A medical laboratory equipment downtime early warning and consumable optimization system and implementation method

Abstract

This invention discloses a system and method for early warning of equipment downtime and optimization of consumables in pharmaceutical laboratories. The system includes a predictive maintenance module, a fault prediction module, a status monitoring module, a health management module, a resource scheduling module, an inventory optimization module, a scheduling management module, a configuration management module, and an operation and maintenance management module. This invention achieves accurate early warning of equipment faults and dynamic assessment of equipment health status by constructing an equipment health status prediction model and monitoring operational data in real time; it dynamically adjusts spare parts and consumable reserves through structured bill of materials management and intelligent inventory optimization; it maximizes equipment utilization efficiency and ensures continuous and stable experimental processes through intelligent resource scheduling, collaborative production scheduling, and adaptive optimization of configuration parameters; and finally, through integrated operation and maintenance management, it comprehensively reduces equipment downtime risks and management costs, bringing significant cost reduction and efficiency improvement benefits to pharmaceutical laboratories.

Description

Technical Field

[0001] This invention relates to the field of pharmaceutical laboratory equipment management technology, and more specifically, to a pharmaceutical laboratory equipment downtime early warning and consumables optimization system and its implementation method. Background Technology

[0002] With the rapid development of medical technology, the use of medical laboratory equipment is becoming increasingly widespread. However, existing equipment management methods still have many problems. Traditional maintenance methods mainly rely on manual records and regular maintenance, which is not only inefficient but also prone to omissions and errors. In addition, the equipment bill of materials (BOM) lacks structure, and the data exists in unstructured or semi-structured form, making it difficult to quickly and accurately locate the required information during maintenance resource management, equipment maintenance, and troubleshooting, thus affecting maintenance efficiency and decision-making quality.

[0003] Currently, the maintenance and management of pharmaceutical laboratory equipment mainly adopts a combination of scheduled maintenance and fault-based repair. This approach suffers from problems such as insufficient or excessive maintenance, as well as serious stockpiling of spare parts. Furthermore, the lack of effective predictive maintenance strategies makes it difficult to promptly identify and address potential equipment failure risks, leading to increased equipment downtime and impacting the normal operation of the laboratory.

[0004] To address these issues, a system capable of resource allocation, inventory optimization, scheduling, configuration, and operation and maintenance is urgently needed. This system should be able to monitor instrument status in real time through predictive maintenance and fault prediction technologies, optimize reagent inventory and equipment usage schedules, thereby improving equipment management efficiency, reducing maintenance costs, and ensuring the reliable operation of pharmaceutical laboratory equipment.

[0005] Several invention patents have been issued to address the issues of equipment downtime warning and consumable optimization in pharmaceutical laboratories. For example, CN114118548A discloses a method and system for collaborative scheduling of maintenance resources for tunnel boring machines. This method includes steps such as information resource acquisition, information storage, information processing and analysis, and collaborative scheduling service application for maintenance resources. The system includes an information acquisition module, a data transmission and storage module, an information processing and analysis module, and a maintenance resource scheduling module. The maintenance resource scheduling module includes a project resource management subsystem, a resource scheduling collaborative decision-making subsystem, and a supplier subsystem, enabling health assessment and collaborative scheduling of maintenance resources for tunnel boring machine equipment. However, this patent still faces the challenge of dynamically adjusting spare parts inventory based on the project department's maintenance needs and inventory status to reduce inventory costs and improve inventory efficiency. CN120278852A discloses a big data-based property maintenance prediction system. This system monitors equipment and the environment in real time using intelligent sensors, transmits data to the cloud or local storage via wireless network, and uses a distributed architecture for historical data processing and cleaning. The system uses machine learning to predict equipment failure risks, establishes customized prediction models for different equipment, analyzes the lifecycle, and predicts equipment lifespan. For aging equipment, a lifespan extension model is built using historical fault data. The system generates intelligent decision-making suggestions, pushes equipment health status to property management personnel, automatically formulates maintenance plans and priorities, and dispatches maintenance personnel. However, this patent still faces the challenge of how to dynamically adjust the inventory of consumables based on actual demand and stock levels to achieve cost reduction and efficiency improvement in consumable inventory management.

[0006] In summary, the existing technology has the following drawbacks: 1. Traditional maintenance methods mainly rely on manual recording and regular maintenance, which is inefficient, prone to omissions and errors, and affects maintenance quality and decision-making accuracy; 2. The equipment bill of materials (BOM) lacks structure, and the data exists in an unstructured or semi-structured form, which makes it impossible to quickly and accurately locate the required information during maintenance resource management, equipment maintenance, and troubleshooting, thus affecting maintenance efficiency and decision-making quality. 3. The existing maintenance and management methods combine scheduled maintenance and fault repair, which have problems such as insufficient maintenance, excess maintenance, and serious backlog of spare parts. It is impossible to dynamically adjust the spare parts inventory according to the project department's maintenance needs and inventory status, which affects inventory efficiency. 4. The lack of effective predictive maintenance strategies makes it impossible to detect and address potential equipment failure risks in a timely manner, leading to increased equipment downtime and affecting the normal operation of the laboratory; 5. The existing system has shortcomings in resource scheduling, inventory optimization, scheduling, configuration and operation and maintenance. It is difficult to dynamically adjust the reserve of consumables according to the actual demand and inventory status, and thus cannot achieve cost reduction and efficiency improvement in consumable inventory management.

[0007] In response to the above problems, how to solve the problems of low maintenance efficiency of pharmaceutical laboratory equipment, unreasonable inventory management, and lack of effective predictive maintenance strategies in existing technologies has become an urgent technical problem to be solved. Summary of the Invention

[0008] In view of the above-mentioned technical problems in related technologies, the present invention proposes a medical laboratory equipment downtime early warning and consumable optimization system and implementation method, which can overcome the above-mentioned shortcomings of the prior art.

[0009] To achieve the above-mentioned technical objectives, the technical solution of the present invention is implemented as follows: A medical laboratory equipment downtime early warning and consumables optimization system includes: The predictive maintenance module is used to acquire historical operating data of the equipment and build a predictive model of the equipment's health status based on machine learning algorithms; The status monitoring module is used to collect the operating data of the equipment in real time through sensors, and transmit the preprocessed operating data. The fault prediction module, connected to the predictive maintenance module and the status monitoring module, is used to input the real-time operating data collected by the status monitoring module into the equipment health status prediction model to assess the current health status of the equipment and predict future fault risks, and generate fault warning information accordingly. The health management module, connected to the fault prediction module, is used to formulate and output equipment maintenance plans and repair schemes based on the fault warning information. The inventory optimization module is used to analyze the historical consumption data and failure data of the equipment, predict the demand for consumables in future periods, and dynamically adjust the inventory of spare parts and consumables based on the prediction results, maintenance plans and repair schemes. The resource scheduling module, connected to the health management module, is used to intelligently schedule maintenance personnel and equipment resources according to the maintenance plan and repair scheme. The scheduling management module is used to dynamically adjust the scheduling plan for experiments or production based on the experimental production needs and the equipment operating status obtained from the fault prediction module. The configuration management module is used to optimize and adjust the configuration parameters of the device based on the device operation data collected by the status monitoring module; and The operation and maintenance management module is used to formulate and output long-term and short-term maintenance plans for equipment based on equipment failures and maintenance records; The various modules communicate and share data through a unified data exchange standard to form a complete equipment preventive maintenance system.

[0010] Furthermore, the predictive maintenance module further includes a demand learning unit, which employs a Stacking ensemble learning method to fuse the prediction results of multiple heterogeneous primary learners and train the equipment health status prediction model.

[0011] Furthermore, the process by which the predictive maintenance module constructs a predictive model for equipment health status includes: The historical operational data is subjected to feature engineering processing, including data normalization and feature selection; K-fold cross-validation is employed to train multiple primary learners using the processed data, and the outputs of each primary learner are used as meta-features. A univariate learner is then used to train these meta-features, learning the weight coefficients of each primary learner to obtain the final equipment health status assessment function: H(x) = σ, where... For the first The predicted output value of a primary learner for the device state; To be assigned to the The weight coefficients of each learner; For bias term (Bias), force; The Sigmoid activation function maps the result to... Interval.

[0012] Furthermore, the fault warning information includes at least the predicted fault type, the probability of fault occurrence, and the time window of fault occurrence.

[0013] Furthermore, the status monitoring module also includes a data cleaning unit, which is used to remove outliers and clean invalid data from the collected real-time operating data, and send the cleaned data to the cloud server for storage via the MQTT protocol.

[0014] Furthermore, the health management module is further configured to trigger an immediate repair command when the fault probability indicated by the fault warning information exceeds a first preset threshold; and to generate a maintenance plan according to a preset fixed cycle when the fault probability is lower than a second preset threshold.

[0015] Furthermore, the resource scheduling module employs a genetic algorithm to optimize the scheduling of maintenance personnel and equipment resources with the optimization objectives of workload balancing and minimizing total time cost, and supports collaborative scheduling with external maintenance resources.

[0016] Furthermore, the inventory optimization module uses the ARIMA time series forecasting model to predict future inventory demand and includes an alarm unit to issue an inventory alarm when the inventory level falls below a preset minimum threshold. The inventory optimization module automatically determines the optimal parameter combination of the ARIMA model using a grid search method based on the Akaike information criterion or the Bayesian information criterion. Furthermore, the scheduling management module adopts a sliding window algorithm and uses an optimization algorithm within each window to dynamically adjust the production schedule with the goal of minimizing the total waiting time and delay.

[0017] A method for implementing the aforementioned medical laboratory equipment downtime early warning and consumables optimization system includes the following steps: S1. Through the predictive maintenance module, a predictive model for equipment health status is trained and generated using historical operating data; S2. The status monitoring module collects and preprocesses the equipment's operating data in real time. S3. The fault prediction module inputs the preprocessed real-time operating data into the equipment health status prediction model and outputs the equipment health status assessment results and fault warning information. S4. The health management module formulates maintenance plans and repair schemes based on the fault warning information; S5. The inventory optimization module dynamically adjusts the consumables inventory based on historical data and the maintenance plan and repair scheme. S6. The resource scheduling module, scheduling management module, configuration management module and operation and maintenance management module shall, in accordance with the fault warning information, maintenance plan, repair scheme and inventory information, collaboratively perform operations of resource scheduling, scheduling optimization, parameter configuration and maintenance plan formulation.

[0018] The beneficial effects of this invention are as follows: By constructing a predictive model for equipment health status and monitoring operational data in real time, this invention achieves accurate early warning of equipment failures and dynamic assessment of equipment health status, significantly improving the accuracy and timeliness of maintenance decisions; through structured bill of materials management and intelligent inventory optimization, it dynamically adjusts spare parts and consumable reserves, effectively solving the problems of over-maintenance, inventory backlog, and supply shortages; through intelligent resource scheduling, collaborative production scheduling, and adaptive optimization of configuration parameters, it maximizes equipment utilization efficiency and ensures continuous and stable experimental processes; and finally, through integrated operation and maintenance management, it comprehensively reduces the risk of equipment downtime and management costs, bringing significant cost reduction and efficiency improvement benefits to pharmaceutical laboratories. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0020] A medical laboratory equipment downtime early warning and consumable optimization system according to an embodiment of the present invention includes a predictive maintenance module, a fault prediction module, a status monitoring module, a health management module, a resource scheduling module, an inventory optimization module, a scheduling management module, a configuration management module, and an operation and maintenance management module.

[0021] The predictive maintenance module employs machine learning technology to analyze historical equipment failure data and build a predictive model of equipment health status, enabling predictive maintenance of production equipment. This module also includes a demand learning unit that uses a stacking ensemble learning method to fuse the prediction results of multiple primary learners, improving the model's generalization ability.

[0022] The fault prediction module, based on an equipment health status prediction model, monitors the equipment's operating status in real time, predicts future fault risks, and generates fault warning information. This module also includes an intelligent analysis unit that further improves the accuracy of fault prediction through in-depth analysis of equipment operating data.

[0023] The status monitoring module collects equipment operation data in real time through intelligent sensors and transmits the data to a cloud server for processing and analysis. This module also includes a data cleaning unit, responsible for preprocessing and cleaning the collected data to ensure its accuracy and usability.

[0024] The health management module develops equipment maintenance plans and repair schemes based on equipment health status prediction models and fault early warning information. This module also includes a resource allocation unit, responsible for rationally allocating maintenance resources according to the maintenance and repair plans.

[0025] The resource scheduling module enables intelligent scheduling of equipment and personnel. Through production scheduling optimization algorithms, this module rationally allocates production tasks and optimizes the workload of equipment and personnel. Simultaneously, this module also supports collaborative scheduling of external maintenance resources, achieving resource sharing and optimization.

[0026] The inventory optimization module analyzes historical data on equipment consumption and failures to predict equipment consumption over a future period and dynamically adjusts spare parts inventory levels based on production schedules and maintenance plans. This module also includes an alarm unit that promptly issues an inventory alert when inventory levels fall below a preset threshold.

[0027] The scheduling management module dynamically adjusts the production schedule based on production needs and equipment operating status. This module further optimizes the production process and improves efficiency by analyzing waiting times and delays during production.

[0028] The configuration management module enables intelligent management of device configurations. By monitoring the device's usage status in real time, this module adjusts device configuration parameters promptly to ensure stable and reliable device operation.

[0029] The operation and maintenance management module analyzes equipment failures and repair status to develop long-term and short-term equipment maintenance plans. This module also includes a training unit responsible for providing regular training to equipment operators to improve their equipment operation and maintenance skills.

[0030] The various modules achieve data interoperability and sharing through a unified data exchange standard, forming a complete preventive maintenance system for production equipment. The system also includes a user-friendly human-machine interface to display equipment status, inventory information, and maintenance plans, facilitating viewing and management by administrators.

[0031] A method for implementing the aforementioned medical laboratory equipment downtime early warning and consumables optimization system includes the following steps: S1. Through the predictive maintenance module, a predictive model for equipment health status is trained and generated using historical operating data; S2. The status monitoring module collects and preprocesses the equipment's operating data in real time. S3. The fault prediction module inputs the preprocessed real-time operating data into the equipment health status prediction model and outputs the equipment health status assessment results and fault warning information. S4. The health management module formulates maintenance plans and repair schemes based on the fault warning information; S5. The inventory optimization module dynamically adjusts the consumables inventory based on historical data and the maintenance plan and repair scheme. S6. The resource scheduling module, scheduling management module, configuration management module and operation and maintenance management module shall, in accordance with the fault warning information, maintenance plan, repair scheme and inventory information, collaboratively perform operations of resource scheduling, scheduling optimization, parameter configuration and maintenance plan formulation.

[0032] To facilitate understanding of the above technical solutions of the present invention, the following detailed description of the above technical solutions of the present invention will be provided through specific usage methods.

[0033] In practical use, the medical laboratory equipment downtime early warning and consumable optimization system according to the present invention includes a predictive maintenance module, a fault prediction module, a status monitoring module, a health management module, a resource scheduling module, an inventory optimization module, a scheduling management module, a configuration management module, and an operation and maintenance management module.

[0034] The predictive maintenance module employs machine learning technology to build a predictive model for equipment health status by analyzing historical equipment failure data. Specifically, the module first acquires historical operating data of the equipment, extracts its operating parameters and environmental parameters to form a training set, and then uses this training set to train the equipment health status prediction model using a random forest algorithm. This model can predict the future health status of the equipment based on its current operating state.

[0035] The fault prediction module, based on an equipment health status prediction model, monitors the equipment's operating status in real time and predicts future fault risks. Specifically, the module first collects real-time operating data from the equipment's operating status monitoring system; then, it inputs this data into the health status prediction model, which outputs a health status assessment result; finally, based on the assessment result, the module generates fault warning information. This warning information includes fault type, fault probability, and fault time window.

[0036] The status monitoring module collects equipment operating data in real time through intelligent sensors and transmits the data to a cloud server for processing and analysis. Specifically, the status monitoring module uses a data acquisition unit to collect the equipment's operating parameters, environmental parameters, and status parameters; then, the collected data is sent to the cloud server via a network transmission protocol; finally, the data is cleaned and stored in the cloud.

[0037] The health management module formulates equipment maintenance plans and repair schemes based on equipment health status prediction models and fault early warning information. Specifically, the health management module first obtains the equipment health status assessment results and fault early warning information; then, based on the assessment results and early warning information, it generates equipment maintenance and repair plans; and finally, it allocates the corresponding maintenance resources through the resource allocation unit.

[0038] The resource scheduling module enables intelligent scheduling of equipment and personnel. Specifically, the resource scheduling module first obtains the production schedule and equipment operating status; then, it uses a genetic algorithm to optimize the work schedule of equipment and personnel to ensure a balanced workload for both; finally, it coordinates with external maintenance resources through the collaborative scheduling module.

[0039] The inventory optimization module analyzes historical data on equipment consumption and failures to predict equipment consumption over a future period and dynamically adjusts spare parts inventory levels based on production schedules and maintenance plans. Specifically, the module first acquires historical data on equipment consumption and failures; then uses time series analysis to predict future inventory demand; and finally adjusts spare parts inventory levels based on the predicted values ​​and actual inventory levels.

[0040] The scheduling management module dynamically adjusts the production schedule based on production demand and equipment operating status. Specifically, the scheduling management module first obtains the production demand plan; then, it adjusts the production schedule based on equipment operating status and fault warning information; and finally, it optimizes production waiting time and delays using queue theory in statistics.

[0041] The configuration management module enables intelligent management of device configurations. Specifically, the module first acquires device operating status data; then, based on historical data of device operating parameters, it adjusts the device configuration parameters; and finally, it optimizes the configuration parameters through machine learning algorithms to make the device operation more stable and reliable.

[0042] The operation and maintenance management module analyzes equipment failures and repair status to develop long-term and short-term equipment maintenance plans. Specifically, the module first obtains equipment failure and repair records; then, it uses regression analysis in statistics to predict future equipment maintenance needs; and finally, it develops maintenance plans and assigns them to equipment maintenance personnel.

[0043] The various modules achieve data interoperability and sharing through a unified data exchange standard, forming a complete preventive maintenance system for production equipment. The system also includes a user-friendly human-machine interface to display equipment status, inventory information, and maintenance plans, facilitating viewing and management by administrators.

[0044] The specific implementation steps of a medical laboratory equipment downtime early warning and consumables optimization system are as follows: Step 1: The predictive maintenance module trains the equipment health status prediction model. The model construction follows the logic of "multi-algorithm integration" and "feature weighting." The specific generation process is as follows: 1. Data Preprocessing and Feature Engineering First, obtain the historical operating data of the equipment from the status monitoring module. Let the dataset be denoted as , which contains feature vectors of dimensions such as temperature, pressure, and current (e.g., normal, sub-healthy, faulty).

[0045] 1) Normalization: Standardize data of different dimensions to ensure model convergence; 2) Feature selection: Using principal component analysis (PCA) or Pearson correlation coefficient, redundant environmental parameters are removed, and the Top-N features with the highest correlation to the fault are retained.

[0046] 2. Initial learner training A stacking ensemble learning architecture is adopted. The first layer deploys multiple heterogeneous primary learners L1, L2, ... k (such as random forest, support vector machine, etc.).

[0047] 1) Cross-validation training: Perform K-fold cross-validation on 60% of the training set; 2) Generate meta-features: Each primary learner predicts a sample, and its output will be used as the input feature of the second-layer meta-learner; 3. Meta-learner fusion The second layer uses logistic regression or a lightweight gradient booster as the meta-learner.

[0048] 1) Model Fusion: Let the predicted output of the primary learner be , and the meta-learner learns the weight coefficients ω to obtain the final health status assessment function: H(x) = σ in: : No. The predicted output value of the device state by a primary learner (such as random forest, SVM); : Assigned to the The weight coefficients of each learner represent the contribution of the algorithm to the model; Bias is used to adjust the overall offset of the model and improve its fitting ability. The Sigmoid activation function maps the result to... The interval represents the health probability score.

[0049] 4. Model Evaluation and Persistence The accuracy, recall, and F1 score of the model were validated using 40% of the prediction set. When the model's performance on the test set exceeded the preset threshold (F1) of 0.85, the model parameters were serialized and deployed to the production environment for real-time use by the fault prediction module.

[0050] First, historical operating data of the equipment is collected, and its operating parameters and environmental parameters are extracted to form a training set. Then, using the Stacking ensemble learning method, multiple decision tree algorithms are used as primary learners to train a health status prediction model for the equipment. Specifically, 60% of the dataset is used as the training set, and 40% as the test set. For each primary learner, the top 10% of the dataset is used as bootstrap cluster samples, and the remaining dataset is used as bagged samples. After training, the prediction results from multiple primary learners are fused using the Stacking ensemble learning method to obtain the final health status prediction model.

[0051] Step 2: The fault prediction module predicts equipment faults. First, it collects real-time operating data from the equipment through the equipment operation status monitoring system, including operating parameters, environmental parameters, and status parameters. Then, the real-time data is input into the health status prediction model, which outputs the equipment's health status assessment results. Based on the assessment results, the fault prediction module generates fault warning information, including fault type (sudden fault, deterioration fault, temporary fault, permanent fault, etc.), fault probability (0-1), and fault time window (unit: hours). For example, for equipment with a fault probability greater than 0.8, a fault warning is generated immediately.

[0052] Step 3: The status monitoring module performs data acquisition and cleaning. First, the data acquisition unit collects the equipment's operating parameters, environmental parameters, and status parameters, including temperature, humidity, pressure, flow rate, current, and voltage. Then, the collected data is sent to the cloud server via the MQTT protocol. Finally, the data is cleaned and stored in the cloud, removing outliers and invalid data.

[0053] Step 4: The health management module develops a maintenance plan. First, it obtains the equipment's health status assessment results and fault warning information. Then, based on the assessment results and warning information, it generates a maintenance and repair plan for the equipment. For example, for equipment with a "moderate" health status assessment result and a fault probability greater than 0.6, repairs are scheduled immediately; for equipment with a "good" health status assessment result, maintenance is performed according to a fixed schedule. Finally, the resource allocation unit allocates the corresponding maintenance resources.

[0054] Step 5: The resource scheduling module performs intelligent scheduling. First, it obtains the production schedule and equipment operating status. Then, it uses a genetic algorithm to optimize the work schedule of equipment and personnel, ensuring a balanced workload for both. Specifically, each device is encoded with 5 genes, each gene representing a time period. The crossover probability is 0.8, and the mutation rate is 0.01. The optimization objectives are overall load balancing and minimizing total time cost. Finally, the module coordinates with external maintenance resources through a collaborative scheduling module to achieve resource sharing.

[0055] Step 6: The inventory optimization module manages inventory. First, it acquires historical data on equipment consumption and failures. Then, it uses an ARIMA model to predict future inventory demand. Specifically, it employs an ARIMA(p,d,q) model, where p is the number of autoregressive terms, d is the difference frequency, and q is the number of moving average terms. It uses the AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion) minimization principle. The system uses automated grid search to iterate through combinations of p, d, and q within a certain range, selecting the parameters with the least information loss. For the time span, it selects historical consumption data from the past 12-24 months to capture seasonal trends. For the forecast period, it sets two days (T+7 for weekly forecasts) or T+30 for monthly forecasts to match the procurement lead time for medical consumables.

[0056] The optimal model parameters are sought through trial and error. Finally, the spare parts inventory is adjusted based on the predicted values ​​and actual inventory levels. When the inventory level falls below a minimum threshold, an inventory alarm is generated.

[0057] Step 7: The scheduling management module optimizes the production schedule. First, the production demand plan is obtained. Then, the production schedule is adjusted based on equipment operating status and fault warning information. Specifically, a sliding window algorithm is used, with a window size of 7 days. Within each window, the CPLEX optimization algorithm is used to optimize the production schedule, with the objective function being to minimize the sum of total waiting time and delay. Finally, waiting time and delay in the production process are analyzed using queue theory in statistics to further optimize the production process.

[0058] Step 8: The configuration management module optimizes device configuration. First, it acquires device operating status data. Then, based on historical data of device operating parameters, it adjusts the device configuration parameters. Specifically, it uses a gradient boosting algorithm for parameter optimization, with each iteration step size of 0.1. After training, it uses machine learning algorithms to further optimize configuration parameters, making the device operation more stable and reliable. For example, for the temperature parameter that needs to be controlled, the optimization range is 20-40℃, with an accuracy of 0.1℃.

[0059] Step 9: The Operations and Maintenance Management module develops a maintenance plan. First, it obtains equipment failure and repair records. Then, it uses regression analysis to predict future equipment maintenance needs. Specifically, a multinomial regression model is used, considering factors such as equipment usage time, failure type, and environment. Finally, a maintenance plan is developed and issued to equipment maintenance personnel, including maintenance time, maintenance content, and maintenance requirements.

[0060] The specific application scenarios of a medical laboratory equipment downtime early warning and consumables optimization system are as follows: Predictive maintenance module. The system predicts the health status of spectrometers in the laboratory. First, it collects historical operating data from the spectrometers, extracting operating parameters including temperature, humidity, pressure, current, and voltage. Then, using the Stacking ensemble learning method, it trains a health status prediction model for the spectrometers using algorithms such as decision trees, logistic regression, and support vector machines as primary learners. After training, the system can predict the future health status of the spectrometers based on their real-time operating status.

[0061] Fault Early Warning Module. The system provides fault early warnings for the ultrasonic cleaning equipment in the laboratory. When the ultrasonic cleaning equipment exhibits abnormal operating parameters, the fault early warning module immediately generates fault warning information, including the fault type (temporary fault), fault probability (0.8), and fault time window (2 hours). Simultaneously, the fault warning information is sent to equipment maintenance personnel via SMS and email.

[0062] Status monitoring module. The system collects real-time operating parameters of the refrigeration equipment in the laboratory, including temperature, humidity, and pressure, through sensors. The collected data is sent to a cloud server via the MQTT protocol. Data cleaning and storage are performed in the cloud, removing outliers and invalid data. For example, outliers in temperature data (such as -20℃) and outliers in humidity data (such as 95%) are removed.

[0063] Health Management Module. Based on the spectrometer's health status assessment results and fault warning information, the system formulates a maintenance plan. For example, for spectrometers with a "moderate" health status assessment result and a fault probability greater than 0.6, repairs are arranged immediately; for spectrometers with a "good" health status assessment result, maintenance is performed on a fixed weekly schedule. Appropriate maintenance resources are allocated through the resource allocation unit, including the arrangement of maintenance personnel and spare parts.

[0064] Resource scheduling module. The system optimizes the work schedule of laboratory equipment and personnel using a genetic algorithm. First, the production schedule and equipment operating status are obtained. Then, each piece of equipment is encoded using 5 genes, with each gene representing a time period. The crossover probability is 0.8, and the mutation rate is 0.01. The optimization objectives are overall load balancing and minimizing total time cost. Resource sharing is achieved through collaborative scheduling with external maintenance resources via a collaborative scheduling module.

[0065] Inventory optimization module. The system predicts the demand for consumables used in the laboratory using an ARIMA model. First, it acquires historical data on equipment consumption and failures. Then, it uses an ARIMA(2,1,2) model to predict future inventory demand. Optimal model parameters are sought through trial and error. Finally, based on the predicted values ​​and actual inventory levels, spare parts inventory is adjusted. When inventory levels fall below a minimum threshold, an inventory alarm is generated.

[0066] The scheduling management module dynamically adjusts the production schedule based on experimental requirements and equipment operating status. First, it obtains the production demand plan. Then, it adjusts the production schedule based on equipment operating status and fault warning information. A sliding window algorithm is used, with a window size of 7 days. Within each window, the CPLEX optimization algorithm is used to optimize the production schedule, with the objective function being to minimize the sum of total waiting time and delay.

[0067] Configuration Management Module. The system optimizes the configuration parameters of laboratory equipment using machine learning algorithms. First, it acquires equipment operating status data. Then, based on historical data of equipment operating parameters, it adjusts the equipment configuration parameters. A gradient boosting algorithm is used for parameter optimization, with each iteration step size of 0.1. After training, the machine learning algorithm optimizes the configuration parameters, making the equipment operation more stable and reliable. For example, for the temperature parameter that needs to be controlled, the optimization range is 20-40℃, with an accuracy of 0.1℃.

[0068] Operations and Maintenance Management Module. The system uses regression analysis to develop equipment maintenance plans. First, it acquires equipment fault and repair records. Then, it uses regression analysis to predict future equipment maintenance needs. Finally, it develops and assigns maintenance plans to equipment maintenance personnel, including maintenance time, content, and requirements.

[0069] In summary, by utilizing the technical solutions described above, this invention enables precise early warning of equipment failures and dynamic assessment of health status through the construction of an equipment health status prediction model and real-time monitoring of operational data, significantly improving the accuracy and timeliness of maintenance decisions. Through structured bill of materials management and intelligent inventory optimization, spare parts and consumable reserves are dynamically adjusted, effectively addressing issues of over-maintenance, inventory backlog, and supply shortages. Intelligent resource scheduling, collaborative production scheduling, and adaptive optimization of configuration parameters maximize equipment utilization efficiency and ensure continuous and stable experimental processes. Finally, integrated operation and maintenance management comprehensively reduces equipment downtime risks and management costs, bringing significant cost reduction and efficiency improvement benefits to pharmaceutical laboratories.

[0070] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A medical laboratory equipment downtime early warning and consumables optimization system, characterized in that, include: The predictive maintenance module is used to acquire historical operating data of the equipment and build a predictive model of the equipment's health status based on machine learning algorithms; The status monitoring module is used to collect the operating data of the equipment in real time through sensors, and transmit the preprocessed operating data. The fault prediction module, connected to the predictive maintenance module and the status monitoring module, is used to input the real-time operating data collected by the status monitoring module into the equipment health status prediction model to assess the current health status of the equipment and predict future fault risks, and generate fault warning information accordingly. The health management module, connected to the fault prediction module, is used to formulate and output equipment maintenance plans and repair schemes based on the fault warning information. The inventory optimization module is used to analyze the historical consumption data and failure data of the equipment, predict the demand for consumables in future periods, and dynamically adjust the inventory of spare parts and consumables based on the prediction results, maintenance plans and repair schemes. The resource scheduling module, connected to the health management module, is used to intelligently schedule maintenance personnel and equipment resources according to the maintenance plan and repair scheme. The scheduling management module is used to dynamically adjust the scheduling plan for experiments or production based on the experimental production needs and the equipment operating status obtained from the fault prediction module. The configuration management module is used to optimize and adjust the configuration parameters of the device based on the device operation data collected by the status monitoring module. as well as The operation and maintenance management module is used to formulate and output long-term and short-term maintenance plans for equipment based on equipment failures and maintenance records; The various modules communicate and share data through a unified data exchange standard to form a complete equipment preventive maintenance system.

2. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The predictive maintenance module further includes a demand learning unit, which employs a Stacking ensemble learning method to fuse the prediction results of multiple heterogeneous primary learners and train the device health status prediction model.

3. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The process by which the predictive maintenance module constructs a predictive model for equipment health status includes: The historical operational data is subjected to feature engineering processing, including data normalization and feature selection; K-fold cross-validation is employed to train multiple primary learners using the processed data, and the outputs of each primary learner are used as meta-features. A univariate learner is then used to train these meta-features, learning the weight coefficients of each primary learner to obtain the final equipment health status assessment function: H(x) = σ, where... For the first The predicted output value of a primary learner for the device state; To be assigned to the The weight coefficients of each learner; For bias term (Bias), force; The Sigmoid activation function maps the result to... Interval.

4. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The fault warning information includes at least the predicted fault type, the probability of fault occurrence, and the time window of fault occurrence.

5. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The status monitoring module also includes a data cleaning unit, which is used to remove outliers and clean invalid data from the collected real-time operating data, and send the cleaned data to the cloud server for storage via the MQTT protocol.

6. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The health management module is further configured to trigger an immediate repair command when the fault probability indicated by the fault warning information exceeds a first preset threshold; and to generate a maintenance plan according to a preset fixed cycle when the fault probability is lower than a second preset threshold.

7. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The resource scheduling module uses a genetic algorithm to optimize the scheduling of maintenance personnel and equipment resources with the optimization objectives of workload balancing and minimizing total time cost, and supports collaborative scheduling with external maintenance resources.

8. The medical laboratory equipment downtime early warning and consumables optimization system according to claim 1, characterized in that, The inventory optimization module uses the ARIMA time series forecasting model to predict future inventory demand and includes an alarm unit to issue an inventory alarm when the inventory level is lower than a preset minimum threshold. The inventory optimization module automatically determines the optimal parameter combination of the ARIMA model using a grid search method based on the Akaike information criterion or the Bayesian information criterion. The pharmaceutical laboratory equipment downtime early warning and consumables optimization system according to claim 1 is characterized in that the scheduling management module adopts a sliding window algorithm, and uses an optimization algorithm within each window to dynamically adjust the production schedule with the goal of minimizing the total waiting time and delay.

9. A method for implementing a medical laboratory equipment downtime early warning and consumables optimization system based on any one of claims 1 to 10, characterized in that, Includes the following steps: S1. Through the predictive maintenance module, a predictive model for equipment health status is trained and generated using historical operating data; S2. The status monitoring module collects and preprocesses the equipment's operating data in real time. S3. The fault prediction module inputs the preprocessed real-time operating data into the equipment health status prediction model and outputs the equipment health status assessment results and fault warning information. S4. The health management module formulates maintenance plans and repair schemes based on the fault warning information; S5. The inventory optimization module dynamically adjusts the consumables inventory based on historical data and the maintenance plan and repair scheme. S6. The resource scheduling module, scheduling management module, configuration management module and operation and maintenance management module shall, in accordance with the fault warning information, maintenance plan, repair scheme and inventory information, collaboratively perform operations of resource scheduling, scheduling optimization, parameter configuration and maintenance plan formulation.