Integrating Predictive Maintenance Analytics into a Cloud-based CMMS for Smart Work Order Management and Resource Allocation

在维护实践中产生了大量与建筑物、设施和基础设施组件相关的数据。这些数据大多被动地存储在数据库中作为维护记录。数据分析技术提供了一个机会，可以将大量数据转化为有用的信息，并生成知识，以提高未来维护实践的性能、可靠性和效率。在这方面，预测分析可以用于故障预测和诊断（确保物理系统的可靠性）之外，并用于维护活动的智能（智能和预测性）管理，以提高资源效率。拟议项目的主要目标是建立研发，将人工智能技术整合到基于云的维护管理流程中，重点关注维护工作订单管理和资源分配。使用一组选择的聚类方法，一组区分属性（功能）和最佳（最小）数量的集群被确定分类工作订单管理模式。此外，设计了一个基于规则的智能资源分配系统，以智能的方式将员工分配到工作中，从而降低资源强度，提高成本效率。使用决策树分类器，从历史数据中识别出的模式被用来生成一组决策规则，为未来的维护项目，建立一个预测的方法对资源分配。最后，将开发一个基于KPI的评估系统，为上述基于AI的算法提供反馈，以进行验证和改进。

There is a massive amount of data generated in the maintenance practices as related to components of buildings, facilities, and infrastructure. This data is mostly passively stored in databases as maintenance records. Data analytics technologies present an opportunity to turn this large volume of data into useful information and generate knowledge to enhance the performance, reliability, and efficiency of future maintenance practices. In this regard, predictive analytics can be utilized beyond failure prediction and diagnosis (ensuring reliability of physical systems) and towards an intelligent (smart and predictive) management of maintenance activities to enhance resource efficiency. The main objective of the proposed project is to establish R&D into the integration of AI technologies into a cloud-based maintenance administration process with a focus on maintenance work-order management and resource allocation. Using a select set of clustering methods, the set of differentiating attributes (features) and an optimal (minimum) number of clusters are identified to classify work-order management patterns. In addition, a smart resource allocation rule-based systems is designed to assign the staff to the jobs in an intelligent manner such that to reduce resource intensity and improve the cost efficiency. Using decision tree classifiers, the identified patterns from historical data are utilized to generate a set of decision rules for future maintenance projects, establishing a predictive approach toward resource allocation. Finally, a KPIs-based assessment system will be developed as a means of providing feedback to the above AI-based algorithms for validation and improvement purposes.