Student laboratory chemical operation control method and system

Abstract

The invention relates to the technical field of chemical operation control, and provides a student laboratory chemical operation control method and system, and the method comprises the following steps: obtaining real-time image data and environment parameter data; identifying and numbering the identities of the students; identifying the real-time image data by using a chemical operation identification model to generate real-time operation state information; identifying the real-time operation state information based on a multi-dimensional risk assessment model to obtain a chemical operation risk level of each student; chemical operation risk levels are identified through the decision control model, different early warning signals are started, and safety control measures are executed according to the early warning signals; and extracting operation characteristics of each student through a behavior analysis model, generating a student operation characteristic vector, updating the student operation behavior characteristic library, and generating an experiment operation score and an operation improvement suggestion corresponding to each student. According to the invention, the safety management level of the laboratory is improved, and management and control of chemical operation of students are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical operation control, and in particular to a method and system for chemical operation control in a student laboratory. Background Art

[0002] As an important part of laboratory safety management, the safety management of chemical operations in student laboratories is directly related to the quality of experimental teaching and the personal safety of students. There are many safety hazards in the process of chemical operations, such as improper operation, inadequate protection, and abnormal environment. If these problems are not discovered and handled in time, they may lead to safety accidents.

[0003] Existing technologies generally use a single video surveillance or sensor detection method to manage the safety of student laboratories. Although a single detection method can obtain certain types of data information, it cannot fully reflect the safety status of students' chemical operations. At the same time, existing technologies lack systematic risk assessment and intelligent early warning mechanisms, and often use a simple manual inspection mode, resulting in low management efficiency. After the data collection process, there is insufficient analysis of the information retained by students' experimental operations, and no personalized chemical operation suggestions are provided for students, which cannot achieve accurate analysis and personalized guidance of students' individual chemical operation behaviors. Summary of the invention

[0004] In view of this, the present invention proposes a method and system for controlling chemical operations in student laboratories, which solves the problem that the prior art lacks analysis of the information retained by students' experimental operations after the data collection process, does not provide personalized chemical operation suggestions for students, and cannot achieve accurate analysis and personalized guidance of students' individual chemical operation behaviors.

[0005] The technical solution of the present invention is implemented as follows: In the first aspect, the present invention provides a method for controlling chemical operation in a student laboratory, comprising the following steps:

[0006] Through monitoring cameras and sensor networks, real-time image data and environmental parameter data in the laboratory are obtained respectively;

[0007] Identify and number students entering the laboratory;

[0008] The real-time image data is identified by using a chemical operation identification model to obtain the student's operation behavior information and protective equipment wearing information, and real-time operation status information is generated by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information;

[0009] Identify the real-time operation status information based on a multi-dimensional risk assessment model to obtain the chemical operation risk level of each student;

[0010] Identify the chemical operation risk level through a decision control model, activate a corresponding early warning signal when the chemical operation risk level reaches different preset thresholds, and execute safety control measures according to the early warning signal;

[0011] Establishing a student operation behavior feature library, storing each student's number, real-time operation status information and environmental parameter data in the student operation behavior feature library, extracting each student's operation feature through a behavior analysis model, generating a student operation feature vector, and updating the student operation behavior feature library based on the student operation feature vector;

[0012] Based on the student operation behavior feature library, generate experimental operation scores and operation improvement suggestions for each student.

[0013] On the basis of the above technical solution, preferably, the real-time image data and environmental parameter data in the laboratory are obtained through monitoring cameras and sensor networks, specifically including:

[0014] A dual-camera surveillance camera is set up at each laboratory bench in the laboratory, wherein the dual-camera surveillance camera includes a wide-angle surveillance camera and a close-up surveillance camera. The wide-angle surveillance camera is used to capture the overall image of the laboratory bench, and the close-up camera is used to capture the detailed image of the chemical operation area;

[0015] A sensor network is arranged in the chemical operation area to collect environmental parameter data. The sensor network includes a chemical concentration sensor, a weight sensor, a pressure sensor and a temperature sensor. The chemical concentration sensor is used to monitor the concentration of chemicals in the air, the weight sensor is used to monitor the amount of chemicals used, the pressure sensor is used to monitor the operation force, and the temperature sensor is used to monitor the change of reaction temperature. The chemical operation area includes a laboratory bench, a reagent cabinet and a fume hood.

[0016] On the basis of the above technical solution, preferably, the real-time image data is identified by using a chemical operation recognition model to obtain the student's operation behavior information and protective equipment wearing information, and real-time operation status information is generated by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information, specifically including:

[0017] A chemical operation recognition model is constructed through a deep learning model, and real-time image data is recognized based on the chemical operation recognition model to generate operation behavior information, wherein each feature vector element in the operation behavior information represents the recognition probability of a specific operation behavior;

[0018] The calculation formula of the chemical operation identification model is:

[0019]

[0020] α(t)=1-λ·|P(A i |I t )-P(A i I t-1 )|;

[0021]

[0022] Among them, P(A i |I) is the probability of judging the i-th type of operation behavior under a given image I, CNN(I) is the image feature of image I extracted by convolutional neural network, exp(·) is the exponential function, and f(CNN(I)) i is the normalized probability of the i-th type of operation behavior, f(CNN(I)) j is the normalized probability of the jth type of operation behavior, N is the number of categories of operation behavior, α(t) is the time series adjustment factor, β(c) is the environmental constraint factor, λ is the time series smoothing coefficient, P(A i |I t ) is a given image I t The probability of judging it as the i-th type of operation behavior, P(A i |I t-1 ) is a given image I t-1 The probability of being judged as the i-th type of operation behavior, I t and I t-1 are the images at time t and time t-1, c k is the normalized value of the kth environmental parameter, γ k c k The corresponding influence coefficient, M is the number of environmental parameters;

[0023] The wearing status of protective equipment is detected through image processing algorithms, and protective equipment wearing information is generated. In combination with environmental parameter data, real-time operation status information is generated through weighted algorithms;

[0024] The calculation formula of the real-time operation status information is:

[0025]

[0026] Among them, Z is the normalization factor, S is the real-time operation status information, is the i1th element of the operation behavior information, is the weight coefficient of the i1th element of the operation behavior information, L1 is the number of elements of the operation behavior information, is the i2th element of the protective equipment wearing information, is the weight coefficient of the i2th element of the protective equipment wearing information, L2 is the number of elements of the protective equipment wearing information, is the i3th element of the environmental parameter data, is the weight coefficient of the i3th element of the environmental parameter data, and L3 is the number of elements of the environmental parameter data.

[0027] On the basis of the above technical solution, preferably, the real-time operation status information is identified based on the multi-dimensional risk assessment model to obtain the chemical operation risk level of each student, specifically including:

[0028] Dynamically extract features of real-time operation status information through a spatiotemporal attention mechanism to construct a risk feature vector, wherein the risk feature vector includes an operation behavior risk feature sub-vector, a protective equipment risk feature sub-vector, and an environmental risk feature sub-vector;

[0029] A multi-level risk assessment model is established using the analytic hierarchy process. The comprehensive impact of each risk feature is calculated through a combined weight algorithm. Combined with the risk event records in the historical database, the risk assessment parameters are dynamically adjusted to output the chemical operation risk level.

[0030] On the basis of the above technical solution, preferably, the chemical operation risk level is identified by the decision control model, and when the chemical operation risk level reaches different preset thresholds, a corresponding early warning signal is activated, and safety control measures are executed according to the early warning signal, specifically including:

[0031] A deep Q-learning network is used to build an early warning decision model, which takes the chemical operation risk level, environmental status and historical early warning records as input, and generates an early warning decision vector through state-action value evaluation;

[0032] The calculation formula of the early warning decision vector is:

[0033]

[0034] Among them, P is the warning decision vector, ω1, ω2, and ω3 are the weight coefficients of the state-action value function, the time series correlation function, and the environmental sensitivity function, respectively. is the Q value of the i6th decision state, is the immediate reward of the i6th decision state, L6 is the number of decision states, μ is the temperature parameter, σ(·) is the sigmoid activation function, is the impact factor of the i7th historical warning event, L7 is the time length of the historical warning event, θ is the state-action time attenuation coefficient, T0 is the current time, For the historical moment, is the value of the i8th environment parameter, is the safety threshold of the i8th environmental parameter, tanh(·) is the hyperbolic tangent activation function, is the sensitivity coefficient of the i8th environmental parameter, L8 is the number of environmental parameters;

[0035] A multi-level early warning execution mechanism is established based on fuzzy control theory, which triggers early warning signals of different levels according to the threshold of the early warning decision vector, and executes the safety control measures corresponding to the early warning level of the early warning signal;

[0036] The rules for determining the warning level include: when |P|<0.3, the warning signal corresponding to the low-level warning is triggered, and an sound and light alarm is given; when 0.3≤|P|<0.7, the warning signal corresponding to the intermediate warning is triggered, and ventilation adjustment is started; when |P|≥0.7, the warning signal corresponding to the high-level warning is triggered, and emergency response measures are implemented.

[0037] On the basis of the above technical solution, preferably, the student operation behavior feature library is established, the number, real-time operation status information and environmental parameter data of each student are stored in the student operation behavior feature library, the operation characteristics of each student are extracted through the behavior analysis model, the student operation feature vector is generated, and the student operation behavior feature library is updated based on the student operation feature vector, which specifically includes:

[0038] Establish a student operation behavior feature library, index the real-time operation status information and environmental parameter data by timestamp and student number, and store them in the student operation behavior feature library, use the behavior analysis model to process the stored data, identify the student's operation mode and behavior characteristics, generate student operation feature vectors, and update the student operation behavior feature library according to the change trend of the student operation feature vectors.

[0039] On the basis of the above technical solution, preferably, the generation of the experimental operation score and operation improvement suggestions corresponding to each student based on the student operation behavior feature library specifically includes:

[0040] A multi-layer scoring mechanism is used to comprehensively evaluate students from multiple dimensions, including operational standardization, safety awareness, and emergency response capabilities. The experimental operation score of each student is calculated through a weighted fusion algorithm, and a student operation ability development curve is generated based on historical data analysis.

[0041] Construct a chemical operation knowledge graph, which includes standard operating procedures, safety specifications and expert experience. Combined with students' personal characteristics and operating habits, targeted improvement suggestions are generated through rule reasoning and similarity matching algorithms, and the improvement suggestions are dynamically optimized based on student feedback.

[0042] In a second aspect, the present invention further provides a student laboratory chemical operation control system, the system comprising:

[0043] A data acquisition module is used to acquire real-time image data and environmental parameter data in the laboratory through monitoring cameras and sensor networks;

[0044] The identification and numbering module is used to identify and number students entering the laboratory;

[0045] An image recognition module is used to recognize the real-time image data using a chemical operation recognition model to obtain the student's operation behavior information and protective equipment wearing information, and generate real-time operation status information by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information;

[0046] A risk level module, used to identify the real-time operation status information based on a multi-dimensional risk assessment model to obtain the chemical operation risk level of each student;

[0047] An operation warning module, used to identify the risk level of chemical operation through a decision control model, and when the risk level of chemical operation reaches different preset thresholds, activate the corresponding warning signal and execute safety control measures according to the warning signal;

[0048] An operation analysis module is used to establish a student operation behavior feature library, store each student's number, real-time operation status information and environmental parameter data in the student operation behavior feature library, extract each student's operation feature through a behavior analysis model, generate a student operation feature vector, and update the student operation behavior feature library based on the student operation feature vector;

[0049] The operation improvement module is used to generate experimental operation scores and operation improvement suggestions for each student based on the student operation behavior feature library.

[0050] In a third aspect, the present invention further provides an electronic device, comprising: at least one processor, at least one memory, a communication interface and a bus;

[0051] The processor, memory, and communication interface communicate with each other via the bus, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement steps of a student laboratory chemical operation control method.

[0052] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions enable a computer to implement steps such as a method for controlling chemical operations in a student laboratory.

[0053] The student laboratory chemical operation control method and system of the present invention has the following beneficial effects compared with the prior art:

[0054] (1) By integrating surveillance cameras and sensor networks, combined with chemical operation recognition models and multi-dimensional risk assessment models, the system can identify students' operation behaviors and protective equipment wearing conditions, generate real-time operation status information, and evaluate the risk level of chemical operations. Based on database storage and behavior analysis models, it can extract students' operation characteristics, establish an operation behavior characteristic library, and then generate personalized experimental operation scores and improvement suggestions. It realizes real-time collection and monitoring of images and environmental data in the laboratory, improves the safety management level and efficiency of the laboratory, and realizes all-round and intelligent control of students' chemical operations.

[0055] (2) A chemical operation recognition model was constructed through a deep learning model to identify students’ operation behavior information and generate the recognition probability of each feature vector element, which improved the accuracy of operation behavior recognition and ensured the safety of laboratory operations. The image processing algorithm was combined to detect the wearing of protective equipment, further enhancing the reliability of laboratory safety management.

[0056] (3) A deep Q-learning network is used to construct an early warning decision model, and an early warning decision vector is generated through state-action value evaluation. A multi-level early warning execution mechanism is established based on fuzzy control theory to dynamically identify the risk level of chemical operations. When the risk reaches different preset thresholds, the corresponding early warning signals and safety control measures are triggered, which improves the accuracy and response speed of the early warning, ensures the effectiveness of laboratory operations, and can effectively reduce the risk of laboratory accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 A flow chart of a method for controlling chemical operations in student laboratories according to the present invention;

[0059] Figure 2 This is a structural diagram of a student laboratory chemical operation control system of the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0061] See also Figure 1 The present invention provides a method for controlling chemical operation in a student laboratory, comprising the following steps:

[0062] Through monitoring cameras and sensor networks, real-time image data and environmental parameter data in the laboratory are obtained respectively;

[0063] Identify and number students entering the laboratory;

[0064] The real-time image data is identified by using a chemical operation identification model to obtain the student's operation behavior information and protective equipment wearing information, and real-time operation status information is generated by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information;

[0065] Identify the real-time operation status information based on a multi-dimensional risk assessment model to obtain the chemical operation risk level of each student;

[0066] Identify the chemical operation risk level through a decision control model, activate a corresponding early warning signal when the chemical operation risk level reaches different preset thresholds, and execute safety control measures according to the early warning signal;

[0067] Establishing a student operation behavior feature library, storing each student's number, real-time operation status information and environmental parameter data in the student operation behavior feature library, extracting each student's operation feature through a behavior analysis model, generating a student operation feature vector, and updating the student operation behavior feature library based on the student operation feature vector;

[0068] Based on the student operation behavior feature library, generate experimental operation scores and operation improvement suggestions for each student.

[0069] Specifically, this embodiment integrates surveillance cameras and sensor networks, combines chemical operation recognition models and multi-dimensional risk assessment models, identifies students' operating behaviors and the wearing of protective equipment, generates real-time operating status information, and evaluates the risk level of chemical operations. Based on database storage and behavior analysis models, it extracts student operating characteristics, establishes an operating behavior feature library, and then generates personalized experimental operation scores and improvement suggestions, realizing real-time collection and monitoring of images and environmental data in the laboratory, improving the safety management level and efficiency of the laboratory, and realizing all-round and intelligent control of students' chemical operations.

[0070] The real-time image data and environmental parameter data in the laboratory are obtained through monitoring cameras and sensor networks, including:

[0071] A dual-camera surveillance camera is set up at each laboratory bench in the laboratory. The dual-camera surveillance camera includes a wide-angle surveillance camera and a close-up surveillance camera. The wide-angle surveillance camera is used to capture the overall image of the laboratory bench, and the close-up camera is used to capture the detailed image of the chemical operation area. The dual-camera surveillance camera has an infrared night vision function to ensure the image acquisition quality when the light is insufficient;

[0072] A sensor network is arranged in the chemical operation area to collect environmental parameter data. The sensor network includes a chemical concentration sensor, a weight sensor, a pressure sensor and a temperature sensor. The chemical concentration sensor is used to monitor the concentration of chemicals in the air, the weight sensor is used to monitor the amount of chemicals used, the pressure sensor is used to monitor the operation force, and the temperature sensor is used to monitor the change of reaction temperature. The chemical operation area includes a laboratory bench, a reagent cabinet and a fume hood.

[0073] Specifically, this embodiment realizes the unification of macro-overall monitoring and micro-detail monitoring of experimental operations by configuring dual-position monitoring cameras, thereby improving the comprehensiveness of monitoring. The infrared night vision function ensures continuous monitoring capabilities under low-light conditions and ensures the all-weather availability of the monitoring system.

[0074] Through the coordinated cooperation of multiple sensors, all-round monitoring of key parameters in the chemical operation process is achieved. Air safety is monitored through chemical concentration sensors, chemical usage is accurately controlled through weight sensors, operation compliance is monitored through pressure sensors, and dangerous reactions are prevented through temperature sensors.

[0075] Comprehensive monitoring of key areas such as laboratory benches, reagent cabinets and fume hoods ensures safe management of the entire process of chemical use, forms a complete chemical operation safety monitoring network, and avoids blind spots in monitoring.

[0076] Step S2 specifically includes:

[0077] A facial recognition camera and card swiping device are installed at the entrance of the laboratory, and a dual authentication mechanism of "facial recognition + student ID card swiping" is adopted to establish a laboratory access personnel database to store student photos, identity information, etc.

[0078] A multi-level coding method of "year + major + class + student number" is adopted to assign a unique operation serial number to each experimental operation, and the student number is associated with the experimental project number to form a complete experimental record identification.

[0079] Match student identity information with experimental arrangements, establish a correspondence between students, experimental projects, and time periods, and update students' entry and exit records in the laboratory in real time.

[0080] The real-time image data is identified by using a chemical operation identification model to obtain the student's operation behavior information and protective equipment wearing information, and real-time operation status information is generated by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information, specifically including:

[0081] A chemical operation recognition model is constructed through a deep learning model, and real-time image data is recognized based on the chemical operation recognition model to generate operation behavior information, wherein each feature vector element in the operation behavior information represents the recognition probability of a specific operation behavior;

[0082] The calculation formula of the chemical operation identification model is:

[0083]

[0084] α(t)=1-λ·|P(A i |I t )-P(A i I t-1 )|;

[0085]

[0086] Among them, P(A i |I) is the probability of judging the i-th type of operation behavior under a given image I, CNN(I) is the image feature of image I extracted by convolutional neural network, exp(·) is the exponential function, and f(CNN(I)) i is the normalized probability of the i-th type of operation behavior, f(CNN(I)) j is the normalized probability of the jth type of operation behavior, N is the number of categories of operation behavior, α(t) is the time series adjustment factor, β(c) is the environmental constraint factor, λ is the time series smoothing coefficient, P(A i |I t ) is a given image I tThe probability of judging it as the i-th type of operation behavior, P(A i |I t-1 ) is a given image I t-1 The probability of being judged as the i-th type of operation behavior, I t and I t-1 are the images at time t and time t-1, c k is the normalized value of the kth environmental parameter, γ k c k The corresponding influence coefficient, M is the number of environmental parameters;

[0087] The wearing status of protective equipment is detected through image processing algorithms, and protective equipment wearing information is generated. In combination with environmental parameter data, real-time operation status information is generated through weighted algorithms;

[0088] The calculation formula of the real-time operation status information is:

[0089]

[0090] Among them, Z is the normalization factor, S is the real-time operation status information, is the i1th element of the operation behavior information, is the weight coefficient of the i1th element of the operation behavior information, L1 is the number of elements of the operation behavior information, is the i2th element of the protective equipment wearing information, is the weight coefficient of the i2th element of the protective equipment wearing information, L2 is the number of elements of the protective equipment wearing information, is the i3th element of the environmental parameter data, is the weight coefficient of the i3th element of the environmental parameter data, and L3 is the number of elements of the environmental parameter data.

[0091] In a specific embodiment, the process of constructing the chemical operation identification model is as follows:

[0092] ResNet-50 is adopted as the basic convolutional neural network architecture, and a labeled chemical operation video dataset is used for model training. A classification system containing 20 types of standard operation behaviors is established. Image features are extracted in real time and the probability of operation behaviors is calculated. A timing adjustment mechanism is introduced to eliminate short-term fluctuations, and constraint optimization is performed in combination with environmental parameters.

[0093] The specific steps for implementing protective equipment testing are:

[0094] The YOLO target detection algorithm is used to identify protective equipment, a protective equipment detection list (such as goggles, gloves, masks, etc.) is set, the wearing position and status of protective equipment are detected in real time, and false detection results are filtered out through confidence thresholds.

[0095] The mechanisms for generating real-time operational status information include:

[0096] Establish a three-layer weighted fusion architecture, design an adaptive weight adjustment algorithm, achieve millisecond-level status information updates, and establish a data caching mechanism to improve processing efficiency.

[0097] Specifically, this embodiment achieves an operation behavior recognition accuracy of over 95%, a protective equipment detection accuracy of over 98%, effectively reduces false alarm and missed alarm rates, and supports stable recognition in complex scenarios; it achieves millisecond-level recognition response speed, smooth video stream processing capabilities, fast status information updates, and low-latency system feedback; it can adapt to different lighting conditions, has strong anti-interference capabilities, stable long-term operating performance, and good environmental adaptability.

[0098] The real-time operation status information is identified based on the multi-dimensional risk assessment model to obtain the chemical operation risk level of each student, specifically including:

[0099] Dynamically extract features of real-time operation status information through a spatiotemporal attention mechanism to construct a risk feature vector, which includes an operation behavior risk feature sub-vector, a protective equipment risk feature sub-vector, and an environmental risk feature sub-vector. The elements of each sub-vector represent the risk contribution of the corresponding dimension.

[0100] A multi-level risk assessment model is established using the analytic hierarchy process. The comprehensive impact of each risk feature is calculated through a combined weight algorithm. Combined with the risk event records in the historical database, the risk assessment parameters are dynamically adjusted to output the chemical operation risk level.

[0101] The calculation formula for the chemical operation risk level is:

[0102]

[0103] Among them, R is the risk level of chemical operation, is the i4th element of the risk feature vector, is the weight coefficient of the i4th element of the risk feature vector, F is the risk feature vector, max(F) is the maximum value of the risk feature vector, L4 is the number of elements of the risk feature vector, α0, β0, and γ0 are the risk level balance factor, time decay balance factor, and environmental dynamic balance factor, respectively. is the risk level record at the historical time t0, T is the length of the historical data, t1 is the current time, λ0 is the time attenuation coefficient, D is the environmental dynamic factor, is the value of the i5th environmental parameter, is the safety threshold of the i5th environmental parameter, is the influence coefficient of the i5th environmental parameter, and L5 is the number of environmental parameters.

[0104] Specifically, this embodiment realizes dynamic feature extraction of operation status information through the spatiotemporal attention mechanism, decomposes risk characteristics into three dimensions: operation behavior, protective equipment, and environment, improves the accuracy of risk assessment, realizes quantitative representation of risk contribution, and facilitates accurate measurement of risk level.

[0105] The multi-level risk assessment model provides a more comprehensive risk analysis capability. The combined weight algorithm ensures the reasonable weight distribution of each risk factor. The dynamic adjustment mechanism of historical data improves the accuracy of risk assessment. The introduction of balance factors improves the stability of the assessment results, while supporting the rapid processing of large-scale data and real-time response.

[0106] Adaptive adjustment of risk assessment parameters is achieved through the historical database, the sensitivity of assessment results to environmental changes is ensured through environmental dynamic factors, and the time decay mechanism is used to ensure the reasonable decay of the impact of historical data.

[0107] The step of identifying the risk level of chemical operation through the decision control model, activating the corresponding early warning signal when the risk level of chemical operation reaches different preset thresholds, and executing safety control measures according to the early warning signal specifically includes:

[0108] A deep Q-learning network is used to build an early warning decision model, which takes the chemical operation risk level, environmental status and historical early warning records as input, and generates an early warning decision vector through state-action value evaluation;

[0109] The calculation formula of the early warning decision vector is:

[0110]

[0111] Among them, P is the warning decision vector, ω1, ω2, and ω3 are the weight coefficients of the state-action value function, the time series correlation function, and the environmental sensitivity function, respectively. is the Q value of the i6th decision state, is the immediate reward of the i6th decision state, L6 is the number of decision states, μ is the temperature parameter, σ(·) is the sigmoid activation function, is the impact factor of the i7th historical warning event, L7 is the time length of the historical warning event, θ is the state-action time attenuation coefficient, T0 is the current time, For the historical moment, is the value of the i8th environment parameter, is the safety threshold of the i8th environmental parameter, tanh(·) is the hyperbolic tangent activation function, is the sensitivity coefficient of the i8th environmental parameter, L8 is the number of environmental parameters;

[0112] A multi-level early warning execution mechanism is established based on fuzzy control theory, which triggers early warning signals of different levels according to the threshold of the early warning decision vector, and executes the safety control measures corresponding to the early warning level of the early warning signal;

[0113] The rules for determining the warning level include: when |P|<0.3, the warning signal corresponding to the low-level warning is triggered, and an sound and light alarm is given; when 0.3≤|P|<0.7, the warning signal corresponding to the intermediate warning is triggered, and ventilation adjustment is started; when |P|≥0.7, the warning signal corresponding to the high-level warning is triggered, and emergency response measures are implemented.

[0114] In a specific embodiment, the implementation steps of using a deep Q learning network to build an early warning decision model include:

[0115] The input layer receives risk levels (floating point numbers between 0 and 1), environmental conditions (sensor data such as temperature and humidity), and historical warning data (warning records for the past 30 days). The hidden layer uses a three-layer neural network, each layer contains 128 neurons, and the activation function uses ReLU. The output layer generates a warning decision vector with an output dimension of 3, corresponding to low, medium, and high warning levels.

[0116] The warning decision vector is implemented through the following steps:

[0117] The state-action value evaluation uses the Q-value update formula, with the learning rate set to 0.01 and the discount factor set to 0.9; the time series correlation analysis considers the impact of the past five warning events, and the time attenuation coefficient is set to 0.8; the environmental sensitivity calculation adjusts the weight according to the rate of change of environmental parameters, and the sensitivity coefficient ranges from 0.1 to 0.5; the weight coefficient is dynamically adjusted according to real-time data, and the initial weight is set to 0.3.

[0118] The multi-level early warning execution mechanism includes:

[0119] For low-level warnings (|P|<0.3): the sound and light warning system is triggered, the alarm volume is set to 70 decibels, and a reminder message is sent to the laboratory management system.

[0120] For intermediate warning (0.3≤|P|<0.7): start the ventilation system, set the wind speed to 2 meters per second, send a warning message to the laboratory director, and record the warning event in the database for subsequent analysis.

[0121] For high-level warnings (|P|≥0.7): Execute emergency measures, including shutting down experimental equipment, evacuating personnel to a safe area within 2 minutes, shutting down related equipment, and ensuring laboratory safety.

[0122] Specifically, this embodiment realizes intelligent early warning decision-making through deep Q learning, with an accuracy rate of more than 95%. Multi-dimensional input improves the accuracy of early warning judgment and reduces false positives and missed positives. Through millisecond-level early warning decision generation, rapid response is ensured, rapid early warning signal triggering is achieved, the average response time is less than 1 second, and safety measures are implemented in a timely manner to ensure laboratory safety.

[0123] The early warning decision model can continuously optimize the early warning strategy through learning, adapt to changes in different experimental environments, dynamically adjust decision parameters according to historical early warning records, improve system flexibility, achieve accurate division of early warning levels, avoid excessive intervention and waste of resources, and improve management efficiency.

[0124] The step of establishing a student operation behavior feature library, storing each student's ID, real-time operation status information, and environmental parameter data in the student operation behavior feature library, extracting each student's operation feature through a behavior analysis model, generating a student operation feature vector, and updating the student operation behavior feature library based on the student operation feature vector, specifically includes:

[0125] Establish a student operation behavior feature library, index the real-time operation status information and environmental parameter data by timestamp and student number, and store them in the student operation behavior feature library to ensure efficient storage and fast retrieval of data, support incremental update of real-time data and batch processing of historical data, use the behavior analysis model to process the stored data, identify the student's operation mode and behavior characteristics, generate student operation feature vectors, and update the student operation behavior feature library according to the change trend of the student operation feature vectors;

[0126] The calculation formula of the student operation feature vector is:

[0127]

[0128] in, is the i9th element of the operation behavior feature, is the weight coefficient of the i9th element of the operation behavior feature, max(X) is the maximum value of the operation behavior feature vector X, L9 is the number of dimensions of the operation behavior feature, is the i-th environmental parameter characteristic 10 elements, is the i-th environmental parameter characteristic 10 The weight coefficient of each element, max(Y) is the maximum value of the environmental parameter feature vector Y, L 10 is the number of dimensions of the environmental parameter characteristics.

[0129] Specifically, this embodiment realizes the systematic management of student operation data by establishing a student operation behavior feature library, indexes the data by timestamp and student number, ensures efficient storage and fast retrieval of data, supports incremental update of real-time data and batch processing of historical data, and improves the flexibility and efficiency of data management.

[0130] By processing the stored data using the behavior analysis model, students' operation modes and behavior characteristics can be identified, which helps to discover students' habitual behaviors and potential problems in experimental operations. The generated student operation feature vector provides data support for subsequent personalized teaching and improvement suggestions. By analyzing the changing trend of students' operation feature vectors and dynamically updating the student operation behavior feature library, it helps to timely reflect the changes in students' learning and operation abilities, provide personalized operation improvement suggestions for each student, and improve the pertinence of teaching. Through continuous monitoring and analysis of students' operation behaviors, irregular operations or potential safety hazards can be identified in a timely manner, thereby improving the overall safety management level of the laboratory.

[0131] The experimental operation score and operation improvement suggestions corresponding to each student are generated based on the student operation behavior feature library, specifically including:

[0132] A multi-layer scoring mechanism is used to comprehensively evaluate students from multiple dimensions, including operational standardization, safety awareness, and emergency response capabilities. The experimental operation score of each student is calculated through a weighted fusion algorithm, and a student operation ability development curve is generated based on historical data analysis.

[0133] Construct a chemical operation knowledge graph, which includes standard operating procedures, safety specifications and expert experience. Combined with students' personal characteristics and operating habits, targeted improvement suggestions are generated through rule reasoning and similarity matching algorithms, and the improvement suggestions are dynamically optimized based on student feedback.

[0134] Specifically, this embodiment uses a multi-layer scoring mechanism to comprehensively evaluate students from multiple dimensions such as operational standardization, safety awareness, and emergency response capabilities, accurately reflecting students' comprehensive operational capabilities. A weighted fusion algorithm is used to ensure the reasonable weight distribution of each evaluation dimension, making the scoring results more objective. By constructing a knowledge graph for chemical operations, combined with standard operating procedures, safety specifications, and expert experience, a wealth of knowledge resources are provided for students. Through rule reasoning and similarity matching algorithms, personalized improvement suggestions are generated based on students' personal characteristics and operating habits, thereby improving the pertinence of teaching. Dynamically optimize improvement suggestions based on student feedback to ensure the practicality and timeliness of the suggestions, which helps to continuously improve students' operational capabilities and safety awareness.

[0135] See also Figure 2The present invention also provides a student laboratory chemical operation control system, the system comprising:

[0136] A data acquisition module is used to acquire real-time image data and environmental parameter data in the laboratory through monitoring cameras and sensor networks;

[0137] The identification and numbering module is used to identify and number students entering the laboratory;

[0138] An image recognition module is used to recognize the real-time image data using a chemical operation recognition model to obtain the student's operation behavior information and protective equipment wearing information, and generate real-time operation status information by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information;

[0139] A risk level module, used to identify the real-time operation status information based on a multi-dimensional risk assessment model to obtain the chemical operation risk level of each student;

[0140] An operation warning module, used to identify the risk level of chemical operation through a decision control model, and when the risk level of chemical operation reaches different preset thresholds, activate the corresponding warning signal and execute safety control measures according to the warning signal;

[0141] An operation analysis module is used to establish a student operation behavior feature library, store each student's number, real-time operation status information and environmental parameter data in the student operation behavior feature library, extract each student's operation feature through a behavior analysis model, generate a student operation feature vector, and update the student operation behavior feature library based on the student operation feature vector;

[0142] The operation improvement module is used to generate experimental operation scores and operation improvement suggestions for each student based on the student operation behavior feature library.

[0143] Specifically, a student laboratory chemical operation control system of this embodiment realizes comprehensive monitoring and management of student experimental operations by integrating data acquisition, identification number, image recognition, risk level, operation warning, operation analysis and operation improvement modules. The system can identify students' operating behaviors and protective equipment wearing in real time, evaluate the risk level of chemical operations, and trigger warning signals and safety control measures when the risk reaches a preset threshold. At the same time, by establishing a student operation behavior feature library and generating personalized operation scores and improvement suggestions, the system effectively improves the laboratory safety management level and teaching quality, and ensures the standardization and safety of experimental operations.

[0144] The present invention also discloses an electronic device, comprising: at least one processor, at least one memory communication interface and a bus: wherein the processor, memory and communication interface communicate with each other through the bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement a distributed photovoltaic power station drone inspection method.

[0145] The present invention also discloses a computer-readable storage medium, which stores computer instructions, and the computer instructions enable the computer to implement all or part of the steps of a distributed photovoltaic power station drone inspection method described in an embodiment of the present invention. The storage medium includes: a U disk, a mobile hard disk, a read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk, and other media that can store program codes.

[0146] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for controlling chemical operations in student laboratories, characterized in that: The following steps are involved: Through monitoring cameras and sensor networks, real-time image data and environmental parameter data in the laboratory are obtained respectively; Identify and number students entering the laboratory; The real-time image data is identified by using a chemical operation identification model to obtain the student's operation behavior information and protective equipment wearing information, and real-time operation status information is generated by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information; Identify the real-time operation status information based on a multi-dimensional risk assessment model to obtain the chemical operation risk level of each student; Identify the chemical operation risk level through a decision control model, activate a corresponding early warning signal when the chemical operation risk level reaches different preset thresholds, and execute safety control measures according to the early warning signal; Establishing a student operation behavior feature library, storing each student's number, real-time operation status information and environmental parameter data in the student operation behavior feature library, extracting each student's operation feature through a behavior analysis model, generating a student operation feature vector, and updating the student operation behavior feature library based on the student operation feature vector; Based on the student operation behavior feature library, generate experimental operation scores and operation improvement suggestions for each student.

2. A method for controlling chemical operations in student laboratories according to claim 1, characterized in that: The real-time image data and environmental parameter data in the laboratory are obtained through monitoring cameras and sensor networks, including: A dual-camera surveillance camera is set up at each laboratory bench in the laboratory, wherein the dual-camera surveillance camera includes a wide-angle surveillance camera and a close-up surveillance camera. The wide-angle surveillance camera is used to capture the overall image of the laboratory bench, and the close-up camera is used to capture the detailed image of the chemical operation area; A sensor network is arranged in the chemical operation area to collect environmental parameter data. The sensor network includes a chemical concentration sensor, a weight sensor, a pressure sensor and a temperature sensor. The chemical concentration sensor is used to monitor the concentration of chemicals in the air, the weight sensor is used to monitor the amount of chemicals used, the pressure sensor is used to monitor the operation force, and the temperature sensor is used to monitor the change of reaction temperature. The chemical operation area includes a laboratory bench, a reagent cabinet and a fume hood.

3. A method for controlling chemical operations in student laboratories according to claim 1, characterized in that: The real-time image data is identified by using a chemical operation identification model to obtain the student's operation behavior information and protective equipment wearing information, and real-time operation status information is generated by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information, specifically including: A chemical operation recognition model is constructed through a deep learning model, and real-time image data is recognized based on the chemical operation recognition model to generate operation behavior information, wherein each feature vector element in the operation behavior information represents the recognition probability of a specific operation behavior; The calculation formula of the chemical operation identification model is: α(t)=1-λ·|P(A i |I t )-P(A i I t-1 )|; Among them, P(A i |I) is the probability of judging the i-th type of operation behavior under a given image I, CNN(I) is the image feature of image I extracted by convolutional neural network, exp(·) is the exponential function, and f(CNN(I)) i is the normalized probability of the i-th type of operation behavior, f(CNN(I)) j is the normalized probability of the jth type of operation behavior, N is the number of categories of operation behavior, α(t) is the time series adjustment factor, β(c) is the environmental constraint factor, λ is the time series smoothing coefficient, P(A i |I t ) is a given image I t The probability of judging it as the i-th type of operation behavior, P(A i |I t-1 ) is a given image I t-1 The probability of being judged as the i-th type of operation behavior, I t and I t-1 are the images at time t and time t-1, c k is the normalized value of the kth environmental parameter, γ k c k The corresponding influence coefficient, M is the number of environmental parameters; The wearing status of protective equipment is detected through image processing algorithms, and protective equipment wearing information is generated. In combination with environmental parameter data, real-time operation status information is generated through weighted algorithms; The calculation formula of the real-time operation status information is: Among them, Z is the normalization factor, S is the real-time operation status information, is the i1th element of the operation behavior information, is the weight coefficient of the i1th element of the operation behavior information, L1 is the number of elements of the operation behavior information, is the i2th element of the protective equipment wearing information, is the weight coefficient of the i2th element of the protective equipment wearing information, L2 is the number of elements of the protective equipment wearing information, is the i3th element of the environmental parameter data, is the weight coefficient of the i3th element of the environmental parameter data, and L3 is the number of elements of the environmental parameter data.

4. A method for controlling chemical operations in student laboratories according to claim 3, characterized in that: The real-time operation status information is identified based on the multi-dimensional risk assessment model to obtain the chemical operation risk level of each student, specifically including: Dynamically extract features of real-time operation status information through a spatiotemporal attention mechanism to construct a risk feature vector, wherein the risk feature vector includes an operation behavior risk feature sub-vector, a protective equipment risk feature sub-vector, and an environmental risk feature sub-vector; A multi-level risk assessment model is established using the analytic hierarchy process. The comprehensive impact of each risk feature is calculated through a combined weight algorithm. Combined with the risk event records in the historical database, the risk assessment parameters are dynamically adjusted to output the chemical operation risk level.

5. A method for controlling chemical operations in student laboratories according to claim 1, characterized in that: The step of identifying the risk level of chemical operation through the decision control model, activating the corresponding early warning signal when the risk level of chemical operation reaches different preset thresholds, and executing safety control measures according to the early warning signal specifically includes: A deep Q-learning network is used to build an early warning decision model, which takes the chemical operation risk level, environmental status and historical early warning records as input, and generates an early warning decision vector through state-action value evaluation; The calculation formula of the early warning decision vector is: Among them, P is the warning decision vector, ω1, ω2, and ω3 are the weight coefficients of the state-action value function, the time series correlation function, and the environmental sensitivity function, respectively. is the Q value of the i6th decision state, is the immediate reward of the i6th decision state, L6 is the number of decision states, μ is the temperature parameter, σ(·) is the sigmoid activation function, is the impact factor of the i7th historical warning event, L7 is the time length of the historical warning event, θ is the state-action time attenuation coefficient, T0 is the current time, For the historical moment, is the value of the i8th environment parameter, is the safety threshold of the i8th environmental parameter, tanh(·) is the hyperbolic tangent activation function, is the sensitivity coefficient of the i8th environmental parameter, L8 is the number of environmental parameters; A multi-level early warning execution mechanism is established based on fuzzy control theory, which triggers early warning signals of different levels according to the threshold of the early warning decision vector, and executes the safety control measures corresponding to the early warning level of the early warning signal; The rules for determining the warning level include: when |P|<0.3, the warning signal corresponding to the low-level warning is triggered, and an sound and light alarm is given; when 0.3≤|P|<0.7, the warning signal corresponding to the intermediate warning is triggered, and ventilation adjustment is started; when |P|≥0.7, the warning signal corresponding to the high-level warning is triggered, and emergency response measures are implemented.

6. A method for controlling chemical operations in student laboratories according to claim 5, characterized in that: The step of establishing a student operation behavior feature library, storing each student's ID, real-time operation status information, and environmental parameter data in the student operation behavior feature library, extracting each student's operation feature through a behavior analysis model, generating a student operation feature vector, and updating the student operation behavior feature library based on the student operation feature vector, specifically includes: Establish a student operation behavior feature library, index the real-time operation status information and environmental parameter data by timestamp and student number, and store them in the student operation behavior feature library, use the behavior analysis model to process the stored data, identify the student's operation mode and behavior characteristics, generate student operation feature vectors, and update the student operation behavior feature library according to the change trend of the student operation feature vectors.

7. A method for controlling chemical operations in student laboratories according to claim 1, characterized in that: The experimental operation score and operation improvement suggestions corresponding to each student are generated based on the student operation behavior feature library, specifically including: A multi-layer scoring mechanism is used to comprehensively evaluate students from multiple dimensions, including operational standardization, safety awareness, and emergency response capabilities. The experimental operation score of each student is calculated through a weighted fusion algorithm, and a student operation ability development curve is generated based on historical data analysis. Construct a chemical operation knowledge graph, which includes standard operating procedures, safety specifications and expert experience. Combined with students' personal characteristics and operating habits, targeted improvement suggestions are generated through rule reasoning and similarity matching algorithms, and the improvement suggestions are dynamically optimized based on student feedback.

8. A student laboratory chemical operation control system, characterized in that: The system comprises: A data acquisition module is used to acquire real-time image data and environmental parameter data in the laboratory through monitoring cameras and sensor networks; The identification and numbering module is used to identify and number students entering the laboratory; An image recognition module is used to recognize the real-time image data using a chemical operation recognition model to obtain the student's operation behavior information and protective equipment wearing information, and generate real-time operation status information by combining the environmental parameter data, the operation behavior information and the protective equipment wearing information; A risk level module, used to identify the real-time operation status information based on a multi-dimensional risk assessment model to obtain the chemical operation risk level of each student; An operation warning module, used to identify the risk level of chemical operation through a decision control model, and when the risk level of chemical operation reaches different preset thresholds, activate the corresponding warning signal and execute safety control measures according to the warning signal; An operation analysis module is used to establish a student operation behavior feature library, store each student's number, real-time operation status information and environmental parameter data in the student operation behavior feature library, extract each student's operation feature through a behavior analysis model, generate a student operation feature vector, and update the student operation behavior feature library based on the student operation feature vector; The operation improvement module is used to generate experimental operation scores and operation improvement suggestions for each student based on the student operation behavior feature library.

9. An electronic device, characterized in that: include: at least one processor, at least one memory, a communication interface, and a bus; The processor, memory, and communication interface communicate with each other via the bus, the memory stores program instructions executable by the processor, and the processor calls the program instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions enable a computer to implement the method according to any one of claims 1 to 7.

Images