A method and system for predicting carbon emissions of steel projects based on historical projects

Abstract

The present invention belongs to the technical field of carbon emission prediction, and discloses a method and system for predicting carbon emissions of steel engineering based on historical projects. The method comprises the following steps: collecting a number of historical full life cycle data; performing dimensionality reduction of historical data; constructing an artificial intelligence model; obtaining current real-time full life cycle data; performing dimensionality reduction of real-time data; predicting carbon emissions of steel engineering; generating a model using a carbon emission management strategy; continuing to collect real-time project stage data of the next project stage to form the next real-time full life cycle data; returning to the real-time data dimensionality reduction step. The present invention solves the problems of low utilization of historical data, low prediction accuracy, low efficiency, poor effect, poor data processing capability, lack of comprehensiveness and feedback mechanism of carbon emission prediction results existing in the prior art.

Description

Technical Field

[0001] The present invention belongs to the technical field of carbon emission prediction, and specifically relates to a method and system for predicting carbon emission of steel projects based on historical projects. Background Art

[0002] Steel projects generate a large amount of carbon emissions during the production process, which has an impact on the environment. In order to effectively control and reduce carbon emissions, it is necessary to accurately predict and manage carbon emissions from steel projects. However, the existing steel project carbon emission prediction technology has the following defects:

[0003] 1) Existing steel engineering carbon emission prediction technology only relies on the analysis of real-time steel engineering project data, ignoring the rich information and guiding role of historical steel engineering projects, resulting in the inability to fully utilize the laws and trends in historical data, wasting the value of historical steel engineering project data, and low utilization of historical data, which reduces the accuracy of prediction;

[0004] 2) Most of the existing carbon emission prediction technologies for steel projects rely on human analysis of relevant data, which has poor data processing capabilities and is unable to discover the key information hidden in the data. In addition, the efficiency is low and the prediction effect of carbon emission for steel projects is poor, which cannot meet the needs of large-scale data prediction;

[0005] 3) The existing carbon emission prediction technology for steel projects only analyzes the production and operation data after the delivery of real-time steel project and uses it as the data basis for carbon emission prediction for steel projects. It ignores the factors related to carbon emission prediction for steel projects throughout the entire life cycle of steel projects and lacks comprehensiveness and feedback mechanism for carbon emission prediction results. Summary of the invention

[0006] In order to solve the problems of low historical data utilization, low prediction accuracy, low efficiency, poor effect, poor data processing capability, lack of comprehensiveness and feedback mechanism of carbon emission prediction results in the prior art, the present invention aims to provide a steel engineering carbon emission prediction method and system based on historical projects.

[0007] The technical solution adopted by the present invention is:

[0008] According to a first aspect of an embodiment of the present invention, a method for predicting carbon emissions of a steel project based on historical projects is provided, comprising the following steps:

[0009] Step 1: Collect historical full life cycle data of several historical steel engineering projects including at least one project phase, and perform preprocessing to obtain several preprocessed historical full life cycle data;

[0010] Step 2: Perform data dimensionality reduction on a number of preprocessed historical full life cycle data to obtain a corresponding principal component set and a number of historical full life cycle data after dimensionality reduction;

[0011] Step 3: Based on several historical life cycle data after dimensionality reduction, use artificial intelligence algorithms to build a steel engineering carbon emission prediction model and a carbon emission management strategy generation model;

[0012] Step 4: Collect the real-time project phase data of the real-time steel engineering project at the current project phase, and all the real-time project phase data of other project phases corresponding to the current project phase, to form the current real-time full life cycle data; other project phases corresponding to the current project phase indicate the project phases before the current project phase in the real-time steel engineering project;

[0013] Step 5: According to the principal component set, the current real-time full life cycle data is reduced in dimension to obtain the current real-time full life cycle data after dimension reduction;

[0014] Step 6: Based on the current real-time full life cycle data after dimensionality reduction, use the steel engineering carbon emission prediction model to predict the carbon emissions of the steel engineering and obtain the current real-time carbon emission prediction results;

[0015] Step 7: Based on the current real-time carbon emission prediction results, use the carbon emission management strategy generation model to generate a carbon emission management strategy to obtain the current real-time carbon emission management strategy;

[0016] Step 8: Continue to collect the real-time project phase data of the real-time steel engineering project in the next project phase, and all the real-time project phase data of other project phases corresponding to the next project phase, to form the next real-time full life cycle data; the other project phases corresponding to the next project phase indicate the project phases before the next project phase in the real-time steel engineering project;

[0017] Step 9: Use the next real-time full life cycle data as the current real-time full life cycle data and return to the real-time data dimensionality reduction step, that is, return to execute steps 5 to 8 until the next real-time full life cycle data includes real-time project stage data of all project stages.

[0018] Furthermore, preprocessing includes data mapping and transformation, data cleaning, and normalization.

[0019] Furthermore, the PCA method is used to perform data dimensionality reduction on a number of preprocessed historical full life cycle data to obtain a corresponding principal component set and a number of historical full life cycle data after dimensionality reduction.

[0020] Furthermore, based on several historical life cycle data after dimensionality reduction, an artificial intelligence algorithm is used to construct a steel engineering carbon emission prediction model and a carbon emission management strategy generation model, including the following steps:

[0021] Based on several historical life cycle data after dimensionality reduction, a fusion algorithm is used to build a carbon emission prediction model for steel projects and generate several historical carbon emission prediction results;

[0022] Based on several historical carbon emission prediction results, a carbon emission management strategy generation model is constructed using a reinforcement learning algorithm, and several historical carbon emission management strategy generation experiences are generated.

[0023] Furthermore, based on several dimensionality-reduced historical life cycle data, a fusion algorithm is used to construct a steel engineering carbon emission prediction model and generate several historical carbon emission prediction results, including the following steps:

[0024] Using a deep learning algorithm, an initial steel engineering carbon emission prediction model is constructed, and the initial network parameters of the initial steel engineering carbon emission prediction model are used as the optimization target. A swarm intelligence optimization algorithm is used to search and optimize to obtain the optimal initial network parameters of the initial steel engineering carbon emission prediction model; the optimal initial network parameters indicate the initial network parameters obtained when the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements;

[0025] Labels are added to a number of historical full life cycle data after dimensionality reduction to obtain a number of steel engineering carbon emission prediction samples with real labels, and the number of steel engineering carbon emission prediction samples with real labels are divided into a model training set and a model test set according to a preset ratio;

[0026] Based on the optimal initial network parameters, the initial steel engineering carbon emission prediction model is optimized to obtain an optimized steel engineering carbon emission prediction model, and the optimized steel engineering carbon emission prediction model is trained and optimized according to the model training set to obtain a trained steel engineering carbon emission prediction model;

[0027] According to the model test set, the trained steel engineering carbon emission prediction model is tested to obtain several historical prediction labels, which are compared and counted with several real labels in the model test set to obtain the model test accuracy;

[0028] If the model test accuracy is greater than the accuracy threshold, the trained steel engineering carbon emission prediction model is output as the optimal steel engineering carbon emission prediction model, so that the optimal steel engineering carbon emission prediction model becomes the constructed steel engineering carbon emission prediction model; otherwise, the training optimization of the steel engineering carbon emission prediction model continues.

[0029] Furthermore, the deep learning algorithm is the LSTM-Attention-cGAN algorithm;

[0030] The swarm intelligence optimization algorithm is the ICPO algorithm.

[0031] Furthermore, based on several historical carbon emission prediction results, a reinforcement learning algorithm is used to construct a carbon emission management strategy generation model, and several historical carbon emission management strategy generation experiences are generated, including the following steps:

[0032] The carbon emission management strategy generation problem is used as the simulation environment for the reinforcement learning algorithm to build an intelligent agent and experience replay pool;

[0033] Define the state space of the reinforcement learning algorithm based on the carbon emission prediction state contained in the historical carbon emission prediction results;

[0034] Define the action space of the reinforcement learning algorithm according to the preset actions included in the carbon emission management strategy, and define the reward function based on the action space;

[0035] Construct a network structure of a neural network including an input layer and an output layer, connect the input layer to the state space, and connect the output layer to the action space;

[0036] Construct an initial carbon emission management strategy generation model based on action space, state space, reward function, neural network, intelligent agent, and experience replay pool;

[0037] According to several historical carbon emission prediction results, the neural network and intelligent agent of the initial carbon emission management strategy generation model are optimized and trained until the optimization training results meet the preset conditions, and the final carbon emission management strategy generation model is obtained. The final carbon emission management strategy generation model is used as the constructed carbon emission management strategy generation model, and several historical carbon emission management strategy generation experiences are generated;

[0038] Several historical carbon emission management strategy generation experiences are stored in the experience playback pool of the carbon emission management strategy generation model.

[0039] Furthermore, the reinforcement learning algorithm is a DQN algorithm, and the neural network is a deep Q network.

[0040] Further, according to the current real-time carbon emission prediction result, the carbon emission management strategy generation model is used to generate the carbon emission management strategy to obtain the current real-time carbon emission management strategy, including the following steps:

[0041] According to the current real-time carbon emission prediction results, extract several matching historical carbon emission management strategy generation experiences from the experience playback pool of the carbon emission management strategy generation model;

[0042] According to the current real-time carbon emission prediction results, the state space of the carbon emission management strategy generation model is updated to obtain an updated state space, and according to a number of matching historical carbon emission management strategy generation experiences, the action space of the carbon emission management strategy generation model is updated to obtain an updated action space;

[0043] Input the updated state space into the deep Q network of the carbon emission management strategy generation model, and use the deep Q network to generate the Q value of each possible action in the updated action space based on the updated state space;

[0044] Using the agent, according to the reward function, obtain the reward value of each possible action in the updated action space, and according to the reward value, update the Q value of the possible action to obtain the updated Q value of each possible action;

[0045] Repeat the above Q-value update steps until the number of iterations reaches the iteration number threshold, and use the greedy strategy to select the possible action with the highest updated Q-value in each iteration as the execution action;

[0046] Using intelligent agents, several execution actions are integrated to obtain the current real-time carbon emission management strategy.

[0047] According to a second aspect of an embodiment of the present invention, a carbon emission prediction system for steel projects based on historical projects is provided, which is used to implement a carbon emission prediction method for steel projects. The system includes a historical project data collection unit, a historical data dimensionality reduction unit, an artificial intelligence model building unit, a real-time data collection unit, a real-time data dimensionality reduction unit, a steel project carbon emission prediction unit, a carbon emission management strategy generation unit, a real-time data continuous collection unit and a step return unit, which are connected in sequence.

[0048] The beneficial effects of the present invention are:

[0049] The present invention provides a method and system for predicting carbon emissions from steel projects based on historical projects. By processing and effectively utilizing the historical full life cycle data of historical projects, the present invention retains the rich information and guiding role contained in historical steel projects, makes full use of the laws and trends in historical data, avoids the waste of value of historical steel project data, and improves the accuracy of carbon emission prediction for steel projects. The carbon emission prediction model for steel projects constructed using an artificial intelligence algorithm can mine deep key information from steel project data and perform automated and intelligent carbon emission prediction for steel projects, thereby improving the accuracy of carbon emission prediction and the effect of carbon emission prediction, and is suitable for carbon emission prediction for large-scale and large-scale steel projects. The constructed carbon emission management strategy generation model can automatically generate and adjust management strategies according to the carbon emission forecast of each project stage, guide the carbon emission management work of the next project stage, and realize the feedback mechanism of carbon emission forecast results; it uses the full life cycle data of the steel engineering project as the data basis, retains all factors related to the carbon emission forecast of the steel engineering project throughout the life cycle of the steel engineering project, and improves the comprehensiveness of analysis and forecasting; by supplementing the data of the next project stage to improve the real-time data continuous collection mechanism of the full life cycle data, the forecast results and management strategies are continuously updated, and the continuous forecast and management of carbon emissions of steel engineering are realized, which can timely reflect changes in project operation and improve the real-time and accuracy of forecasting and management strategies.

[0050] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flowchart of the carbon emission prediction method for steel engineering based on historical projects in the present invention.

[0052] Figure 2 It is a structural block diagram of the steel engineering carbon emission prediction system based on historical projects in the present invention. DETAILED DESCRIPTION

[0053] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Embodiment 1:

[0055] like Figure 1 As shown, this embodiment provides a method for predicting carbon emissions of steel projects based on historical projects, including the following steps:

[0056] Step 1: Collect historical full life cycle data of several historical steel engineering projects including at least one project phase, and perform preprocessing to obtain several preprocessed historical full life cycle data;

[0057] Step 2: Perform data dimensionality reduction on a number of preprocessed historical full life cycle data to obtain a corresponding principal component set and a number of historical full life cycle data after dimensionality reduction;

[0058] Step 3: Based on several historical life cycle data after dimensionality reduction, use artificial intelligence algorithms to build a steel engineering carbon emission prediction model and a carbon emission management strategy generation model;

[0059] Step 4: Collect the real-time project phase data of the real-time steel engineering project at the current project phase, and all the real-time project phase data of other project phases corresponding to the current project phase, to form the current real-time full life cycle data; other project phases corresponding to the current project phase indicate the project phases before the current project phase in the real-time steel engineering project;

[0060] Step 5: According to the principal component set, the current real-time full life cycle data is reduced in dimension to obtain the current real-time full life cycle data after dimension reduction;

[0061] Step 6: Based on the current real-time full life cycle data after dimensionality reduction, use the steel engineering carbon emission prediction model to predict the carbon emissions of the steel engineering and obtain the current real-time carbon emission prediction results;

[0062] Step 7: Based on the current real-time carbon emission prediction results, use the carbon emission management strategy generation model to generate a carbon emission management strategy to obtain the current real-time carbon emission management strategy;

[0063] Step 8: Continue to collect the real-time project phase data of the real-time steel engineering project in the next project phase, and all the real-time project phase data of other project phases corresponding to the next project phase, to form the next real-time full life cycle data; the other project phases corresponding to the next project phase indicate the project phases before the next project phase in the real-time steel engineering project;

[0064] Step 9: Based on the next real-time full life cycle data as the current real-time full life cycle data, return to the real-time data dimensionality reduction step, that is, return to execute steps 5 to 8 until the next real-time full life cycle data includes real-time project stage data of all project stages.

[0065] In a specific embodiment, the method for predicting carbon emissions from steel projects based on historical projects includes the following steps:

[0066] S1: Collecting historical full life cycle data of several historical steel engineering projects including at least one project phase, and preprocessing them to obtain several preprocessed historical full life cycle data, including the following steps:

[0067] S1-1: Collect heterogeneous historical life cycle data of several historical steel engineering projects, including at least one project phase;

[0068] Among them, the project stages include, in chronological order, the design stage, the procurement stage, the construction stage, the commissioning and trial production stage, the production operation stage, and the maintenance and transformation stage;

[0069] Project phase data during the design phase include forecasted energy consumption and forecasted carbon emissions;

[0070] Project phase data in the procurement phase includes raw material attributes, procurement equipment data, and supplier information;

[0071] Project phase data during the construction phase include energy consumption of construction equipment and carbon emissions of building materials;

[0072] Project phase data during the commissioning and trial production phase include preliminary production data, commissioning and trial production energy consumption, and commissioning and trial production carbon emission data;

[0073] Project phase data for the production and operation phase include production and operation energy consumption, raw material usage, production process parameters, and production and operation emission data;

[0074] The project phase data of the maintenance and renovation phase include maintenance and renovation equipment performance data, maintenance and renovation energy efficiency, and maintenance and renovation emission data;

[0075] S1-2: Map and convert a number of heterogeneous historical full life cycle data to obtain a number of homogeneous converted historical full life cycle data;

[0076] Specifically, the data systems used in different project stages are different, resulting in different data formats for historical life cycle data, which affects subsequent model building and analysis and prediction. By mapping and converting heterogeneous data into homogeneous data, data support can be provided for subsequent model building and analysis and prediction.

[0077] S1-3: Perform data cleaning on a number of converted historical full life cycle data of the same structure to obtain a number of cleaned historical full life cycle data;

[0078] Among them, data cleaning includes: missing value processing, filling or deleting missing data, outlier detection and processing, identifying and processing abnormal or outlier values, duplicate data processing, deleting duplicate records, and improving data quality;

[0079] S1-4: normalize a number of cleaned historical full life cycle data to obtain a number of pre-processed historical full life cycle data;

[0080] Among them, normalization processing unifies the magnitude of different data to eliminate the accuracy impact caused by magnitude differences;

[0081] S2: Using the principal component analysis (PCA) method, the data dimension reduction is performed on a number of pre-processed historical full life cycle data to obtain a corresponding principal component set and a number of reduced-dimension historical full life cycle data, including the following steps:

[0082] S2-1: Perform matrix conversion on some pre-processed historical life cycle data to obtain the corresponding historical life cycle data matrix. The formula is:

[0083] X=[x1,x2,...x p ,...,x n ] T

[0084] Among them, x p is the row vector of the p-th preprocessed historical full life cycle data, p is the row vector indicator, and n is the total number of preprocessed historical full life cycle data;

[0085] The initial row vector of the historical full life cycle data matrix is ​​the preprocessed historical full life cycle data, and the initial column vector of the historical full life cycle data matrix is ​​the historical full life cycle indicator data, each indicator data corresponds to the data of each column in the historical full life cycle data matrix, and the indicator data corresponds to various data contained in the project stage data in the project stage, for example, the types of indicator data include raw material attributes, procurement equipment data, preliminary production data, debugging and trial production energy consumption and other indicator data;

[0086] S2-2: Standardize the historical full life cycle data matrix to obtain the corresponding standardized historical full life cycle data matrix;

[0087] The formula is:

[0088]

[0089] Where X' is the historical full life cycle data matrix after standardization; μ is the mean of the historical full life cycle data matrix; σ is the variance of the historical full life cycle data matrix;

[0090] S2-3: Obtain the covariance matrix of the standardized historical full life cycle data matrix, and obtain the corresponding candidate principal component matrix based on the standardized historical full life cycle data matrix and the covariance matrix; the candidate row vectors of the candidate principal component matrix are the preprocessed historical full life cycle data, and the candidate column vectors of the candidate principal component matrix are the historical full life cycle indicator data;

[0091] The formula is:

[0092]

[0093] Where D is the covariance matrix of the historical full life cycle data matrix after standardization; Y is the candidate principal component matrix; P is the conversion matrix; E is the unit eigenvector matrix; n is the total number of historical full life cycle data after preprocessing;

[0094] Y=PX'

[0095] In the formula, Y is the candidate principal component matrix; P is the transformation matrix; X' is the historical full life cycle data matrix after standardization;

[0096] S2-4: Set the candidate principal component matrix Y = [y1, y2, ...y l ,...,y L ] Several candidate column vectors y' of the top 90% of the cumulative contribution rate of variance l As the corresponding principal component column vectors, the reduced-dimensional historical full-life cycle data matrix Y'=[y'1,y'2,...y' l ,...,y K ] and the principal component set;

[0097] The key row vector of the historical full life cycle data matrix after dimensionality reduction is the historical full life cycle data after dimensionality reduction, and the data index corresponding to the key column vector of the historical full life cycle data matrix after dimensionality reduction is the principal component, and the principal component set is obtained;

[0098] The formula is:

[0099]

[0100] In the formula, G is the cumulative contribution rate of variance; l The lth candidate principal component y l The variance of ; l is the indicator of the candidate principal component; L is the total number of candidate principal components; K is the total number of principal components;

[0101] The data indicators included in the principal component set, for example, the principal components include raw material attribute indicators, preliminary production indicators, commissioning and trial production energy consumption indicators, etc., which are highly correlated with the carbon emissions of steel projects, while the purchased equipment indicators, as non-principal components, are less correlated with the carbon emissions of steel projects;

[0102] S2-5: Perform an inverse matrix transformation on the historical full life cycle data matrix after dimension reduction to obtain a number of historical full life cycle data after dimension reduction;

[0103] S3: Based on several historical life cycle data after dimensionality reduction, an artificial intelligence algorithm is used to build a steel engineering carbon emission prediction model and a carbon emission management strategy generation model, including the following steps:

[0104] S3-1: Based on several historical life cycle data after dimensionality reduction, the Improved Crested Porcupine Optimizer (ICPO)-Long Short-Term Memory (LSTM)-Attention-Conditional Generative Adversarial Network (cGAN) algorithm is used to build a steel engineering carbon emission prediction model and generate several historical carbon emission prediction results, including the following steps:

[0105] S3-1-1: Use the LSTM-Attention-cGAN algorithm to build an initial steel engineering carbon emission prediction model, and use the initial network parameters of the steel engineering carbon emission prediction model as the optimization target. Use the ICPO algorithm to search and optimize to obtain the optimal initial network parameters of the initial steel engineering carbon emission prediction model, including the following steps:

[0106] S3-1-1-1: Use the LSTM-Attention-cGAN algorithm to build an initial steel engineering carbon emission prediction model;

[0107] The input layer is used to receive the full life cycle data. According to the number of input features of the full life cycle data including all project phase data, the neuron structure of the input layer is set. According to the data structure of the full life cycle data including all project phase data, the missing bits of all full life cycle data are filled with 0 values. For example, the project phase data of the steel engineering project in the design project phase is collected. At this time, if the project phase data of other phases is missing, the missing bits will be filled with 0 values ​​to unify the input feature structure of the data. The LSTM network realizes feature extraction of the data through multiple gating mechanisms and has a memory function, which improves the accuracy and stability of the model. It is used to extract features from time series data and convert the processed data into a hidden state sequence. The Attention mechanism realizes the splicing of key features through preset attention weights, integrates scattered features, and enhances the key features for the standard. The influence of signature prediction is increased, which increases the flexibility and explanatory power of the model. The cGAN network generates data based on conditional input, simulates different operating scenarios or generates anomaly detection samples, which are used to input the attention weights and LSTM network outputs into the cGAN as conditions for adversarial training. Adversarial training helps the model learn more complex distributions and improve generalization ability. The cGAN generator tries to generate data similar to the real data distribution, while the discriminator tries to distinguish between the real data and the data generated by the generator. The softmax classification function is used according to the output of the adversarial training module to perform feature classification and obtain the corresponding steel engineering carbon emission prediction results. According to the preset steel engineering carbon emission prediction situation, the steel engineering carbon emission prediction label is encoded to obtain the steel engineering carbon emission prediction coding vector, and the number of neurons in the classification layer and the output layer is set according to the steel engineering carbon emission prediction coding vector.

[0108] S3-1-1-2: Taking the initial network parameters of the steel engineering carbon emission prediction model as the optimization target, set the ICPO individual decoding vector, fitness function, ICPO population parameters and maximum number of iterations;

[0109] The formula for the fitness value is:

[0110]

[0111] Where fit is the fitness function; MSE is the mean square error function; Y n' is the true output value of the n'th classification layer neuron; is the ideal output value of the n'th classification layer neuron; n' is the classification layer neuron indicator; N is the total number of classification layer neurons;

[0112] Since deep learning models are sensitive to initial values ​​and prone to premature maturation, they are prone to overfitting. In order to solve the above problems, the ICPO algorithm is used to search for initial values.

[0113] S3-1-1-3: According to the coding vector of the ICPO individual and the ICPO population parameters, the Circle chaotic mapping sequence is used for initialization to generate several initial ICPO individuals of the initial ICPO population;

[0114] The formula is:

[0115]

[0116] In the formula, is the initial ICPO individual of the Circle chaos map, i.e. the initial solution; is the randomly generated initial ICPO individual; i' is the ICPO individual indicator;

[0117] S3-1-1-4: Introduce a cyclic population reduction mechanism to limit the number of individuals in the ICPO population parameters and obtain updated ICPO population parameters for the next iteration;

[0118] The formula is:

[0119]

[0120] In the formula, S t+1 is the number of individuals in the ICPO population parameter of the t+1th iteration; S t is the number of individuals in the ICPO population parameter of the tth iteration; S min is the minimum number of individuals in the ICPO population parameter; a' is the function evaluation parameter; V is the function evaluation cycle parameter; V max is the maximum function evaluation loop parameter; t is the number of iterations indicator;

[0121] S3-1-1-5: Calculate the initial fitness value of the initial ICPO individuals in the initial ICPO population according to the fitness function;

[0122] S3-1-1-6: According to the initial fitness value and the updated ICPO population parameters, the first defense strategy, the second defense strategy, the third defense strategy and the fourth defense strategy are used to update the initial ICPO population to obtain an updated ICPO population;

[0123] The formula for the first defense strategy is:

[0124]

[0125] In the formula, For the updated ICPO individuals within the first defense range; is the initial ICPO individual within the first defense range; τ1 is a random number based on normal distribution; τ2 is a random value in the interval [0,1]; It is the optimal solution within the first defense range; is the vector generated between the true optimal solution within the first defense range and the optimal solution randomly selected from the ICPO population; i' is the ICPO individual indicator; t is the iteration indicator;

[0126] The formula for the second defense strategy is:

[0127]

[0128] In the formula, For the updated ICPO individuals within the second defense range; The initial ICPO individual within the second defense range; is the search upper limit vector of the second defense range; τ3 is a random value in the interval [0,1]; are the r1th and r2th initial ICPO individuals respectively; r1 and r2 are two random integers between [1, S]; is the vector generated between the true optimal solution within the second defense range and the optimal solution randomly selected from the ICPO population;

[0129] The formula for the third defense strategy is:

[0130]

[0131] In the formula, For the updated ICPO individuals within the third defense range; The initial ICPO individual within the third defense range; is the search upper limit vector of the third defense range; are the r2th and r3th initial ICPO individuals respectively; r3 is a random integer between [1, S]; The odor diffusion factor defined for the fitness function; λ t It is a defense factor; Control parameters for search direction;

[0132] The formula for the fourth defense strategy is:

[0133]

[0134] In the formula, For the updated ICPO individuals within the fourth defense range; The initial ICPO individual within the fourth defense range; is the optimal solution within the fourth defense range; τ4 and τ5 are both random values ​​in the interval [0,1]; λt It is a defense factor; Control parameters for search direction; is the average force affecting the search direction; a' is the convergence speed factor;

[0135] S3-1-1-7: Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated ICPO population to generate a dynamic reverse ICPO population;

[0136] The formula is:

[0137]

[0138] In the formula, is the ICPO individual in dynamic reverse direction; γ' is the decreasing inertia coefficient; L max , L min are the maximum and minimum values ​​of the vector space respectively; For the updated ICPO individual;

[0139] S3-1-1-8: According to the fitness function, calculate the fitness values ​​of all ICPO individuals in the updated ICPO population and the dynamically reversed ICPO population, take the ICPO individual with the minimum fitness value as the optimal individual, and retain the optimal individual;

[0140] S3-1-1-9: If the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal solution corresponding to the optimal individual is output to obtain the optimal initial network parameters of the initial steel engineering carbon emission prediction model;

[0141] S3-1-2: adding labels to a number of historical full life cycle data after dimensionality reduction to obtain a number of steel engineering carbon emission prediction samples with real labels, and dividing the number of steel engineering carbon emission prediction samples with real labels into a model training set and a model test set according to a preset ratio; preferably, the preset ratio can be set to 7:3;

[0142] S3-1-3: Based on the optimal initial network parameters, the initial steel engineering carbon emission prediction model is optimized to obtain an optimized steel engineering carbon emission prediction model, and the optimized steel engineering carbon emission prediction model is trained and optimized according to the model training set to obtain a trained steel engineering carbon emission prediction model;

[0143] S3-1-4: Based on the model test set, the trained steel engineering carbon emission prediction model is tested to obtain several historical prediction labels, which are compared and counted with several real labels in the model test set to obtain the model test accuracy;

[0144] S3-1-5: If the model test accuracy is greater than the accuracy threshold, the trained steel engineering carbon emission prediction model is output as the optimal steel engineering carbon emission prediction model, so that the optimal steel engineering carbon emission prediction model becomes the constructed steel engineering carbon emission prediction model; otherwise, continue to train and optimize the steel engineering carbon emission prediction model, that is, repeat steps S3-1-1 to S3-1-4 until the model test accuracy is greater than the accuracy threshold;

[0145] S3-2: Based on several historical carbon emission prediction results, a reinforcement learning algorithm is used to build a carbon emission management strategy generation model and generate several historical carbon emission management strategy generation experiences, including the following steps:

[0146] S3-2-1: Use the carbon emission management strategy generation problem as the simulation environment for the reinforcement learning algorithm to build an intelligent agent and experience replay pool;

[0147] S3-2-2: Define the state space of the reinforcement learning algorithm based on the carbon emission prediction state contained in the historical carbon emission prediction results;

[0148] S3-2-3: Define the action space of the reinforcement learning algorithm based on the preset actions included in the carbon emission management strategy, and define the reward function based on the action space;

[0149] S3-2-4: Construct a network structure including an input layer and an output layer of a neural network, connect the input layer to the state space, and connect the output layer to the action space;

[0150] S3-2-5: Construct an initial carbon emission management strategy generation model based on action space, state space, reward function, neural network, intelligent agent, and experience replay pool;

[0151] S3-2-6: Based on several historical carbon emission prediction results, the neural network and intelligent agent of the initial carbon emission management strategy generation model are optimized and trained until the optimization training results meet the preset conditions, and the final carbon emission management strategy generation model is obtained. The final carbon emission management strategy generation model is used as the constructed carbon emission management strategy generation model, and several historical carbon emission management strategy generation experiences are generated;

[0152] S3-2-7: Store several historical carbon emission management strategy generation experiences into the experience playback pool of the carbon emission management strategy generation model;

[0153] The reinforcement learning algorithm is the Deep Q Network (DQN) algorithm, and the neural network is the Deep Q Network;

[0154] S4: Collect the real-time project phase data of the real-time steel engineering project in the current project phase, and all the real-time project phase data of other project phases corresponding to the current project phase, to form the current real-time full life cycle data; the other project phases corresponding to the current project phase indicate the project phases before the current project phase in the real-time steel engineering project;

[0155] For example, real-time project phase data of a real-time steel engineering project in the construction phase is collected and combined with all real-time project phase data of the corresponding design phase and procurement phase to form the current real-time full life cycle data;

[0156] S5: According to the principal component set, the current real-time full life cycle data is reduced in dimension to obtain the current real-time full life cycle data after dimension reduction;

[0157] Specifically, the PCA method transforms the original data into a set of linearly independent representations in each dimension through linear transformation, which is used to remove noise and redundancy in the data, making the data easier to understand and process;

[0158] S6: Based on the current real-time full life cycle data after dimensionality reduction, use the steel engineering carbon emission prediction model to predict the carbon emissions of the steel engineering and obtain the current real-time carbon emission prediction results;

[0159] For example, the current carbon emissions in the design phase are normal, the carbon emissions in the procurement phase are normal, and the carbon emissions in the construction phase exceed the standard, etc. obtained in this embodiment can reflect the current carbon emissions in the entire life cycle;

[0160] S7: Based on the current real-time carbon emission prediction results, a carbon emission management strategy generation model is used to generate a carbon emission management strategy to obtain the current real-time carbon emission management strategy, including the following steps:

[0161] S7-1: Based on the current real-time carbon emission prediction results, extract several matching historical carbon emission management strategy generation experiences from the experience playback pool of the carbon emission management strategy generation model;

[0162] S7-2: Update the state space of the carbon emission management strategy generation model according to the current real-time carbon emission prediction results, and obtain the updated state space S'=[s'1,...,s' i" ,...,s' I' ], and based on several matching historical carbon emission management strategy generation experiences, update the action space of the carbon emission management strategy generation model to obtain the updated action space A'=[a'1,...,a' j" ,...,a' I ], where s' i" is the updated i-th state value, i" is the state indicator, I' is the total number of state space dimensions, a'j" is the updated j-th "action value", j" is the action indicator, and I is the total number of action space dimensions;

[0163] The updated state space includes the carbon emission prediction states of normal carbon emission in the design phase, normal carbon emission in the procurement phase, and excessive carbon emission in the construction phase. The updated action space includes carbon emission management actions in the commissioning and trial production phases, maintenance and transformation equipment performance adjustment, maintenance and transformation energy efficiency adjustment, and maintenance and transformation emission control;

[0164] S7-3: Input the updated state space into the deep Q network of the carbon emission management strategy generation model, and use the deep Q network to generate the Q value of each possible action in the updated action space based on the updated state space;

[0165] S7-4: Using the agent, according to the reward function, obtain the reward value of each possible action in the updated action space, and according to the reward value, update the Q value of the possible action to obtain the updated Q value of each possible action;

[0166] The formula is:

[0167] Q(s' p' ,a' p' )=(1-α")·Q(s p' ,a p' )+α"·(R(s p' ,a p' ,s' p' )+γ·Q max (s p' ,a p' ))

[0168] In the formula, Q(s' p' ,a' p' ) is the updated state value s' p' and the updated action value a' p' The corresponding updated Q value; Q(s p' ,a p' ) is the state value s p' and action value a p' The corresponding predicted Q value; α" is the learning rate; Q max (s p' ,a p' ) is the highest predicted Q value; p' is the comprehensive indicator; γ is the update parameter;

[0169] S7-5: Repeat the above Q value update step, that is, repeat step S7-4, until the number of iterations reaches the iteration number threshold, and use the greedy strategy to select the possible action with the highest updated Q value in each iteration as the execution action;

[0170] S7-6: Use the intelligent agent to integrate several execution actions to obtain the current real-time carbon emission management strategy, including maintenance and transformation equipment performance improvement, maintenance and transformation energy efficiency improvement, and maintenance and transformation emission reduction;

[0171] S8: Continue to collect the real-time project phase data of the next project phase of the real-time steel engineering project, that is, the real-time project phase data of the debugging and trial production phase, and all the real-time project phase data of other project phases corresponding to the next project phase, to form the next real-time full life cycle data; the other project phases corresponding to the next project phase indicate the project phases before the next project phase in the real-time steel engineering project;

[0172] Specifically, through the phased and continuous collection of project data, the carbon emission forecast is continuously improved, and the real-time carbon emission forecast results obtained can reflect the carbon emission conditions of all project stages that have been collected, and can be used to guide the carbon emission management of the next project stage and control the overall carbon emissions;

[0173] S9: Use the next real-time full life cycle data as the current real-time full life cycle data and return to the real-time data dimensionality reduction step, that is, return to execute steps S5-S8 until the next real-time full life cycle data includes real-time project stage data of all project stages.

[0174] Embodiment 2:

[0175] like Figure 2 As shown, this embodiment provides a carbon emission prediction system for steel engineering based on historical projects, which is used to implement a carbon emission prediction method for steel engineering. The system includes a historical project data collection unit, a historical data dimension reduction unit, an artificial intelligence model construction unit, a real-time data collection unit, a real-time data dimension reduction unit, a steel engineering carbon emission prediction unit, a carbon emission management strategy generation unit, a real-time data continuous collection unit, and a step return unit connected in sequence;

[0176] A historical project data collection unit is used to collect historical full life cycle data of several historical steel engineering projects including at least one project phase, and perform preprocessing to obtain several preprocessed historical full life cycle data;

[0177] A historical data dimension reduction unit is used to perform data dimension reduction on a number of pre-processed historical full life cycle data to obtain a corresponding principal component set and a number of dimension-reduced historical full life cycle data;

[0178] An artificial intelligence model building unit is used to build a steel engineering carbon emission prediction model and a carbon emission management strategy generation model based on a number of historical life cycle data after dimensionality reduction using artificial intelligence algorithms;

[0179] The real-time data collection unit is used to collect the real-time project stage data of the real-time steel engineering project at the current project stage, and all the real-time project stage data of other project stages corresponding to the current project stage, to form the current real-time full life cycle data; the other project stages corresponding to the current project stage indicate the project stages before the current project stage in the real-time steel engineering project;

[0180] The real-time data dimension reduction unit is used to perform data dimension reduction on the current real-time full life cycle data according to the principal component set to obtain the current real-time full life cycle data after dimension reduction;

[0181] The steel engineering carbon emission prediction unit is used to predict the carbon emission of the steel engineering based on the current real-time full life cycle data after dimensionality reduction and use the steel engineering carbon emission prediction model to obtain the current real-time carbon emission prediction results;

[0182] A carbon emission management strategy generation unit is used to generate a carbon emission management strategy based on the current real-time carbon emission prediction result using a carbon emission management strategy generation model to obtain the current real-time carbon emission management strategy;

[0183] The real-time data continuous collection unit is used to continue to collect the real-time project stage data of the real-time steel engineering project in the next project stage, and all the real-time project stage data of other project stages corresponding to the next project stage constitute the next real-time full life cycle data; the other project stages corresponding to the next project stage indicate the project stages before the next project stage in the real-time steel engineering project;

[0184] The step return unit is used to take the next real-time full life cycle data as the current real-time full life cycle data and return to the real-time data dimensionality reduction step, that is, return to execute the real-time data dimensionality reduction unit, the steel engineering carbon emission prediction unit, the carbon emission management strategy generation unit and the real-time data continuous collection unit in sequence until the next real-time full life cycle data includes the real-time project stage data of all project stages.

[0185] The present invention provides a method and system for predicting carbon emissions from steel projects based on historical projects. By processing and effectively utilizing the historical full life cycle data of historical projects, the present invention retains the rich information and guiding role contained in historical steel projects, makes full use of the laws and trends in historical data, avoids the waste of value of historical steel project data, and improves the accuracy of carbon emission prediction for steel projects. The carbon emission prediction model for steel projects constructed using an artificial intelligence algorithm can mine deep key information from steel project data and perform automated and intelligent carbon emission prediction for steel projects, thereby improving the accuracy of carbon emission prediction and the effect of carbon emission prediction, and is suitable for carbon emission prediction for large-scale and large-scale steel projects. The constructed carbon emission management strategy generation model can automatically generate and adjust management strategies according to the carbon emission forecast of each project stage, guide the carbon emission management work of the next project stage, and realize the feedback mechanism of carbon emission forecast results; it uses the full life cycle data of the steel engineering project as the data basis, retains all factors related to the carbon emission forecast of the steel engineering project throughout the life cycle of the steel engineering project, and improves the comprehensiveness of analysis and forecasting; by supplementing the data of the next project stage to improve the real-time data continuous collection mechanism of the full life cycle data, the forecast results and management strategies are continuously updated, and the continuous forecast and management of carbon emissions of steel engineering are realized, which can timely reflect changes in project operation and improve the real-time and accuracy of forecasting and management strategies.

[0186] The present invention is not limited to the above optional implementations, and anyone can derive other various forms of products under the enlightenment of the present invention. The above specific implementations should not be understood as limiting the scope of protection of the present invention. The scope of protection of the present invention should be based on the definition in the claims, and the description can be used to interpret the claims.

Claims

1. A method for predicting carbon emissions from steel projects based on historical projects, characterized by: The steps include: Step 1: Collect historical full life cycle data of several historical steel engineering projects including at least one project phase, and perform preprocessing to obtain several preprocessed historical full life cycle data; Step 2: Performing data dimensionality reduction on the preprocessed historical full life cycle data to obtain a corresponding principal component set and a number of dimensionality-reduced historical full life cycle data; Step 3: Based on the historical life cycle data after dimension reduction, use artificial intelligence algorithms to build a steel engineering carbon emission prediction model and a carbon emission management strategy generation model; Step 4: Collect the real-time project phase data of the real-time steel engineering project at the current project phase, and collect all the real-time project phase data of other project phases corresponding to the current project phase to form the current real-time full life cycle data; the other project phases corresponding to the current project phase indicate the project phases before the current project phase in the real-time steel engineering project; Step 5: According to the principal component set, the current real-time full life cycle data is subjected to data dimension reduction to obtain the current real-time full life cycle data after dimension reduction; Step 6: Based on the current real-time full life cycle data after dimensionality reduction, use the steel engineering carbon emission prediction model to perform steel engineering carbon emission prediction to obtain the current real-time carbon emission prediction result; Step 7: Based on the current real-time carbon emission prediction result, use the carbon emission management strategy generation model to generate a carbon emission management strategy to obtain the current real-time carbon emission management strategy; Step 8: Continue to collect the real-time project phase data of the real-time steel engineering project at the next project phase, and all the real-time project phase data of other project phases corresponding to the next project phase, to form the next real-time full life cycle data; the other project phases corresponding to the next project phase indicate the project phases before the next project phase in the real-time steel engineering project; Step 9: Use the next real-time full life cycle data as the current real-time full life cycle data and return to the real-time data dimensionality reduction step, that is, return to execute steps 5 to 8 until the next real-time full life cycle data includes real-time project stage data of all project stages.

2. The method for predicting carbon emissions from steel projects based on historical projects according to claim 1, characterized in that: The preprocessing described in step 1 includes data mapping and conversion, data cleaning and normalization.

3. The method for predicting carbon emissions from steel projects based on historical projects according to claim 1, characterized in that: The PCA method is used to perform data dimensionality reduction on some of the pre-processed historical full life cycle data to obtain the corresponding principal component set and some of the reduced-dimensional historical full life cycle data.

4. A method for predicting carbon emissions from steel projects based on historical projects according to any one of claims 1 to 3, characterized in that: The method of constructing a steel engineering carbon emission prediction model and a carbon emission management strategy generation model based on a number of historical life cycle data after dimensionality reduction using an artificial intelligence algorithm includes the following steps: Based on the reduced-dimensional historical full-life cycle data, a fusion algorithm is used to construct the steel engineering carbon emission prediction model and generate several historical carbon emission prediction results; Based on some of the historical carbon emission prediction results, a reinforcement learning algorithm is used to construct the carbon emission management strategy generation model, and generate some historical carbon emission management strategy generation experiences.

5. The method for predicting carbon emissions from steel projects based on historical projects according to claim 4, characterized in that: The method of constructing the steel engineering carbon emission prediction model based on the historical full life cycle data after dimension reduction using a fusion algorithm and generating a number of historical carbon emission prediction results includes the following steps: Using a deep learning algorithm, an initial steel engineering carbon emission prediction model is constructed, and taking the initial network parameters of the initial steel engineering carbon emission prediction model as the optimization target, a swarm intelligence optimization algorithm is used to search and optimize to obtain the optimal initial network parameters of the initial steel engineering carbon emission prediction model; the optimal initial network parameters indicate the initial network parameters obtained when the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements; Labeling the plurality of the historical full life cycle data after dimensionality reduction to obtain a plurality of steel engineering carbon emission prediction samples with real labels, and dividing the plurality of steel engineering carbon emission prediction samples with real labels into a model training set and a model test set according to a preset ratio; Based on the optimal initial network parameters, the initial steel engineering carbon emission prediction model is optimized to obtain an optimized steel engineering carbon emission prediction model, and according to the model training set, the optimized steel engineering carbon emission prediction model is trained and optimized to obtain a trained steel engineering carbon emission prediction model; According to the model test set, the trained steel engineering carbon emission prediction model is tested to obtain a number of historical prediction labels, and the labels are compared and counted with a number of real labels of the model test set to obtain the model test accuracy; If the model test accuracy is greater than the accuracy threshold, the trained steel engineering carbon emission prediction model is output as the optimal steel engineering carbon emission prediction model, so that the optimal steel engineering carbon emission prediction model becomes the constructed steel engineering carbon emission prediction model; otherwise, the training optimization of the steel engineering carbon emission prediction model continues.

6. The method for predicting carbon emissions from steel projects based on historical projects according to claim 5, characterized in that: The deep learning algorithm is LSTM-Attention-cGAN algorithm; The swarm intelligence optimization algorithm is the ICPO algorithm.

7. The method for predicting carbon emissions from steel projects based on historical projects according to claim 4, characterized in that: The carbon emission management strategy generation model is constructed based on the historical carbon emission prediction results using a reinforcement learning algorithm, and a number of historical carbon emission management strategy generation experiences are generated, including the following steps: The carbon emission management strategy generation problem is used as the simulation environment of the reinforcement learning algorithm to construct an intelligent agent and experience replay pool; Defining a state space of the reinforcement learning algorithm according to the carbon emission prediction state contained in the historical carbon emission prediction results; Defining an action space of the reinforcement learning algorithm according to the preset actions included in the carbon emission management strategy, and defining a reward function according to the action space; Constructing a network structure of a neural network including an input layer and an output layer, connecting the input layer to the state space, and connecting the output layer to the action space; Constructing an initial carbon emission management strategy generation model according to the action space, the state space, the reward function, the neural network, the intelligent agent, and the experience replay pool; According to the several historical carbon emission prediction results, the neural network and the intelligent agent of the initial carbon emission management strategy generation model are optimized and trained until the optimization training results meet the preset conditions, and the final carbon emission management strategy generation model is obtained, and the final carbon emission management strategy generation model is used as the constructed carbon emission management strategy generation model, and several historical carbon emission management strategy generation experiences are generated; The plurality of historical carbon emission management strategy generation experiences are stored in the experience playback pool of the carbon emission management strategy generation model.

8. The method for predicting carbon emissions from steel projects based on historical projects according to claim 7, characterized in that: The reinforcement learning algorithm is the DQN algorithm, and the neural network is a deep Q network.

9. The method for predicting carbon emissions from steel projects based on historical projects according to claim 8, characterized in that: The method of generating a carbon emission management strategy based on the current real-time carbon emission prediction result and using the carbon emission management strategy generation model to obtain the current real-time carbon emission management strategy includes the following steps: According to the current real-time carbon emission prediction result, extracting a number of matching historical carbon emission management strategy generation experiences from the experience playback pool of the carbon emission management strategy generation model; According to the current real-time carbon emission prediction result, the state space of the carbon emission management strategy generation model is updated to obtain an updated state space, and according to the matching historical carbon emission management strategy generation experience, the action space of the carbon emission management strategy generation model is updated to obtain an updated action space; Inputting the updated state space into the deep Q network of the carbon emission management strategy generation model, and generating the Q value of each possible action in the updated action space using the deep Q network according to the updated state space; Using the agent, according to the reward function, obtaining a reward value for each possible action in the updated action space, and updating the Q value of the possible action according to the reward value to obtain an updated Q value for each possible action; Repeat the above Q-value update steps until the number of iterations reaches the iteration threshold, and use the greedy strategy to select the possible action with the highest updated Q-value in each iteration as the execution action; Using the intelligent agent, several of the execution actions are integrated to obtain the current real-time carbon emission management strategy.

10. A carbon emission prediction system for steel projects based on historical projects, used to implement the carbon emission prediction method for steel projects as claimed in any one of claims 1 to 9, characterized in that: The system includes a historical project data collection unit, a historical data dimensionality reduction unit, an artificial intelligence model building unit, a real-time data collection unit, a real-time data dimensionality reduction unit, a steel engineering carbon emission prediction unit, a carbon emission management strategy generation unit, a real-time data continuous collection unit and a step return unit, which are connected in sequence.

Images