Laboratory intelligent control system

Abstract

The invention relates to the technical field of laboratory intelligent management, and discloses a laboratory intelligent control system, which comprises an equipment monitoring module, a sample full life cycle management module, a data acquisition and analysis module, a process management module, a multi-mode abnormal event causal association analysis module and a mobile terminal and cloud module. Multi-modal heterogeneous data from equipment, samples and processes are integrated through the data acquisition and analysis module and standardized processing is carried out, and the system realizes comprehensive perception of a laboratory environment. Wherein the multi-modal abnormal event causal correlation analysis module is used as a core, and the root cause of an abnormal event is identified by constructing a domain knowledge graph and based on a specific scoring mechanism, so that the efficiency and accuracy of root cause analysis are remarkably improved. The method corresponds to various functions of the system, and intelligent control over the laboratory is achieved through systematized steps. The laboratory operation efficiency, the data reliability, the traceability and the safety are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent laboratory management technology, specifically to an intelligent laboratory control system. Background Technology

[0002] Currently, with the increasing demands for experimental efficiency and data accuracy in scientific research and industrial production, the level of automation and informatization in laboratories is gradually improving. In modern laboratories, the collaboration among various sophisticated experimental equipment, diverse sample management, and complex experimental procedures is becoming increasingly close. To address this trend, laboratories have introduced digital tools such as equipment monitoring systems, sample management systems, and experimental information management systems, aiming to improve the standardization of experimental processes and the efficiency of data management.

[0003] However, existing technological solutions still face numerous challenges in practical applications. Various independently operating digital systems typically lack effective data integration and interaction mechanisms, leading to information silos formed from key data such as equipment operation data, sample flow information, experimental operation records, and environmental parameters. This data dispersion and heterogeneity makes it difficult to comprehensively and timely gain insight into the overall operational status of the laboratory from massive amounts of data. When abnormal events occur, such as equipment failure, abnormal sample storage environment, or deviation from experimental procedures, the lack of a unified data view and correlation analysis capabilities often makes it difficult to quickly and accurately identify the deep connections between events, resulting in low efficiency in root cause location and affecting the timeliness of problem solving. Furthermore, traditional human experience and rule-based judgment are easily influenced by subjectivity and are inefficient when dealing with multi-source and complex anomalies. This not only increases laboratory operating costs but also hinders the reliability of experimental data, the traceability of processes, and the intelligence of decision-making. Therefore, there is an urgent need for an intelligent laboratory control system and method that can effectively integrate multimodal data and achieve intelligent correlation analysis and root cause identification. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent control system for laboratories, which solves the problems of insufficient multimodal data integration, low efficiency in identifying the root causes of abnormal events, and limited level of intelligent control in existing laboratory management.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent laboratory control system, comprising:

[0006] The first aspect of this invention provides an intelligent laboratory control system, comprising:

[0007] The equipment monitoring module is configured to monitor the operating parameters of the experimental equipment and generate equipment abnormality events when the operating parameters are abnormal.

[0008] The sample lifecycle management module is configured to manage the status and storage environment of experimental samples, and generate sample environment abnormality events when the storage environment is abnormal.

[0009] The data acquisition and analysis module is configured to acquire and analyze experimental data, and generate a data anomaly event when the experimental data is abnormal.

[0010] The process management module is configured to monitor the execution of the experimental process and generate a process deviation event when a deviation occurs in the execution.

[0011] The multimodal abnormal event causal correlation analysis module is connected to the equipment monitoring module, sample life cycle management module, data acquisition and analysis module and process management module respectively. It is configured to receive the equipment abnormal events, sample environment abnormal events, data abnormal events and process deviation events, and perform multimodal abnormal event causal correlation analysis on the received events to determine the root cause of the target event.

[0012] The mobile and cloud modules are configured to interact with the user and present the root cause determined by the multimodal abnormal event causal correlation analysis module.

[0013] Through the above technical solution, this invention unifies the monitoring of information from four different modalities during laboratory operation: equipment, samples, data, and processes, and transforms the anomalies occurring in each modality into structured events. The multimodal anomaly causal correlation analysis module aggregates these events from different sources and performs correlation analysis through an algorithmic model. This allows it to penetrate information barriers, reveal the inherent causal relationships between different anomalies, and achieve automated root cause localization of experimental problems.

[0014] As a further technical solution of the first aspect of the present invention, the multimodal abnormal event causal correlation analysis module includes an abnormal event standardization unit and a causal correlation inference engine. The abnormal event standardization unit is configured to convert the equipment abnormal events, sample environment abnormal events, data abnormal events, and process deviation events into a unified standardized data structure. The causal correlation inference engine is configured to construct a causal correlation weighted directed graph based on the standardized data structure, and infer the root cause by performing a graph search on the graph.

[0015] As a further technical solution of the first aspect of the present invention, the weights of the edges in the causal association weighted directed graph are determined by the causal association inference engine based on the lift value of the association relationship between different events. Specifically, for any two standardized anomalous events... and The formula for calculating its lift is:

[0016] ;

[0017] in, and Representing events respectively and The probability of occurring independently, This represents the joint probability of two events occurring consecutively within a preset time window.

[0018] As a further technical solution of the first aspect of the present invention, the device monitoring module is configured to generate the device abnormal event when the monitored device operating parameters deviate from the preset physical threshold or statistical baseline.

[0019] As a further technical solution of the first aspect of the present invention, the sample life cycle management module is configured to generate an abnormal sample environment event when the monitored sample storage environment parameters exceed a preset range.

[0020] As a further technical solution of the first aspect of the present invention, the data acquisition and analysis module is configured to generate the data anomaly event when the deviation between the analyzed experimental data and its historical statistical data exceeds a preset threshold.

[0021] As a further technical solution of the first aspect of the present invention, the process management module is configured to generate the process deviation event when the actual execution progress of the experiment deviates from the preset plan or the execution order violates the preset dependency relationship.

[0022] As a further technical solution of the first aspect of the present invention, the mobile terminal and cloud module includes a mobile terminal application configured to receive and present the root cause in the form of an alert or a visual report.

[0023] A second aspect of the present invention provides an intelligent control method for a laboratory, comprising the following steps:

[0024] S1: Use the equipment monitoring module to monitor the operating parameters of the experimental equipment, and generate an equipment abnormality event when the operating parameters are abnormal;

[0025] S2: Use the sample lifecycle management module to manage the status and storage environment of experimental samples, and generate sample environment abnormality events when the storage environment is abnormal;

[0026] S3: Use the data acquisition and analysis module to collect and analyze experimental data, and generate a data anomaly event when the experimental data is abnormal;

[0027] S4: Use the process management module to monitor the execution of the experimental process and generate a process deviation event when a deviation occurs in the execution;

[0028] S5: Utilize the multimodal abnormal event causal correlation analysis module to receive the equipment abnormal events, sample environment abnormal events, data abnormal events, and process deviation events, and perform multimodal abnormal event causal correlation analysis on the events to determine the root cause of the target event.

[0029] S6: Present the underlying reasons to the user using mobile and cloud modules.

[0030] In summary, this application includes at least one of the following beneficial technical effects:

[0031] 1. This invention integrates multimodal heterogeneous data from equipment monitoring, sample management and process control through a data acquisition and analysis module, and constructs a domain knowledge graph using a multimodal abnormal event causal correlation analysis module. This module identifies root causes based on a precise scoring formula, and can deeply analyze the causal chain between abnormal events, rather than just staying at the surface phenomenon. This significantly improves the efficiency and accuracy of abnormal event root cause localization and reduces the time and cost of manual investigation.

[0032] 2. This invention uses a sample lifecycle management module to uniquely identify, monitor the environment, and record the flow of samples, ensuring the traceability and integrity of all sample data. The data acquisition and analysis module standardizes multi-source data to ensure data consistency. In addition, the mobile terminal and cloud module use encryption protocols during data transmission and encrypt the stored data. Combined with multi-factor authentication and access control mechanisms, the confidentiality, integrity, and availability of experimental data are fully guaranteed.

[0033] 3. This invention achieves standardized and automated management of experimental processes through the workflow definition, task scheduling, and resource allocation functions provided by the process management module. The mobile and cloud modules provide a convenient user interface, supporting task execution, data entry, and real-time status monitoring. Even in unstable network environments, offline operation and data synchronization can be performed. This not only reduces manual intervention and operational errors but also improves the execution efficiency of experimental tasks and the overall operational efficiency of the laboratory. Attached Figure Description

[0034] Figure 1 This is a system architecture diagram of the present invention;

[0035] Figure 2 This is a flowchart of the method of the present invention.

[0036] Explanation of reference numerals in the attached diagram: 100, Equipment monitoring module; 200, Sample lifecycle management module; 300, Data acquisition and analysis module; 400, Process management module; 500, Multimodal abnormal event causal correlation analysis module; 600, Mobile terminal and cloud module. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] See appendix Figure 1-2 This invention provides an intelligent laboratory control system, comprising: an equipment monitoring module 100, a sample lifecycle management module 200, a data acquisition and analysis module 300, a process management module 400, a multimodal abnormal event causal correlation analysis module 500, and a mobile terminal and cloud module 600.

[0039] During system operation, the process management module 400 schedules equipment and sample resources according to the experimental tasks. During the experiment, the equipment monitoring module 100 collects equipment operation data in real time, the sample lifecycle management module 200 tracks sample information and storage environment, and the data acquisition and analysis module 300 collects and processes the above multi-source data in a unified manner.

[0040] Simultaneously, the equipment monitoring module 100, the sample lifecycle management module 200, and the process management module 400 perform anomaly detection on the equipment, samples, and processes, respectively. Detected anomalies are sent to the multimodal anomaly causal correlation analysis module 500. This module 500 integrates multi-source anomalies and performs causal correlation analysis based on a domain knowledge graph to identify the root causes of the anomalies.

[0041] The root causes and related anomaly information obtained from the analysis were used to generate early warnings, which were then communicated to relevant personnel via mobile devices and the cloud module 600. After the experimental task was completed, all process data, analysis results, and anomaly event records were archived as part of the system knowledge base for subsequent system optimization.

[0042] This invention provides an intelligent control method for a laboratory, comprising the following steps:

[0043] S1: Experiment Task Initiation and Resource Preparation. Researchers initiate experimental tasks via mobile and cloud modules 600. The process management module 400 customizes the experimental process and coordinates with the equipment monitoring module 100 and the sample lifecycle management module 200 to prepare the necessary equipment and sample resources.

[0044] S2: Real-time acquisition of multi-source data. During the experiment, the equipment monitoring module 100, the sample lifecycle management module 200, and the data acquisition and analysis module 300 acquire real-time data on equipment operation, sample transfer, storage environment, and other relevant experimental data.

[0045] S3: Parallel Anomaly Detection and Event Aggregation. The system detects anomalies in equipment, samples, processes, and data in parallel. All detected standardized anomaly events are sent in real time to the multimodal anomaly event causal correlation analysis module 500.

[0046] S4: Multimodal Anomaly Event Causal Correlation Analysis. The Multimodal Anomaly Event Causal Correlation Analysis module 500 integrates the received anomaly event information, uses association rules and time series analysis algorithms to infer causal relationships and identify root causes.

[0047] S5: Intelligent Early Warning and Closed-Loop Management. Based on anomaly detection results and causal analysis reports, the system sends intelligent early warnings to personnel via mobile and cloud modules 600. Simultaneously, the data acquisition and analysis module 300 provides data analysis reports to support decision-making. After the experimental task is completed, all experimental process data, analysis results, and anomaly event records are archived, achieving closed-loop task management.

[0048] The equipment monitoring module 100 includes an equipment status acquisition unit and a centralized monitoring and anomaly detection unit.

[0049] The equipment status acquisition unit is responsible for establishing data connections with various experimental devices within the laboratory and acquiring their operating parameters in real time. This unit acquires data through multiple interfaces, such as direct connection to temperature, pressure, and vibration sensors installed on the equipment, or through communication interfaces with the equipment's built-in programmable logic controller (PLC) or distributed control system. For older equipment lacking standard interfaces, data can be digitized by adding external data acquisition cards or gateways. The parameters acquired by this unit form a multi-dimensional state vector, describing the instantaneous operating status of the equipment.

[0050] The centralized monitoring and anomaly detection unit receives and processes data from the equipment status acquisition unit. This unit first parses and standardizes the received multi-source heterogeneous data, and then performs real-time anomaly detection on key operating parameters. The specific anomaly detection steps are as follows:

[0051] Equipment status data modeling. For any given device... At any moment The operating state is modeled as a multidimensional feature vector. :

[0052] ;

[0053] in, Indicates device The Each operating parameter at time The real-time collected values.

[0054] Dynamic baseline calculation. To accommodate normal parameter fluctuations under different operating conditions, this unit employs a statistical method based on a sliding time window to calculate the dynamic baseline. For any parameter... Its time window dynamic mean within and standard deviation The calculation is as follows:

[0055] ;

[0056] ;

[0057] in, The preset sliding time window size.

[0058] Anomaly detection and event generation. When a certain parameter... When the deviation of the current value from its dynamic mean exceeds a preset threshold, the unit is determined to be abnormal. The specific determination criteria are as follows:

[0059] ;

[0060] in, This is a preset sensitivity coefficient. Once this condition is met, the centralized monitoring and anomaly detection unit generates a device anomaly event. After the event is standardized, it is sent to the multimodal abnormal event causal correlation analysis module 500 for subsequent correlation analysis.

[0061] The working principle of the equipment monitoring module 100 is to realize the complete process from equipment data acquisition to abnormal event generation.

[0062] The process begins with the device status acquisition unit, which establishes a connection with the target experimental device through a preset communication protocol. The acquired raw data packets, along with the acquisition timestamp, are formatted into a unified data structure and sent to the centralized monitoring and anomaly detection unit.

[0063] After receiving the data, the centralized monitoring and anomaly detection unit performs real-time anomaly detection on key operating parameters. This detection is based on a dual verification mechanism. On one hand, the unit compares the real-time value of the parameter with a pre-set static physical threshold based on equipment safety specifications; for example, the internal temperature of the reactor must not exceed the maximum withstand temperature of its materials. On the other hand, the unit compares the parameter value with its dynamically calculated statistical baseline, which reflects the normal fluctuation range of the equipment under current operating conditions.

[0064] When the real-time value of any parameter violates the static physical threshold, or deviates from the dynamic baseline beyond the preset statistical range, the centralized monitoring and anomaly detection unit determines that an anomaly has occurred. At this time, the unit generates a structured equipment anomaly event. The data structure of this event includes a unique event identifier, an event timestamp, the ID of the device from which the anomaly originated, the name of the anomaly parameter, the real-time parameter value, the threshold that was triggered, and the specific value of the threshold.

[0065] Ultimately, this device anomaly event contains complete context information. The data is pushed to the multimodal anomaly event causal correlation analysis module 500 as one of the data inputs for causal inference. This workflow ensures rapid detection and accurate characterization of equipment anomalies.

[0066] The sample lifecycle management module 200 includes a sample identification unit, a sample transfer record unit 220, and a sample storage management unit.

[0067] The sample identification unit is responsible for assigning a unique electronic identifier to each experimental sample and establishing a link between this identifier and the sample's metadata. Specifically, this unit can use RFID tags, QR codes, or barcodes as the carriers of the electronic identifier. When a sample is stored, a reader reads the identifier information and binds this information to the basic sample information entered into the system, thereby completing the digital archiving of the sample.

[0068] The sample flow record unit is responsible for tracking all physical location changes and state transitions of samples throughout their actual lifecycle. Fixed or handheld reading devices are deployed at key points in the laboratory, such as storage entrances, workbenches, and waste disposal areas. When a sample passes through these points, its electronic tag is read, and the system records the operator, time, location, and specific operation type. These records form a complete traceability chain and are stored in the system database.

[0069] The sample storage management unit 230 is responsible for monitoring the stability of the sample storage environment. This unit is connected to an environmental sensor installed inside the sample storage device. The unit continuously acquires real-time data of environmental parameters and compares it with the safe storage range set for the sample. When any environmental parameter is detected to exceed the preset range, the unit generates a sample environment anomaly event. The event is then sent to the multimodal abnormal event causal correlation analysis module 500.

[0070] The working principle of the Sample Lifecycle Management Module 200 is based on the digital modeling and standardized management of each sample's state. This module abstracts the sample's lifecycle into a finite state machine model.

[0071] This finite state machine model includes a predefined set of states, such as inventory, in use, awaiting inspection, archived, and discarded. The model also includes a set of operational events that trigger state transitions, such as requisition, return, weighing, and disposal. The state transitions of the sample follow a predefined state transition function. :

[0072] ;

[0073] in, Indicates the current state of the sample. This represents an operational event that acts on the sample. This indicates the new state after the transfer.

[0074] Specifically, when the reading device of the sample transfer record unit captures the unique identifier provided by the sample identification unit at a specific operation point, and combines it with the corresponding operation event input by the operator... When this happens, the system calls the state transition function. The function, based on preset rules, determines the current status of the sample in the system database. Update to new status For example, a sample that is in stock will change its status to "in use" after a requisition operation event is performed.

[0075] All state transition operations, along with contextual information such as the operator, time, and location, are recorded, thus constructing a complete and traceable full lifecycle history record for each sample.

[0076] Meanwhile, the sample storage management unit continuously monitors the environment in which the sample is located. When environmental parameters deviate from the preset range, this unit generates an independent sample environment anomaly event. The event is then sent to the multimodal abnormal event causal association analysis module 500. The generation of this event does not directly change the state of the sample in the finite state machine, but it serves as important contextual information for subsequent causal association analysis.

[0077] The data acquisition and analysis module 300 includes a data acquisition unit and a data analysis unit.

[0078] The data acquisition unit is responsible for collecting various types of data generated during the experiment from other modules of the system and external experimental instruments. This unit acquires data in multiple ways, such as through application programming interfaces (APIs) to obtain equipment operating parameters from the equipment monitoring module 100 and sample transfer information from the sample lifecycle management module 200. For experimental instruments not directly integrated with this system, this unit can obtain experimental results data by parsing their output data files or through standard data interfaces. All acquired data is associated with a corresponding experimental task ID, sample ID, and timestamp.

[0079] The data analysis unit receives the data integrated by the data acquisition unit and processes and analyzes it. Internally, this unit includes a data preprocessing engine, a statistical analysis engine, and an anomaly detection engine. The data preprocessing engine is responsible for data cleaning, format conversion, and normalization. The statistical analysis engine performs descriptive statistics, correlation analysis, regression analysis, and other calculations. The anomaly detection engine is responsible for identifying outliers or patterns in the data that do not conform to a preset model, either in real-time or in batches. When an anomaly is detected in the experimental data, for example, if a key indicator of a batch of products significantly deviates from the historical mean, this engine generates a data anomaly event. It is then sent to the multimodal abnormal event causal correlation analysis module 500.

[0080] The working principle of the data acquisition and analysis module 300 is to uniformly aggregate, process, and deeply mine multi-source heterogeneous data.

[0081] The workflow begins with the data acquisition unit, which obtains data streams from data sources such as the equipment monitoring module 100, the sample lifecycle management module 200, and external experimental instruments through pre-defined interfaces and protocols. The acquired data is appended with metadata such as experimental task ID, sample ID, and timestamps to ensure contextual integrity and traceability. These data streams with metadata are then integrated and transmitted to the data analysis unit.

[0082] After receiving the integrated data, the data analysis unit's internal data preprocessing engine first processes the data. This processing includes imputing missing values, uniformly converting data units from different sources, and normalizing numerical data.

[0083] The preprocessed data is then fed into the statistical analysis engine and the anomaly detection engine. The statistical analysis engine performs pre-defined analysis tasks, such as calculating the correlation coefficient between experimental parameters and product yield, or performing time series trend analysis on specific parameters. The anomaly detection engine monitors key performance indicators (KPIs) in real time. For example, for a key performance indicator... The engine will set its current value Compare the data with historical statistics for this type of experiment. Data is considered outlier when the following conditions are met:

[0084] ;

[0085] in and These are the historical mean and standard deviation of the key performance indicator. It is the preset sensitivity coefficient.

[0086] Once a data anomaly is detected, the anomaly detection engine generates a structured data anomaly event containing information such as the anomaly indicator name, current value, and historical statistics. The event is then sent to the multimodal abnormal event causal correlation analysis module 500.

[0087] The process management module 400 includes a process customization unit and a process execution monitoring unit.

[0088] The workflow customization unit is responsible for the digital modeling and configuration of experimental workflows. Specifically, this unit abstracts a complete experimental workflow as a directed acyclic graph (DAG). Each node in the graph represents a specific experimental step, such as sample preparation, heating reaction, or data acquisition. The directed edges in the graph define the execution order and dependencies between experimental steps. Each step node records information such as the required equipment, samples, operating parameters, and planned execution time for that step. Researchers can create new experimental workflows or call existing standard workflows through the graphical interface or templates provided by this unit.

[0089] The process execution monitoring unit is responsible for comparing the actual progress of the process with the preset process during experiment execution and identifying deviations. This unit receives and records the actual start and end times of each experimental step in real time. By comparing the actual execution time with the planned time set by the process customization unit, when a significant deviation from the plan is detected, or the execution order of steps violates the dependencies defined by the directed acyclic graph, the unit generates a process deviation event. The event was then sent to the multimodal anomaly event causal correlation analysis module 500.

[0090] The working principle of the process management module 400 is to transform the digital experimental process model into an executable and monitorable actual operation.

[0091] At the start of the experiment, the process management module 400 uses the directed acyclic graph defined in the process customization unit as the execution blueprint. The process execution monitoring unit tracks the execution status of each experimental step based on this blueprint. This unit receives manual confirmation from the experimenter regarding the start and end of steps via the mobile terminal and cloud module 600, or automatically obtains the execution status of steps through linkage with other modules. For example, when the device monitoring module 100 reports that a certain associated device has started running, this can be considered the start of the corresponding step.

[0092] The process execution monitoring unit mainly performs two types of deviation detection:

[0093] The first method is time deviation detection. For any step in the experimental procedure... This unit records its actual end time. and the planned end time A comparison is made. A time deviation is determined when the following conditions are met:

[0094] ;

[0095] in, It is a preset time tolerance threshold.

[0096] The second type is sequence deviation detection. This unit checks the step dependencies defined in the directed acyclic graph. If the system receives the start signal of the subsequent step before the preceding step has been completed, it is determined to be a sequence deviation.

[0097] Once any of the above types of deviations is detected, the process execution monitoring unit generates a structured process deviation event. The event contains information such as the ID of the deviation step, the deviation type, and the specific value of the deviation. This event is then sent to the multimodal anomaly event causal correlation analysis module 500 for subsequent causal analysis.

[0098] The multimodal abnormal event causal correlation analysis module 500 includes an abnormal event standardization unit 510 and a causal correlation inference engine 520.

[0099] The abnormal event standardization unit is responsible for receiving various heterogeneous abnormal events from the equipment monitoring module 100, the sample lifecycle management module 200, the data acquisition and analysis module 300, and the process management module 400, such as equipment abnormal events. Abnormal events in the sample environment Data anomaly events Process deviation events This unit transforms event information from diverse sources and with varying structures into a unified, standardized data structure.

[0100] A standardized exception event It can be defined as a tuple containing the following fields: unique event identifier, event timestamp, event type, event source identifier, and a set of key-value pairs containing specific parameters. Through this standardization process, abnormal events of different modalities can be compared and analyzed on the same dimension.

[0101] The causal association inference engine receives a standardized stream of anomalous events output by the anomalous event standardization unit. Internally, the engine contains a knowledge base for storing historical association rules and causal knowledge, as well as a computational core for executing inference algorithms. Based on the received event stream and information from the knowledge base, the engine performs association analysis and causal chain inference; the specific algorithm implementation will be described later.

[0102] The causal correlation inference engine works by inferring causal chains between events from a synchronously received multimodal anomalous event stream using quantitative analysis and graph theory methods. This inference process specifically includes the following steps:

[0103] This engine quantifies the strength of causal relationships between events based on association rules. It first mines association rules between different types of anomalous events from historical data. For any two standardized anomalous events... and The strength of the causal relationship between these events, which occur sequentially in time, is quantified using lift. The formula for calculating lift is:

[0104] ;

[0105] in, and Representing events respectively and The probability of occurring independently, This represents the joint probability of two events occurring consecutively within a preset time window. These probabilities are obtained through statistical analysis of historical anomalous event datasets. A lift value greater than 1 indicates that the event... The occurrence of the event The occurrence of [a certain value] has a promoting effect, and the larger the value, the stronger the correlation.

[0106] Construct a weighted directed graph of causal relationships. Based on the calculated lift values, the engine builds a global weighted directed graph of causal relationships. In this graph, the set of nodes Represents all predefined standardized exception event types. Edge set It represents a possible causal relationship between events. If and only if... When the value exceeds a preset threshold, there exists a path from node [the node in the graph]. Pointing to node Directed edges. Weight set. The weight of each edge is equal to its corresponding lift value. This diagram constitutes the system's causal knowledge base.

[0107] The engine uses graph search to infer root causes. When one or more new anomalous events are reported, the engine identifies the most critical or ultimately occurring event as the target event. Subsequently, within a preset backtracking time window, the engine retrieves all events that occurred during that period. Previous abnormal events. Finally, the engine uses a causal weighted directed graph. In the middle, starting with these preceding events, With the target event as the endpoint, search for all possible causal paths. The total score for each path is calculated by multiplying the weights of all edges along that path; the path with the highest score is identified as the one leading to the target event. The most likely causal chain that occurred. The starting node of this path is then determined as the root cause. For graph search algorithms, those skilled in the art can use Dijkstra's algorithm or its variants, the implementation of which is well-known in the field and will not be described in detail here.

[0108] The mobile and cloud module 600 includes a mobile application, a cloud storage unit, and a cloud computing unit.

[0109] The mobile application serves as the primary interface for researchers to interact with the system. This application runs on mobile devices such as smartphones or tablets, and its functions include, but are not limited to, real-time data display, receiving alerts for abnormal events, and confirming operations. For example, researchers can use the application to view the real-time operating status of the equipment, receive root cause analysis results pushed by the multimodal abnormal event causal correlation analysis module 500, or scan the QR code on a sample to confirm its use.

[0110] The cloud storage unit serves as the central data warehouse for this system. This unit employs a hybrid storage architecture; for example, it uses a relational database to store structured configuration information, while simultaneously using a time-series database to store frequently collected device sensor data and various event logs. This unit provides persistent data storage services to all modules of the system.

[0111] The cloud computing unit provides computing power support for the core analysis and computation tasks of this system. Specifically, this unit deploys and runs computationally intensive components such as the device monitoring module 100, the data acquisition and analysis module 300, and the multimodal abnormal event causal correlation analysis module 500. This unit provides services to the outside world through application programming interfaces (APIs), and mobile applications use these APIs to obtain data and submit operation commands.

[0112] The working principle of the Mobile and Cloud Module 600 lies in building a distributed architecture that connects front-end interaction with back-end computing and analysis.

[0113] Researchers interact with the system through a mobile application running on their mobile devices. When a researcher performs an operation, such as scanning a sample QR code to confirm receipt or manually confirming the completion of an experimental step, the mobile application encapsulates the operation instruction and related data into an API request via a secure communication protocol and sends it to the service interface deployed on the cloud computing unit. Upon receiving the request, the cloud computing unit distributes it to the corresponding business module for processing; for example, it hands over the sample operation information to the sample lifecycle management module 200 to update the sample status.

[0114] Conversely, when the multimodal anomaly causal correlation analysis module 500 within the cloud computing unit infers the root cause analysis results, these results are formatted and proactively sent to the corresponding mobile application via push notification service. Upon receiving the push notification, the mobile application presents the causal correlation analysis results to the experimenters in the form of an alert pop-up or a visual report.

[0115] During this process, all operation records submitted through the mobile application, abnormal events generated by each module, and analysis results are written to the cloud storage unit in real time or near real time for persistent storage to ensure data integrity and traceability.

[0116] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Identical components are represented by the same reference numerals. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be covered within the scope of protection of this application.

Claims

1. An intelligent control system for a laboratory, characterized in that, include: The equipment monitoring module is configured to monitor the operating parameters of the experimental equipment and generate equipment abnormality events when the operating parameters are abnormal. The sample lifecycle management module is configured to manage the status and storage environment of experimental samples, and generate sample environment abnormality events when the storage environment is abnormal. The data acquisition and analysis module is configured to acquire and analyze experimental data, and generate a data anomaly event when the experimental data is abnormal. The process management module is configured to monitor the execution of the experimental process and generate a process deviation event when a deviation occurs in the execution. The multimodal abnormal event causal correlation analysis module is connected to the equipment monitoring module, sample life cycle management module, data acquisition and analysis module and process management module respectively. It is configured to receive the equipment abnormal events, sample environment abnormal events, data abnormal events and process deviation events, and perform multimodal abnormal event causal correlation analysis on the received events to determine the root cause of the target event. The mobile and cloud modules are configured to interact with the user and present the root cause determined by the multimodal abnormal event causal correlation analysis module.

2. The system according to claim 1, characterized in that, The multimodal abnormal event causal correlation analysis module includes: The abnormal event standardization unit is configured to convert the equipment abnormal events, sample environment abnormal events, data abnormal events, and process deviation events into a unified standardized data structure.

3. The system according to claim 2, characterized in that, The multimodal abnormal event causal correlation analysis module also includes: The causal association inference engine is configured to construct a causal association weighted directed graph based on the standardized data structure, and infer the root cause by performing a graph search on the graph.

4. The system according to claim 3, characterized in that, The weights of the edges in the causal association weighted directed graph are determined by the causal association inference engine based on the lift value of the association between different events.

5. The system according to claim 1, characterized in that, The device monitoring module is configured to generate an abnormal device event when the monitored device operating parameters deviate from a preset physical threshold or statistical baseline.

6. The system according to claim 1, characterized in that, The sample lifecycle management module is configured to generate an abnormal sample environment event when the monitored sample storage environment parameters exceed a preset range.

7. The system according to claim 1, characterized in that, The data acquisition and analysis module is configured to generate the data anomaly event when the deviation between the analyzed experimental data and its historical statistical data exceeds a preset threshold.

8. The system according to claim 1, characterized in that, The process management module is configured to generate the process deviation event when the actual execution progress of the experiment deviates from the preset plan or the execution order violates the preset dependency relationship.

9. The system according to claim 1, characterized in that, The mobile and cloud module includes a mobile application configured to receive and present the root cause in the form of an alert or visual report.

10. A laboratory intelligent control method, comprising a laboratory intelligent control system according to any one of claims 1-9, characterized in that, Includes the following steps: S1: Use the equipment monitoring module to monitor the operating parameters of the experimental equipment, and generate an equipment abnormality event when the operating parameters are abnormal; S2: Use the sample lifecycle management module to manage the status and storage environment of experimental samples, and generate sample environment abnormality events when the storage environment is abnormal; S3: Use the data acquisition and analysis module to collect and analyze experimental data, and generate a data anomaly event when the experimental data is abnormal; S4: Use the process management module to monitor the execution of the experimental process and generate a process deviation event when a deviation occurs in the execution; S5: Utilize the multimodal abnormal event causal correlation analysis module to receive the equipment abnormal events, sample environment abnormal events, data abnormal events, and process deviation events, and perform multimodal abnormal event causal correlation analysis on the events to determine the root cause of the target event. S6: Present the underlying reasons to the user using mobile and cloud modules.

Images