Artificial Intelligence-Based Laboratory Safety Management System

Abstract

The present invention discloses a laboratory safety management system based on artificial intelligence, including a data acquisition module, a laboratory safety management model establishment module, and a laboratory safety management module. The present invention belongs to the field of data processing, specifically referring to a laboratory safety management system based on artificial intelligence. In this solution, an interval type-2 asymmetric Gaussian membership function is introduced to adapt to the dynamic changes of the laboratory environment in real time and avoid false alarms; a variable interaction weight and a time activation mechanism with a time window are introduced to record the long-term risk trend and accurately identify multi-variable risks in complex environments, thereby improving the laboratory safety management effect; a rule self-adaptive adjustment unit is designed to generate or remove rules in real time based on feature importance and rule usage, automatically adapting to risk changes; a dynamic interval reduction coefficient is introduced to balance the credibility of the upper and lower bound activation intensities according to historical data, reduce the impact of uncertainty on risk assessment, and thus improve the early warning accuracy.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and specifically refers to a laboratory safety management system based on artificial intelligence. Background Art

[0002] A laboratory safety management system is a comprehensive management platform that combines various technical means to monitor, evaluate, and warn the environmental status, equipment operation, personnel behavior, and potential risks of a laboratory in real time. It aims to ensure the safety of laboratory operations, reduce the accident rate, and respond to abnormal or emergency situations in a timely manner through dynamic data collection, risk modeling, and intelligent decision-making, so as to protect the safety of laboratory personnel, equipment, and the environment. However, the general laboratory safety management system has serious problems of being severely affected by noise, being unable to adapt to the changes of laboratory distribution data in real time, and being unable to identify risks in complex environments, resulting in poor safety management effects; the general laboratory safety management system has poor adaptability to risk changes and inaccurate warnings due to being affected by noise. Summary of the Invention

[0003] In view of the above situation, to overcome the defects of the prior art, the present invention provides a laboratory safety management system based on artificial intelligence. Aiming at the problems that the general laboratory safety management system is severely affected by noise, unable to adapt to the changes of laboratory distribution data in real time, and unable to identify risks in complex environments, resulting in poor safety management effects, this solution introduces an interval type-2 asymmetric Gaussian membership function to adapt to the dynamic changes of the laboratory environment in real time and avoid false alarms; introduces variable interaction weights and a time activation mechanism with a time window to record long-term risk trends and accurately identify multi-variable risks in complex environments, thereby improving the laboratory safety management effect; aiming at the problems that the general laboratory safety management system has poor adaptability to risk changes and inaccurate warnings due to being affected by noise, this solution designs a rule self-adaptive adjustment unit to generate or remove rules in real time based on feature importance and rule usage, and automatically adapt to risk changes; introduces a dynamic interval reduction coefficient to weigh the credibility of the upper and lower bound activation intensities according to historical data, reduce the impact of uncertainty on risk assessment, and thereby improve the warning accuracy.

[0004] The technical solution adopted by the present invention is as follows: The laboratory safety management system based on artificial intelligence provided by the present invention includes a data collection module, a laboratory safety management model establishment module, and a laboratory safety management module;

[0005] The data collection module collects historical laboratory management data;

[0006] The laboratory safety management model establishment module realizes the establishment of a laboratory safety management model based on the dynamic environment of the laboratory through fuzzyification, a double activation mechanism, rule self-adaptive adjustment, risk assessment, and parameter optimization.

[0007] The laboratory safety management module performs laboratory safety management on the real-time collected laboratory management data based on the established laboratory safety management model.

[0008] Furthermore, in the data acquisition module, the historical laboratory management data includes environmental sensor data, equipment status monitoring data, personnel behavior data, time, and laboratory safety status evaluation levels; the laboratory safety status evaluation level is used as a data label; after cleaning, transforming, and standardizing the collected data, input data is obtained.

[0009] Furthermore, the laboratory safety management model establishment module specifically includes the following:

[0010] Fuzzification layer design unit; using interval type-2 asymmetric Gaussian membership functions to handle uncertainties and noises in environmental monitoring, and defining the fuzzy range of input variables based on the upper membership function and the lower membership function, expressed as: ; Dynamically adjusting the fuzzy membership center to adapt to dynamic laboratory distribution data, expressed as: ; where is the upper bound of the fuzzy membership degree of the j-th input variable in the k-th rule; t is time; is the input variable; and are the left and right bounds of the fuzzy membership center of the input variable; and are the left and right standard deviations of the fuzzy range; is the learning rate; and are respectively the left bound values of the fuzzy membership center of the input variable at the (t + 1)-th iteration and the t-th iteration;

[0011] Dual activation mechanism design unit; designing a dual activation mechanism so that the laboratory safety management model can analyze the current state and time trend simultaneously; spatial activation represents the matching degree between the current input and the fuzzy rule through the activation intensity of the fuzzy rule, introducing variable interaction weights and time window, expressed as: ; ; ; Time activation captures the time dependence of the input data and reflects the risk trend, expressed as: ; where and are the upper and lower bound activation intensities of the matching degree between the input data at the current time t and the fuzzy rule; is the lower bound of the fuzzy membership degree; N is the total number of input variables; is the time activation intensity; is the LSTM output gate; is the LSTM cell state, representing the cumulative historical risk trend; is the element-wise product; is the hyperbolic tangent function; the fuzzy rules are used to describe the conditional result relationship of the laboratory state; τ is the time index; j and m are variable indices; is the upper bound of the fuzzy membership of the m-th input variable; and are input variables; and are the corresponding standard deviations of the variables; Cov(·) is the covariance; is the time window length;

[0012] Rule Adaptive Adjustment Unit; Dynamically generate new safety rules to ensure that the laboratory safety management model adapts to data changes. When the dynamic feature importance evaluation value is greater than the feature importance threshold, the system generates a new rule, expressed as: ; ; ; When the rule is not activated beyond the time threshold or the rule usage is lower than the usage threshold, the rule is deleted and the historical impact of increasing the time activation value is introduced; expressed as: ; ; where is the proximity of the input to the rule center ; is the current input; is the sum of the squares of the data accumulations, representing the overall distribution; is the normalized proximity, used to judge the rule generation condition; is the input data at the i-th moment; is the usage of the k-th rule; is the normalized activation value; is the rule creation time step; is the single-value risk index; is the total number of rules for the j-th variable; Var(·) is the variance; is the input of the j-th variable at time t; is the activation weight;

[0013] Risk Assessment Unit; Achieve real-time risk assessment through interval reduction and fuzzy inference; Introduce the dynamic interval reduction coefficient , generate a single-value risk index by reducing the uncertainty interval; expressed as: ; ; Through defuzzification by the weighted average method, convert the fuzzy inference result into a specific laboratory safety status evaluation level, expressed as: ; where is the evaluation level of the laboratory safety status; is the weight of the k-th fuzzy rule for the q-th output variable; is the time activation intensity of the k-th rule for the q-th output; K is the total number of rules;

[0014] Parameter optimization unit; defines the error of the laboratory safety risk, expressed as: ; introducing a momentum factor, the parameter update is expressed as: ; ; where E is the laboratory safety risk error function; and RL are the actual value and the predicted value of the laboratory safety status respectively; is the parameter weight of the k-th rule for the s-th output variable; is the weight change amount; is the momentum coefficient;

[0015] Model determination unit; divides the input data into a test set and a training set; pre-sets a maximum number of iterations and a test threshold. When the laboratory safety management model converges based on the training set loss or reaches the maximum number of training times, the training of the laboratory safety management model is completed. If the prediction accuracy rate of the trained laboratory safety management model for the test set is higher than the test threshold, the laboratory safety management model is established; otherwise, re-divide the data set and adjust the parameters to train the laboratory safety management model.

[0016] Furthermore, the laboratory safety management module is based on the established laboratory safety management model, collects laboratory management data in real time, and conducts laboratory safety management based on the laboratory safety management model.

[0017] The beneficial effects obtained by the present invention using the above solution are as follows:

[0018] (1) Aiming at the problems of the general laboratory safety management system being seriously affected by noise, unable to adapt to the changes of laboratory distribution data in real time, and unable to identify risks in complex environments, resulting in poor safety management effects, this solution introduces an interval type-2 asymmetric Gaussian membership function to adapt to the dynamic changes of the laboratory environment in real time and avoid false alarms; introduces variable interaction weights and a time activation mechanism with a time window to record the long-term risk trend and accurately identify multi-variable risks in complex environments, thereby improving the laboratory safety management effect.

[0019] (2) Aiming at the problems of poor adaptability to risk changes and inaccurate early warnings caused by noise in general laboratory safety management systems, this solution designs a rule adaptive adjustment unit that generates or removes rules in real time based on feature importance and rule usage to automatically adapt to risk changes. A dynamic interval reduction coefficient is introduced to weigh the credibility of the upper and lower bound activation intensities according to historical data, reduce the impact of uncertainty on risk assessment, and thus improve the accuracy of early warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic flow chart of the laboratory safety management system based on artificial intelligence provided by the present invention;

[0021] Figure 2 It is a schematic flow chart of the laboratory safety management model establishment module.

[0022] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc. indicating the orientation or position relationship are based on the orientation or position 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 thus cannot be understood as a limitation to the present invention.

[0025] Embodiment 1. Refer to Figure 1 The laboratory safety management system based on artificial intelligence provided by the present invention includes a data acquisition module, a laboratory safety management model establishment module, and a laboratory safety management module;

[0026] The data acquisition module collects historical laboratory management data and sends the data to the laboratory safety management model establishment module;

[0027] The laboratory safety management model establishment module establishes a laboratory safety management model based on the dynamic laboratory environment through fuzzification, dual activation mechanism, rule adaptive adjustment, risk assessment, and parameter optimization; and sends the data to the laboratory safety management module;

[0028] The laboratory safety management module conducts laboratory safety management on the laboratory management data collected in real time based on the established laboratory safety management model.

[0029] Example 2, refer to Figure 1 , this example is based on the above example. In the data collection module, the historical laboratory management data includes environmental sensor data, equipment status monitoring data, personnel behavior data, time, and laboratory safety status evaluation level; the laboratory safety status evaluation level is used as the data label; after data cleaning, data conversion, and standardization processing of the collected data, input data is obtained; the environmental sensor data includes real-time monitored temperature, humidity, vibration, and gas concentration; the equipment status monitoring data includes current, voltage, and vibration status; the personnel behavior data includes personnel location, behavior, and violation operation conditions; the laboratory safety status includes normal, slightly abnormal, moderately abnormal, and severely abnormal.

[0030] Example 3, refer to Figure 1 and Figure 2 , this example is based on the above example. The laboratory safety management model establishment module includes the following:

[0031] Fuzzification layer design unit; the fuzzification layer allows the laboratory safety management model to perform inference in an uncertain environment, improving the robustness to noise interference; using an interval type-2 asymmetric Gaussian membership function to handle the uncertainty and noise in environmental monitoring, and defining the fuzzy range of the input variable based on the upper membership function and the lower membership function, expressed as: ; dynamically adjusting the fuzzy membership center to adapt to the dynamic laboratory distribution data, expressed as: ; where, is the upper bound of the fuzzy membership degree of the jth input variable in the kth rule; t is time; is the input variable; and are the left and right bounds of the fuzzy membership center of the input variable; and are the left and right standard deviations of the fuzzy range; is the learning rate; and are respectively the left boundary values of the fuzzy membership center of the input variable at the (t + 1)th iteration and the tth iteration;

[0032] Dual activation mechanism design unit; design a dual activation mechanism to enable the laboratory safety management model to analyze the current state and time trend simultaneously to cope with the dynamic risk prediction of the laboratory. When the gas concentration gradually increases, the LSTM records the time dependence. When the equipment vibration gradually increases, the time activation value gradually increases; the spatial activation represents the matching degree between the current input and the fuzzy rule through the activation intensity of the fuzzy rule, and introduces the variable interaction weight and the time window, expressed as: ; ; ; The time activation captures the time dependence of the input data and reflects the risk trend, expressed as: ; where, and are the upper and lower bound activation intensities of the matching degree between the input data at the current time t and the fuzzy rule; is the lower bound of the fuzzy membership degree; N is the total number of input variables; is the time activation intensity; is the LSTM output gate; is the LSTM cell state, representing the cumulative historical risk trend; is the element-wise product; is the hyperbolic tangent function; the fuzzy rule is used to describe the conditional result relationship of the laboratory state; τ is the time index; j and m are variable indices; is the upper bound of the fuzzy membership degree of the m-th input variable; and are the input variables; and are the corresponding standard deviations of the variables; Cov(·) is the covariance; is the time window length.

[0033] By performing the above operations, for the problems existing in the general laboratory safety management system, such as being seriously affected by noise, unable to adapt to the changes of laboratory distribution data in real time, unable to identify risks in complex environments, and thus resulting in poor safety management effects, this solution introduces an interval type-2 asymmetric Gaussian membership function to adapt to the dynamic changes of the laboratory environment in real time and avoid false alarms; introduces variable interaction weights and a time activation mechanism with a time window to record long-term risk trends, accurately identify multi-variable risks in complex environments, and thus improve the laboratory safety management effect.

[0034] Example 4, refer to Figure 1 and Figure 2 , based on the above example, the laboratory safety management model establishment module further includes the following content:

[0035] Rule Adaptive Adjustment Unit; constructing a dynamic structure learning enables the laboratory safety management model to adjust the rule set online, reduce redundant rules, and improve the scalability of the system. When a new chemical reagent leakage is detected, the laboratory safety management model automatically generates a new gas concentration threshold rule; dynamically generates new safety rules to ensure that the laboratory safety management model adapts to data changes. When the dynamic feature importance evaluation value is greater than the feature importance threshold, the system generates a new rule, expressed as: ; ; ; When a rule is not activated beyond the time threshold or the rule usage degree is lower than the usage threshold, the rule is deleted and the historical influence of increasing the time activation value is introduced; expressed as: ; ; where, is the proximity between the input and the rule center ; is the current input; is the sum of the squares of the data accumulations, representing the overall distribution; is the normalized proximity, used to judge the rule generation condition; is the input data at the i-th moment; is the usage degree of the k-th rule; is the normalized activation value; is the rule creation time step; is the single-value risk index; is the total number of rules for the j-th variable; Var(·) is the variance; is the input of the j-th variable at time t; is the activation weight;

[0036] Risk Assessment Unit; realizes real-time risk assessment through interval reduction and fuzzy inference; introduces a dynamic interval reduction coefficient , generates a single-value risk index by reducing the uncertainty interval, reduces the impact of uncertainty on the alarm result, and improves the alarm accuracy; expressed as: ; ; Through defuzzification by the weighted average method, the fuzzy inference result is converted into a specific laboratory safety status evaluation level, expressed as: ; where, is the laboratory safety status evaluation level; is the weight of the k-th fuzzy rule for the q-th output variable; is the time activation intensity of the k-th rule for the q-th output; K is the total number of rules;

[0037] Parameter Optimization Unit; defines the error of the laboratory safety risk, expressed as: ; Introduces a momentum factor, and the parameter update is expressed as: ; ; where E is the laboratory safety risk error function; and RL are the actual value and predicted value of the laboratory safety status respectively; is the parameter weight of the k-th rule for the s-th output variable; is the weight change amount; is the momentum coefficient;

[0038] Model determination unit: divides the input data into a test set and a training set; there are preset maximum iteration times and test thresholds. When the laboratory safety management model converges based on the training set loss or reaches the maximum training times, the training of the laboratory safety management model is completed. If the prediction accuracy rate of the trained laboratory safety management model for the test set is higher than the test threshold, the laboratory safety management model is established; otherwise, re-divide the data set and adjust the parameters to train the laboratory safety management model.

[0039] By performing the above operations, aiming at the problems of poor adaptability to risk changes and inaccurate early warnings caused by noise in general laboratory safety management systems, this solution designs a rule adaptive adjustment unit to generate or remove rules in real time based on feature importance and rule usage, and automatically adapt to risk changes; introduces a dynamic interval reduction coefficient to balance the credibility of the activation intensity of the upper and lower bounds according to historical data, reduce the impact of uncertainty on risk assessment, and thus improve the accuracy of early warnings.

[0040] Example 5, refer to Figure 1 , this example is based on the above example. The laboratory safety management module is based on the established laboratory safety management model, collects laboratory management data in real time, and conducts laboratory safety management based on the laboratory safety management model. When the risk assessment unit outputs a laboratory safety status assessment level of moderate abnormality and severe abnormality for the real-time collected laboratory management data, early warning processing is performed.

[0041] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0042] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

[0043] The present invention and its embodiments have been described above. Such description is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. In summary, if those of ordinary skill in the art are inspired by it and, without departing from the purpose of the present invention, design similar structural modes and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of the present invention.

Claims

1. Laboratory safety management system based on artificial intelligence, characterized by: The system includes a data acquisition module, a laboratory safety management model building module and a laboratory safety management module; The data acquisition module collects historical laboratory management data; The laboratory safety management model establishment module realizes the establishment of a laboratory safety management model based on the laboratory dynamic environment through fuzzification, dual activation mechanism, rule adaptive adjustment, risk assessment and parameter optimization; The laboratory safety management module performs laboratory safety management on the laboratory management data collected in real time based on the established laboratory safety management model; The laboratory safety management model establishment module specifically includes the following contents: Fuzzy layer design unit; interval type II asymmetric Gaussian membership function is used to deal with uncertainty and noise in environmental monitoring. The fuzzy range of the input variable is defined based on the upper membership function and the lower membership function. The upper membership function is expressed as: ; Dynamically adjust the fuzzy membership center to adapt to the dynamic laboratory distribution data, expressed as: ;in, is the upper bound of the fuzzy membership of the jth input variable in the kth rule; t is the time; is the input variable; and are the left and right bounds of the fuzzy membership center of the input variable; and is the left and right standard deviation of the fuzzy range; is the learning rate; and are the left bounds of the fuzzy membership center of the input variable at the t+1th iteration and the tth iteration respectively; Dual activation mechanism design unit; designing a dual activation mechanism enables the laboratory safety management model to analyze the current state and time trend simultaneously; spatial activation is to express the degree of match between the current input and the fuzzy rule through the activation intensity of the fuzzy rule, and introduce variable interaction weights and time window, expressed as: ; ; ; Temporal activation captures the temporal dependency of input data and reflects the risk trend, expressed as: ;in, and is the activation strength of the upper and lower bounds of the matching degree between the input data at the current time t and the fuzzy rule; is the lower bound of fuzzy membership; N is the total number of input variables; is the temporal activation intensity; is the LSTM output gate; is the LSTM unit state, representing the accumulated historical risk trend; is the element-wise product; is the hyperbolic tangent function; fuzzy rules are used to describe the condition-result relationship of laboratory status; τ is the time index; j and m are variable indexes; is the upper bound of the fuzzy membership of the mth input variable; and is the input variable; and is the standard deviation of the variable; Cov(·) is the covariance; is the time window length; Rule adaptive adjustment unit; Risk Assessment Unit; Parameter optimization unit; Model judgment unit; divide the input data into a test set and a training set; pre-set the maximum number of iterations and the test threshold. When the laboratory safety management model converges based on the training set loss or reaches the maximum number of training times, the laboratory safety management model training is completed. If the prediction accuracy of the trained laboratory safety management model for the test set is higher than the test threshold, the laboratory safety management model is established; otherwise, the data set is re-divided and the parameters are adjusted to train the laboratory safety management model.

2. The laboratory safety management system based on artificial intelligence according to claim 1 is characterized in that: The rule adaptive adjustment unit dynamically generates new safety rules to ensure that the laboratory safety management model adapts to data changes. When it is greater than the feature importance threshold, the system generates a new rule, expressed as: ; ; ; When a rule exceeds the time threshold and is not activated or the rule usage is lower than the usage threshold, the rule is deleted and the historical impact of increasing the time activation value is introduced; expressed as: ; ;in, It is the input and rule center proximity; is the current input; is the cumulative sum of squares of the data, representing the overall distribution; is the normalized proximity, which is used to determine the rule generation conditions; is the input data at the i-th moment; is the usage degree of the kth rule; is the normalized activation value; is the rule creation time step; It is a single-value risk indicator; is the total number of rules for the jth variable; Var(·) is the variance; is the input of the jth variable at time t; is the activation weight.

3. The laboratory safety management system based on artificial intelligence according to claim 2 is characterized in that: The risk assessment unit realizes real-time risk assessment through interval reduction and fuzzy reasoning; introduces dynamic interval reduction coefficient , by reducing the uncertainty interval, a single-value risk indicator is generated; it is expressed as: ; ; The fuzzy reasoning results are converted into specific laboratory safety status assessment levels through weighted average method, which is expressed as: ;in, It is the laboratory safety status assessment level; is the weight of the kth fuzzy rule on the qth output variable; K is the total number of rules; is the temporal activation strength of the k-th rule on the q-th output.

4. The laboratory safety management system based on artificial intelligence according to claim 3 is characterized in that: The parameter optimization unit is the error that defines the laboratory safety risk, expressed as: ; Introducing the momentum factor, the parameter update is expressed as: ; ; Among them, E is the laboratory safety risk error function; and RL are the actual and predicted values ​​of laboratory safety status, respectively; is the parameter weight of the k-th rule on the s-th output variable; is the weight change; is the momentum coefficient.

5. The laboratory safety management system based on artificial intelligence according to claim 4 is characterized in that: In the data acquisition module, the historical laboratory management data includes environmental sensor data, equipment status monitoring data, personnel behavior data, time and laboratory safety status assessment level; the laboratory safety status assessment level is used as a data label; the collected data is cleaned, converted and standardized to obtain input data.

Images