 293 is expected to grow further. However, increased competi-294 tion and economic crises have reduced the market prices of 295 LEMAs to unexpectedly low levels. The prevailing market 296 prices are much lower than the forecasts for different periods. 297 Taking the example of LEMA with roller screw transmission 298 for spot contact welding, Figure 1 illustrates the actual market 299 price, forecast market price, and target production cost [52], 300 [53], [54]. In 2012, the market price was EUR 4500, and 301 the market forecast indicated that the price would decrease 302 for approximately 15% and stabilize in the coming years. 303 However, the market price followed a markedly downward 304 trend compared to the expected price, forcing industries to 305 review their target costs significantly in 2017. In turn, the 306 reduction in the target production cost makes it necessary to 307 revise the product design and production process. In partic-308 ular, the market price for 2012 allowed the manufacturing 309 of customized products in small and medium batches and 310 offered additional technical advantages to the product, such 311 as longer service life and an integrated system of lubrica-312 tion replacement. In 2015, the business landscape changed 313 remarkably with the arrival of Asian manufacturers, forcing 314 European and American car manufacturers to reduce their 315 costs. This development led to a further reduction in the 316 market price of LEMA as part of the spot welding equipment 317 [55]. Since price is the primary factor affecting consumer 318 decisions, the presence of additional characteristics such as 319 an increased life cycle and additional features ceases to be a 320 competitive advantage, and a customized product with tech-321 nological superiority over competitors is the only means to 322 sustain the competition. With a current target cost, which has 323 decreased by 2.5 times, the unified modular design of LEMA 324 using reconfigurable assembly lines and integrated automa-325 tion and digitalization of the production cycle is a plausible 326 means to remain competitive. The next section describes the 327  The following methodology is based on Diakont's experi-367 ence in achieving successful hyperautomation for sustainable 368 manufacturing of LEMAs at Diakont premises in Lucignano, 369 AR, Italy.
One of the key aspects of sustainable manufacturing of 373 LEMAs is identifying areas that can offer synergistic advan-374 tages. In several scenarios, manufacturing a complete CPS 375 rather than just LEMAs can provide a synergistic advantage 376 in terms of cost and technical quality. For example, the CPS 377 involved in robotic resistance spot welding consists of several 378 components, including an industrial robot, a welding gun 379 with electrodes, and LEMA that controls the gun during weld-380 ing. Analysis of the system reveals two tasks to be solved: 381 the first is the delivery of a welding toolset to the welding 382 point using an industrial robot, and the second is the primary 383 technological cycle of welding performed by the welding 384 gun and actuator. While the first task is auxiliary and can be 385 performed by any five-or six-axis industrial robot, the second 386 task is critical and impacts the performance and quality of the 387 entire process. The technological setup that implements the 388 primary welding cycle is currently not independent of CPS. 389 The control of the welding cycle was assigned to the control 390 system of the robotic arm, which sent a signal to the welding 391 current controller to perform welding and control the actuator 392 that provided the closure of the gun with a given force. The 393 required timing diagram of the force is provided by either the 394 servo drive of the robot or a separate actuator servo drive. Any 395 type of architecture implies certain restrictions caused by the 396 need for motor feedback: the robot controller may not interact 397 with any external servo drive, and the robot servo drive may 398 not interact with any sensor. Thus, it is advisable to create 399 a ''smart tool'' as a separate CPS that would implement the 400 primary technological cycle of welding, an additional ''7th'' 401 axis of an industrial robot. 402 Such a system consists of an actuator, position and force 403 sensors, a control device implementing the functions of a 404 controller, and a servo drive that will be integrated with 405 the welding controller [56], [57]. The advantages of such a 406 system include independence from the industrial robot that is 407 needed to deliver the tool to the welding point and simplified 408 interaction with an industrial robot. It sends a signal that the 409 necessary position is taken and that the welding can be started 410 and receives a response, which means that the welding cycle 411 is completed. This decentralized control approach is particu-412 larly relevant for upgrading existing welding lines when the 413 robot has already been defined, and before that, a pneumatic 414 solution was used as a gun actuator. The price advantage 415 of the integrated solution over the currently used analogs 416 is the use of a less expensive and designed specifically for 417 this application ''smart device'' (controller and servo drive),   allowing quick customization and delivery of LEMAs with 474 desired specifications without adversely affecting the produc-475 tion cost, time, and technical features. Rather than focusing 476 on the composite design of LEMAs, in the hyperautomation 477 approach used at Diakont, the design and engineering stage 478 of integrated R&D begins by identifying the common compo-479 nents of different types of LEMAs. Furthermore, achievable 480 sales and production volumes are regularly updated based 481 on the new information received from the market analysis or 482 from within the company. Technical analysis of the LEMAs 483 revealed that roller screw transmission is the key element for 484 all types of LEMAs. Therefore, the focus was on develop-485 ing a range of roller screw designs that cover the technical 486 specifications desired for various applications. As a matter of 487 fact, the form in which the unification of the product / pro-488 duction design goes, in connection with the new information 489 received from the marked, provided to and from the company 490 engineering department, in a digitalized and interconnected 491 management information system is enhancing the novelty of 492 the hyperautomation approach.

493
Once the designs are ready, the next stage of R&D is to 494 determine the optimal manufacturing technology to produce 495 roller screw components with the advantages of flexibility, 496 versatility, and performance over conventional manufacturing 497 technology. The entire range of products that can be manufac-498 tured using the common key element is allocated to a single 499 class of devices based on common principles. The modular 500 design concept was implemented using common design and 501 technological solutions (Figure 2).

502
As a matter of fact, the Responsive R&D addressed here 503 above means not only ''responsive to the marked change'', 504 but also means ''interconnected'', because through this R&D 505 not only a product is developed, but also the process tech-506 nology and methods for the production organization are 507 designed, which are supported by the result of the analysis 508 provided by the MCOFP. The concept of Repsonsive R&D 509 is novel by itself, including into R&D not only problems and 510 tasks of the product design, but technology, process, factory 511 building, equipment and processes, as well as support to 512 automation and management decision making processes.

513
Three basic unified modules -the roller screw, the rotor, 514 and the stator make up to 75% of the added value of the prod-515 uct and that do not require changes during the development of 516 a new product or product customization. The unified design 517 includes 20 different variants of fastening and connecting 518 elements, 4 dimension types, 6 types of feedback sensors 519 for actuator control, and 6 external options, providing the 520 possibility of creating up more than a thousand variants of 521 the final product. Upgrading the product with properties not 522 provided for in the basic universal design can be carried out 523 in the process of minimum customization (refinement) of 524 additional parts and can be implemented in a short time with 525 minimal cost. This approach resulted in a low cost of produc-526 tion owing to the mass production of basic unified modules, 527 and it allowed the use of high-performance manufacturing 528 technologies, such as thread whirling, circular grinding using 529 VOLUME 10, 2022   Figure. 3 [61], [62], [63], [64]. The factories of  domestic markets and provides cost advantages for several 553 components. Figure 3 shows the geographically distributed 554 production and supply chain created by the Diakont. One 555 of the key tasks in the concept of ''three-level factories'' 556 is to provide rational inventory management, which would 557 ensure the minimum amount of goods necessary to maintain 558 production under unstable demand conditions.

559
The real implementation of the ''three-level factories'' 560 concept, where the production system is considered as a 561 combination of 3 different logic levels, physically combined 562 or not, inside one integrated (digitalized and networked) 563 management and supply chain system, allows, among others, 564 a real-time and visible presence for the customer, shortening 565 and optimizing schedules / terms and lower costs. This is 566 rather novel and is an essential requirement to be fulfilled by 567 the hyperautomation framework.   also covers the strategic and management decision-making 628 level (through its ACS (Accounting and Control System), 629 where the models, described below as parts of the DTs are 630 performed).  The iEMS had three structural levels: strategic, opera-697 tional, and execution (Table 3). A feedback-driven inter-698 action between the DT of the production process and the 699 TABLE 3. Structural organization and main tasks of the intelligent enterprise management system. produc-tion cost plays a key role in the complete automation 700 of the production process ( Figure 5 and 6). The Model of 701 Evaluation of Solutions on the Organization of Production 702 (MESOP) provides a comparative analysis of preferable orga-703 nizational decisions, covering individual technologies and the 704 overall parameters of the manufacturing system. Depend-705 ing on the optimal manufacturing technology predicted by 706 MESOP, a Model of Calculation, analysis, and Optimiza-707 tion of the Financial-economic Parameters of production 708 (MCOFP) determines the number of personnel, equipment, 709 and machines required for automation to reach a given sales 710 volume. MCOFP also predicts fixed and variable costs and 711 the average value of production costs using input cost data 712 such as salaries and costs of materials and types of machin-713 ery, energy resources, and services obtained from a third 714 party. The calculation is repeated cyclically with a specified 715 increase in sales volume. Finally, based on the results of 716 the model, a declining curve is formed that characterizes the 717 average cost under specified conditions. Autonomous production control is achieved using an iEMS 720 that integrates the planning and forecasting system (PFS) with 721 The end-to-end traceability of products in the manufac-740 turing process is organized such that the PES is the digital 741 shadow of the product instance in the manufacturing lifecycle 742 phase [79]. In manufacturing, updated information can be Following the identification of the primary needs related to 758 the development of a new product and its associated business, 759 it is critical to identify the links between several natural 760 problems, tasks, and parameters. The next step is to quan-761 tify and qualitatively analyze the technicality and underlying 762 financial imperatives of these interconnections. This is fol-763 lowed by the development of a collection of models and tools 764 for implementing all the interconnections and computations 765 in a digital environment, while also devel-oping tools for 766 early data gathering and incorporating these tools into the 767 core processes. Solving the research/study problem was an 768 important part of the process; thus, a large set of case data 769 was analyzed, and then, based on the case data, a group of 770 problem points that were more representative were selected 771 and deeply analyzed, with the goal of creating a solution 772 that avoids all representative problems. Indeed, a distinctive 773 part of the research and innovation methodology is the direct 774 application to the entire design and manufacturing process 775 of a critical component in diverse cyberphysical systems, the 776 Linear Electromechanical Actuator (LEMA).

777
To ensure a larger and higher effect on the existing indus-778 trial ecosystem, a major feature of the innovative approach 779 is that the outcomes of development and implementation, 780 as well as the related business framework, are completely 781 aligned with commonly used industrial reference specifica-782 tions. Following the digital transformation impulse carried 783 out by two major representatives, industrial digitalization 784 and networking initiatives such as the Industrial Internet of 785 Things and Industry 4.0, it was decided to position the hyper-786 automation infrastructure within the Reference Architecture 787 Model for Industry 4.0 (RAMI4.0) (DIN SPEC 91345:2016-788 05) [80], [81], [82], [2]. It is important to recall here that 789 RAMI4.0 aims to formally specify industrial assets and 790 asset combinations positioning them within a 3-dimensional 791 space covering its/their position (1)

806
The hyperautomation approach described in this work has 807 a significant positive impact on production cost and overall 808 returns on investment; however, the benefits of this approach 809 VOLUME 10, 2022  in earlier stages) and larger volumes in the long term. Taken 848 together, these factors provide the possibility of achieving 849 sales volumes corresponding to the production volume of 850 50,000 units of standard products per year. The dependence 851 of production costs on sales volumes is shown in Figure 9 for 852 the various scenarios. These results were obtained using the 853 MCOFP method described in the previous section.
854 Figure 9 (a) shows the production cost when universal 855 Computer Numerical Control (CNC) machines are used, and 856 Figure 9 (b) shows a scenario in which mass customization 857 is performed with the use of specialized CNC machines. 858 Figure 9 (c) represents the production cost when mass cus-859 tomization using specialized CNC machines is achieved with 860 a high level of automation. Figure 9 (d) shows the produc-861 tion cost when the concept of a 3-level factory is imple-862 mented (intra-enterprise), along with the scenario mentioned 863 in Figure 8 (c).

864
Brief analysis of Figure 9 shows that from an annual 865 production volume bigger than 1,000 units per year (iden-866 tified as point A), the use of universal CNC machines 867 becomes less preferable than the other three described 868 alternatives. With an annual production of more than 869 2,500 units the more feasible scenario is production using 870 within the Supply Chain. This paper has provided the knowl-908 edge background, scientific, technical and business and tech-909 nical background, which is basically necessary for achieving 910 sustainable hyperautomation, with a real industrial applica-911 tion scenario in the manufacturing of cyberphysical actuators. 912 The hyperautomation approach outlined in this study allows 913 efficient interconnections between various control, automa-914 tion, and business functions facilitated by the digitaliza-915 tion and networking of different industrial tools. Moreover, 916 by uncovering and automating previously inaccessible data 917 and processes, this approach also shows the unique benefit of 918 creating a set of Digital Twins, provided by these tools and 919 positioned within the real industrial engineering, automation 920 and management infrastructure inside a real industrial organi-921 zation. It has also been shown that the promotion of definitive 922 designs based on unified solutions for typical and special-923 ized applications is an effective strategy for achieving mass 924 customization while retaining a sufficiently high production 925 volume. The paper presented a set of industrial-mature digital 926 tools formally positioned within the company infrastructure, 927 according to the DIN SPEC 91345 RAMI4.0. This digital-928 ization approach also allows online monitoring of market 929 changes and provides optimized feedback that can be used for 930 responsive digital modeling of the entire production process 931 and associated businesses, within the company and out of the 932 company within the connected digital world represented by 933 the Supply Chain.

934
At this point, it is important to reinforce the fact that the 935 applicability and particularly the impact of the hyperautoma-936 tion approach discussed in this paper has been validated 937 based on real experience with hyperautomation implementa-938 tion at Diakont premises. Moreover, shown in the Table 1, 939 the authors were able to compare the properties of fre-940 quently reported hyperautomation approaches with the major 941 characteristics of the framework provided in this study for 942 sustainable hyperautomation, identifying the differences and 943 highlighting the novel aspects introduced in this manuscript. 944 On a holistic level, this paper has illustrated the key aspects 945 of attaining long-term sustainable growth in the manufac-946 turing of high-tech equipment by using mass customiza-947 tion, cutting-edge technology in the production process, and 948 system-wide interconnected digital tools. The use of mathe-949 matical models for market forecasting, which are regularly 950 updated by feedback from the market and customers, pro-951 vides valuable knowledge on potential pro-duction volume 952 and product cost estimates, enabling a fine balance in demand 953 and supply. Responsive R&D, which can rapidly adapt to 954 changes in production requirements and product features, 955 is another central feature of the proposed approach. Mass 956 customization is supported by a modular design approach 957 that can provide up to 1000 variants of the product, whereas 958 individual designs are used to produce definitive segments. 959 Hyperautomation using digital twins of product and pro-960 duction processes, forecasting models, and interconnected 961 enterprise management systems provides all-pervasive syn-962 ergy across the entire business function. Although the present 963 VOLUME 10, 2022 study focused mainly on hyperautoma-tion in the manufac- the maximum (or minimum) of the function under the given 1019 constraints, we use the simplex method, which allows us to 1020 perform calculations rather fast even at a very large dimension 1021 of the problem.

1022
This approach, based on a single database of available 1023 optional solutions, allows, for example:

1024
• To choose the routes for manufacturing of product com-1025 ponents that will provide minimum costs at maximum 1026 throughput among other available routes, considering 1027 the available production resources and infrastructure, the 1028 organization of production processes and the level of 1029 automation;

1030
• To calculate and optimize such parameters as the number 1031 of production staff, the number and modes of operation 1032 of machine tools considering the available production 1033 infrastructure, ways of organization of production pro-1034 cesses and the level of automation;

1035
• To compare several options for the organization and 1036 automation of production processes;

1037
• To check the sustainability of the production system to 1038 the changes in batch sizes and product customization 1039 grade.

1040
To obtain all these solutions, we use a number of optimiza-1041 tion problems from linear programming, united by a common 1042 set of constraints. Also, the model for evaluationg solutions 1043 on the organization of production performs the function of 1044 checking the balance of production resources under condi-1045 tions of multiple-machine service mode of the production 1046 operators. In our opinion, such addition is essential for orga-1047 nization of automated production of the smart factory, since 1048 with a high degree of automation operators are no longer 1049 assigned to the machines and act like an additional restriction 1050 when scheduling the operations. No scheduling system can 1051 provide perfect schedules that would ensure accurate timing 1052 of production tasks for machines and personnel, which can 1053 lead to either machine downtime while waiting for the setup 1054 or excessive staff and overstating production costs when oper-1055 ators maintain machines in multiple-machine mode. Thus, 1056 there is a problem of defining the balance of production 1057 resources.

1058
The developed model of checking the balance of produc-1059 tion resources for modeling possible losses uses an approach 1060 based on Markov's theorem. A Markov process with N dis-1061 crete states is modeled for the considered production area. 1062 One of the states corresponds to the need for interven-1063 tion of the operator into the process, while all the others 1064 correspond to the maintenance of one of the machine units 1065 and a combination of machines in the service queue. A system 1066 of Kolmogorov's equations is compiled: