Biochemical warfare agent degradation effect AI evaluation system

Abstract

The invention relates to the technical field of biochemical safety and artificial intelligence, in particular to a biochemical warfare agent degradation effect AI evaluation system which comprises a multi-source data acquisition unit, an intelligent feature processing unit, an artificial intelligence middleware unit and an intelligent decision and feedback optimization unit. According to the method, the quality of multi-source data is guaranteed through differential preprocessing, a data type barrier is broken through in combination with cross-modal dynamic weighted fusion, then exclusive features are extracted and standardized based on a degradation mechanism, accurate and adaptive feature input is provided for subsequent evaluation, and on the basis of the feature input, the quality of the multi-source data is improved. Dynamic calculation of the degradation rate, future trend prediction and risk index quantification are realized in combination with a kinetic model and security constraints, and verification is performed through historical data, so that the accuracy, real-time performance and security of degradation effect evaluation of the biochemical warfare agent are remarkably improved, and reliable and efficient data support is provided for degradation process optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biochemical safety and artificial intelligence, in particular to a biochemical warfare agent degradation effect AI evaluation system. BACKGROUND

[0002] Biochemical warfare agents (such as nerve agent sarin, blister agent mustard gas, biological warfare agent anthrax bacillus, etc.) have high toxicity, high diffusion and long-term pollution, and once released, they will pose a serious threat to human health and the ecological environment. Therefore, scientific means are needed to achieve safe disposal - this process is biochemical warfare agent degradation, which specifically refers to the operation of converting highly toxic biochemical warfare agents into low-toxic or non-toxic substances, or destroying their molecular structure and biological activity, so that they lose their original killing and pollution ability, and ultimately eliminate the harm to the human body and the environment. The process is widely used in key scenarios such as biochemical emergency disposal, contaminated site remediation, and national defense security protection. In the prior art, traditional biochemical warfare agent degradation effect evaluation relies on time-consuming and lagging manual sampling detection, single evaluation dimension is easy to misjudge, relies on human experience and is highly subjective, results have no verification and are low in reliability, and cannot provide real-time and reliable data support for process optimization.

[0003] Based on this, the present application provides a biochemical warfare agent degradation effect AI evaluation system to solve the above-mentioned technical problems. SUMMARY

[0004] The present application aims to provide a biochemical warfare agent degradation effect AI evaluation system. The present application guarantees the quality of multi-source data through differential preprocessing, breaks through the data type barrier by combining cross-modal dynamic weighted fusion, extracts exclusive features and standardizes them based on degradation mechanism, provides accurate and adaptive feature input for subsequent evaluation, and based on this feature input, combines kinetic models and safety constraints to realize degradation rate dynamic calculation, future trend prediction and risk quantification, and checks through historical data, thereby significantly improving the accuracy, real-time performance and safety of biochemical warfare agent degradation effect evaluation, and providing reliable and efficient data support for degradation process optimization.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions: The present application provides a biochemical warfare agent degradation effect AI evaluation system, which comprises a multi-source data acquisition unit, an intelligent feature processing unit, an artificial intelligence middleware unit, and an intelligent decision-making and feedback optimization unit, wherein: The multi-source data acquisition unit is used to acquire multi-dimensional raw data of agent concentration, intermediate product content and reaction environment parameters during the degradation process; The intelligent feature processing unit is configured to construct domain-specific feature engineering fused with cross-modal data according to biochemical warfare agent degradation chemical mechanism, clean, fuse and extract features of collected data, and generate standardized input features. The artificial intelligence middleware unit is configured to perform dynamic analysis on the standardized input features based on a machine learning-based multi-dimensional data analysis model, fuse warfare agent degradation kinetic characteristics and safety constraints, and output degradation rate, intermediate product risk index and comprehensive safety level. The intelligent decision-making and feedback optimization unit is configured to visualize the evaluation results and provide data support and intelligent suggestions for degradation process parameter adjustment and optimization.

[0006] The multi-source data acquisition unit includes a multi-type sensor adaptation module, a data real-time acquisition and transmission module, and an acquisition state monitoring module. The multi-type sensor adaptation module is configured to access warfare agent concentration sensors, intermediate product detection sensors, and environmental sensors for temperature, humidity and pH value. The data real-time acquisition and transmission module is configured to acquire sensor data at a preset frequency and transmit the data to the intelligent feature processing unit through wired / wireless means. The acquisition state monitoring module is configured to monitor the working state of the sensors and trigger a prompt when the equipment is abnormal or data is interrupted.

[0007] The intelligent feature processing unit includes a data preprocessing module, a cross-modal data fusion module, and a domain-specific feature generation module. The data preprocessing module is configured to denoise, fill in missing values, and remove outliers from the collected raw data. The cross-modal data fusion module is configured to associate and fuse numerical parameters with spectral / chromatographic curve data to obtain associated features between multi-dimensional data. The domain-specific feature generation module is configured to extract reaction kinetic characteristics and safety warning features based on biochemical warfare agent degradation mechanism, and generate a standardized feature vector suitable for machine learning model.

[0008] In the cross-modal data fusion module, numerical parameters and spectral / chromatographic curve data are associated and fused to obtain associated features between multi-dimensional data, which are implemented as follows: A1: The input numerical parameter set A, including warfare agent concentration, reaction temperature and pH value, is normalized to generate a standardized numerical feature vector A', and the normalization formula is: ; In the formula, is the original value of the numerical parameter, respectively the historical minimum value, maximum value of the parameter, and the characteristic peak extraction is performed on the spectral / chromatographic curve data set B including the wavelength corresponding to the absorbance value, the characteristic peak interval reflecting the change of the molecular structure of the war agent is retained, and a curve feature vector B' is generated; A2: based on the timestamp, the normalized numerical feature vector A' and the curve feature vector B' are time-aligned to form a feature pair after alignment . A3: based on the attention mechanism, the dynamic weight of the numerical feature and the curve feature is calculated, and the weight calculation formula is: . . In the formula, is the dynamic weight of the numerical feature, is the dynamic weight of the curve feature, t is the time series index, , respectively the variance of A' and B' in the time series t, the greater the variance, the higher the sensitivity of the modal data to the change of the degradation state, and the greater the weight; A4: the aligned features are weighted and summed according to the dynamic weight to obtain a multi-dimensional correlation feature vector F, and the weighted sum formula is: . In the formula, is the cross-modal correlation feature corresponding to time t.

[0009] The field-specific feature generation module extracts reaction kinetics features and safety warning features in combination with the degradation mechanism of biochemical war agents, generates a standardized feature vector suitable for machine learning models, and the specific operations are as follows: B1: in combination with the degradation chemical mechanism of the target biochemical war agent, the core parameters of war agent concentration c(t), reaction temperature T, pH value at different time nodes are determined, and the real-time concentration of toxic intermediate product , the concentration of degradation end product . B2: based on the first-order degradation reaction kinetics mechanism, the kinetic characteristics of reaction rate constant, half-life, and reaction completion degree are calculated, and the specific formula is as follows: Ⅰ. Reaction rate constant feature k: the first-order reaction equation is fitted through the change of war agent concentration with time , and the solution is: , in the formula, is the initial concentration of the war agent, and t is the degradation time; Ⅱ. Half-life feature : based on the reaction rate constant calculation, reflecting the degradation speed of the war agent, the formula is: . Ⅲ、Reaction completion degree characteristics : Calculate the ratio of the current concentration to the initial concentration, which reflects the degradation progress, and the formula is: ; B3: Combine the degradation product toxicity mechanism to generate intermediate product risk coefficient, degradation product stability safety warning characteristics: Ⅰ, Intermediate product risk coefficient : Compare the real-time concentration of intermediate product with the preset safety threshold , the formula is: ; Ⅱ, Degradation product stability characteristics S: calculated by the concentration change rate of degradation end product within 24h, the formula is: ; In the formula, is the concentration of the degradation end product at time t, , and the smaller the product change is, the closer S is to 1; B4: Map the above kinetic characteristics and safety warning characteristics to the [0, 1] interval through Min-Max standardization to obtain a standardized feature set , and finally build a standardized feature vector for adaptive machine learning model.

[0010] The artificial intelligence middleware unit includes a multi-dimensional data dynamic analysis module, an evaluation result output and verification module, wherein: The multi-dimensional data dynamic analysis module: for inputting standardized features, combining degradation kinetics and safety threshold constraints to calculate degradation rate and intermediate product risk index; The evaluation result output and verification module: for generating a comprehensive safety level report and comparing and verifying with historical data.

[0011] In the multi-dimensional data dynamic analysis module, input the standardized features, combine the degradation kinetics and safety threshold constraints to calculate the degradation rate and intermediate product risk index, and the specific operation is as follows: C1: Receive the standardized feature vector output by the intelligent feature processing unit, extract the standardized reaction completion degree characteristics , reaction rate constant standardized characteristics , call the preset first-order degradation reaction kinetics model for fitting, dynamically calculate and output the current degradation rate and the prediction of future period degradation process; C2: Extract the intermediate product risk coefficient standardized feature from the standardized feature vector, combine the preset safety threshold library and the intermediate product concentration change trend to calculate the quantitative intermediate product risk index; C3: The fusion degradation rate, the predicted results of future degradation process, and the risk index of intermediate products are combined and the final comprehensive safety level is output based on the fusion results and the preset safety level rules.

[0012] The specific formulas in C1 and C2 are as follows: Current degradation rate calculation: based on standardized reaction completion characteristics The degree of completeness of the reduction reaction is obtained. The degradation rate was calculated using the following formula: ; In the formula, This represents the current degradation rate. Future degradation process prediction: Based on the reaction rate constant k, the agent concentration at future times is calculated, and the predicted degradation rate is obtained. The formula is as follows: ; ; In the formula, For the future Predicted degradation rate at time, To predict duration, Predicting agent concentrations for future moments; Intermediate Product Risk Index Calculation: Based on Intermediate Product Risk Coefficient The trend of fusion concentration change is expressed by the formula: ; In the formula, A quantified risk index for intermediate products. , For preset weighting coefficients, This represents the rate of change in the concentration of the intermediate product. The preset maximum concentration rise slope, This only considers the impact of an upward trend on risk.

[0013] The assessment result output and verification module generates a comprehensive security level report and compares it with historical data for verification. The specific operations are as follows: D1: Based on the degradation rate, intermediate product risk index, and overall safety level output by the multidimensional data dynamic analysis module, a structured assessment report is generated; D2: Retrieve historical assessment results from the historical degradation case database that match the current agent type, initial concentration, and reaction conditions as a comparison benchmark; D3: Perform a consistency check between the current evaluation result and the historical benchmark. If the deviation exceeds the preset tolerance range, mark it as abnormal and trigger an early warning.

[0014] The intelligent decision-making and feedback optimization unit includes an evaluation result visualization module, a process parameter recommendation module, and a data storage and backtracking module, wherein: The assessment result visualization module is used to display changes in degradation rate and risk index through charts. The process parameter recommendation module is used to adjust the parameter adjustment scheme based on the evaluation results and the output parameters from the process database. The data storage and backtracking module is used to store all evaluation results and process adjustment records, and supports historical data query and backtracking.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention ensures the quality of multi-source data through differentiated preprocessing, breaks down data type barriers by combining cross-modal dynamic weighted fusion, extracts and standardizes specific features based on degradation mechanisms, and provides accurate and suitable feature inputs for subsequent evaluation. Based on these feature inputs, it achieves dynamic calculation of degradation rate, prediction of future trends, and quantification of risk index by combining kinetic models and safety constraints. At the same time, it is verified by historical data, thereby significantly improving the accuracy, real-time performance, and safety of biochemical warfare agent degradation effect evaluation, and providing reliable and efficient data support for degradation process optimization. Attached Figure Description

[0016] Fig. 1 This is a system diagram of an AI evaluation system for the degradation effect of biochemical warfare agents according to the present invention.

[0017] Fig. 2 This is a flowchart of cross-modal data fusion in an AI evaluation system for the degradation effect of biochemical warfare agents according to the present invention.

[0018] Explanation of icon numbers: 100. Multi-source data acquisition unit; 101. Multi-type sensor adaptation module; 102. Real-time data acquisition and transmission module; 103. Acquisition status monitoring module; 200. Intelligent feature processing unit; 201. Data preprocessing module; 202. Cross-modal data fusion module; 203. Domain-specific feature generation module; 300. Artificial intelligence middleware unit; 301. Multi-dimensional data dynamic analysis module; 302. Evaluation result output and verification module; 400. Intelligent decision-making and feedback optimization unit; 401. Evaluation result visualization module; 402. Process parameter recommendation module; 403. Data storage and backtracking module. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Example: like Figs. 1-2 As shown in the figure, this embodiment provides an AI evaluation system for the degradation effect of biochemical warfare agents, including a multi-source data acquisition unit 100, an intelligent feature processing unit 200, an artificial intelligence middleware unit 300, and an intelligent decision-making and feedback optimization unit 400. Specifically: the multi-source data acquisition unit 100 is used to collect multi-dimensional raw data on agent concentration, intermediate product content, and reaction environment parameters during the degradation process; the intelligent feature processing unit 200 is used to construct a domain-specific feature engineering that integrates cross-modal data based on the chemical mechanism of biochemical warfare agent degradation, cleans, fuses, and extracts features from the collected data, and generates standardized input features; the artificial intelligence middleware unit 300 is a multi-dimensional data analysis model based on machine learning, which integrates agent degradation kinetic characteristics and safety constraints, dynamically analyzes the standardized input features, and outputs the degradation rate, intermediate product risk index, and overall safety level; the intelligent decision-making and feedback optimization unit 400 is used to visualize the evaluation results and provide data support and intelligent suggestions for adjusting and optimizing degradation process parameters.

[0021] It should be noted that the multi-source data acquisition unit 100 acquires raw monitoring data, which is then fused across modes and constructed with domain-specific features based on chemical mechanisms by the intelligent feature processing unit 200 to generate standardized input. Subsequently, the artificial intelligence middleware unit 300 combines kinetic laws and safety constraints to perform dynamic analysis of the features and output degradation rate, risk index and safety level. Finally, the intelligent decision-making and feedback optimization unit 400 visualizes the evaluation results and drives the intelligent optimization of process parameters.

[0022] In this embodiment, it should also be noted that the multi-source data acquisition unit 100 includes a multi-type sensor adaptation module 101, a real-time data acquisition and transmission module 102, and an acquisition status monitoring module 103, wherein: the multi-type sensor adaptation module 101 is used to connect to a combat agent concentration sensor, an intermediate product detection sensor, and an environmental sensor for temperature, humidity / pH value; the real-time data acquisition and transmission module 102 is used to acquire data from each sensor at a preset frequency and transmit it to the intelligent feature processing unit 200 via wired / wireless means; the acquisition status monitoring module 103 is used to monitor the working status of the sensors and trigger a prompt when the device is abnormal or the data is interrupted.

[0023] It should be noted that the multi-type sensor adaptation module 101 is connected to the agent concentration, intermediate product and environmental parameter sensors. The data is collected and transmitted to the intelligent feature processing unit 200 at a preset frequency by the real-time data acquisition and transmission module 102. At the same time, the acquisition status monitoring module 103 monitors the sensor operation status in real time and triggers an alarm when the equipment is abnormal or the data is interrupted.

[0024] Furthermore, it should be noted that the multi-type sensor adapter module 101 includes the following sensors: a war agent concentration sensor: employing a surface-enhanced Raman scattering sensor with a detection limit ≤0.1μg / mL and a response time ≤10s, capable of identifying nerve agents such as sarin and VX, as well as biological warfare agents such as anthrax; an intermediate product detection sensor: employing a high-performance liquid chromatography sensor, capable of separating and detecting degradation intermediates with a detection accuracy ≤0.05μg / mL; environmental sensors: a temperature and humidity sensor with a measurement range of -20℃ to 60℃ and 0 to 100%RH, with an error ≤±0.5℃ / ±2% RH; and a pH sensor with a measurement range of 0 to 14, with an error ≤±0.02pH, all equipped with a PTFE coating resistant to war agent corrosion.

[0025] The data acquisition and transmission module 102 can dynamically adjust the acquisition frequency according to the degradation stage: the acquisition frequency is 1 time / minute in the early stage of degradation (high agent concentration and violent reaction), 1 time / 5 minutes in the middle stage of degradation (concentration steadily decreases), and 1 time / 15 minutes in the late stage of degradation (concentration approaches 0); the transmission method prioritizes 5G private network (latency ≤10ms), with wired Ethernet as backup, to ensure that no data is lost.

[0026] The abnormal judgment criteria in the data acquisition status monitoring module 103 are: sensor power supply voltage exceeding 5±0.2V, no data transmission for 3 consecutive acquisition cycles, and single-cycle data fluctuation >20% (e.g., pH value suddenly changes from 7.0 to 8.5); alarm methods include: local audible and visual alarm (buzzer sounds + red light flashes), and remote push alarm (pushed to the administrator terminal via SMS / system message), while automatically switching to the backup sensor.

[0027] In this embodiment, it should also be noted that the intelligent feature processing unit 200 includes a data preprocessing module 201, a cross-modal data fusion module 202, and a domain-specific feature generation module 203, wherein: the data preprocessing module 201 is used to perform noise reduction, missing value imputation, and outlier removal on the collected raw data; the cross-modal data fusion module 202 is used to correlate and fuse numerical parameters with spectral / chromatographic curve data to obtain correlation features between multi-dimensional data; the specific operation is as follows: A1: The input set of numerical parameters A, including agent concentration, reaction temperature, and pH value, is normalized to generate a standardized numerical feature vector A', and the normalization formula is: ; In the formula, These are the original values ​​of the numerical parameters. The historical minimum and maximum values ​​of the parameter are respectively, and the characteristic peaks of the spectral / chromatographic curve data set B, including the absorbance values ​​corresponding to the wavelength, are extracted, and the characteristic peak ranges reflecting the changes in the molecular structure of the combat agent are retained to generate the curve feature vector B'. A2: Based on the timestamp, the standardized numerical feature vector A' and the curve feature vector B' are time-series aligned to form an aligned feature pair. ; A3: The dynamic weights of numerical and curve features are calculated based on an attention mechanism. The weight calculation formula is as follows: ; ; In the formula, The dynamic weights for numerical features. represents the dynamic weights of the curve features, and t is the time series index. , A' and B' are the variances of A' and B' within the time series t, respectively. The larger the variance, the higher the sensitivity of the modality data to changes in degradation state, and the greater the weight. A4: The aligned features are summed according to dynamic weights to obtain the multi-dimensional associated feature vector F. The weighted summation formula is: ; In the formula, This represents the cross-modal correlation features corresponding to time t. Domain-specific feature generation module 203: Used to combine the degradation mechanism of biochemical warfare agents, extract reaction kinetic features and safety warning features, and generate standardized feature vectors adapted to machine learning models. Specific operations are as follows: B1: Combining the degradation chemical mechanism of the target biochemical warfare agent, determine the core parameters that need to be collected at different time points, including agent concentration c(t), reaction temperature T, pH value, and the real-time concentration of toxic intermediate products. Concentration of final degradation products B2: Based on the kinetic mechanism of first-order degradation reaction, the kinetic characteristics of the reaction rate constant, half-life, and reaction completeness are calculated. The specific formulas are as follows: I. Reaction rate constant characteristic k: Fitting the first-order reaction equation by the change of agent concentration over time. Solving for the problem, we get: In the formula, The initial concentration of the agent is given by t, and the degradation time is given by t. II. Half-life characteristics Based on the reaction rate constant, it reflects the degradation rate of the war agent, and the formula is: III. Characteristics of Reaction Completeness Calculate the ratio of the current concentration to the initial concentration to reflect the degradation progress. The formula is: B3: Based on the toxicity mechanism of degradation products, safety warning characteristics for the risk coefficient of intermediate products and the stability of degradation products: I. Risk coefficient of intermediate products Compare the real-time concentration of intermediate products with the preset safety threshold. The formula is: ; II. Stability characteristics of degradation products S: Calculated by the rate of change in the concentration of the final degradation products within 24 hours, using the following formula: ; In the formula, Let t be the concentration of the final degradation products at time t. Furthermore, the smaller the product change, the closer S is to 1; B4: The above kinetic characteristics With security early warning characteristics By standardizing and mapping to the [0,1] interval using Min-Max, we obtain the standardized feature set. Ultimately, a standardized feature vector adapted to machine learning models is constructed.

[0028] It should be noted that the data preprocessing module 201 cleans and repairs the original multi-source data. The cross-modal data fusion module 202 merges the numerical parameters and spectral / chromatographic curve data through normalization, feature peak extraction, time alignment, and dynamic weighted fusion based on variance sensitivity to generate a multi-dimensional associated feature vector. Then, the domain-specific feature generation module 203 combines the chemical mechanism of biochemical warfare agent degradation to further extract reaction kinetic features and safety warning features from the fused features, and standardizes and constructs a domain-specific feature vector adapted to the machine learning model.

[0029] Furthermore, it should be noted that the data preprocessing module 201 employs different preprocessing methods for "numerical data (temperature, pH)" and "spectral / chromatographic curve data": For numerical data (temperature, pH): outliers are removed using the 3σ criterion (data exceeding the mean ± 3 times the standard deviation are considered outliers), and missing values ​​are filled using either "linear interpolation" (suitable for missing values ​​≤ 5 minutes) or "historical similar data filling" (suitable for missing values ​​> 5 minutes, matching historical data of the same degradation stage and environmental parameters); For spectral / chromatographic curve data: "wavelet transform noise reduction" is used (selecting the db4 wavelet basis and decomposing into 3 layers), baseline drift is removed (the baseline is fitted using a 5th-order polynomial and then subtracted from the baseline value), ensuring that characteristic peaks are free from baseline interference.

[0030] Feature peak extraction in A1: For spectral data, the "adaptive threshold method" is used to extract feature peaks (threshold = baseline mean + 2 times baseline standard deviation), retaining the peak range of "peak height ≥ 0.5 AU (absorbance unit) and half-width ≤ 5 nm", and removing impurity peaks that have no molecular structural significance.

[0031] In B1, the corresponding degradation mechanisms are selected for different war agents as follows: nerve agents (sarin, VX) are adapted to the "hydrolysis reaction mechanism", with the core parameters being pH value and temperature (which affect the hydrolysis rate); biological warfare agents (bacillus anthracis) are adapted to the "oxidative inactivation mechanism", with the core parameters being oxidant concentration and reaction time.

[0032] In this embodiment, it should also be noted that the artificial intelligence middleware unit 300 includes a multi-dimensional data dynamic analysis module 301 and an evaluation result output and verification module 302, wherein: the multi-dimensional data dynamic analysis module 301 is used to input standardized features, combine degradation kinetics laws and safety threshold constraints, and calculate the degradation rate and intermediate product risk index; the specific operation is as follows: C1: receives the standardized feature vector output by the intelligent feature processing unit 200, and extracts the standardized reaction completeness feature therein. Normalized characteristics of reaction rate constant The system uses a pre-defined first-order degradation reaction kinetic model for fitting, dynamically calculates and outputs the current degradation rate and a prediction of the degradation process in future periods; the current degradation rate is calculated based on standardized reaction completion characteristics. The degree of completeness of the reduction reaction is obtained. The degradation rate was calculated using the following formula: ; In the formula, The current degradation rate; future degradation prediction: Based on the reaction rate constant k, the agent concentration at future times is calculated, and the predicted degradation rate is obtained, using the following formula: ; ; In the formula, For the future Predicted degradation rate at time, To predict duration, C2: Predicting agent concentrations for future moments; C2: Extracting intermediate product risk coefficients from standardized feature vectors. Based on a pre-defined safety threshold library and the concentration trends of intermediate products, a quantified intermediate product risk index is calculated. The intermediate product risk index is calculated based on the intermediate product risk coefficient. The trend of fusion concentration change is expressed by the formula: ; In the formula, A quantified risk index for intermediate products. , For preset weighting coefficients, This represents the rate of change in the concentration of the intermediate product. The preset maximum concentration rise slope, To consider only the impact of the upward trend on risk. C3: Integrates the degradation rate, future degradation process prediction results, and intermediate product risk index, and outputs the final comprehensive safety level based on the integration results and preset safety level rules. Evaluation result output and verification module 302: Used to generate a comprehensive safety level report and compare and verify it with historical data. The specific operation is as follows: D1: Generate a structured evaluation report based on the degradation rate, intermediate product risk index, and comprehensive safety level output by the multidimensional data dynamic analysis module 301; D2: Retrieve historical evaluation results that match the current agent type, initial concentration, and reaction conditions from the historical degradation case library as a comparison benchmark; D3: Verify the consistency between the current evaluation results and the historical benchmark. If the deviation exceeds the preset tolerance range, mark it as abnormal and trigger an early warning.

[0033] It should be noted that the multidimensional data dynamic analysis module 301 and the evaluation result output and verification module 302 work together: the former is based on the standardized features output by the intelligent feature processing unit 200, combined with the degradation kinetic model and safety threshold constraints, to dynamically calculate the current degradation rate, predict the future degradation trend and quantify the risk index of intermediate products, and then integrate multidimensional indicators to output a comprehensive safety level; the latter is based on this evaluation result to generate a structured report, and verifies the consistency with similar historical cases, triggering an anomaly warning when the deviation exceeds the limit.

[0034] Furthermore, it should be noted that the fusion value calculation in C3 uses a "weighted summation," with the weights allocated as follows: current degradation rate 0.4 + future predicted probability of achieving the target 0.2 + risk index reverse value 0.4. Safety level rules: fusion value ≥ 0.8 is "safe," 0.5~0.8 is "warning," and < 0.5 is "dangerous."

[0035] The structured report in D1 contains the following: ① Basic information (agent type, degradation start time, experiment number); ② Real-time data (current degradation rate, risk index, environmental parameters); ③ Predicted results (degradation curve for the next 24 hours, predicted time to reach target); ④ Verification conclusions (deviation from historical data, whether it is normal). The preset tolerance range in D3 is 10%; the deviation calculation formula is as follows: ; Anomaly handling: When the deviation exceeds the limit, a "deviation cause analysis" will be pushed out simultaneously (such as whether the degradation rate is too low due to a sudden change in pH value) and a "parameter verification prompt" will be triggered (it is recommended to check whether the sensor is drifting or whether the reaction environment is abnormal).

[0036] In this embodiment, it should also be noted that the intelligent decision-making and feedback optimization unit 400 includes an evaluation result visualization module 401, a process parameter recommendation module 402, and a data storage and backtracking module 403, wherein: the evaluation result visualization module 401 is used to display the degradation rate and risk index changes through charts; the process parameter recommendation module 402 is used to output parameter adjustment schemes based on the evaluation results and the process database; and the data storage and backtracking module 403 is used to store evaluation results and process adjustment records, supporting historical data query and backtracking.

[0037] It should be noted that the evaluation result visualization module 401 intuitively presents the dynamic changes of degradation rate and risk index, the process parameter recommendation module 402 generates optimization suggestions based on the evaluation results and combined with the process database, and the data storage and backtracking module 403 synchronously saves all evaluation results and adjustment records to support historical queries and process traceability.

[0038] Furthermore, it should be noted that the parameter adjustment scheme in the process parameter recommendation module 402 is based on the "process-effect correlation model" (trained through 1000+ sets of degradation experiments): if the current degradation rate is lower than expected (e.g., 20%, lower than the historical average of 30% in the same period), and the pH value is 6.0 (slightly acidic), it is recommended to "adjust the pH value to 7.5~8.0", with an expected degradation rate increase of 15% within 1 hour; if the intermediate product risk index exceeds the standard (e.g., 2.8), and the temperature is 25℃, it is recommended to "increase the temperature to 30℃".

[0039] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0040] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An AI-based evaluation system for the degradation effect of biological warfare agents, characterized in that, It includes a multi-source data acquisition unit (100), an intelligent feature processing unit (200), an artificial intelligence middleware unit (300), and an intelligent decision-making and feedback optimization unit (400), wherein: The multi-source data acquisition unit (100) is used to collect multi-dimensional raw data on agent concentration, intermediate product content, and reaction environment parameters during the degradation process. The intelligent feature processing unit (200) is used to construct a domain-specific feature engineering that integrates cross-modal data based on the degradation chemical mechanism of biochemical warfare agents, clean, fuse and extract features from the collected data, and generate standardized input features. The artificial intelligence middleware unit (300) is a multi-dimensional data analysis model based on machine learning. It integrates the degradation kinetics characteristics of the combat agent with safety constraints, performs dynamic analysis on standardized input features, and outputs degradation rate, intermediate product risk index and comprehensive safety level. The intelligent decision-making and feedback optimization unit (400) is used to visualize the evaluation results and provide data support and intelligent suggestions for adjusting and optimizing degradation process parameters.

2. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 1, characterized in that, The multi-source data acquisition unit (100) includes a multi-type sensor adaptation module (101), a real-time data acquisition and transmission module (102), and an acquisition status monitoring module (103), wherein: The multi-type sensor adapter module (101) is used to connect to a combat agent concentration sensor, an intermediate product detection sensor, and an environmental sensor for temperature, humidity / pH value. The real-time data acquisition and transmission module (102) is used to acquire data from each sensor at a preset frequency and transmit it to the intelligent feature processing unit (200) via wired / wireless means. The data acquisition status monitoring module (103) is used to monitor the working status of the sensor and trigger a prompt when the device malfunctions or the data is interrupted.

3. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 1, characterized in that, The intelligent feature processing unit (200) includes a data preprocessing module (201), a cross-modal data fusion module (202), and a domain-specific feature generation module (203), wherein: The data preprocessing module (201) is used to perform noise reduction, missing value imputation, and outlier removal on the collected raw data. The cross-modal data fusion module (202) is used to correlate and fuse numerical parameters with spectral / chromatographic curve data to obtain correlation features between multi-dimensional data. The domain-specific feature generation module (203) is used to combine the degradation mechanism of biochemical warfare agents, extract reaction kinetic features and safety warning features, and generate standardized feature vectors that are adapted to machine learning models.

4. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 3, characterized in that, The cross-modal data fusion module (202) correlates and fuses numerical parameters with spectral / chromatographic curve data to obtain the correlation features between multi-dimensional data. The specific operation is as follows: A1: Normalize the input set of numerical parameters A, including agent concentration, reaction temperature, and pH value, to generate a standardized numerical feature vector A'. The normalization formula is: ； In the formula, These are the original values ​​of the numerical parameters. The historical minimum and maximum values ​​of the parameter are respectively, and the characteristic peaks of the spectral / chromatographic curve data set B, including the absorbance values ​​corresponding to the wavelength, are extracted, and the characteristic peak ranges reflecting the changes in the molecular structure of the combat agent are retained to generate the curve feature vector B'. A2: Based on the timestamp, the standardized numerical feature vector A' and the curve feature vector B' are time-series aligned to form an aligned feature pair. ; A3: The dynamic weights of numerical and curve features are calculated based on an attention mechanism. The weight calculation formula is as follows: ； ； In the formula, The dynamic weights for numerical features. represents the dynamic weights of the curve features, and t is the time series index. , , , are the variances of A' and B' within the time series t, respectively. The larger the variance, the higher the sensitivity of the modal data to changes in the degradation state, and the greater the weight. A4: The aligned features are weighted and summed according to dynamic weights to obtain the multi-dimensional associated feature vector F. The weighted summation formula is: ； In the formula, This represents the cross-modal correlation feature corresponding to time t.

5. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 3, characterized in that, The domain-specific feature generation module (203) combines the degradation mechanism of biochemical warfare agents to extract reaction kinetic features and safety warning features, and generates standardized feature vectors adapted to machine learning models. The specific operations are as follows: B1: Based on the degradation chemical mechanism of the target biochemical warfare agent, determine the core parameters that need to be collected at different time points, including agent concentration c(t), reaction temperature T, pH value, and the real-time concentration of toxic intermediate products. Concentration of final degradation products ; B2: Based on the kinetic mechanism of first-order degradation reaction, the kinetic characteristics of the reaction rate constant, half-life, and reaction completeness are calculated. The specific formulas are as follows: I. Reaction rate constant characteristic k: Fitting the first-order reaction equation by the change of agent concentration over time. Solving for the problem, we get: In the formula, t represents the initial concentration of the agent and t represents the degradation time. II. Half-life characteristics Based on the reaction rate constant, it reflects the degradation rate of the war agent, and the formula is: ; III. Characteristics of Reaction Completeness Calculate the ratio of the current concentration to the initial concentration to reflect the degradation progress. The formula is: ； B3: Safety warning features based on the toxicity mechanism of degradation products, the risk coefficient of intermediate products, and the stability of degradation products: I. Risk coefficient of intermediate products Compare the real-time concentration of intermediate products with the preset safety threshold. The formula is: ； II. Stability characteristics of degradation products S: Calculated by the rate of change in the concentration of the final degradation products within 24 hours, using the following formula: ； In the formula, Let t be the concentration of the final degradation products at time t. Furthermore, the smaller the product change, the closer S is to 1; B4: The above dynamic characteristics With security early warning characteristics By standardizing and mapping to the [0,1] interval using Min-Max, we obtain the standardized feature set. Ultimately, a standardized feature vector adapted to machine learning models is constructed.

6. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 1, characterized in that, The artificial intelligence middleware unit (300) includes a multi-dimensional data dynamic analysis module (301) and an evaluation result output and verification module (302), wherein: The multidimensional data dynamic analysis module (301) is used to input standardized features, combine degradation kinetics with safety threshold constraints, and calculate degradation rate and intermediate product risk index. The evaluation result output and verification module (302) is used to generate a comprehensive security level report and compare and verify it with historical data.

7. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 6, characterized in that, The multidimensional data dynamic analysis module (301) inputs standardized features, combines degradation kinetics with safety threshold constraints, and calculates the degradation rate and intermediate product risk index. The specific operation is as follows: C1: Receives the standardized feature vector output by the intelligent feature processing unit (200) and extracts the standardized response completeness feature from it. Normalized characteristics of reaction rate constant The system calls a preset first-order degradation reaction kinetic model for fitting, dynamically calculates and outputs the current degradation rate and a prediction of the degradation process in future periods. C2: Extracting intermediate product risk coefficients from standardized feature vectors. Based on a pre-set safety threshold library and the concentration change trend of intermediate products, a quantitative risk index of intermediate products is calculated. C3: The fusion degradation rate, the predicted results of future degradation process, and the risk index of intermediate products are combined and the final comprehensive safety level is output based on the fusion results and the preset safety level rules.

8. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 7, characterized in that, The specific formulas in C1 and C2 are as follows: Current degradation rate calculation: based on standardized reaction completion characteristics The degree of completeness of the reduction reaction is obtained. The degradation rate was calculated using the following formula: ； In the formula, This represents the current degradation rate. Future degradation process prediction: Based on the reaction rate constant k, the agent concentration at future times is calculated, and the predicted degradation rate is obtained. The formula is as follows: ； ； In the formula, For the future Predicted degradation rate at time, To predict duration, Predicting agent concentrations for future moments; Intermediate Product Risk Index Calculation: Based on Intermediate Product Risk Coefficient The trend of fusion concentration change is expressed by the formula: ； In the formula, A quantified risk index for intermediate products. , For preset weighting coefficients, This represents the rate of change in the concentration of the intermediate product. The preset maximum concentration rise slope, This only considers the impact of an upward trend on risk.

9. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 7, characterized in that, The evaluation result output and verification module (302) generates a comprehensive security level report and compares and verifies it with historical data. The specific operation is as follows: D1: Based on the degradation rate, intermediate product risk index and comprehensive safety level output by the multidimensional data dynamic analysis module (301), a structured assessment report is generated; D2: Retrieve historical assessment results from the historical degradation case database that match the current agent type, initial concentration, and reaction conditions as a comparison benchmark; D3: Perform a consistency check between the current evaluation result and the historical benchmark. If the deviation exceeds the preset tolerance range, mark it as abnormal and trigger an early warning.

10. The AI ​​evaluation system for the degradation effect of biological warfare agents according to claim 1, characterized in that, The intelligent decision-making and feedback optimization unit (400) includes an evaluation result visualization module (401), a process parameter recommendation module (402), and a data storage and backtracking module (403), wherein: The evaluation result visualization module (401) is used to display changes in degradation rate and risk index through charts; The process parameter recommendation module (402) is used to adjust the parameter adjustment scheme based on the evaluation results and the output parameters of the process database. The data storage and backtracking module (403) is used to store all evaluation results and process adjustment records, and supports historical data query and backtracking.

Images