Multi-target collaborative predictive maintenance system and method for industrial dehumidifier

Abstract

The invention relates to the technical field of industrial air conditioners, in particular to an industrial dehumidifier multi-target collaborative predictive maintenance system and method.The industrial dehumidifier multi-target collaborative predictive maintenance system comprises a data access module, a data feature module, a digital management module and a multi-target prediction engine; the data access module is configured to obtain multi-source real-time operation data of the industrial dehumidifier from an external sensor network and a data collector through a preset application programming interface, and the multi-source real-time operation data comprises vibration, current, voltage, temperature, pressure, differential pressure and temperature and humidity data; the data feature module is connected to the data access module and is configured to perform data noise reduction processing and feature extraction on the acquired multi-source real-time operation data to generate a standardized feature vector. According to the method, early warning can be carried out in advance at the initial stage of part performance degradation or before potential faults occur, so that a user can easily plan and execute maintenance activities, and the production interruption risk and high loss caused by unplanned shutdown are greatly avoided.

Description

Technical Field

[0001] This invention relates to the field of industrial air conditioning technology, and in particular to a predictive maintenance system and method for multi-objective collaborative operation of industrial dehumidifiers. Background Technology

[0002] Industrial dehumidifiers are currently key equipment for ensuring stable humidity in precision manufacturing, warehousing, and logistics environments, and their operational reliability directly impacts production quality and continuity. Traditional maintenance methods mainly rely on periodic planned maintenance and reactive fault repair, which have significant shortcomings: periodic maintenance is prone to over-maintenance or under-maintenance, leading to resource waste or sudden downtime; reactive repair results in unplanned production stoppages and high losses. With the development of the Internet of Things and sensor technology, some equipment has been connected to sensors for status monitoring, but existing early warning systems are mostly limited to single-parameter threshold alarms or isolated life predictions for single components, lacking in-depth analysis of the coupling failure mechanisms between key components such as compressors, fans, heat exchangers, and filters, and failing to coordinate and optimize multiple objectives such as maintenance costs, equipment availability, and production plans. In addition, most systems have a disconnect between prediction and decision-making, failing to form a closed loop from data perception to strategy execution to model optimization, resulting in a decline in prediction accuracy over time and insufficient real-time and adaptability of maintenance strategies.

[0003] Therefore, there is an urgent need for an intelligent predictive maintenance system that can integrate multi-source data, understand the coupling relationship between components, and generate the optimal maintenance strategy under multiple constraints, so as to achieve a fundamental transformation in the operation and maintenance of industrial dehumidifiers from passive response to proactive and precise decision-making. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides a multi-objective, collaborative predictive maintenance system and method for industrial dehumidifiers that can provide early warnings at the initial stage of component performance degradation or before potential failures occur, enabling users to plan and execute maintenance activities with ease. This greatly avoids the production interruption risks and high losses caused by unplanned downtime.

[0005] The technical solution adopted in this invention is: a multi-objective collaborative predictive maintenance system for industrial dehumidifiers, comprising a data access module, a data feature module, a digital management module, and a multi-objective prediction engine. The data access module is configured to acquire multi-source real-time operating data of the industrial dehumidifier from an external sensor network and data acquisition device through a preset application programming interface. The multi-source real-time operating data includes vibration, current, voltage, temperature, pressure, differential pressure, and temperature and humidity data. The data feature module is connected to the data access module and is configured to perform data noise reduction and feature extraction on the acquired multi-source real-time running data to generate a standardized feature vector. The digital management module is used to create and maintain a virtual model that corresponds one-to-one with the physical industrial dehumidifier. The input of the virtual model is the standardized feature vector, and the output is a real-time status mapping of the equipment. The multi-objective prediction engine is connected to the digital management module, and the multi-objective prediction engine includes a component life prediction unit, a collaborative evaluation unit, a maintenance generation module, and a strategy feedback module. The component life prediction unit is configured to invoke a pre-trained machine learning model to independently predict the remaining lifespan of the compressor, fan, heat exchanger, and filter based on the standardized feature vector. The collaborative evaluation unit is configured to analyze the coupling impact of predicted component life decay on overall performance, energy consumption and other related components based on a predefined system coupling relationship knowledge base, and generate a system-level health index and risk warning signal. The maintenance generation module is connected to the multi-objective prediction engine and is configured to solve the maintenance strategy based on the system-level health index, risk warning signal, preset maintenance cost library and production plan information obtained from external systems, through a multi-objective optimization algorithm. The strategy feedback module is configured to send the generated maintenance strategy to an external computerized maintenance management system or enterprise resource planning system through a standard interface, and to receive feedback data after the strategy is executed for model iterative optimization.

[0006] A multi-objective collaborative predictive maintenance method for industrial dehumidifiers, based on a multi-objective collaborative predictive maintenance system, wherein the method is executed by a software system and includes: Obtain multi-source real-time operating data of industrial dehumidifiers from external sources through data access service interfaces; The operational data is cleaned, transformed, and its features are extracted to generate standardized feature vectors; The standardized feature vector is input into the digital twin virtual model of the corresponding device to update its operating status; A pre-trained machine learning model is invoked to predict the remaining service life of key components based on the standardized feature vectors; Based on a system-coupled knowledge base, the synergistic impact of component lifespan degradation on the overall health of the machine is assessed, and a health index and risk level are generated. Based on health index, risk level, maintenance cost and production plan, the optimal maintenance strategy is generated through a multi-objective optimization algorithm. The optimal maintenance strategy is distributed to the external management system for execution through a standard interface, and feedback data is collected to optimize the model.

[0007] The beneficial effects of this invention are: This invention collects real-time, multi-dimensional operational data of industrial dehumidifiers through a data access module and uses a pre-trained machine learning model to accurately predict the remaining lifespan of core components, completely changing the traditional, reactive approach of relying on periodic inspections or repairs after a failure. This system can provide early warnings at the initial stage of component performance degradation or before potential failures occur, allowing users to plan and execute maintenance activities with greater ease, significantly reducing the risk of production interruptions and high losses caused by unplanned downtime.

[0008] The maintenance generation module of this invention is based on a multi-objective optimization algorithm. When formulating maintenance strategies, it simultaneously and comprehensively considers multiple factors such as system health status, maintenance costs, spare parts inventory, and external production plans. Its output is no longer a single maintenance prompt, but a scientific decision-making solution that finds the optimal balance among multiple conflicting objectives such as "lowest maintenance cost," "highest system availability," and "minimum production impact." This helps enterprises shift from extensive maintenance to refined and lean maintenance, significantly reducing the total lifecycle maintenance costs.

[0009] This invention, through a strategy feedback module, enables the system to collect data on the actual execution effect of maintenance strategies and utilize this feedback data to continuously iterate and optimize the prediction model and knowledge base. This allows the system to become increasingly accurate and "intelligent" over time, effectively overcoming the prediction drift problem caused by equipment aging and changes in operating conditions in traditional models, and ensuring the accuracy and reliability of long-term predictions.

[0010] This invention seamlessly integrates with external sensors, data acquisition devices, and upper-level management systems through standardized data interfaces and application programming interfaces, enabling automatic data acquisition and automatic distribution of maintenance strategies. This reduces manual intervention, lowers the human error rate, and establishes a fully digital management process from condition monitoring to maintenance execution, significantly improving the automation level and overall efficiency of equipment management.

[0011] This invention integrates intelligent prediction, collaborative evaluation, optimized decision-making, and closed-loop feedback, providing a forward-looking, systematic, and cost-effective new solution for the operation and maintenance management of industrial dehumidifiers. It has extremely high practical value and promising prospects for promotion.

[0012] This invention presents a multi-objective collaborative predictive maintenance method for industrial dehumidifiers, executed by a software system. This method abandons the traditional passive maintenance model that relies on manual experience, periodic inspections, or reactive measures after a failure occurs. Instead, it automates the entire process from data collection, processing, and analysis to decision generation through software. The method utilizes machine learning models to accurately predict component lifespan and generates maintenance strategies based on multi-objective optimization algorithms. This transforms the decision-making process from "experience-driven" to "data-driven" and "model-driven," significantly enhancing the predictability and scientific rigor of maintenance activities and avoiding under- or over-maintenance. This method does not view the health status of individual components in isolation but introduces a system-coupled knowledge base to quantify and analyze the cascading effects of individual component performance degradation on overall machine performance, energy consumption, and related components. Collaborative assessment provides a comprehensive understanding of the equipment system's health status, accurately identifying the critical weaknesses that have the greatest impact on system stability and energy efficiency. This guides users to prioritize addressing the most critical and economical issues, achieving a leap from "component health" management to "system health" management. This invention continuously optimizes and calibrates the prediction model and knowledge base by collecting feedback data after strategy execution, enabling the system to learn and become increasingly accurate over time. It effectively overcomes the problem of decreased model prediction accuracy caused by equipment aging, operating condition drift, or environmental changes, ensuring that the system provides reliable, adaptive predictive maintenance services throughout its entire lifecycle, demonstrating strong long-term viability and practicality. This invention provides an advanced maintenance method for industrial dehumidifiers that integrates intelligent sensing, collaborative evaluation, optimized decision-making, and closed-loop learning, providing solid technical support for achieving high-reliability, low-cost, and intelligent operation and maintenance of equipment, and possessing significant industrial application value. Attached Figure Description

[0013] Figure 1 This is a connection diagram of the multi-objective collaborative predictive maintenance system for industrial dehumidifiers according to the present invention; Figure 2 This is a flowchart illustrating the multi-objective collaborative predictive maintenance method for industrial dehumidifiers according to the present invention.

[0014] Explanation of reference numerals in the attached drawings: Structural frame 1, First module area 11, Second module area 12, Compressor refrigeration and dehumidification module 2, Compressor 21, Condenser 22, Throttling device 23, Evaporator 24, Solution dehumidification module 3, Dehumidification unit 31, Regeneration unit 32, Regeneration air inlet 321, Regeneration air outlet 322, Solution circulation pipeline 33, Solution pump 34, Air handling channel 4, Air inlet 41, Air outlet 42, Mode air valve 43, Intelligent control system 5, Humidity sensor 51, Temperature sensor 52, PLC controller 53, Human-machine interface 54, Heat recovery device 6, Filter module 7, Fan module 8, Humidification module 9. Detailed Implementation

[0015] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0016] It should be noted that when a component is said to be "fixed to" another component, it can be directly attached to the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Figures 1-2 As shown, in one embodiment of the present invention, a predictive maintenance system for industrial dehumidifiers with multi-objective collaboration is provided, including a data access module 1, a data feature module 2, a digital management module 3, and a multi-objective prediction engine 4.

[0018] The data access module 1 is configured to acquire multi-source real-time operating data of the industrial dehumidifier from an external sensor network and data acquisition device through a preset application programming interface. The multi-source real-time operating data includes vibration, current, voltage, temperature, pressure, pressure difference, and temperature and humidity data. The data feature module 2 is connected to the data access module 1 and is configured to perform data noise reduction and feature extraction on the acquired multi-source real-time running data to generate a standardized feature vector. The digital management module 3 is used to create and maintain a virtual model that corresponds one-to-one with the physical industrial dehumidifier. The input of the virtual model is the standardized feature vector, and the output is the real-time status mapping of the equipment. The multi-objective prediction engine 4 is connected to the digital management module 3. The multi-objective prediction engine 4 includes a component life prediction unit 41, a collaborative evaluation unit 42, a maintenance generation module 43, and a strategy feedback module 44. The component life prediction unit 41 is configured to call a pre-trained machine learning model to independently predict the remaining lifespan of the compressor, fan, heat exchanger, and filter based on the standardized feature vector. The collaborative evaluation unit 42 is configured to analyze the coupling impact of predicted component life decay on overall performance, energy consumption and other related components based on a predefined system coupling relationship knowledge base, and generate a system-level health index and risk warning signal. The maintenance generation module 43 is connected to the multi-objective prediction engine 4 and is configured to solve the maintenance strategy based on the system-level health index, risk warning signal, preset maintenance cost library and production plan information obtained from external systems, through a multi-objective optimization algorithm. The strategy feedback module 44 is configured to send the generated maintenance strategy to an external computerized maintenance management system or enterprise resource planning system through a standard interface, and to receive feedback data after the strategy is executed for model iterative optimization.

[0019] The system provided in this embodiment constructs a complete closed loop of "data perception - intelligent prediction - collaborative evaluation - decision generation - feedback optimization" through the collaborative work of four core modules. Its comprehensive technical effect lies in completely transforming the maintenance mode of industrial dehumidifiers from traditional passive response and periodic maintenance to data-driven proactive prediction and multi-objective optimization decision-making. The system can provide early warnings of component failure risks weeks or even months in advance and offer the optimal maintenance plan that comprehensively considers economy, reliability, and production planning, achieving intelligent and lean operation and maintenance management, and effectively ensuring the high-reliability, low-energy-consumption, and continuous operation of industrial dehumidifiers.

[0020] Data access module 1 is used to: establish connections with sensors and data acquisition devices deployed on the industrial dehumidifier body via Modbus TCP / IP, OPC UA, or MQTT communication protocols; receive and parse time-series data packets uploaded by sensors and data acquisition devices, and perform data format unification and timestamp alignment. This embodiment, by adopting industrial standard protocols such as Modbus TCP / IP and OPC UA, ensures that the system can seamlessly and stably access various heterogeneous sensors and data acquisition devices in the field, solving the problem of data interconnection and interoperability among multi-source devices. Parsing and timestamp alignment of time-series data packets ensures the consistency of data in time for subsequent analysis, laying a reliable data foundation for accurate time-series feature extraction and status analysis.

[0021] Data feature module 2 is specifically used for: The original time-series data were smoothed and denoised using a sliding window method and median filtering. Multidimensional data are normalized using the maximum-minimum method or Z-Score standardization to eliminate the influence of dimensions; Features including root mean square, amplitude, and frequency components are extracted from the time domain, frequency domain, and time-frequency domain to construct a feature vector of equipment health status.

[0022] This embodiment effectively removes impulse noise and random interference from the raw sensor data through median filtering using a sliding window method, significantly improving data quality. The use of the minimax method or Z-score normalization eliminates the negative impact of different physical dimensions such as vibration, current, and temperature on model training, enabling subsequent machine learning models to converge faster and more accurately. Comprehensive features are extracted from multiple domains (time domain and frequency domain) to construct feature vectors that deeply characterize the health status of equipment, greatly enhancing the model's ability to identify early signs of failure.

[0023] The system coupling relationship knowledge base is stored in matrix form, quantifying the impact coefficients of compressor efficiency decline on condenser load, evaporator scaling on fan power consumption, and filter blockage on system airflow and overall energy efficiency ratio. This embodiment, by constructing a quantified system coupling relationship knowledge base (stored in matrix form), transforms engineers' domain knowledge (such as the impact of compressor efficiency on condenser load) into computer-recognizable and computable model parameters. This enables the system to scientifically assess the cascading effects of individual component performance degradation on the entire system, avoiding the limitations of isolated predictions and achieving true system-level health status assessment and collaborative early warning, resulting in more comprehensive and reliable decision-making.

[0024] The maintenance generation module 43 employs a non-dominated sorting genetic algorithm (NSGA-II) for multi-objective optimization. Its objective function simultaneously minimizes total maintenance cost, maximizes system availability, and minimizes the impact of maintenance activities on production planning. This embodiment uses NSGA-II to solve the maintenance strategy. Its advantage lies in its ability to simultaneously balance the three conflicting objectives of "maintenance cost," "system availability," and "production impact," automatically finding a set of optimal Pareto solutions. This ensures that the final generated maintenance strategy is not merely a single-objective optimization, but a comprehensive optimal solution under multiple practical constraints, helping enterprises maximize operational efficiency.

[0025] The strategy feedback module 44 is also connected to a visualization dashboard component 441, which graphically displays the real-time status of the equipment, lifespan prediction curves, health trends, risk warnings, and recommended maintenance strategies to the user, and provides an interface for manual confirmation and work order dispatch. In this embodiment, the visualization dashboard connected to the strategy feedback module 44 presents complex prediction data, model results, and strategy suggestions to the user in a graphical and visual manner. Its effect is to greatly reduce the barrier to entry for the system, enabling maintenance personnel to intuitively and quickly grasp the health status of the equipment, understand the prediction results, confirm maintenance strategies, and dispatch work orders, thereby improving the efficiency of human-computer interaction and the transparency of decision-making.

[0026] The data access module 1 also connects to an external meteorological data interface to obtain future temperature and humidity forecast data for the equipment deployment location. The multi-objective prediction engine 4 uses this future temperature and humidity forecast data to correct the lifespan predictions of components such as filters and heat exchangers that are significantly affected by environmental factors. This embodiment innovatively incorporates future environmental factors (temperature and humidity) into the prediction model by accessing external meteorological data. Its technical effect is that the system can anticipate the impact of environmental changes on equipment components (such as filters being more prone to clogging under high temperature and humidity), thereby dynamically correcting the lifespan prediction results, making the prediction results more consistent with actual operating conditions, and improving the accuracy and adaptability of long-term predictions.

[0027] like Figures 1-2 As shown, a multi-objective collaborative predictive maintenance method for industrial dehumidifiers is based on a multi-objective collaborative predictive maintenance system. The method is executed by a software system and includes: Obtain multi-source real-time operating data of industrial dehumidifiers from external sources through data access service interfaces; The operational data is cleaned, transformed, and its features are extracted to generate standardized feature vectors; The standardized feature vector is input into the digital twin virtual model of the corresponding device to update its operating status; A pre-trained machine learning model is invoked to predict the remaining service life of key components based on the standardized feature vectors; Based on a system-coupled knowledge base, the synergistic impact of component lifespan degradation on the overall health of the machine is assessed, and a health index and risk level are generated. Based on health index, risk level, maintenance cost and production plan, the optimal maintenance strategy is generated through a multi-objective optimization algorithm. The optimal maintenance strategy is distributed to the external management system for execution through a standard interface, and feedback data is collected to optimize the model.

[0028] The method described in this embodiment defines a clear and executable set of steps, putting the system's functions into practice. This method provides a standardized and automated industrial dehumidifier maintenance and management process. By sequentially executing data processing, state mapping, lifespan prediction, collaborative evaluation, optimization decision-making, and closed-loop feedback, it ensures the value transformation from data to decision-making, realizing the replicability and scalability of the entire predictive maintenance process.

[0029] Training methods for pre-trained machine learning models include: Extract equipment operation data and corresponding component maintenance and failure records from historical databases; Perform the same cleaning, transformation, and feature extraction operations on historical data to construct a labeled training sample set; The model is trained and cross-validated using neural networks or ensemble learning algorithms; Deploy the trained model to the production environment and establish an online learning mechanism for continuous optimization.

[0030] This embodiment details the training method for the machine learning model, which ensures that the deployed predictive model has high accuracy and reliability. By learning real failure modes from historical data and continuously optimizing them through an online learning mechanism, the model can adapt to the aging characteristics and changing operating conditions of the equipment, thus guaranteeing the core capability of the predictive maintenance system to remain effective in the long term.

[0031] Generating optimal maintenance strategies through multi-objective optimization algorithms includes: Define a multi-objective optimization function with maintenance cost, system reliability, and production impact as its core considerations; The predicted lifespan and health index of each component are used as constraints. A multi-objective evolutionary algorithm is used to search for the Pareto optimal front in the solution space; The final execution strategy is selected from the Pareto solution set based on the decision-maker's preferences.

[0032] This embodiment details the specific implementation steps of the multi-objective optimization algorithm. Its effect is to transform the complex multi-objective decision-making problem into a computable and solvable mathematical optimization model. By searching the Pareto front and providing decision-makers with a set of optimal choices, the algorithm's scientific rigor is ensured while retaining flexibility for human intervention in decision-making, resulting in a final strategy that is both intelligent and meets actual management needs.

[0033] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A predictive maintenance system for industrial dehumidifiers with multi-objective collaboration, characterized in that: It includes a data access module, a data feature module, a digital management module, and a multi-objective prediction engine. The data access module is configured to acquire multi-source real-time operating data of the industrial dehumidifier from an external sensor network and data acquisition device through a preset application programming interface. The multi-source real-time operating data includes vibration, current, voltage, temperature, pressure, differential pressure, and temperature and humidity data. The data feature module is connected to the data access module and is configured to perform data noise reduction and feature extraction on the acquired multi-source real-time running data to generate a standardized feature vector. The digital management module is used to create and maintain a virtual model that corresponds one-to-one with the physical industrial dehumidifier. The input of the virtual model is the standardized feature vector, and the output is a real-time status mapping of the equipment. The multi-objective prediction engine is connected to the digital management module, and the multi-objective prediction engine includes a component life prediction unit, a collaborative evaluation unit, a maintenance generation module, and a strategy feedback module. The component life prediction unit is configured to invoke a pre-trained machine learning model to independently predict the remaining lifespan of the compressor, fan, heat exchanger, and filter based on the standardized feature vector. The collaborative evaluation unit is configured to analyze the coupling impact of predicted component life decay on overall performance, energy consumption and other related components based on a predefined system coupling relationship knowledge base, and generate a system-level health index and risk warning signal. The maintenance generation module is connected to the multi-objective prediction engine and is configured to solve the maintenance strategy based on the system-level health index, risk warning signal, preset maintenance cost library and production plan information obtained from external systems, through a multi-objective optimization algorithm. The strategy feedback module is configured to send the generated maintenance strategy to an external computerized maintenance management system or enterprise resource planning system through a standard interface, and to receive feedback data after the strategy is executed for model iterative optimization.

2. The predictive maintenance system for multi-objective collaborative operation of industrial dehumidifiers according to claim 1, characterized in that: The data access module is used to: establish a connection with the sensors and data acquisition devices deployed on the industrial dehumidifier body through Modbus TCP / IP, OPCUA or MQTT communication protocols; receive and parse the time-series data packets uploaded by the sensors and data acquisition devices, and perform data format unification and timestamp alignment.

3. The predictive maintenance system for multi-objective collaborative operation of industrial dehumidifiers according to claim 2, characterized in that: The data feature module is specifically used for: The original time-series data were smoothed and denoised using a sliding window method and median filtering. Multidimensional data are normalized using the maximum-minimum method or Z-Score standardization to eliminate the influence of dimensions; Features including root mean square, amplitude, and frequency components are extracted from the time domain, frequency domain, and time-frequency domain to construct a feature vector of equipment health status.

4. The predictive maintenance system for multi-objective collaborative operation of industrial dehumidifiers according to claim 1, characterized in that: The system coupling relationship knowledge base is stored in matrix form, quantifying the impact coefficient of compressor efficiency reduction on condenser load, the impact coefficient of evaporator scaling on fan power consumption, and the impact coefficient of filter blockage on system air volume and overall energy efficiency ratio.

5. The predictive maintenance system for multi-objective collaborative operation of industrial dehumidifiers according to claim 1, characterized in that: The maintenance generation module employs a non-dominated sorting genetic algorithm as its multi-objective optimization algorithm. Its objective function simultaneously minimizes the total maintenance cost, maximizes system availability, and minimizes the impact of maintenance activities on production planning.

6. The predictive maintenance system for multi-objective collaborative operation of industrial dehumidifiers according to claim 1, characterized in that: The strategy feedback module is also connected to a visualization dashboard component, which is used to graphically display the real-time status of the equipment, life prediction curve, health trend, risk warning and recommended maintenance strategy to the user, and provides a manual confirmation and work order dispatch interface.

7. The predictive maintenance system for multi-objective collaborative operation of industrial dehumidifiers according to claim 1, characterized in that: The data access module also connects to an external meteorological data interface to obtain future temperature and humidity forecast data for the equipment deployment site. The multi-objective prediction engine uses the future temperature and humidity forecast data to perform life prediction corrections for components such as filters and heat exchangers that are significantly affected by environmental factors.

8. A method for multi-objective collaborative predictive maintenance of industrial dehumidifiers based on the multi-objective collaborative predictive maintenance system of industrial dehumidifiers according to claim 1, characterized in that: The method is executed by a software system and includes: Obtain multi-source real-time operating data of industrial dehumidifiers from external sources through data access service interfaces; The operational data is cleaned, transformed, and its features are extracted to generate standardized feature vectors; The standardized feature vector is input into the digital twin virtual model of the corresponding device to update its operating status; A pre-trained machine learning model is invoked to predict the remaining service life of key components based on the standardized feature vectors; Based on a system-coupled knowledge base, the synergistic impact of component lifespan degradation on the overall health of the machine is assessed, and a health index and risk level are generated. Based on health index, risk level, maintenance cost and production plan, the optimal maintenance strategy is generated through a multi-objective optimization algorithm. The optimal maintenance strategy is distributed to the external management system for execution through a standard interface, and feedback data is collected to optimize the model.

9. The multi-objective collaborative predictive maintenance method for industrial dehumidifiers according to claim 8, characterized in that: The training methods for the pre-trained machine learning model include: Extract equipment operation data and corresponding component maintenance and failure records from historical databases; Perform the same cleaning, transformation, and feature extraction operations on historical data to construct a labeled training sample set; The model is trained and cross-validated using neural networks or ensemble learning algorithms; Deploy the trained model to the production environment and establish an online learning mechanism for continuous optimization.

10. The multi-objective collaborative predictive maintenance method for industrial dehumidifiers according to claim 8, characterized in that: The process of generating the optimal maintenance strategy using a multi-objective optimization algorithm includes: Define a multi-objective optimization function with maintenance cost, system reliability, and production impact as its core considerations; The predicted lifespan and health index of each component are used as constraints. A multi-objective evolutionary algorithm is used to search for the Pareto optimal front in the solution space; The final execution strategy is selected from the Pareto solution set based on the decision-maker's preferences.

Images