Multi-energy-consumption carbon emission prediction method, system, equipment, medium and product

Abstract

The invention relates to the technical field of power systems, and discloses a multi-energy-consumption carbon emission prediction method, system, device, medium and product, and the method comprises the steps: obtaining power operation sample data and carbon emission sample data of a plurality of energy consumption types of a power enterprise, and carrying out the feature extraction of the power operation sample data; using the operation sample feature data and the carbon emission sample data as mapping samples, constructing a training data set, constructing an extreme learning machine based on an AdaBoost algorithm, and training the extreme learning machine by using the training data set to obtain a trained carbon emission prediction model; and the carbon emission prediction value of the current calculation time period is predicted through the trained carbon emission prediction model, so that the accuracy and reliability of carbon emission prediction are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method, system, device, medium and product for predicting carbon emissions from multiple energy consumption sources. Background Technology

[0002] A multi-energy consumption prediction method driven by electricity data for high-energy-consuming enterprises has significant practical implications. High-energy-consuming enterprises have complex energy structures, involving multiple energy sources such as electricity, steam, natural gas, and coal, with significant energy coupling effects. Electricity data, as the most real-time and detailed energy data carrier for enterprises, can reflect changes in production load, energy consumption patterns, and their interaction with the external environment.

[0003] However, current methods for predicting carbon emissions are inaccurate and unreliable. Summary of the Invention

[0004] In view of this, the present invention provides a method, system, device, medium and product for predicting carbon emissions from multiple energy consumption sources, which solves the technical problems of poor accuracy and low reliability of current carbon emission prediction methods.

[0005] The first aspect of this invention provides a method for predicting carbon emissions from multiple energy consumption sources, comprising:

[0006] Acquire sample data of power operation and carbon emissions from power companies across multiple energy consumption types;

[0007] Feature extraction is performed on the power operation sample data to obtain operation sample feature data, and the operation sample feature data and the carbon emission sample data are used as mapping samples to construct a training dataset.

[0008] An extreme learning machine based on the AdaBoost algorithm optimized by combining the eGSK shared knowledge optimization algorithm is constructed. The extreme learning machine is trained using the training dataset to obtain a trained carbon emission prediction model.

[0009] The operating characteristic data corresponding to the power company's power operation data in the current calculation period is input into the trained carbon emission prediction model, so that the trained carbon emission prediction model outputs the carbon emission prediction value for the current calculation period.

[0010] Preferably, the energy consumption types include raw coal, natural gas, gasoline, and diesel; the feature extraction of the power operation sample data to obtain operation sample feature data includes:

[0011] Principal component analysis was used to reduce the dimensionality of the power operation sample data to obtain the dimensionality-reduced sample data.

[0012] Feature extraction is performed on the dimensionality-reduced sample data to obtain running sample feature data; wherein, the running sample feature data includes at least one of running sample mean data, running sample variance data, and running sample extreme value data.

[0013] Preferably, the construction of the extreme learning machine based on the AdaBoost algorithm includes:

[0014] Initialize the basic parameters of the extreme learning machine, including the number of input layer nodes, the number of hidden layer nodes, and the number of output layer nodes;

[0015] Multiple weak predictors are constructed based on the AdaBoost algorithm, and each weak predictor adopts an extreme learning machine structure.

[0016] Multiple weak predictors are linearly combined to form a strong predictor based on the AdaBoost algorithm, thus obtaining the extreme learning machine based on the AdaBoost algorithm.

[0017] Preferably, the hyperparameters of the extreme learning machine are optimized using the eGSK shared knowledge optimization algorithm.

[0018] Preferably, the optimization of the hyperparameters of the extreme learning machine using the eGSK shared knowledge optimization algorithm includes:

[0019] Set the initial parameters for the eGSK heuristic optimization algorithm, including population size, maximum number of iterations, and knowledge rate;

[0020] An initial population is randomly generated within the parameter space, with each individual representing a set of hyperparameters of the extreme learning machine;

[0021] Calculate the fitness value of each individual in the initial population, and select the best individual based on the fitness value;

[0022] Knowledge is shared and updated according to the knowledge rate and iteration progress to generate a new generation of population;

[0023] Based on the new generation population, the process returns to calculating the fitness value of each individual in the initial population, selecting the optimal individual based on the fitness value, comparing the fitness value of the new generation population with that of the population before the update, and retaining the population with better fitness and its corresponding optimal individual, until the preset maximum number of iterations is reached. The hyperparameters corresponding to the individual with the highest fitness value are then output as the optimal hyperparameter configuration for the extreme learning machine.

[0024] Preferably, the step of sharing and updating knowledge according to knowledge rate and iteration progress to generate a new generation of population includes:

[0025] Determine the number of dimensions to be updated at each stage based on the knowledge rate and iteration progress;

[0026] Based on the determined number of dimensions, at each stage, the best individual in the current population is selected for knowledge sharing, and the dimension information of the best individual is transmitted to other individuals in the current population.

[0027] The individuals who receive the knowledge are updated to generate a new generation of population.

[0028] Secondly, the present invention also provides a multi-energy consumption carbon emission prediction system, comprising:

[0029] The data acquisition module is used to acquire power operation sample data and carbon emission sample data of power companies for multiple energy consumption types.

[0030] The dataset construction module is used to extract features from the power operation sample data to obtain operation sample feature data, and to construct a training dataset by using the operation sample feature data and the carbon emission sample data as mapping samples.

[0031] The prediction model training module is used to construct an extreme learning machine based on the AdaBoost algorithm optimized by combining the eGSK shared knowledge optimization algorithm. The extreme learning machine is trained using the training dataset to obtain a trained carbon emission prediction model.

[0032] The carbon emission prediction module is used to input the operating characteristic data corresponding to the power company's power operation data in the current calculation period into the trained carbon emission prediction model, so that the trained carbon emission prediction model outputs the carbon emission prediction value for the current calculation period.

[0033] Thirdly, the present invention also provides an electronic device, the electronic device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the multi-energy carbon emission prediction method as described in the first aspect.

[0034] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the steps of the multi-energy carbon emission prediction method as described in the first aspect.

[0035] Fifthly, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein, when the program instructions are executed by a computer, the computer performs the steps of the multi-energy carbon emission prediction method as described in the first aspect.

[0036] As can be seen from the above technical solution, this invention acquires power operation sample data and carbon emission sample data of multiple energy consumption types from power companies, extracts features from the power operation sample data, and uses the operation sample feature data and carbon emission sample data as mapping samples to construct a training dataset. A Extreme Learning Machine (ELM) based on the AdaBoost algorithm is then constructed, and the ELM is trained using the training dataset to obtain a trained carbon emission prediction model. This model is then used to predict the carbon emission value for the current calculation period. Thus, by utilizing the power operation data features of multiple energy consumption types in the power data, the accuracy and reliability of carbon emission prediction are effectively improved. Furthermore, by introducing the eGSK shared knowledge optimization algorithm to optimize the hyperparameters of the ELM, the predictive performance and stability of the model are further enhanced. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 An application environment diagram of a multi-energy consumption carbon emission prediction method provided in an embodiment of the present invention;

[0039] Figure 2 A flowchart illustrating a multi-energy consumption carbon emission prediction method provided in this embodiment of the invention;

[0040] Figure 3 This is a schematic diagram of the structure of a multi-energy consumption carbon emission prediction system provided in an embodiment of the present invention;

[0041] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0042] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] The multi-energy consumption carbon emission prediction method provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, terminal 101 communicates with server 102 via a network. A data storage system can store the data that server 102 needs to process. The data storage system can be integrated onto server 102 or placed on a cloud or other network server. Terminal 101 or server 102 acquires power operation sample data and carbon emission sample data of multiple energy consumption types from the power company; it extracts features from the power operation sample data to obtain operation sample feature data, and uses the operation sample feature data and carbon emission sample data as mapping samples to construct a training dataset; it constructs an extreme learning machine based on the AdaBoost algorithm, trains the extreme learning machine using the training dataset, and obtains a trained carbon emission prediction model; it inputs the operation feature data corresponding to the power company's power operation data in the current calculation period into the trained carbon emission prediction model, causing the trained carbon emission prediction model to output the predicted carbon emission value for the current calculation period.

[0044] Terminal 101 can be, but is not limited to, various personal computers, laptops, smartphones, and tablets.

[0045] Server 102 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0046] like Figure 2 As shown in the embodiments of this application, a method for predicting carbon emissions from multiple energy consumption sources is provided, which is then applied to... Figure 1 Taking terminal 101 or server 102 as an example, the explanation includes the following steps S1 to S4. Wherein:

[0047] Step S1: Obtain power operation sample data and carbon emission sample data for multiple energy consumption types from power companies.

[0048] The power operation sample data is collected from the historical operation records, real-time monitoring systems, or relevant energy management platforms of power companies. This data covers energy consumption at different time scales. Energy consumption types include raw coal, natural gas, gasoline, and diesel.

[0049] Step S2: Extract features from the power operation sample data to obtain operation sample feature data, and use the operation sample feature data and carbon emission sample data as mapping samples to construct a training dataset.

[0050] Among these steps, feature extraction is performed on power operation sample data to obtain operation sample feature data, including:

[0051] Step S201: Principal component analysis is used to reduce the dimensionality of the power operation sample data to obtain the dimensionality-reduced sample data.

[0052] Step S202: Extract features from the dimensionality-reduced sample data to obtain running sample feature data; wherein, the running sample feature data includes at least one of running sample mean data, running sample variance data, and running sample extreme value data.

[0053] The running sample mean data is obtained by averaging the dimensionality-reduced sample data at various time points or across various observation dimensions, reflecting the overall average level of the sample data. The running sample variance data measures the dispersion of the sample data around the mean, reflecting the volatility of the data. The running sample extreme value data includes the maximum and minimum values, which represent the upper and lower limits of the sample data within the observation range, respectively, helping to understand extreme cases. By extracting these feature data and combining them with carbon emission sample data as mapping samples, a more accurate and reliable training dataset can be constructed.

[0054] Step S3: Construct an extreme learning machine based on the AdaBoost algorithm optimized by combining the eGSK shared knowledge optimization algorithm, and train the extreme learning machine using the training dataset to obtain a trained carbon emission prediction model.

[0055] Among them, the eGSK shared knowledge optimization algorithm is used to optimize the hyperparameters of the extreme learning machine.

[0056] The hyperparameters include the number of hidden layer nodes, regularization coefficients, and activation functions. The eGSK (Enhanced Group Search Optimizer with Knowledge Sharing) algorithm improves global search capability and convergence speed by introducing a knowledge-sharing mechanism. Specifically, the eGSK algorithm simulates the information interaction behavior among individuals in a biological population, dynamically maintaining a shared knowledge base during the search process. This knowledge base stores high-quality solutions discovered by outstanding individuals in the population. In each iteration, the algorithm not only updates its position based on its own search experience but also refers to the globally optimal information in the shared knowledge base to guide the search direction. This two-layer optimization mechanism effectively avoids the pitfalls of traditional population algorithms that easily get trapped in local optima. Simultaneously, the eGSK algorithm employs an adaptive knowledge-sharing strategy, dynamically adjusting the intensity of knowledge sharing according to population diversity and search progress. It maintains high exploratory capability in the early stages of the search and enhances development capability in the later stages, thus balancing the algorithm's global search and local development performance.

[0057] This application's embodiments introduce the eGSK shared knowledge optimization algorithm into ELM-AdaBoost. Utilizing its global optimization and fast convergence characteristics, it automatically searches for and determines the optimal hyperparameter combination in each iteration, thereby fully leveraging the advantages of AdaBoost ensemble and significantly improving the model's accuracy, stability, and generalization ability.

[0058] Extreme Learning Machine (ELM) is a single-hidden-layer feedforward neural network algorithm with advantages such as fast learning speed and good generalization performance. Its key feature is that the input weights and hidden layer biases are randomly generated and fixed, while the output weights are obtained analytically in a single step, thus avoiding the computational overhead of repeated iterations in traditional gradient descent training. Its basic mechanism is as follows: first, a linear mapping is performed between the input layer and the hidden layer, and a non-linear activation function is applied to form a high-dimensional feature representation. Then, the output weights are obtained analytically from the hidden layer to the output layer using the least squares method or a regularized method, achieving rapid prediction.

[0059] The Extreme Learning Machine consists of an input layer, a hidden layer, and an output layer.

[0060] The input layer receives the feature vectors corresponding to the feature data of the running samples, and the hidden layer contains L nodes, each with a corresponding input weight vector. and bias The output layer multiplies the hidden layer output by the weight matrix to obtain the final predicted Y:

[0061]

[0062] In the formula, H is the hidden layer output matrix (also called the feature mapping matrix), with dimensions N×L, where N is the number of samples; β is the weight matrix from the hidden layer to the output layer.

[0063] Hidden layer mapping is the core step of Extreme Learning Machine. The idea is to randomly generate and fix the input weights and biases, and project the original input samples into a high-dimensional feature space through a non-linear activation function, thereby forming the hidden layer output matrix H. This mapping can enhance the expressive power of features, making linearly inseparable problems easier to fit by linear models in high-dimensional space.

[0064] Since the hidden layer mapping is fixed, the training task simplifies to a linear regression problem:

[0065]

[0066] The goal is to minimize the error:

[0067]

[0068] Where λ is the regularization coefficient (optional, used to prevent overfitting).

[0069] The learning process of the Extreme Learning Machine mainly includes three steps: First, the input weights and biases of the hidden layer are randomly generated and fixed; second, the input samples are mapped to the hidden layer through a non-linear activation function to construct the feature matrix; finally, the output layer weights are obtained in one step through least squares or analytical solutions with regularization, thereby completing the training quickly without relying on backpropagation and repeated iterations as in traditional neural networks.

[0070] Traditional Extreme Learning Machines (ELMs), while maintaining the advantage of rapid training, still suffer from performance instability due to random generation of hidden layer parameters, susceptibility to noise, and insufficient generalization ability. To overcome these shortcomings...

[0071] This invention combines Extreme Learning Machine (ELM) with the Adaptive Ensemble Boosting Framework (AdaBoost). By iteratively constructing multiple weak learners and dynamically adjusting sample weights, the model can focus more on samples with large prediction errors, thereby effectively improving the overall learning performance. The weighted fusion of multiple weak learners during the prediction phase significantly reduces the volatility and bias of a single model, enhancing its ability to fit complex nonlinear relationships and its robustness to outliers. Therefore, ELM combined with AdaBoost not only maintains its fast training speed but also possesses higher prediction accuracy and stronger stability, making it particularly suitable for high-precision scenarios such as multi-input multi-output regression and energy consumption prediction.

[0072] Specifically, constructing an extreme learning machine based on the AdaBoost algorithm includes:

[0073] Step S301: Initialize the basic parameters of the extreme learning machine. The basic parameters include the number of input layer nodes, the number of hidden layer nodes, and the number of output layer nodes.

[0074] Step S302: Construct multiple weak predictors based on the AdaBoost algorithm, with each weak predictor employing an extreme learning machine structure;

[0075] Step S303: Linearly combine multiple weak predictors to form a strong predictor based on the AdaBoost algorithm, thus obtaining the extreme learning machine based on the AdaBoost algorithm.

[0076] Understandably, Extreme Learning Machines (ELMs) based on the AdaBoost algorithm improve overall prediction performance by integrating multiple weak predictors (each a separate ELM). During construction, the basic parameters of the ELM must first be initialized. These parameters include the number of input layer nodes, hidden layer nodes, and output layer nodes, which collectively determine the model's structure and complexity. The number of input layer nodes typically matches the number of input features to ensure sufficient reception and processing of all relevant input information. The choice of the number of hidden layer nodes requires a trade-off between model complexity and computational efficiency; too many nodes may lead to overfitting, while too few nodes may fail to adequately capture complex patterns in the data. The number of output layer nodes corresponds to the number of prediction targets, ensuring the model can output the desired prediction results.

[0077] The process of training an Extreme Learning Machine (ELM) using a training dataset includes: dividing the training dataset into multiple subsets, each containing a portion of running sample feature data and corresponding carbon emission sample data. This increases the diversity of the training data, preventing the model from over-relying on any single data set and thus improving its generalization ability. For each subset, an initialized ELM is trained to obtain a weak predictor. During training, the ELM uses the input running sample feature data, through hidden layer mapping and output layer calculations, to output the corresponding predicted carbon emission value.

[0078] Then, this predicted value is compared with actual carbon emission sample data to calculate the prediction error. Based on the prediction error of each weak predictor, the weights of samples in the corresponding subset are adjusted. Samples with larger prediction errors are assigned higher weights, so that in subsequent training, the model will pay more attention to these difficult-to-predict samples, thereby striving to improve the prediction accuracy for these samples. Multiple weak predictors are iteratively built until a preset number of iterations is reached or the model's prediction performance meets the requirements.

[0079] In each iteration, the extreme learning machine is retrained based on the new sample weights to obtain new weak predictors, and the sample weights are updated. All the weak predictors obtained in all iterations are linearly combined to form a strong predictor based on the AdaBoost algorithm.

[0080] Specifically, each weak predictor is assigned a weight based on its predictive performance, with better-performing weak predictors receiving larger weights. Then, the predictions from all weak predictors are weighted and summed to obtain the final carbon emission prediction. In this way, by integrating the advantages of multiple weak predictors, the strong predictor achieves higher prediction accuracy and greater stability.

[0081] Step S4: Input the operating characteristic data corresponding to the power operation data of the power company in the current calculation period into the trained carbon emission prediction model, so that the trained carbon emission prediction model outputs the carbon emission prediction value for the current calculation period.

[0082] In practical applications, power companies can collect their power operation data in real time or periodically and extract corresponding operational characteristic data. This characteristic data is then input into a pre-trained carbon emission prediction model. Based on the learned mapping relationships, the model outputs a predicted carbon emission value for the current calculation period. This prediction can provide strong decision support for power companies' carbon emission management, helping them better understand their carbon emission situation and formulate corresponding emission reduction strategies.

[0083] It should be noted that this embodiment of the application obtains power operation sample data and carbon emission sample data of multiple energy consumption types from power companies, extracts features from the power operation sample data, and uses the operation sample feature data and carbon emission sample data as mapping samples to construct a training dataset. An extreme learning machine based on the AdaBoost algorithm is then constructed, and the extreme learning machine is trained using the training dataset to obtain a trained carbon emission prediction model. This model is then used to predict the carbon emission value for the current calculation period. By utilizing the power operation data features of multiple energy consumption types in the power data, the accuracy and reliability of carbon emission prediction are effectively improved. Furthermore, by introducing the eGSK shared knowledge optimization algorithm to optimize the hyperparameters of the extreme learning machine, the predictive performance and stability of the model are further enhanced.

[0084] In some embodiments, the eGSK shared knowledge optimization algorithm is used to optimize the hyperparameters of the extreme learning machine, including:

[0085] Step S501: Set the initial parameters of the eGSK heuristic optimization algorithm. The initial parameters include population size, maximum number of iterations, and knowledge rate.

[0086] Step S502: Randomly generate an initial population in the parameter space, where each individual represents a set of hyperparameters of the extreme learning machine.

[0087] Step S503: Calculate the fitness value of each individual in the initial population, and select the optimal individual based on the fitness value.

[0088] The fitness function uses metrics such as mean squared error or mean absolute error to measure the degree of deviation between the prediction results of the Extreme Learning Machine and the actual carbon emission sample data. Specifically, for each individual in the population (i.e., each combination of hyperparameters), it is applied to the Extreme Learning Machine and trained, and the trained model is used to predict the validation set.

[0089] Step S504: Share and update knowledge according to the knowledge rate and iteration progress to generate a new generation of population.

[0090] The knowledge-sharing process involves extracting information about the current optimal solution from a shared knowledge base and updating the position by combining it with the individual's own search experience. Specifically, the algorithm determines the proportion of information retrieved from the shared knowledge base based on the knowledge rate, while dynamically adjusting the intensity of knowledge sharing considering the current iteration progress. In the early stages of the search, the knowledge-sharing intensity is low, and individuals mainly rely on their own search experience for exploration; as iterations progress, the knowledge-sharing intensity gradually increases, and individuals refer more to the globally optimal information in the shared knowledge base to guide their search direction. Through this adaptive knowledge-sharing strategy, the eGSK algorithm can effectively balance global exploration and local exploitation capabilities. After updating the position, a new generation of the population is generated, with each individual representing a new set of hyperparameter combinations.

[0091] In some embodiments, knowledge sharing and updating are performed according to knowledge rate and iteration schedule to generate a new generation of population, including:

[0092] Step S5041: Determine the number of dimensions to be updated in each stage according to the knowledge rate and iteration progress;

[0093] Step S5042: Based on the determined number of dimensions, select the best individual in the current population at each stage for knowledge sharing, and transmit the dimension information of the best individual to other individuals in the current population.

[0094] Step S5043: Update the individuals that receive the knowledge transfer to generate a new generation of population.

[0095] Specifically, the number of dimensions updated at each stage is dynamically calculated based on the knowledge rate K and the iteration progress:

[0096]

[0097]

[0098] In the formula, G is the current algebra, and GEN is the maximum algebra. The original number of dimensions. This represents the change in the number of dimensions. This represents the updated number of dimensions.

[0099] In the Junior stage (early to mid-stage, acquiring and sharing knowledge), individuals primarily learn from neighboring superior or inferior individuals and from random entities, and according to probability. To decide whether to update, follow these steps:

[0100]

[0101] In the formula, For the updated individual,

[0102] In the Senior stage (mid-to-late stage, acquiring and sharing knowledge), individuals learn from the excellent, average, and poor groups, updating their knowledge according to the following formula:

[0103]

[0104] In the formula, Randomly selected from the top group, Randomly selected from the bottom group. Randomly selected from the middle group.

[0105] Step S505: Based on the new generation population, return to the step of calculating the fitness value of each individual in the initial population, selecting the optimal individual based on the fitness value, comparing the fitness value of the new generation population with the fitness value of the population before the update, and retaining the population with better fitness and its corresponding optimal individual, until the preset maximum number of iterations is reached, and outputting the hyperparameters corresponding to the individual with the highest fitness value as the optimal hyperparameter configuration of the extreme learning machine.

[0106] Among them, the sum of the fitness of the old and new populations is compared, and the following strategy is used for selection:

[0107]

[0108] In the formula, For the old individual, For the updated individual.

[0109] Based on the same inventive concept, this application also provides a multi-energy carbon emission prediction system for implementing the multi-energy carbon emission prediction method mentioned above.

[0110] The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more multi-energy carbon emission prediction system embodiments provided below can be found in the limitations of the multi-energy carbon emission prediction method above, and will not be repeated here.

[0111] like Figure 3 As shown in the figure, this application provides a multi-energy carbon emission prediction system, including:

[0112] The data acquisition module 100 is used to acquire power operation sample data and carbon emission sample data of power companies for multiple energy consumption types.

[0113] The dataset construction module 200 is used to extract features from the power operation sample data to obtain the operation sample feature data, and to use the operation sample feature data and carbon emission sample data as mapping samples to construct the training dataset.

[0114] The prediction model training module 300 is used to build an extreme learning machine based on the AdaBoost algorithm optimized by combining the eGSK shared knowledge optimization algorithm. The extreme learning machine is trained using the training dataset to obtain a trained carbon emission prediction model.

[0115] The carbon emission prediction module 400 is used to input the operating characteristic data corresponding to the power operation data of the power company in the current calculation period into the trained carbon emission prediction model, so that the trained carbon emission prediction model outputs the carbon emission prediction value for the current calculation period.

[0116] like Figure 4 As shown, this application provides an electronic device. The electronic device 10 includes a memory 20 and a processor 30. The memory 20 stores a computer program. When the computer program is executed by the processor 30, the processor 30 performs the steps of the multi-energy consumption carbon emission prediction method as described in the above embodiment.

[0117] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed, implements the steps of the multi-energy consumption carbon emission prediction method as described in the above embodiments.

[0118] This application provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the steps of the multi-energy consumption carbon emission prediction method as described in the above embodiments.

[0119] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, electronic devices, computer storage media, and computer program products described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0120] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products or devices.

[0121] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0122] In the several embodiments provided by this invention, it should be understood that the disclosed systems, electronic devices, computer storage media, computer program products, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.

[0123] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0124] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0125] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods described in the various embodiments of the present invention through a computer device (which may be a personal computer, a server, or a network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0126] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for multi-energy carbon emission prediction, characterized in that, The method comprises the following steps: obtaining power operation sample data of multiple energy consumption types of a power enterprise and carbon emission sample data; performing feature extraction on the power operation sample data to obtain operation sample feature data, and taking the operation sample feature data and the carbon emission sample data as mapping samples to construct a training data set; constructing an extreme learning machine based on an AdaBoost algorithm optimized by an eGSK shared knowledge optimization algorithm, training the extreme learning machine by using the training data set, and obtaining a trained carbon emission prediction model; inputting operation feature data corresponding to power operation data of the power enterprise in a current calculation period into the trained carbon emission prediction model, and enabling the trained carbon emission prediction model to output a carbon emission prediction value of the current calculation period.

2. The method of claim 1, wherein, The energy consumption types include raw coal, natural gas, gasoline and diesel oil; the feature extraction on the power operation sample data to obtain operation sample feature data comprises the following steps: performing dimension reduction processing on the power operation sample data by using a principal component analysis method to obtain reduced sample data; performing feature extraction on the reduced sample data to obtain operation sample feature data; wherein the operation sample feature data comprises at least one of operation sample mean value data, operation sample variance data and operation sample extreme value data.

3. The method of claim 1, wherein, The construction of the extreme learning machine based on the AdaBoost algorithm comprises the following steps: initializing basic parameters of the extreme learning machine, wherein the basic parameters include the number of input layer nodes, the number of hidden layer nodes and the number of output layer nodes; constructing multiple weak predictors based on the AdaBoost algorithm, wherein each weak predictor adopts an extreme learning machine structure; linearly combining the multiple weak predictors to form a strong predictor based on the AdaBoost algorithm, and obtaining the extreme learning machine based on the AdaBoost algorithm.

4. The method of claim 1, wherein, The hyperparameters of the extreme learning machine are optimized by using an eGSK shared knowledge optimization algorithm.

5. The method of claim 4, wherein, The optimization of the hyperparameters of the extreme learning machine by using the eGSK shared knowledge optimization algorithm comprises the following steps: setting initial parameters of an eGSK heuristic optimization algorithm, wherein the initial parameters include the population size, the maximum number of iterations and the knowledge rate; randomly generating an initial population in a parameter space, wherein each individual represents a set of hyperparameters of the extreme learning machine; calculating the fitness value of each individual in the initial population, and selecting an optimal individual according to the fitness value; performing knowledge sharing and updating according to the knowledge rate and the iteration progress to generate a new generation population; based on the new generation population, returning to the step of calculating the fitness value of each individual in the initial population, selecting an optimal individual according to the fitness value, comparing the fitness values of the new generation population and the population before updating, and retaining the population with a better fitness value and the corresponding optimal individual until a preset maximum number of iterations is reached, outputting the hyperparameters corresponding to the individual with the highest fitness value as the optimal hyperparameter configuration of the extreme learning machine.

6. The method of claim 5, wherein, The knowledge sharing and updating according to the knowledge rate and the iteration progress to generate a new generation population comprises the following steps: According to the knowledge rate and the iteration progress, the number of dimensions updated in each stage is determined; According to the determined number of dimensions, the optimal individual in the current population is selected in each stage for knowledge sharing, and the dimension information of the optimal individual is transmitted to other individuals in the current population; The individuals receiving knowledge transmission are updated to generate a new generation of population.

7. A multi-energy carbon emission prediction system, characterized by, Comprise: The data acquisition module is used for acquiring power operation sample data of multiple energy consumption types of the power enterprise and carbon emission sample data; The data set construction module is used for extracting features from the power operation sample data to obtain operation sample feature data, and constructing a training data set by taking the operation sample feature data and the carbon emission sample data as mapping samples; The prediction model training module is used for constructing an extreme learning machine based on an AdaBoost algorithm optimized by combining an eGSK shared knowledge optimization algorithm, training the extreme learning machine by using the training data set, and obtaining a trained carbon emission prediction model; The carbon emission prediction module is used for inputting operation feature data corresponding to power operation data of the power enterprise in a current calculation period into the trained carbon emission prediction model, so that the trained carbon emission prediction model outputs a carbon emission prediction value of the current calculation period.

8. An electronic device, comprising: The electronic device includes a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor, so that the processor executes the steps of the multi-energy consumption carbon emission prediction method according to any one of claims 1-6.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed to realize the steps of the multi-energy consumption carbon emission prediction method according to any one of claims 1-6.

10. A computer program product, characterised in that, The computer program product comprises a computer program stored on a non-transitory computer readable storage medium, and the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the steps of the multi-energy consumption carbon emission prediction method according to any one of claims 1-6.

Images