Intelligent park electric power carbon emission prediction method and system based on artificial intelligence

Abstract

The invention discloses a smart park electric power carbon emission prediction method and system based on artificial intelligence, and relates to the field of carbon emission prediction, and the method comprises the steps: obtaining internal and external power purchase activity data and direct power utilization data of a smart park, calculating outsourcing electric power carbon emission and direct electric power carbon emission in the smart park by using an electric power carbon emission accounting standard to obtain a total amount of carbon emission; constructing a relation model between the total carbon emission amount and the energy consumption of the electric equipment, predicting the total power carbon emission amount generated by the electric equipment in the smart park under any operation strategy, and judging the emission efficiency corresponding to the total power carbon emission amount; and generating a power use distribution model based on the total power carbon emission amount and the emission efficiency, and determining a power carbon emission reduction strategy in the smart park. The method can optimize the power use structure of the park according to the total power carbon emission amount and the emission efficiency, helps the smart park to achieve a low-carbon target, and optimizes the energy use.

Description

Technical Field

[0001] The present invention relates to the field of carbon emission prediction, and particularly to a method and system for predicting the carbon emissions of electric power in a smart park based on artificial intelligence. Background Art

[0002] The carbon emission prediction models in the prior art cannot be combined with the energy consumption conditions of electrical equipment for analysis, and it is difficult to reveal the carbon emission characteristics under different energy consumption levels. This limits the optimization effect of low-carbon strategies. Therefore, there are potential negative impacts. Due to factors such as a decrease in equipment energy efficiency and load fluctuations, the carbon emissions may instead increase, thus reducing the actual effect of low-carbon strategies. Summary of the Invention

[0003] In order to solve the above problems, the present invention proposes a method and system for predicting the carbon emissions of electric power in a smart park based on artificial intelligence, so as to achieve the purpose of optimizing the power usage structure of the park according to the total carbon emissions of electric power and the emission efficiency.

[0004] To achieve the above purpose, the present invention adopts the following technical solutions: In the first aspect, the present invention provides a method for predicting the carbon emissions of electric power in a smart park based on artificial intelligence, and the method includes: Obtain the data of the external power purchase activities and direct power consumption in the smart park, and calculate the carbon emissions of the externally purchased power and the direct power consumption in the smart park by using the carbon emission accounting standard for electric power to obtain the total carbon emissions; Construct a relationship model between the total carbon emissions and the energy consumption of electrical equipment, predict the total carbon emissions of the electrical equipment in the smart park under any operation strategy, and judge the emission efficiency corresponding to the total carbon emissions of the electric power; Generate a power usage allocation model based on the total carbon emissions of electric power and the emission efficiency, determine the carbon emission reduction strategy for the electric power in the smart park, and evaluate the energy conservation and emission reduction effect of the smart park after using the carbon emission reduction strategy for the electric power according to the total carbon emissions of the electric power.

[0005] Preferably, constructing a relationship model between the total carbon emissions and the energy consumption of electrical equipment, predicting the total carbon emissions of the electrical equipment in the smart park under any operation strategy, and judging the emission efficiency corresponding to the total carbon emissions of the electric power includes: Obtain the data of the total carbon emissions and the energy consumption of electrical equipment corresponding to the historical operation and maintenance process in the smart park, and perform dimensionality reduction and decoupling processing on the power consumption data to obtain a number of main dimension vectors that are orthogonally distributed; Based on the mean clustering technology, sequentially constrain each main dimension vector, randomly combine to generate energy consumption conditions, and combine with the carbon emissions to analyze the comprehensive energy consumption level of the energy consumption conditions; Obtain the relationship model between carbon emissions and the energy consumption of electrical equipment according to the Pearson correlation coefficient between the main dimension vector and the comprehensive level of energy consumption, and generate the function relationship result between carbon emissions and energy consumption; Based on the feature selection algorithm and the Shapley additive explanation technique, obtain the key factors affecting the energy consumption change of electrical equipment under any operating strategy, generate a numerical set to predict the energy consumption value, and judge the total amount of electricity carbon emissions; Use the carbon emission factor to analyze the corresponding duration when the total amount of electricity carbon emissions is completely discharged, and evaluate the emission efficiency of the power supply equipment in the smart park according to the duration result.

[0006] Preferably, based on the feature selection algorithm and the Shapley additive explanation technique, obtaining the key factors affecting the energy consumption change of electrical equipment under any operating strategy, generating a numerical set to predict the energy consumption value, and judging the total amount of electricity carbon emissions includes: Select the influencing factors including the load fluctuation mode, the environmental impact mode, and the operating cycle mode according to the operating strategy of the electrical equipment, and judge the total influence component value of the influencing factors on the electrical equipment; Obtain the characteristic information of the total influence component value, and use the characteristic information as the feature set and the total influence component value as the data set to initialize the number of iterations and the selected number of neighbors; Randomly select a set of influence component values of an influencing factor from the data set as the selection object, and at the same time select the influence component values of the other two influencing factors in the same dimension as the heterogeneous objects, and repeat the selection until the number of iterative selections of the object is reached; Analyze the distance between the selected object and the heterogeneous object in terms of characteristic information, and combine the Shapley additive explanation technique to determine the key factors affecting the energy consumption change, and generate a numerical set of the key factors; Calculate the energy consumption of the electrical equipment under the numerical set, and judge the total amount of electricity carbon emissions according to the function relationship between the energy consumption of the electrical equipment and the carbon emissions.

[0007] Preferably, the calculation formula for the total influence component value is: ; In the formula, β represents the total influence component value, β F represents the influence component value under the load fluctuation mode, β Y represents the influence component value under the operating cycle mode, β H represents the influence component value under the environmental impact mode, a represents the accumulation order, c b represents the regression coefficient of the influence component under the load fluctuation mode, D b represents the bThe load fluctuation value under an operating strategy, f 1e and f 2e represents the regression coefficient of the influencing component in the operating cycle mode, t represents the cumulative number of days of the current operating strategy from the start observation date to the end observation date, e represents a constant, i h represents the regression coefficient of the influencing component in the environmental impact mode, J h represents the h th aging factor.

[0008] Preferably, analyze the distance between the selected object and the heterogeneous object in terms of characteristic information, and combine the Shapley additive explanation technique to determine the key factors affecting the energy consumption change, and generate a numerical set of key factors including: Analyze the importance value of the selected object based on the distance result, and repeat the analysis of the importance value of the object according to the number of iterative selections, and take the maximum importance value as the corresponding characteristic weight; Analyze the Shapley value of each influencing factor, calculate the average of the absolute values of the Shapley values to obtain the characteristic contribution value of each influencing factor, and collectively call the combination of the characteristic contribution value and the characteristic weight the factor contribution value; Analyze the cumulative contribution rate of each influencing factor contribution value in the total factor contribution value, calculate the cumulative contribution rate difference of each influencing factor according to the cumulative contribution rate, and compare the cumulative contribution rate difference with the cumulative contribution rate difference threshold; Select the influencing factors with a cumulative contribution rate difference greater than the cumulative contribution rate difference threshold as the key factors affecting the energy consumption change, and generate a numerical set of the electrical equipment under different operating strategies according to the key factors.

[0009] Preferably, use the carbon emission factor to analyze the corresponding duration when the total power carbon emission is completely discharged, and evaluate the emission efficiency of the power supply equipment in the smart park according to the duration result, including: Regard the total power carbon emission as the total direct power consumption carbon emission and the total purchased power carbon emission respectively, set the carbon emission rate according to the corresponding carbon emission factor, and obtain the emission duration of the direct power consumption equipment and the emission duration of the purchased power equipment; Obtain the decision-making factors affecting the carbon emissions of the direct power consumption equipment and the purchased power equipment, establish a direct influence relationship matrix, and perform a normalization process on the direct influence relationship matrix to obtain a normalized influence matrix; Obtain the comprehensive influence matrix of the normalized influence matrix, add the factors in each row of the comprehensive influence matrix to obtain the influence degree, and at the same time add the factors in each column of the comprehensive influence matrix to obtain the degree of being influenced; Calculate the centrality and cause degree of each factor based on the influence degree and the influenced degree, determine the evaluation index system of the emission duration according to the calculation results, and evaluate the relative efficiency of carbon emissions of direct electricity-consuming equipment and purchased electricity equipment respectively by combining the data envelopment analysis model.

[0010] Preferably, generate a power usage allocation model based on the total power carbon emissions and emission efficiency, determine the power carbon emission reduction strategy in the smart park, and evaluate the energy conservation and emission reduction effect of the smart park after the application of the power carbon emission reduction strategy, including: Obtain the electricity consumption details, time granularity data and historical electricity consumption data in the smart park, and combine the time series model to identify the trend prediction and seasonal changes of electricity consumption, and determine the peak and off-peak periods of electricity consumption; Analyze the total electricity consumption cost and regional electricity consumption cost in the smart park according to the electricity price data, and analyze the equipment usage situation during the peak electricity consumption period in combination with the electricity consumption details to obtain the electricity demand in the smart area; Construct a multi-objective integrated planning model with the minimum power generation cost as the objective function, and generate the optimal output combination corresponding to the dynamic allocation of direct electricity consumption and purchased electricity by combining the total power carbon emissions and emission efficiency; Obtain the direct electricity consumption and purchased electricity consumption in the smart park according to the optimal output combination, determine the power carbon emission reduction strategy in the smart park, and obtain the total power carbon emissions of the smart park after the application of the optimal output combination; Compare the difference between the total power carbon emissions obtained after the application of the optimal output combination and the initial total power carbon emissions, and evaluate the energy conservation and emission reduction effect of the smart park after the application of the power carbon emission reduction strategy.

[0011] Preferably, construct a multi-objective integrated planning model with the minimum power generation cost as the objective function, and generate the optimal output combination corresponding to the dynamic allocation of direct electricity consumption and purchased electricity by combining the total power carbon emissions and emission efficiency, including: Calculate the total power generation cost under the condition of meeting the electricity demand based on the electricity demand and electricity consumption cost in the smart park, and analyze the power generation cost value required for direct electricity consumption or purchased electricity to meet the electricity demand at any time period in the smart park according to the total power generation cost; Take the minimum power generation cost value and the minimization of carbon emissions as the objective function, and the capacity of the power generation equipment as the constraint condition, and combine the single objective function and the constraint condition to construct a multi-objective integrated planning model; Allocate weights to the objective function by the weighted sum method, and find a compromise based on Pareto optimality to obtain the output balance point of the multi-objective integrated planning model, and determine the combined allocation optimization result of the multi-objective integrated planning; Taking the total power carbon emissions and emission efficiency as the basis for judging the fairness of the output combination, the fairness of the combined allocation optimization result is verified using the calculation formula for the fairness of the output combination allocation, and the optimal output combination corresponding to the dynamic allocation of direct power consumption and purchased power is output according to the fairness result.

[0012] Preferably, the calculation formula for the fairness of the output combination allocation is: ; In the formula, max Z represents the fairness of the output combination allocation, α represents the electricity demand of the electrical equipment in the smart park, A represents the set of electricity demands of the electrical equipment in the smart park, δ represents the time point, B represents the set of time points, ε represents the relative efficiency of carbon emissions, N represents the set of carbon emission efficiencies, σ represents the transmission point in the power supply process in the smart park, Q represents the set of transmission points in the power supply process in the smart park, ρ represents the transmission path in the power supply process in the smart park, P represents the set of transmission paths in the power supply process in the smart park, S σ,α,δ,ρ represents the δ th σ transmission point in the ρ th α transmission path to meet the electricity demand of the R α,δ represents the capacity required for the power generation equipment when meeting the electricity demand of the δ th α electrical equipment in the λ α,δ At the δ th α time period, the required power transmission duration when meeting the electricity demand of the

[0013] In the second aspect, the present invention also provides a smart park power carbon emissions prediction system based on artificial intelligence, and the system includes: A total carbon emissions calculation unit, configured to obtain the external power purchase activity data and direct power consumption data in the smart park, and calculate the external power purchase carbon emissions and direct power consumption carbon emissions in the smart park using the power carbon emissions accounting standard to obtain the total carbon emissions; A carbon emission efficiency judgment unit is used to construct a relationship model between the total carbon emissions and the energy consumption of electrical equipment, predict the total power carbon emissions generated by the electrical equipment in the smart park under any operating strategy, and judge the emission efficiency corresponding to the total power carbon emissions. An energy-saving strategy execution evaluation unit is used to generate a power usage allocation model based on the total power carbon emissions and the emission efficiency, determine the power carbon emission reduction strategy in the smart park, and evaluate the energy-saving and emission-reduction effect of the smart park after using the power carbon emission reduction strategy according to the total power carbon emissions.

[0014] The beneficial effects of the present invention are as follows: 1. By obtaining the data of the external power purchase activities and the direct power consumption data in the smart park, the present invention can accurately calculate the total carbon emissions, provide reliable data support for subsequent analysis, construct a relationship model between the total carbon emissions and the energy consumption of electrical equipment, can scientifically predict the power carbon emissions under different operating strategies, and at the same time generate a power usage allocation model according to the total power carbon emissions and the emission efficiency, optimize the power usage structure of the park, help the smart park achieve the low-carbon goal, optimize the energy usage, and improve the overall operating efficiency, and through regular evaluation and adjustment, ensure the long-term effect of the energy-saving and emission-reduction measures.

[0015] 2. By obtaining the real total carbon emissions and the energy consumption data of electrical equipment during the historical operation and maintenance process in the smart park, the present invention reflects the energy consumption performance under the operating conditions of various different equipment in the smart park, and combines the energy consumption conditions and the carbon emissions for analysis, can reveal the carbon emission characteristics under different energy consumption levels, help to understand the total carbon emissions generated under specific conditions, and thus provide a basis for the carbon emission reduction strategy.

[0016] 3. By analyzing the historical power consumption data through a time series model, the present invention can accurately identify the trends and seasonal variations of the power consumption, and at the same time construct a multi-objective integrated planning model with the minimum power generation cost as the objective function, generate the optimal output combination corresponding to the dynamic allocation of direct power consumption and external power purchase in combination with the total power carbon emissions and the emission efficiency, so that while ensuring the lowest power generation cost, it can also take into account the minimization of carbon emissions, and thus can minimize carbon emissions to the greatest extent while ensuring power supply and optimize the energy usage efficiency. Description of the Drawings

[0017] The specification drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0018] Figure 1 It is a flowchart of a method for predicting the power carbon emissions in a smart park based on artificial intelligence according to an embodiment of the present invention. Figure 2It is a schematic block diagram of an intelligent park power carbon emission prediction system based on artificial intelligence according to an embodiment of the present invention.

[0019] In the figure: 1. Total carbon emission calculation unit; 2. Carbon emission efficiency judgment unit; 3. Energy-saving strategy execution evaluation unit. Specific implementation manners

[0020] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0021] It should be noted that the following detailed description is exemplary and is intended to provide further description of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0022] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0023] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0024] Please refer to Figure 1 , the present invention provides an intelligent park power carbon emission prediction method based on artificial intelligence, and the method includes: Step S1, obtain the external power purchase activity data and direct power consumption data of the intelligent park, and calculate the external power purchase carbon emissions and direct power consumption carbon emissions in the intelligent park by using the power carbon emission accounting standard to obtain the total carbon emissions.

[0025] It should be explained that in the process of obtaining the total carbon emissions, based on the latest released international, national, and local provincial and municipal power carbon emission accounting standards, the power carbon emission accounting method and the reference value of the power carbon emission factor are determined, and a power carbon emission algorithm model for external power purchase during the operation and maintenance period of the intelligent park is constructed.

[0026] In the process of calculating the total carbon emissions, it is necessary to construct an artificial intelligence spatio-temporal prediction model sample library for the electricity carbon emissions in the smart park. The goal of constructing the sample library is to collect the sample data for predicting the electricity carbon emissions in different regions (spatial dimension) and different time periods (time dimension) in the smart park.

[0027] Collect electricity consumption data from various IoT weak current sensors deployed in the park to obtain the real-time electricity consumption data and historical energy consumption data of the equipment. Calculate the electricity carbon emission data by selecting the corresponding electricity carbon emission factors. In addition, it is necessary to collect the environmental parameter information in the park for assisting the model prediction, including meteorological data such as temperature, humidity, wind speed, weather, season, etc., which may affect the electricity demand; in addition, the operation information of the park, such as weekdays, weekends, holidays, etc., all of which will have an impact on the electricity consumption pattern. Step S2: Construct a relationship model between the total carbon emissions and the energy consumption of the electrical equipment, predict the total electricity carbon emissions generated by the electrical equipment in the smart park under any operation strategy, and judge the emission efficiency corresponding to the total electricity carbon emissions.

[0028] In one embodiment, in the process of constructing a relationship model between the total carbon emissions and the energy consumption of the electrical equipment, predicting the total electricity carbon emissions generated by the electrical equipment in the smart park under any operation strategy, and judging the emission efficiency corresponding to the total electricity carbon emissions, the corresponding total carbon emissions and energy consumption data of the electrical equipment in the historical operation and maintenance process of the smart park can be obtained, and the dimensionality reduction and decoupling processing are performed on the electricity consumption data to obtain a number of main dimension vectors that are orthogonally distributed; based on the mean clustering technology, each main dimension vector is sequentially constrained, and the energy consumption conditions are randomly combined and generated, and the comprehensive energy consumption level of the energy consumption conditions is analyzed in combination with the carbon emissions; according to the Pearson correlation coefficient between the main dimension vector and the comprehensive energy consumption level, the relationship model between the carbon emissions and the energy consumption of the electrical equipment is obtained, and the function relationship result between the carbon emissions and the energy consumption is generated; based on the feature selection algorithm and the Shapley additive explanation technology, the key factors affecting the energy consumption change under any operation strategy of the electrical equipment are obtained, and a numerical set is generated to predict the energy consumption value and judge the total electricity carbon emissions; use the carbon emission factor to analyze the corresponding duration when the total electricity carbon emissions are completely discharged, and evaluate the emission efficiency of the power supply equipment in the smart park according to the duration result.

[0029] It should be explained that in the process of obtaining the relationship model between the carbon emissions and the energy consumption of the electrical equipment, first obtain the historical operation and maintenance data of the smart park, including the total carbon emissions (calculated by the electricity carbon emission factor) and the energy consumption data related to all electrical equipment in the park. After obtaining the original data, it is necessary to clean the data, process the missing values, outliers and problems with inconsistent data formats. For the energy consumption data, it is usually necessary to organize it according to the time granularity (such as hourly, daily or monthly).

[0030] To simplify the electricity consumption data and remove redundant information, dimensionality reduction methods such as principal component analysis (PCA) or independent component analysis (ICA) can be used. PCA can map multi-dimensional data to a lower-dimensional space and retain most of the data variance.

[0031] The specific steps are as follows: First, standardize the energy consumption data so that the mean of each feature is 0 and the standard deviation is 1; calculate the covariance matrix of the standardized data to reflect the correlation between various features; perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors, and the eigenvectors are the main dimensions; select the first few principal components (i.e., the main dimension vectors with larger eigenvalues) to represent the energy consumption data; the main dimension vectors generated by PCA are orthogonal, that is, uncorrelated, which provides a basis for subsequent analysis; obtain several main dimension vectors, representing different aspects of the electricity consumption data, and these vectors are orthogonal, that is, the vectors are numerically independent of each other, avoiding information redundancy.

[0032] Take each main dimension vector as a data point and perform mean clustering analysis in multiple dimensions to find data groups with similarity in the energy consumption data. For example, K-means clustering can be performed according to the numerical characteristics of each main dimension vector, and devices with similar energy consumption patterns can be classified into one category.

[0033] At the same time, select the best K value, that is, the number of clusters, according to experience or the Silhouette Method. At the same time, randomly select K center points. According to the distance metric of the main dimension vectors (such as Euclidean distance), assign each data point to the closest center point. Based on the data points in each cluster, recalculate the center point of each cluster until the clustering result converges. Based on the result of mean clustering, randomly combine different categories of energy consumption patterns (i.e., each category represents an energy consumption condition), and combine these conditions with the carbon emissions according to the power carbon emission data. Each combination will generate a specific energy consumption level and correspond to a carbon emission. Combine the carbon emissions of each energy consumption condition with the comprehensive energy consumption level (i.e., the sum of energy consumption in all dimensions) for analysis to help understand the energy consumption under different equipment configurations and operating modes and its impact on carbon emissions.

[0034] The Pearson correlation coefficient is used to measure the linear correlation between two variables, and its value ranges from -1 to 1. The closer the value is to 1, the stronger the positive correlation. At the same time, use the Pearson correlation coefficient to calculate the correlation between the main dimension vectors (energy consumption) and the comprehensive energy consumption level to help understand which main dimensions have a greater impact on the total energy consumption and reveal the relationship between energy consumption changes and carbon emissions.

[0035] Pearson correlation coefficient ω Calculation formula: ; In the formula,x i and y i respectively represent the values of the main dimension vector and the comprehensive energy consumption level. and represents the mean value of the corresponding variable.

[0036] Based on the calculated Pearson correlation coefficient, a relationship model between carbon emissions and the energy consumption of electrical equipment can be constructed. For example, if the correlation between the main dimension and carbon emissions is strong, it indicates that this dimension plays a key role in carbon emission control. Using these correlations, a function model can be generated to describe the quantitative relationship between carbon emissions and energy consumption. If the linear correlation is strong, it can be represented by a linear regression model.

[0037] In one embodiment, in the process of obtaining the key factors affecting energy consumption changes under any operating strategy of electrical equipment based on the feature selection algorithm and the Shapley additive explanation technique, and generating a numerical set to predict the energy consumption value and judge the total power carbon emissions, the influencing factors including the load fluctuation mode, the environmental impact mode, and the operating cycle mode can be selected according to the operating strategy of the electrical equipment, and the total influence component value of the influencing factors on the electrical equipment can be judged; the characteristic information of the total influence component value is obtained, and the characteristic information is used as the feature set, and the total influence component value is used as the data set to initialize the number of iterations and the number of selected neighbors; a set of influence component values of an influencing factor is randomly selected from the data set as the selection object, and at the same time, the influence component values of the same dimension of the other two influencing factors are selected as the heterogeneous objects, and the selection is repeated until the number of iterative selections of the object is reached; the distance between the selection object and the heterogeneous object in the characteristic information is analyzed, and the key factors affecting energy consumption changes are determined in combination with the Shapley additive explanation technique, and a numerical set of the key factors is generated; the energy consumption of the electrical equipment under the numerical set is calculated, and the total power carbon emissions are judged according to the function relationship between the energy consumption of the electrical equipment and the carbon emissions.

[0038] Among them, the calculation formula for the total influence component value is: ; In the formula, β represents the total influence component value, β F represents the influence component value under the load fluctuation mode, β Y represents the influence component value under the operating cycle mode, β H represents the influence component value under the environmental impact mode, a represents the accumulation order, c b represents the regression coefficient of the influence component under the load fluctuation mode, D b represents the bThe load fluctuation value under a running strategy f 1e and f 2e represents the regression coefficient of the influencing component under the operating cycle mode t represents the cumulative number of days from the start observation date to the end observation date of the current running strategy e represents a constant i h represents the regression coefficient of the influencing component under the environmental impact mode J h represents the h th aging factor

[0039] It should be noted that in the process of obtaining the key factors affecting energy consumption changes, the influencing factors are first selected, including the load fluctuation mode, the environmental impact mode, and the operating cycle mode Specific implementation Load fluctuation pattern: The load fluctuation pattern represents the change of the device load. For example, the load fluctuation of the device in different time periods. This pattern can be constructed based on historical power consumption data and the change of the device load. By using methods such as time series analysis or FFT (Fast Fourier Transform), the frequency characteristics of the device load are extracted to obtain the load fluctuation pattern

[0040] Environmental impact pattern: The environmental impact pattern considers the influence of external environmental factors such as temperature, humidity, and air pressure on the device operation efficiency and energy consumption. For example, an increase in temperature may lead to an increase in the energy consumption of the device

[0041] Operating cycle pattern: The operating cycle pattern refers to the operating law of the device within a certain cycle. For example, the energy consumption of some devices is relatively high at startup and relatively low during periodic operation. Considering the start-stop frequency and continuous working duration of the device, this pattern can help analyze the energy consumption fluctuation of the device in different working states. According to the start-stop records of the device, the operating cycle pattern is extracted. Different cycles can be divided by using clustering algorithms (such as K-means), and the energy consumption changes of the device under different cycles can be analyzed

[0042] For each influencing factor (load fluctuation, environmental impact, and operating cycle), the Shapley Additive Explanations (SHAP) technique is used to evaluate the contribution of this factor to the device energy consumption. The SHAP method comes from cooperative game theory and is used to measure the marginal contribution of features to the model output

[0043] Shapley Value: For each influencing factor, the SHAP method calculates a "contribution score" or influence value by computing the contribution of this factor in different subsets. This influence value represents the relative influence of the factor on the change in the total energy consumption of the device. Calculation method of SHAP: Feature Selection: First, select the specific features of each influencing factor (load fluctuation, environmental impact, operating cycle). For example, load fluctuation can include load fluctuation frequency and amplitude; environmental impact can include temperature and humidity; operating cycle can include start time and duration, etc.

[0044] Calculate the Shapley Value: For each feature, calculate its influence value in all possible feature combinations. This means combining this feature with other features and calculating the marginal contribution of this feature by comparing the impact on the model output (energy consumption) with and without this feature.

[0045] Therefore, for each influencing factor, a corresponding influence component value can be obtained finally, which reflects its influence degree on energy consumption. Then, take the Shapley value of each influencing factor as the influence component value, and sum up the influence component values of all influencing factors to get the "total influence component value". This value represents the change in the device energy consumption under the combined action of various influencing factors.

[0046] During the initialization of the feature set and the data set, the feature set contains the feature data of all influencing factors (such as load fluctuation patterns, environmental impact patterns, operating cycle patterns). These feature data constitute the input for subsequent model training; take the total influence component value as the target value (label) of the data set for training the model. The data set will be used in the subsequent iterative process for optimizing the modeling of the influencing factors and energy consumption changes. Set the initial number of iterations and the number of nearest neighbors selected for subsequent feature selection and optimization algorithms (such as KNN, support vector machine, etc.). The number of iterations determines the training accuracy of the model, and the number of nearest neighbors is used to adjust the distance metric during training, thereby affecting the performance of the model.

[0047] Randomly select a group of influence component values of influencing factors from the data set as the selection object, that is, select a sample of a specific influencing factor. For the other two influencing factors, randomly select their influence component values as heterogeneous objects in the same dimension (such as the same time, environment, or cycle state). It can be understood as combining different behavior patterns (load, environment, cycle) together for comparative analysis. Through iteration, continuously randomly select a group of "selection objects" and two groups of "heterogeneous objects" from the data set. In each iteration, the influencing factor and its corresponding influence component value will be used as the data points for training the model. This iterative process can help generate more feature combinations and enhance the generalization ability of the model to ensure that the model can adapt to various different operating strategies and different electricity consumption environments.

[0048] Through iterative selection, a dataset containing multiple selected objects and heterogeneous objects is finally obtained. Each data point represents a specific device operation mode, and these data points will be used to train a machine learning model to help predict energy consumption changes and carbon emissions.

[0049] In one embodiment, during the process of analyzing the distance between the selected objects and the heterogeneous objects in terms of feature information, and combining the Shapley additive explanation technique to determine the key factors affecting energy consumption changes and generate a numerical set of key factors, the importance value of the selected objects can be analyzed based on the distance results, and the importance value of the analysis objects can be repeated according to the number of iterative selections. The maximum importance value is used as the corresponding feature weight; the Shapley values of each influencing factor are analyzed, and the average of the absolute values of the Shapley values is obtained to get the feature contribution value of each influencing factor, and the feature contribution value and the feature weight are collectively referred to as the factor contribution value; the cumulative contribution rate of each influencing factor contribution value in the total factor contribution value is analyzed, and the cumulative contribution rate difference of each influencing factor is calculated according to the cumulative contribution rate. The cumulative contribution rate difference is compared with the cumulative contribution rate difference threshold; the influencing factors with a cumulative contribution rate difference greater than the cumulative contribution rate difference threshold are selected as the key factors affecting energy consumption changes, and a numerical set of the electrical equipment under different operation strategies is generated according to the key factors.

[0050] In one embodiment, during the process of analyzing the corresponding duration when the total electricity carbon emissions are completely discharged by using the carbon emission factor and evaluating the emission efficiency of the power supply equipment in the smart park according to the duration result, the total electricity carbon emissions can be respectively regarded as the total direct electricity carbon emissions and the total purchased electricity carbon emissions, and the carbon emission rate is set according to the corresponding carbon emission factor to obtain the emission duration of the direct electricity equipment and the emission duration of the purchased electricity equipment; the decision-making factors affecting the carbon emissions of the direct electricity equipment and the purchased electricity equipment are obtained, a direct influence relationship matrix is established, and the direct influence relationship matrix is normalized to obtain a normalized influence matrix; the comprehensive influence matrix of the normalized influence matrix is obtained, the factors in each row of the comprehensive influence matrix are added to get the influence degree, and at the same time, the factors in each column of the comprehensive influence matrix are added to get the influenced degree; based on the influence degree and the influenced degree, the centrality and cause degree of each factor are calculated, and according to the calculation results, an emission duration evaluation index system is determined, and the relative efficiency of the carbon emissions of the direct electricity equipment and the purchased electricity equipment is respectively evaluated by combining the data envelopment analysis model.

[0051] It should be noted that during the process of relative efficiency evaluation, the total electricity carbon emissions of the smart park can be split into two parts: Total direct electricity carbon emissions: The carbon emissions generated by the electricity directly provided by the self-owned power generation equipment in the park (such as wind energy, solar energy, coal-fired generators, etc.); Total purchased electricity carbon emissions: The carbon emissions generated by the electricity purchased through the external power grid.

[0052] For each power source (direct power consumption and purchased electricity), different carbon emission factors need to be set. These factors represent the carbon emissions generated per unit of power consumption. Generally, the carbon emission factor of purchased electricity depends on the overall energy composition of the power grid (such as the proportion of thermal power generation, etc.), while the carbon emission factor of direct power consumption may be set based on factors such as equipment type and operating efficiency.

[0053] The carbon emission rate can be calculated by the following formula: Carbon emission rate = Time / Total carbon emissions. Based on the total carbon emissions of direct power consumption and purchased electricity and the set carbon emission factors, the carbon emission rates of these two parts of power supply can be calculated.

[0054] Identification of decision factors: Decision factors for carbon emissions of direct power consumption equipment: For example, power demand of the equipment, load fluctuation, operating time, equipment energy efficiency, etc.; Decision factors for carbon emissions of purchased electricity equipment: For example, power source of the power grid (whether mainly coal-fired power), external electricity price, external power grid load, etc.

[0055] The influence relationship matrix describes the mutual influence relationships among decision factors. The elements in the matrix represent the direct influence degree between decision factors. The larger the value, the stronger the influence between two factors. For example, if the energy efficiency of the equipment has a greater impact on the carbon emissions of direct power consumption equipment, a larger weight will be assigned to the corresponding position in the matrix. The purpose of normalization is to standardize all values in the influence relationship matrix to the same dimension (such as values between 0 and 1) for better comparison of the influence intensities of different factors; The comprehensive influence matrix can be obtained by matrix weighted summation of the normalized influence matrix, reflecting the mutual influence degree among all factors: The influence degree of each row factor (decision factor) on other factors is obtained by summing each row; The degree of being influenced of each column factor (decision factor) by other factors is obtained by summing each column.

[0056] Centrality is an index to measure the importance of a factor in the influence network. The greater the influence degree, the stronger the influence of this factor, so it can be considered that this factor is "central"; Causality is a measure of the degree to which a factor affects other factors as a "cause", measured based on the "degree of being influenced" of this factor, that is, the degree to which this factor affects other factors.

[0057] Based on the calculation results of the influence degree and the degree of being influenced, an evaluation index system for emission duration is established. This system can include factors such as influence degree, causality, carbon emission rate, etc., comprehensively reflecting the carbon emission efficiency of the equipment. Data Envelopment Analysis (DEA) is a technology based on efficiency evaluation, which evaluates the relative efficiency by comparing the inputs and outputs of different equipment.

[0058] In this embodiment, the DEA model can be used to evaluate the carbon emission efficiency of direct electricity-consuming equipment and externally purchased electricity equipment. Through the DEA model, the relative efficiency between different equipment (input carbon emissions and output electricity) is compared to identify the most efficient equipment operation strategy.

[0059] DEA model calculation: Inputs: total carbon emissions, total energy consumption, etc.; Outputs: electricity supply, system efficiency, etc.

[0060] DEA will calculate the relative efficiency value of each equipment based on these input and output data, providing a reference for the best practices of equipment carbon emissions and energy consumption.

[0061] To facilitate the understanding of the above technical solution of the present invention, the following will provide a detailed description of the evaluation method of the emission efficiency in the actual process of the present invention.

[0062] Step 1: Mean clustering and energy consumption condition analysis; (1) Data collection and preprocessing: Collect the historical total carbon emissions and energy consumption data of the intelligent park's electricity-consuming equipment. The following are some example data: Total carbon emissions: Calculated through the electricity carbon emission factor, with the unit of ton CO 2 .

[0063] Assume that the electricity carbon emission factor of the intelligent park is 0.4 ton CO 2 / kWh (thousand kilowatt-hours). If the intelligent park consumes 100,000 kWh of electricity in a month, the total carbon emissions are: Total carbon emissions = 100,000 kWh × 0.4 (ton CO 2 / kWh) = 40 tons CO 2 ; Electricity-consuming equipment energy consumption data: Assume that there are the following types of equipment in the park: Air conditioning system: The energy consumption in January is 10,000 kWh, and in February is 12,000 kWh; Lighting system: In January is 5,000 kWh, and in February is 4,800 kWh; Elevator system: In January is 3,500 kWh, and in February is 3,700 kWh; Computer equipment: In January is 6,000 kWh, and in February is 6,100 kWh; Clean the energy consumption data collected from electrical equipment to handle issues such as missing values, outliers, and inconsistent data formats. For example, if the energy consumption data of certain equipment is missing in a specific month, it needs to be filled (methods such as mean filling and previous value filling can be used). Outliers (such as extreme peak values) can be judged based on the equipment operation conditions whether they are data entry errors or abnormal fluctuations and adjusted accordingly.

[0064] To simplify the electrical energy consumption data and remove redundant information, principal component analysis (PCA) is used to reduce the dimensionality of the data. Assume the electrical data has the following characteristics: Load fluctuation frequency: The number of fluctuations of the equipment load per unit time; Energy consumption: The monthly electricity consumption of the equipment (unit: kWh); Environmental factors: For example, the impact of temperature and humidity on energy consumption; Through PCA, perform eigenvalue decomposition on the covariance matrix to obtain the principal dimension vectors. Select the first few principal components, which retain most of the variance in the data. Assume two main components (principal dimension vectors) are obtained for subsequent analysis. Perform mean clustering analysis on each principal dimension vector to find similar energy consumption patterns. Assume K = 3 clusters are selected for clustering analysis and the K - means algorithm is used.

[0065] In the clustering process, for each principal dimension vector, randomly select the initial center point, use the Euclidean distance metric to assign the data points to the closest center point, and recalculate the center point of the cluster until the clustering result converges. Through clustering analysis, three typical energy consumption working conditions are identified: Low - energy consumption mode: The equipment operates at a low load state with low energy consumption; Medium - energy consumption mode: The equipment operates at a standard load state with moderate energy consumption; High - energy consumption mode: The equipment operates at a high load state with high energy consumption; Combine each energy consumption mode with the corresponding carbon emissions to analyze the impact of energy consumption working conditions on carbon emissions.

[0066] Step 2: Analysis of key factors affecting energy consumption changes; According to the PCA dimensionality reduction results and clustering analysis, calculate the Pearson correlation coefficient between each principal dimension vector and carbon emissions. Assume the correlation coefficients obtained through calculation are: Correlation coefficient between principal dimension 1 and carbon emissions: 0.85; Correlation coefficient between principal dimension 2 and carbon emissions: 0.75; According to the Pearson correlation coefficient, construct a relationship model between carbon emissions and energy consumption. Assume an LSTM model is used for modeling: Carbon emissions = τ 1 × principal dimension 1 + τ 2 × principal dimension 2 + ν , where, τ1 and τ 2 are the regression coefficients, ν is the intercept term. By solving the regression coefficients through the least squares method, the functional relationship between carbon emissions and energy consumption is obtained.

[0067] Through the feature selection algorithm and the Shapley Additive Explanation technique (SHAP), analyze the contributions of the load fluctuation pattern, environmental impact pattern, and operation cycle pattern to the change in equipment energy consumption. Assuming different operation strategies, the following impact component values are obtained through SHAP calculation: Impact component value of the load fluctuation pattern: 0.60; Impact component value of the environmental impact pattern: 0.25; Impact component value of the operation cycle pattern: 0.15; reflecting the relative contributions of each impact factor to the change in equipment energy consumption.

[0068] Step 3: Emission efficiency assessment; Based on the constructed carbon emission model and the analysis results of the impact factors, the total electricity carbon emissions of the smart park can be predicted under any operation strategy. For example, assuming that the operation strategy of the equipment leads to the following changes in energy consumption: Change in the main dimension 1: increase of 200 kWh; Change in the main dimension 2: decrease of 100 kWh; According to the model, predict the carbon emissions: Predicted carbon emissions = τ 1 × 200 + τ 2 × (-100) + ν ; Based on the carbon emission rate calculated from the carbon emission factor, evaluate the emission efficiency of the power supply equipment. According to the set carbon emission factor (0.4 tons of CO 2 / kWh), we get: Direct electricity carbon emission rate: 0.4 tons of CO 2 / kWh; Carbon emission rate of purchased electricity: Assume the carbon emission factor of purchased electricity is 0.5 tons of CO 2 / kWh.

[0069] Through the Data Envelopment Analysis (DEA) model, evaluate the carbon emissions and energy consumption of each equipment, and calculate the relative efficiency value of each equipment. For example: Emission efficiency of direct electricity equipment: 0.8 (higher efficiency); Emission efficiency of purchased electricity equipment: 0.6 (lower efficiency).

[0070] Therefore, based on the above steps, a relationship model between the total carbon emissions and the energy consumption of electrical equipment can be accurately constructed to predict the total carbon emissions of electrical equipment in the smart park under different operation strategies, and to judge the emission efficiency corresponding to the total carbon emissions of electricity. Finally, based on these analysis results, the park managers can optimize the equipment operation strategies, reduce carbon emissions, and improve energy efficiency.

[0071] Step S3: Generate an electricity usage allocation model based on the total carbon emissions of electricity and the emission efficiency, determine the electricity carbon emission reduction strategy in the smart park, and evaluate the energy conservation and emission reduction effect of the smart park after the implementation of the electricity carbon emission reduction strategy according to the total carbon emissions of electricity.

[0072] In one embodiment, in the process of generating an electricity usage allocation model based on the total carbon emissions of electricity and the emission efficiency, determining the electricity carbon emission reduction strategy in the smart park, and evaluating the energy conservation and emission reduction effect of the smart park after the implementation of the electricity carbon emission reduction strategy, the electricity consumption details, time granularity data, and historical electricity consumption data in the smart park can be obtained, and the trend prediction and seasonal changes of electricity consumption can be identified by combining with the time series model to determine the peak and off-peak periods of electricity consumption; analyze the total electricity cost and regional electricity cost in the smart park according to the electricity price data, and analyze the equipment usage situation during the peak electricity consumption period in combination with the electricity consumption details to obtain the electricity demand in the smart area; construct a multi-objective integrated planning model with the minimum power generation cost as the objective function, and generate the optimal output combination corresponding to the dynamic allocation of direct electricity and purchased electricity by combining the total carbon emissions of electricity and the emission efficiency; obtain the direct electricity consumption and purchased electricity consumption in the smart park according to the optimal output combination, determine the electricity carbon emission reduction strategy in the smart park, and obtain the total carbon emissions of electricity in the smart park after the application of the optimal output combination; compare the difference between the total carbon emissions of electricity obtained after the application of the optimal output combination and the initial total carbon emissions of electricity to evaluate the energy conservation and emission reduction effect of the smart park after the implementation of the electricity carbon emission reduction strategy.

[0073] In one embodiment, in the process of constructing a multi-objective integration planning model with the minimum power generation cost as the objective function and generating an optimal output combination corresponding to the dynamic allocation of direct power consumption and externally purchased power by combining the total power carbon emissions and emission efficiency, the total power generation cost under the condition of meeting the power consumption demand can be calculated based on the power consumption demand and power consumption cost of the smart park, and the power generation cost value required to meet the power consumption demand by direct power consumption or externally purchased power at any time period within the smart park can be analyzed according to the total power generation cost; taking the minimum power generation cost value and the minimization of carbon emissions as the objective function and the capacity of the power generation equipment as the constraint condition, a multi-objective integration planning model is constructed by combining the single objective function and the constraint condition; weights are assigned to the objective function by the weighted sum method, and the output balance point of the multi-objective integration planning model is obtained by finding a compromise based on Pareto optimality, and the combined allocation optimization result of the multi-objective integration planning is determined; taking the total power carbon emissions and emission efficiency as the fair judgment basis for the output combination, the fairness of the combined allocation optimization result is verified by using the calculation formula of the fairness of the output combination allocation, and the optimal output combination corresponding to the dynamic allocation of direct power consumption and externally purchased power is output according to the fairness result.

[0074] It should be noted that in the process of outputting the optimal output combination, the goal is to minimize the total power generation cost of the smart park under the condition of meeting the power consumption demand. Here, the power generation cost involves two parts: direct power consumption and externally purchased power.

[0075] Direct power consumption cost: The cost of power supply generated by the power generation equipment within the smart park; Externally purchased power cost: The cost of power purchased from the external power grid; Minimization of carbon emissions: The goal is to reduce the total carbon emissions, and the emissions are reduced by optimizing the power consumption method, such as by selecting power generation equipment or externally purchased power with a lower carbon emission factor.

[0076] The power consumption demand constraint is to ensure that the sum of direct power consumption and externally purchased power can meet the power consumption demand of the park at any time period, that is: D ( t ) = P direct ( t ) + P purchase ( t ), where P direct ( t ) represents the direct power consumption at time t , P purchase ( t ) represents the externally purchased power consumption, D ( t ) represents the power consumption demand at time t .

[0077] Power generation equipment capacity constraint: Each power generation equipment has a maximum power generation capacity, which cannot exceed its design capacity: P direct ( t ) ≤ P max,direct , where P max,direct represents the maximum power generation capacity of the direct power consumption source.

[0078] Construct a multi-objective optimization model that includes two objective functions (minimizing power generation cost and minimizing carbon emissions). To make the model operable, the weighted sum method is used to transform the two objective functions into a single objective function: Objective function = U 1 × Power generation cost + U 2 × Total carbon emissions target function; U 1 and U 2 are the weight coefficients of the objective function, reflecting the relative importance between power generation cost and carbon emissions. The weights can be determined through expert evaluation or sensitivity analysis, and the power generation cost and total carbon emissions can be calculated based on the equipment and power carbon emission factors in the smart park.

[0079] For each time period, the cost of direct power consumption and the cost of purchased electricity are calculated based on factors such as the fuel cost of the power generation equipment, the electricity market price, and the efficiency of the direct power consumption equipment. The carbon emissions of each power generation equipment can be calculated by multiplying the electricity consumed by the equipment by the carbon emission factor of the equipment, Carbon emissions = ∑ l P l ( t ) × Carbon emission factor l where P l ( t ) is the power generation of equipment l at time t , and the carbon emission factor l is the carbon emission factor of the equipment.

[0080] In multi-objective optimization, the objective functions are conflicting with each other. Therefore, it is necessary to find a balance point to achieve a compromise effect. To obtain the optimal solution, the concept of Pareto optimality can be used, that is, under the given constraints, find a solution that is not dominated by any other solution.

[0081] Dominance relationship: For two solutions E and G , if E is not inferior to Gand is superior to on at least one objective G then E dominates G In multi-objective optimization, the Pareto optimal solution means that there is no solution that can be superior to this solution on all objectives. To obtain the Pareto optimal solution, some common optimization algorithms can be used. For example: Genetic algorithm: By simulating the biological evolution process, operations such as crossover and mutation are used to find the optimal solution; Particle Swarm Optimization (PSO): Simulates the collective behavior of bird flocks or fish schools to find the optimal solution.

[0082] In the weighted sum method, by giving weights U 1 and U 2 to set the relative importance between objectives. For example: If U 1 is larger, it means that the power generation cost takes precedence over carbon emissions; if U 2 is larger, it means that reducing carbon emissions takes precedence over the power generation cost. The weights can be adjusted according to different operation strategies, policy requirements or expert opinions, and sensitivity analysis can be used to evaluate the changes in the model results under different weights.

[0083] Judge the fairness of the distribution of direct power consumption and purchased power through the total carbon emissions and emission efficiency. The core idea of fairness evaluation is: while meeting the power demand, try to make the carbon emission efficiencies of the two power consumption methods close to avoid excessive dependence on one method, resulting in too high carbon emissions.

[0084] Finally, the optimal output combination of direct power consumption and purchased power is output through the above process, that is, how to reasonably allocate the power demand of the park within a certain period of time to meet the power demand and minimize the power generation cost and carbon emissions at the same time. The distribution of direct power consumption and purchased power will be dynamically adjusted to adapt to actual demand fluctuations, energy cost changes and carbon emission requirements. The output results can provide optimization suggestions for park managers to select the optimal power supply method at different time periods.

[0085] Among them, the calculation formula for the fairness of output combination distribution is: ; In the formula, max Z represents the fairness of output combination distribution, α represents the power demand of electrical equipment in the smart park, A represents the set of power demands of electrical equipment in the smart park, δ represents the time point, B represents the set of time points, ε represents the relative efficiency of carbon emissions, N represents the set of carbon emission efficiencies, σRepresents the transfer point during the power supply process in the smart park, Q Represents the set of transfer points during the power supply process in the smart park, ρ Represents the transfer path during the power supply process in the smart park, P Represents the set of transfer paths during the power supply process in the smart park, S σ,α,δ,ρ Represents the δ th time period, the σ th transfer point, through the ρ th transfer path, to meet the electricity demand of the α th electrical equipment, the required power generation cost value, R α,δ Represents the capacity required for the power generation equipment when meeting the electricity demand of the δ th electrical equipment in the α th time period, λ α,δ During the δ th time period, when meeting the electricity demand of the α th electrical equipment, the required power transmission duration.

[0086] It should be noted that during the operation of the power carbon emission reduction strategy, according to the carbon emission quota set by the park, different warning levels are defined (such as yellow warning, orange warning, red warning), and each level corresponds to a different degree of over-limit risk. Considering factors such as seasonal fluctuations and special activities, the warning threshold is adjusted in a timely manner to better adapt to the actual situation.

[0087] Classification suggestions: Low-level warning (yellow): Inform users that the current carbon emissions are approaching the quota ceiling, encourage them to take some simple energy-saving measures, such as turning off unnecessary electrical equipment, and provide tips or links on energy conservation and emission reduction to help users understand how to effectively reduce personal or departmental carbon emissions.

[0088] Medium-level warning (orange): For specific areas or facilities, give a detailed energy-saving plan, such as adjusting the air-conditioning temperature setting, optimizing the lighting schedule, etc., and impose temporary power restrictions or operation mode switching on key energy-consuming equipment to avoid a sharp increase in carbon emissions in the short term.

[0089] High-level warning (red): Activate the emergency plan, such as mobilizing the emergency team to check and repair problem points that may cause high carbon emissions; when necessary, turn off non-essential production equipment or services, evaluate the deficiencies in the existing energy management system, and formulate medium- and long-term improvement measures, such as introducing more renewable energy and upgrading energy storage facilities.

[0090] To facilitate the understanding of the above technical solutions of the present invention, the following will detail the determination method of the power carbon emission reduction strategy in the smart park during the actual process of the present invention.

[0091] Step 1: Data Preparation Before implementing the power carbon emission reduction strategy, various data of the smart park need to be collected and analyzed first. The following are some key data examples: Electricity consumption details data: Total system electricity consumption (unit: kWh): For example, the total electricity consumption of the park on a certain day is 10,000 kWh; Regional electricity consumption: For example, the park is divided into three regions A, B, and C. The electricity consumption of region A is 4,000 kWh, that of region B is 3,000 kWh, and that of region C is 3,000 kWh.

[0092] Sub-system electricity consumption: For example, the electricity consumption of the air conditioning system is 2,000 kWh, and that of the lighting system is 1,500 kWh.

[0093] Electricity price data: Daily electricity price (unit: yuan / kWh). For example, the electricity price during peak hours is 1.0 yuan / kWh, and the electricity price during off-peak hours is 0.5 yuan / kWh; Carbon emission data: Carbon emission factor (unit: ton CO 2 / kWh). The carbon emission factor of direct electricity-consuming equipment is 0.4 ton CO 2 / kWh, and the carbon emission factor of purchased electricity is 0.5 ton CO 2 / kWh.

[0094] Total carbon emissions: Assume the total carbon emissions of the smart park are 4 tons CO 2 .

[0095] Historical electricity consumption data: High electricity consumption areas: For example, region A is often a high electricity consumption area in summer; Low electricity consumption areas: Region C is often a low electricity consumption area in summer; Energy-saving measure evaluation data: Energy-saving measures: Adjusting air conditioning temperature, using high-efficiency lighting, etc., and the energy-saving effect is 10%.

[0096] Step 2: Power Usage Allocation Model It is necessary to optimize the power usage allocation in the smart park according to factors such as electricity demand, carbon emission factors, and electricity price data. The goal is to reduce the total power generation cost and carbon emissions by reducing peak-hour power consumption and optimizing the power source.

[0097] (1) Determine electricity demand: Assume that in a specific time period (such as 24 hours a day), the total electricity consumption of 10,000 kWh needs to be met, and the electricity demand of each region is as follows: Region A: 4,000 kWh; Region B: 3,000 kWh; Region C: 3,000 kWh; (2)Analysis of electricity price and carbon emissions: The electricity price has a greater impact on the cost of electricity consumption, especially during peak periods: The electricity price during peak hours is 1.0 yuan / kWh; The electricity price during off-peak hours is 0.5 yuan / kWh; At the same time, different power sources (direct electricity consumption and purchased electricity) also have different carbon emission factors: Direct electricity consumption: 0.4 tons of CO 2 / kWh; Purchased electricity: 0.5 tons of CO 2 / kWh.

[0098] Step 3: Analysis of electricity consumption cost; Based on the electricity price data, the electricity consumption costs in different regions can be analyzed. For example: The electricity consumption cost in Region A is 4000 kWh × 1.0 yuan / kWh = 4000 yuan (assuming all are peak hours); The electricity consumption cost in Region B is 3000 kWh × 0.5 yuan / kWh = 1500 yuan (assuming all are off-peak hours); The electricity consumption cost in Region C is 3000 kWh × 1.0 yuan / kWh = 3000 yuan (assuming all are peak hours).

[0099] Step 4: Optimize the power usage allocation based on carbon emissions; It is necessary to reduce the total carbon emissions in the park through optimization strategies. The following is a possible strategy: During the peak electricity period (1.0 yuan / kWh), try to use purchased electricity as much as possible (emission factor is 0.5 tons of CO 2 / / kWh); During the off-peak electricity period (0.5 yuan / kWh), give priority to using direct electricity (emission factor is 0.4 tons of CO 2 / / kWh), and reduce the electricity demand during peak hours through system scheduling.

[0100] Assume that in the off-peak period, Region A uses 2000 kWh of direct electricity and Region B uses 3000 kWh of purchased electricity; during the peak period, Region A uses 2000 kWh of purchased electricity, Region B uses 0 kWh of purchased electricity, and Region C uses 3000 kWh of purchased electricity.

[0101] The goal is to minimize the power generation cost and carbon emissions. A single objective function is constructed through the weighted sum method to balance these two objectives: Objective function = U 1 × Power generation cost + U 2 × Total carbon emissions where: U 1 = 0.6 (reflecting a cost-priority strategy);U 2 = 0.4 reflects the priority of carbon emission reduction.

[0102] In multi-objective optimization, genetic algorithms or particle swarm optimization are used to solve the Pareto optimal solution, that is, to find a compromise point between power generation cost and carbon emissions.

[0103] Dominance relationship: Suppose two solutions are obtained E and G , if solution E is superior to solution G in terms of cost, and is also superior to solution G in terms of carbon emissions, then solution E will dominate solution G , and then gradually adjust the power source and the output of power generation equipment until a solution that meets the power demand and maximally reduces costs and carbon emissions is found.

[0104] Step 4. Apply the optimal output combination; The results of the optimal output combination may be as follows: Region A: Direct power consumption 2000 kWh, purchased power 2000 kWh; Region B: Direct power consumption 0 kWh, purchased power 3000 kWh; Region C: Direct power consumption 0 kWh, purchased power 3000 kWh; Assume that these output combinations are optimal in terms of balancing costs and carbon emissions.

[0105] According to the comparison between the optimal output combination and the total initial power carbon emissions, evaluate the energy conservation and emission reduction effect: Total initial carbon emissions: 10000 kWh × 0.4 tons of CO 2 / kWh = 4 tons of CO 2 ; Total carbon emissions under the optimal output combination: Assume that the total carbon emissions under the optimal output combination are 3.5 tons of CO 2 ; The energy conservation and emission reduction effect is: Emission reduction = 4 tons of CO 2 - 3.5 tons of CO 2 = 0.5 tons of CO 2 , which means that by applying the optimal power usage allocation model, the carbon emissions of the smart park are reduced by 0.5 tons of CO 2 .

[0106] According to the evaluation data of the energy conservation measures, the effects of energy conservation measures (such as adjusting air conditioner temperature, optimizing lighting, etc.) at different time periods can be further analyzed and compared with the same period in the past.

[0107] For example, during a certain energy-saving cycle, the park saved 10% of its power consumption by adjusting the air-conditioning temperature, reducing 0.2 tons of CO 2 emissions. The final result was that through optimizing power distribution and implementing energy-saving measures, the smart park achieved the goal of reducing 0.7 tons of CO 2 emissions.

[0108] Therefore, through the above process, a power usage allocation model based on the total power carbon emissions and emission efficiency was constructed, and the power carbon emission reduction strategy for the smart park was determined. With the support of real-time data monitoring and optimization algorithms, it can provide the best power usage strategy for the park, not only effectively reducing carbon emissions, but also achieving energy conservation and emission reduction, and providing strong support for sustainable development.

[0109] In a second aspect, the present invention also provides a smart park power carbon emission prediction system based on artificial intelligence. The system includes: A total carbon emission calculation unit 1, which is used to obtain the data of the park's externally purchased power activities and direct power consumption data, and calculate the carbon emissions of the externally purchased power and direct power consumption in the smart park using the power carbon emission accounting standard to obtain the total carbon emissions; A carbon emission efficiency judgment unit 2, which is used to construct a relationship model between the total carbon emissions and the energy consumption of the electrical equipment, predict the total power carbon emissions generated by the electrical equipment in the smart park under any operation strategy, and judge the emission efficiency corresponding to the total power carbon emissions; An energy-saving strategy execution evaluation unit 3, which is used to generate a power usage allocation model based on the total power carbon emissions and emission efficiency, determine the power carbon emission reduction strategy in the smart park, and evaluate the energy conservation and emission reduction effect of the smart park after using the power carbon emission reduction strategy according to the total power carbon emissions.

[0110] In summary, by means of the above technical solutions of the present invention, the present invention can accurately calculate the total carbon emissions by obtaining the data of the park's externally purchased power activities and direct power consumption data, providing reliable data support for subsequent analysis, constructing a relationship model between the total carbon emissions and the energy consumption of the electrical equipment, providing a scientific prediction for the power carbon emissions under different operation strategies, and at the same time generating a power usage allocation model to optimize the power usage structure of the park according to the total power carbon emissions and emission efficiency, helping the smart park achieve low-carbon goals, optimize energy use, and improve the overall operation efficiency. And through regular evaluation and adjustment, the long-term effect of energy conservation and emission reduction measures is ensured.

[0111] The present invention obtains the actual total carbon emissions and power consumption data of electrical equipment during the historical operation and maintenance of an intelligent park, reflects the energy consumption performance of various different devices under different operating conditions in the intelligent park, and combines the energy consumption conditions with the carbon emissions for analysis, which can reveal the carbon emission characteristics under different energy consumption levels, help understand the total carbon emissions generated under specific conditions, and thus provide a basis for carbon emission reduction strategies. The present invention analyzes the historical power consumption data through a time series model, can accurately identify the trends and seasonal variations of power consumption, and constructs a multi-objective integrated planning model with the minimum power generation cost as the objective function. By combining the total power carbon emissions and emission efficiency, it generates the optimal output combination corresponding to the dynamic allocation of direct power consumption and purchased power, so that while ensuring the lowest power generation cost, it can also take into account the minimization of carbon emissions, and thus can minimize carbon emissions to the greatest extent while ensuring power supply and optimize the energy use efficiency.

[0112] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0113] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, it is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that based on the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative efforts are still within the protection scope of the present invention.

Claims

1. A method for predicting carbon emissions from electricity in smart parks based on artificial intelligence, characterized in that: The method includes: Obtain the data on purchased electricity activities and direct electricity consumption in the smart park, and use the electricity carbon emission accounting standard to calculate the carbon emissions of purchased electricity and direct electricity consumption in the smart park to obtain the total carbon emissions; Construct a relationship model between total carbon emissions and energy consumption of electrical equipment, predict the total carbon emissions of electricity generated by electrical equipment in the smart park under any operation strategy, and determine the emission efficiency corresponding to the total carbon emissions of electricity; Based on the total amount of electricity carbon emissions and emission efficiency, an electricity usage allocation model is generated to determine the electricity carbon emission reduction strategy within the smart park. Then, based on the total amount of electricity carbon emissions, the energy saving and emission reduction effect of the smart park after the use of the electricity carbon emission reduction strategy is evaluated.

2. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 1 is characterized in that: The relationship model between the total carbon emissions and the energy consumption of electrical equipment is constructed to predict the total carbon emissions of electricity generated by electrical equipment in the smart park under any operation strategy, and to determine the emission efficiency corresponding to the total carbon emissions of electricity, including: Obtain the total carbon emissions and energy consumption data of electrical equipment corresponding to the historical operation and maintenance of the smart park, and perform dimensionality reduction and decoupling processing on the electrical energy consumption data to obtain several main dimension vectors with orthogonal distribution; Based on the mean clustering technology, each main dimension vector is constrained in turn, and energy consumption conditions are generated by random combination, and the comprehensive energy consumption level of the energy consumption conditions is analyzed in combination with carbon emissions; According to the Pearson correlation coefficient between the main dimension vector and the comprehensive level energy consumption level, the relationship model between carbon emissions and the energy consumption of electrical equipment is obtained, and the functional relationship between carbon emissions and energy consumption is generated; Based on the feature selection algorithm and Shapley additive interpretation technology, the key factors that affect the energy consumption of power equipment under any operation strategy are obtained, and the energy consumption value is predicted by a numerical set to determine the total carbon emissions of electricity. The carbon emission factor is used to analyze the time required for the total amount of electricity carbon emissions to be completely discharged, and the emission efficiency of the power supply equipment in the smart park is evaluated based on the time results.

3. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 2 is characterized in that: The feature selection algorithm and Shapley additive interpretation technology are used to obtain the key factors that affect the energy consumption of electrical equipment under any operation strategy, and generate a numerical set to predict the energy consumption value, and determine the total amount of carbon emissions from electricity, including: Select influencing factors including load fluctuation mode, environmental impact mode and operation cycle mode according to the operation strategy of the power-consuming equipment, and determine the total impact component value of the influencing factors on the power-consuming equipment; Obtain feature information of the total impact component value, and use the feature information as the feature set and the total impact component value as the data set to initialize the number of iterations and the number of selected neighbors; A set of impact component values ​​of impact factors are randomly selected from the data set as the selected objects, and the impact component values ​​of the other two sets of impact factors under the same dimension are selected as heterogeneous objects. The selection is repeated until the number of objects selected by iteration is reached. Analyze the distance between the selected object and the heterogeneous objects in feature information, and combine the Shapley additive interpretation technique to determine the key factors affecting the energy consumption change, and generate a numerical set of key factors; Calculate the energy consumption of electrical equipment under a set of numerical values, and determine the total amount of carbon emissions from electricity based on the functional relationship between the energy consumption of electrical equipment and carbon emissions.

4. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 3 is characterized in that: The calculation formula of the total impact component value is: ; In the formula, β Represents the total impact component value, β F Indicates the impact component value in load fluctuation mode, β Y Indicates the impact component value in the operating cycle mode, β H Indicates the impact component value under the environmental impact model, a represents the accumulation order, c b represents the regression coefficient of the influencing component under load fluctuation mode, D b Indicates b Load fluctuation value under each operation strategy: f 1e and f 2e represents the regression coefficient of the influencing component in the operating cycle mode, t Indicates the cumulative number of days from the start observation date to the end observation date of the current operation strategy. e represents a constant, i h represents the regression coefficient of the impact component under the environmental impact model, J h Indicates h A time factor.

5. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 4 is characterized in that: The analysis selects the distance between the object and the heterogeneous object in feature information, and combines the Shapley additive interpretation technique to determine the key factors affecting the energy consumption change, and generates a numerical set of key factors including: The importance value of the selected object is analyzed based on the distance result, and the importance value of the object is repeatedly analyzed according to the number of iterative selections, and the maximum importance value is used as the corresponding feature weight for selection; Analyze the Shapley value of each influencing factor, average the absolute value of the Shapley value to get the characteristic contribution value of each influencing factor, and the combination of the characteristic contribution value and the characteristic weight is collectively referred to as the factor contribution value; Analyze the cumulative contribution rate of each influencing factor contribution value in the total factor contribution value, calculate the cumulative contribution rate difference of each influencing factor based on the cumulative contribution rate, and compare the cumulative contribution rate difference with the cumulative contribution rate difference threshold; The influencing factors whose cumulative contribution rate difference is greater than the cumulative contribution rate difference threshold are selected as the key factors affecting the change of energy consumption, and the value sets of the power-consuming equipment under different operation strategies are generated according to the key factors.

6. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 5 is characterized in that: The use of carbon emission factors to analyze the time corresponding to the total amount of carbon emissions from electricity being completely discharged and the evaluation of the emission efficiency of power supply equipment in the smart park based on the time results include: The total carbon emissions from electricity are taken as the total carbon emissions from direct electricity consumption and the total carbon emissions from purchased electricity, and the carbon emission rate is set according to the corresponding carbon emission factors to obtain the emission duration of direct electricity consumption equipment and the emission duration of purchased electricity equipment; Obtain the decision factors that affect the carbon emissions of direct power-consuming equipment and purchased power equipment, establish a direct impact relationship matrix, and normalize the direct impact relationship matrix to obtain a normalized impact matrix; Obtain a comprehensive influence matrix of the normalized influence matrix, add up the factors in each row of the comprehensive influence matrix to obtain the influence degree, and add up the factors in each column of the comprehensive influence matrix to obtain the influence degree; The centrality and causality of each factor are calculated based on the influence and the influence. The emission duration evaluation index system is determined according to the calculation results. The data envelopment analysis model is used to evaluate the relative efficiency of carbon emissions from direct power equipment and purchased power equipment.

7. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 1 is characterized in that: The power usage allocation model is generated based on the total power carbon emissions and emission efficiency, the power carbon emission reduction strategy in the smart park is determined, and the energy saving and emission reduction effect of the smart park after the power carbon emission reduction strategy is used is evaluated according to the total power carbon emissions, including: Obtain electricity consumption details, time granularity data, and historical electricity consumption data within the smart park, and use time series models to identify trend forecasts and seasonal changes in electricity consumption, and determine peak and trough periods of electricity consumption; Analyze the total electricity cost and regional electricity cost in the smart park based on electricity price data, and analyze the equipment usage during peak hours based on electricity usage details to obtain the electricity demand in the smart area; A multi-objective integrated planning model is constructed with the minimum power generation cost as the objective function, and the optimal output combination corresponding to the dynamic allocation of direct electricity consumption and purchased electricity consumption is generated by combining the total carbon emissions of electricity and emission efficiency; Obtain the direct electricity consumption and purchased electricity consumption in the smart park based on the optimal output combination, determine the electricity carbon emission reduction strategy in the smart park, and obtain the total electricity carbon emission of the smart park after the optimal output combination is applied; The total electricity carbon emissions obtained after applying the optimal output combination are compared with the initial total electricity carbon emissions, and the energy-saving and emission reduction effect of the smart park after the use of the electricity carbon emission reduction strategy is evaluated.

8. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 7 is characterized in that: The multi-objective integrated planning model is constructed with the minimum power generation cost as the objective function, and the optimal output combination corresponding to the dynamic allocation of direct power consumption and purchased power consumption is generated by combining the total amount of power carbon emissions and emission efficiency. The optimal output combination includes: Based on the electricity demand and electricity cost of the smart park, the total power generation cost under the condition of meeting the electricity demand is calculated, and based on the total power generation cost, the power generation cost value required to directly use electricity or purchase electricity to meet the electricity demand in any time period in the smart park is analyzed; Taking the minimum power generation cost value and the minimum carbon emission as the objective function and the capacity of the power generation equipment as the constraint condition, a multi-objective integrated planning model is constructed by combining the single objective function with the constraint condition; The weights are assigned to the objective functions by the weighted sum method, and the output equilibrium point of the multi-objective integrated planning model is obtained by finding a compromise based on Pareto optimality to determine the combined allocation optimization result of the multi-objective integrated planning. The total amount of carbon emissions from electricity and emission efficiency are used as the basis for judging the fairness of the output combination. The fairness of the combination allocation optimization results is verified by using the calculation formula for the fairness of the output combination allocation. Based on the fairness results, the optimal output combination corresponding to the dynamic allocation of direct electricity consumption and purchased electricity is output.

9. The method for predicting carbon emissions of power in a smart park based on artificial intelligence according to claim 8 is characterized in that: The calculation formula for the fairness of the output combination allocation is: ; In the formula, max Z Indicates the fairness of output combination allocation, α Indicates the power demand of electrical equipment in the smart park. A It represents the power demand of the electrical equipment in the smart park. δ Indicates a point in time, B represents a set of time points, ε represents the relative efficiency of carbon emissions, N represents the carbon emission efficiency set, σ Indicates the transmission point in the power supply process within the smart park. Q Represents the collection of transmission points in the power supply process of the smart park. ρ Indicates the transmission path during the power supply process in the smart park. P It represents the set of transmission paths in the power supply process of the smart park. S σ,α,δ,ρ Indicates δ In the time period σ The delivery point ρ The transport path meets the α The power generation cost value required for each power-consuming device to meet its power demand. R α,δ Indicated in δ The time period meets the α The capacity of the power generation equipment required when each power-consuming device demands electricity. λ α,δ In the δ The time period meets the α The length of time it takes for electricity to be delivered to each electrical device when it needs it.

10. An artificial intelligence-based smart park electricity carbon emissions prediction system, used to implement the artificial intelligence-based smart park electricity carbon emissions prediction method described in any one of claims 1-9, characterized in that: The system includes: The total carbon emission calculation unit is used to obtain the data on purchased electricity activities and direct electricity consumption in the smart park, and calculate the carbon emissions of purchased electricity and direct electricity consumption in the smart park using the electricity carbon emission accounting standard to obtain the total carbon emissions; The carbon emission efficiency judgment unit is used to construct a relationship model between the total carbon emissions and the energy consumption of electrical equipment, predict the total carbon emissions of electricity generated by electrical equipment in the smart park under any operation strategy, and judge the emission efficiency corresponding to the total carbon emissions of electricity; The energy-saving strategy execution evaluation unit is used to generate an electricity usage distribution model based on the total amount of electricity carbon emissions and emission efficiency, determine the electricity carbon emission reduction strategy within the smart park, and evaluate the energy-saving and emission reduction effect of the smart park after the electricity carbon emission reduction strategy is used based on the total amount of electricity carbon emissions.

Images