A laboratory intelligent comprehensive management method and system

Abstract

The present invention relates to the technical field of data processing, and discloses a laboratory intelligent comprehensive management method and system for improving the efficiency of laboratory intelligent comprehensive management. The method includes: performing experimental phase identification and workflow mapping on the laboratory ecological state data stream to obtain a dynamic laboratory activity hologram; performing experimental key path analysis on the dynamic laboratory activity hologram to obtain a multidimensional health index spectrum of laboratory operation; performing intelligent comparison and abnormal pattern recognition on the laboratory ecological state data collected in real time and the multidimensional health index spectrum to obtain a laboratory sub-health event feature chain; performing source tracing analysis and development trend analysis on the laboratory sub-health event feature chain to obtain a laboratory risk evolution path map; generating a laboratory safety early warning signal based on the laboratory risk evolution path map, and triggering a corresponding emergency response strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a laboratory intelligent comprehensive management method and system. Background Art

[0002] Existing laboratory management methods mainly rely on manual monitoring and regular inspections, which makes it difficult to grasp the operating status of the laboratory in real time and comprehensively. Traditional methods usually use fixed checklists and preset safety procedures, which cannot adapt to complex and changing experimental environments. At the same time, data collection is often limited to a single dimension, such as temperature and humidity monitoring or gas concentration detection, and lacks a comprehensive perception of the overall ecological status of the laboratory. In addition, existing anomaly detection and risk assessment methods are mostly based on static thresholds, which make it difficult to capture potential sub-health conditions and dynamic risk evolution processes.

[0003] These traditional methods have several major shortcomings: first, they lack the ability to comprehensively analyze multimodal data and cannot fully reflect the true status of the laboratory; second, they cannot achieve dynamic identification and intelligent early warning of the experimental process, resulting in the failure to timely discover potential risks; third, they lack the ability to quantitatively evaluate the health status of the laboratory and predict trends, making it difficult to provide forward-looking management decision support; finally, emergency response strategies are often preset and lack the ability to flexibly adjust to specific situations. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a laboratory intelligent integrated management method and system for improving the efficiency of laboratory intelligent integrated management.

[0005] The present invention provides a laboratory intelligent comprehensive management method, comprising: performing multimodal perception and collection on the acoustic characteristics of laboratory instrument equipment operation, chemical environment odor fingerprints and biorhythm data of researchers in the laboratory to obtain a laboratory ecological status data stream; performing experimental stage identification and workflow mapping on the laboratory ecological status data stream to obtain a dynamic laboratory activity hologram; performing experimental key path analysis on the dynamic laboratory activity hologram to obtain a multidimensional health index spectrum of laboratory operation; performing intelligent comparison and abnormal pattern recognition on the laboratory ecological status data collected in real time and the multidimensional health index spectrum to obtain a laboratory sub-health event feature chain; performing source tracing analysis and development trend analysis on the laboratory sub-health event feature chain to obtain a laboratory risk evolution path map; generating a laboratory safety early warning signal based on the laboratory risk evolution path map, and triggering a corresponding emergency response strategy.

[0006] In the present invention, the step of performing multimodal perception and collection on the acoustic characteristics of the laboratory instrument and equipment operation, the chemical environment odor fingerprint and the biorhythm data of the researchers in the laboratory to obtain the laboratory ecological status data stream includes: performing spectrum analysis on the acoustic signals of the laboratory instrument and equipment to obtain the acoustic characteristic spectrum diagram of the equipment operation, and performing peak extraction on the acoustic characteristic spectrum diagram of the equipment operation to obtain the characteristic point set of the equipment operation status; performing gas sampling on the chemical environment of the laboratory to obtain multi-component gas concentration data, and performing principal component analysis on the multi-component gas concentration data to obtain the chemical environment odor fingerprint; collecting the physiological data of the researchers in the laboratory to obtain the heart rate variability and skin conductivity time series, and performing wavelet transformation on the heart rate variability and skin conductivity time series to obtain the biorhythm characteristic spectrum of the researchers; performing timestamp alignment on the equipment operation status characteristic point set, the chemical environment odor fingerprint and the biorhythm characteristic spectrum of the researchers to obtain a synchronized multimodal data sequence, and performing data fusion on the synchronized multimodal data sequence to obtain the laboratory ecological status characteristic vector; performing time series reconstruction on the laboratory ecological status characteristic vector to obtain the laboratory ecological status data stream.

[0007] In the present invention, the step of performing experimental phase identification and workflow mapping on the laboratory ecological status data stream to obtain a dynamic laboratory activity hologram includes: performing sliding window segmentation on the laboratory ecological status data stream to obtain experimental activity time segments of fixed duration, and extracting statistical features and frequency domain features from the experimental activity time segments to obtain a multidimensional experimental activity feature vector; performing cluster analysis on the multidimensional experimental activity feature vector by a K-means algorithm to obtain an experimental activity type cluster, and performing semantic matching on the experimental activity type cluster according to a preset experimental type dictionary to obtain an experimental phase identifier; constructing an experimental activity transfer matrix based on the experimental phase identifier to obtain an experimental workflow sequence, and converting the experimental workflow sequence into a directed graph structure to obtain an experimental workflow diagram; calculating the degree centrality and betweenness centrality of each node in the experimental workflow diagram to obtain a set of key experimental nodes, and assigning weights to the edges of the experimental workflow diagram according to the centrality index of the key experimental node set to obtain a weighted experimental workflow network; mapping the weighted experimental workflow network to the laboratory physical space and time dimensions to obtain the dynamic laboratory activity hologram.

[0008] In the present invention, the step of performing experimental critical path analysis on the dynamic laboratory activity hologram to obtain a multidimensional health index spectrum of laboratory operation includes: performing graph structure analysis on the dynamic laboratory activity hologram to identify experimental key nodes and paths to obtain an experimental critical path set, and arranging the experimental critical path set in time series to obtain an experimental critical stage sequence; extracting time, space and resource occupancy characteristics for each stage in the experimental critical stage sequence to obtain a multidimensional experimental feature matrix, and performing principal component analysis on the multidimensional experimental feature matrix to obtain an experimental critical factor set; constructing a laboratory operation status evaluation index system based on the experimental critical factor set to obtain an initial health index scoring table, and adjusting the weight of the initial health index scoring table to obtain a weighted health index scoring standard; retrospectively evaluating historical experimental data based on the weighted health index scoring standard to obtain a laboratory historical health status data set, and The laboratory's historical health status data set is subjected to time series analysis to obtain a health status change trend graph; periodic patterns and abnormal fluctuations are identified based on the health status change trend graph to obtain laboratory health status periodic characteristics, and future health status is predicted through the laboratory health status periodic characteristics to obtain a health status prediction curve; the weighted health index scoring standard is digitally encoded to obtain a health index scoring vector, and the health status change trend graph is converted into time series data to obtain a health trend sequence; the health status prediction curve is discretized and sampled to obtain a predicted state point set, and the health index scoring vector, the health trend sequence and the predicted state point set are aligned according to the time dimension to construct a multidimensional time series matrix; spatiotemporal feature extraction is performed on the multidimensional time series matrix to obtain a health status feature tensor, and dimensionality reduction processing is performed on the health status feature tensor to obtain a multidimensional health index spectrum of the laboratory operation.

[0009] In the present invention, the steps of intelligently comparing and identifying abnormal patterns of the laboratory ecological status data collected in real time with the multidimensional health index spectrum to obtain the characteristic chain of laboratory sub-health events include: performing data preprocessing and standardization on the laboratory ecological status data collected in real time to obtain a normalized real-time status feature vector, and performing feature matching on the normalized real-time status feature vector and the multidimensional health index spectrum to obtain a feature difference matrix; performing threshold analysis and anomaly detection on the feature difference matrix to identify indicators that deviate from the normal range to obtain a preliminary abnormal indicator set, and performing context correlation analysis on the preliminary abnormal indicator set through laboratory historical data to obtain an associated abnormal indicator group; constructing an abnormal event candidate set based on the associated abnormal indicator group, and performing multi-dimensional evaluation and sorting on the abnormal event candidate set to obtain an abnormal event priority list; performing time series correlation analysis on the events in the abnormal event priority list to identify the causal relationship and evolution law between events to obtain an abnormal event association The invention relates to a network, and extracts key abnormal event paths based on the abnormal event association network to obtain a core abnormal event chain; semantically annotates and extracts features of the core abnormal event chain to generate an abnormal event descriptor, and performs pattern matching on the abnormal event descriptor and a predefined sub-health pattern library to obtain a sub-health event type identifier; matches each event node in the core abnormal event chain with the corresponding abnormal event descriptor one by one to obtain an event description correspondence table, and performs association analysis on each event description pair with the sub-health event type identifier to obtain a triple set; arranges the triple set in chronological order to construct a time series association graph, and performs causal reasoning on the time series association graph to obtain a causal link structure; performs data fusion on the causal link structure and the original event time series information to construct a multi-level event network, and performs visualization processing and data structure optimization on the multi-level event network to obtain the laboratory sub-health event feature chain.

[0010] In the present invention, the step of conducting source tracing analysis and development trend analysis on the characteristic chain of laboratory sub-health events to obtain a laboratory risk evolution path map includes: performing time series decomposition on the characteristic chain of laboratory sub-health events, extracting the time, space and type characteristics of the event occurrence, obtaining a sub-health event characteristic matrix, and constructing an event association graph based on the sub-health event characteristic matrix to obtain a sub-health event network structure; performing topological analysis on the sub-health event network structure to identify key nodes and core paths to obtain a sub-health event propagation chain, and tracing the sub-health event propagation chain through laboratory historical data to obtain a sub-health event source set; performing causal relationship analysis based on the sub-health event source set to construct an event causal network to obtain A causal chain of sub-health events is obtained, and risk assessment is performed on the causal chain of sub-health events to obtain a risk level distribution map; trend extrapolation is performed on the sub-health event feature matrix through time series prediction technology to obtain a sub-health event development prediction curve, and risk weighting is performed on the prediction results through the risk level distribution map to obtain a risk evolution prediction map; a multidimensional risk evolution model is constructed based on the sub-health event causal chain and the risk evolution prediction map to obtain an initial risk evolution path set, and a multi-scenario simulation is performed on the initial risk evolution path set to obtain a risk evolution scenario tree; the sub-health event network structure, the sub-health event causal chain, the risk evolution prediction map and the risk evolution scenario tree are visualized and integrated to obtain the laboratory risk evolution path map.

[0011] In the present invention, the laboratory safety early warning signal is generated based on the laboratory risk evolution path map, and the corresponding emergency response strategy steps are triggered, including: performing risk level assessment on the laboratory risk evolution path map to obtain a risk level matrix, and setting a multi-level early warning threshold based on the risk level matrix to obtain a laboratory safety early warning threshold table; real-time monitoring of key nodes in the laboratory risk evolution path map, extracting current risk state characteristics, obtaining a real-time risk state vector, and comparing the real-time risk state vector with the laboratory safety early warning threshold table to obtain an early warning trigger signal; generating early warning information including risk type, risk level and impact range according to the early warning trigger signal to obtain a laboratory safety early warning signal. number, and classify the laboratory safety warning signal to obtain a hierarchical warning information set; based on the hierarchical warning information set, query the preset emergency response rule library, extract the corresponding emergency disposal plan, obtain the initial emergency response strategy set, and conduct a feasibility assessment on the initial emergency response strategy set according to the current resource status of the laboratory to obtain a list of executable emergency plans; perform multi-objective optimization on the plans in the executable emergency plan list to obtain a preferred emergency response plan, and convert the preferred emergency response plan into an operation instruction sequence to obtain an emergency response execution list; merge the laboratory safety warning signal and the emergency response execution list into a comprehensive emergency instruction package, and trigger the corresponding emergency response strategy through the comprehensive emergency instruction package.

[0012] The present invention also provides a laboratory intelligent integrated management system, comprising:

[0013] The acquisition module is used to perform multimodal perception and acquisition of the laboratory instrument and equipment operating acoustic characteristics, chemical environment odor fingerprints, and researchers' biorhythm data to obtain the laboratory ecological status data stream;

[0014] A mapping module, used for performing experimental phase identification and workflow mapping on the laboratory ecological status data stream to obtain a dynamic laboratory activity hologram;

[0015] An analysis module, used to perform experimental critical path analysis on the dynamic laboratory activity hologram to obtain a multi-dimensional health index spectrum of laboratory operation;

[0016] An identification module is used to intelligently compare the real-time collected laboratory ecological status data with the multi-dimensional health index spectrum and identify abnormal patterns to obtain a characteristic chain of laboratory sub-health events;

[0017] A traceability module is used to perform traceability analysis and development trend analysis on the characteristic chain of laboratory sub-health events to obtain a laboratory risk evolution path map;

[0018] The early warning module is used to generate laboratory safety early warning signals based on the laboratory risk evolution path map and trigger the corresponding emergency response strategy. In the technical solution provided by the present invention, the laboratory intelligent comprehensive management method realizes comprehensive and real-time monitoring of the laboratory operation status through multimodal perception acquisition and data fusion technology, significantly improving the accuracy and comprehensiveness of data acquisition. Through the comprehensive analysis of the acoustic characteristics of laboratory instrument operation, chemical environment odor fingerprints and researchers' biorhythm data, it is possible to capture minor anomalies that are difficult to find by traditional methods, thereby greatly improving the sensitivity and predictability of laboratory safety management. The introduction of experimental stage identification and workflow mapping enables accurate tracking of experimental progress and identification of key nodes, which not only optimizes resource allocation, but also provides important support for the early detection of abnormal events. The construction of a dynamic laboratory activity hologram provides managers with an intuitive and dynamic laboratory operation status visualization tool, which effectively improves the timeliness and accuracy of decision-making. The introduction of a multidimensional health index spectrum has enabled the laboratory operation status assessment to shift from a single indicator to a multi-dimensional, comprehensive assessment, greatly enhancing the scientificity and comprehensiveness of the assessment. Through intelligent comparison of real-time data and health index spectrum and abnormal pattern recognition, potential sub-health status can be quickly identified, providing strong support for preventive maintenance and risk management. The construction and analysis of the characteristic chain of sub-health events enable an in-depth understanding of the occurrence mechanism and evolution law of abnormal events, providing a scientific basis for the formulation of targeted preventive measures. The generation of risk evolution path maps provides laboratory managers with a global view of risk development, which helps to formulate long-term risk management and control strategies. The design of multi-level early warning mechanisms and adaptive emergency response strategies has greatly improved the laboratory's ability and efficiency in responding to emergencies. Through multi-objective optimization of emergency plan selection, the best balance can be found between safety, resource utilization efficiency and experimental progress. The automatic generation and distribution of comprehensive emergency instruction packages significantly shortens the time from risk identification to response implementation, and improves the efficiency and accuracy of emergency disposal. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 A flow chart of a laboratory intelligent comprehensive management method in an embodiment of the present invention;

[0021] Figure 2 Schematic diagram of a laboratory intelligent integrated management system in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The technical solution of the present invention will be described clearly and completely below through the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0024] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , Figure 1 is a flow chart of a laboratory intelligent comprehensive management method according to an embodiment of the present invention, such as Figure 1 As shown, the following steps are included:

[0026] S101, perform multimodal sensing and collection of the laboratory's laboratory instrument and equipment operating acoustic characteristics, chemical environment odor fingerprints, and researchers' biorhythm data to obtain the laboratory ecological status data stream;

[0027] S102, performing experimental phase identification and workflow mapping on the laboratory ecological status data stream to obtain a dynamic laboratory activity hologram;

[0028] S103. Conduct experimental critical path analysis on the dynamic laboratory activity hologram to obtain a multi-dimensional health index spectrum of laboratory operation;

[0029] S104, performing intelligent comparison and abnormal pattern recognition on the laboratory ecological status data collected in real time and the multi-dimensional health index spectrum to obtain a characteristic chain of laboratory sub-health events;

[0030] S105. Conduct source tracing and development trend analysis on the characteristic chain of laboratory sub-health events to obtain a laboratory risk evolution path map;

[0031] S106. Generate laboratory safety warning signals based on the laboratory risk evolution path map and trigger corresponding emergency response strategies.

[0032] It should be noted that the laboratory intelligent comprehensive management method obtains comprehensive laboratory data through multimodal sensing and acquisition technology, including acoustic characteristics of laboratory instrument operation, chemical environment odor fingerprints and researchers' biorhythm data. After processing, these data form a laboratory ecological status data stream, providing a basis for subsequent analysis. Acoustic feature acquisition uses a high-precision microphone array to capture the sound signal when the equipment is running, and extracts features through spectrum analysis. The chemical environment odor fingerprint uses electronic nose technology and a gas sensor array to detect changes in chemical composition in the air. The researchers' biorhythm data collects physiological indicators such as heart rate and body temperature through wearable devices.

[0033] The experimental phase identification and workflow mapping of the laboratory ecological status data stream are performed to generate a dynamic laboratory activity hologram. This process first segments the data stream into time series and extracts the feature vector of each time segment. Then, similar feature vectors are classified through cluster analysis to identify different experimental phases. Based on the identified experimental phases, an experimental workflow diagram is constructed and mapped to the laboratory physical space and time dimensions to form a dynamic hologram.

[0034] The experimental critical path analysis of the dynamic laboratory activity hologram was performed to obtain the multidimensional health index spectrum of laboratory operation. First, the key nodes and paths in the hologram were identified, and then the time, space and resource occupation characteristics of each key stage were extracted to construct a multidimensional feature matrix. The principal component analysis was used to reduce the dimension and obtain the set of experimental key factors. Based on these factors, the health index scoring criteria were constructed, and the historical data were retrospectively evaluated to generate the health status change trend chart and prediction curve. Finally, this information was integrated to form a multidimensional health index spectrum.

[0035] The laboratory ecological status data collected in real time is intelligently compared with the multidimensional health index spectrum and abnormal pattern recognition is performed to obtain the characteristic chain of laboratory sub-health events. First, the real-time data is standardized and feature matched with the health index spectrum to generate a feature difference matrix. Threshold analysis and anomaly detection are used to identify indicators that deviate from the normal range, and context correlation analysis is performed in combination with historical data to construct a candidate set of abnormal events. The candidate set is evaluated and sorted in multiple dimensions to identify the causal relationship between events and extract the core abnormal event chain. Finally, the event chain is semantically annotated and feature extracted, and matched with the predefined sub-health pattern library to form a complete characteristic chain of sub-health events.

[0036] The characteristic chain of laboratory sub-health events is traced and analyzed for its development trend, and a laboratory risk evolution path map is obtained. First, the characteristic chain is decomposed into a time series, the time, space and type characteristics of the event are extracted, and an event association diagram is constructed. Key nodes and core paths are identified through topological analysis, and historical data are combined for source tracing. Then, causal relationship analysis is performed, an event causal network is constructed, and risk assessment is performed. Time series prediction technology is used to extrapolate the trend of the characteristic matrix, and a risk evolution prediction diagram is generated in combination with the risk assessment results. Finally, multi-scenario simulation is performed through scenario analysis methods, all information is integrated and visualized, and a risk evolution path map is formed.

[0037] Generate laboratory safety warning signals based on the laboratory risk evolution path map and trigger corresponding emergency response strategies. First, evaluate the risk level of the risk evolution path map and set multi-level warning thresholds. Monitor the risk status of key nodes in real time, compare it with the warning threshold, and generate warning trigger signals. Generate warning information containing risk type, level and impact range according to the warning signal, and classify it. Query the preset emergency response rule library, extract the corresponding disposal plan, and conduct a feasibility assessment based on the current resource status. Perform multi-objective optimization on the executable plan, consider the response timeliness, resource consumption and risk reduction effect, select the optimal plan and convert it into a specific operation instruction sequence. Finally, integrate the warning signal and the emergency response execution list into a comprehensive emergency instruction package, distribute it to relevant execution units, and trigger the emergency response strategy.

[0038] For example, in a laboratory, an acoustic sensor detected an abnormal peak in the operating sound spectrum of a centrifuge, while a chemical sensor detected a slightly increased concentration of acidic gases in the air, and the heart rate variability data of the researchers showed an increase in stress levels. After processing, these data were identified as a potential risk stage in the experiment. The system then conducted an in-depth analysis of this stage and found that the centrifuge may have a bearing failure and the sample may have leaked. Based on historical data and preset rules, the system assessed that this situation may cause equipment damage and personnel exposure risks, so it generated a medium-level safety warning signal. At the same time, the system automatically triggered the emergency response strategy of equipment shutdown, ventilation enhancement and personnel evacuation, effectively preventing potential accidents.

[0039] By executing the above steps, the laboratory intelligent comprehensive management method realizes comprehensive and real-time monitoring of the laboratory operation status through multimodal sensing acquisition and data fusion technology, significantly improving the accuracy and comprehensiveness of data acquisition. Through the comprehensive analysis of the acoustic characteristics of laboratory equipment operation, chemical environment odor fingerprints and researchers' biorhythm data, it is possible to capture minor anomalies that are difficult to detect with traditional methods, thereby greatly improving the sensitivity and predictability of laboratory safety management. The introduction of experimental stage identification and workflow mapping enables accurate tracking of experimental progress and identification of key nodes, which not only optimizes resource allocation, but also provides important support for the early detection of abnormal events. The construction of dynamic laboratory activity holograms provides managers with an intuitive and dynamic laboratory operation status visualization tool, which effectively improves the timeliness and accuracy of decision-making. The introduction of multidimensional health index spectrum enables the laboratory operation status assessment to shift from a single indicator to a multi-dimensional, comprehensive assessment, greatly enhancing the scientificity and comprehensiveness of the assessment. Through the intelligent comparison of real-time data and health index spectrum and abnormal pattern recognition, potential sub-health states can be quickly identified, providing strong support for preventive maintenance and risk management. The construction and analysis of the characteristic chain of sub-health events enables an in-depth understanding of the occurrence mechanism and evolution law of abnormal events, providing a scientific basis for the formulation of targeted preventive measures. The generation of risk evolution path maps provides laboratory managers with a global view of risk development, which helps to formulate long-term risk management and control strategies. The design of multi-level early warning mechanisms and adaptive emergency response strategies has greatly improved the laboratory's ability and efficiency in responding to emergencies. Through the selection of emergency plans based on multi-objective optimization, the best balance can be found between safety, resource utilization efficiency and experimental progress. The automatic generation and distribution of comprehensive emergency instruction packages significantly shortens the time from risk identification to response implementation, and improves the efficiency and accuracy of emergency disposal.

[0040] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0041] (1) Perform spectrum analysis on the acoustic signals of laboratory instruments and equipment to obtain the equipment operation acoustic feature spectrum, and perform peak extraction on the equipment operation acoustic feature spectrum to obtain the equipment operation status feature point set;

[0042] (2) Sampling the gas in the laboratory's chemical environment to obtain multi-component gas concentration data, and performing principal component analysis on the multi-component gas concentration data to obtain the chemical environment odor fingerprint;

[0043] (3) Collect the physiological data of researchers in the laboratory to obtain the time series of heart rate variability and skin conductivity, and perform wavelet transform on the time series of heart rate variability and skin conductivity to obtain the biorhythm characteristic spectrum of the researchers;

[0044] (4) Align the time stamps of the equipment operation status feature point set, the chemical environment odor fingerprint, and the researchers' biorhythm feature spectrum to obtain a synchronized multimodal data sequence, and perform data fusion on the synchronized multimodal data sequence to obtain the laboratory ecological status feature vector;

[0045] (5) Reconstruct the time series of the laboratory ecological status feature vector to obtain the laboratory ecological status data stream.

[0046] Specifically, the core of the laboratory intelligent comprehensive management method lies in the collection and fusion of multimodal data. First, the acoustic signals of laboratory instruments and equipment are spectrally analyzed. This process uses fast Fourier transform (FFT) to convert the time domain signal into the frequency domain to obtain the acoustic characteristic spectrum of the equipment operation. The spectrum shows the energy distribution of the signal at each frequency and can reflect the operating status of the equipment. Peak extraction is performed on the spectrum to identify the main frequency components and their amplitudes. These peaks constitute the characteristic point set of the equipment operation status. For example, for a centrifuge, there may be obvious peaks at 50Hz, 100Hz, and 200Hz during normal operation, while bearing faults may produce additional peaks in high frequency bands (such as 1kHz-5kHz).

[0047] Gas sampling is performed on the chemical environment of the laboratory to obtain multi-component gas concentration data. Sampling usually uses electronic nose technology, which contains multiple different types of gas sensors, each of which is sensitive to a specific gas or gas combination. The data obtained is a multidimensional vector, with each dimension corresponding to the concentration of a gas. Principal component analysis (PCA) of these data can reduce the data dimension, extract the main components that best represent environmental changes, and form a chemical environment odor fingerprint. PCA projects the original data onto a set of orthogonal new coordinate axes through linear transformation. These new coordinate axes are arranged from large to small according to the data variance. The first few principal components usually contain the main information of the data. At the same time, physiological data are collected from researchers in the laboratory, focusing on heart rate variability and skin conductivity. Heart rate variability reflects the regulatory ability of the autonomic nervous system, while skin conductivity is related to emotions and stress levels. These data form a time series and are then analyzed by wavelet transform. Wavelet transform can provide local characteristics of signals in both time and frequency domains, and is particularly suitable for processing non-stationary signals. By selecting appropriate wavelet basis functions and decomposition scales, the physiological rhythm characteristics of different frequency bands can be extracted and the researchers' biological rhythm characteristic spectrum can be obtained.

[0048] In order to comprehensively analyze these three types of data, timestamp alignment is required. This step ensures the temporal consistency of data from different sources, so that subsequent analysis can accurately reflect the overall status of the laboratory at a specific moment. After alignment, a synchronized multimodal data sequence is obtained. Then data fusion is performed, and this process can adopt feature-level fusion or decision-level fusion methods. Feature-level fusion directly splices the feature vectors of different modes, while decision-level fusion processes the data of each modality separately and then fuses their respective decision results. The laboratory ecological status feature vector obtained after fusion comprehensively reflects the status of equipment, environment and personnel.

[0049] Finally, the laboratory ecological state feature vector is reconstructed in time series to obtain the laboratory ecological state data stream. The reconstruction process can use phase space reconstruction technology, such as the delayed coordinate method. This method can reconstruct the multidimensional dynamic characteristics of the system from the observed univariate time series, which helps to capture the dynamic changes and potential abnormal patterns of the laboratory state.

[0050] For example, in a biological laboratory, a high-speed centrifuge was acoustically monitored with a sampling frequency of 10kHz and an FFT analysis performed once per second. During normal operation, a main peak with an amplitude of 0.8 was observed at 50Hz, and secondary peaks with amplitudes of 0.5 and 0.3 were observed at 100Hz and 200Hz, respectively. At the same time, the electronic nose collected concentration data of 10 gases, including oxygen, carbon dioxide, ammonia, and volatile organic compounds. Through PCA analysis, the first three principal components explained 95% of the data variance, among which the first principal component was highly correlated with CO2 concentration, reflecting the ventilation conditions of the laboratory. The heart rate variability data of the researchers were subjected to wavelet transform, and rhythmic features related to breathing were found in the frequency band of 0.04-0.15Hz. These data were aligned and fused according to a 10-second time window to obtain a laboratory ecological status feature vector with one sample point every 10 seconds. Through time series reconstruction, a data stream reflecting the changes in the overall state of the laboratory over time was finally formed, providing a basis for subsequent anomaly detection and risk assessment.

[0051] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0052] (1) Perform sliding window segmentation on the laboratory ecological status data stream to obtain experimental activity time segments of fixed duration, and extract statistical features and frequency domain features from the experimental activity time segments to obtain a multi-dimensional experimental activity feature vector;

[0053] (2) Cluster analysis of multi-dimensional experimental activity feature vectors is performed using the K-means algorithm to obtain experimental activity type clusters, and semantic matching of experimental activity type clusters is performed based on the preset experimental type dictionary to obtain experimental stage identifiers;

[0054] (3) Construct the experimental activity transfer matrix based on the experimental stage identification to obtain the experimental workflow sequence, and convert the experimental workflow sequence into a directed graph structure to obtain the experimental workflow diagram;

[0055] (4) Calculate the degree centrality and betweenness centrality of each node in the experimental workflow diagram to obtain the key experimental node set, and assign weights to the edges of the experimental workflow diagram according to the centrality index of the key experimental node set to obtain a weighted experimental workflow network;

[0056] (5) Map the weighted experimental workflow network to the laboratory physical space and time dimensions to obtain a dynamic laboratory activity hologram.

[0057] Specifically, the processing of the laboratory ecological status data stream first uses the sliding window technology to segment the continuous data stream into experimental activity time segments of fixed length. The size of the sliding window is set according to the experimental characteristics. For example, 5 minutes can be selected as the window size, and each slide is 1 minute. Statistical features and frequency domain features are extracted for each time segment. Statistical features include mean, variance, skewness, and kurtosis, and frequency domain features are obtained through fast Fourier transform (FFT). Assume that the time segment is , where n is the number of sampling points, then the mean and variance The calculation formula is:

[0058] ;

[0059] ;

[0060] The spectrum obtained after FFT can be represented by F(k), where k is the frequency index. These features are combined to form a multidimensional experimental activity feature vector. Next, the K-means algorithm is used to perform cluster analysis on the multidimensional experimental activity feature vector. The K-means algorithm divides the data points into K clusters through an iterative process, and each iteration calculates the cluster center and redistributes the data points. Assume There are K clusters, is the corresponding cluster center, then the objective function is:

[0061] ;

[0062] The algorithm iterates minimizeJ until convergence. The clustering results are experimental activity type clusters, which are then semantically matched according to the preset experimental type dictionary to obtain the experimental stage identification.

[0063] Based on the experimental stage identification, the experimental activity transfer matrix is ​​constructed. The elements in the transfer matrix P are represents the transition probability from state i to state j, and the calculation formula is:

[0064] ;

[0065] in is the number of observed transitions from state i to state j. The experimental workflow sequence is obtained through the transition matrix and converted into a directed graph structure to form an experimental workflow diagram.

[0066] In the experimental workflow diagram, the degree centrality and betweenness centrality of each node are calculated. Degree centrality reflects the number of direct connections of a node, and betweenness centrality measures the importance of a node as a "bridge". Let G=(V,E) be a graph, V be a node set, and E be an edge set. The degree centrality C_D(v) and betweenness centrality C_B(v) of node v are calculated as follows:

[0067] ;

[0068] ;

[0069] in is the degree of node v, is the number of shortest paths from s to t, is the number of shortest paths from s to t that pass through v. Based on these centrality indicators, the key experimental node sets are identified and weights are assigned to the edges of the experimental workflow diagram to obtain a weighted experimental workflow network. Finally, the weighted experimental workflow network is mapped to the laboratory physical space and time dimensions to generate a dynamic laboratory activity hologram. This step involves combining the abstract network structure with the actual laboratory layout and timeline to form a visual and dynamic representation of experimental activities.

[0070] For example, in a biological laboratory, the continuous 24-hour ecological status data stream is segmented by a 5-minute sliding window (1 minute per slide), resulting in a total of 1436 time segments. Ten statistical features and five main frequency domain features are extracted for each segment to form a 15-dimensional feature vector. These vectors are clustered using the K-means algorithm (K=8) to obtain eight clusters of experimental activity types. By matching with the preset dictionary, these eight clusters are identified as: preparation, culture medium preparation, inoculation, cultivation, observation, data recording, cleaning, and rest. The constructed transition matrix shows that the transition probability from "culture medium preparation" to "inoculation" is 0.85, while the transition probability from "cultivation" to "observation" is 0.3. In the experimental workflow diagram, the degree centrality of the "cultivation" node is 0.6 and the betweenness centrality is 0.4, identifying it as a key node. The final dynamic hologram shows the distribution and transformation of these activities in different areas of the laboratory within 24 hours, providing an intuitive visualization tool for laboratory management.

[0071] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0072] (1) Perform graph structure analysis on the dynamic laboratory activity hologram to identify the key nodes and paths of the experiment, obtain the set of key paths of the experiment, and arrange the set of key paths of the experiment in time sequence to obtain the sequence of key stages of the experiment;

[0073] (2) Extract the time, space and resource occupancy characteristics of each stage in the experimental key stage sequence to obtain a multidimensional experimental feature matrix, and perform principal component analysis on the multidimensional experimental feature matrix to obtain the experimental key factor set;

[0074] (3) Construct a laboratory operation status evaluation index system based on the experimental key factor set to obtain an initial health index score table, and adjust the weight of the initial health index score table to obtain a weighted health index score standard;

[0075] (4) Based on the weighted health index scoring standard, a retrospective evaluation of historical experimental data was performed to obtain a laboratory historical health status data set, and a time series analysis of the laboratory historical health status data set was performed to obtain a health status change trend chart;

[0076] (5) Identify periodic patterns and abnormal fluctuations based on the health status change trend graph to obtain the laboratory health status cycle characteristics, and predict the future health status based on the laboratory health status cycle characteristics to obtain the health status prediction curve;

[0077] (6) Digitally encode the weighted health index scoring standard to obtain a health index scoring vector, and convert the health status change trend graph into time series data to obtain a health trend sequence;

[0078] (7) Discretely sample the health status prediction curve to obtain the prediction status point set, and align the health index score vector, health trend sequence and prediction status point set according to the time dimension to construct a multidimensional time series matrix;

[0079] (8) Extract spatiotemporal features from the multidimensional time series matrix to obtain a health status feature tensor. Perform dimensionality reduction on the health status feature tensor to obtain a multidimensional health index spectrum for laboratory operation.

[0080] Specifically, the processing of dynamic laboratory activity holograms first conducts graph structure analysis, and identifies experimental key nodes and paths through complex network analysis methods. The identification of key nodes is based on indicators such as degree centrality, betweenness centrality, and eigenvector centrality of nodes, while the critical path is determined by the shortest path algorithm and the critical path method (CPM). These key nodes and paths constitute the set of experimental critical paths, which are then arranged in chronological order to form a sequence of experimental key stages. For each stage in the sequence of experimental key stages, time, space, and resource occupancy characteristics are extracted. Time features include duration and start and end time, space features involve the use of laboratory areas, and resource occupancy features include the use of equipment, materials, and human resources. These features form a multidimensional experimental feature matrix, in which each row of the matrix represents an experimental stage and each column represents a feature. Principal component analysis (PCA) is performed on this matrix. By calculating eigenvalues ​​and eigenvectors, several principal components with the largest contribution rates are selected to obtain the set of experimental key factors.

[0081] Based on the set of experimental key factors, a laboratory operation status evaluation index system is constructed. This system converts key factors into quantifiable indicators, such as equipment utilization, energy efficiency, safety index, etc., to form an initial health index score table. The initial score table is weighted and adjusted. Considering the importance of different indicators, the weights are determined by the analytic hierarchy process (AHP) or Delphi method to obtain the weighted health index scoring standard. The weighted health index scoring standard is used to retrospectively evaluate historical experimental data, calculate the laboratory health status score over a period of time, and form a laboratory historical health status data set. This data set is subjected to time series analysis, including trend analysis, seasonal decomposition, and autocorrelation analysis, to obtain a health status change trend chart. This trend chart intuitively shows the long-term changes and short-term fluctuations in the health status of the laboratory.

[0082] Based on the health status change trend chart, periodic patterns and abnormal fluctuations are identified. Periodic patterns may reflect the regular operating rhythm of the laboratory, such as the difference between weekdays and weekends, or seasonal changes. Abnormal fluctuations may indicate potential problems or special events. These characteristics constitute the laboratory health status cycle characteristics and are used to predict future health status. The prediction uses time series prediction methods, such as ARIMA model or exponential smoothing method, to generate health status prediction curves.

[0083] Next, the weighted health index scoring standard is digitally encoded, and the qualitative description is converted into a quantitative indicator to obtain a health index scoring vector. At the same time, the health status change trend graph is converted into time series data to form a health trend sequence. Discrete sampling is performed on the health status prediction curve, and the predicted values ​​at key time points are selected to obtain a predicted state point set. The three sets of data (health index scoring vector, health trend sequence, and predicted state point set) are aligned according to the time dimension to construct a multidimensional time series matrix. Finally, the multidimensional time series matrix is ​​subjected to spatiotemporal feature extraction, and the health status feature tensor is obtained by considering the time correlation and spatial distribution of the data. This tensor is a three-dimensional data structure that contains information in three dimensions: time, space, and features. To reduce the dimensionality of the health status feature tensor, tensor decomposition methods such as Tucker decomposition or CP decomposition can be used to extract the core feature patterns, and finally a multidimensional health index spectrum for laboratory operation is obtained.

[0084] For example, in a biosafety laboratory, five key nodes (sample processing, PCR analysis, cell culture, data analysis, and disinfection) were identified by analyzing the dynamic laboratory activity hologram, forming three key paths. Fifteen features were extracted for each key stage, including duration, energy consumption, number of personnel, etc., to form a 5x15 multidimensional experimental feature matrix. Through PCA analysis, the first three principal components were selected as key factors, which explained 85% of the total variance. Based on these factors, a health index scoring standard including safety index, efficiency index, and quality index was constructed. A retrospective evaluation of the data from the past six months was performed to generate a 180-day health status data set. Time series analysis showed a periodic pattern on weekdays and weekends, as well as a monthly fluctuation related to equipment maintenance. The ARIMA (1,1,1) model was used to predict the health status for the next 30 days. These data were integrated into a 180x3x5 tensor (180 days, 3 indices, 5 key nodes), and the core pattern was extracted by decomposition, finally obtaining a multidimensional health index spectrum that can reflect the overall operation status of the laboratory. This spectrum intuitively demonstrates the laboratory's performance in the three dimensions of safety, efficiency and quality, and provides a scientific basis for laboratory management decisions.

[0085] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0086] (1) Preprocess and standardize the real-time collected laboratory ecological status data to obtain a normalized real-time status feature vector, and then perform feature matching between the normalized real-time status feature vector and the multidimensional health index spectrum to obtain a feature difference matrix;

[0087] (2) Perform threshold analysis and anomaly detection on the feature difference matrix to identify indicators that deviate from the normal range and obtain a preliminary abnormal indicator set. Perform context correlation analysis on the preliminary abnormal indicator set through laboratory historical data to obtain a group of related abnormal indicators.

[0088] (3) Construct a candidate set of abnormal events based on the associated abnormal indicator group, and perform multi-dimensional evaluation and sorting on the candidate set of abnormal events to obtain a priority list of abnormal events;

[0089] (4) Perform temporal correlation analysis on the events in the abnormal event priority list, identify the causal relationship and evolution law between events, obtain the abnormal event correlation network, and extract the key abnormal event path based on the abnormal event correlation network to obtain the core abnormal event chain;

[0090] (5) Perform semantic annotation and feature extraction on the core abnormal event chain to generate an abnormal event descriptor, and then perform pattern matching between the abnormal event descriptor and the predefined sub-health pattern library to obtain the sub-health event type identification;

[0091] (6) Match each event node in the core abnormal event chain with the corresponding abnormal event descriptor one by one to obtain an event description correspondence table, for each event description pair in the event description correspondence table;

[0092] (7) Perform association analysis on the sub-health event type identifier based on each event description to obtain a set of triples;

[0093] (8) Arrange the triple sets in chronological order, construct a temporal association graph, and perform causal reasoning on the temporal association graph to obtain a causal link structure;

[0094] (9) The causal link structure is integrated with the original event sequence information to construct a multi-level event network. The multi-level event network is then visualized and the data structure optimized to obtain the characteristic chain of laboratory sub-health events.

[0095] Specifically, the construction process of the laboratory sub-health event feature chain begins with the preprocessing and standardization of the laboratory ecological status data collected in real time. This includes removing noise, filling missing values ​​and normalization, unifying data of different dimensions to the same scale, and obtaining a standardized real-time state feature vector. Subsequently, this feature vector is feature matched with the multidimensional health index spectrum, and the difference in each feature dimension is calculated to form a feature difference matrix. This matrix reflects the degree of deviation between the current laboratory state and the health standard. Threshold analysis and anomaly detection are performed on the feature difference matrix to identify indicators that significantly deviate from the normal range and form a preliminary abnormal indicator set. The setting of the threshold is based on statistical principles, such as the 3σ principle or the quartile rule. Next, the context correlation analysis of these preliminary abnormal indicators is performed using laboratory historical data, considering the mutual influence and potential causal relationship between the indicators, so as to obtain a group of associated abnormal indicators. This step helps to reduce false alarms and improve the accuracy of anomaly detection.

[0096] Based on the associated abnormal indicator group, a candidate set of abnormal events is constructed. Each candidate event is defined by a set of related abnormal indicators. These candidate events are evaluated in multiple dimensions, considering factors such as severity, scope of impact, and urgency, and sorted to obtain a priority list of abnormal events. This list provides an important basis for subsequent processing, ensuring that the most critical issues can be handled in a timely manner. The events in the priority list of abnormal events are analyzed for temporal correlation to identify the causal relationship and evolution law between events. This process uses time series analysis techniques, such as Granger causality test or dynamic time warping algorithm, to construct an abnormal event correlation network. In this network, nodes represent abnormal events, and edges represent the strength of correlation between events. Based on this network, the key abnormal event paths are extracted to obtain the core abnormal event chain, which reflects the most important abnormal event sequence and its evolution process.

[0097] The core abnormal event chain is semantically annotated and feature extracted to generate abnormal event descriptors. The descriptor contains key information such as the type, severity, and duration of the event. These descriptors are pattern matched with the predefined sub-health pattern library to obtain the sub-health event type identification. This step converts quantitative abnormal indicators into qualitative event descriptions, which is convenient for managers to understand and make decisions. Next, each event node in the core abnormal event chain is matched one by one with the corresponding abnormal event descriptor to obtain an event description correspondence table. Each event description pair contains quantitative indicators and qualitative descriptions of the event. Then, an association analysis is performed on each event description pair with the sub-health event type identification to obtain a set of triples. Each triple contains an event, a descriptor, and a type identification, which fully describes the characteristics of an abnormal event.

[0098] Arrange these triplets in chronological order to construct a time series association graph. This graph intuitively shows the development context of abnormal events. Perform causal reasoning on the time series association graph and use Bayesian networks or structural equation models to obtain a causal link structure. This structure reveals the deep-level associations and impact mechanisms between abnormal events. Finally, the causal link structure is fused with the original event time series information to construct a multi-level event network. This network contains information in multiple dimensions such as time, causality, and type. Visualize the multi-level event network and optimize the data structure to finally obtain the characteristic chain of laboratory sub-health events. This characteristic chain is a comprehensive data structure that fully reflects the evolution process and internal mechanism of the laboratory sub-health state.

[0099] For example, in a biological laboratory, the real-time data collected include 20 indicators such as temperature, humidity, air pressure, and air quality. After preprocessing and standardization, a 20-dimensional real-time state feature vector is obtained. By comparing this vector with the health index spectrum, it is found that the temperature and air quality indicators deviate significantly from the normal range, forming a preliminary abnormal indicator set. Through context analysis, it is found that these two indicators are related to the abnormality of the ventilation system, forming a group of associated abnormal indicators.

[0100] Based on this indicator group, the abnormal event "Decrease in ventilation system efficiency" was identified and listed as a high-priority event. Time series analysis showed that before this event, a low-level abnormality of "filter blockage" occurred, and the two formed a core abnormal event chain. This chain was semantically annotated to obtain event descriptors, such as "Decrease in ventilation system efficiency: severity - high, duration - 2 hours, impact range - the entire laboratory".

[0101] Events are matched with descriptors to form event description pairs, such as "filter blockage-slight blockage, lasting 8 hours" and "ventilation system efficiency reduction-severe reduction, lasting 2 hours". These description pairs are associated with the sub-health event type "environmental control system abnormality" to form triplets. These triplets are arranged in chronological order to construct a time series association graph from "filter blockage" to "ventilation system efficiency reduction". Causal reasoning shows that "filter blockage" is the direct cause of "ventilation system efficiency reduction", forming a causal link structure. Finally, this information is integrated into a multi-level event network, which intuitively shows how the initial filter problem evolves into a serious ventilation system abnormality, and the impact of this process on the laboratory environment. This network constitutes a characteristic chain of laboratory sub-health events, providing laboratory managers with comprehensive problem diagnosis and decision support information.

[0102] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0103] (1) Perform time series decomposition on the characteristic chain of laboratory sub-health events, extract the time, space and type characteristics of the events, obtain the sub-health event characteristic matrix, and construct an event association graph based on the sub-health event characteristic matrix to obtain the sub-health event network structure;

[0104] (2) Perform topological analysis on the network structure of sub-health events, identify key nodes and core paths, obtain the transmission chain of sub-health events, and trace the transmission chain of sub-health events through laboratory historical data to obtain the source set of sub-health events;

[0105] (3) Based on the source set of sub-health events, causal relationship analysis is performed to construct an event causal network, obtain the causal chain of sub-health events, and conduct risk assessment on the causal chain of sub-health events to obtain a risk level distribution map;

[0106] (4) The trend of the sub-health event feature matrix is ​​extrapolated through time series prediction technology to obtain the sub-health event development prediction curve, and the prediction results are risk-weighted through the risk level distribution map to obtain the risk evolution prediction map;

[0107] (5) Based on the causal chain of sub-health events and the risk evolution prediction diagram, a multidimensional risk evolution model is constructed to obtain the initial risk evolution path set, and a multi-scenario simulation is performed on the initial risk evolution path set to obtain a risk evolution scenario tree;

[0108] (6) The sub-health event network structure, sub-health event causal chain, risk evolution prediction diagram and risk evolution scenario tree are visualized and integrated to obtain a laboratory risk evolution path map.

[0109] Specifically, the construction process of the laboratory risk evolution path map begins with the time series decomposition of the laboratory sub-health event feature chain. By extracting the occurrence time, spatial location and type characteristics of each event, a sub-health event feature matrix is ​​formed. In this matrix, each row represents a sub-health event, and the column contains the timestamp, spatial coordinates and event type code. Based on this feature matrix, an event association graph is constructed, in which the nodes represent sub-health events and the edges represent the strength of association between events, thereby obtaining the sub-health event network structure. The sub-health event network structure is topologically analyzed to identify key nodes and core paths. Key nodes are determined by calculating indicators such as degree centrality and betweenness centrality of nodes, while core paths are identified by finding the maximum weight path in the network. These key nodes and core paths constitute the sub-health event propagation chain, reflecting the propagation process of sub-health status in the laboratory. Subsequently, the laboratory historical data is used to trace the source of this propagation chain, find the initial triggering factors that trigger a series of sub-health events, and obtain the source set of sub-health events.

[0110] Based on the source set of sub-health events, an in-depth causal analysis is conducted. This step uses Bayesian networks or structural equation models to construct event causal networks and reveal causal links between events. The resulting causal chain of sub-health events not only shows the order of events, but also reflects the intensity of causal influence between events. Risk assessment is performed on this causal chain, considering the severity and scope of impact of each event, and a risk level distribution map is generated to intuitively display the distribution of risks in different laboratory areas and time periods. Next, the time series prediction technology is used to perform trend extrapolation on the sub-health event feature matrix. In this step, the ARIMA model or long short-term memory network (LSTM) can be used to predict the development trend of sub-health events in the future and obtain a sub-health event development prediction curve. Then, this prediction result is combined with the risk level distribution map obtained previously, and the prediction result is risk-weighted to generate a risk evolution prediction map. This prediction map not only shows the future development trend of sub-health events, but also reflects the degree of risk that these events may bring.

[0111] Based on the causal chain of sub-health events and the risk evolution prediction diagram, a multidimensional risk evolution model is constructed. This model comprehensively considers the causal relationship, time evolution and risk level of the event to form an initial risk evolution path set. Multi-scenario simulation is performed on this path set, considering different intervention measures and changes in external conditions to generate a risk evolution scenario tree. This scenario tree shows the possible evolution paths and results of laboratory risks under different circumstances. Finally, the sub-health event network structure, sub-health event causal chain, risk evolution prediction diagram and risk evolution scenario tree are visualized and integrated to obtain a laboratory risk evolution path map. This map is a comprehensive visualization tool that intuitively shows the formation, spread, evolution and potential risks of laboratory sub-health status, providing comprehensive decision-making support for laboratory management and risk prevention and control.

[0112] For example, in a biosafety laboratory, 50 sub-health events were extracted by analyzing the characteristic chain of sub-health events for one month, forming a 50x5 characteristic matrix (50 events, each event contains time, x-coordinate, y-coordinate, z-coordinate and event type code). The event association diagram constructed based on this matrix shows that there are 3 highly correlated event clusters, corresponding to the laboratory's culture area, PCR area and waste disposal area. Topological analysis identified that "bacterial incubator temperature anomaly" is a key node, which is connected to multiple subsequent events. Traceability tracking found that this temperature anomaly was caused by a failure in the refrigeration system.

[0113] Causal analysis showed that the failure of the refrigeration system caused the incubator temperature to rise, which in turn affected the growth rate of bacteria and ultimately caused abnormal experimental data. The risk assessment results of this causal chain showed that the culture area had the highest risk level, reaching level 4 (the severity level on the 5-level scale). Time series prediction showed that if no intervention was taken, the frequency of related sub-health events would increase by 50% in the next week. Combining this prediction result with the risk level, the resulting risk evolution prediction map showed that the risk in the culture area could rise to level 5 within 3 days. The multi-scenario simulation considered three situations: no intervention, partial repair, and complete replacement of the refrigeration system. The simulation results formed a three-branch risk evolution scenario tree, in which the branch of complete replacement of the refrigeration system showed that the risk could be reduced to level 2 within a week. The final risk evolution path map intuitively shows the changes from the initial refrigeration system failure to the risk level of each area, as well as the risk evolution path under different intervention measures. This map provides laboratory managers with a clear picture of the risk and helps to formulate the optimal intervention strategy.

[0114] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0115] (1) Perform risk level assessment on the laboratory risk evolution path map to obtain a risk level matrix, and set multi-level warning thresholds based on the risk level matrix to obtain a laboratory safety warning threshold table;

[0116] (2) Monitor the key nodes in the laboratory risk evolution path map in real time, extract the current risk status characteristics, obtain the real-time risk status vector, and compare the real-time risk status vector with the laboratory safety warning threshold table to obtain the warning trigger signal;

[0117] (3) Generate warning information including risk type, risk level and impact scope according to the warning trigger signal to obtain laboratory safety warning signal, and classify the laboratory safety warning signal to obtain a hierarchical warning information set;

[0118] (4) Based on the hierarchical warning information set, query the preset emergency response rule library, extract the corresponding emergency disposal plan, obtain the initial emergency response strategy set, and conduct a feasibility assessment on the initial emergency response strategy set based on the current resource status of the laboratory to obtain a list of executable emergency plans;

[0119] (5) Perform multi-objective optimization on the plans in the executable emergency plan list to obtain the optimal emergency response plan, and convert the optimal emergency response plan into an operation instruction sequence to obtain an emergency response execution list;

[0120] (6) Integrate laboratory safety warning signals and emergency response execution checklists into a comprehensive emergency command package, and trigger corresponding emergency response strategies through the comprehensive emergency command package.

[0121] Specifically, the construction of the laboratory safety early warning system begins with a comprehensive risk level assessment of the laboratory risk evolution path map. This process generates a risk level matrix by analyzing the risk level and impact range of each node in the map. The matrix is ​​a multidimensional data structure that contains the level information of different experimental areas and different types of risks. Based on this risk level matrix, multi-level early warning thresholds are set to form a laboratory safety early warning threshold table. This threshold table defines the trigger conditions under different risk levels and provides judgment criteria for subsequent real-time monitoring. Continuous real-time monitoring of key nodes in the laboratory risk evolution path map is carried out. These key nodes are usually important hubs or high-risk areas for risk transmission, such as key equipment, hazardous materials storage areas, etc. Through various sensors and data acquisition devices, the current risk state characteristics are extracted to form a real-time risk state vector. This vector contains risk indicators in multiple dimensions, such as temperature, pressure, gas concentration, etc. This real-time risk state vector is compared with the previously formulated laboratory safety early warning threshold table. If some indicators exceed the preset threshold, an early warning trigger signal will be generated.

[0122] The early warning trigger signal is the key to start the safety early warning mechanism. According to this signal, detailed early warning information including risk type, risk level and impact scope is generated to form a laboratory safety early warning signal. This early warning signal is not a simple binary alarm, but a data packet containing rich information. Subsequently, this safety early warning signal is classified and classified to obtain a hierarchical early warning information set. This information set classifies the early warning information according to factors such as severity and impact scope, which facilitates subsequent accurate response. Based on the hierarchical early warning information set, the system will query the pre-set emergency response rule base. This rule base contains standard disposal plans for different types and levels of risks. By matching the early warning information and the rule base, the corresponding emergency disposal plan is extracted to form an initial emergency response strategy set. However, these initial strategies also need to be evaluated for feasibility based on the current resource status of the laboratory. This step takes into account actual factors such as human resources, equipment status, and emergency material reserves to obtain a list of executable emergency plans.

[0123] Multi-objective optimization of the options in the executable emergency plan list is a key step to ensure the response effect. This process takes into account multiple objectives, such as minimizing risks, maximizing resource utilization efficiency, and minimizing the impact on normal experimental activities. The optimal emergency response plan is obtained through multi-objective optimization algorithms, such as genetic algorithms or particle swarm optimization. This plan is then converted into a specific sequence of operation instructions to form an emergency response execution list. This list lists in detail each step that needs to be performed, including information such as the execution order, responsible persons, and required resources. Finally, the laboratory safety warning signal and the emergency response execution list are integrated to form a comprehensive emergency instruction package. This instruction package is a complete emergency response plan that contains all information from risk identification to specific response measures. Through this comprehensive emergency instruction package, the corresponding emergency response strategy is triggered and the actual emergency response process is started.

[0124] For example, in a biosafety laboratory, the risk evolution path map shows that there is a potential risk of virus contamination in the PCR area. The risk level assessment sets this risk as level 4 (the severity level in the 5-level scale). Based on this, multi-level warning thresholds are set. For example, when the concentration of virus particles in the air exceeds 100 / m³, a level 3 warning is triggered, and when it exceeds 500 / m³, a level 4 warning is triggered. The real-time monitoring system detects that the concentration of virus particles in the PCR area reaches 600 / m³, and immediately generates a warning trigger signal. The system generates warning information including "biological contamination", "level 4 risk", and "scope of impact: PCR area and adjacent areas".

[0125] The emergency response rule base was queried to extract the corresponding disposal plan, including steps such as area blockade, ultraviolet disinfection, and personnel evacuation. Considering that there are only two personnel with biosafety level 4 protection qualifications on duty in the laboratory, the system adjusted the initial plan to form an executable list of emergency plans. Through multi-objective optimization, a plan that balances safety and experimental progress was selected: immediately block the PCR area, start ultraviolet disinfection, and preventively disinfect adjacent areas, and suspend non-emergency experimental activities. This plan was converted into 15 specific operational steps, including starting the alarm system, shutting down the ventilation system in related areas, and wearing protective equipment. Finally, a comprehensive emergency instruction package containing complete warning information and detailed response steps was generated and distributed to relevant personnel through the laboratory's emergency response system, which quickly started the emergency response process.

[0126] The present invention also provides a laboratory intelligent integrated management system. Figure 2 As shown, the laboratory intelligent integrated management system specifically includes:

[0127] The acquisition module 201 is used to perform multimodal sensing and acquisition of the acoustic characteristics of laboratory equipment operation, chemical environment odor fingerprints and researcher biorhythm data in the laboratory to obtain a laboratory ecological status data stream;

[0128] A mapping module 202 is used to identify the experimental stage and map the workflow of the laboratory ecological status data flow to obtain a dynamic laboratory activity hologram;

[0129] Analysis module 203, used to perform experimental critical path analysis on the dynamic laboratory activity hologram to obtain a multi-dimensional health index spectrum of laboratory operation;

[0130] The identification module 204 is used to perform intelligent comparison and abnormal pattern recognition on the laboratory ecological status data collected in real time and the multi-dimensional health index spectrum to obtain a characteristic chain of laboratory sub-health events;

[0131] The traceability module 205 is used to perform traceability analysis and development trend analysis on the characteristic chain of the laboratory sub-health events to obtain a laboratory risk evolution path map;

[0132] The early warning module 206 is used to generate a laboratory safety early warning signal based on the laboratory risk evolution path map and trigger a corresponding emergency response strategy.

[0133] Through the collaborative work of the above modules, the intelligent comprehensive management method of the laboratory realizes comprehensive and real-time monitoring of the laboratory operation status through multimodal sensing acquisition and data fusion technology, significantly improving the accuracy and comprehensiveness of data acquisition. Through the comprehensive analysis of the acoustic characteristics of laboratory equipment operation, the chemical environment odor fingerprint and the biorhythm data of researchers, it is possible to capture minor anomalies that are difficult to detect with traditional methods, thereby greatly improving the sensitivity and predictability of laboratory safety management. The introduction of experimental stage identification and workflow mapping enables accurate tracking of experimental progress and identification of key nodes, which not only optimizes resource allocation, but also provides important support for the early detection of abnormal events. The construction of dynamic laboratory activity holograms provides managers with an intuitive and dynamic visualization tool for laboratory operation status, effectively improving the timeliness and accuracy of decision-making. The introduction of multidimensional health index spectrum enables the evaluation of laboratory operation status to shift from a single indicator to a multi-dimensional, comprehensive evaluation, greatly enhancing the scientificity and comprehensiveness of the evaluation. Through the intelligent comparison and abnormal pattern recognition of real-time data and health index spectrum, potential sub-health status can be quickly identified, providing strong support for preventive maintenance and risk management. The construction and analysis of the characteristic chain of sub-health events enables an in-depth understanding of the occurrence mechanism and evolution law of abnormal events, providing a scientific basis for the formulation of targeted preventive measures. The generation of risk evolution path maps provides laboratory managers with a global view of risk development, which helps to formulate long-term risk management and control strategies. The design of multi-level early warning mechanisms and adaptive emergency response strategies has greatly improved the laboratory's ability and efficiency in responding to emergencies. Through the selection of emergency plans based on multi-objective optimization, the best balance can be found between safety, resource utilization efficiency and experimental progress. The automatic generation and distribution of comprehensive emergency instruction packages significantly shortens the time from risk identification to response implementation, and improves the efficiency and accuracy of emergency disposal.

[0134] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the embodiments, a person skilled in the art should understand that the specific implementation modes of the present invention can still be modified or replaced by equivalents, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A laboratory intelligent comprehensive management method, characterized in that: include: Multimodal sensing and collection of laboratory instrument and equipment operating acoustic characteristics, chemical environment odor fingerprints, and researchers' biorhythm data are performed to obtain laboratory ecological status data streams; The laboratory ecological status data stream is subjected to experimental stage identification and workflow mapping to obtain a dynamic laboratory activity hologram, including: performing sliding window segmentation on the laboratory ecological status data stream to obtain experimental activity time segments of fixed duration, and extracting statistical features and frequency domain features from the experimental activity time segments to obtain a multidimensional experimental activity feature vector; performing cluster analysis on the multidimensional experimental activity feature vector by a K-means algorithm to obtain an experimental activity type cluster, and performing semantic matching on the experimental activity type cluster according to a preset experimental type dictionary to obtain an experimental stage identifier; constructing an experimental activity transfer matrix based on the experimental stage identifier to obtain an experimental workflow sequence, and converting the experimental workflow sequence into a directed graph structure to obtain an experimental workflow diagram; calculating the degree centrality and betweenness centrality of each node in the experimental workflow diagram to obtain a set of key experimental nodes, and assigning weights to the edges of the experimental workflow diagram according to the centrality index of the key experimental node set to obtain a weighted experimental workflow network; mapping the weighted experimental workflow network to the laboratory physical space and time dimensions to obtain the dynamic laboratory activity hologram; Conducting experimental critical path analysis on the dynamic laboratory activity hologram to obtain a multi-dimensional health index spectrum of laboratory operation; Intelligently compare and identify abnormal patterns of laboratory ecological status data collected in real time with the multi-dimensional health index spectrum to obtain a characteristic chain of laboratory sub-health events; Conduct source tracing and development trend analysis on the characteristic chain of laboratory sub-health events to obtain a laboratory risk evolution path map; A laboratory safety early warning signal is generated based on the laboratory risk evolution path map, and a corresponding emergency response strategy is triggered.

2. The laboratory intelligent comprehensive management method according to claim 1 is characterized in that: The steps of performing multimodal sensing and acquisition of the acoustic characteristics of laboratory equipment operation, chemical environment odor fingerprints, and biorhythm data of researchers in the laboratory to obtain the laboratory ecological status data flow include: Performing spectrum analysis on the acoustic signals of the laboratory instrument and equipment to obtain an acoustic characteristic spectrum diagram of the equipment operation, and performing peak extraction on the acoustic characteristic spectrum diagram of the equipment operation to obtain a characteristic point set of the equipment operation status; Sampling the chemical environment of the laboratory to obtain multi-component gas concentration data, and performing principal component analysis on the multi-component gas concentration data to obtain a chemical environment odor fingerprint; The physiological data of researchers in the laboratory are collected to obtain heart rate variability and skin conductivity time series, and the heart rate variability and skin conductivity time series are subjected to wavelet transformation to obtain the biorhythm characteristic spectrum of the researchers; Performing time stamp alignment on the equipment operation status feature point set, the chemical environment odor fingerprint and the researcher's biorhythm feature spectrum to obtain a synchronized multimodal data sequence, and performing data fusion on the synchronized multimodal data sequence to obtain a laboratory ecological status feature vector; The laboratory ecological state feature vector is reconstructed in time series to obtain the laboratory ecological state data stream.

3. The laboratory intelligent comprehensive management method according to claim 1 is characterized in that: The steps of performing experimental key path analysis on the dynamic laboratory activity hologram to obtain a multidimensional health index spectrum of laboratory operation include: Performing a graph structure analysis on the dynamic laboratory activity hologram, identifying key nodes and paths of the experiment, obtaining a set of key paths of the experiment, and arranging the set of key paths of the experiment in time sequence to obtain a sequence of key stages of the experiment; Extracting time, space and resource occupancy characteristics from each stage in the experimental key stage sequence to obtain a multidimensional experimental feature matrix, and performing principal component analysis on the multidimensional experimental feature matrix to obtain an experimental key factor set; Constructing a laboratory operation status evaluation index system according to the experimental key factor set to obtain an initial health index scoring table, and performing weight adjustment on the initial health index scoring table to obtain a weighted health index scoring standard; Performing a retrospective evaluation on the historical experimental data based on the weighted health index scoring standard to obtain a laboratory historical health status data set, and performing a time series analysis on the laboratory historical health status data set to obtain a health status change trend graph; Identify periodic patterns and abnormal fluctuations according to the health status change trend graph to obtain laboratory health status cycle characteristics, and predict future health status through the laboratory health status cycle characteristics to obtain a health status prediction curve; Digitally encoding the weighted health index scoring standard to obtain a health index scoring vector, and converting the health status change trend graph into time series data to obtain a health trend sequence; Discrete sampling is performed on the health status prediction curve to obtain a prediction status point set, and the health index score vector, the health trend sequence and the prediction status point set are aligned according to the time dimension to construct a multi-dimensional time series matrix; The multidimensional time series matrix is ​​subjected to spatiotemporal feature extraction to obtain a health status feature tensor, and the health status feature tensor is subjected to dimensionality reduction processing to obtain a multidimensional health index spectrum of the laboratory operation.

4. The laboratory intelligent comprehensive management method according to claim 1, characterized in that: The steps of intelligently comparing the real-time collected laboratory ecological status data with the multi-dimensional health index spectrum and identifying abnormal patterns to obtain a laboratory sub-health event feature chain include: Preprocessing and standardizing the laboratory ecological status data collected in real time to obtain a normalized real-time status feature vector, and matching the normalized real-time status feature vector with the multidimensional health index spectrum to obtain a feature difference matrix; Performing threshold analysis and anomaly detection on the feature difference matrix to identify indicators that deviate from the normal range to obtain a preliminary abnormal indicator set, and performing context correlation analysis on the preliminary abnormal indicator set through laboratory historical data to obtain a group of associated abnormal indicators; Constructing an abnormal event candidate set according to the associated abnormal indicator group, and performing multi-dimensional evaluation and sorting on the abnormal event candidate set to obtain an abnormal event priority list; Performing time series correlation analysis on the events in the abnormal event priority list, identifying the causal relationship and evolution law between the events, obtaining an abnormal event correlation network, and extracting key abnormal event paths based on the abnormal event correlation network to obtain a core abnormal event chain; Performing semantic annotation and feature extraction on the core abnormal event chain to generate an abnormal event descriptor, and performing pattern matching on the abnormal event descriptor and a predefined sub-health pattern library to obtain a sub-health event type identifier; Match each event node in the core abnormal event chain with the corresponding abnormal event descriptor one by one to obtain an event description corresponding table, and for each event description pair in the event description corresponding table; According to each event description, an association analysis is performed with the sub-health event type identifier to obtain a triplet set; Arranging the triplet set in chronological order, constructing a time series association graph, and performing causal relationship reasoning on the time series association graph to obtain a causal link structure; The causal link structure is data-fused with the original event time sequence information to construct a multi-level event network, and the multi-level event network is visualized and data structure optimized to obtain the laboratory sub-health event feature chain.

5. The laboratory intelligent comprehensive management method according to claim 1, characterized in that: The step of tracing the source and developing trend of the characteristic chain of the laboratory sub-health events to obtain a laboratory risk evolution path map includes: Performing time series decomposition on the laboratory sub-health event feature chain, extracting the time, space and type characteristics of the event, obtaining a sub-health event feature matrix, and constructing an event association graph based on the sub-health event feature matrix to obtain a sub-health event network structure; Performing topological analysis on the network structure of the sub-health events, identifying key nodes and core paths, obtaining a sub-health event propagation chain, and tracing the sub-health event propagation chain through laboratory historical data to obtain a sub-health event source set; Performing causal relationship analysis based on the sub-health event source set, constructing an event causal network, obtaining a sub-health event causal chain, and performing risk assessment on the sub-health event causal chain to obtain a risk level distribution map; Performing trend extrapolation on the sub-health event feature matrix through time series prediction technology to obtain a sub-health event development prediction curve, and performing risk weighting on the prediction results through the risk level distribution map to obtain a risk evolution prediction map; A multidimensional risk evolution model is constructed based on the sub-health event causal chain and the risk evolution prediction graph to obtain an initial risk evolution path set, and a multi-scenario simulation is performed on the initial risk evolution path set to obtain a risk evolution scenario tree; The sub-health event network structure, the sub-health event causal chain, the risk evolution prediction diagram and the risk evolution scenario tree are visualized and integrated to obtain the laboratory risk evolution path map.

6. The laboratory intelligent comprehensive management method according to claim 5, characterized in that: The generating of laboratory safety early warning signals based on the laboratory risk evolution path map and triggering corresponding emergency response strategy steps include: Performing a risk level assessment on the laboratory risk evolution path map to obtain a risk level matrix, and setting a multi-level warning threshold based on the risk level matrix to obtain a laboratory safety warning threshold table; Performing real-time monitoring on key nodes in the laboratory risk evolution path map, extracting current risk state characteristics, obtaining a real-time risk state vector, and comparing the real-time risk state vector with the laboratory safety early warning threshold table to obtain an early warning trigger signal; Generate warning information including risk type, risk level and impact scope according to the warning trigger signal to obtain a laboratory safety warning signal, and classify the laboratory safety warning signal to obtain a hierarchical warning information set; Based on the hierarchical warning information set, a preset emergency response rule base is queried to extract the corresponding emergency disposal plan to obtain an initial emergency response strategy set, and the feasibility of the initial emergency response strategy set is evaluated according to the current resource status of the laboratory to obtain an executable emergency plan list; Performing multi-objective optimization on the plans in the executable emergency plan list to obtain a preferred emergency response plan, and converting the preferred emergency response plan into an operation instruction sequence to obtain an emergency response execution list; The laboratory safety early warning signal and the emergency response execution list are integrated into a comprehensive emergency instruction package, and the corresponding emergency response strategy is triggered by the comprehensive emergency instruction package.

7. A laboratory intelligent integrated management system, used to execute the laboratory intelligent integrated management method according to any one of claims 1 to 6, characterized in that: include: The acquisition module is used to perform multimodal perception and acquisition of the laboratory instrument and equipment operating acoustic characteristics, chemical environment odor fingerprints, and researchers' biorhythm data to obtain the laboratory ecological status data stream; A mapping module is used to perform experimental phase identification and workflow mapping on the laboratory ecological status data stream to obtain a dynamic laboratory activity hologram, including: performing sliding window segmentation on the laboratory ecological status data stream to obtain experimental activity time segments of fixed duration, and extracting statistical features and frequency domain features from the experimental activity time segments to obtain a multidimensional experimental activity feature vector; performing cluster analysis on the multidimensional experimental activity feature vector by a K-means algorithm to obtain an experimental activity type cluster, and performing semantic matching on the experimental activity type cluster according to a preset experimental type dictionary to obtain an experimental phase identifier; constructing an experimental activity transfer matrix based on the experimental phase identifier to obtain an experimental workflow sequence, and converting the experimental workflow sequence into a directed graph structure to obtain an experimental workflow diagram; calculating the degree centrality and betweenness centrality of each node in the experimental workflow diagram to obtain a set of key experimental nodes, and assigning weights to the edges of the experimental workflow diagram according to the centrality index of the key experimental node set to obtain a weighted experimental workflow network; mapping the weighted experimental workflow network to the laboratory physical space and time dimensions to obtain the dynamic laboratory activity hologram; An analysis module, used to perform experimental critical path analysis on the dynamic laboratory activity hologram to obtain a multi-dimensional health index spectrum of laboratory operation; An identification module is used to intelligently compare the real-time collected laboratory ecological status data with the multi-dimensional health index spectrum and identify abnormal patterns to obtain a characteristic chain of laboratory sub-health events; A traceability module is used to perform traceability analysis and development trend analysis on the characteristic chain of laboratory sub-health events to obtain a laboratory risk evolution path map; The early warning module is used to generate a laboratory safety early warning signal based on the laboratory risk evolution path map and trigger a corresponding emergency response strategy.

Images