Improved PSR model carbon emission early warning system

Abstract

The invention belongs to the field of environmental monitoring and prediction, and particularly relates to an improved PSR (Pressure State Response) model carbon emission early warning system, which is used for carrying out improved PSR model carbon emission early warning based on a biological immune system principle and comprises a data acquisition module, a PSR model, a biological immune algorithm model and an early warning output module, the method is used for identifying abnormal changes of carbon emission and realizing dynamic early warning, and has the advantages of high adaptability, high detection capability, multi-factor comprehensive analysis, identification of potential problems and optimal regulation and control paths existing in the system, and realization of accurate prediction and early warning of carbon emission.

Description

Technical Field

[0001] The present invention belongs to the field of environmental monitoring and prediction, and specifically relates to a carbon emission early warning system based on an improved Pressure - State - Response (PSR) model according to the principle of biological immune system, which is used to identify abnormal changes in carbon emissions and achieve dynamic early warning. Background Art

[0002] Currently, global climate change and carbon emission problems are becoming increasingly serious. Scientifically and effectively conducting carbon emission early warning has become an important means for environmental protection and policy - making. Traditional carbon emission prediction and early warning methods (such as grey system analysis, regression analysis, etc.) have the following deficiencies when dealing with multi - factor dynamic complex systems: 1. Too strong linear assumption: Traditional methods are difficult to capture the non - linear relationship between carbon emission factors.

[0003] 2. Lack of self - adaptability: Unable to respond to dynamic impacts such as policy adjustments and energy structure changes in real time.

[0004] 3. Limited anomaly detection ability: Weak ability to identify sudden carbon emission anomalies.

[0005] 4. Only reflect the current data or the state at the time of data collection, and also unable to identify potential problems and regulation paths existing in the system.

[0006] The biological immune system has the advantages of self - adaptability, dynamic learning and anomaly detection, can identify potential problems and optimal regulation paths existing in the system. Combining with the improved PSR model, it can more accurately capture the multi - dimensional dynamic characteristics of the carbon emission system, providing an efficient and flexible new method for carbon emission early warning. Summary of the Invention

[0007] The present invention aims to provide a carbon emission early warning system based on an improved PSR model using a biological immune algorithm, constructing a model with real - time dynamic update, self - adaptability and multi - factor comprehensive analysis capabilities, so as to identify potential problems and optimal regulation paths existing in the system, and achieve accurate prediction and early warning of carbon emissions.

[0008] The technical solution adopted by the present invention is as follows: A carbon emission early warning system based on an improved PSR model, including a data acquisition module, a PSR model, a biological immune algorithm model, and a warning output module; The data acquisition module obtains real - time data related to carbon emissions through the Internet of Things and big data platform; The PSR model pre - processes and standardizes the real - time data according to three dimensions of pressure, state and response; The bio - immune algorithm model constructs an antibody from data in three dimensions: pressure, state, and response. It takes the average value of zero - carbon - emission or low - risk areas as the target antigen, calculates the affinity between the antibody and the target antigen, and obtains the comprehensively risk - optimized index through cloning and mutation operations on the antibody. The early - warning output module outputs the early - warning level according to the comprehensive risk index.

[0009] Preferably, the data - acquisition module obtains real - time data related to carbon emissions through the Internet of Things and big - data platform, specifically including: Energy - consumption data, industrial - production data, urbanization data; Total carbon - emission data, environmental - condition data; Policy - implementation data, new - energy - utilization data, technological - innovation data.

[0010] Preferably, the PSR model pre - processes and standardizes the real - time data according to the three dimensions of pressure, state, and response, specifically including: Data cleaning, which eliminates outliers and fills in missing values in the collected original data; Data standardization, which standardizes all data using the range normalization method; Data fusion, which aligns the pressure, state, and response data according to a unified time dimension; Data dynamic update, which updates the real - time data daily through the Internet of Things and API interfaces.

[0011] Preferably, constructing an antibody from data in three dimensions of pressure, state, and response specifically includes: Representing each group of samples in the dataset as an antibody to form an antibody population , each antibody is composed of data in the dimensions of pressure (P), state (S), and response (R), , where n is the initial population size.

[0012] Preferably, n≥3.

[0013] Preferably, calculating the affinity between each antibody and the target antigen specifically includes: Taking the average value of zero - carbon - emission or low - risk areas as the target antigen, the affinity between each antibody and the target antigen The calculation formula is: ; : The - th feature value of the - th antibody.

[0014] : Target antigen.

[0015] Preferably, through the cloning and mutation operations of antibodies, specifically including: According to the affinity of the antibodies, different cloning numbers are proportionally allocated. The higher the affinity of the antibody, the greater the probability of being selected, achieving the effect of replicating high-quality antibodies. ; Wherein, is the cloning number of the i-th antibody, N is the total number of clones, n is the initial population size, is the affinity of the i-th antibody; The cloned antibodies are randomly mutated within a small range. The mutation formula is: ; Wherein, is the mutation amplitude, is the random perturbation factor, is The antibody after mutation.

[0016] Preferably, through the cloning and mutation operations of antibodies, it also includes: Inhibition operation: Remove the antibodies with high similarity in the population after mutation, avoid redundancy, and maintain population diversity; Similarity formula: ; If 5, then remove one of the antibodies.

[0017] Preferably, obtaining the comprehensively risk-index optimized by immunity specifically includes: N antibodies are generated. When calculating the comprehensive risk index, we need to consider the affinity of all antibodies with the target antigen. The optimized comprehensive risk index Calculation formula: .

[0018] Preferably, outputting the warning level according to the comprehensive risk index specifically includes: According to the comprehensive risk index The risks are divided into four different levels: : Low risk, green warning; : Medium risk, yellow warning; : High risk, orange warning; : Extremely high risk, red warning.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: Strong adaptability: The immune algorithm can dynamically learn new carbon emission patterns; Strong anomaly detection ability: It can quickly identify abnormal changes in carbon emissions; Comprehensive multi-factor analysis: Combining the PSR model to quantitatively analyze the impact of multiple driving factors on carbon emissions; Dynamic optimization: Realize the real-time optimization of complex multi-variable carbon emission systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 : It is a schematic diagram of a carbon emission early warning system of an improved PSR model provided by an embodiment of the present invention; Figure 2 : It is a schematic diagram of the process for preprocessing and standardizing the real-time data provided by an embodiment of the present invention; Figure 3 : It is a schematic diagram of the process of the biological immune algorithm model provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0023] The present invention provides a carbon emission early warning system based on an improved PSR model by a biological immune algorithm, constructs a model with real-time dynamic update, adaptability and comprehensive multi-factor analysis capabilities, so as to achieve accurate prediction and early warning of carbon emissions.

[0024] As Figure 1 , the technical solution adopted by the present invention is as follows: A carbon emission early warning system of an improved PSR model, including a data acquisition module, a PSR model, a biological immune algorithm model, and a warning output module; The data acquisition module obtains real-time data related to carbon emissions through the Internet of Things and a big data platform; The PSR model preprocesses and standardizes the real-time data according to the three dimensions of pressure, state, and response; The bio-immune algorithm model constructs an antibody from the data of the three dimensions of pressure, state, and response, takes the average value of zero carbon emissions or low-risk areas as the target antigen, calculates the affinity between the antibody and the target antigen, and obtains the comprehensively risk index optimized by immunity through the cloning and mutation operations of the antibody; The early warning output module outputs the early warning level according to the comprehensive risk index.

[0025] Preferably, the data acquisition module obtains real-time data related to carbon emissions through the Internet of Things and big data platforms, specifically including: Energy consumption data, industrial production data, urbanization data; Total carbon emissions data, environmental condition data; Policy implementation data, new energy utilization data, technological innovation data.

[0026] Specifically, the data can be collected in the following ways: The Internet of Things sensors installed in industrial facilities and energy enterprises upload real-time data related to carbon emissions; Use meteorological and environmental monitoring satellites to collect carbon dioxide concentration data in urban and industrial areas; Automatically capture policy, statistical, and planning data on public platforms such as governments, statistical bureaus, and energy bureaus; Obtain transaction data of participating enterprises from the carbon trading market and record the usage of carbon emission quotas.

[0027] The following is the specific data refinement table and its processing method of the data acquisition module in the carbon emission early warning system. Each data type provides the collection source, main indicators, collection frequency, processing method, and example data.

[0028] I. Pressure data Data Type Main Indicator Data Source Collection Frequency Processing Method Example Data Energy Consumption Data Coal Consumption (tons) Energy Bureau, Enterprise Sensors Daily Outlier Removal, Normalization 1,200,000 tons Oil Consumption (barrels) Energy Bureau, Industrial Sector Monthly Data Imputation, Standardization Processing 1,500,000 barrels Natural Gas Consumption (cubic meters) Energy Company, Local Government Monthly Missing Values Complemented by Interpolation 32 million cubic meters Industrial Production Data Total Industrial Output Value (100 million yuan) National Bureau of Statistics, Enterprise Reports Monthly Moving Average to Process Fluctuating Data 89 billion yuan Proportion of Heavy Industry (%) National Bureau of Statistics Quarterly Keep Two Decimal Places 45.67% Urbanization Data Urbanization Rate (%) Ministry of Land and Resources, Satellite Remote Sensing System Annually Convert Image Data to Statistical Values 63.5% Newly Added Floor Area (square meters) Ministry of Land and Resources, Local Planning Departments Quarterly Grouped and Statistically Analyzed by Region 10,000,000 square meters II. State data Data Type Main Indicator Data Source Collection Frequency Processing Method Example Data Total Carbon Emission Data Total Carbon Emissions (tons) Carbon Trading Market, Monitoring System Daily Regional Aggregation 250,000 tons Per Capita Carbon Emissions (tons / person) Carbon Trading Market Quarterly Calculate Total Carbon Emissions Divided by Total Population 8 tons / person Environmental Condition Data <![CDATA[CO2 concentration (ppm)]]> Environmental Monitoring Station Hourly Time Series Smoothing Processing 415 ppm PM2.5 Concentration (μg / m³) Environmental Monitoring Station Hourly Use 24-Hour Average Value 60 μg / m³ III. Response data Data Type Main Indicator Data Source Collection Frequency Processing Method Example Data Policy Implementation Data Carbon Trading Market Participation Rate (%) Carbon Trading Market Quarterly National Distribution Statistics 80% Total Carbon Tax Amount (100 million yuan) Finance Department, Enterprise Reports Quarterly Summed by Industry Classification 12 billion yuan New Energy Utilization Data Wind Power Generation (10,000 kWh) Power Grid Company Monthly Data Smoothing Processing 3 million kWh Solar Power Generation (10,000 kWh) New Energy Enterprises, Local Energy Departments Monthly Missing Value Imputation 2 million kWh Proportion of New Energy Generation (%) Power Grid Company Quarterly Calculate the Proportion of Wind and Solar Power Generation in Total Power Generation 15% Technological Innovation Data Number of Emission Reduction Technology Patents (items) Patent Database, Ministry of Science and Technology Annually Extract the Latest Published Patent Quantity 500 items New Technology Coverage Rate (%) National Bureau of Statistics Quarterly Dynamically Calculate the Proportion of the Total Number of Technologically Covered Regions 25% As Figure 2 shown, preferably, the PSR model preprocesses and standardizes the real-time data according to the three dimensions of pressure, state, and response, specifically including: Data cleaning, removing outliers and filling in missing values from the collected raw data; Data standardization, standardizing all data using the range normalization method; Data fusion, aligning pressure, status, and response data along a unified time dimension; Data dynamic update, updating real-time data daily through the Internet of Things and API interfaces.

[0029] Specifically, the collected raw data has redundancy, missing values, or noise, and the following cleaning steps are required: Outlier removal: Use the 3σ principle or Z-score method to delete abnormal data. For example, if the coal consumption data far exceeds the historical average (such as 10 times), it is regarded as abnormal.

[0030] Missing value imputation: For short-term missing data, use interpolation method to complete; for long-term missing data, use moving average method for prediction.

[0031] Specifically, the range normalization method is used to standardize all data. The calculation formula for range normalization is as follows: ; For example, for the normalization of coal consumption, assuming the collected data range is [1,000,000, 1,500,000] tons and the current value is 1,200,000 tons, then after standardization: .

[0032] Specifically, the goal of data fusion is to integrate pressure, status, and response data from different sources into a unified time dimension for subsequent analysis and prediction. This operation is divided into steps such as data alignment, time synchronization, and interpolation.

[0033] 1. Data alignment and time synchronization Since pressure, status, and response data may come from different data sources, these data may be collected at different time points, and the data collection frequencies may also be different. Therefore, time alignment and synchronization are required. The specific operation steps are as follows: Determine a unified time scale: Select a unified time dimension, for example, summarize the data by hour or by minute. Usually, the time scale is determined according to the real-time nature of the data source. If it is very frequent data (such as real-time carbon emissions), it can be summarized by minute; for meteorological data, etc., it may be summarized by hour.

[0034] Unify timestamps: Unify the timestamp format of all data, such as: YYYY-MM-DD HH:MM:SS. Ensure that each data record has a clear time mark. Align data with different time scales. For example, if some data is collected once a minute and other data is collected once an hour, use interpolation or data aggregation methods to convert them to a unified time scale.

[0035] Interpolation or data filling: For data at different times, interpolation methods (such as linear interpolation, spline interpolation, etc.) are used to fill in the missing data to ensure that complete pressure, status, and response data are available at each time point.

[0036] Example: If pressure data is collected every minute and status data is collected every hour, linear interpolation can be performed on the status data to fill in the data points for each minute.

[0037] Data aggregation: For time series data with a high collection frequency (such as carbon emissions), aggregation methods (such as average, maximum, or sum) can be used to summarize the data to a unified time scale.

[0038] Example: If carbon emission data is updated every second and temperature data is updated every hour, the carbon emissions can be converted to hourly units by calculating the average carbon emissions per hour.

[0039] 2. Data fusion and integration Align the pressure, status, and response data according to the timestamp and merge them into a unified dataset to ensure that complete information on pressure, status, and response is available at each time point.

[0040] Specifically, for data dynamic update, to ensure that the carbon emission warning system can reflect the latest carbon emission dynamics in real time, real-time data update needs to be implemented. The following is the specific operation process of dynamic update: 1. Data source and update frequency The update method of data mainly depends on Internet of Things devices and API interfaces: Data from Internet of Things devices: Data such as carbon emissions, temperature, and humidity uploaded in real time by Internet of Things sensors need to be obtained through the Internet of Things platform. The update frequency is usually every minute or every second.

[0041] API interfaces: Data provided by public platforms such as the government, statistical bureaus, and energy bureaus can be obtained through API interfaces. According to the type of data, the update frequency is usually daily or hourly.

[0042] 2. Automatic update mechanism To ensure the real-time and automatic nature of the data, the following data dynamic update process can be designed: Data scraping: Use crawlers or API interfaces to regularly scrape public data from relevant platforms. For example, use the requests library or APIs in Python for data scraping.

[0043] For example, use the API to regularly obtain the latest usage of carbon emission quotas and trading data in the carbon trading market.

[0044] Scheduled task scheduling: Use task scheduling tools (such as Cron Jobs, Airflow, etc.) to set the task cycle. The scheduled task will automatically execute the data scraping script at the set time interval and obtain the latest real-time data.

[0045] Data storage and update: Upload the data scraped each time to the database through the API or the Internet of Things data interface. Each time the data is updated, the database will automatically update the corresponding records or insert new data. Ensure the integration of historical data and real-time data.

[0046] Data cleaning and verification: Perform data cleaning and verification during each update. The content of cleaning includes removing invalid data, correcting error data (such as outliers caused by sensor failures), and filling in missing values.

[0047] Update log and monitoring: After each data update, record the update log to ensure that the system can trace the data source and update time. Monitor the data update process in real time to ensure no errors or anomalies.

[0048] Appendix Figure 3 It is a schematic flow diagram of the bio-immune algorithm model provided by the embodiment of the present invention, specifically as follows: Preferably, the data in three dimensions of pressure, state, and response form an antibody, specifically including: Represent each group of samples in the dataset as an antibody , forming an antibody population , each antibody 's characteristics are composed of data in the dimensions of pressure (P), state (S), and response (R), , where n is the initial population size.

[0049] Preferably, n = 3 or n > 3.

[0050] Specifically, the input vector of each antibody will include the following content: Pressure dimension: Coal consumption, oil consumption, natural gas consumption, total industrial output value, proportion of heavy industry, urbanization rate, newly added building area; State dimension: Total carbon emissions, per capita carbon emissions, CO2 concentration, PM2.5 concentration; Response dimension: Carbon trading market participation rate, total amount of carbon tax, wind power generation, solar power generation, proportion of new energy generation, number of emission reduction technology patents, new technology coverage rate.

[0051] Each data point will be optimized as an antibody. Assuming the antibody is composed of 6 variables (2 variables in each dimension), then the structure of the antibody can be similar to: =[Coal consumption, proportion of heavy industry, total carbon emissions, PM2.5 concentration, carbon trading market participation rate, proportion of new energy].

[0052] Target antigen setting: The target antigen usually represents the ideal state, such as the target carbon emissions, target CO2 concentration, etc. Assume the target is: Indicator Target Antigen Value Normalized Data Target Coal Consumption (tons) 1,100,000 tons 0.2 Target Proportion of Heavy Industry (%) 45% 0.45 Target Total Carbon Emissions (tons) 100,000 tons 0.1 Target PM2.5 Concentration (μg / m³) 35 μg / m³ 0.35 Target Carbon Trading Market Participation Rate (%) 60% 0.6 Target Proportion of New Energy (%) 40% 0.4 These target values will be used as the optimization objectives of the immune algorithm to calculate the affinity between the antibody and the antigen, and find the optimal solution through cloning and mutation operations.

[0053] Suppose we have a specific input data sample as follows:

[0054] Sample 1 Indicator Target Antigen Value Normalized Data Coal Consumption (tons) 1,200,000 tons 0.4 Proportion of Heavy Industry (%) 45.67% 0.46 Total Carbon Emissions (tons) 150,000 tons 0.35 PM2.5 Concentration (μg / m³) 41.8 μg / m³ 0.42 Carbon Trading Market Participation Rate (%) 50% 0.5 Proportion of New Energy (%) 31% 0.31

[0055] Sample 2 Indicator Target Antigen Value Normalized Data Coal Consumption (tons) 1,350,000 tons 0.7 Heavy industry ratio (%) 50% 0.5 Total carbon emissions (tons) 200,000 tons 0.6 Target PM2.5 concentration (μg / m³) 50 μg / m³ 0.5 Carbon trading market participation rate (%) 42% 0.42 New energy proportion (%) 25% 0.25

[0056] Sample 3 Indicator Target antigen value Normalized data Coal consumption (tons) 1,300,000 tons 0.6 Heavy industry ratio (%) 48.23% 0.48 Total carbon emissions (tons) 160,000 tons 0.4 Target PM2.5 concentration (μg / m³) 38 μg / m³ 0.38 Carbon trading market participation rate (%) 48% 0.48 New energy proportion (%) 28% 0.28 The data of the sample can be equivalent to the data of different regions in a to-be-forewarned system, or can be equivalent to the data of a to-be-forewarned system at different time periods. These data will be used as the input of the antibody for the biological immune algorithm to optimize and find the optimal carbon emission management plan.

[0057] Preferably, calculate the affinity between each antibody and the target antigen, specifically including: Take the average value of the zero-carbon emission or low-risk area as the target antigen, and the affinity calculation formula between each antibody and the target antigen is: ; : The th feature value of the th antibody.

[0058] : Target antigen.

[0059] Specifically, calculate the affinity between Sample 1 and the target antigen: ; ; Specifically, calculate the affinity between Sample 2 and the target antigen: ; ; Specifically, calculate the affinity between Sample 3 and the target antigen: ; ; Through affinity calculation, we found that: The affinity of sample 2 (0.76) is slightly higher than that of samples 1 and 3 (0.35 and 0.53), indicating that sample 2 is closer to the target antigen (the ideal carbon emission state).

[0060] This means that sample 2 has a better carbon emission management state relative to the target value and can better meet the target value.

[0061] In addition, the initial risk index can also be calculated through antibody affinity , .

[0062] When the antibody data represents the data of different regions in a system to be warned or the data of a system to be warned at different time periods, the initial risk index only reflects the state at the time of data collection. For system warning, it should be used to identify potential problems and optimal regulation paths in the system. For example, if the optimized risk index shows that the carbon emission risk in a certain area is still high, it means that the current policy or corresponding measures are insufficient and need to be further adjusted. Specifically, the process of calculating the comprehensive risk index is as follows: Preferably, through cloning and mutation operations on antibodies, specifically including: According to the affinity of the antibody, different cloning numbers are proportionally allocated. The higher the affinity of the antibody, the greater the probability of being selected, achieving the effect of replicating high-quality antibodies. ; Among them, is the cloning number of the i-th antibody, N is the total cloning number, n is the initial population size, is the affinity of the i-th antibody; Perform small-range random mutation on the cloned antibody, and the mutation formula is: ; Among them, is the mutation amplitude, is the random perturbation factor, is the antibody after mutation.

[0063] Specifically, take the total cloning number N = 10, and calculate the cloning numbers according to the above 3 samples as follows: , take 2.

[0064] , take 5.

[0065] , take 3.

[0066] The cloning result is: .

[0067] Then, according to the mutation amplitude and the random perturbation factor, the mutated antibody population is obtained: .

[0068] The cloning and mutation operations help the immune algorithm optimize the solution space and improve the search efficiency. By preferentially cloning antibodies with low affinity and locally randomly mutating the cloned antibodies, the algorithm can be prevented from falling into local optimal solutions and the solution set can gradually approach the target antigen.

[0069] In the biological immune algorithm, the calculation of the comprehensive risk index aims to evaluate the closeness between all antibodies optimized by cloning and mutation and the target antigen. The comprehensive risk index combines the affinity information of all antibodies and can reflect the risk status of the overall system.

[0070] Preferably, through the cloning and mutation operations on antibodies, it also includes: Inhibition operation: Remove antibodies with high similarity in the compiled population to avoid redundancy and maintain population diversity; Similarity formula: ; If 5, then remove one of the antibodies.

[0071] Preferably, to obtain the comprehensive risk index after immune optimization, it specifically includes: N antibodies are generated. When calculating the comprehensive risk index, we need to consider the affinity between all antibodies and the target antigen. The optimized comprehensive risk index Calculation formula: .

[0072] Specifically, assuming that we have generated N antibodies through the cloning and mutation operations and the inhibition operation, and calculated their affinities. Next, the calculation formula for the optimized comprehensive risk index R is: ;

[0073] Where: is the affinity between the th antibody and the target antigen.

[0074] is the total number of antibodies.

[0075] This value represents the average affinity of all antibodies relative to the target antigen. The lower the value of the comprehensive risk index, the lower the overall risk of the antibody population and the higher the degree of proximity to the target antigen.

[0076] The comprehensive risk index optimized through cloning and mutation is no longer just an assessment of the current situation, but the optimal risk assessment value adjusted by the immune algorithm.

[0077] Furthermore, through multiple immunological optimizations, the comprehensive risk index is made to meet the convergence conditions, and the obtained optimal antibody data represents the optimal solution of the current system's carbon emission risk among all possible data combinations, which can be used to identify potential problems and optimal regulation paths in the system.

[0078] Preferably, the warning level is output according to the comprehensive risk index, specifically including: According to the comprehensive risk index The risks are divided into four different levels: : Low risk, green warning; : Medium risk, yellow warning; : High risk, orange warning; : Extremely high risk, red warning.

[0079] Specifically, a risk threshold (for example, 0.5) can be set for the comprehensive risk index. If the comprehensive risk index is greater than this threshold, it indicates that the system has a relatively high risk and may require further optimization or adjustment of strategies; if the comprehensive risk index is less than this threshold, it indicates that the system has been optimized to a relatively low risk state and is close to the ideal state.

[0080] It should be noted that the specific values for dividing the risk levels in this application are not unique and can be determined in combination with historical data or through expert experience.

[0081] If the comprehensive risk index is higher than the preset threshold, the immune algorithm can further perform optimization operations, such as: Increasing the diversity of antibodies and adopting stronger mutation operations; Adjusting the cloning operation so that more antibodies with low affinity are cloned; Adjusting the mutation amplitude to explore new solution spaces.

[0082] In this way, the biological immune algorithm can gradually converge to the best carbon emission management strategy.

[0083] The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. An improved carbon emission early warning system based on the PSR model, characterized in that It includes a data acquisition module, a PSR model, a biological immune algorithm model, and an early warning output module. Among them, the data acquisition module is used to obtain real-time data related to carbon emissions through the Internet of Things and big data platform; the PSR model is used to preprocess and standardize the real-time data according to the three dimensions of pressure, state, and response; the biological immune algorithm model is used to form an antibody from the data of the three dimensions of pressure, state, and response, use the average value of zero carbon emissions or low-risk areas as the target antigen, calculate the affinity between the antibody and the target antigen, and obtain the comprehensively risk index after immune optimization through the cloning and mutation operations of the antibody; the early warning output module outputs the early warning level according to the comprehensive risk index.

2. The carbon emission early warning system of an improved PSR model according to claim 1, characterized in that: The data acquisition module is used to obtain real-time data related to carbon emissions through the Internet of Things and big data platform, specifically including: energy consumption data, industrial production data, urbanization data; total carbon emission data, environmental condition data; policy implementation data, new energy utilization data, technological innovation data.

3. An improved PSR model-based carbon emission warning system according to claim 2, characterized in that: The PSR model is used to preprocess and standardize the real-time data according to the three dimensions of pressure, state, and response, specifically including: data cleaning, removing outliers and filling in missing values for the collected original data; data standardization, standardizing all data using the range normalization method; data fusion, aligning the pressure, state, and response data according to the unified time dimension; data dynamic update, updating the real-time data daily through the Internet of Things and API interfaces.

4. An improved carbon emission warning system based on the PSR model according to claim 1, characterized in that: The biological immune algorithm model is used to form an antibody from the data of the three dimensions of pressure, state, and response, specifically including: Represent each group of samples in the dataset as an antibody , forming an antibody population . The characteristics of each antibody are composed of data in the dimensions of pressure (P), state (S), and response (R). , where n is the size of the initial population.

5. An improved carbon emission warning system of the PSR model according to claim 4, characterized in that: calculating the affinity between each antibody and the target antigen, specifically including: using the average value of zero carbon emissions or low-risk areas as the target antigen, and the affinity calculation formula between each antibody and the target antigen is: ； is the k-th eigenvalue of the th antibody; is the k-th eigenvalue of the target antigen; d is the total number of eigenvalues.

6. An improved carbon emission warning system based on the PSR model according to claim 5, characterized in that: through the cloning and mutation operations of the antibody, specifically including: allocating different cloning numbers proportionally according to the affinity of the antibody, ； Among them, is the cloning number of the i-th antibody, N is the total number of clones, and n is the initial population size. is the affinity of the i-th antibody; performing a small-range random mutation on the cloned antibody, and the mutation formula is: ； Among them, is the mutation range, is the random perturbation factor, is the antibody after mutation.

7. An improved PSR model-based carbon emission early warning system according to claim 6, characterized in that: through the cloning and mutation operations of the antibody, it also includes: inhibition operation: removing the antibodies with high similarity in the compiled population; Similarity formula: ; If the first threshold, then remove one of the antibodies.

8. An improved carbon emission warning system based on the PSR model according to claim 7, characterized in that: obtaining the comprehensively risk index after immune optimization, specifically including: Generated N antibodies, comprehensive risk index The calculation formula is: 。 9. An improved carbon emission early warning system of the PSR model according to claim 8, characterized in that: n = 3, d = 6, N = 10.

10. An improved PSR model-based carbon emission warning system according to claim 9, characterized in that: Outputting the early warning level according to the comprehensive risk index, specifically including: According to the comprehensive risk index the risks are divided into four different levels: : Low risk, green warning; : Medium risk, yellow warning; : High risk, orange warning; : Extremely high risk, red warning.

Images