A regional building carbon emissions optimization system and method

Abstract

The application discloses a regional building carbon emission optimization system and method, aiming at solving the problem of multiple optimization targets of regional building balanced carbon emission, economic cost and user comfort. The system includes a perception layer, a transmission layer, a platform layer and an application layer. The perception layer collects multi-source data such as building energy consumption, environmental state and regional energy. The energy load prediction module of the platform layer predicts future load based on historical data. The multi-objective collaborative optimization module constructs a target function with the minimum regional total carbon emission, the lowest total energy cost and the highest average user comfort as the target. The non-dominated sorting genetic algorithm is used to solve the optimal regional energy scheduling strategy. Finally, the strategy is executed through the regional energy hub module. Through data-driven collaborative scheduling, the application scientifically balances the contradiction between emission reduction, economy and comfort, and significantly improves the comprehensive energy efficiency and low-carbon operation level of the regional building group.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of regional building energy management and carbon emission control, in particular to a regional building carbon emission optimization system and method. BACKGROUND

[0002] Under the driving of global climate change and China's "double carbon" strategic goal, green and low-carbon transformation of various industries in society has become a trend. Carbon emissions in the building sector are one of the main sources of carbon emissions, accounting for more than 30% of global total carbon emissions. In order to achieve the goal of "carbon peak and carbon neutrality", building energy management and carbon emission control has become the core direction of current research.

[0003] At present, building carbon emission optimization is mostly aimed at single buildings, for example, adjusting the air conditioning and lighting systems of single buildings, reducing carbon emissions by reducing energy consumption of single buildings, but there are the following key deficiencies:

[0004] Lack of regional coordination: the energy resources shared among multiple buildings in the region, such as photovoltaic power from photovoltaic power stations, charging and discharging of energy storage batteries, and waste heat from waste heat recovery systems, are not considered, resulting in low utilization rate of regional energy resources;

[0005] Insufficient data integration: the data collection of existing systems is scattered in different devices or systems, and lacks unified preprocessing and integration, making it difficult to support real-time dynamic optimization;

[0006] Single optimization target: existing energy-saving control strategies often take the lowest energy consumption or the lowest operating cost as the single optimization target, making it difficult to effectively quantify and balance the three mutually contradictory and conflicting targets of carbon emissions, economic cost and user comfort.

[0007] Therefore, there is an urgent need for a regional building carbon emission optimization system and method that can integrate regional energy resources, balance multiple targets, and support real-time dynamic optimization to address the deficiencies of existing technology. SUMMARY

[0008] In order to solve the problems existing in the background art, the present application proposes a regional building carbon emission optimization system and method to solve the problems of lack of regional coordination, low resource utilization, and lack of practicality of optimization scheme caused by single building optimization, scattered data, and single target orientation in regional building energy management, and to realize unified scheduling of energy resources among multiple buildings in the region, integration and preprocessing of multi-source data, and multi-target dynamic balance of carbon emissions, energy cost, and user comfort.

[0009] To achieve the above purpose, the present application adopts the following technical solutions:

[0010] The application relates to a regional building carbon emission optimization system applied to a smart city community, comprising a perception layer, a transmission layer, a platform layer and an application layer connected in sequence, wherein:

[0011] The perception layer is used for collecting multi-source data of buildings in a region, and the multi-source data comprises building energy consumption data, environmental state data, regional energy resource data and user behavior data;

[0012] The transmission layer is used for receiving the multi-source data collected by the perception layer, performing edge preprocessing on the multi-source data, and transmitting the preprocessed data to the platform layer through communication protocol adaptation;

[0013] The platform layer is used for storing, processing and making optimization decisions on the data transmitted by the transmission layer, and comprises:

[0014] The data storage module is used for storing the preprocessed data;

[0015] The carbon emission accounting module calculates the carbon emission of a single building and the total carbon emission of a region based on building energy consumption data and carbon emission factors;

[0016] The energy consumption load prediction module predicts the energy consumption load of buildings in a region in a future period based on building energy consumption data and environmental state data;

[0017] The multi-target collaborative optimization module balances carbon emission, energy cost and comfort degree targets based on the energy consumption load prediction result and in combination with user comfort degree constraints, and generates a regional energy scheduling strategy;

[0018] The regional energy hub module is used for executing the regional energy scheduling strategy and scheduling the transmission and distribution of energy among buildings in a region;

[0019] The application layer is used for displaying the results of the platform layer, and comprises a visual interface used for presenting regional carbon emission data, energy consumption load prediction values in a future period, energy cost benefits and user comfort degree states.

[0020] Specifically, in the perception layer, the building energy consumption data comprises electric consumption data, gas consumption data and heat consumption data of each building in a region, and is collected through NB-IoT intelligent electric meters, gas meters and heat meters;

[0021] The environmental state data comprises indoor environmental data of buildings and outdoor meteorological data of a region, and the indoor environmental data comprises temperature data, humidity data, Concentration data of the indoor environment of buildings are collected through LoRa wireless sensors, and the outdoor meteorological data comprises illumination intensity data, wind speed data and outdoor temperature data of the outdoor environment of a region and are obtained through a regional meteorological station;

[0022] Regional energy resource data includes photovoltaic power resource data of photovoltaic power plants in the region, charging and discharging power data of energy storage batteries, and waste heat resource data of waste heat recovery systems, which are collected through Modbus and OPC UA standardized interfaces.

[0023] User behavior data includes user comfort preference data, which is collected through smart terminal devices within the building.

[0024] Specifically, the transmission layer preprocesses the multi-source data collected by the perception layer, including data cleaning and data aggregation. Data cleaning removes outliers from the multi-source data, and data aggregation aggregates minute-level data from the multi-source data into 15-minute or hour-level data. The preprocessed multi-source data is transmitted to the platform layer through an adapted communication protocol, including 5G, LoRa, and NB-IoT protocols.

[0025] Specifically, the carbon emission accounting module of the platform layer is used to calculate the carbon emissions of a single building and the total carbon emissions of the region. The formula for calculating the carbon emissions of a single building is as follows:

[0026] ;

[0027] in, For the first Electricity consumption data for the building As a carbon emission factor for electricity, For the first Gas consumption data for the building Carbon emission factors from fuel gas For the first Heat consumption data of the building Thermal carbon emission factor;

[0028] The formula for calculating total regional carbon emissions is:

[0029] ;

[0030] in This represents the number of buildings within the area.

[0031] Specifically, the energy load forecasting module of the platform layer includes a data processing unit, a model training unit, and a load forecasting unit connected in sequence, wherein:

[0032] The data processing unit is used to standardize historical building energy consumption data and outdoor meteorological data collected by the sensing layer and stored in the data storage module. The processing steps specifically include data alignment and normalization. The data alignment step will standardize the data within the region. Historical building energy consumption data and corresponding outdoor meteorological data for each building are synchronized and aligned with timestamps at fixed intervals to form a dataset linking energy consumption and the environment. The normalization process uses Min-Max normalization to map the aligned dataset to... Intervals, eliminating dimensional differences, are expressed by the following formula:

[0033] ;

[0034] in, The raw data includes building energy consumption data and outdoor meteorological data. , These are the minimum and maximum values ​​of the data sequence, respectively. The data is after normalization;

[0035] The model training unit uses a long short-term memory neural network to train the energy load forecasting model. The input is a historical feature vector, and the output is the predicted energy load value for future periods. For the th... For each building, the historical feature vector at time step t is:

[0036] ;

[0037] in, For the first Building The building energy consumption data vector after the data processing unit has normalized it for the specified time period. For the first Building Electricity consumption data for different time periods For the first Building Gas consumption data for different time periods For the first Building Thermal energy consumption data for different time periods They are respectively The outdoor temperature, light intensity, and wind speed data for the specified time period are normalized by the data processing unit. Training samples are constructed using a sliding window method with a 24-hour window length. Each sample includes a historical feature vector sequence from the past 24 hours. The horizontal prediction is for the next 6 hours, with one prediction point per hour. The loss function is as follows:

[0038]

[0039]

[0040] in, The number of training samples. The first Building Periodic electric, gas, and heat energy consumption load prediction values, For the Building Periodic electric, gas, and heat energy consumption data true values;

[0041] The load prediction unit uses the trained energy consumption load prediction model to predict the energy consumption load of the buildings in the region in the future period:

[0042] ;

[0043] Among them, , For the Building Periodic energy consumption load prediction value vector; finally, each prediction value in the energy consumption load prediction value vector is converted into actual building energy consumption prediction data by inverse normalization.

[0044] Specifically, the multi-objective collaborative optimization module of the platform layer is used to generate a regional energy scheduling strategy, and the workflow includes:

[0045] (1) Receive the necessary input data for multi-objective optimization, including energy consumption load prediction values from the energy consumption load prediction module, and user comfort preference data and regional energy resource data collected by the perception layer;

[0046] (2) Individual encoding is performed on the regional energy scheduling strategy, and the population size is set. A number of individuals are generated by uniform distribution random initialization, and each individual represents a scheduling strategy. The encoding structure is:

[0047] ;

[0048] Among them, is the residual heat transmission proportion matrix, represents the proportion of residual heat transmitted by the factory to the building to the total residual heat of the factory, satisfying , , , is the number of factories in the region, is the number of buildings in the region; is the photovoltaic power distribution proportion matrix, represents the proportion of photovoltaic power distributed by the photovoltaic power station to the building to the total photovoltaic power of the photovoltaic power station, satisfying , , is the number of photovoltaic power stations in the region; is the energy storage battery charging power matrix, representing buildings for the charging power of the energy storage battery , is the discharging power matrix of the energy storage battery, representing the energy storage battery to the building , , is the number of energy storage batteries in the region;

[0049] (3) Construct a multi-objective optimization objective function with the goals of minimizing the total carbon emissions of the future region, minimizing the total energy cost of the future region, and maximizing the average comfort of users:

[0050] ;

[0051] ;

[0052] ;

[0053] ;

[0054] wherein, is the building electricity carbon emission, and the calculation formula is:

[0055]

[0056] is the total photovoltaic power resource of the photovoltaic power station , is the electricity carbon emission factor, K is the number of future predicted time periods, is the time period index, is the time step, representing the length of each predicted time period;

[0057] is the building gas carbon emission, and the calculation formula is:

[0058] ;

[0059] is the total waste heat resource of the factory , is the gas carbon emission factor;

[0060] is the building heating carbon emission, and the calculation formula is:

[0061] ; is the total waste heat resource of the factory

[0062] is the thermal carbon emission factor;

[0063] , , are the electricity cost, gas cost and heat cost of the building, respectively, the electricity cost is calculated by the following formula:

[0064]

[0065] is the real-time electricity price of the power grid;

[0066] is the user comfort score of the building, is calculated by the following formula:

[0067] ;

[0068] wherein, , , are the comfort index weights, , , are the indoor environmental data of temperature, humidity and carbon dioxide concentration, , , are the boundary values allowed by the indoor environment, , , are the comfort preference data set by the user;

[0069] (4) After decoding the scheduling strategy of each individual, the corresponding objective function value is calculated;

[0070] (5) The non-dominated sorting and crowding distance are used to evaluate the fitness of the individual, and according to the objective function value, the individuals in the population are divided into different non-dominated layers, and the crowding distance of each individual in the layer is calculated:

[0071] ;

[0072] wherein, correspond to the objective functions , is the jth objective value of the individual, is the maximum value and the minimum value of the jth objective value in the layer where the individual is located;

[0073] ​​​​​​(6) randomly select two individuals from the population, select the individual with better non-dominated layer or larger crowding distance into the next generation, repeat the step until all individuals in the population are selected to complete;

[0074] Perform single-point crossover on the selected individuals two by two, and the specific logic of single-point crossover is: randomly select a crossover point between the individual codes, and exchange the part of the code after the crossover point of the two individuals;

[0075] Perform Gaussian mutation on the crossed individuals, and the mutation probability is set to 0.1, and the mutation logic is: randomly select a code parameter and add Gaussian noise to the parameter;

[0076] (7) output the optimal solution in the first non-dominated layer when the number of iterations reaches the maximum value, and select the optimal individual according to the user priority to decode the regional energy scheduling strategy.

[0077] A regional building carbon emission optimization method based on the above system, comprising the following steps:

[0078] S1, collect multi-source data of buildings in the region, the multi-source data comprising:

[0079] Building energy consumption data: electric consumption data, gas consumption data and heat consumption data of each building in the region;

[0080] Environmental state data: indoor environmental data of the building and outdoor meteorological data of the region, the indoor environmental data including temperature data, humidity data, Concentration data in the building, and outdoor meteorological data including light intensity data, wind speed data and outdoor temperature data in the region;

[0081] Regional energy resource data: photovoltaic power resource data of photovoltaic power stations in the region, charge and discharge power data of energy storage batteries and waste heat resource data of waste heat recovery systems;

[0082] User behavior data: user comfort preference data;

[0083] S2, edge pre-processing is performed on the multi-source data, and the outliers in the multi-source data are removed, the minute-level data is aggregated into 15-minute-level or hour-level data, and the data transmission amount is reduced;

[0084] S3, store the pre-processed multi-source data;

[0085] S4, based on the building energy consumption data and the carbon emission factor, calculate the carbon emission of a single building and the total carbon emission of the region;

[0086] S5, based on the historical building energy consumption data and outdoor meteorological data, predict the energy load in the future period by using long short-term memory neural network (LSTM);

[0087] S6, based on the energy load prediction result, combining the user comfort constraint, balancing carbon emission, energy cost and comfort target, generating regional energy scheduling strategy through non-dominated sorting genetic algorithm;

[0088] S7, executing the regional energy scheduling strategy, scheduling the transmission and distribution of energy among buildings in the region;

[0089] S8, displaying regional carbon emission data, future period energy load prediction value, energy cost benefit and user comfort state through a visual interface.

[0090] In summary, the beneficial technical effects of the present application are:

[0091] 1. Regional collaborative optimization, improving energy resource utilization: by integrating regional photovoltaic power stations, energy storage batteries, waste heat recovery systems and other energy resources, realizing energy sharing among multiple buildings, reducing energy waste.

[0092] 2. Multi-source data integration, supporting real-time dynamic optimization: through the perception layer, multi-source data such as building energy consumption, environmental state, regional energy resources and user behavior are collected, and through edge preprocessing and unified storage, data support is provided for carbon emission accounting, energy load prediction and optimization, realizing dynamic scheduling strategy update.

[0093] 3. Multi-objective balance, improving the practicability of the scheme: multi-objective collaborative optimization is carried out by using non-dominated sorting genetic algorithm, minimizing the total carbon emission of the region, reducing energy cost and maintaining user comfort, avoiding the problem of user experience decline caused by single target optimization. BRIEF DESCRIPTION OF DRAWINGS

[0094] Figure 1 is a system overview diagram of the present application;

[0095] Figure 2 is a method flowchart of the present application. DETAILED DESCRIPTION

[0096] In order to make the technical means, creative features, purposes and effects realized by the present application more clear and easy to understand, the present application is further described below in combination with the drawings and specific embodiments.

[0097] EMBODIMENT

[0098] As shown in Figure 1 The regional building carbon emission optimization system provided by the present application specifically comprises a perception layer, a transmission layer, a platform layer and an application layer connected in sequence, wherein:

[0099] The perception layer is used for collecting multi-source data of buildings in the region, including building energy consumption data, environmental state data, regional energy resource data and user behavior data;

[0100] The building energy consumption data includes electricity consumption data, gas consumption data and heat consumption data of each building in the region, which are collected through NB-IoT smart electricity meters, gas meters and heat meters;

[0101] The environmental state data includes indoor environmental data of the building and outdoor meteorological data of the region, the indoor environmental data includes temperature data, humidity data, Concentration data are collected through LoRa wireless sensors, and outdoor meteorological data includes light intensity data, wind speed data and outdoor temperature data of the region, which are obtained through regional meteorological stations;

[0102] The regional energy resource data includes photovoltaic power resource data of photovoltaic power stations, charge and discharge power data of energy storage batteries and waste heat resource data of waste heat recovery systems in the region, which are collected through Modbus and OPC UA standardized interfaces;

[0103] The user behavior data includes user comfort preference data, which are collected through intelligent terminal devices in the building.

[0104] The transmission layer is used to receive the multi-source data collected by the perception layer, and to perform edge preprocessing on the multi-source data, including data cleaning and data aggregation. Data cleaning removes outliers in multi-source data, and data aggregation aggregates minute-level data in multi-source data into 15-minute or hour-level data. The preprocessed multi-source data is transmitted to the platform layer through an adaptive communication protocol, including 5G protocol, LoRa protocol and NB-IoT protocol.

[0105] The platform layer is used to store, process and make optimized decisions on the data transmitted by the transmission layer, including:

[0106] The data storage module is used to store the preprocessed data;

[0107] The carbon emission accounting module calculates the carbon emissions of a single building and the total carbon emissions of the region based on building energy consumption data and carbon emission factors;

[0108] The carbon emission calculation formula of a single building is:

[0109] ;

[0110] Wherein, is the electricity consumption data of the th building, is the electricity carbon emission factor, is the gas consumption data of the th building, is the gas carbon emission factor, is the heat consumption data of the th building, a thermal carbon emission factor;

[0111] The total carbon emission calculation formula of the region is:

[0112] ;

[0113] wherein is the number of buildings in the region.

[0114] The energy consumption load prediction module comprises a data processing unit, a model training unit and a load prediction unit connected in sequence, wherein:

[0115] The data processing unit is used for standardizing the historical building energy consumption data and outdoor meteorological data collected by the perception layer and stored in the data storage module, and the processing steps specifically include data alignment and normalization processing. The data alignment step synchronously aligns the historical building energy consumption data of each building in the region and the outdoor meteorological data of the corresponding period according to the time stamp of the fixed interval time point, forming an energy consumption and environment related data set. The normalization processing step uses Min-Max normalization to map the aligned data set to the interval, eliminating the dimensional difference, and the formula is:

[0116] ;

[0117] wherein, is the original data, including building energy consumption data and outdoor meteorological data, , are the minimum value and the maximum value of the data sequence respectively, is the normalized data;

[0118] The model training unit uses a long short-term memory neural network to train an energy consumption load prediction model, the input is a historical feature vector, and the output is a future period energy consumption load prediction value. For the i-th building, the historical feature vector of each time step t is:

[0119] ;

[0120] wherein, is the i-th building period normalized building energy consumption data vector processed by the data processing unit, is the i-th building period electric energy consumption data, is the i-th building period gas energy consumption data, is the i-th building period heating energy consumption data, is the i-th building period cooling energy consumption data, is the i-th building ​​​The thermal energy data of the period, respectively The outdoor temperature data, the light intensity data, and the wind speed data of the period are normalized by the data processing unit. The training sample is constructed by the sliding window method using the energy consumption load prediction model. The window length is 24 hours. Each sample includes a historical feature vector sequence of the past 24 hours. The prediction horizon is 6 hours in the future. There is one prediction point per hour. The loss function is as follows:

[0121]

[0122]

[0123] wherein, is the number of training samples, is the building The electric, gas, and thermal energy consumption load prediction values of the period; is the building The electric, gas, and thermal energy consumption data true values of the period;

[0124] The load prediction unit uses the trained energy consumption load prediction model to predict the energy consumption load of the building in the future period in the region:

[0125] ;

[0126] wherein, , is the building The energy consumption load prediction value vector of the period; finally, each prediction value in the energy consumption load prediction value vector is converted into actual building energy consumption prediction data through reverse normalization.

[0127] The multi-objective collaborative optimization module is used to generate the regional energy scheduling strategy. The working process includes:

[0128] (1) receiving necessary input data for executing multi-objective optimization, including the energy consumption load prediction value from the energy consumption load prediction module, and the user comfort preference data and the regional energy resource data collected by the perception layer;

[0129] (2) individual coding is performed on the regional energy scheduling strategy, and the population size is set. A plurality of individuals are generated by adopting uniform distribution random initialization. Each individual represents a scheduling strategy. The coding structure is as follows:

[0130] ;

[0131] wherein, is the waste heat transmission proportion matrix, representing factories to buildings The proportion of the waste heat transmitted to the total waste heat of the factory meets , , , is the number of factories in the region, is the number of buildings in the region; is a photovoltaic power distribution proportion matrix, representing photovoltaic power stations to buildings The proportion of the photovoltaic power distributed to the total photovoltaic power of the photovoltaic power station meets , , is the number of photovoltaic power stations in the region; is a storage battery charging power matrix, representing buildings The charging power of the storage battery to , is a storage battery discharging power matrix, representing storage batteries The discharging power of the storage battery to buildings , , is the number of storage batteries in the region;

[0132] (3) A multi-objective optimization objective function is constructed with the minimum total carbon emission of the future region, the minimum total energy cost of the future region, and the maximum average comfort of users as targets:

[0133] ;

[0134] ;

[0135] ;

[0136] ;

[0137] wherein, is the carbon emission of the building , and the calculation formula is:

[0138]

[0139] is the total photovoltaic power resource of the photovoltaic power station , is the carbon emission factor of the power, K is the number of time periods predicted in the future, in the embodiment of the present application, K = 6, is a time period index, is the time step, representing the length of each prediction period, which is 1 hour in the embodiment of the present application;

[0140] is the building gas carbon emission, calculated by:

[0141] ;

[0142] is the building total waste heat resource, is the gas carbon emission factor;

[0143] is the building heating carbon emission, calculated by:

[0144] ;

[0145] is the heating carbon emission factor;

[0146] , , are the electricity cost, gas cost and heating cost of the building respectively, the electricity cost is calculated by:

[0147]

[0148] is the real-time electricity price of the power grid;

[0149] is the user comfort score of the building , calculated by:

[0150] ;

[0151] wherein, , , are the comfort index weights, , , are the indoor environmental data of temperature, humidity and carbon dioxide concentration, , , are the boundary values allowed by the indoor environment, , , are the user-set comfort preference data;

[0152] (4) After decoding the scheduling strategy for each individual, calculate the corresponding objective function value;

[0153] (5) Individual fitness is assessed using non-dominated ordination and crowding distance. Based on the objective function value, individuals in the population are divided into different non-dominated layers, and the crowding distance of each individual in its layer is calculated:

[0154] ;

[0155] in, Corresponding to the objective function , For individuals The One target value, For individuals The first in the layer The maximum and minimum values ​​of the target values;

[0156] (6) Randomly select two individuals from the population, and select the individual with better non-dominant layer or greater crowding distance to enter the next generation. Repeat this step until all individuals in the population have been selected.

[0157] The selected individuals are subjected to single-point cross-intersection in pairs. The specific logic of single-point cross-intersection is as follows: randomly select a cross-intersection point between the individual codes and swap the partial codes after the cross-intersection point of the two individuals.

[0158] Gaussian mutation is performed on the crossover individuals with a mutation probability of 0.1. The mutation logic is as follows: randomly select an encoding parameter and add Gaussian noise to the parameter.

[0159] (7) When the number of iterations reaches the maximum value, output the optimal solution in the first non-dominated layer, and select the optimal individual from them according to the user priority to decode the regional energy scheduling strategy.

[0160] The regional energy hub module is used to execute regional energy dispatch strategies and dispatch the transmission and distribution of energy between buildings within the region;

[0161] The application layer is used to display the results of the platform layer, including a visualization interface, which presents regional carbon emission data, energy load forecasts for future periods, energy cost-effectiveness, and user comfort status.

[0162] like Figure 2 As shown, the present invention provides a method for optimizing regional building carbon emissions based on the above system, comprising the following steps:

[0163] S1. Collect multi-source data of buildings within the area, the multi-source data including:

[0164] Building energy consumption data: electricity consumption data, gas consumption data and heat consumption data of each building in the region;

[0165] Environmental state data: indoor environmental data of the building and outdoor meteorological data of the region, the indoor environmental data including temperature data, humidity data, Concentration data, the outdoor meteorological data including light intensity data, wind speed data and outdoor temperature data of the region;

[0166] Regional energy resource data: photovoltaic power resource data of photovoltaic power stations, charging and discharging power data of energy storage batteries and waste heat resource data of waste heat recovery systems in the region;

[0167] User behavior data: comfort preference data of the user;

[0168] S2, edge preprocessing of multi-source data, removing outliers in multi-source data, aggregating minute-level data into 15-minute-level or hour-level data, and reducing data transmission amount;

[0169] S3, storing the preprocessed multi-source data;

[0170] S4, calculating carbon emissions of a single building and total carbon emissions of the region based on building energy consumption data and carbon emission factors;

[0171] S5, predicting energy consumption load in a future period based on historical building energy consumption data and outdoor meteorological data through a long short-term memory neural network (LSTM);

[0172] S6, balancing carbon emissions, energy costs and comfort targets based on energy consumption load prediction results and user comfort constraints, and generating a regional energy scheduling strategy through a non-dominated sorting genetic algorithm;

[0173] S7, executing the regional energy scheduling strategy to schedule the transmission and distribution of energy among buildings in the region;

[0174] S8, displaying regional carbon emission data, future period energy consumption load prediction values, energy cost benefits and user comfort state through a visual interface.

[0175] Therefore, the regional building carbon emission optimization system and method provided by the application integrates regional photovoltaic, energy storage, waste heat and other energy resources, multi-source data preprocessing and dynamic scheduling, and multi-objective collaborative optimization, solves the problems of regional collaboration loss, real-time optimization difficulty and single target scheme practicability shortage caused by traditional single building optimization, realizes efficient use of regional energy resources, real-time dynamic carbon emission management, and balance of carbon emissions, costs and user comfort, improves energy utilization efficiency of regional building clusters and practicability of optimization schemes, and provides an efficient solution for carbon emission control of smart city communities.

[0176] The embodiments of the present application are described in detail above with reference to the accompanying drawings, and the above embodiments are only used to illustrate the technical solutions of the present application but not to limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalent replaced without departing from the purpose and scope of the technical solutions of the present application, and all should be covered in the scope of the claims of the present application.

Claims

1. A regional building carbon emission optimization system, applied to smart city communities, characterized in that, It includes the perception layer, transport layer, platform layer, and application layer connected in sequence, wherein: The perception layer is used to collect multi-source data of buildings in the area, including building energy consumption data, environmental status data, regional energy resource data, and user behavior data. The transport layer is used to receive multi-source data collected by the perception layer, perform edge preprocessing on the multi-source data, and transmit the preprocessed data to the platform layer through communication protocol adaptation. The platform layer is used to store, process, and optimize the data transmitted by the transport layer, including: The data storage module is used to store the preprocessed data; The carbon emission accounting module calculates the carbon emissions of a single building and the total carbon emissions of the region based on building energy consumption data and carbon emission factors. The energy load forecasting module predicts the energy load of buildings in the area for future periods based on building energy consumption data and environmental status data. The multi-objective collaborative optimization module, based on energy load forecasting results and combined with user comfort constraints, balances carbon emissions, energy costs, and comfort objectives to generate a regional energy dispatch strategy. The regional energy hub module is used to execute regional energy dispatch strategies and dispatch the transmission and distribution of energy between buildings within the region; The application layer is used to display the results of the platform layer, including a visualization interface, which presents regional carbon emission data, energy load forecasts for future periods, energy cost-effectiveness, and user comfort status. In the sensing layer, building energy consumption data includes electricity consumption data, gas consumption data, and heat consumption data for each building in the area, which are collected through NB-IoT smart meters, gas meters, and heat meters. Environmental data includes building indoor environmental data and regional outdoor meteorological data. Indoor environmental data includes building indoor temperature and humidity data. Concentration data was collected via LoRa wireless sensors, while outdoor meteorological data, including regional outdoor light intensity, wind speed, and outdoor temperature data, was obtained from regional weather stations. Regional energy resource data includes photovoltaic power resource data of photovoltaic power plants in the region, charging and discharging power data of energy storage batteries, and waste heat resource data of waste heat recovery systems, which are collected through Modbus and OPC UA standardized interfaces. User behavior data includes user comfort preference data, collected through smart terminal devices within the building; The energy load forecasting module of the platform layer includes a data processing unit, a model training unit, and a load forecasting unit connected in sequence, wherein: The data processing unit is used to standardize historical building energy consumption data and outdoor meteorological data collected by the sensing layer and stored in the data storage module. The processing steps specifically include data alignment and normalization. The data alignment step will standardize the data within the region. Historical building energy consumption data and corresponding outdoor meteorological data for each building are synchronized and aligned with timestamps at fixed intervals to form a dataset linking energy consumption and the environment. The normalization process uses Min-Max normalization to map the aligned dataset to... Intervals, eliminating dimensional differences, are expressed by the following formula: ； in, The raw data includes building energy consumption data and outdoor meteorological data. , These are the minimum and maximum values ​​of the data sequence, respectively. The data is after normalization; The model training unit uses a long short-term memory neural network to train the energy load forecasting model. The input is a historical feature vector, and the output is the predicted energy load value for future periods. For the th... For each building, the historical feature vector at time step t is: ； in, For the first Building The building energy consumption data vector after the data processing unit has normalized it for the specified time period. For the first Building Electricity consumption data for different time periods For the first Building Gas consumption data for different time periods For the first Building Thermal energy consumption data for different time periods They are respectively The outdoor temperature, light intensity, and wind speed data for the specified time period are normalized by the data processing unit. Training samples are constructed using a sliding window method with a 24-hour window length. Each sample includes a historical feature vector sequence from the past 24 hours. The horizontal prediction is for the next 6 hours, with one prediction point per hour. The loss function is as follows: in, The number of training samples. The first Building Forecast values ​​of electricity, gas, and heat energy consumption during the specified time period. For the first Building The actual values ​​of electricity, gas, and heat consumption data for each time period; The load forecasting unit uses a trained energy load forecasting model to predict the energy load of buildings in the area for future periods. ； in, , For the first Building The energy load forecast vector for each time period is generated; finally, the forecast values ​​in the energy load forecast vector are converted into actual building energy consumption forecast data through inverse normalization. The multi-objective collaborative optimization module of the platform layer is used to generate regional energy dispatch strategies, and its workflow includes: (1) Receive input data for performing multi-objective optimization. The input data includes the energy load forecast value from the energy load forecasting module, as well as user comfort preference data and regional energy resource data collected by the perception layer. (2) Individual coding is performed on the regional energy dispatch strategy, and the population size is set. Several individuals are generated by uniformly distributed random initialization. Each individual represents a dispatch strategy. The coding structure is as follows: ； in, This is the waste heat transfer ratio matrix. Indicates factory Towards Architecture The proportion of transferred waste heat to the total waste heat of the factory meets the requirements. , , , The number of factories in the region. The number of buildings in the area; A photovoltaic power distribution ratio matrix. Indicates photovoltaic power station Towards Architecture The proportion of allocated photovoltaic power to the total photovoltaic power generation of the photovoltaic power station meets the requirements. , , The number of photovoltaic power stations in the region; Power matrix for charging energy storage batteries, Represents architecture For energy storage batteries The charging power, This is the energy storage battery discharge power matrix. Indicates energy storage battery Towards Architecture The discharge power, , The number of energy storage batteries in the region; (3) To minimize the total carbon emissions, the total energy cost, and the average user comfort in the future region, a multi-objective optimization objective function is constructed: ； ； ； ； in, For architecture Carbon emissions from electricity are calculated using the following formula: For photovoltaic power station Total photovoltaic power resources As a carbon emission factor for electricity, K The number of time periods predicted for the future. For time period index, The time step represents the length of each prediction period; For architecture Carbon emissions from natural gas are calculated using the following formula: ； For the factory Total waste heat resources Carbon emission factors from fuel gas; For architecture The formula for calculating thermal carbon emissions is: ； Thermal carbon emission factor; , , Buildings Electricity costs, gas costs, and heating costs; electricity costs The calculation formula is: Real-time electricity price for the power grid; For architecture User comfort rating The calculation formula is: ； in, , , As a weighting factor for comfort index, , , This includes indoor environmental data such as temperature, humidity, and carbon dioxide concentration. , , These are the allowable boundary values ​​for the indoor environment. , , Comfort preference data set for users; (4) After decoding the scheduling strategy for each individual, calculate the corresponding objective function value; (5) Individual fitness is assessed using non-dominated ordination and crowding distance. Based on the objective function value, individuals in the population are divided into different non-dominated layers, and the crowding distance of each individual in its layer is calculated: ； in, Corresponding to the objective function , For individuals The One target value, For individuals The first in the layer The maximum and minimum values ​​of the target values; (6) Randomly select two individuals from the population, and select the individual with better non-dominant layer or greater crowding distance to enter the next generation. Repeat this step until all individuals in the population have been selected. The selected individuals are subjected to single-point cross-intersection in pairs. The specific logic of single-point cross-intersection is as follows: randomly select a cross-intersection point between the individual codes and swap the partial codes after the cross-intersection point of the two individuals. Gaussian mutation is performed on the crossover individuals with a mutation probability of 0.

1. The mutation logic is as follows: randomly select an encoding parameter and add Gaussian noise to the parameter. (7) When the number of iterations reaches the maximum value, output the optimal solution in the first non-dominated layer, and select the optimal individual from them according to the user priority to decode the regional energy scheduling strategy.

2. The regional building carbon emission optimization system according to claim 1, characterized in that, The transmission layer preprocesses the multi-source data collected by the perception layer, including data cleaning and data aggregation. Data cleaning removes outliers from the multi-source data, and data aggregation aggregates minute-level data from the multi-source data into 15-minute or hour-level data. The preprocessed multi-source data is transmitted to the platform layer through adapted communication protocols, including 5G, LoRa, and NB-IoT protocols.

3. The regional building carbon emission optimization system according to claim 1, characterized in that, The carbon emission accounting module of the platform layer is used to calculate the carbon emissions of a single building and the total carbon emissions of the region. The formula for calculating the carbon emissions of a single building is as follows: ； in, For the first Electricity consumption data for the building As a carbon emission factor for electricity, For the first Gas consumption data for the building Carbon emission factors from fuel gas For the first Heat consumption data of the building Thermal carbon emission factor; The formula for calculating total regional carbon emissions is: ； in This represents the number of buildings within the area.

4. A method for optimizing regional building carbon emissions based on a regional building carbon emission optimization system according to any one of claims 1-3, characterized in that, Includes the following steps: S1. Collect multi-source data of buildings within the area, including: Building energy consumption data: Electricity consumption data, gas consumption data, and heat consumption data for each building in the area; Environmental status data: indoor environmental data of the building and outdoor meteorological data of the area. Indoor environmental data includes indoor temperature data, humidity data, etc. Concentration data, outdoor meteorological data including regional outdoor light intensity data, wind speed data and outdoor temperature data; Regional energy resource data: photovoltaic power resources data of photovoltaic power plants in the region, charging and discharging power data of energy storage batteries, and waste heat resources data of waste heat recovery systems; User behavior data: User comfort preference data; S2. Perform edge preprocessing on multi-source data to remove outliers and aggregate minute-level data into 15-minute or hour-level data to reduce data transmission volume. S3. Store preprocessed multi-source data; S4. Calculate the carbon emissions of a single building and the total carbon emissions of the region based on building energy consumption data and carbon emission factors; S5. Based on historical building energy consumption data and outdoor meteorological data, predict future energy load using a long short-term memory neural network (LSTM). S6. Based on the energy load forecast results and combined with user comfort constraints, balance carbon emissions, energy costs and comfort objectives, and generate a regional energy dispatch strategy through a non-dominated sorting genetic algorithm. S7. Implement regional energy dispatching strategies to dispatch the transmission and distribution of energy between buildings within the region; S8 displays regional carbon emission data, future energy load forecasts, energy cost-effectiveness, and user comfort status through a visual interface.

Images