Laboratory intelligent inspection and maintenance method based on cooperation of Internet of Things and robot

Abstract

The invention discloses a laboratory intelligent inspection and maintenance method based on cooperation of the Internet of Things and a robot, and relates to the technical field of laboratory management, and the maintenance method comprises the following specific steps: S100, data collection and transmission: deploying a plurality of Internet of Things sensors at key positions of a laboratory, collecting environment data and equipment operation state data in real time, and sending the environment data and the equipment operation state data to a server; according to the method, data models of different disciplines are constructed by combining knowledge of multiple disciplines of chemistry, physics and biology, real-time data are analyzed and monitored in real time, abnormal conditions in operation of a laboratory can be found in time, and in a knowledge graph fault traceability stage, the fault traceability of the laboratory is improved. According to the method, the map covering laboratory equipment, experimental process and environmental factor knowledge is constructed, and the comprehensive association degree of the fault features and related entities in the knowledge map is evaluated, so that the fault generation root can be quickly judged, the fault processing time is shortened, the fault processing accuracy is improved, and the fault processing efficiency is improved. Powerful support is provided for the operation of a laboratory.

Description

Technical Field

[0001] The present invention relates to the technical field of laboratory management, and particularly to an intelligent inspection and maintenance method for laboratories based on the collaboration of the Internet of Things and robots. Background Art

[0002] In today's era of rapid technological innovation, laboratories, as the core places for scientific research exploration and technological breakthroughs, the efficient, safe, and intelligent operation management has become the goal pursued by scientific research institutions and enterprises. The Internet of Things technology, as an important part of the new generation of information technology, through various sensors, RFID tags, and wireless communication technology means, realizes the deep integration of the physical world and the digital world. At the same time, with the continuous progress of robot technology, its applications in automation and intelligence are becoming increasingly widespread, providing new possibilities for laboratory management. Against this background, an intelligent inspection and maintenance method for laboratories that combines the Internet of Things and robot technology has emerged, aiming to comprehensively improve the operation efficiency and safety of laboratories through intelligent means.

[0003] Although traditional laboratory management methods have played a certain role in ensuring the smooth progress of scientific research activities, there are still many deficiencies. On the one hand, traditional methods mainly rely on manual inspections, which are not only time-consuming and laborious, but also difficult to achieve comprehensive and real-time monitoring of the laboratory environment. Problems can often only be discovered and processed after they occur, lacking predictability and initiative. On the other hand, traditional methods have limitations in data processing and analysis, and it is difficult to deeply explore the potential laws and abnormal information in laboratory operation data, thus restricting the improvement of laboratory management levels. In addition, traditional methods lack intelligent means in fault tracing and maintenance decision-making, resulting in low fault handling efficiency and high maintenance costs.

[0004] Therefore, developing an intelligent inspection and maintenance method for laboratories based on the collaboration of the Internet of Things and robots will effectively improve the operation efficiency and safety of laboratories and provide a more stable and reliable environmental support for scientific research activities. Summary of the Invention

[0005] The purpose of the present invention is to make up for the deficiencies of the existing technology and provide an intelligent inspection and maintenance method for laboratories based on the collaboration of the Internet of Things and robots. This method uses Internet of Things sensors to collect laboratory environment and equipment data in real time, and uses inspection robots to obtain multi-modal data to achieve comprehensive monitoring. Combining the analysis of different disciplinary models, a model is constructed for anomaly detection, and fault tracing reasoning is carried out through a knowledge graph to quickly locate problems.

[0006] In order to solve the above technical problems, the present invention provides the following technical solution: An intelligent inspection and maintenance method for laboratories based on the collaboration of the Internet of Things and robots. The specific steps of this maintenance method are as follows:

[0007] S100, Data Acquisition and Transmission: By deploying a variety of Internet of Things sensors at key positions in the laboratory, environmental data such as temperature, humidity, gas concentration, pressure, and vibration, as well as equipment operation status data, are collected in real time. At the same time, inspection robots equipped with a variety of sensors patrol according to a preset path to obtain multi-modal data such as images, thermal images, and sounds. The collected data is synchronously transmitted to the data processing center, where the data is preprocessed, standardized, and stored;

[0008] S200, Disciplinary Modeling and Anomaly Warning: Extract relevant data from the data center, combine the knowledge of multiple disciplines such as chemistry, physics, and biology in the laboratory to build data models for different disciplines. Using the built disciplinary data models, real-time data is analyzed and monitored in real time to detect abnormal situations during the operation of the laboratory. When the prediction results of the model show abnormalities, an alarm is automatically issued, including the type of abnormality, the location of occurrence, and the affected content, and relevant personnel are notified via text message, email, and pop-up windows in the laboratory management system;

[0009] S300, Knowledge Graph Fault Tracing: Build a knowledge graph covering laboratory equipment, experimental procedures, and environmental factors. When fault characteristics appear, evaluate the comprehensive correlation degree between the fault characteristics and the equipment usage history, experimental operation records, and environmental condition changes in the knowledge graph. By comparing the comprehensive correlation degrees of different entities, the root cause of the fault is determined;

[0010] S400, Intelligent Decision-making and Task Allocation: Collect real-time environment and equipment data of the laboratory through sensors, generate appropriate inspection and maintenance strategies in combination with historical data, and generate the task execution sequence according to the inspection and maintenance strategies and the status information of the robot;

[0011] S500, Robot Task Execution: The inspection robot that receives the task goes to the designated area as required, uses its own sensors and detection equipment to detect the environment and equipment in detail, uploads data in real time, automatically maintains simple faults, and reports complex faults to the Internet of Things platform for managers to arrange professional personnel to handle;

[0012] S600, Data Update and Optimization Feedback: After the task is completed, the new data and maintenance results are updated to the Internet of Things platform, disciplinary data models, and knowledge graphs. At the same time, according to the task execution situation and effect indicators, the inspection and maintenance strategies and task execution sequences are feedback-optimized.

[0013] Furthermore, the Internet of Things sensors equipped in the laboratory in S100, Data Acquisition and Transmission are: temperature sensors, humidity sensors, gas sensors, pressure sensors, vibration sensors; the sensors equipped on the robot are: high-definition cameras, infrared thermal imagers, gas detectors, and sound sensors.

[0014] Further, in the construction of the model in the chemical field in S200, subject modeling and anomaly warning, let the reaction rate be R, and the concentrations of reactants A, B, and C be C A 、C B 、C C , the environmental temperature be T, the pH value of the reaction system be pH, and the catalyst activity be S. The model construction formula is: where k is the reaction rate constant, m1, m2, and m3 are the reaction orders of reactants A, B, and C respectively, q is the temperature sensitivity coefficient, T0 is the standard reaction temperature, r is the pH influence coefficient, pH0 is the optimal reaction pH, and s is the catalyst activity index.

[0015] Further, in the construction of the model in the physical field in S200, subject modeling and anomaly warning, let the energy loss of the device be E, the current passing through be I, the magnetic field strength be B, the resistance of the device itself be R, and the environmental temperature be T. The model construction formula is: E = I 2 R×(1 + αβ)×(1 + β(T - T ref ))), where α is the magnetic field influence coefficient, β is the temperature coefficient, and T ref is the reference temperature.

[0016] Further, in the construction of the model in the biological field in S200, subject modeling and anomaly warning, let the cell proliferation rate be G, the concentration of nutrients in the culture medium be N, the culture environment temperature be T, the light intensity be L, and the initial density of the cells themselves be D0. The model construction formula is: where g is the basic cell proliferation rate constant, H is the nutrient concentration influence index, a is the temperature sensitive factor, T opt is the optimal culture temperature, J is the light influence coefficient, and k is the initial cell density influence factor.

[0017] Further, in S200, the determination of abnormal situations during laboratory operation in subject modeling and anomaly warning:

[0018] For the calculation result R of the chemical field model, let the normal reaction rate range be [R min , R max . When the reaction rate R real obtained by real-time calculation exceeds this range, it is determined as an abnormal situation;

[0019] For the calculation result E of the physical field model, let the normal energy loss range be [E min , E max . When the energy loss E real monitored in real time exceeds this range, it is determined as an abnormal situation;

[0020] For the calculation result G of the biological field model, the normal energy loss range is set as [G min , G max . When the real-time calculated G real exceeds this range, it is determined as an abnormal situation.

[0021] Furthermore, in S300, the specific steps of knowledge graph construction in knowledge graph fault tracing are as follows:

[0022] (1) Determine that the knowledge graph is used for laboratory fault tracing reasoning, covering the knowledge scope of equipment, experimental procedures, and environmental factors;

[0023] (2) Collect relevant data on equipment, experimental procedures, and environmental factors from multiple channels such as the equipment management system, researchers, and environmental monitoring system;

[0024] (3) Define laboratory equipment, experimental procedures, and environmental factors as entities respectively, and determine their attributes and unique identifiers;

[0025] (4) Define the used relationship between equipment and experimental procedures, the affected relationship between equipment and environmental factors, the dependent relationship between experimental procedures and environmental factors, the collaborative relationship between equipment and equipment, the usage relationship between experimental procedures and chemical reagents, and their attributes;

[0026] (5) Store the knowledge graph through a graph database and display it using Gephi.

[0027] Furthermore, in S300, in knowledge graph fault tracing, the comprehensive correlation degree between fault features and the equipment usage history, experimental operation records, and environmental condition changes in the knowledge graph is evaluated through a multi-dimensional fault association comprehensive evaluation formula. Let the fault feature vector be F, and the feature vectors corresponding to the equipment entity E, experimental procedure entity P, and environmental factor entity Q in the knowledge graph be E V , P V , Q V respectively, the equipment usage history correlation degree R E , the experimental operation record correlation degree R P , the environmental condition change correlation degree R Q . The calculation formula is: Among them, w1, w2, and w3 are the weight coefficients of the equipment usage history, experimental operation records, and environmental condition changes in fault reasoning respectively, and w1 + w2 + w3 = 1. R Total represents the comprehensive correlation degree, and the larger the value, the closer the comprehensive connection between the fault feature and the relevant entities in the knowledge graph.

[0028] Further, in the S400, in intelligent decision-making and task allocation, an inspection and maintenance strategy is generated through a historical association strategy evaluation formula. Let the real-time state vector of the laboratory be L = [l1, l2, …, l n , where l i represents each real-time index. At the same time, assume that there are k historical state samples in the historical experience library, and the state vector of the jth sample is H j = [h j1 , h j2 , …, h jn , and the corresponding inspection and maintenance strategy is marked as P1 j . Define the similarity function between the real-time state and the historical state as: where ω i is the weight of the ith index, and a comprehensive score is assigned to each historical strategy. The higher the score, the more the strategy matches the current real-time state. The formula is: Select the strategy corresponding to the historical strategy with the highest score as the current inspection and maintenance strategy.

[0029] Further, in the S400, in intelligent decision-making and task allocation, arrange the task execution order and path. Let the robot state vector be S R = [b, c, s], which respectively represent the battery power, load capacity, and remaining task duration. Score j is the calculated task score Score j . For the jth task, calculate the adaptation value M j of the task to the robot. The formula is: M j = Score j × (λ1b + λ2c + λ3s), where λ1, λ2, and λ3 are weight coefficients, and λ1 + λ2 + λ3 = 1. Assume that there are n tasks in total, and the task execution order is π = (π1, π2, …, π n ). Calculate the total revenue T of task execution. The formula is: where d(π i , π i+1 ) represents the length of the path passed by the robot when completing task π i and then going to execute task π i+1 . Arrange the tasks by maximizing the total revenue T.

[0030] Compared with the prior art, the intelligent inspection and maintenance method for laboratories based on the cooperation of the Internet of Things and robots has the following beneficial effects:

[0031] I. By deploying a variety of Internet of Things sensors and inspection robots, the present invention can collect real-time multi-dimensional environmental data such as temperature, humidity, gas concentration, pressure, and vibration in the laboratory, as well as equipment operation status data, achieving comprehensive monitoring of the laboratory environment. At the same time, the inspection robot can perform autonomous inspections along a preset path, obtaining multi-modal data such as images, thermal images, and sounds, further improving the accuracy and comprehensiveness of inspections. This intelligent inspection method not only reduces the burden of manual inspections but also improves the efficiency and precision of inspections, providing strong guarantee for the safe operation of the laboratory.

[0032] II. By combining knowledge of multiple disciplines such as chemistry, physics, and biology, the present invention constructs data models for different disciplines, performs real-time analysis and monitoring of real-time data, and can timely detect abnormal situations in the operation of the laboratory. In the knowledge graph fault tracing stage, by constructing a graph covering laboratory equipment, experimental procedures, and environmental factors knowledge, and evaluating the comprehensive correlation degree between fault characteristics and relevant entities in the knowledge graph, the present invention can quickly determine the root cause of the fault, not only shortening the time for fault handling but also improving the accuracy of fault handling, providing strong support for the operation of the laboratory.

[0033] Other advantages, objectives, and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0035] Figure 1 It is a flowchart of a laboratory intelligent inspection and maintenance method based on the cooperation of the Internet of Things and robots;

[0036] Figure 2 It is a process framework diagram of a laboratory intelligent inspection and maintenance method based on the cooperation of the Internet of Things and robots. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention objective, the following will, in conjunction with the accompanying drawings and preferred embodiments, describe in detail the specific embodiments, structures, features, and their effects of the present invention as follows.

[0038] Embodiment 1

[0039] Intelligent Patrol and Maintenance of Chemical Laboratories

[0040] Deploy temperature sensors, humidity sensors, gas sensors (detecting the concentration of harmful gases), pressure sensors, and vibration sensors at key positions in the chemical laboratory to collect environmental data and equipment operation status data in real time. For example, deploy temperature sensors and pressure sensors near the chemical reactor to monitor the temperature and pressure changes during the reaction process.

[0041] The patrol robot is equipped with a high-definition camera, an infrared thermal imager, a gas detector, and a sound sensor, and patrols according to a preset path to obtain multi-modal data of images, thermal images, and sounds of reaction equipment. For example, during the patrol, the robot takes pictures of the appearance of experimental instruments through the high-definition camera, detects the temperature distribution of the equipment through the infrared thermal imager, detects whether there is leakage of harmful gases in the laboratory air through the gas detector, and listens for abnormal sounds during equipment operation through the sound sensor.

[0042] Extract relevant data from the data center and construct a chemical experiment model in combination with knowledge in the chemical field. Let the reaction rate be R, and the concentrations of reactants A, B, and C be C A 、C B 、C C , the environmental temperature be T, the pH value of the reaction system be pH, and the catalyst activity be S. The model construction formula is: For the calculation result R of the domain model, assume that the range of normal reaction rate is [R min , R max . When the reaction rate R real calculated in real time exceeds this range, it is determined as an abnormal situation. For example, in a certain experiment, it is monitored in real time that the reaction rate suddenly drops significantly and is lower than R real , and the system determines it as abnormal.

[0043] Construct a knowledge graph covering the knowledge of laboratory equipment, experimental procedures, and environmental factors, determine the knowledge scope of the knowledge graph covering chemical laboratory equipment (such as reaction kettles, distillation devices), experimental procedures (such as synthesis steps, separation and purification processes), and environmental factors (such as temperature, humidity, ventilation conditions), obtain data on equipment models, service life, and maintenance records from the equipment management system, collect information on experimental operation steps and chemical reagents used from researchers, obtain data on real-time temperature, humidity, and gas concentration in the laboratory from the environmental monitoring system, define laboratory equipment, experimental procedures, and environmental factors as entities respectively, and determine their attributes and unique identifiers. For example, the attributes of a reaction kettle include capacity, material, and heating method, the attributes of an experimental procedure include reaction steps and reaction conditions, and the attributes of environmental factors include temperature value, humidity value, and harmful gas concentration value. Define the usage relationship between equipment and experimental procedures (such as a reaction kettle is used in a certain chemical reaction process), the affected relationship between equipment and environmental factors (such as too high temperature will affect the service life of the reaction kettle), the dependence relationship between experimental procedures and environmental factors (such as certain chemical reactions require a specific temperature and humidity environment), the collaborative relationship between equipment and equipment (such as a distillation device and a condensation device work together), the usage relationship between experimental procedures and chemical reagents (such as a certain reaction process uses specific chemical reagents) and their attributes. Store the knowledge graph through a graph database and display it using Gephi. When a fault feature (such as abnormal reaction products) appears, evaluate the comprehensive correlation degree of the fault feature with the equipment usage history, experimental operation records, and environmental condition changes in the knowledge graph. Let the fault feature vector be F, and the feature vectors corresponding to the equipment entity E, experimental procedure entity P, and environmental factor entity Q in the knowledge graph be E V , P V , Q V , the correlation degree of equipment usage history R E , the correlation degree of experimental operation records R P , the correlation degree of environmental condition changes R Q , and the calculation formula is: Judge the most likely root cause of the fault. For example, it is found that the impurity content of the reaction product is too high. Through knowledge graph analysis, it is found that the chemical reagent supplier was recently changed, and it may be the reagent purity problem that caused the fault.

[0044] Collect real-time environmental and equipment data in the laboratory through sensors, and generate appropriate inspection and maintenance strategies in combination with historical data. For example, according to historical data, when the reaction temperature fluctuates greatly, the inspection frequency needs to be increased and the heating system needs to be checked. Let the real-time state vector of the laboratory be L = [l1, l2,..., l n , where l i represents each real-time index. At the same time, assume that there are k historical state samples in the historical experience library, and the state vector of the jth sample is H j = [h j1, h j2 , …, h jn , the corresponding inspection and maintenance strategy is marked as P1 j , define the similarity function of the real-time state and the historical state as: Assign a comprehensive score to each historical strategy. The higher the score, the more the strategy matches the current real-time state. The formula is: Select the strategy corresponding to the historical strategy with the highest score as the current inspection and maintenance strategy. Generate the task execution order according to the inspection and maintenance strategy combined with the status information of the robot. Let the robot status vector be S R = [b, c, s], representing the battery power, load capacity, and remaining task duration respectively. Score j is the calculated task score Score j , for the jth task, calculate the adaptation value M of the task and the robot j , the formula is: M j = Score j ×(λ1b + λ2c + λ3s), where λ1, λ2, and λ3 are weight coefficients, and λ1 + λ2 + λ3 = 1. Suppose there are n tasks in total, and the task execution order is π = (π1, π2, …, π n ), calculate the total revenue T of task execution. The formula is: Sort the tasks by maximizing the total revenue T. For example, when it is found that the temperature of the reactor is abnormal, the system arranges the robot to first go to detect whether the temperature sensor of the reactor is faulty, then check the operation of the heating system, and at the same time select the most suitable robot to execute the task according to the battery power and load of the robot, and plan the optimal inspection path.

[0045] The inspection robot that receives the task goes to the specified area as required, uses its own sensors and detection equipment to detect the environment and equipment in detail, and uploads data in real time. For example, after the robot arrives at the position of the reactor, it uses a high-definition camera to take pictures of the appearance of the reactor, checks for signs of leakage, uses an infrared thermal imager to detect the surface temperature distribution of the reactor to judge whether there is local overheating, uses a gas detector to detect the concentration of harmful gases in the surrounding air, uses a sound sensor to listen to whether the operation sound of the reactor is normal, and uploads this data to the Internet of Things platform in real time. For simple faults, automatic maintenance is performed. For example, if it is found that the temperature sensor of the reactor is loose and causes inaccurate temperature measurement, the robot automatically adjusts the position of the sensor and tightens the screws. For complex faults, they are reported to the Internet of Things platform for the management personnel to arrange professional personnel to handle. For example, if it is detected that the heating system of the reactor fails and the robot cannot repair it by itself, the fault information is reported, and the management personnel arrange maintenance personnel to repair it.

[0046] After the task is completed, update the new data and maintenance results to the Internet of Things platform, the disciplinary data model, and the knowledge graph. For example, update the reactor temperature and pressure change data obtained during this inspection and maintenance process, as well as the equipment status data after maintenance, to the corresponding systems for subsequent analysis and reference. According to the task execution situation and effectiveness indicators, feedback and optimize the inspection and maintenance strategies and the task execution order. For example, if it is found that the previously formulated inspection strategy fails to detect potential faults in a timely manner in some cases, adjust the strategy according to the actual situation, increase the inspection frequency of key equipment or optimize the weight of inspection indicators. If the task execution order results in low robot work efficiency, re-plan the task execution order to improve the overall work efficiency.

[0047] In summary, in the chemical laboratory, this invention collects data by means of a variety of sensors and inspection robots, discovers problems in a timely manner through analysis by different disciplinary models and anomaly detection, accurately traces the faults based on the knowledge graph, and after intelligent decision-making to allocate tasks, the robot executes and provides feedback for optimization. This series of processes effectively guarantees the safe and stable operation of the chemical laboratory, and improves the experimental efficiency and the level of management intelligence.

[0048] Example Two:

[0049] Intelligent Inspection and Maintenance of Biological Laboratories

[0050] Deploy temperature sensors, humidity sensors, gas sensors (detecting oxygen and carbon dioxide concentrations), pressure sensors, and vibration sensors in the biological laboratory to collect environmental data and the operating status data of cell culture equipment, centrifuges, and PCR equipment. For example, install temperature sensors and carbon dioxide concentration sensors in the cell culture incubator to monitor the culture environment parameters in real time. The inspection robot is equipped with a high-definition camera, an infrared thermal imager, a gas detector, and a sound sensor, and patrols according to a preset path to obtain multi-modal data of images, thermal images, and sounds of cell culture equipment and biological sample storage equipment. For example, when the robot patrols the cell culture area, it takes high-definition images of the cell growth status in the cell culture flask, detects the temperature uniformity of the culture equipment through the infrared thermal imager, uses the gas detector to detect whether the gas concentration in the incubator is normal, and uses the sound sensor to listen for any abnormalities in the equipment operation sound.

[0051] Extract relevant data from the data center and construct a biological experiment model in combination with biological domain knowledge. Let the cell proliferation rate be G, the nutrient concentration in the culture medium be N, the culture environment temperature be T, the light intensity be L, and the initial density of the cells themselves be D0. The model construction formula is: For the calculation result G of the biological domain model, assume that the range of normal energy loss is [G min , G max , when the real-time calculated G realWhen deviating from this range, it is determined as an abnormal situation. For example, during cell culture, if it is monitored that the cell proliferation rate is significantly lower than the normal range, it may indicate problems with the cell culture environment or nutrient supply.

[0052] Clarify the knowledge scope of the knowledge graph covering biological laboratory equipment (such as cell incubators, centrifuges, microscopes), experimental procedures (such as cell culture steps, gene amplification procedures), and environmental factors (such as temperature, humidity, light, cleanliness). Obtain information on the purchase date, calibration records, and fault history of equipment from the equipment management system, collect experimental operation procedures, biological reagents used, and sample information from researchers, and obtain real-time laboratory environmental data from the environmental monitoring system. Define laboratory equipment, experimental procedures, and environmental factors as entities respectively, and determine their attributes and unique identifiers. For example, the attributes of a cell incubator include temperature control accuracy, humidity adjustment range, and carbon dioxide concentration control range; the attributes of an experimental procedure include culture time, passage number, types and times of added reagents; the attributes of environmental factors include real-time temperature value, humidity value, light intensity value, and cleanliness level. Define the usage relationships between equipment and experimental procedures (such as a cell incubator is used for the cell culture experimental procedure), the affected relationships between equipment and environmental factors (such as substandard cleanliness will affect cell growth in the cell incubator), the dependency relationships between experimental procedures and environmental factors (such as certain cell culture procedures require specific temperature and light conditions), the collaborative relationships between equipment and equipment (such as the collaborative use of a centrifuge and a pipette during sample processing), the usage relationships between experimental procedures and biological reagents (such as specific culture media and sera are used in the cell culture procedure) and their attributes. Store the knowledge graph through a graph database and display it using Gephi. When a fault feature (such as cell contamination, abnormal cell growth) appears, evaluate the comprehensive correlation degree between the fault feature and the equipment usage history, experimental operation records, and environmental condition changes in the knowledge graph. Let the fault feature vector be F, and the feature vectors corresponding to the equipment entity E, experimental procedure entity P, and environmental factor entity Q in the knowledge graph be E V 、P V 、Q V , the correlation degree of equipment usage history R E 、the correlation degree of experimental operation records R P 、the correlation degree of environmental condition changes R Q , and the calculation formula is: Judge the root cause of the fault. For example, if it is found that the cell culture is contaminated, through knowledge graph analysis, it is found that recently the experimental personnel did not strictly follow the aseptic operation specifications during the operation, or the filter of the cell incubator may need to be replaced.

[0053] Based on the real-time environmental and device data collected by sensors, combined with historical data, inspection and maintenance strategies are generated. For example, according to historical experience, when the temperature fluctuation of the cell incubator exceeds a certain range, it is necessary to check the temperature control system and the refrigeration and heating components. The inspection and maintenance strategies are generated through the historical correlation strategy evaluation formula. Let the real-time state vector of the laboratory be L = [l1, l2, …, l n , where l i represents each real-time index. At the same time, assume that there are k historical state samples in the historical experience library, and the state vector of the jth sample is H j = [h j1 , h j2 , …, h jn , and the corresponding inspection and maintenance strategy is marked as P1 j . Define the similarity function between the real-time state and the historical state as: where ω i is the weight of the ith index. A comprehensive score is assigned to each historical strategy. The higher the score, the more the strategy matches the current real-time state. The formula is: Select the strategy corresponding to the historical strategy with the highest score as the current inspection and maintenance strategy, and generate the task execution order in combination with the robot state information. For example, when it is found that the temperature of the cell incubator is abnormal, the system arranges the robot to detect the temperature sensor of the incubator, check the refrigeration and heating system, and plan the optimal path according to the robot's power, load capacity, and remaining task duration. Let the robot state vector be S R = [b, c, s], which represent power, load capacity, and remaining task duration respectively. Score j is the calculated task score Score j . For the jth task, calculate the adaptation value M j of the task to the robot. The formula is: M j = Score j ×(λ1b + λ2c + λ3s), where λ1, λ2, and λ3 are weight coefficients, and λ1 + λ2 + λ3 = 1. Assume there are n tasks in total, and the task execution order is π = (π1, π2, …, π n ). Calculate the total benefit T of task execution. The formula is: Order the tasks by maximizing the total benefit T.

[0054] The robot travels to the designated area according to the task requirements, uses its own sensors and detection equipment to detect the environment and equipment, and uploads data in real time. For example, after the robot arrives at the position of the cell incubator, it uses a high-definition camera to check the placement of cell culture bottles in the incubator and the cell growth status, uses an infrared thermal imager to detect whether the internal temperature distribution of the incubator is uniform, uses a gas detector to detect whether the carbon dioxide concentration is normal, and uses a sound sensor to monitor the running sound of the incubator fan components, and uploads the data in real time. For simple faults, it performs automatic maintenance. For example, if the incubator door is not closed tightly, the robot automatically closes the incubator door. For complex faults, it reports them, and the management personnel arranges professional personnel to handle them. For example, if a temperature control system fault of the cell incubator is detected and the robot cannot repair it, it reports the fault information and waits for professional maintenance personnel to repair it.

[0055] After the task is completed, the data of the Internet of Things platform, the disciplinary data model, and the knowledge graph are updated. For example, the data on temperature, humidity, cell growth status changes during the cell culture process, and equipment maintenance conditions are updated to the corresponding systems. According to the task execution situation and effectiveness indicators, the inspection and maintenance strategies and task execution sequences are optimized. For example, if it is found that the equipment failure rate in a certain area is relatively high, the inspection frequency of that area is increased. If the path planning during the robot's task execution is unreasonable and causes too long time, the path planning strategy is re-planned to improve the overall inspection and maintenance efficiency.

[0056] In summary, for a biological laboratory, the present invention also realizes the effective monitoring and maintenance of the biological experimental environment and equipment through data collection, different disciplinary model analysis, knowledge graph reasoning, intelligent decision-making and robot collaboration, and data update and optimization links. It can timely detect abnormalities in cell culture experiments, accurately locate faults, and improve the operation reliability and intelligent management ability of the biological laboratory.

[0057] The above is only a preferred embodiment of the present invention, and it is not intended to limit the present invention in any form. Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the above-disclosed technical content to be equivalent embodiments with equivalent changes, but as long as it does not depart from the technical content of the present invention, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots, characterized in that: The specific steps of this maintenance method are: S100, data collection and transmission: By deploying a variety of IoT sensors at key locations in the laboratory, environmental data such as temperature, humidity, gas concentration, pressure, vibration, and equipment operation status data are collected in real time. At the same time, inspection robots equipped with a variety of sensors are used to inspect along preset routes to obtain multi-modal data such as images, thermal imaging, and sound. The collected data is simultaneously transmitted to the data processing center for pre-processing, standardization, and storage. S200, discipline modeling and abnormal warning: extract relevant data from the data center, combine the knowledge of chemistry, physics, and biology in the laboratory, build data models of different disciplines, use the constructed discipline data models to analyze and monitor real-time data in real time, detect abnormal situations in laboratory operations, and automatically send out an abnormality including the type of abnormality, location, and impact content, and notify relevant personnel through SMS, email, and pop-up windows in the laboratory management system when the model predicts an abnormality. S300, knowledge graph fault tracing: Build a knowledge graph covering laboratory equipment, experimental processes, and environmental factors. When a fault feature appears, evaluate the comprehensive correlation between the fault feature and the equipment usage history, experimental operation records, and environmental condition changes in the knowledge graph. By comparing the comprehensive correlation of different entities, the root cause of the fault is determined; S400, intelligent decision-making and task allocation: collects real-time environment and equipment data of the laboratory through sensors, generates adaptive inspection and maintenance strategies based on historical data, and generates task execution sequence based on the inspection and maintenance strategies and the robot's status information; S500, robot execution: the inspection robot that receives the task goes to the designated area as required, uses its own sensors and detection equipment to detect the environment and equipment in detail, uploads data in real time, automatically maintains simple faults, and reports complex faults to the IoT platform for management personnel to arrange professional personnel to handle; S600, data update and optimization feedback: After the task is completed, the new data and maintenance results are updated to the IoT platform, subject data model and knowledge graph. At the same time, based on the task execution status and effect indicators, the inspection and maintenance strategies and task execution sequence are fed back and optimized.

2. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: The IoT sensors equipped in the S100 data collection and transmission laboratory are: temperature sensor, humidity sensor, gas sensor, pressure sensor, vibration sensor; the sensors equipped on the robot are: high-definition camera, infrared thermal imager, gas detector, sound sensor.

3. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: S200, the construction of the chemical domain model in subject modeling and abnormal warning, assumes that the reaction rate is R, and the concentrations of reactants A, B, and C are C A , C B , C C , the ambient temperature is T, the pH of the reaction system is pH, the catalyst activity is S, and the model construction formula is: Among them, k is the reaction rate constant, m1, m2, m3 are the reaction orders of reactants A, B, C respectively, q is the temperature sensitivity coefficient, T0 is the standard reaction temperature, r is the pH influence coefficient, pH0 is the optimal reaction pH, and s is the catalyst activity index.

4. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: In the above S200, the construction of the physical domain model in the subject modeling and abnormal warning, the energy loss of the equipment is E, the current passing through is I, the magnetic field strength is B, the resistance of the equipment itself is R, and the ambient temperature is T. The model construction formula is: E=I 2 R×(1+αB)×(1+β(TT ref )), where α is the magnetic field influence coefficient, β is the temperature coefficient, T ref is the reference temperature.

5. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: S200, the construction of the biological field model in subject modeling and abnormal warning, assumes that the cell proliferation rate is G, the nutrient concentration in the culture medium is N, the culture environment temperature is T, the light intensity is L, and the initial density of the cell itself is D0. The model construction formula is: Among them, g is the basic rate constant of cell proliferation, H is the influence index of nutrient concentration, a is the temperature sensitivity factor, T opt is the optimum culture temperature, J is the light influence coefficient, and k is the initial cell density influence factor.

6. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: S200, determination of abnormal conditions in laboratory operation in subject modeling and abnormal warning: For the calculation result R of the chemical domain model, the range of normal reaction rate is [R min , R max ], when the reaction rate R calculated in real time real When it exceeds this range, it is considered an abnormal situation; For the calculation result E of the physical domain model, the range of normal energy loss is [E min , E max ], when the real-time monitored energy loss E real When it exceeds this range, it is considered an abnormal situation; For the calculation result G of the biological domain model, the range of normal energy loss is [G min , G max ], when the real-time calculation of G real When it exceeds this range, it is considered an abnormal situation.

7. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: In S300, the specific steps of constructing the knowledge graph in the knowledge graph fault tracing are: (1) Determine the knowledge graph used for laboratory fault tracing reasoning, covering the knowledge scope of equipment, experimental procedures, and environmental factors; (2) Collect relevant data on equipment, experimental processes, and environmental factors from multiple channels including equipment management systems, scientific researchers, and environmental monitoring systems; (3) Define laboratory equipment, experimental procedures, and environmental factors as entities, and determine their attributes and unique identifiers; (4) Define the usage relationship between equipment and experimental process, the influence relationship between equipment and environmental factors, the dependency relationship between experimental process and environmental factors, the synergistic relationship between equipment, and the usage relationship between experimental process and chemical reagents and their properties; (5) Store the knowledge graph in a graph database and display it using Gephi.

8. According to claim 1, a laboratory intelligent inspection and maintenance method based on the collaboration of the Internet of Things and robots is characterized in that: In S300, in the knowledge graph fault tracing, a multi-dimensional fault correlation comprehensive evaluation formula is used to evaluate the comprehensive correlation between the fault characteristics and the equipment usage history, experimental operation records, and environmental condition changes in the knowledge graph. Let the fault feature vector be F, and the feature vectors corresponding to the equipment entity E, the experimental process entity P, and the environmental factor entity Q in the knowledge graph be E. V , P V , Q V , equipment usage history correlation R E , Experimental operation record correlation R P 、Correlation degree of environmental condition change R Q , the calculation formula is: Among them, w1, w2, and w3 are the weight coefficients of equipment usage history, experimental operation records, and environmental condition changes in fault reasoning, and w1+w2+w3=1, R Total It represents the comprehensive correlation. The larger the value, the closer the comprehensive connection between the fault feature and the related entities in the knowledge graph.

9. The method for intelligent laboratory inspection and maintenance based on collaboration between the Internet of Things and robots according to claim 1 is characterized in that: In the S400, in the intelligent decision-making and task allocation, the inspection and maintenance strategy is generated by the historical association strategy evaluation formula, and the real-time state vector of the laboratory is L=[l1, l2, ..., l n ], where l i Represents various real-time indicators. At the same time, suppose there are k historical state samples in the historical experience database, and the state vector of the jth sample is H j =[h j1 ,h j2 ,…,h jn ], the corresponding inspection and maintenance strategy is marked as P1 j , define the similarity function between the real-time state and the historical state as: Among them, ω i is the weight of the ith indicator. A comprehensive score is assigned to each historical strategy. The higher the score, the more the strategy matches the current real-time status. The formula is: The strategy corresponding to the historical strategy with the highest score is selected as the current inspection and maintenance strategy.

10. The method for intelligent laboratory inspection and maintenance based on the collaboration of the Internet of Things and robots according to claim 1 is characterized in that: In S400, the task execution order and path are arranged in the intelligent decision-making and task allocation. Assume that the robot state vector is S R = [b, c, s], representing power, load capacity, and remaining task duration, respectively. Score j The calculated task score Score j , for the jth task, calculate the adaptation value M between the task and the robot j , the formula is: N j =Score j ×(λ1b+λ2c+λ3s), where λ1, λ2, and λ3 are weight coefficients, and λ1+λ2+λ3=1. Suppose there are n tasks in total, and the task execution order is π=(π1, π2, …, π n ), calculate the total revenue T of task execution, the formula is: Among them, d(π i , π i+1 ) indicates that the robot completes task π i Then go to perform the task π i+1 The length of the path traversed when the task is completed is the same as the length of the path traversed when the task is completed, and the tasks are ordered by maximizing the total benefit T.

Images