Facilitating AI-Based Solutions Integration in Production Planning: Development of an Interoperable Data Model

Facilitating AI-Based Solutions Integration in Production Planning: Development of an Interoperable Data Model Beatriz Andres(B), Juan Pablo Fiesco, Raul Poler, and Miguel A. Mateo-Casali Research Centre on Production Management and Engineering (CIGIP), Universitat Politècnica de València (UPV), Camino de Vera s/n, 46022 Valencia, Spain {bandres,jfiesco,rpoler,mmateo}@cigip.upv.es Abstract. Industries strive to improve efficiency, minimise environmental impacts and ensure operational continuity. Enterprises digitisation favours the achievement of these challenges. AI-based tools play a significant role in improving enterprise digitisation by offering capabilities that enhance efficiency and decision making across various functions of an organization’s operations, processes, and systems. This article delves into planning processes by proposing the use of an interoperable data model to integrate AI-based solutions developed to support the optimisation of production plans. The proposed data model facilitates interoperability between enterprise legacy systems and AI-based manufacturing solutions to support seamless communication between them, thus, assist replenishment and manufacturing decisions. Keywords: Artificial Intelligence · Sustainable Manufacturing · Interoperability · Industrial food inspection equipment · Industry 4.0 1 Introduction In an increasingly interconnected and competitive world, companies face the constant challenge of improving efficiency by focusing on environmental impacts and ensuring the continuity of their operations. In this context, the European AI-Driven Industrial Equipment Product Life Cycle Boosting Agility, Sustainability and Resilience (AIDEAS, 2022) project focuses on developing a set of artificial intelligence (AI) based solutions to improve various industrial machinery production aspects. AIDEAS proposes a set of AI tools that covers the entire lifecycle of these machines, including the: (i) design phase; (ii) production phase; (iii) use phase; (iv) repair/reuse/recycling phase. The project aims to strengthen the competitiveness, sustainability and resilience of European machinery manufacturers and users. Four industrial pilots participate in the AIDEAS project, in which AI-based solutions are demonstrated. The four pilots are manufacturers of industrial equipment and belong to different sectors, and provide wide-ranging case studies in which different problems of the f our lifecycle phases are addressed. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025 R. Carrasco-Gallego et al. (Eds.): CIO 2024, LNDECT 239, pp. 220–225, 2025. https://doi.org/10.1007/978-3-031-82334-3_38 Facilitating AI-Based Solutions Integration in Production Planning 221 Industry 4.0 (I4.0) and its technologies have triggered a major transformation of processes in the manufacturing industry (Oliveira-Dias et al., 2023). The digitisation of information is an unfinished task for many smalland medium-sized enterprises (SMEs), especially in industrial equipment companies that handle large volumes of data related to bills of materials, suppliers, inventory and manufacturing process status (Doyle and Cosgrove, 2019; Shah et al., 2024). However, limited resources, the high costs associated with implementing I4.0 technology projects, deficiencies in data and knowledge management and lack of trained personnel can significantly affect their ability to innovate, compete in the market and adapt to today’s rapidly changing business environment. To address the data management problem, this paper identifies in the literature a data model for collaborative manufacturing environments (CMData) proposed by Andres, Poler and Sanchis (2021) for the exchange of data among AI-based solutions developed to support the optimisation of production plans. The proposed interoperable data model plays a critical role in bridging the gap between legacy systems (LS) and modern AI-based manufacturing solutions, leading to enhanced interoperability, optimized production plans, cost efficiency, scalability, improved decision-making, and the facilitation of AI adoption. To validate the paper results, a pilot case study is presented. Accordingly, the paper is organised as follows: Sect. 2 introduces the AI-based solutions to support the manufacturing plans within the European AIDEAS project scope; Sect. 3 conceptualises the interoperable data model to support communication of production planning AI-based solutions. The proposed model interoperates not only at the solutions level, procurement, production and scheduling planning, but also at the enterprise LS level; Sect. 4 provides the conclusions and future research lines. 2 AI Solutions to Support the Manufacturing Phase Of the four product lifecycle phases, namely design, manufacturing, use and disposal, this paper focuses on the manufacturing phase, which encompasses the procurement, fabrication and delivery processes. Optimising these processes requires analysing and processing large amounts of dynamic data, such as suppliers, manufacturers and customers’ information, material supply, customer demand or resource capacity, and less dynamic data, such as bills of materials or product catalogues. These data are usually stored in enterprise LS. To make these data interoperable for any interorganisational or intraorganisational management system, it is necessary to define a normalised data model. To improve the resilience of supply chain processes, it is vital that enterprise LSs are open and secure to foster collaboration between partners and to be more accurate in decision making (Andres, Poler and Sanchis, 2021). Another crucial aspect is to ensure the quality and connectivity of the data to be used for efficient supply chain programming and control plans (Burggräf et al., 2018). This paper focuses on the definition of the interoperable data model for the Procurement Optimiser (AI-PO) and the Fabrication Optimiser (AI-FO) solutions of the AIDEAS project. AI-PO optimises the purchase of materials by minimising inventory costs and improving company profits. This solution analyses information, such as supplier quotations, bill of materials and production requirements for purchasing decisions. 222 B. Andres et al. AI-FO supports informed decision making on allocating resources, reducing downtime and increasing productivity by allowing manufacturers to quickly adapt to changes in demand and operating conditions. Both solutions leverage AI algorithms to increase agility and productivity in the manufacturing phase. In the AI-FO solution, there are two AI-based algorithms: one that solves the master production plan (MPP) and the scheduling plan (SP): (i) AI-FOMPP considers account factors, such as committed or forecast demand, operator availability, inventory levels and other constraints to optimise resource allocation and to meet customer demand efficiently; (ii) AI-FOSP sets out the detailed schedule for executing the production activities specified in the AI-FOMPP.It determines when each production task should begin and end, what resources (machines, equipment, operators) should be allocated to each task and the order in which they should be performed. Production planning is an iterative and critical task in the manufacturing sector that requires many data. Production levels are planned for each period in production management so that all orders can be delivered on time (Charles et al., 2022). The complexity of production planning depends on the policies of the production system and the amount of data to be considered (Karimi et al., 2003). Therefore, to integrate AI-PO and AI-FO, it is necessary to analyse the policies of the company’s production system and to see how this communication between solutions can take place. 3 Data Model for Production Planning AI-Based Solutions To communicate data from enterprise LSs to the AI-based manufacturing solutions developed in the AIDEAS project, an interoperable data model is needed. Based on the paper of Andres, Poler and Sanchis (2021), this section describes the use of the data model for the CMData proposed by the authors. This paper proposes the integration of the CMData to facilitating AI-based Solutions Integration in Production Planning, including AI-PO, AI-FOMPP and AI-FOSP solutions. The CMData tables contain a homogenised nomenclature of the input data and objectives required to feed AI-based solutions. Each CMData table is composed of different attributes that characterise it. Hence the CMData table Part refers to a product’s generic terminology, including a firm’s raw materials, components or final products. The CMData Table Part has several attributes, such as code, description, inventory cost, initial inventory, purchase cost, selling price, current inventory, dimensions, weight, location in inventory, among others. AI-based manufacturing solutions have the particularity that they communicate with one another and work in an integrated model environment. The relations between solutions and how they communicate through a centralised database, the AIDEAS datastore (AI-DS), is depicted in Fig. 1. Within this framework, enterprise LS establish a link with the AI-DS database through a querying process. This querying mechanism serves as the conduit through which enterprise LS access and interact with the data stored in the AI-DS (see flow 1-Enterprise Mapping in Fig. 1). Once the data are in the AI-DS and according to the CMData structure, AI-FOMPP accesses the AI-DS to obtain t he required input data (2AI_FOMPP_InputData). Subsequently, the MPP is calculated and the output data are obtained (3-AI-FOMPP_OutputData) to present the number of products to be produced during each period by considering each product’s start and end manufacturing date. The Facilitating AI-Based Solutions Integration in Production Planning 223 AI-FOMPP output is stored in the AI-DS according to the data structure defined by the CMData. Furthermore, the AI-FOMPP output serves as input to the AI-FOSP and AI-PO solutions. Additionally, the output from AI-FOMPP is mapped back through a query by generating a spreadsheet with fields translated into the nomenclature preferred by the company for data interpretation. In the pilot phase, the production planner reviews the spreadsheet data. If the company wishes to integrate these data into its LS, additional query generation is required. The AI-PO solution explodes the materials needed to manufacture according to the MPP computed by AI_FOMPP (3-AI_FOMPP_OutputData) and generates the purchasing plan (4-AI_PO_OutputData), which is stored in the AI-DS. If there is any temporal infeasibility, it is communicated to the AI-FOMPP solution. In this way, the AI-FOMPP solution r ecalculates the MPP by considering the new material arrival dates specified in 4-AI_PO_OutputData, which will cause the production start dates to be delayed. Moreover, the AI-PO output is mapped back through the query by generating a spreadsheet with fields translated into t he nomenclature preferred by the company Fig. 1. Data Flow Architecture 224 B. Andres et al. for data interpretation. In the pilot phase, the purchase planner reviews the data on the spreadsheet. If the company desires these data to appear in its LS, further write queries must be generated. The AI-FOSP SP solution is fed with 3-AI_FOMPP_OutputData and 4-AI_PO_Output_Data, which enables the production schedule calculation that considers the start and finish dates of the MPP and the availability of the materials calculated by the procurement plan. The AI_FOSP solution (5-AI_FOSP_Output) output is stored in the AI-DS with information about the operators that are going to carry out different operations in certain work centres to fulfil the defined MPP. It is possible that 4-AI_SP_Output may s chedule a series of materials or operators that cannot be fulfilled in the short term. This will prompt a feedback loop into the process to recalculate the MPP by AI-FOMPP. Consequently, if the necessary materials are unavailable, the procurement plan will be recalculated to generate new 3-AI_FOMPP_Output_Data and 4AI_PO_Output_Data solutions. Additionally, the AI-FOSP SP output is mapped again through the query and generates a spreadsheet with fields translated into the nomenclature preferred by the company for data interpretation. In the pilot phase, the scheduler reviews the data on the spreadsheet. If the company wishes these data to appear in its LS, other write queries must be generated. 4 Conclusion and Future Research The AIDEAS project represents a significant improvement towards enhancing sustainability and efficiency in the manufacturing of industrial food inspection equipment using interoperable AI solutions. Considering the challenges of digitisation and the integration of I4.0 technologies into SMEs, this paper conceptualises communication among three solutions developed in AIDEAS. This communication occurs t hrough the AI-DS tool built with an interoperable data model following the CMData standard, and designed to feed integrated planning models, including procurement, production and scheduling, and to facilitate data exchange not only between AI-based solutions, but also with the LS company. Standardising data and integrating enterprise LS with advanced AI technologies enable agile and accurate decision makingby improving the path towards a more sustainable and efficient future in the industrial equipment industry. As future research lines, experiments with real company data are proposed to be carried out by evaluating the quality and viability of solutions in company operations. Acknowledgements. The research received funding from the Horizon Europe Framework Programme (HORIZON) with Grant Agreement No. 101057294 “AI-Driven Industrial Equipment Product Life Cycle Boosting Agility, Sustainability, and Resilience (AIDEAS)”. The Regional Department of Innovation, Universities, Science, and Digital Society of the Generalitat Valenciana “Programa Investigo” (ref. INVEST/2022/330), which the European Union supported - NextGenerationEU with Plan de Recuperación, Transformación y Resiliencia. The Generalitat Valenciana with program “Subvenciones para la realización de estancias de personal investigador doctor en empresas de la Comunitat Valenciana” (ref. CIAEST/2022/39). Facilitating AI-Based Solutions Integration in Production Planning 225 References AIDEAS. AI driven industrial equipment product life cycle boosting agility, sustainability and resilience. European Union’s Horizon Europe re-search and innovation programme under grant agreement no. 1010572942022. https://doi.org/10.3030/101057294 Andres, B., Poler, R., Sanchis, R.: A data model for collaborative manufacturing environments. Comput. Ind. 126 (2021).https://doi.org/10.1016/j.compind.2021.103398 Burggräf, P., Dannapfel, M., Förstmann, R., Adlon, T., Fölling, C.: Data quality-based process enabling: application to logistics supply processes in low-volume ramp-up context. In: 2018 International Conference on Information Management and Processing (ICIMP). pp. 36–41 (2018). https://doi.org/10.1109/ICIMP1.2018.8325838 Charles, M., Dauzère-Pérès, S., Kedad-Sidhoum, S., Mazhoud, I.: Motivations and analysis of the capacitated lot-sizing problem with setup times and minimum and maximum ending inventories. Eur. J. Oper. Res. 302, 203–220 (2022). https://doi.org/10.1016/j.ejor.2021.12.017 Doyle, F., Cosgrove, J.: Steps towards digitization of manufacturing in an SME environment. Procedia Manuf. 38, 540–547 (2019). https://doi.org/10.1016/j.promfg.2020.01.068 Karimi, B., Ghomi, S., Wilson, J.M.: The capacitated lot sizing problem: a review of models and algorithms. Omega-Int. J. Manage. Sci. 31, 365–378 (2003). https://doi.org/10.1016/S03050483(03)00059-8 Oliveira-Dias, D., de, Maqueira-Marin, J.M., Moyano-Fuentes, J., Carvalho, H.: Implications of using Industry 4.0 base technologies for lean and agile supply chains and performance. Int. J. Prod. Econ. 262 (2023). https://doi.org/10.1016/j.ijpe.2023.108916 Shah, S., Hussain Madni, S.H., Hashim, S.Z.B.M., Ali, J., Faheem, M.: Factors influencing the adoption of industrial internet of things for the manufacturing and production small and medium enterprises in developing countries. IET Collabor. Intell. Manuf. 6 (2024). https://doi.org/10. 1049/cim2.12093