Linen intelligent decision system and method for internet of things multi-source data fusion

Abstract

This invention relates to an intelligent decision-making system and method for linens based on multi-source data fusion in the fields of industrial IoT, big data analytics, and asset management. The system comprises: a sensing and data acquisition layer that collects multi-source data in real time; an edge processing and communication layer that processes the multi-source data and outputs standardized data packets; a cloud-based data fusion and analysis center that receives and stores data, creating a digital twin for each piece of linen, and performing mining analysis on the digital twin cluster and real-time data stream to calculate key operational indicators; and an intelligent decision-making and optimization layer that dynamically generates optimal linen allocation plans and delivery routes for future cycles, outputs washing program suggestions, periodically assesses the remaining lifespan index of each linen digital twin, and generates cost analysis reports. This invention significantly reduces reliance on human experience, improves overall operational efficiency and economic benefits, achieves refined cost and process control, empowers full lifecycle value management of assets, and revolutionizes asset management transparency.

Description

Technical Field

[0001] This invention relates to the fields of industrial Internet of Things (IoT), big data analytics, and asset management, and more specifically to an intelligent decision-making system and method for linens based on multi-source data fusion via IoT. Background Technology

[0002] In industries such as railway passenger transport, highway passenger transport, single-train services, hospitals, catering, and laundry rental, large-scale and professional management of linens is the core of operations. Currently, the industry generally uses RFID-based tracking systems to achieve the identification of individual linens and basic circulation records. However, existing RFID-based tracking systems have the following significant bottlenecks:

[0003] (1) Single data dimension, forming information silos: The existing tracking system based on RFID technology only records the spatiotemporal location of linens (when and where they are read), which is completely separated from the physical parameters of the washing process (such as water temperature and time), resource consumption (such as water, electricity and detergent), the condition of linen material and upstream business needs (such as room occupancy rate). It lacks the correlation of multi-dimensional data and cannot gain a deep understanding of the overall operation.

[0004] (2) Management is passive and lacks intelligent decision-making: The existing tracking system based on RFID technology is limited to post-event query, statistics and reports. It is a record-keeping system and cannot use data for prediction (such as inventory shortage prediction, equipment failure warning) and optimization (such as delivery route optimization, washing program personalization). Decision-making is highly dependent on human experience, which is inefficient and costly.

[0005] (3) The process control is rough and it is difficult to balance cost and quality: The linen washing process usually adopts fixed or empirical procedures, which cannot be dynamically adjusted according to the actual degree of soiling, material and color of the linen. This often leads to the dilemma of over-washing (such as wasting water, electricity and detergent and accelerating the wear and tear of linens) or under-washing (such as hygiene risks), and it is difficult to scientifically balance operating costs and washing quality. Summary of the Invention

[0006] The purpose of this invention is to provide a smart decision-making system and method for linen products based on multi-source data fusion from the Internet of Things (IoT). This addresses the technical problems of existing RFID-based tracking systems, such as single data dimensions, information silos, passive response, lack of intelligent decision-making, and rudimentary process control, making it difficult to balance cost and quality. This invention, based on RFID, IoT devices, and multi-source fusion of business data, realizes a smart decision-making system and method for the linen washing and rental industry. Through data fusion analysis and intelligent decision-making, it proactively provides quantitative decision support for inventory management, logistics and distribution, washing process optimization, lifespan prediction, and sustainable operation, promoting the leap of linen asset management from informatization to digitalization and intelligence.

[0007] The technical solution adopted by this invention to solve the technical problem is as follows:

[0008] This invention provides an IoT multi-source data fusion-based intelligent decision-making system for linens, comprising:

[0009] The sensing and data acquisition layer is used to collect multi-source data in real time;

[0010] The edge processing and communication layer is used to process the collected multi-source data and output standardized data packets;

[0011] The cloud-based data fusion and analysis hub is used to receive and store all time-stream and event-stream data, to create and maintain a dynamically updated digital twin for each piece of linen, and to mine and analyze the digital twin cluster and real-time data stream to calculate key operational metrics.

[0012] The intelligent decision-making and optimization layer is used to dynamically generate optimal linen allocation plans and delivery routes for future cycles, to output washing program suggestions, to periodically assess the remaining life index of each digital twin of linen, and to generate cost analysis reports.

[0013] The visualization and human-computer interaction layer is used to graphically display the global operation map, inventory heat map, early warning center, decision suggestion dashboard and in-depth analysis report in real time.

[0014] Furthermore, the sensing and data acquisition layer includes an RFID sensing network, an IoT sensing unit for devices, and environmental and business data interfaces. The RFID sensing network is deployed across all nodes of the linen circulation process, continuously generating identity-location-time event stream data of the linens and uploading it to the cloud-based data fusion and analysis center. The IoT sensing unit for devices is integrated into or connected to industrial washing equipment, collecting and uploading core data to the cloud-based data fusion and analysis center in real time. The environmental and business data interfaces acquire predictive demand data and upload it to the cloud-based data fusion and analysis center.

[0015] Furthermore, the core data includes process parameters and resource consumption data. The process parameters include water temperature at each stage of washing, main wash and rinsing time, drum speed, steam pressure, drying temperature, and ironing speed and temperature. The resource consumption data includes the total water flow rate for each washing cycle, the precise quantitative dosage of chemicals, and electricity consumption data.

[0016] Furthermore, the edge processing and communication layer includes an edge computing device and an industrial gateway; the edge computing device is used to clean, preprocess, convert protocols, and perform local lightweight analysis on the data, and output standardized data packets; the industrial gateway is used to upload the standardized data packets to the cloud data fusion and analysis center via a wired or wireless network.

[0017] Furthermore, the cloud-based data fusion and analysis hub includes a unified data lake, a digital twin engine, and a correlation analysis module; the unified data lake is used to receive and store all time-stream data and event-stream data; the digital twin engine is used to create and maintain a dynamically updated digital twin for each piece of linen; and the correlation analysis module is used to mine and analyze the digital twin cluster and real-time data stream to calculate key operational indicators.

[0018] Furthermore, the core of the digital twin is a lifecycle event chain strictly ordered by timestamps, including at least one event log entry that records washing events, as well as location information from RFID events, washing program parameters from device IoT data, and water and detergent usage information for this wash.

[0019] Furthermore, the intelligent decision-making and optimization layer includes a dynamic inventory and logistics optimization model, an adaptive washing strategy recommendation model, a precise lifespan prediction and health management model, and a sustainable operation and cost accounting module. The dynamic inventory and logistics optimization model, based on predictive demand data and the real-time location distribution of digital twins of linens at each network point, combined with constraints, uses operations research algorithms to dynamically generate the optimal linen allocation plan and delivery route map for future cycles. The adaptive washing strategy recommendation model, based on historical data and washing mechanisms, aims to minimize overall costs and linen fiber damage, outputs washing program suggestions, and automatically sends them to washing equipment. The precise lifespan prediction and health management model periodically evaluates the remaining lifespan index of each digital twin of linens and provides early warnings when the remaining lifespan index falls below a preset threshold, while also outputting suggestions for downgrading use or planned disposal, achieving predictive asset management. The sustainable operation and cost accounting module, based on detailed resource consumption data recorded by the digital twins, accurately calculates green indicators and generates cost analysis reports.

[0020] Furthermore, the visualization and human-computer interaction layer includes a web-based and a mobile-based cockpit.

[0021] This invention provides a smart decision-making method for linens based on multi-source data fusion in the Internet of Things, comprising the following steps:

[0022] Step 1: Synchronously collect the identity-location-time event stream data, process parameters and resource consumption data, and predictive demand data of the linens; process the collected data, output standardized data packets, and upload them to the unified data lake.

[0023] Step 2: Create and maintain a digital twin for each physical linen item; based on event logic and timestamps, merge and associate the standardized data packets and continuously add them to the lifecycle event chain of the corresponding digital twin;

[0024] Step 3: Perform aggregate analysis on the digital twin cluster, and calculate and update global and local operational indicators in real time;

[0025] Step 4: Parallel Processing of Intelligent Decision-Making

[0026] (1) Inventory and logistics decision: When the calculation results of step three show that the inventory safety factor of a certain area is lower than the preset threshold and the predictive demand data shows an upward trend, the dynamic inventory and logistics optimization model is automatically triggered to generate the optimal linen allocation plan and delivery route map, and push it to the logistics scheduling terminal.

[0027] (2) Washing process decision: When the RFID sensing network identifies a batch of linens at the entrance of the laundry, it automatically triggers the adaptive washing strategy recommendation model: retrieves the digital twin information of the batch of linens, generates washing program suggestions, and sends them to the washing equipment for execution;

[0028] (3) Asset health decision: Regularly scan the remaining life index of all active linens. For linens with a remaining life index below the preset threshold, automatically generate downgrade or planned disposal suggestions and update their digital twin status.

[0029] Step 5: Decision Feedback and Model Self-Learning;

[0030] The actual results after the intelligent decision-making is executed are used as new training data and fed back to the corresponding adaptive washing strategy recommendation model. The model parameters are iteratively optimized using machine learning algorithms to form a closed-loop optimization system of perception-decision-execution-learning.

[0031] Furthermore, in step two, a digital twin is initialized for each piece of linen using the RFID EPC code as the primary key; based on event logic and timestamps, the standardized data packets are merged, associated, and continuously added to the lifecycle event chain of the corresponding digital twin to ensure that the digital twin is synchronized with the physical entity.

[0032] The beneficial effects of this invention are:

[0033] 1. In response to the problem of single data dimension and information silos in existing technologies, this invention collects multi-source data and seamlessly integrates data such as asset identification, process technology, resource consumption, and business requirements, achieving deep integration of all elements of data and thus breaking down the information silos in existing technologies.

[0034] 2. To address the issues of passive management response and lack of intelligent decision-making in existing technologies, this invention designs an intelligent decision-making and optimization layer to dynamically generate optimal linen allocation plans and delivery routes for future cycles, and outputs washing program suggestions. At the same time, it regularly evaluates the remaining lifespan index of each digital twin of linen and uses it to generate cost analysis reports. It proactively provides optimized decisions covering core aspects such as inventory, logistics, washing, and disposal, significantly reducing reliance on human experience, improving overall operational efficiency and economic benefits, and thus driving operations from "experience-driven" to "data-intelligent driven".

[0035] 3. In response to the problems of extensive process control and difficulty in balancing cost and quality in existing technologies, this invention uses an adaptive washing strategy recommendation model to accurately recommend and control water and detergent. This can directly reduce variable operating costs and reduce chemical emissions while ensuring quality, support green operation, and achieve refined cost and process control.

[0036] 4. This invention, through precise lifespan prediction and health management models, can transform disposal recommendations from "based on intuition" to "based on evidence." Combined with cost accounting data, it can accurately calculate the full life cycle cost and return on investment of a single piece of linen, optimize procurement and asset replacement strategies, and empower asset full life cycle value management.

[0037] 5. This invention achieves a revolutionary improvement in asset management transparency by creating and maintaining a dynamically updated digital twin for each piece of linen, and by mining and analyzing the digital twin cluster and real-time data stream to calculate key operational indicators. Attached Figure Description

[0038] Figure 1 The present invention provides a structural block diagram of an IoT multi-source data fusion intelligent decision-making system for linens.

[0039] Figure 2 This is an example flowchart for the intelligent decision-making and optimization layer.

[0040] In the diagram, the components are: Sensing and Data Acquisition Layer 1, RFID Sensing Network 101, Device Internet of Things (IoT) Sensing Unit 102, Environmental and Business Data Interface 103, Edge Processing and Communication Layer 2, Edge Computing Device 201, Industrial Gateway 202, Cloud Data Fusion and Analysis Hub 3, Unified Data Lake 301, Digital Twin Engine 302, Correlation Analysis Module 303, Intelligent Decision-Making and Optimization Layer 4, Dynamic Inventory and Logistics Optimization Model 401, Adaptive Washing Strategy Recommendation Model 402, Precise Life Prediction and Health Management Model 403, Sustainable Operation and Cost Accounting Module 404, Cloud Data Fusion and Analysis Hub 5, Web Terminal 501, and Mobile Terminal Cockpit 502. Detailed Implementation

[0041] In a first aspect, the present invention provides a smart decision-making system for linens based on the fusion of multi-source data from the Internet of Things.

[0042] See Figure 1 This invention provides an IoT-based intelligent decision-making system for linens, incorporating multi-source data. The system mainly comprises the following modules: a sensing and data acquisition layer 1, an edge processing and communication layer 2, a cloud-based data fusion and analysis center 3, an intelligent decision-making and optimization layer 4, and a cloud-based data fusion and analysis center 5. Their specific functions and implementation principles are as follows:

[0043] The sensing and data acquisition layer 1 mainly includes an RFID sensing network 101, an Internet of Things (IoT) sensing unit 102, and an environmental and business data interface 103.

[0044] The RFID sensing network 101 is deployed across all nodes of the linen circulation process (warehouses, logistics vehicles, customer points, laundry entrances / exits, sorting lines). Specifically, the RFID sensing network 101 can use fixed or handheld RFID readers and writers, and can continuously generate "identity-location-time" event stream data of the linens and upload it to the unified data lake 301 in the cloud data fusion and analysis hub 3.

[0045] The device Internet of Things (IoT) sensing unit 102 is integrated into or connected to industrial washing equipment (washing machine, dryer, ironing machine). Specifically, the device IoT sensing unit 102 can adopt the form of a sensor and controller working together to collect and upload two types of core data in real time to the unified data lake 301 in the cloud data fusion and analysis hub 3. These two types of core data are process parameters and resource consumption data. Among them, process parameters include, but are not limited to, water temperature at each stage of washing, main wash and rinsing time, drum speed, steam pressure, drying temperature, and ironing speed and temperature. Resource consumption data includes, but is not limited to, the total water flow (water consumption) for each washing cycle, the precise quantitative dosage of detergent / fabricating agent and other chemicals, and power consumption data.

[0046] Preferably, the device Internet of Things (IoT) sensing unit 102 can be a combination of a flow meter, a smart feeding pump and a PLC controller to collect process parameters and resource consumption data in real time.

[0047] The environmental and business data interface 103 is mainly used to connect external warehouse temperature and humidity sensors and to a train property management system (PMS), hospital information system (HIS), order management system, etc., through the application programming interface (API) to obtain predictive demand data (such as future room occupancy rate, surgical arrangements) and upload it to the unified data lake 301 in the cloud data fusion and analysis hub 3.

[0048] The edge processing and communication layer 2 mainly includes edge computing devices 201 and industrial gateways 202, which are deployed in data-intensive sites such as laundry factories.

[0049] Edge computing device 201 is mainly responsible for cleaning (filtering RFID misreads), preprocessing (format standardization), protocol conversion (converting Modbus and PLC protocols to MQTT / HTTP) the raw data (the "identity-location-time" event stream data of linens, process parameters and resource consumption data, and predictive demand data), and performing local lightweight analysis, thereby outputting standardized data packets.

[0050] The industrial gateway 202 is mainly responsible for uploading standardized data packets to the unified data lake 301 in the cloud data fusion and analysis hub 3 via wired or wireless network.

[0051] The cloud-based data fusion and analysis hub 3 mainly includes a unified data lake 301, a digital twin engine 302, and a correlation analysis module 303.

[0052] The Unified Data Lake 301 is primarily responsible for receiving and storing all time-stream and event-stream data, providing high-throughput read and write capabilities.

[0053] The Digital Twin Engine 302 creates and maintains a virtual mapping entity—a digital twin—for each physical linen item (identified by a unique RFID tag) in the system. This digital twin is dynamic, its core being a strictly time-stamped chain of lifecycle events. It includes at least one event log entry recording a washing event, along with location information from the RFID event, washing program parameters from the device's IoT data, and information on water and detergent usage. Specifically, each event not only records what happened (e.g., being washed), but also integrates multi-dimensional details through data fusion. For example, a washing completion event will integrate: RFID reader point (laundry station exit), device ID, washing program name, total water usage, detergent usage, water temperature profile, energy consumption, etc. Thus, this digital twin becomes a complete and traceable digital archive of the linen's entire lifecycle.

[0054] Taking a pure cotton bed sheet with EPC "301.ABC123" as an example, its digital twin includes static attributes and a dynamic event chain. Static attributes include: material: pure cotton, size: 200×600cm, initial state: unwashed. The detailed payload of the 25th washing event (event number: 25) in the dynamic event chain demonstrates data fusion: location from RFID sensing network 101: "washing machine #03 exit", water_used: 85.5L, detergent_A: 320ml, temp_peak: 70℃ from device Internet of Things (IoT) sensing unit 102, and customer from the business system: "a train A".

[0055] The correlation analysis module 303 is primarily responsible for mining and analyzing the digital twin cluster and real-time data streams (time stream data and event stream data) to calculate key operational indicators (KPIs). Examples include: dynamic saturation of linen inventory at each outlet, linen turnover rate, average washing cost per piece of linen (such as water consumption, detergent, and electricity consumption), resource efficiency ratio of different washing programs, and overall equipment efficiency (OEE).

[0056] The intelligent decision-making and optimization layer 4, also known as the value output layer, mainly includes the dynamic inventory and logistics optimization model 401, the adaptive washing strategy recommendation model 402, the accurate lifespan prediction and health management model 403, and the sustainable operation and cost accounting module 404.

[0057] The Dynamic Inventory and Logistics Optimization Model 401 is mainly used to dynamically generate the optimal linen allocation plan and delivery route map for future periods (such as daily) based on predictive demand data and the real-time location distribution of linen digital twins at each network point, combined with constraints such as vehicle load and route distance, using operations research algorithms, in order to minimize total logistics costs and ensure service levels.

[0058] The adaptive washing strategy recommendation model 402 is one of the core functional layers of the intelligent decision-making and optimization layer 4. When a batch of linens enters the washing process, the system invokes the adaptive washing strategy recommendation model 402 based on the digital twin information of the linens (category, material, usage scenario, and previous washing intensity) and the preliminary judgment of soiling by optical sensors. Based on historical data and washing mechanisms, the adaptive washing strategy recommendation model 402, while ensuring cleanliness and hygiene standards, aims to minimize overall costs (water, detergent, energy consumption) and linen fiber damage, outputting a customized washing program suggestion for that batch of linens. This washing program suggestion includes, but is not limited to, the main wash temperature, water level, washing time, and precise detergent dosage. Simultaneously, this washing program suggestion can be automatically sent to the washing equipment for execution via the environmental and business data interface 103.

[0059] The Precision Lifetime Prediction and Health Management Model 403 integrates multi-dimensional stress data, including washing intensity (temperature, mechanical force), chemical exposure (cumulative detergent usage), and thermal history (drying and ironing temperatures), to construct a nonlinear lifetime decay model for linens of different materials. The Model 403 periodically assesses the remaining lifetime index of each linen's digital twin and provides early warnings when the remaining lifetime index falls below a preset threshold. It also outputs recommendations for degraded use or planned disposal, thus achieving predictive asset management.

[0060] The Sustainable Operations and Cost Accounting Module 404 uses detailed resource consumption data recorded by digital twins to accurately calculate green indicators such as average water consumption per piece of linen, detergent cost ratio, and carbon emission estimates from multiple dimensions, including single item, single customer, single laundry plant, and the company as a whole. It generates cost analysis reports to provide management with a refined management perspective and decision-making basis.

[0061] The sustainable operations and cost accounting module 404 runs regularly, and the generated cost analysis report shows that for customer "Train B," the average comprehensive cost per wash for all bed sheets in the past quarter was RMB 0.45 (including RMB 0.08 for water, RMB 0.22 for detergent, and RMB 0.15 for electricity), and this cost decreased by 8% compared to the previous quarter through program optimization. The cost analysis report also points out that using linens from a specific supplier results in a 15% higher "cost per wash" compared to other suppliers, providing direct data support for procurement decisions.

[0062] The visualization and human-computer interaction layer 5 mainly includes the web terminal 501 and the mobile terminal cockpit 502, which graphically display the global operation map, inventory heat map, early warning center, decision suggestion dashboard and in-depth analysis report in real time.

[0063] Secondly, the present invention provides an IoT-based multi-source data fusion-based intelligent decision-making method for linens, implemented by the IoT multi-source data fusion intelligent decision-making system provided in the first aspect.

[0064] The present invention provides a smart decision-making method for linens based on multi-source data fusion in the Internet of Things, the specific implementation process of which is as follows:

[0065] Step 1: Synchronous acquisition and edge preprocessing of multi-source heterogeneous data;

[0066] The RFID sensing network 101, the device Internet of Things (IoT) sensing unit 102, and the environmental and business data interface 103 are started in parallel to collect the "identity-location-time" event stream data, process parameters and resource consumption data, and predictive demand data of the linens. The edge processing and communication layer 2 is started and receives the raw data collected in parallel. The data cleaning (filtering RFID misreads), preprocessing (format standardization), protocol conversion (converting Modbus and PLC protocols to MQTT / HTTP) and local lightweight analysis are completed. The standardized data packets are output and uploaded to the unified data lake 301 in the cloud data fusion and analysis hub 3.

[0067] Step 2: Construct and dynamically update the digital twin of the linens;

[0068] The digital twin engine 302 is activated to create and maintain a digital twin for each physical linen item in the system (identified by a unique RFID tag). Using the RFID EPC code as the primary key, a digital twin is initialized for each linen item. Based on event logic and timestamps, standardized data packets are merged, associated, and continuously appended to the lifecycle event chain of the corresponding digital twin, ensuring that the digital twin's state is synchronized with the physical entity.

[0069] Step 3: Correlation analysis and real-time situation calculation based on fused data;

[0070] Activate the correlation analysis module 303 to perform aggregate analysis on the digital twin cluster, and calculate and update global and local operational indicators (KPIs) in real time.

[0071] Step 4: Triggering and executing intelligent decisions; this is a parallel processing flow:

[0072] (1) Inventory and logistics decision: When the real-time situation calculation results show that the inventory safety factor of a certain area is lower than the preset threshold and the predictive demand data shows an upward trend, the dynamic inventory and logistics optimization model 401 is automatically triggered to generate the optimal linen allocation plan and delivery route map, and push it to the logistics scheduling terminal.

[0073] (2) Washing process decision: When the RFID sensing network 101 identifies a batch of linens at the entrance of the laundry, it automatically triggers the adaptive washing strategy recommendation model 402. The adaptive washing strategy recommendation model 402 retrieves the digital twin information of the batch of linens, generates a washing program recommendation containing customized detergent dosage and water level suggestions within milliseconds, and automatically or after confirmation sends it to the washing equipment through the environmental and business data interface 103.

[0074] See Figure 2 As shown, when a batch of linens (mainly white sheets) is delivered to the laundry, the RFID sensing network 101 reads their EPC codes in batches. The system immediately retrieves the digital twins of these linens, discovering that they come from the same gym and have been washed an average of 80 times. Simultaneously, optical sensors assess their soiling level as "moderate." The adaptive washing strategy recommendation model 402 is triggered, recommending an optimized washing program based on historical data for "gym - moderately soiled - white sheets": 65℃ main wash, medium water level, 380ml of detergent B (degreasing type) (lower than the standard program's 450ml). This optimized washing program recommendation is automatically sent to the washing equipment for execution after operator confirmation at the terminal. After washing, the actual water consumption and resource consumption data are recorded and fed back to the adaptive washing strategy recommendation model 402 for future optimization.

[0075] (3) Asset health decision-making: Regularly (e.g., daily) run the Precision Life Prediction and Health Management Model 403 to scan the remaining life index of all active linens. For linens with a remaining life index below a preset threshold, automatically generate downgrade or planned disposal recommendations and update their digital twin status to "needs attention".

[0076] Step 5: Decision Feedback and Model Self-Learning;

[0077] The actual results after the intelligent decision is executed (such as the cleanliness test results of linens after washing according to the recommended washing program, and the deviation between the actual water consumption and the predicted value) are used as new training data and fed back to the corresponding adaptive washing strategy recommendation model 402. The model parameters are iteratively optimized using machine learning algorithms to form a closed-loop optimization system of "perception-decision-execution-learning".

[0078] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A smart decision-making system for linens based on multi-source data fusion from the Internet of Things, characterized in that: include: The sensing and data acquisition layer is used to collect multi-source data in real time; The edge processing and communication layer is used to process the collected multi-source data and output standardized data packets; The cloud-based data fusion and analysis hub is used to receive and store all time-stream and event-stream data, to create and maintain a dynamically updated digital twin for each piece of linen, and to mine and analyze the digital twin cluster and real-time data stream to calculate key operational metrics. The intelligent decision-making and optimization layer is used to dynamically generate optimal linen allocation plans and delivery routes for future cycles, to output washing program suggestions, to periodically assess the remaining life index of each digital twin of linen, and to generate cost analysis reports. The visualization and human-computer interaction layer is used to graphically display the global operation map, inventory heat map, early warning center, decision suggestion dashboard and in-depth analysis report in real time.

2. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 1, characterized in that, The sensing and data acquisition layer includes an RFID sensing network, an IoT sensing unit for devices, and environmental and business data interfaces. The RFID sensing network is deployed at all nodes of the linen circulation process, continuously generating identity-location-time event stream data of the linens and uploading it to the cloud-based data fusion and analysis center. The IoT sensing unit for devices is integrated into or connected to industrial washing equipment, collecting and uploading core data to the cloud-based data fusion and analysis center in real time. The environmental and business data interfaces acquire predictive demand data and upload it to the cloud-based data fusion and analysis center.

3. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 2, characterized in that, The core data includes process parameters and resource consumption data. The process parameters include water temperature at each stage of washing, main wash and rinsing time, drum speed, steam pressure, drying temperature, and ironing speed and temperature. The resource consumption data includes the total water flow rate for each washing cycle, the precise amount of chemicals added, and electricity consumption data.

4. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 1, characterized in that, The edge processing and communication layer includes edge computing devices and industrial gateways; the edge computing devices are used to clean, preprocess, convert protocols, and perform local lightweight analysis on data, and output standardized data packets; the industrial gateways are used to upload the standardized data packets to the cloud data fusion and analysis center via wired or wireless networks.

5. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 1, characterized in that, The cloud-based data fusion and analysis hub includes a unified data lake, a digital twin engine, and a correlation analysis module. The unified data lake is used to receive and store all time-stream data and event-stream data. The digital twin engine is used to create and maintain a dynamically updated digital twin for each piece of linen. The correlation analysis module is used to mine and analyze the digital twin cluster and real-time data stream to calculate key operational indicators.

6. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 5, characterized in that, The core of the digital twin is a lifecycle event chain strictly ordered by timestamps, including at least one event log entry that records washing events, as well as location information from RFID events, washing program parameters from device IoT data, and water and detergent usage information for this wash.

7. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 1, characterized in that, The intelligent decision-making and optimization layer includes a dynamic inventory and logistics optimization model, an adaptive washing strategy recommendation model, an accurate lifespan prediction and health management model, and a sustainable operation and cost accounting module. The dynamic inventory and logistics optimization model, based on predictive demand data and the real-time location distribution of digital twins of linens at each outlet, combined with constraints, uses operations research algorithms to dynamically generate optimal linen allocation plans and delivery routes for future cycles. The adaptive washing strategy recommendation model, based on historical data and washing mechanisms, aims to minimize overall costs and linen fiber damage, outputting washing program suggestions and automatically sending them to washing equipment. The accurate lifespan prediction and health management model periodically assesses the remaining lifespan index of each digital twin of linens and provides early warnings when the remaining lifespan index falls below a preset threshold, while also outputting suggestions for degraded use or planned disposal, achieving predictive asset management. The sustainable operation and cost accounting module, based on detailed resource consumption data recorded by the digital twins, accurately calculates green indicators and generates cost analysis reports.

8. The IoT multi-source data fusion intelligent decision-making system for linens according to claim 1, characterized in that, The visualization and human-computer interaction layer includes a web-based and a mobile-based cockpit.

9. A method for intelligent decision-making on linens based on IoT multi-source data fusion, implemented using the IoT multi-source data fusion intelligent decision-making system according to any one of claims 1-8, characterized in that: Includes the following steps: Step 1: Synchronously collect the identity-location-time event stream data, process parameters and resource consumption data, and predictive demand data of the linens; process the collected data, output standardized data packets, and upload them to the unified data lake. Step 2: Create and maintain a digital twin for each physical linen item; based on event logic and timestamps, merge and associate the standardized data packets and continuously add them to the lifecycle event chain of the corresponding digital twin; Step 3: Perform aggregate analysis on the digital twin cluster, and calculate and update global and local operational indicators in real time; Step 4: Parallel Processing of Intelligent Decision-Making (1) Inventory and logistics decision: When the calculation results of step three show that the inventory safety factor of a certain area is lower than the preset threshold and the predictive demand data shows an upward trend, the dynamic inventory and logistics optimization model is automatically triggered to generate the optimal linen allocation plan and delivery route map, and push it to the logistics scheduling terminal. (2) Washing process decision: When the RFID sensing network identifies a batch of linens at the entrance of the laundry, it automatically triggers the adaptive washing strategy recommendation model: retrieves the digital twin information of the batch of linens, generates washing program suggestions, and sends them to the washing equipment for execution; (3) Asset health decision: Regularly scan the remaining life index of all active linens. For linens with a remaining life index below the preset threshold, automatically generate downgrade or planned disposal suggestions and update their digital twin status. Step 5: Decision Feedback and Model Self-Learning; The actual results after the intelligent decision-making is executed are used as new training data and fed back to the corresponding adaptive washing strategy recommendation model. The model parameters are iteratively optimized using machine learning algorithms to form a closed-loop optimization system of perception-decision-execution-learning.

10. The IoT multi-source data fusion intelligent decision-making method for linens according to claim 9, characterized in that, In step two, a digital twin is initialized for each piece of linen using the RFID EPC code as the primary key. Based on event logic and timestamps, the standardized data packets are merged, associated, and continuously added to the lifecycle event chain of the corresponding digital twin to ensure that the digital twin is synchronized with the physical entity.

Images