Carbon emission dynamic measuring and calculating method based on public building operation and related device

Abstract

The invention discloses a carbon emission dynamic measurement and calculation method based on public building operation and a related device, and relates to the technical field of power system optimization operation and dispatch. After operation data of a public building are obtained, the operation data are input into a carbon emission dynamic measurement and calculation model based on a time sequence; and obtaining a carbon emission prediction value of each operation stage of the public building. The model is constructed hierarchically according to operation stages of the public building, submodules are divided for each hierarchy unit, a carbon emission prediction value of the corresponding operation stage of the public building is calculated by adopting a carbon emission calculation unit in each submodule, and the whole-flow carbon emission prediction of the public building operation is completed. The carbon emission calculation unit adopts an improved particle swarm optimization algorithm to explore key parameters, namely weights and offsets, in the deep neural network in the calculation process, the optimal weights and offsets obtained through exploration are used for optimizing the deep neural network, the optimized deep neural network is adopted for carbon emission prediction, and the accuracy of carbon emission prediction is improved.

Description

Technical Field

[0001] The present application relates to the technical field of optimized operation and dispatching of power systems, and in particular to a method and related devices for dynamically calculating carbon emissions based on public building operations. Background Art

[0002] With the increasing concern about global climate change, all industries are facing the pressure of energy conservation and emission reduction. Accordingly, the carbon emissions of public buildings, as intensive places of energy consumption and material circulation, including office buildings, commercial buildings, tourism buildings, science and education buildings, communication buildings and transportation buildings, have attracted much attention.

[0003] However, the existing technology uses traditional meter readings to calculate carbon emissions from public buildings. This method can only make a general estimate of carbon emissions from public buildings, but cannot distinguish the carbon emissions contributions of different operating links in detail. Therefore, the existing technology has the problems of lag and low accuracy in calculating carbon emissions from public buildings. Summary of the invention

[0004] In view of the above problems, this application provides a method and related device for dynamic measurement of carbon emissions based on public building operations to achieve the purpose of accurately calculating carbon emissions of public buildings. The specific scheme is as follows:

[0005] The first aspect of the present application provides a method for dynamically calculating carbon emissions based on public building operations, comprising:

[0006] Obtain operational data of public buildings collected by smart terminals;

[0007] Input the operation data into the dynamic carbon emission calculation model based on time series to obtain the carbon emission forecast value of each operation stage of public buildings;

[0008] The operating framework of the carbon emission dynamic calculation model includes at least four operation level modules, and the at least four operation level modules correspond to each operation stage of the public building respectively. Each operation level module includes at least one sub-module, and each sub-module includes a carbon emission calculation unit; the carbon emission calculation unit is constructed according to the energy consumption type, material flow process and carbon emission mechanism of the corresponding operation level, and is used for: while processing operation data based on a deep neural network to predict carbon emissions, a particle swarm optimization algorithm is used to explore with the weights and biases in the deep neural network as particles until the exploration stop condition is met, and the deep neural network is optimized with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value of the corresponding operation stage of the public building.

[0009] In a possible implementation, the exploration stop condition is that the change in the fitness value of the particle's current position compared to the fitness value of the previous position is not greater than a preset threshold, or the number of iterations of the particle reaches a maximum number of iterations;

[0010] The particle swarm optimization algorithm is used to explore the weights and biases in the deep neural network as particles until the exploration stop conditions are met, including:

[0011] The starting position of the generated particles;

[0012] Calculate the fitness value of the particle at the starting position, use the preset weight calculation formula to calculate the inertia weight of the particle at the starting position, and determine the next position of the particle based on the inertia weight; use the next position as the starting position of the particle, and the inertia weight is determined by the current number of iterations and the maximum number of iterations of the particle;

[0013] Iteratively calculate the fitness value of the particle at the starting position; use a preset weight calculation formula to calculate the inertia weight of the particle at the starting position; determine the next position of the particle based on the inertia weight, and use the next position as the starting position of the particle until the fitness value of the particle at the starting position changes by no more than a preset threshold compared to the fitness value of the particle at the previous position; or the number of iterations of the particle reaches the maximum number of iterations.

[0014] In a possible implementation, the method further includes:

[0015] A particle swarm optimization algorithm is used to explore the hyperparameters and / or network structure characteristics of the deep neural network as particles until the exploration stopping condition is met. The deep neural network is optimized with the optimal particles that meet the exploration stopping condition to obtain an optimized deep neural network. Carbon emissions are predicted using the optimized deep neural network to obtain carbon emission prediction values.

[0016] In one possible implementation, carbon emission prediction based on deep neural network processing of operational data includes:

[0017] While using deep neural networks to process operational data, carbon emissions predictions are made by combining energy carbon emission factors and special carbon emission factors that match public buildings; special carbon emission factors include at least the carbon emission factors corresponding to refrigerant leakage of cold chain equipment in public buildings, the use and disposal of disposable plastic products, and wastewater treatment of fresh food processing.

[0018] In a possible implementation, the method further includes:

[0019] The energy carbon emission factor is updated according to a first preset period, and the special carbon emission factor is updated according to a second preset period; the time interval between each period in the first preset period is shorter than the time interval between each period in the second preset period.

[0020] In a possible implementation, the method further includes:

[0021] The carbon emission forecast value is adjusted and outputted using the market dynamic information corresponding to public buildings.

[0022] The second aspect of the present application provides a carbon emission dynamic calculation device based on public building operation, comprising:

[0023] An acquisition unit, used to acquire the operation data of public buildings collected by the intelligent terminal;

[0024] The carbon emission dynamic calculation unit is used to input the operation data into the carbon emission dynamic calculation model based on time series to obtain the carbon emission prediction value of each operation stage of the public building;

[0025] The operating framework of the carbon emission dynamic calculation model includes at least four operation level modules, and the at least four operation level modules correspond to each operation stage of the public building respectively. Each operation level module includes at least one sub-module, and each sub-module includes a carbon emission calculation unit; the carbon emission calculation unit is constructed according to the energy consumption type, material flow process and carbon emission mechanism of the corresponding operation level, and is used for: while processing operation data based on a deep neural network to predict carbon emissions, a particle swarm optimization algorithm is used to explore with the weights and biases in the deep neural network as particles until the exploration stop condition is met, and the deep neural network is optimized with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value of the corresponding operation stage of the public building.

[0026] The third aspect of the present application provides a carbon emission dynamic calculation device based on public building operation, comprising at least one processor and a memory connected to the processor, wherein:

[0027] The memory is used to store computer programs;

[0028] The processor is used to execute a computer program so that the device for dynamic measurement of carbon emissions based on the operation of public buildings can implement any of the methods for dynamic measurement of carbon emissions based on the operation of public buildings described above.

[0029] The fourth aspect of the present application provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the methods for dynamically calculating carbon emissions based on public building operations as described above.

[0030] The fifth aspect of the present application provides a computer storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the dynamic carbon emission measurement methods based on public building operations as described above.

[0031] By means of the above technical scheme, the present application provides a method and related devices for dynamic carbon emission calculation based on public building operation, after obtaining the operation data of the public building collected by the intelligent terminal, the operation data is input into the carbon emission dynamic calculation model based on time series to obtain the carbon emission prediction value of each operation stage of the public building. The carbon emission dynamic calculation model is constructed hierarchically according to the operation stage of the public building, and then each hierarchical unit is divided into submodules. Finally, the carbon emission calculation unit in each submodule is used to calculate the carbon emission prediction value of the corresponding operation stage of the public building, and the carbon emission prediction of the whole process of public building operation is completed. The carbon emission calculation unit uses the improved particle swarm optimization algorithm to explore the key parameters in the deep neural network: weights and biases during the calculation process, and optimizes the deep neural network with the optimal weights and biases explored, and uses the optimized deep neural network to predict carbon emissions, thereby improving the accuracy of carbon emission prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0033] Figure 1 A flow chart of a dynamic carbon emission calculation method based on public building operation provided for this application;

[0034] Figure 2 This is an example diagram of the hierarchical division of the commercial supermarket operation stage in the carbon emission dynamic calculation model provided in this application;

[0035] Figure 3 A flow chart of a dynamic carbon emission calculation method based on public building operation provided for this application;

[0036] Figure 4 A schematic diagram of the structure of a carbon emission dynamic calculation device based on public building operation provided for this application;

[0037] Figure 5 Schematic diagram of the structure of the carbon emission dynamic measurement equipment based on public building operation provided for this application. DETAILED DESCRIPTION

[0038] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0039] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0040] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0041] Public buildings include office buildings, such as office buildings, office buildings, etc.; commercial buildings, such as commercial supermarkets, financial buildings, etc.; tourist buildings, such as hotels, entertainment venues, etc.; science, education, culture, health buildings, such as culture, education, scientific research, medical care, health, sports buildings, etc.; communication buildings, such as post and telecommunications, communications, broadcasting rooms, etc.; transportation rooms, such as airports and station buildings, etc.

[0042] The public buildings mentioned above are places with intensive energy consumption and material circulation, and their carbon emissions have attracted much attention.

[0043] However, the existing technology generally uses electricity meter readings to count carbon emissions from public buildings. This method can only make a general estimate of the carbon emissions of public buildings and has very low accuracy.

[0044] Next, taking a commercial supermarket as an example, we will introduce the problems existing in the existing technology of using electricity meter readings to calculate carbon emissions:

[0045] Insufficient data accuracy: Traditional electricity meters can only count electricity consumption, while carbon emissions need to be calculated in combination with carbon emission factors. In addition, the source of electricity will change over time, and a fixed electricity-carbon factor will lead to accounting errors. For example, when the proportion of thermal power is high during a certain period of time, the actual carbon emissions may be underestimated.

[0046] Limited coverage: Commercial supermarkets’ carbon emissions include not only electricity consumption, but also important sources such as refrigerant leakage and supply chain emissions. However, electricity meter readings can only reflect electricity-related emissions and cannot cover other important sources. For example, carbon emissions caused by refrigerant leakage in the refrigeration system of large supermarkets cannot be monitored through electricity meters.

[0047] Timeliness lag: Traditional accounting relies on fixed electricity carbon factors that are updated annually / quarterly and cannot reflect short-term changes in the electricity structure.

[0048] High management complexity: Carbon emission accounting requires a unified standard system and multi-dimensional data integration. Relying solely on meter readings requires the establishment of an additional complex conversion model and is easily affected by human factors.

[0049] In order to solve the above problems, this application provides a dynamic carbon emission measurement method and related devices based on public building operations, and also provides integrated carbon emission detection equipment and supporting systems specially customized for public buildings, so as to meet the needs of public building managers to grasp carbon emission data in real time, formulate precise emission reduction strategies and demonstrate environmental protection results to the public.

[0050] Optional, see Figure 1 , a flow chart of a method for dynamic calculation of carbon emissions based on public building operations provided in this application.

[0051] like Figure 1 As shown in the figure, the dynamic calculation method of carbon emissions based on public building operations includes the following steps:

[0052] Step 101: Acquire the operation data of public buildings collected by the smart terminal.

[0053] It should be noted that smart terminals are mainly used to collect operational data of public buildings. Specifically, smart terminals integrate various types of sensors and devices to collect energy and environmental operation parameters of public buildings in an all-round way, so that data can be synchronized in real time and provide first-hand materials for back-end analysis. Sensors include but are not limited to multi-function meters and temperature and humidity sensors.

[0054] The existence of smart terminals can ensure the all-round collection of operational data of public buildings, ensuring that there are no blind spots in the data, and providing a solid data foundation for carbon emission accounting and control.

[0055] Optionally, the backend obtains operational data of public buildings from various types of sensors and devices integrated into smart terminals.

[0056] It should be noted that for the various types of sensors and devices integrated into smart terminals, the devices are mainly used for energy consumption monitoring, such as multi-function electricity meters, gas meters, water meters, etc. These devices not only have traditional metering functions, but also have built-in high-precision sensors that can collect parameters such as voltage, current, power factor, gas flow pressure, water flow temperature, etc. in real time. Specifically, the data can be encrypted and transmitted to the cloud data center through a wireless transmission module. The device can also have a local data storage function, which can cache at least 7 days of data when the network is interrupted to ensure data integrity.

[0057] Various types of sensors are mainly used to sense environmental and operational parameters. For example, temperature and humidity sensors, light sensors, human infrared sensors, passenger flow counters, logistics vehicle positioning trackers and other equipment are deployed in key areas inside and outside public buildings. Temperature and humidity sensors provide a basis for optimizing air-conditioning systems; light sensors cooperate with lighting control systems to achieve automatic dimming and energy saving; human infrared sensors work with passenger flow counters to automatically shut down equipment in unmanned areas to reduce energy consumption; logistics vehicle positioning trackers facilitate the optimization of delivery routes and reduce transportation carbon emissions. All these sensor data and energy data are synchronized and converged in real time.

[0058] Step 102: Input the operation data into a carbon emission dynamic calculation model based on time series to obtain the carbon emission forecast value of each operation stage of the public building.

[0059] First, the carbon emission dynamic estimation model based on time series provided in this application is introduced.

[0060] Based on time series, it mainly refers to the introduction of the time dimension in the carbon emission calculation model provided in this application. Specifically, a dynamic calculation model of carbon emissions for public buildings can be established for the whole year, quarter, month, day or even hour. At the same time, the model also takes into account the impact of different seasonal climates on air conditioning energy consumption in public buildings. Taking commercial supermarkets as an example, the energy consumption of commercial supermarkets for cooling in summer and heating in winter, the impact of holiday promotions on customer flow and logistics demand, and the impact of business hours adjustments on lighting and equipment market changes are taken into account as time dimensions. At the same time, the historical operating data of commercial supermarkets (such as operating data from the past 3 to 5 years) is used for machine learning training to obtain a dynamic calculation model for carbon emissions that can predict different periods in the future.

[0061] Next, taking a commercial supermarket as an example, the overall framework of the carbon emission dynamic measurement model is introduced.

[0062] The operating framework of the carbon emission dynamic measurement model includes at least four operation level modules, which correspond to the various operation stages of commercial supermarkets.

[0063] Optionally, the four operation level modules may be a building energy consumption operation level module, a logistics distribution operation level module, a commodity sales and service operation level module, and a waste treatment operation level module divided according to the operation stages of the commercial supermarket.

[0064] The reason for the division into these four operational level modules is that commercial supermarkets contain a variety of complex activities, and these four operational level modules can basically cover all aspects of commercial supermarkets. Building energy consumption involves electricity consumption for lighting, air conditioning and other equipment, which is an important part of energy consumption; logistics and distribution cover transportation, warehousing and handling, which affects the carbon emissions of material circulation; commodity sales and services consider commodity packaging, in-store processing, etc., which involve carbon emissions in the sales process; waste disposal involves carbon emissions from solid waste treatment and recyclable transportation. Dividing the operational stages of commercial supermarkets into the four operational levels mentioned above can improve the carbon emission accounting of the operational links and ensure that there are no omissions in the accounting of carbon emissions.

[0065] Specifically, each operation level module includes at least one sub-module.

[0066] For example, the building energy consumption operation level module is subdivided into lighting, air conditioning, elevators, refrigeration and freezing equipment and other sub-modules; the logistics distribution operation level module is subdivided into self-owned fleet transportation, third-party logistics entrustment, purchase warehousing and handling and other sub-modules; the commodity sales and service operation level module is subdivided into commodity packaging, in-store processing, electronic equipment use and other sub-modules; the waste treatment operation level module is subdivided into solid waste landfill incineration, recyclables recycling and transportation and other sub-modules.

[0067] Each submodule includes a carbon emission calculation unit, which is constructed based on the energy consumption type, material flow process and carbon emission mechanism of the corresponding operation level. The carbon emission calculation unit is mainly used to calculate the carbon emission forecast value of the corresponding operation stage of the commercial supermarket.

[0068] There are many benefits to building a dynamic carbon emission calculation model in a hierarchical manner. Take the dynamic carbon emission calculation model used in commercial supermarkets as an example to introduce the benefits.

[0069] First: This model can realize refined control over the entire process of commercial supermarket operations.

[0070] Specifically, commercial supermarkets involve multiple complex links. By constructing a hierarchical model, the operation links of commercial supermarkets are divided into four major operation level modules, and the four major operation level modules are further subdivided into their corresponding sub-modules, so that each operation link of the commercial supermarket can be deeply analyzed and each carbon emission point can be accurately located. Each sub-module independently calculates carbon emissions, which effectively avoids the general estimation using electricity meters in the existing technology, allowing commercial supermarket managers to know the contribution of each carbon emission source, thereby realizing refined control of the entire process of commercial supermarket operations.

[0071] Second: The hierarchical construction of a dynamic carbon emission calculation model can improve the accuracy of carbon emission calculations. Each carbon emission calculation unit is constructed based on the energy consumption type, material flow process and carbon emission mechanism of its corresponding operating level. When calculating carbon emissions, real-time detection or statistical operation data, such as equipment operating time, power consumption, etc., are input, and then combined with refined carbon emission factors to achieve accurate calculation of carbon emissions in each submodule and improve the accuracy of carbon emission calculations.

[0072] Third: The hierarchical construction of a dynamic carbon emission measurement model facilitates supermarket managers to make targeted management decisions. Supermarket managers can formulate more targeted energy-saving and emission reduction strategies based on the carbon emissions of different operating levels and sub-modules. For example, if it is found that the air-conditioning sub-module in the building energy consumption operation level module has high carbon emissions, supermarket managers can carry out energy-saving transformation or optimize operation management for air-conditioning equipment; if the carbon emissions of the logistics distribution module are high, supermarket managers can optimize logistics distribution routes and upgrade transportation equipment to effectively reduce overall carbon emissions and improve operational management efficiency.

[0073] For example, see Figure 2 , an example diagram of the hierarchical division of the commercial supermarket operation stage in the carbon emission dynamic calculation model provided in this application.

[0074] like Figure 2 As shown in the figure, the operation of commercial supermarkets is divided into four operation level modules: building energy consumption, logistics distribution, commodity sales and services, and waste disposal. Then, each operation level is divided into sub-modules: building energy consumption is divided into sub-modules such as air conditioning and lighting; logistics distribution is divided into sub-modules such as self-owned fleet transportation and third-party logistics entrustment; commodity sales and services are divided into sub-modules such as commodity packaging and electronic equipment use; waste disposal is divided into sub-modules such as recyclable recycling and transportation and solid waste landfill incineration.

[0075] For each carbon emission calculation unit used to calculate the carbon emission forecast value of the corresponding operation stage of the commercial supermarket, the operation process includes:

[0076] While processing operational data based on deep neural networks to predict carbon emissions, a particle swarm optimization algorithm is used to explore with the weights and biases in the deep neural network as particles until the exploration stopping conditions are met. The deep neural network is optimized with the optimal particles that meet the exploration stopping conditions to obtain the optimized deep neural network. Carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value of the commercial supermarket in the corresponding operation stage.

[0077] It should be noted that the prediction of carbon emissions is mainly completed by deep neural network models and improved particle swarm optimization algorithms.

[0078] Among them, the improved particle swarm optimization algorithm mainly fine-tunes the parameters used by the deep neural network while the deep neural network processes operational data. After obtaining the optimized deep neural network, more accurate carbon emission prediction values ​​are further obtained.

[0079] In this application, an improved particle swarm optimization algorithm is mainly used to explore the optimal combination of key parameters in the deep neural network: weights and biases, until the exploration stopping conditions are met, and the weights and biases that meet the stopping conditions are used as the optimal combination to optimize the deep neural network to obtain the optimized deep neural network, and the optimized deep neural network is further used to predict carbon emissions to obtain the carbon emission prediction value of the commercial supermarket in the corresponding operation stage.

[0080] Next, the improved particle swarm optimization algorithm is introduced.

[0081] Particle Swarm Optimization (PSO) is one of the modern innovative heuristic algorithms, which has the characteristics of wide application and easy implementation. The PSO algorithm carefully studies the collective behavior of animals and looks to the future as a reliable method to solve optimization problems in a wide range of applications. In PSO, each possible solution is represented by a particle, and the algorithm as a whole is represented by the entire swarm. Due to the collaboration and information sharing of the swarm, the particles are able to learn and develop to the highest level of efficiency.

[0082] The improved particle swarm optimization algorithm has three main improvements.

[0083] Improvement point 1: Dynamically adjust the inertia weight w.

[0084] The inertia weight in the particle swarm optimization algorithm represents the "strength" or "flexibility" of adjusting the pace of particles when they move. Its function is to balance the global exploration of particles in the search space (extensively trying different directions) and local optimization (fine-tuning to the best position).

[0085] This application provides an innovative way to calculate the inertia weight: the inertia weight is determined by using the current number of iterations and the maximum number of iterations of the particle, that is, the inertia weight changes according to the current number of iterations and the maximum number of iterations. The calculation formula of the inertia weight can be as follows:

[0086] ;

[0087] Among them, t is the current iteration number, t max is the maximum number of iterations, and w is the inertia weight.

[0088] It should be noted that in the process of exploring the best position of particles, dynamically adjusting the inertia weight of particles can enable the improved particle swarm optimization algorithm to search more widely in the early stage to avoid falling into the local optimal solution, and to search more finely in the later stage, accelerate the convergence speed, and improve the optimization efficiency of the improved particle swarm optimization algorithm.

[0089] Improvement point 2: Improve the calculation of fitness value.

[0090] The fitness value in the particle swarm optimization algorithm represents the quality of the particle's current position. For this scheme, the particle's fitness value is a score of the particle's current position's optimization effect on the deep neural network. When the fitness value is high, it means that the current particle has a relatively good optimization effect on the deep neural network. When the fitness value is low, it means that the current particle has a relatively poor optimization effect on the deep neural network.

[0091] Taking the jth particle as an example, the calculation formula of the fitness value provided in this application is:

[0092] ;

[0093] Among them, Fitness j is the fitness value of the jth particle, RMLSE j Indicates a certain error indicator, which needs to be determined based on the scenario.

[0094] It should be noted that the method provided by the present application is used to measure the quality of particles and guide the particles to search in the direction of a better solution, so that the improved particle swarm optimization algorithm can better find the optimal solution that meets the goal, that is, the optimal particle that meets the exploration stop condition.

[0095] Improvement point three: Combined with deep neural networks, the key parameters involved in the process of deep neural networks processing operational data for carbon emission prediction are optimized: weights and biases. The weights and biases are regarded as the positions of particles in the search space. The improved particle swarm optimization algorithm is used to explore the optimal combination of weights and biases to minimize the loss function and improve the performance of deep neural networks.

[0096] After introducing the improved particle swarm optimization algorithm, the application process of the improved particle swarm optimization algorithm in this application is introduced below:

[0097] Step 1: Generate the starting position of the particles. The particles here are the key parameters in the deep neural network: weights and biases.

[0098] Specifically, starting particles located at different starting positions are randomly generated in the population.

[0099] Step 2: Calculate the fitness value of the particle at the starting position, use a preset weight calculation formula to calculate the inertia weight of the particle at the starting position, determine the next position of the particle based on inertia, and use the next position as the starting position of the particle.

[0100] Specifically, the fitness value of each particle in the population can be calculated by using the fitness value calculation formula mentioned above.

[0101] The calculation formula of fitness value can be as follows:

[0102] ;

[0103] Then, the inertia weight of each particle is calculated using the inertia weight calculation formula mentioned above.

[0104] The calculation formula for the inertia weight of each particle can be as follows:

[0105] ;

[0106] Next, calculate the velocity of each particle.

[0107] First, suppose the particle swarm is n-dimensional, and each particle in the particle swarm is represented by express.

[0108] The particle velocity can be calculated as follows:

[0109] ;

[0110] in, is the speed of the current iteration, is the speed of the previous iteration, is the inertia weight, and is a random variable uniformly distributed in the range of [0,1], and the acceleration coefficient that has a greater impact on the effectiveness of the PSO algorithm is defined as and . is the best position of the particle before the previous iteration, is the best position of the entire particle swarm before the same iteration. As a form of memory, it stores the best prior position that the particle has reached as .the term It means that particles act according to the information they have learned from the group, guided by the optimal position of the group.

[0111] Finally, the next position of each particle is determined based on the particle's velocity and inertia weight. The calculation formula for the next position can be as follows:

[0112] ;

[0113] in, is the current location, For the previous position, To determine the speed of each particle in the current iteration.

[0114] Use the next position of each particle as the starting position of that particle.

[0115] Step 3: Iterate the above steps and step 3 until the exploration stop condition is reached.

[0116] It should be noted that the exploration stop condition is: the change in the fitness value of the particle's current position compared to the fitness value of the previous position is higher than a preset threshold, or the number of iterations of the particle reaches the maximum number of iterations.

[0117] Specifically, the first iteration stopping condition is that the change in the fitness value of the particle at the starting position compared to the fitness value of the particle at the previous position is not higher than a preset threshold.

[0118] Based on the fitness value, the fitness value of each particle is calculated. During the algorithm iteration process, the fitness values ​​of the particles are constantly compared. When the fitness value of a particle no longer increases in multiple iterations, or the increase is very small and tends to be stable, it indicates that the particle may have found the optimal solution in the current search space. At this time, the weight and bias combination represented by the particle may be the optimal combination, and the corresponding fitness value can be used as an important basis for judging whether the optimal value has been found.

[0119] The second iteration stopping condition is: the number of iterations of the particle reaches the maximum number of iterations.

[0120] It can be the stop condition set by the improved particle swarm optimization algorithm itself. When the algorithm execution process meets the stop condition, the relative optimal value is found by default. The stop condition can be related to the number of iterations. For example, when the algorithm reaches the preset maximum number of iterations, the algorithm stops. Within the preset number of iterations, the particles continuously update their positions and try to find a better solution. When the maximum number of iterations is reached, the solution corresponding to the particle position (i.e., the value of weight and bias) can be regarded as the optimal value under the current conditions.

[0121] In summary, in deep neural networks, weights and biases are the key factors that determine model performance. The improved particle swarm optimization algorithm regards weights and biases as the positions of particles in the search space. Each particle represents a set of possible combinations of weights and biases. Particles move in the search space and continuously update their positions (i.e., the values ​​of weights and biases) to find the optimal combination, thereby minimizing the loss function.

[0122] For example, see Figure 3 , a flow chart of a method for dynamic calculation of carbon emissions based on public building operations provided in this application.

[0123] like Figure 3 As shown, the carbon emission data is first input into the deep neural network for processing to determine whether the current processing process of the deep neural network meets the stopping criteria. When the stopping criteria are not met, the improved particle swarm optimization algorithm is used to update the carbon emission data that needs to be input into the deep neural network until the current processing process of the deep neural network meets the stopping criteria and then the prediction result is output.

[0124] After introducing the improved particle swarm optimization algorithm and the combination of the improved particle swarm optimization algorithm and the deep neural network, the deep neural network and the process of processing operational data using the deep neural network optimized by the improved particle swarm optimization algorithm are introduced next.

[0125] A deep neural network is a powerful nonlinear modeling tool. It is an artificial neural network with several layers between the input layer and the output layer. A deep neural network has two hidden layers, and the neurons in each layer are connected to the neurons in the previous layer. Before generating an output, each neuron processes the information from the neurons in its previous layer using weights and activation functions. The following equation describes the output of neuron j.

[0126] ;

[0127] in, is the output of the neuron, is the ith input, is the weight of i input and neuron, n is the total number of inputs, is the threshold of the neuron, and f is the activation function.

[0128] Optionally, in the dynamic carbon emission measurement model, while using deep neural networks to process operational data, carbon emissions predictions are made in combination with energy carbon emission factors and special carbon emission factors that match public buildings.

[0129] Taking commercial supermarkets as an example, special carbon emission factors include at least the carbon emission factors corresponding to refrigerant leakage of cold chain equipment in commercial supermarkets, the use and disposal of disposable plastic products, and wastewater treatment of fresh food processing.

[0130] It should be noted that the energy carbon emission factor includes electricity, natural gas, diesel, etc. based on energy types.

[0131] For electricity, the basic emission factor is 0.86 kg CO per unit 2 e, this value is obtained by referring to the national average thermal power emission factor, and its calibration basis and data source are:

[0132] Combined with the power generation structure of the power grid in the area where the commercial supermarket is located, the proportion of thermal power [X]%, the proportion of hydropower [Y]%, and the proportion of wind power [Z]%, etc., obtain real-time power quality parameters (such as sulfur content, power generation efficiency, etc.) through the data interface with the power supply bureau, and calculate and calibrate based on the weighted carbon emission coefficient of power generation from different energy sources. If the proportion of local hydropower clean power generation is high, the emission factor will be reduced accordingly.

[0133] For natural gas, the base emission factor is 2.16 kg CO per unit. 2 e, this value is the default value of the IPCC guidelines, and its calibration basis and data source are:

[0134] Analyze the composition of local natural gas sources, compare the difference with the standard composition, refer to the gas quality report provided by the gas company (such as methane content, impurity ratio, etc.), and correct the emission factor. If the methane content is higher than the standard, the emission factor may be slightly adjusted and reduced.

[0135] For diesel, the basic emission factor is 3.16 kg CO per unit 2 e, this value is obtained based on the domestic general oil emission factor, and its calibration basis and data source are:

[0136] Taking into account the changes in oil quality standards, based on the diesel batch test report provided by the gas station (such as sulfur content, cetane number and other indicators), combined with the actual combustion efficiency of diesel in commercial supermarket logistics vehicles and standby power generation equipment, the carbon emission adjustment is calculated through the combustion chemical reaction formula. If the sulfur content is low and the combustion efficiency is high, the emission factor will be appropriately reduced.

[0137] Special carbon emission factors include emission factors of refrigerant leakage of cold chain equipment in commercial supermarkets, emission factors of use of disposable plastic products, and emission factors of wastewater treatment of discarded and fresh food processing.

[0138] For the refrigerant leakage emission factor of commercial supermarket cold chain equipment, taking R134a as an example, the determination method is:

[0139] In the operating environment of cold chain equipment in commercial supermarkets, different temperatures (-20℃ to 10℃), pressures (2 - 8bar) and equipment aging stages (new equipment, 3-year-old equipment, 5-year-old equipment, etc.) were simulated to measure the refrigerant leakage rate. The emission factor was calculated based on the global warming potential (GWP = 1430) of R134a and the leakage amount.

[0140] The emission factor is 1.43 kg CO2 per 1 g R134a leak. 2 e. The actual leakage rate is determined by experimental measurement.

[0141] For disposable plastic products, emission factors are used. Taking ordinary plastic bags as an example, the determination method is:

[0142] Study the entire life cycle of plastic bags from production (raw material acquisition, energy consumption during processing) to disposal (greenhouse gas emissions from landfill or incineration). Consider the recycling stage. If the local recycling rate is [M]%, the emission reduction benefit after recycling is [X] kg CO per kilogram of plastic bag. 2 e. The emission factor is calculated comprehensively, assuming that the plastic bag weighs 50g / bag.

[0143] The specific value of carbon emissions generated by each plastic bag in its entire life cycle is 0.12 kg CO 2 e.

[0144] For fresh food processing wastewater treatment emission factors, the determination method is:

[0145] Analyze the fresh food processing wastewater treatment process (such as anaerobic fermentation, aerobic treatment, etc.), monitor the amount of methane emission and other greenhouse gases (such as nitrous oxide) produced during the treatment process. Calculate the emission factor based on the gas emissions and the corresponding global warming potential (methane GWP = 28, nitrous oxide GWP = 265), with the calculation unit per ton of fresh food processing wastewater.

[0146] The carbon emissions from the treatment of each ton of fresh food processing wastewater can be specifically 5.2 kg CO 2 e.

[0147] For the energy carbon emission factors and special carbon emission factors mentioned above, these carbon emission factors will be finely customized before application in this application. The fine customization mainly includes multi-source data fusion calibration and special accounting of special emission sources.

[0148] For multi-source data fusion calibration, we first collect the national authoritative energy carbon emission database, international greenhouse gas accounting standards (such as IPCC guidelines), and domestic regional energy structure reports to establish a basic carbon emission factor library. For energy and materials commonly used in commercial supermarkets, such as electricity, natural gas, diesel, and refrigerants, we determine the preliminary carbon emission factors.

[0149] Then, combined with the power generation energy structure of the power grid in the area where the commercial supermarket is located (such as the proportion of thermal power, hydropower, and wind power), the differences in the composition of local natural gas sources, and changes in oil quality standards, real-time monitoring data (such as real-time energy quality parameters obtained through the data interface with the power supply bureau and gas company) are used to dynamically calibrate the basic factors to ensure that they reflect the actual local carbon emission levels.

[0150] For special accounting of special emission sources, special experiments are carried out specifically for special emission sources such as refrigerant leakage of cold chain equipment in commercial supermarkets, use and disposal of disposable plastic products, and wastewater treatment in fresh food processing. The leakage rates of different refrigerants under different temperature, pressure, and equipment aging conditions are measured, and the carbon emission factor is accurately calculated based on its global warming potential (GWP); the carbon emissions of disposable plastic products from production to disposal are studied, and the emission reduction effect of recycling is considered to comprehensively determine its emission factor; greenhouse gas emissions such as methane fugitives in the treatment of fresh food processing wastewater are analyzed, and a corresponding emission factor model is established.

[0151] It should also be noted that energy carbon emission factors and special carbon emission factors will be updated regularly.

[0152] Optionally, the energy carbon emission factor is updated according to a first preset period, and the special carbon emission factor is updated according to a second preset period; the time interval between each period in the first preset period is shorter than the time interval between each period in the second preset period, that is, the energy carbon emission factor is updated more frequently than the special carbon emission factor.

[0153] Specifically, the energy carbon emission factor is reviewed and updated every quarter based on the power supply bureau and gas company's new quarter energy quality data and regional energy structure change report to ensure that it is closely aligned with the actual energy supply carbon emissions situation.

[0154] Special carbon emission factors are re-evaluated every year. For refrigerant leakage factors, they are adjusted according to the equipment aging status and operating parameter changes in the annual maintenance report of cold chain equipment; disposable plastic product factors are updated with market raw materials, production process innovations, local recycling policies, and utilization rate changes; fresh wastewater treatment factors are revised according to wastewater treatment process optimization and upgrades and annual water quality component monitoring results to ensure the timeliness and accuracy of emission factors and provide reliable data support for the precise calculation of carbon emissions in commercial supermarkets.

[0155] In addition, in the dynamic calculation model of carbon emissions, the optimized deep neural network uses the market dynamic information corresponding to public buildings to adjust the carbon emission prediction value and then output it.

[0156] It should be noted that when the public building is a commercial supermarket, market dynamic information includes but is not limited to weather forecasts, advance reporting of promotional plans, etc. Adjusting the carbon emission forecast results in combination with market dynamic information can provide immediate carbon emission reference for the commercial supermarket's operational decisions.

[0157] In summary, the method for dynamic carbon emission calculation based on public building operation provided by the present application obtains the operation data of the public building collected by the intelligent terminal, and then inputs the operation data into the carbon emission dynamic calculation model based on time series to obtain the carbon emission prediction value of each operation stage of the public building. The carbon emission dynamic calculation model is constructed hierarchically according to the operation stage of the public building, and then each hierarchical unit is divided into submodules. Finally, the carbon emission calculation unit in each submodule is used to calculate the carbon emission prediction value of the corresponding operation stage of the public building, which can complete the carbon emission prediction of the whole process of public building operation. Furthermore, the carbon emission calculation unit uses the improved particle swarm optimization algorithm to explore the key parameters in the deep neural network: weights and biases during the calculation process, and optimizes the deep neural network with the optimal weights and biases explored, and uses the optimized deep neural network to predict carbon emissions, thereby improving the accuracy of carbon emission prediction.

[0158] Optionally, for the above-mentioned dynamic carbon emission calculation method based on public building operation, the method further includes:

[0159] A particle swarm optimization algorithm is used to explore the hyperparameters and / or network structure characteristics of the deep neural network as particles until the exploration stopping condition is met. The deep neural network is optimized with the optimal particles that meet the exploration stopping condition to obtain an optimized deep neural network. Carbon emissions are predicted using the optimized deep neural network to obtain carbon emission prediction values.

[0160] Specifically, the hyperparameters are performance parameters of the deep neural network. Exemplarily, the hyperparameters may be: learning rate, number of hidden layers, number of neurons in each layer, etc. The improved particle swarm optimization algorithm can explore the weights and biases of the deep neural network while considering the hyperparameters as the position vectors of the particles. By evaluating the performance of the deep neural network under different hyperparameter combinations (such as the accuracy of the verification set), the search direction of the particle swarm is guided, thereby obtaining the hyperparameter combination that can optimize the performance of the deep neural network, and using the hyperparameter combination to optimize the performance of the deep neural network.

[0161] The network structure characteristics are the characteristics of the network structure that affect the deep neural network. The improved particle swarm optimization algorithm can, on the basis of exploring the weights and biases of the deep neural network, encode certain characteristics of the network structure, such as whether it contains a specific layer, the connection method between layers, and other network structure characteristics as particle position vectors, and update the position of the particle according to the performance of the deep neural network under the combination of network structure characteristics until the optimal combination of network structure characteristics is explored, and use the optimal combination of network structure characteristics to optimize the structure of the deep neural network.

[0162] It is understandable that the improved particle swarm optimization algorithm can also be used in the present application to simultaneously explore the optimal combination of weights and biases of the deep neural network, the optimal combination of hyperparameters, and the optimal combination of network structure features. In addition, the exploration stop condition is the same as the exploration stop condition mentioned above.

[0163] Specifically, when the improved particle swarm optimization algorithm is used to explore the optimal combination of weights and biases of a deep neural network and the optimal combination of hyperparameters, the exploration needs to be stopped when the exploration of weights and biases as particles and the exploration of hyperparameters as particles meet the exploration stop condition. Correspondingly, when the improved particle swarm optimization algorithm is used to explore the optimal combination of weights and biases of a deep neural network and the optimal combination of network structure features, the exploration needs to be stopped when the exploration of weights and biases as particles and the exploration of network structure feature combinations as particles meet the exploration stop condition.

[0164] When the improved particle swarm optimization algorithm is used simultaneously to explore the optimal combination of weights and biases of a deep neural network, the optimal combination of hyperparameters, and the optimal combination of network structure features, the exploration can be stopped when the exploration using weights and biases as particles, the exploration using hyperparameters as particles, and the exploration using network structure feature combinations as particles all meet the exploration stopping conditions.

[0165] In summary, the dynamic calculation method of carbon emissions based on public building operations provided in this application uses a modified particle swarm optimization algorithm to explore relevant parameters that can optimize deep neural networks, thereby greatly improving the accuracy of carbon emission predictions of deep neural networks.

[0166] The above introduces a method for dynamically calculating carbon emissions based on public building operations provided by an embodiment of the present application. The following will introduce a device for executing the above-mentioned method for dynamically calculating carbon emissions based on public building operations.

[0167] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of a carbon emission dynamic calculation device based on public building operation provided in this application. As shown in Figure 4, the device includes: an acquisition unit 10 and a carbon emission dynamic calculation unit 20. Among them:

[0168] An acquisition unit 10 is used to acquire the operation data of the public building collected by the intelligent terminal;

[0169] The carbon emission dynamic calculation unit 20 is used to input the operation data into the carbon emission dynamic calculation model based on the time series to obtain the carbon emission prediction value of each operation stage of the public building;

[0170] The operating framework of the carbon emission dynamic calculation model includes at least four operation level modules, and the at least four operation level modules correspond to each operation stage of the public building respectively. Each operation level module includes at least one sub-module, and each sub-module includes a carbon emission calculation unit; the carbon emission calculation unit is constructed according to the energy consumption type, material flow process and carbon emission mechanism of the corresponding operation level, and is used for: while processing operation data based on a deep neural network to predict carbon emissions, a particle swarm optimization algorithm is used to explore with the weights and biases in the deep neural network as particles until the exploration stop condition is met, and the deep neural network is optimized with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value of the corresponding operation stage of the public building.

[0171] In one embodiment, the exploration stop condition is that the change in the fitness value of the particle's current position compared to the fitness value of the previous position is not higher than a preset threshold, or the number of iterations of the particle reaches a maximum number of iterations. The carbon emission dynamic measurement unit 20 is specifically used to:

[0172] The starting position of the generated particles;

[0173] Calculate the fitness value of the particle at the starting position, use the preset weight calculation formula to calculate the inertia weight of the particle at the starting position, and determine the next position of the particle based on the inertia weight; use the next position as the starting position of the particle, and the inertia weight is determined by the current number of iterations and the maximum number of iterations of the particle;

[0174] Iteratively calculate the fitness value of the particle at the starting position; use a preset weight calculation formula to calculate the inertia weight of the particle at the starting position; determine the next position of the particle based on the inertia weight, and use the next position as the starting position of the particle until the fitness value of the particle at the starting position changes by no more than a preset threshold compared to the fitness value of the particle at the previous position; or the number of iterations of the particle reaches the maximum number of iterations.

[0175] In one embodiment, the above-mentioned carbon emission dynamic calculation device based on public building operation further includes an optimization parameter exploration unit;

[0176] The optimization parameter exploration unit is used to use the particle swarm optimization algorithm to explore the hyperparameters and / or network structure characteristics of the deep neural network as particles until the exploration stop condition is met, and to optimize the deep neural network with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and to use the optimized deep neural network to predict carbon emissions to obtain a predicted carbon emission value.

[0177] In one embodiment, the carbon emission dynamic calculation unit 20 is specifically used to:

[0178] While using deep neural networks to process operational data, carbon emissions predictions are made by combining energy carbon emission factors and special carbon emission factors that match public buildings; special carbon emission factors include at least the carbon emission factors corresponding to refrigerant leakage of cold chain equipment in public buildings, the use and disposal of disposable plastic products, and wastewater treatment of fresh food processing.

[0179] In one embodiment, the above-mentioned carbon emission dynamic calculation device based on public building operation further includes a carbon emission factor updating unit;

[0180] The carbon emission factor updating unit is used to update the energy carbon emission factor according to a first preset period and the special carbon emission factor according to a second preset period; the time interval between each period in the first preset period is shorter than the time interval between each period in the second preset period.

[0181] In an implementation manner, the above-mentioned carbon emission dynamic measurement device based on public building operation further includes a carbon emission prediction value adjustment unit;

[0182] The carbon emission prediction value adjustment unit is used to adjust the carbon emission prediction value by using the market dynamic information corresponding to the public building and then output it.

[0183] The present application also provides a carbon emission dynamic calculation device based on public building operation. Figure 5As shown, it shows a schematic diagram of the structure of the carbon emission dynamic calculation device based on public building operation provided by the present application. The carbon emission dynamic calculation device based on public building operation in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 5 The carbon emission dynamic measurement device based on public building operation shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0184] like Figure 5 As shown, the carbon emission dynamics calculation device based on the operation of public buildings may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage device 608 to the random access memory (RAM) 603. When the carbon emission dynamics calculation device based on the operation of public buildings is powered on, the RAM 603 also stores various programs and data required for the operation of the carbon emission dynamics calculation device based on the operation of public buildings. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. The input / output (I / O) interface 605 is also connected to the bus 604.

[0185] Typically, the following devices can be connected to the I / O interface 605: input devices 606 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 608 including, for example, a memory card, a hard disk, etc.; and communication devices 609. The communication device 609 can allow the carbon emission dynamics calculation device based on public building operations to communicate wirelessly or wired with other devices to exchange data. Although Figure 5 The carbon emission dynamic calculation device based on public building operation with various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0186] Also provided in an embodiment of the present application is a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the dynamic carbon emission measurement methods based on public building operations provided in the embodiments of the present application.

[0187] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When one or more computer programs are executed by an electronic device, the electronic device can implement any one of the dynamic carbon emission measurement methods based on public building operations provided in the embodiments of the present application.

[0188] Optionally, the present application also creates a user interaction platform for use with the above-mentioned dynamic carbon emission measurement method based on public building operations and related devices.

[0189] Specifically, a visual web application platform is created for the management of public buildings, which uses intuitive charts (bar charts, line charts, pie charts, heat maps, etc.) to display real-time carbon emission monitoring data, historical trend analysis, comparison of the effectiveness of different emission reduction projects, future carbon emission forecast curves and other information. The platform has interactive operation functions, and the management can dig into data details through clicks, drags and other operations, such as viewing the detailed energy consumption and carbon emission data of a certain device on a certain day, inputting different operation adjustment plans (such as closing one hour in advance, adding one logistics distribution), and instantly obtaining the corresponding carbon emission change forecast, which assists in formulating energy-saving, emission reduction and cost optimization operation strategies.

[0190] Finally, this application introduces the integrated carbon emission detection equipment and supporting systems customized for public buildings, which are matched with the dynamic carbon emission measurement method based on public building operations and related devices.

[0191] The supporting systems include but are not limited to edge gateway models and cloud system models.

[0192] The edge gateway model is a data model provided by the edge gateway device, which may specifically include a carbon emission calculation model, an information service model, and a device model.

[0193] Among them, the carbon emission calculation model is mainly used to optimize and update the edge data provided by the edge gateway and the dynamic carbon emission factors obtained by the cloud platform, providing dynamic metering support and data correction for the edge gateway; the information service model is mainly used to model edge gateway resources, system services and application management, and provide model support for the deployment and control of information services implemented on the edge gateway; the device model is mainly used to model the edge gateway and its connected secondary devices. The secondary devices are specifically connected during the measurement information and operation status, providing basic support for applications such as equipment situation awareness and online monitoring.

[0194] Taking a commercial supermarket as an example, the configuration of the edge computing gateway mainly includes data preprocessing and preliminary analysis and local early warning and emergency response:

[0195] Data preprocessing and preliminary analysis are carried out by the edge computing gateway located in the local network of the commercial supermarket. It receives data from various monitoring terminals, first cleans the data, removes abnormal values ​​(such as erroneous readings caused by sensor failure or transient electromagnetic interference), and uses data interpolation algorithms to fill in a small number of missing data points. Preliminary analysis is performed on real-time energy consumption data, such as classification and statistics by equipment type, region, and time period, and calculation of characteristic values ​​such as power peak, valley, and average values, to provide concise and high-quality data for further in-depth analysis on the cloud.

[0196] Local early warning and emergency response are realized by built-in intelligent algorithms, which realize real-time early warning locally based on preset carbon emission thresholds, energy consumption ceilings, and equipment failure risk indicators. When the lighting energy consumption in a certain area of ​​a commercial supermarket suddenly soars, the abnormal temperature of cold chain equipment may cause the risk of large-scale refrigerant leakage, and the overall carbon emission rate is close to the red line of exceeding the standard, the edge computing gateway immediately triggers an audible and visual alarm to notify local operation and maintenance personnel, and can automatically take emergency measures, such as remotely shutting down some non-critical equipment and starting standby refrigeration units, etc., to minimize losses and risks.

[0197] The cloud system model is a data model provided by the cloud system, which may include analysis and calculation models, time series prediction models, and plan generation models.

[0198] Among them, the analysis and calculation model mainly calculates the dynamic carbon emission factor and conformity characteristic analysis of the acquired operation data, and provides dynamic metering support for prediction, plan generation and edge gateway; the timing analysis and prediction model is used to generate the output of new energy systems and equipment energy consumption forecast data in the user-side energy-consuming equipment, and provide support for the formulation of operation plans; the plan generation model is used to generate the optimal operation plan by using planning algorithms, improved particle swarm algorithms, improved artificial fish school algorithms, etc., to form control signals for the control module, and send them to the edge gateway to provide support for controlling energy-consuming equipment.

[0199] For the supporting systems mentioned above, they can realize on-site collection, cloud-edge collaboration, and plug-and-play.

[0200] The realization of on-site collection and cloud-edge collaboration functions is specifically reflected in the ability to collect and monitor edge-side data on-site, and support the collaboration between local application functions and edge clusters.

[0201] The realization of the plug-and-play function is specifically reflected in the edge gateway's ability to automatically connect to the network, automatically register with the edge cluster, and send a self-describing model after registration is completed; it has the ability to automatically form a network, discover and automatically connect to various systems or terminals downstream of the edge gateway.

[0202] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0203] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0204] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0205] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a training device or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device or data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integration. Available media can be magnetic media, (e.g., floppy disk, hard disk, tape), optical media (e.g., DVD), or semiconductor media (e.g., solid-state hard disk (SSD)), etc.

Claims

1. A dynamic calculation method for carbon emissions based on public building operations, characterized in that: include: Obtain operational data of public buildings collected by smart terminals; Inputting the operation data into a dynamic carbon emission calculation model based on time series to obtain the carbon emission forecast value of each operation stage of the public building; The operation framework of the carbon emission dynamic calculation model includes at least four operation level modules, and the at least four operation level modules correspond to the various operation stages of the public building respectively. Each operation level module includes at least one submodule, and each submodule includes a carbon emission calculation unit; the carbon emission calculation unit is constructed according to the energy consumption type, material flow process and carbon emission mechanism of the corresponding operation level, and is used to: while processing the operation data based on a deep neural network to predict carbon emissions, a particle swarm optimization algorithm is used to explore with the weights and biases in the deep neural network as particles until the exploration stop condition is met, and the deep neural network is optimized with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value of the corresponding operation stage of the public building.

2. The method for dynamic calculation of carbon emissions based on public building operations according to claim 1 is characterized in that: The exploration stop condition is that the change in the fitness value of the particle at the current position compared to the fitness value at the previous position is not higher than a preset threshold, or the number of iterations of the particle reaches a maximum number of iterations; The particle swarm optimization algorithm is used to explore the weights and biases in the deep neural network as particles until the exploration stop condition is met, including: generating a starting position of the particle; Calculating the fitness value of the particle at the starting position, calculating the inertia weight of the particle at the starting position using a preset weight calculation formula, and determining the next position of the particle according to the inertia weight; using the next position as the starting position of the particle, the inertia weight is determined by the current iteration number of the particle and the maximum iteration number; Iteratively executing the calculation of the fitness value of the particle at the starting position; using a preset weight calculation formula to calculate the inertia weight of the particle at the starting position; determining the next position of the particle according to the inertia weight, and using the next position as the starting position of the particle, until the fitness value of the particle at the starting position changes by no more than the preset threshold value compared to the fitness value of the particle at the previous position; or the number of iterations of the particle reaches the maximum number of iterations.

3. The method for dynamic calculation of carbon emissions based on public building operations according to claim 1 is characterized in that: Also includes: The particle swarm optimization algorithm is used to explore the hyperparameters and / or network structure characteristics of the deep neural network as particles until the exploration stop condition is met, and the deep neural network is optimized with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value.

4. The method for dynamic calculation of carbon emissions based on public building operations according to claim 1 is characterized in that: The method of processing the operation data based on a deep neural network to predict carbon emissions includes: While using the deep neural network to process the operating data, carbon emissions prediction is performed in combination with energy carbon emission factors and special carbon emission factors matching the public building; the special carbon emission factors at least include carbon emission factors corresponding to refrigerant leakage of cold chain equipment in public buildings, use and disposal of disposable plastic products, and wastewater treatment of fresh food processing.

5. The method for dynamic calculation of carbon emissions based on public building operations according to claim 4 is characterized in that: Also includes: The energy carbon emission factor is updated according to a first preset period, and the special carbon emission factor is updated according to a second preset period; The time interval between each cycle in the first preset cycle is shorter than the time interval between each cycle in the second preset cycle.

6. The method for dynamic calculation of carbon emissions based on public building operations according to claim 1 is characterized in that: Also includes: The carbon emission forecast value is adjusted using the market dynamic information corresponding to the public building and then outputted.

7. A carbon emission dynamic calculation device based on public building operation, characterized in that: include An acquisition unit, used to acquire the operation data of public buildings collected by the intelligent terminal; A carbon emission dynamic calculation unit, used to input the operation data into a carbon emission dynamic calculation model based on a time series to obtain a carbon emission forecast value at each operation stage of the public building; The operation framework of the carbon emission dynamic calculation model includes at least four operation level modules, and the at least four operation level modules correspond to the various operation stages of the public building respectively. Each operation level module includes at least one submodule, and each submodule includes a carbon emission calculation unit; the carbon emission calculation unit is constructed according to the energy consumption type, material flow process and carbon emission mechanism of the corresponding operation level, and is used to: while processing the operation data based on a deep neural network to predict carbon emissions, a particle swarm optimization algorithm is used to explore with the weights and biases in the deep neural network as particles until the exploration stop condition is met, and the deep neural network is optimized with the optimal particles that meet the exploration stop condition to obtain an optimized deep neural network, and carbon emissions are predicted by the optimized deep neural network to obtain the carbon emission prediction value of the corresponding operation stage of the public building.

8. A carbon emission dynamic calculation device based on public building operation, characterized in that: The method comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the device for dynamic measurement of carbon emissions based on public building operations can implement the method for dynamic measurement of carbon emissions based on public building operations as described in any one of claims 1 to 6.

9. A computer program product, characterized in that It includes computer-readable instructions, which, when executed on an electronic device, enable the electronic device to implement the method for dynamic calculation of carbon emissions based on public building operations as described in any one of claims 1 to 6.

10. A computer storage medium, characterized in that: The storage medium carries one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the method for dynamic measurement of carbon emissions based on public building operations as described in any one of claims 1 to 6.

Images