Energy low-carbon transformation analysis method based on Internet of Things

Abstract

The invention relates to the technical field of low-carbon emission reduction, and discloses an energy low-carbon transformation analysis method based on Internet of Things, which comprises the following steps: S1, collecting data of different energy types and real-time production, transmission, storage and use data through Internet of Things equipment, and constructing a multi-energy data fusion platform; s2, according to the collected energy data, performing comprehensive analysis on different energy types by using a multi-energy collaborative optimization model, and determining the optimal ratio of different energy in conversion efficiency, use time period and economic cost; and S3, collecting user energy consumption behavior data through the Internet of Things equipment, and constructing a user behavior prediction model. Data of multiple energy types are integrated through the Internet of Things technology, intelligent allocation is achieved through a collaborative optimization model, the energy utilization efficiency is improved through multi-energy collaborative management, carbon emission is reduced, the system can flexibly meet the conversion requirements of different energy sources, the resource utilization rate is maximized, and dependence on a single energy source is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of low-carbon emission reduction, and in particular to an energy low-carbon transformation analysis method based on the Internet of Things. Background Art

[0002] As one of the main sources of carbon emissions, achieving efficient and clean energy management has become a research focus. Traditional energy management methods mostly focus on the use and optimization of a single energy source, ignoring the synergistic effects between different types of energy, resulting in low overall energy utilization efficiency. In addition, the impact of user behavior on energy use is ignored, and there is a lack of personalized energy optimization suggestions, which fails to fully mobilize users' energy-saving awareness and enthusiasm.

[0003] Existing low-carbon transformation technologies focus on carbon emission management of power systems. In the comparative document CN117521973A, the low-carbon transformation of cities is mainly evaluated through power system flow analysis and electricity-carbon perception traceability model. However, this method is limited to the optimization and evaluation of power systems, and ignores the coordinated management of multiple energy types. Since the existing technology does not comprehensively consider the coordinated optimization of multiple energy systems, the system is not flexible enough in responding to energy conversion needs, resulting in reduced resource utilization, and over-reliance on a single energy source, which cannot effectively reduce overall carbon emissions;

[0004] In the comparative document CN117521973A, energy management mainly focuses on network load and generator efficiency within the power system, and fails to introduce user behavior data for analysis, and fails to generate personalized energy-saving suggestions. The user's active participation is low, and the user end is not able to fully play its role in promoting low-carbon transformation.

[0005] In the comparative document CN117521973A, the evaluation method is limited to the evaluation of low-carbon transformation results under fixed conditions of the power system, and does not involve the integration and analysis of meteorological data and energy market prices, resulting in poor flexibility of the energy dispatching system and inability to fully utilize low-cost energy resources.

[0006] Therefore, those skilled in the art provide an energy low-carbon transformation analysis method based on the Internet of Things to solve the problems raised in the above background technology. Summary of the invention

[0007] In view of the deficiencies in the prior art, the present invention provides an energy low-carbon transformation analysis method based on the Internet of Things to solve the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: an energy low-carbon transformation analysis method based on the Internet of Things, comprising the following steps: S1, collecting data of different energy types through Internet of Things devices, producing, transmitting, storing and using data in real time, and building a multi-energy data fusion platform;

[0009] S2. Based on the collected energy data, a multi-energy collaborative optimization model is used to conduct a comprehensive analysis of different energy types to determine the optimal ratio of different energy sources in terms of conversion efficiency, usage period and economic cost. The specific optimization is carried out through the following formula:

[0010] Among them, E opt is the total energy consumption after optimization, C i is the carbon emission coefficient of the i-th energy source, E i is the usage of the i-th energy source, η i is the conversion efficiency of the i-th energy source, P i (t) is the market price of the i-th energy at time t, and n is the number of energy types;

[0011] S3. Collect user energy consumption behavior data through IoT devices, build a user behavior prediction model, use time series analysis and clustering algorithms to refine and classify users' electricity consumption patterns, identify users' peak energy consumption habits and equipment usage patterns, and predict future energy demand through user behavior models;

[0012] S4. Based on user behavior data, use machine learning algorithms to generate personalized energy usage optimization suggestions, thereby optimizing energy consumption patterns and reducing carbon emissions;

[0013] S5. Integrate the meteorological data interface to obtain and analyze the impact of climate change on energy demand in real time, and generate future energy demand forecasts in combination with the meteorological forecast model. The relationship between climate data and demand is determined by the following formula: pred =αT(t)+βP(t)+γH(t), where D pred is the predicted value of future energy demand, T(t) is the temperature-time function, P(t) is the air pressure function, H(t) is the humidity function, α, β, γ are the influencing factors;

[0014] S6. Integrate real-time price data of the energy market, build a price fluctuation prediction model for the energy market through historical price data, and optimize energy procurement and scheduling strategies based on price fluctuations;

[0015] S7. Adjust the usage ratio of different energy types and dynamically allocate various energy resources based on the results of multi-energy collaborative optimization analysis, user behavior prediction models, climate change predictions, and market price fluctuation data;

[0016] S8, through edge computing technology, real-time processing of energy data at each IoT node;

[0017] S9. Real-time monitoring of carbon emissions from various energy uses and equipment energy consumption, calculation of overall carbon emissions, and regular generation of carbon emission statistical reports and feedback based on multi-energy use and user behavior patterns;

[0018] S10. Based on real-time monitoring data feedback, use machine learning algorithms to continuously optimize energy scheduling models and user behavior prediction models.

[0019] Preferably, in the step S2, the multi-energy data fusion platform includes a multidimensional data model based on spatiotemporal analysis, which is used to analyze the usage of different energy types in different time periods and spatial scenarios. The multidimensional data model is set, and the energy consumption and carbon emissions in different time and space are expressed by the following formula: Among them, E total (t,s) is the total energy consumption in the time t and space s dimensions; E i (t,s) is the consumption of the i-th energy at time t and space s; W i (s) is the weight factor associated with space s; η i is the conversion efficiency of the i-th energy source; n is the number of energy types;

[0020] The multidimensional data model is used to analyze energy consumption patterns in different time periods and different geographical spaces by adjusting the weights W i (s) and conversion efficiency η i , which can optimize the use of various energy sources in the time and space dimensions, and achieve dynamic balance and optimization of energy consumption and carbon emissions.

[0021] Preferably, in step S3, the user behavior prediction model is constructed by a convolutional neural network or a long short-term memory network, and the convolutional neural network algorithm formula is:

[0022] in, is the output feature value of position (i, j) in layer l; is the weight of the convolution kernel; is the input data of the previous layer; b (l) is the bias term of the lth layer; f(·) is the activation function; M and N are the convolution kernel sizes;

[0023] The long short-term memory network algorithm formula:

[0024] f t =σ(W f ·[h t-1 ,x t ]+b f),i t =σ(W i ·[h t-1 ,x t ]+b i ),

[0025] o t =σ(W o ·[h t-1 ,x t ]+b o ),

[0026] C t =f t ·C t-1 +i t tanh(W C ·[h t-1 ,x t ]+b C ), h t =o t tanh(C t ),

[0027] Among them, f t is the output of the forget gate; i t is the output of the input gate; o t is the output of the output gate; C t is the cell state at the time step; h t is the hidden state of the time step; W f , W i , W o , W C is the weight matrix; b f , b i , b o , b C is the bias term; x t is the input data of the time step; h t-1 is the hidden state at the previous time step.

[0028] Preferably, in the step S4, climate data integration is combined with IoT sensors through a real-time weather API interface, and a machine learning algorithm is used to generate personalized energy usage optimization suggestions;

[0029] When generating personalized energy optimization recommendations, a machine learning algorithm is used to predict user behavior. The support vector machine formula is as follows:

[0030] Under the constraints: i (w T φ(x i )+b)≥1-ξ i ,ξ i ≥0,

[0031] Where w is the weight vector; b is the bias; ξ i is the slack variable; C is the regularization parameter; y i is the classification label of the i-th user; x i is the behavioral feature vector of the i-th user; φ(x i ) is a nonlinear mapping function;

[0032] By training the support vector machine model, we can i Generate classification label y i , and identify the user's energy usage pattern;

[0033] According to the classification result y i and the eigenvector x i , the following optimization suggestions can be formulated:

[0034] Among them, S opt (x i ) indicates that given user x i Personalized energy optimization strategy; E current (x i ) represents user x i Energy consumption; E optimized (x j ) means that after optimization, user x j Optimal energy consumption achieved.

[0035] Preferably, in the step S3, user behavior data is collected through the Internet of Things device, and the user behavior is analyzed and classified using data mining technology. The algorithm formula of the data mining technology is the K-means clustering algorithm, and its formula is as follows:

[0036] Objective function: The goal of K-means clustering is to minimize the sum of the squared distances from points within the cluster to the centroid. The objective function formula is as follows: Where J is the clustering objective function; k is the number of clusters; x j belongs to the i-th cluster C i The j-th user behavior feature vector; μ i is the centroid of the ith cluster, i.e., the center of user behavior; ∥x j -μ i ∥ 2 Represents the user feature vector x j To the center of mass μ i The Euclidean distance of

[0037] Iterative update: After each round of iteration, the position of the centroid is updated according to the assigned clusters. The formula is as follows:

[0038] Among them, μ i is the new centroid of the ith cluster; |C i | is the number of users in the ith cluster; x j is the user behavior feature belonging to the i-th cluster.

[0039] Preferably, in step S5, climate data prediction is usually implemented by multivariate linear regression or machine learning algorithm, and the formula is as follows:

[0040] The multiple linear regression prediction formula: The impact of climate on energy demand is predicted by the following multiple linear regression model: pred =α1T(t)+α2H(t)+α3W(t)+α4P(t)+∈,

[0041] Among them, D pred is the predicted value of future energy demand; T(t) is the temperature at time t; H(t) is the humidity at time t; W(t) is the wind speed at time t; P(t) is the air pressure at time t; α1, α2, α3, α4 are the coefficients of influence of climate characteristics on energy demand; ∈ is the error term;

[0042] The data integration and optimization formula: Combined with the prediction results of multivariate linear regression, the energy dispatch strategy is adjusted in real time through the following optimization formula: E adjust =D pred -E baseline , where E adjust Indicates the energy adjustment amount that needs to be dispatched; D pred is the energy demand predicted based on climate data; E baseline is the baseline energy consumption.

[0043] Preferably, in step S6, the energy market price fluctuation is predicted by a time series prediction model, and energy procurement and scheduling decisions are made in combination with an optimization model, and the formula is as follows:

[0044] Autoregressive integrated moving average model: used to predict future fluctuations in energy market prices. The formula is as follows:

[0045] Among them, P pred (t) is the predicted energy market price at time t; μ is a constant term; φ i is the coefficient of the autoregressive part; P(ti) is the market price at time ti; θ j is the coefficient of the sliding average part; ∈(tj) is the error term at time tj; P is the order of the autoregressive part; q is the order of the sliding average part;

[0046] Energy procurement and scheduling optimization model: Based on the predicted market price fluctuations, optimize the energy procurement and scheduling strategy. The optimization goal is to minimize the energy procurement cost. The formula is as follows:

[0047] Among them, C opt is the minimized cost of energy procurement and dispatch; P pred (t) is the market price predicted at time t; E(t) is the amount of energy purchased or dispatched at time t; T is the optimization period;

[0048] Data integration and optimization formula: Based on market price forecasts, further optimize the dispatch of different types of energy: Among them, E adjust (t) is the energy dispatch amount after adjustment at time t; E i (t) is the usage of the i-th energy at time t; η i is the efficiency of the ith energy source.

[0049] Preferably, in step S7, the multi-energy dynamic scheduling system realizes the dynamic scheduling of multiple energy sources through linear programming or a multi-objective optimization model. The following is an algorithm formula based on the linear programming model:

[0050] Multi-energy scheduling optimization model: The goal is to minimize the comprehensive energy cost and carbon emissions and meet real-time energy demand. The optimization formula is as follows:

[0051] Constraints: 0≤E i (t)≤E max (i)

[0052] Among them, C i is the cost coefficient of the i-th energy; E i (t) is the usage of the i-th energy at time t; λ is the carbon emission weight factor; η i is the carbon emission coefficient of the i-th energy source; n is the number of energy types; D(t) is the total energy demand at time t; E max (i) is the available amount of the i-th energy source;

[0053] Real-time update formula for dynamic scheduling: To achieve real-time scheduling, the system will adjust the energy usage ratio according to the changing demand and supply situation. The optimization formula is as follows:

[0054] Where, ΔE i (t) is the adjustment amount of the i-th energy at time t; α and β are adjustment coefficients; D(t) is the energy demand at time t; P pred(t) is the predicted market price at time t; P(t-1) is the market price at the previous time;

[0055] Energy usage ratio adjustment formula: Based on the optimization scheduling results of each energy source, the usage ratio of different energy sources is updated. The formula is as follows: Among them, r i (t) is the usage ratio of the i-th energy at time t; E i (t) is the actual usage of the ith energy source.

[0056] Preferably, in step S9, the carbon emissions are calculated by the consumption of each energy source and its corresponding carbon emission coefficient, and the formula is as follows:

[0057] Carbon emissions calculation formula: Among them, C total (t) is the total carbon emissions at time t; E i (t) is the usage of the i-th energy at time t; η i is the carbon emission coefficient of the i-th energy source; n is the number of energy types;

[0058] Carbon emission feedback formula: Feedback carbon emissions into the energy use strategy, calculate the carbon emission efficiency of each energy source through the following formula, and perform optimization feedback:

[0059] Among them, Eff C (t) is the carbon emission efficiency at time t; E low-carbon (t) is the amount of low-carbon energy used at time t; C total (t) is the total carbon emissions at time t;

[0060] Carbon emission control feedback formula: Regulate carbon emissions and optimize energy scheduling strategies based on feedback from carbon emission efficiency. The specific optimization formula is as follows:

[0061] Where, ΔE i (t) is the regulation amount of the i-th energy at time t; γ is the regulation coefficient; η i is the carbon emission coefficient of the i-th energy source; C target is the target carbon emission; C total (t) is the actual carbon emissions at time t.

[0062] Preferably, in step S10, continuous learning and optimization are achieved through an online learning algorithm or a reinforcement learning algorithm. The following is an optimization formula based on reinforcement learning:

[0063] The value function formula of reinforcement learning is:

[0064]

[0065] Where V(s) is the value function in state s; R(s,a) is the immediate reward obtained by taking action a in state s; γ is the discount factor; P(s′|s,a) is the probability of transitioning from state s to the next state s′; V(s′) is the value function of the next state s′;

[0066] Strategy update formula: Among them, θ t is the policy parameter at time t; α is the learning rate; is the policy function J(θ t )’s gradient;

[0067] Model optimization and update formula: Where ΔM(t) is the model update amount at time t; η is the model update rate; D(t) is the actual energy demand or user behavior data; D pred (t) is the energy demand or user behavior data predicted by the model.

[0068] The present invention provides an energy low-carbon transformation analysis method based on the Internet of Things. It has the following beneficial effects:

[0069] 1. The present invention integrates data of various energy types through Internet of Things technology, and realizes intelligent allocation through collaborative optimization model. Its multi-energy collaborative management improves energy utilization efficiency and reduces carbon emissions. This innovation enables the system to flexibly respond to the conversion needs between different energy sources, maximize resource utilization, and reduce dependence on a single energy source.

[0070] 2. The present invention collects users' energy usage behavior data through Internet of Things devices and uses machine learning algorithms to generate personalized energy usage optimization suggestions. The system can dynamically adjust the energy usage plan according to the user's historical behavior to achieve personalized energy-saving suggestions and optimization. The process increases the user's enthusiasm for participating in low-carbon transformation and optimizes the user's energy consumption pattern.

[0071] 3. The present invention can predict the impact of climate change on energy demand by integrating meteorological data and energy market price fluctuation information in real time, and adjust energy procurement and scheduling strategies according to market price fluctuations, so that the system has a high degree of flexibility and can quickly respond to changes in the external environment, ensuring the economy of energy use and minimization of carbon emissions. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0073] In order to make the technical personnel in the technical field understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in combination with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a partial embodiment of the present invention, not a complete embodiment. Based on the embodiment of the present invention, other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0074] The present invention is described in detail below in conjunction with the accompanying drawings:

[0075] Example:

[0076] Please refer to the attached Figure 1 ,The embodiment of the present invention provides an energy low-carbon transformation analysis method based on the Internet of Things, including the following steps: S1, collecting data of different energy types through Internet of Things devices, producing, transmitting, storing and using data in real time, and building a multi-energy data fusion platform;

[0077] S2. Based on the collected energy data, a multi-energy collaborative optimization model is used to conduct a comprehensive analysis of different energy types to determine the optimal ratio of different energy sources in terms of conversion efficiency, usage period and economic cost. The specific optimization is carried out through the following formula:

[0078] Among them, E opt is the total energy consumption after optimization, C i is the carbon emission coefficient of the i-th energy source, E i is the usage of the i-th energy source, η i is the conversion efficiency of the i-th energy source, P i (t) is the market price of the i-th energy at time t, and n is the number of energy types;

[0079] S3. Collect user energy consumption behavior data through IoT devices, build a user behavior prediction model, use time series analysis and clustering algorithms to refine and classify users' electricity consumption patterns, identify users' peak energy consumption habits and equipment usage patterns, and predict future energy demand through user behavior models;

[0080] S4. Based on user behavior data, use machine learning algorithms to generate personalized energy usage optimization suggestions, thereby optimizing energy consumption patterns and reducing carbon emissions;

[0081] S5. Integrate the meteorological data interface to obtain and analyze the impact of climate change on energy demand in real time, and generate future energy demand forecasts in combination with the meteorological forecast model. The relationship between climate data and demand is determined by the following formula: pred =αT(t)+βP(t)+γH(t),

[0082] Among them, D pred is the predicted value of future energy demand, T(t) is the temperature-time function, P(t) is the air pressure function, H(t) is the humidity function, α, β, γ are the influencing factors;

[0083] S6. Integrate real-time price data of the energy market, build a price fluctuation prediction model for the energy market through historical price data, and optimize energy procurement and scheduling strategies based on price fluctuations;

[0084] S7. Adjust the usage ratio of different energy types and dynamically allocate various energy resources based on the results of multi-energy collaborative optimization analysis, user behavior prediction models, climate change predictions, and market price fluctuation data;

[0085] S8, through edge computing technology, real-time processing of energy data at each IoT node;

[0086] S9. Real-time monitoring of carbon emissions from various energy uses and equipment energy consumption, calculation of overall carbon emissions, and regular generation of carbon emission statistical reports and feedback based on multi-energy use and user behavior patterns;

[0087] S10. Based on real-time monitoring data feedback, use machine learning algorithms to continuously optimize energy scheduling models and user behavior prediction models.

[0088] The benefit of S1 is that it helps improve the overall energy efficiency of the system, avoids dependence on a single energy source, and achieves flexible energy scheduling; the benefit of S2 is to achieve the lowest carbon emissions and costs, maximize resource utilization, and balance environmental protection and economic needs; the benefit of S3 is to provide valuable user-side data for energy management, so that the system can tailor energy-saving strategies for different types of users and achieve more efficient energy management and distribution; the benefit of S4 is to reduce users' energy consumption, increase users' enthusiasm for participating in low-carbon transformation, ensure accurate energy-saving effects, and improve user experience; the benefit of S5 is to avoid energy supply shortages or waste and achieve more Accurate energy planning and dynamic response; the benefit of S6 is to increase procurement when energy prices are low, or reduce the use of high-energy-consuming equipment when prices are high, thereby improving energy economy; the benefit of S7 is to achieve dynamic allocation of energy resources, optimize energy scheduling, minimize carbon emissions and maximize economic benefits; the benefit of S8 is to improve the response speed of the system, so that energy scheduling and optimization decisions can be executed more promptly and efficiently; the benefit of S9 is to achieve refined carbon emission management and further promote the achievement of emission reduction targets; the benefit of S10 is to achieve long-term and stable low-carbon transformation effects and ensure the efficiency and foresight of energy management.

[0089] In step S2, the multi-energy data fusion platform includes a multidimensional data model based on spatiotemporal analysis, which is used to analyze the usage of different energy types in different time periods and spatial scenarios. The multidimensional data model is set up, and the energy consumption and carbon emissions in different time and space are expressed by the following formula:

[0090]

[0091] Among them, E total (t,s) is the total energy consumption in the time t and space s dimensions; E i (t,s) is the consumption of the i-th energy at time t and space s; W i (s) is the weight factor associated with space s; η i is the conversion efficiency of the i-th energy source; n is the number of energy types;

[0092] The multidimensional data model is used to analyze energy consumption patterns in different time periods and different geographical spaces by adjusting the weights W i (s) and conversion efficiency η i , which can optimize the use of various energy sources in the time and space dimensions, and achieve dynamic balance and optimization of energy consumption and carbon emissions.

[0093] The purpose is to provide the system with a flexible, time-space-based energy consumption and carbon emission optimization analysis framework. By considering multi-dimensional factors such as time, space, energy type and conversion efficiency, the system can dynamically adjust energy allocation and utilization strategies to improve energy utilization efficiency, reduce carbon emissions, and ultimately achieve the goal of low-carbon energy transformation.

[0094] In step S3, the user behavior prediction model is constructed through a convolutional neural network or a long short-term memory network. The convolutional neural network algorithm formula is:

[0095]

[0096] in, is the output feature value of position (i, j) in layer l; is the weight of the convolution kernel; is the input data of the previous layer; b (l) is the bias term of the lth layer; f(·) is the activation function; M and N are the convolution kernel sizes;

[0097] Long short-term memory network algorithm formula:

[0098] f t =σ(W f ·[h t-1 ,x t ]+b f ),i t =σ(Wi ·[h t-1 ,x t ]+b i ),

[0099] o t =σ(W o ·[h t-1 ,x t ]+b o ),

[0100] C t =f t ·C t-1 +i t tanh(W C ·[h t-1 ,x t ]+b C ), h t =o t tanh(C t ),

[0101] Among them, f t is the output of the forget gate; i t is the output of the input gate; o t is the output of the output gate; C t is the cell state at the time step; h t is the hidden state of the time step; W f , W i , W o , W C is the weight matrix; b f 、b i 、b o 、b C is the bias term; x t is the input data of the time step; h t-1 is the hidden state at the previous time step.

[0102] Convolutional neural networks are mainly used to extract local behavior features from user energy consumption data, help identify user energy consumption patterns, and improve the model's ability to capture user behavior;

[0103] The long short-term memory network is mainly used to capture the temporal dependencies of user energy consumption behaviors, ensuring that the model maintains a high prediction accuracy when processing long time series data, especially when processing complex and periodic behaviors.

[0104] In step S4, climate data integration is combined with IoT sensors through real-time weather API interfaces, and machine learning algorithms are used to generate personalized energy usage optimization recommendations;

[0105] When generating personalized energy optimization recommendations, a machine learning algorithm is used to predict user behavior. The support vector machine formula is as follows:

[0106] Under the constraints: i (w T φ(x i )+b)≥1-ξ i ,ξ i ≥0,

[0107] Where w is the weight vector; b is the bias; ξ i is the slack variable; C is the regularization parameter; y i is the classification label of the i-th user; x i is the behavioral feature vector of the i-th user; φ(x i ) is a nonlinear mapping function;

[0108] By training the support vector machine model, we can i Generate classification label y i , and identify the user's energy usage pattern;

[0109] According to the classification result y i and the eigenvector x i , the following optimization suggestions can be formulated:

[0110]

[0111] Among them, S opt (x i ) indicates that given user x i Personalized energy optimization strategy; E current (x i ) represents user x i Energy consumption; E optimized (x j ) means that after optimization, user x j Optimal energy consumption achieved.

[0112] The role of the support vector machine algorithm is to classify users into different groups by analyzing their energy usage behavior characteristics to identify their energy usage patterns;

[0113] The role of the personalized energy optimization recommendation formula is to provide each user with the best energy optimization strategy based on the classification results, thereby minimizing energy consumption and carbon emissions.

[0114] In step S3, user behavior data is collected through IoT devices, and user behavior is analyzed and classified using data mining technology. The algorithm formula of data mining technology is K-means clustering algorithm, and its formula is as follows:

[0115] Objective function: The goal of K-means clustering is to minimize the sum of the squared distances from points within the cluster to the centroid. The objective function formula is as follows:

[0116] Where J is the clustering objective function; k is the number of clusters; x j belongs to the i-th cluster C i The j-th user behavior feature vector; μ i is the centroid of the ith cluster, i.e., the center of user behavior; ∥x j -μ i ∥ 2 Represents the user feature vector x j To the center of mass μ i The Euclidean distance of

[0117] Iterative update: After each round of iteration, the position of the centroid is updated according to the assigned clusters. The formula is as follows:

[0118] Among them, μ i is the new centroid of the ith cluster; |C i | is the number of users in the ith cluster; x j is the user behavior feature belonging to the i-th cluster.

[0119] The purpose of the K-means objective function is to group users with similar energy consumption behaviors into the same category, maximize the similarity of user behaviors within the cluster, minimize the differences between users of different categories, and provide clear user grouping for subsequent energy optimization;

[0120] The centroid update formula is used to dynamically adjust the clustering results so that various types of user behaviors can be accurately represented, ensuring the accuracy and consistency of clustering, and ultimately providing personalized energy optimization recommendations for various user groups.

[0121] In step S5, climate data prediction is usually achieved through multivariate linear regression or machine learning algorithms, and the formula is as follows:

[0122] Multiple linear regression prediction formula: The impact of climate on energy demand is predicted by the following multiple linear regression model: D pred =α1T(t)+α2H(t)+α3W(t)+α4P(t)+∈,

[0123] Among them, D pred is the predicted value of future energy demand; T(t) is the temperature at time t; H(t) is the humidity at time t; W(t) is the wind speed at time t; P(t) is the air pressure at time t; α1, α2, α3, α4 are the coefficients of influence of climate characteristics on energy demand; ∈ is the error term;

[0124] Data integration and optimization formula: Combined with the prediction results of multiple linear regression, the energy scheduling strategy is adjusted in real time through the following optimization formula: E adjust =D pred -E baseline ,

[0125] Among them, E adjust Indicates the energy adjustment amount that needs to be dispatched; D pred is the energy demand predicted based on climate data; E baseline is the baseline energy consumption.

[0126] The role of the multivariate linear regression prediction formula is to combine climate factors to accurately predict future energy demand and provide data support for energy scheduling in advance. By incorporating multiple climate variables into regression analysis, the system can better understand the impact of climate change on energy demand;

[0127] Data integration and optimization formulas are used to compare predicted energy demand with baseline values ​​and adjust energy scheduling in real time. The formulas help the system dynamically and accurately regulate energy supply under different climate change conditions to achieve optimal utilization and scheduling of energy resources.

[0128] In step S6, the time series forecasting model is used to predict energy market price fluctuations, and the optimization model is combined to make energy procurement and scheduling decisions. The formula is as follows:

[0129] Autoregressive integrated moving average model: used to predict future fluctuations in energy market prices. The formula is as follows:

[0130] Among them, P pred (t) is the predicted energy market price at time t; μ is a constant term; φ i is the coefficient of the autoregressive part; P(ti) is the market price at time ti; θ j is the coefficient of the sliding average part; ∈(tj) is the error term at time tj; P is the order of the autoregressive part; q is the order of the sliding average part;

[0131] Energy procurement and scheduling optimization model: Based on the predicted market price fluctuations, optimize the energy procurement and scheduling strategy. The optimization goal is to minimize the energy procurement cost. The formula is as follows:

[0132] Among them, C opt is the minimized cost of energy procurement and dispatch; P pred (t) is the predicted market price at time t; E(t) is the amount of energy purchased or dispatched at time t; T is the optimization period;

[0133] Data integration and optimization formula: Based on market price forecasts, further optimize the dispatch of different types of energy:

[0134] Among them, E adjust (t) is the energy dispatch amount after adjustment at time t; E i (t) is the usage of the i-th energy at time t; η i is the efficiency of the i-th energy source.

[0135] The role of the autoregressive integrated moving average model is to predict future energy market price fluctuations based on historical price data and errors. Through this model, the system can predict price changes in advance and make reasonable purchasing decisions;

[0136] The role of the energy procurement and scheduling optimization model is to determine the optimal procurement and scheduling strategy based on the predicted market price, minimize the energy procurement cost, and ensure the balance between economy and supply;

[0137] The role of data integration and optimization formulas is to optimize the dispatch volume of various energy sources by combining the market prices and conversion efficiencies of different energy sources, ensuring that under different price environments, the system can intelligently dispatch energy to maximize economic benefits and utilization efficiency.

[0138] In step S7, the multi-energy dynamic scheduling system realizes the dynamic scheduling of multiple energy sources through linear programming or multi-objective optimization model. The following is the algorithm formula based on the linear programming model:

[0139] Multi-energy scheduling optimization model: The goal is to minimize the comprehensive energy cost and carbon emissions and meet real-time energy demand. The optimization formula is as follows:

[0140] Constraints: 0≤E i (t)≤E max (i)

[0141] Among them, C i is the cost coefficient of the i-th energy source; E i (t) is the usage of the i-th energy at time t; λ is the carbon emission weight factor; η i is the carbon emission coefficient of the i-th energy source; n is the number of energy types; D(t) is the total energy demand at time t; E max (i) is the available amount of the i-th energy source;

[0142] Real-time update formula for dynamic scheduling: To achieve real-time scheduling, the system will adjust the energy usage ratio according to the changing demand and supply situation. The optimization formula is as follows:

[0143]

[0144] Where, ΔE i (t) is the adjustment amount of the i-th energy at time t; α and β are adjustment coefficients; D(t) is the energy demand at time t; P pred (t) is the predicted market price at time t; P(t-1) is the market price at the previous time;

[0145] Energy usage ratio adjustment formula: Based on the optimization scheduling results of each energy source, the usage ratio of different energy sources is updated. The formula is as follows: Among them, r i (t) is the usage ratio of the i-th energy at time t; E i (t) is the actual usage of the ith energy source.

[0146] The role of the multi-energy scheduling optimization model is to optimize energy use and find a balance between cost and environmental protection by minimizing comprehensive energy costs and carbon emissions while meeting energy demand;

[0147] The role of the real-time update formula of dynamic scheduling is to adjust the usage of different energy sources in real time in response to dynamic changes in market prices and energy demand, so as to achieve efficient energy scheduling and ensure the system's agile response to demand changes and market fluctuations;

[0148] The function of the energy usage ratio adjustment formula is to update the usage ratio of each energy in the system according to the optimization results of different energy sources, ensuring that the system can always optimally allocate energy resources.

[0149] In step S9, carbon emissions are calculated by the consumption of each energy source and its corresponding carbon emission coefficient. The formula is as follows:

[0150] Carbon emissions calculation formula:

[0151] Among them, C total (t) is the total carbon emissions at time t; E i (t) is the usage of the i-th energy at time t; η i is the carbon emission coefficient of the i-th energy source; n is the number of energy types;

[0152] Carbon emission feedback formula: Feedback carbon emissions into the energy use strategy, calculate the carbon emission efficiency of each energy source through the following formula, and perform optimization feedback:

[0153] Among them, Eff C (t) is the carbon emission efficiency at time t; E low-carbon (t) is the amount of low-carbon energy used at time t; C total(t) is the total carbon emissions at time t;

[0154] Carbon emission control feedback formula: Regulate carbon emissions and optimize energy scheduling strategies based on feedback from carbon emission efficiency. The specific optimization formula is as follows:

[0155] Where, ΔE i (t) is the regulation amount of the i-th energy at time t; γ is the regulation coefficient; η i is the carbon emission coefficient of the i-th energy source; C target is the target carbon emission; C total (t) is the actual carbon emissions at time t.

[0156] The function of the carbon emission calculation formula is to calculate the total carbon emissions at time t through the usage of each energy source and the carbon emission coefficient, providing data support for carbon emission monitoring;

[0157] The function of the carbon emission feedback formula is to calculate the carbon emission efficiency based on the ratio of low-carbon energy usage to total carbon emissions, and feed it back to the system to help the system optimize the low-carbon energy usage strategy;

[0158] The function of the carbon emission control feedback formula is to dynamically adjust the use of different energy sources based on the difference between current carbon emissions and target carbon emissions to ensure that carbon emissions are controlled within the target range.

[0159] In step S10, continuous learning and optimization are achieved through online learning algorithms or reinforcement learning algorithms. The following is the optimization formula based on reinforcement learning:

[0160] The value function formula of reinforcement learning is:

[0161]

[0162] Where V(s) is the value function in state s; R(s,a) is the immediate reward obtained by taking action a in state s; γ is the discount factor; P(s′|s,a) is the probability of transitioning from state s to the next state s′; V(s′) is the value function of the next state s′;

[0163] Strategy update formula: Among them, θ t is the policy parameter at time t; α is the learning rate; is the policy function J(θ t )’s gradient;

[0164] Model optimization and update formula: Where ΔM(t) is the model update amount at time t; η is the model update rate; D(t) is the actual energy demand or user behavior data; Dpred (t) is the energy demand or user behavior data predicted by the model.

[0165] The role of the value function formula of reinforcement learning is to evaluate the rewards that the system can obtain from actions taken in different states, and to guide the system to determine the optimal energy scheduling strategy by maximizing future expected benefits;

[0166] The role of the strategy update formula is to gradually optimize the strategy parameters through the gradient ascent method, so that the system can continuously adjust and optimize the energy scheduling strategy according to feedback, ensuring that the system can respond flexibly to environmental changes;

[0167] The role of the model optimization and update formula is to dynamically update and optimize the system's prediction model through the difference between actual demand and predicted demand, thereby improving the accuracy of energy scheduling and the flexibility of model response.

[0168] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An energy low-carbon transformation analysis method based on the Internet of Things, characterized in that: The steps include: S1, collecting data of different energy types through IoT devices, producing, transmitting, storing and using data in real time, and building a multi-energy data fusion platform; S2. Based on the collected energy data, a multi-energy collaborative optimization model is used to conduct a comprehensive analysis of different energy types to determine the optimal ratio of different energy sources in terms of conversion efficiency, usage period and economic cost. The specific optimization is carried out through the following formula: Among them, E opt is the total energy consumption after optimization, C i is the carbon emission coefficient of the i-th energy source, E i is the usage of the i-th energy, η i is the conversion efficiency of the i-th energy source, P i (t) is the market price of the i-th energy at time t, and n is the number of energy types; S3. Collect user energy consumption behavior data through IoT devices, build a user behavior prediction model, use time series analysis and clustering algorithms to refine and classify users' electricity consumption patterns, identify users' peak energy consumption habits and equipment usage patterns, and predict future energy demand through user behavior models; S4. Based on user behavior data, use machine learning algorithms to generate personalized energy usage optimization suggestions, thereby optimizing energy consumption patterns and reducing carbon emissions; S5. Integrate the meteorological data interface to obtain and analyze the impact of climate change on energy demand in real time, and generate future energy demand forecasts in combination with the meteorological forecast model. The relationship between climate data and demand is determined by the following formula: pred =αT(t)+βP(t)+γH(t), Among them, D pred is the predicted value of future energy demand, T(t) is the temperature-time function, P(t) is the air pressure function, H(t) is the humidity function, α, β, γ are the influencing factors; S6. Integrate real-time price data of the energy market, build a price fluctuation prediction model for the energy market through historical price data, and optimize energy procurement and scheduling strategies based on price fluctuations; S7. Adjust the usage ratio of different energy types and dynamically allocate various energy resources based on the results of multi-energy collaborative optimization analysis, user behavior prediction models, climate change predictions, and market price fluctuation data; S8, through edge computing technology, real-time processing of energy data at each IoT node; S9. Real-time monitoring of carbon emissions from various energy uses and equipment energy consumption, calculation of overall carbon emissions, and regular generation of carbon emission statistical reports and feedback based on multi-energy use and user behavior patterns; S10. Based on real-time monitoring data feedback, use machine learning algorithms to continuously optimize energy scheduling models and user behavior prediction models.

2. According to the method for analyzing energy low-carbon transformation based on the Internet of Things according to claim 1, it is characterized in that: In the step S2, the multi-energy data fusion platform includes a multidimensional data model based on spatiotemporal analysis, which is used to analyze the usage of different energy types in different time periods and spatial scenarios. The multidimensional data model is set, and the energy consumption and carbon emissions in different time and space are expressed by the following formula: Among them, E total (t,s) is the total energy consumption in the time t and space s dimensions; E i (t,s) is the consumption of the i-th energy at time t and space s; W i (s) is the weight factor associated with space s; η i is the conversion efficiency of the i-th energy source; n is the number of energy types; The multidimensional data model is used to analyze energy consumption patterns in different time periods and different geographical spaces by adjusting the weights W i (s) and conversion efficiency η i , which can optimize the use of various energy sources in the time and space dimensions, and achieve dynamic balance and optimization of energy consumption and carbon emissions.

3. According to the method for analyzing low-carbon energy transformation based on the Internet of Things in claim 1, it is characterized in that: In the step S3, the user behavior prediction model is constructed by a convolutional neural network or a long short-term memory network. The convolutional neural network algorithm formula is: in, is the output feature value of position (i, j) in layer l; is the weight of the convolution kernel; is the input data of the previous layer; b (l) is the bias term of the lth layer; f(·) is the activation function; M and N are the convolution kernel sizes; The long short-term memory network algorithm formula: f t =σ(W f ·[h t-1 ,x t ]+b f ),i t =σ(W i ·[h t-1 ,x t ]+b i ), the t =σ(W o ·[h t-1 ,x t ]+b o ), C t =f t ·C t-1 +i t ·tanh(W C ·[h t-1 ,x t ]+b C ),h t =o t ·tanh(C t ), Among them, f t is the output of the forget gate; i t is the output of the input gate; o t is the output of the output gate; C t is the cell state at the time step; h t is the hidden state of the time step; W f , W i , W o , W C is the weight matrix; b f 、b i 、b o 、b C is the bias term; x t is the input data of the time step; h t-1 is the hidden state at the previous time step.

4. According to the method for analyzing low-carbon energy transformation based on the Internet of Things in claim 1, it is characterized in that: In the step S4, climate data integration is combined with IoT sensors through a real-time weather API interface, and a machine learning algorithm is used to generate personalized energy usage optimization recommendations; When generating personalized energy optimization recommendations, a machine learning algorithm is used to predict user behavior. The support vector machine formula is as follows: Under the constraints: Where w is the weight vector; b is the bias; ξ i is the slack variable; C is the regularization parameter; y i is the classification label of the i-th user; x i is the behavioral feature vector of the i-th user; φ(x i ) is a nonlinear mapping function; By training the support vector machine model, we can i Generate classification label y i , and identify the user's energy usage pattern; According to the classification result y i and the eigenvector x i , the following optimization suggestions can be formulated: Among them, S opt (x i ) indicates that given user x i Personalized energy optimization strategy; E current (x i ) represents user x i Energy consumption; E optimized (x j ) means that after optimization, user x j Optimal energy consumption achieved.

5. According to the method for analyzing low-carbon energy transformation based on the Internet of Things in claim 1, it is characterized in that: In the step S3, user behavior data is collected through IoT devices, and user behavior is analyzed and classified using data mining technology. The algorithm formula of the data mining technology is the K-means clustering algorithm, and its formula is as follows: Objective function: The goal of K-means clustering is to minimize the sum of the squared distances from points within the cluster to the centroid. The objective function formula is as follows: Where J is the clustering objective function; k is the number of clusters; x j belongs to the i-th cluster C i The j-th user behavior feature vector; μ i is the centroid of the ith cluster, i.e., the center of user behavior; ∥x j -μ i ∥ 2 Represents the user feature vector x j To the center of mass μ i The Euclidean distance of Iterative update: After each round of iteration, the position of the centroid is updated according to the assigned clusters. The formula is as follows: Among them, μ i is the new centroid of the ith cluster; |C i | is the number of users in the ith cluster; x j is the user behavior feature belonging to the i-th cluster.

6. According to the method for analyzing low-carbon energy transformation based on the Internet of Things in claim 1, it is characterized in that: In step S5, climate data prediction is usually achieved through multivariate linear regression or machine learning algorithms, and the formula is as follows: The multiple linear regression prediction formula: The impact of climate on energy demand is predicted by the following multiple linear regression model: pred =α1T(t)+α2H(t)+α3W(t)+α4P(t)+∈, Among them, D pred is the predicted value of future energy demand; T(t) is the temperature at time t; H(t) is the humidity at time t; W(t) is the wind speed at time t; P(t) is the air pressure at time t; α1, α2, α3, α4 are the coefficients of influence of climate characteristics on energy demand; ∈ is the error term; The data integration and optimization formula: Combined with the prediction results of multivariate linear regression, the energy dispatch strategy is adjusted in real time through the following optimization formula: E adjust =D pred -E baseline , Among them, E adjust Indicates the energy adjustment amount that needs to be dispatched; D pred is the energy demand predicted based on climate data; E baseline is the baseline energy consumption.

7. According to the method for analyzing low-carbon energy transformation based on the Internet of Things in claim 1, it is characterized in that: In step S6, the energy market price fluctuations are predicted by the time series prediction model, and the energy procurement and scheduling decisions are made in combination with the optimization model. The formula is as follows: Autoregressive integrated moving average model: used to predict future fluctuations in energy market prices. The formula is as follows: Among them, P pred (t) is the predicted energy market price at time t; μ is a constant term; φ i is the coefficient of the autoregressive part; P(ti) is the market price at time ti; θ j is the coefficient of the sliding average part; ∈(tj) is the error term at time tj; P is the order of the autoregressive part; q is the order of the sliding average part; Energy procurement and scheduling optimization model: Based on the predicted market price fluctuations, optimize the energy procurement and scheduling strategy. The optimization goal is to minimize the energy procurement cost. The formula is as follows: Among them, C opt is the minimized cost of energy procurement and dispatch; P pred (t) is the market price predicted at time t; E(t) is the amount of energy purchased or dispatched at time t; T is the optimization period; Data integration and optimization formula: Based on market price forecasts, further optimize the dispatch of different types of energy: Among them, E adjust (t) is the energy dispatch amount after adjustment at time t; E i (t) is the usage of the i-th energy at time t; η i is the efficiency of the i-th energy source.

8. According to the method for analyzing low-carbon energy transformation based on the Internet of Things in claim 1, it is characterized in that: In step S7, the multi-energy dynamic scheduling system implements the dynamic scheduling of multiple energy sources through linear programming or a multi-objective optimization model. The following is an algorithm formula based on the linear programming model: Multi-energy scheduling optimization model: The goal is to minimize the comprehensive energy cost and carbon emissions and meet real-time energy demand. The optimization formula is as follows: Constraints: 0≤E i (t)≤E max (i), where C i is the cost coefficient of the i-th energy source; E i (t) is the usage of the i-th energy at time t; λ is the carbon emission weight factor; η i is the carbon emission coefficient of the i-th energy source; n is the number of energy types; D(t) is the total energy demand at time t; E max (i) is the available amount of the i-th energy source; Real-time update formula for dynamic scheduling: To achieve real-time scheduling, the system will adjust the energy usage ratio according to the changing demand and supply situation. The optimization formula is as follows: Where, ΔE i (t) is the adjustment amount of the i-th energy at time t; α and β are adjustment coefficients; D(t) is the energy demand at time t; P pred (t) is the predicted market price at time t; P(t-1) is the market price at the previous time; Energy usage ratio adjustment formula: Based on the optimization scheduling results of each energy source, the usage ratio of different energy sources is updated. The formula is as follows: Among them, r i (t) is the usage ratio of the i-th energy at time t; E i (t) is the actual usage of the ith energy source.

9. The energy low-carbon transformation analysis method based on the Internet of Things according to claim 1 is characterized in that: In step S9, the carbon emissions are calculated by the consumption of each energy source and its corresponding carbon emission coefficient, and the formula is as follows: Carbon emissions calculation formula: Among them, C total (t) is the total carbon emissions at time t; E i (t) is the usage of the i-th energy at time t; η i is the carbon emission coefficient of the i-th energy source; n is the number of energy types; Carbon emission feedback formula: Feedback carbon emissions into the energy use strategy, calculate the carbon emission efficiency of each energy source through the following formula, and perform optimization feedback: Among them, Eff C (t) is the carbon emission efficiency at time t; E low-carbon (t) is the amount of low-carbon energy used at time t; C total (t) is the total carbon emissions at time t; Carbon emission control feedback formula: Regulate carbon emissions and optimize energy scheduling strategies based on feedback from carbon emission efficiency. The specific optimization formula is as follows: Where, ΔE i (t) is the regulation amount of the i-th energy at time t; γ is the regulation coefficient; η i is the carbon emission coefficient of the i-th energy source; C target is the target carbon emission; C total (t) is the actual carbon emissions at time t.

10. The method for analyzing energy low-carbon transformation based on the Internet of Things according to claim 1 is characterized in that: In the step S10, continuous learning and optimization are achieved through an online learning algorithm or a reinforcement learning algorithm. The following is an optimization formula based on reinforcement learning: The value function formula of reinforcement learning is: Where V(s) is the value function in state s; R(s,a) is the immediate reward obtained by taking action a in state s; γ is the discount factor; P(s′|s,a) is the probability of transitioning from state s to the next state s′; V(s′) is the value function of the next state s′; Strategy update formula: Among them, θ t is the policy parameter at time t; α is the learning rate; is the policy function J(θ t )’s gradient; Model optimization and update formula: Where ΔM(t) is the model update amount at time t; η is the model update rate; D(t) is the actual energy demand or user behavior data; D pred (t) is the energy demand or user behavior data predicted by the model.

Images