Comprehensive energy carbon management method and system based on digital twinborn technology
Through digital twin technology and intelligent algorithms, combined with sensor network data, anomaly detectors are built, which solves the accuracy and efficiency of traditional carbon emission detection methods, and achieves more accurate and efficient energy and carbon emission management.
Patent Information
- Application Number
- CN202411062007.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-22
AI Technical Summary
Traditional carbon emission detection methods such as direct measurement method and remote sensing monitoring have equipment pollution and damage resulting in inaccuracy, and manual detection takes a long time and is affected by subjective factors, resulting in uncertainty in monitoring results.
The integrated energy carbon management method based on digital twin technology is adopted, energy use and carbon emission data are collected through sensor networks, and a digital twin model is used for sequence correlation encoding and reasoning, anomaly detector is built to intelligently identify abnormal conditions at carbon emission levels.
Reduce errors caused by equipment failure or environmental factors, provide real-time data and in-depth analysis decision-making support, helping enterprises formulate more effective energy and carbon emission management strategies, and improve detection accuracy and efficiency.
Smart Images

Figure CN119005509A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent management, and more specifically, to a comprehensive energy and carbon management method and system based on digital twin technology. Background Art
[0002] As global climate change and environmental problems become increasingly serious, energy efficiency and carbon emission management are receiving more and more attention. Among them, traditional carbon emission detection, such as direct measurement and remote sensing monitoring, have some problems. Specifically, the direct measurement method usually installs sensors or sampling equipment at the emission source to directly measure the composition and concentration of the emitted gas; however, this method is greatly affected by the equipment. When the sampling equipment is contaminated or damaged, the data will be inaccurate, further affecting the accuracy of carbon emission detection. Remote sensing technology monitors the concentration of greenhouse gases in the atmosphere through sensors carried by satellites or aircraft; however, remote sensing data may be affected by factors such as cloud cover and atmospheric conditions, resulting in uncertainty in monitoring results. In addition, the abnormal detection results of carbon emissions cannot be obtained directly and still need to be performed manually; and this manual detection method is time-consuming and is affected by human subjective factors, which may lead to inaccurate results.
[0003] Digital twin technology creates virtual copies of physical entities to achieve real-time monitoring, analysis and optimization of physical systems. The application of digital twin technology in the field of energy and carbon management provides new ideas for solving or alleviating the above problems. Summary of the invention
[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a comprehensive energy-carbon management method and system based on digital twin technology, which uses digital twin technology and intelligent algorithms to comprehensively utilize the time-series implicit correlation of energy usage data to infer emission levels, and then compares and interacts with the actual detected emission levels in a high-dimensional feature distribution space to obtain a correlation interaction representation and difference information for representing the inferred emission level and the detected emission level, and then uses a classifier to construct an anomaly detector to intelligently identify abnormal conditions in carbon emission level detection. In this way, errors caused by important equipment failures or environmental factors in traditional detection methods are reduced, and decision support based on real-time data and in-depth analysis is provided to help companies and organizations develop more effective energy and carbon emission management strategies.
[0005] According to one aspect of the present application, a comprehensive energy and carbon management method based on digital twin technology is provided, which includes:
[0006] A time series of energy usage and carbon emission data collected by a sensor network deployed in the physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels;
[0007] integrating time series of said energy usage and carbon emissions data into the digital twin model;
[0008] In the digital twin model, the time series of the energy use and carbon emission data are sequence-associated encoded to obtain a temperature value time series associated implicit feature vector, a pressure value time series associated implicit feature vector, a flow value time series associated implicit feature vector, an energy consumption value time series associated implicit feature vector, and an emission level time series associated implicit feature vector;
[0009] In the digital twin model, the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector are input into the emission level reasoner to obtain the emission level time series reasoning representation vector;
[0010] In the digital twin model, a detection result is determined based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection.
[0011] According to another aspect of the present application, a comprehensive energy-carbon management system based on digital twin technology is provided, which includes:
[0012] A data collection module, for collecting time series of energy usage and carbon emission data collected by a sensor network deployed in the physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels;
[0013] A digital twin module for integrating the time series of energy usage and carbon emission data into a digital twin model;
[0014] A sequence association coding module is used to perform sequence association coding on the time series of the energy use and carbon emission data in the digital twin model to obtain a temperature value time series association implicit feature vector, a pressure value time series association implicit feature vector, a flow value time series association implicit feature vector, an energy consumption value time series association implicit feature vector, and an emission level time series association implicit feature vector;
[0015] An emission level time series reasoning representation module is used to input the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector into the emission level reasoner in the digital twin model to obtain an emission level time series reasoning representation vector;
[0016] A detection result determination module is used to determine the detection result in the digital twin model based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection.
[0017] Compared with the prior art, the present application provides a comprehensive energy and carbon management method and system based on digital twin technology, which uses digital twin technology and intelligent algorithms to comprehensively utilize the implicit temporal correlation of energy usage data to infer emission levels, and then compares and interacts with the actual detected emission levels in a high-dimensional feature distribution space to obtain a correlation interaction representation and difference information for representing the inferred emission level and the detected emission level, and then uses a classifier to build an anomaly detector to intelligently identify abnormal conditions in carbon emission level detection. In this way, the errors caused by important equipment failures or environmental factors in traditional detection methods are reduced, and decision support based on real-time data and in-depth analysis is provided to help enterprises and organizations develop more effective energy and carbon emission management strategies. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0019] Figure 1 It is a flow chart of a comprehensive energy and carbon management method based on digital twin technology according to an embodiment of the present application;
[0020] Figure 2 A system architecture diagram of a comprehensive energy and carbon management method based on digital twin technology according to an embodiment of the present application;
[0021] Figure 3 It is a flowchart of sub-step S3 of the comprehensive energy and carbon management method based on digital twin technology according to an embodiment of the present application;
[0022] Figure 4 It is a flowchart of sub-step S5 of the comprehensive energy-carbon management method based on digital twin technology according to an embodiment of the present application;
[0023] Figure 5 It is a block diagram of a comprehensive energy and carbon management system based on digital twin technology according to an embodiment of the present application. DETAILED DESCRIPTION
[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0025] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.
[0026] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.
[0027] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.
[0028] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0029] Traditional carbon emission detection, such as direct measurement and remote sensing monitoring, has some problems. Specifically, the direct measurement method usually installs sensors or sampling equipment at the emission source to directly measure the composition and concentration of the emitted gas; however, this method is greatly affected by the equipment. When the sampling equipment is contaminated or damaged, the data will be inaccurate, further affecting the accuracy of carbon emission detection. Remote sensing technology monitors the concentration of greenhouse gases in the atmosphere through sensors carried by satellites or aircraft; however, remote sensing data may be affected by factors such as cloud cover and atmospheric conditions, resulting in uncertainty in monitoring results. In addition, the abnormal detection results of carbon emissions cannot be obtained directly and still need to be performed manually; this manual detection method is time-consuming and is affected by human subjective factors, which may lead to inaccurate results.
[0030] Digital twin technology creates virtual copies of physical entities to achieve real-time monitoring, analysis and optimization of physical systems. The application of digital twin technology in the field of energy and carbon management provides new ideas for solving or alleviating the above problems.
[0031] In the technical solution of the present application, a comprehensive energy and carbon management method based on digital twin technology is proposed. Figure 1 This is a flow chart of a comprehensive energy and carbon management method based on digital twin technology according to an embodiment of the present application. Figure 2 : is a system architecture diagram of a comprehensive energy and carbon management method based on digital twin technology according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the comprehensive energy and carbon management method based on digital twin technology includes the following steps: S1, collecting the time series of energy usage and carbon emission data through a sensor network deployed in a physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels; S2, integrating the time series of energy usage and carbon emission data into a digital twin model; S3, in the digital twin model, performing sequence association encoding on the time series of energy usage and carbon emission data to obtain a time series association implicit feature vector of the temperature value, a time series association implicit feature vector of the pressure value, a time series association implicit feature vector of the flow value Implicit feature vectors, energy consumption value time series associated implicit feature vectors and emission level time series associated implicit feature vectors; S4, in the digital twin model, the temperature value time series associated implicit feature vectors, the pressure value time series associated implicit feature vectors, the flow value time series associated implicit feature vectors, and the energy consumption value time series associated implicit feature vectors are input into the emission level reasoner to obtain the emission level time series reasoning representation vector; S5, in the digital twin model, based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vectors, determine the detection result, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection.
[0032] In particular, the S1 is a time series of energy usage and carbon emission data collected by a sensor network deployed in a physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels. Here, the physical system refers to an actual physical system or equipment that is used to produce, transport, distribute or use energy in various fields such as industry, construction, and transportation. In an embodiment of the present application, the physical system may be: industrial equipment such as boilers, turbines, compressors, etc.; building systems such as heating, ventilation and air conditioning systems (HVAC); and energy production facilities such as power plants, solar panels, wind turbines, etc. In this way, deploying a sensor network to collect time series of energy usage and carbon emission data in real time can provide an important data basis for subsequent carbon emission reasoning and abnormal state identification.
[0033] In particular, S2 integrates the time series of energy use and carbon emission data into a digital twin model. The digital twin model is a virtual, software-driven representation that simulates the operation of a physical system, allowing analysis, prediction, and optimization without causing interference or risk to the actual system. That is, high-level, high-maturity digital twin technology is used to digitally model and simulate physical energy systems and carbon emission processes to achieve accurate mapping between virtual and reality.
[0034] In particular, in the S3, in the digital twin model, the time series of the energy use and carbon emission data are sequence-associated encoded to obtain a temperature value time series associated implicit feature vector, a pressure value time series associated implicit feature vector, a flow value time series associated implicit feature vector, an energy consumption value time series associated implicit feature vector, and an emission level time series associated implicit feature vector. In particular, in a specific example of the present application, Figure 3 As shown, the S3 includes: S31, regularizing the time series of the energy usage and carbon emission data according to the parameter samples to obtain the time series of temperature values, the time series of pressure values, the time series of flow values, the time series of energy consumption values and the time series of emission levels; S32, inputting the time series of temperature values, the time series of pressure values, the time series of flow values, the time series of energy consumption values and the time series of emission levels into the double-branch structure sequence encoder respectively to obtain the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, the energy consumption value time series associated implicit feature vector and the emission level time series associated implicit feature vector.
[0035] Specifically, in S31, the time series of the energy use and carbon emission data are regularized according to the parameter samples to obtain the time series of temperature values, the time series of pressure values, the time series of flow values, the time series of energy consumption values, and the time series of emission levels. It should be understood that by regularizing the energy use and carbon emission data according to the parameter samples, it is possible to ensure that the data of different parameters are aligned in time, making the data more consistent and comparable.
[0036] Specifically, in S32, the time series of the temperature values, the time series of the pressure values, the time series of the flow values, the time series of the energy consumption values, and the time series of the emission levels are respectively input into a double-branch structure sequence encoder to obtain the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, the energy consumption value time series associated implicit feature vector, and the emission level time series associated implicit feature vector. In particular, in a specific example of the present application, the time series of the temperature values, the pressure value time series, the flow value time series, the energy consumption value time series associated implicit feature vector, and the emission level time series associated implicit feature vector are respectively input into a double-branch structure sequence encoder including a 1DCNN model and a bidirectional gated recurrent unit to obtain the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, the energy consumption value time series associated implicit feature vector, and the emission level time series associated implicit feature vector. That is, make full use of the advantages of 1DCNN (one-dimensional convolutional neural network) model and bidirectional gated recurrent unit in processing time series data, extract and capture local patterns and trends in time series data, and identify the dynamic time series characteristics of temperature data, pressure data, flow data, energy consumption data and emission level data.
[0037] In other specific examples of the present application, the time series of the energy usage and carbon emission data can also be sequence-associated encoded in other ways to obtain temperature value time series associated implicit feature vectors, pressure value time series associated implicit feature vectors, flow value time series associated implicit feature vectors, energy consumption value time series associated implicit feature vectors and emission level time series associated implicit feature vectors, for example: input the time series of the energy usage and carbon emission data; arrange the data in chronological order to ensure that the timestamps of the data are aligned; sequence-associated encode the time series of each parameter to capture the associated information and Pattern; for the sequence association encoding result of each parameter, a feature extraction method (such as LSTM, GRU and other recurrent neural network models) can be used to extract the time series association implicit feature vector; for each parameter, the extracted time series association implicit feature vector can be used as an abstract representation of the parameter in time, which contains the key features and patterns in the time series; output the temperature value time series association implicit feature vector, the pressure value time series association implicit feature vector, the flow value time series association implicit feature vector, the energy consumption value time series association implicit feature vector and the emission level time series association implicit feature vector.
[0038] In particular, in the S4, in the digital twin model, the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector are input into the emission level reasoner to obtain the emission level time series reasoning representation vector. In particular, in a specific example of the present application, the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector are input into the emission level reasoner based on the deep belief network to obtain the emission level time series reasoning representation vector. Among them, the deep belief network (DBN) is a deep neural network composed of multiple layers of restricted Boltzmann machines (RBMs), which can learn high-level feature representations of data. These high-level feature representations can reveal complex time series implicit association feature patterns in energy usage data, thereby inferentially encoding and generating carbon emission level prediction results.
[0039] In particular, in S5, in the digital twin model, based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, a detection result is determined, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection. In particular, in a specific example of the present application, if Figure 4As shown, the S5 includes: S51, inputting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector into an associated interaction analysis network guided by a prior distribution to obtain an emission level reasoning-emission level detection interaction analysis representation vector; S52, inputting the emission level reasoning-emission level detection interaction analysis representation vector as the interaction feature into a classifier-based anomaly detector to obtain the detection result.
[0040] Specifically, in S51, the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector are input into the associated interactive analysis network guided by the prior distribution to obtain the emission level reasoning-emission level detection interactive analysis representation vector. It should be understood that the interactive analysis of the inference emission level characteristics represented by the emission level time series reasoning representation vector and the actual detection emission level characteristics represented by the emission level time series associated implicit feature vector can capture the difference information between the two, thereby providing a basis for the identification of abnormal states. However, due to the high uncertainty of the time series coding reasoning characteristics based on the external associated data of carbon emissions, at the same time, there is a complex nonlinear correlation relationship between this time series coding reasoning feature and the actual detected carbon emission level time series feature pattern. Therefore, in the technical solution of the present application, further, based on the digital twin model, the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector are input into the associated interactive analysis network guided by the prior distribution to obtain the emission level reasoning-emission level detection interactive analysis representation vector. In an embodiment of the present application, the association interaction analysis network guided by the prior distribution uses the first random coefficient and the second random coefficient to perform weighted adjustment on the interaction feature representation after linear interaction (divided by position) to guide and optimize the process of interaction fusion. More specifically, the first random coefficient and the second random coefficient are randomly sampled from a standard normal distribution, which can introduce the characteristics of a normal distribution to a certain extent, thereby integrating the prior knowledge or assumptions of the data into the model to help guide the learning process and help the model better handle noise and changes in practical applications.
[0041] In an embodiment of the present application, the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector are input into an associated interaction analysis network guided by a prior distribution to obtain an emission level reasoning-emission level detection interaction analysis representation vector, including: interacting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector to obtain a first indexed emission level reasoning-emission level detection time series interaction representation vector and a second indexed emission level reasoning-emission level detection time series interaction representation vector; determining a first random coefficient and a second random coefficient; based on the first random coefficient and the second random coefficient, performing an associated fusion based on a prior distribution guided on the first indexed emission level reasoning-emission level detection time series interaction representation vector and the second indexed emission level reasoning-emission level detection time series interaction representation vector to obtain the emission level reasoning-emission level detection interaction analysis representation vector.
[0042] Among them, the process of interacting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector to obtain a first exponential emission level reasoning-emission level detection time series interaction representation vector and a second exponential emission level reasoning-emission level detection time series interaction representation vector includes: calculating the positional division of the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, multiplying the vectors obtained by the positional division with the first weight coefficient and the second weight coefficient respectively to obtain the first emission level reasoning-emission level detection time series interaction representation vector and the second emission level reasoning-emission level detection time series interaction representation vector; calculating an exponential function value with a natural constant as the base and each eigenvalue of the first emission level reasoning-emission level detection time series interaction representation vector as an exponent to obtain the first exponential emission level reasoning-emission level detection time series interaction representation vector; calculating an exponential function value with a natural constant as the base and each eigenvalue of the second emission level reasoning-emission level detection time series interaction representation vector as an exponent to obtain the second exponential emission level reasoning-emission level detection time series interaction representation vector.
[0043] More specifically, based on the first random coefficient and the second random coefficient, the process of performing an associative fusion based on the prior distribution guidance on the first indexed emission level reasoning-emission level detection timing interaction representation vector and the second indexed emission level reasoning-emission level detection timing interaction representation vector to obtain the emission level reasoning-emission level detection interaction analysis representation vector includes: calculating the position-weighted sum of the first indexed emission level reasoning-emission level detection timing interaction representation vector and the second indexed emission level reasoning-emission level detection timing interaction representation vector to obtain the emission level reasoning-emission level detection interaction analysis representation vector, wherein the weight of the first indexed emission level reasoning-emission level detection timing interaction representation vector is the first random coefficient, and the weight of the second indexed emission level reasoning-emission level detection timing interaction representation vector is the second random coefficient.
[0044] In summary, in the above embodiment, the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector are input into the associated interaction analysis network guided by the prior distribution to obtain the emission level reasoning-emission level detection interactive analysis representation vector, including: using the associated interaction analysis network guided by the prior distribution to perform an associated interaction analysis on the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector with the following associated interaction analysis formula to obtain the emission level reasoning-emission level detection interactive analysis representation vector; wherein the associated interaction analysis formula is:
[0045]
[0046] Wherein, v2 and v1 are the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, p1 and p2 are the first weight coefficient and the second weight coefficient, f1 and f2 are the first random coefficient and the second random coefficient randomly sampled from the standard normal distribution, V c is the emission level inference-emission level detection interaction analysis representation vector.
[0047] Specifically, in S52, the emission level reasoning-emission level detection interactive analysis representation vector is input as the interactive feature into a classifier-based anomaly detector to obtain the detection result. In a specific example of the present application, the emission level reasoning-emission level detection interactive analysis representation vector is input into a classifier-based anomaly detector to obtain a detection result, and the detection result is used to indicate whether there is an anomaly in the carbon emission level detection.
[0048] In the technical solution of the present application, the emission level temporal reasoning representation vector is obtained by inputting the temperature value temporal association implicit feature vector, the pressure value temporal association implicit feature vector, the flow value temporal association implicit feature vector, and the energy consumption value temporal association implicit feature vector into the emission level reasoner based on the deep belief network, that is, the emission level temporal reasoning representation vector is essentially the encoded reasoning result based on the external correlation data of carbon emissions. Considering that the reasoning performance of the emission level reasoner based on the deep belief network is limited by the training samples and training strategies, it naturally has the encoding-reasoning feature deviation introduced by the incompleteness of training. This means that when the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector are input into the associated interaction analysis network guided by prior distribution for associated interaction analysis, although the associated interaction analysis network guided by prior distribution utilizes prior distribution characteristics to constrain the characteristic manifold boundary of associated interaction, the encoding-inference feature deviation of the emission level time series reasoning representation vector will cause the emission level reasoning-emission level detection interaction analysis representation vector to have local distribution differences, resulting in a deviation in its compatibility with the predetermined category of the classifier-based anomaly detector, affecting the classification iteration effect.
[0049] In a preferred embodiment, passing the emission level reasoning-emission level detection interactive analysis representation vector through a classifier-based anomaly detector to obtain a detection result comprises the following steps:
[0050] First, the autocorrelation matrix of the emission level reasoning-emission level detection interactive analysis representation vector and its own transposition is calculated, and the inner product of the i-th and j-th row vectors of the autocorrelation matrix is calculated as the matrix value of the (i, j)-th position of the weight matrix to obtain the weight matrix. Then, after matrix multiplication of the weight matrix and the autocorrelation matrix, the matrix-vector multiplication is further performed with the emission level reasoning-emission level detection interactive analysis representation vector to obtain a correction vector. Finally, the correction vector is dot-multiplied with the emission level reasoning-emission level detection interactive analysis representation vector to obtain an optimized emission level reasoning-emission level detection interactive analysis representation vector, and the optimized emission level reasoning-emission level detection interactive analysis representation vector is passed through the classifier-based anomaly detector to obtain the detection result.
[0051] The optimization steps are specifically expressed as follows:
[0052]
[0053] Where V and V ′are the emission level reasoning-emission level detection interactive analysis representation vector and the optimized emission level reasoning-emission level detection interactive analysis representation vector, respectively, (·) T represents the transpose, M is the autocorrelation matrix, the value of L is the length of the emission level inference-emission level detection interaction analysis representation vector, m i,j is the matrix value of the (i, j)th position of the weight matrix, M ′ is the weight matrix, R is the real number field, V i and V j are the i-th and j-th row vectors of the autocorrelation matrix, respectively, represents matrix multiplication, ⊙ represents dot product, R' × ' represents a set of real numbers of size L×L.
[0054] Accordingly, in this preferred embodiment, based on the self-correlation dimension of the emission level reasoning-emission level detection interactive analysis representation vector as the object to be modulated, the correlation expansion is performed based on the spatial sub-dimensional complexity of the high-dimensional feature space of the feature distribution, thereby introducing the decomposable correlation dimension offset of the emission level reasoning-emission level detection interactive analysis representation vector into the heterogeneous correlation embedding space, thereby enhancing its correlation self-consistency through the joint fine-tuning of the decomposable dimension set represented by the heterogeneous correlation embedding space, so as to improve the predetermined category harmony of the feature set of the emission level reasoning-emission level detection interactive analysis representation vector in the associated target classification domain, thereby improving the classification training speed and the accuracy of the classification results of the emission level reasoning-emission level detection interactive analysis representation vector when classified by a classifier-based anomaly detector.
[0055] It is worth mentioning that in other specific examples of the present application, the detection result can be determined based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector in other ways, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection, for example: inputting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector; fusing or interacting the emission level time series reasoning representation vector and the time series associated implicit feature vector to capture the correlation information between the two; inputting the fused features into a learning model, such as a neural network, a support vector machine, etc., to learn the relationship and pattern between the features; using the trained model to predict and classify new emission level data; based on the results of the model output, it can be determined whether there is an abnormality in the emission level; based on the results of the model output, the detection result can be expressed in the form of a binary classification (the presence or absence of an abnormality) or a probability value (indicating the degree of abnormality).
[0056] In summary, the comprehensive energy and carbon management method based on digital twin technology according to the embodiment of the present application is explained, which uses digital twin technology and intelligent algorithms to comprehensively utilize the implicit temporal correlation of energy usage data to infer emission levels, and then compares and interacts with the actual detected emission levels in a high-dimensional feature distribution space to obtain a correlation interaction representation and difference information for representing the inferred emission level and the detected emission level, and then uses a classifier to build an anomaly detector to intelligently identify abnormal conditions in carbon emission level detection. In this way, the errors caused by important equipment failures or environmental factors in traditional detection methods are reduced, and decision support based on real-time data and in-depth analysis is provided to help enterprises and organizations develop more effective energy and carbon emission management strategies.
[0057] Furthermore, a comprehensive energy and carbon management system based on digital twin technology is also provided.
[0058] Figure 5 FIG. 1 is a block diagram of a comprehensive energy and carbon management system based on digital twin technology according to an embodiment of the present application. Figure 5 As shown, according to the embodiment of the present application, the comprehensive energy and carbon management system 300 based on digital twin technology includes: a data acquisition module 310, which is used to collect the time series of energy usage and carbon emission data through the sensor network deployed in the physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels; a digital twin module 320, which is used to integrate the time series of energy usage and carbon emission data into the digital twin model; a sequence association coding module 330, which is used to perform sequence association coding on the time series of energy usage and carbon emission data in the digital twin model to obtain the temperature value time series association implicit feature vector, the pressure value time series association implicit feature vector, the flow value time series association implicit feature vector The digital twin model comprises an emission level time series reasoning representation module 340, which is used to input the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector into the emission level reasoner in order to obtain the emission level time series reasoning representation vector; a detection result determination module 350 is used to determine the detection result in the digital twin model based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection.
[0059] As described above, the comprehensive energy-carbon management system 300 based on digital twin technology according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a comprehensive energy-carbon management algorithm based on digital twin technology. In a possible implementation, the comprehensive energy-carbon management system 300 based on digital twin technology according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or hardware module. For example, the comprehensive energy-carbon management system 300 based on digital twin technology can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the comprehensive energy-carbon management system 300 based on digital twin technology can also be one of the many hardware modules of the wireless terminal.
[0060] Alternatively, in another example, the comprehensive energy-carbon management system 300 based on digital twin technology and the wireless terminal may also be separate devices, and the comprehensive energy-carbon management system 300 based on digital twin technology may be connected to the wireless terminal through a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.
[0061] In the specific example of this application, the integrated energy and carbon management application platform based on digital twin technology is a system that integrates advanced technologies such as digital twin, Internet of Things, cloud computing, big data and artificial intelligence, aiming to provide enterprises with comprehensive integrated energy and carbon emission management solutions. The following are the main features and applications of the platform:
[0062] 1. System features:
[0063] 1. Integration of "data, energy, and carbon" flows, fusion and innovation: Based on high-level, high-maturity digital twin technology, by converging "data" flows → optimizing "energy" flows → controlling "carbon" flows, unified and comprehensive management of accurate monitoring and intelligent optimization of the operating status of physical entities and their comprehensive energy usage can be achieved. Based on the three flows of "data, energy, and carbon", using interdisciplinary and cross-field advanced technologies, we can achieve deep integration and innovation in the development of digital technology and the energy and carbon industries.
[0064] 2. Digital twin technology: Use high-level, high-maturity digital twin technology to digitally model and simulate physical energy systems and carbon emission processes to achieve accurate mapping between virtual and reality. Through real-time data updates, ensure the synchronization of digital twin models and physical systems, and provide accurate decision support for energy and carbon management.
[0065] 3. Multi-source heterogeneous data fusion: Integrate multi-source heterogeneous data from physical entities, operating systems, sensors, etc., including simulation models, environmental data, physical object design data, maintenance data, operation data, etc. Through data fusion processing, provide rich data support for the digital twin model and drive the dynamic operation of various parts of the simulation model.
[0066] 4. Data-driven precise mapping: Based on twin data, the mapping between physical entity objects and digital world model objects is realized, including models, behavioral logic, business processes, and state changes caused by parameter adjustments. The state and behavior of physical entities are fully presented, accurately expressed, and dynamically monitored in the digital world.
[0067] 5. Intelligent analysis to assist decision-making: With rich historical and real-time data and advanced algorithm models, the state and behavior of physical objects can be efficiently reflected in the digital world. Through simulation tests and analytical predictions, it provides a basis for decision-making for issuing instructions for physical objects and further optimizing the process system, greatly improving the efficiency of analytical decision-making.
[0068] 2. Application functions:
[0069] 1. Energy monitoring and management: Real-time monitoring of the use of various energy sources such as coal, oil, natural gas, electricity, heat, and renewable energy in the region. Analyze and process energy data to achieve digital and visual management of energy supply and consumption. Provide functions such as energy consumption forecasting, energy load forecasting, and energy system adjustment to optimize energy efficiency.
[0070] 2. Carbon emission monitoring and management: real-time tracking and monitoring of carbon emissions, and management of various carbon emission calculation factors. Assist enterprises to improve their carbon asset management, carbon footprint management, carbon sink project management and other capabilities. Support enterprises to flexibly participate in carbon trading, submit application documents through the carbon emission management system, and achieve carbon emission reduction offset and trading.
[0071] 3. Energy efficiency optimization and emission reduction strategies: Based on digital twin models and data analysis, identify waste in energy use and unreasonable links in carbon emissions. Propose targeted energy efficiency optimization and emission reduction strategies to reduce energy consumption and carbon emission intensity. Evaluate and optimize the implementation effect of the strategy and continuously improve the level of energy and carbon management.
[0072] 4. Intelligent early warning and emergency response: The digital twin model monitors the state changes of the energy system and carbon emission process in real time. Once an abnormal situation or potential risk is found, the intelligent early warning mechanism is immediately triggered to notify relevant personnel to handle it. Emergency response plans and suggestions are provided to help enterprises quickly respond to emergencies and reduce losses and risks.
[0073] The comprehensive energy-carbon management application platform based on digital twin technology integrates a variety of advanced technologies and management models to achieve comprehensive monitoring, precise mapping, intelligent analysis and optimized management of energy and carbon emissions. The platform provides enterprises with effective energy-carbon management solutions, which helps reduce energy consumption and carbon emission intensity, and improve energy efficiency and environmental performance.
[0074] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.
Claims
1. Comprehensive energy and carbon management method based on digital twin technology, It is characterized in that include: A time series of energy usage and carbon emission data collected by a sensor network deployed in the physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels; integrating time series of said energy usage and carbon emissions data into the digital twin model; In the digital twin model, the time series of the energy use and carbon emission data are sequence-associated encoded to obtain a temperature value time series associated implicit feature vector, a pressure value time series associated implicit feature vector, a flow value time series associated implicit feature vector, an energy consumption value time series associated implicit feature vector, and an emission level time series associated implicit feature vector; In the digital twin model, the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector are input into the emission level reasoner to obtain the emission level time series reasoning representation vector; In the digital twin model, a detection result is determined based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection.
2. According to claim 1, the comprehensive energy and carbon management method based on digital twin technology, It is characterized in that In the digital twin model, the time series of the energy use and carbon emission data are sequence-associated encoded to obtain a temperature value time series associated implicit feature vector, a pressure value time series associated implicit feature vector, a flow value time series associated implicit feature vector, an energy consumption value time series associated implicit feature vector, and an emission level time series associated implicit feature vector, including: Regularizing the time series of the energy use and carbon emission data according to the parameter samples to obtain a time series of temperature values, a time series of pressure values, a time series of flow values, a time series of energy consumption values, and a time series of emission levels; The time series of the temperature values, the time series of the pressure values, the time series of the flow values, the time series of the energy consumption values and the time series of the emission levels are respectively input into the double-branch structure sequence encoder to obtain the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, the energy consumption value time series associated implicit feature vector and the emission level time series associated implicit feature vector.
3. According to claim 2, the comprehensive energy and carbon management method based on digital twin technology, It is characterized in that The time series of the temperature value, the time series of the pressure value, the time series of the flow value, the time series of the energy consumption value and the time series of the emission level are respectively input into a double-branch structure sequence encoder to obtain the time series associated implicit feature vector of the temperature value, the time series associated implicit feature vector of the pressure value, the time series associated implicit feature vector of the flow value, the time series associated implicit feature vector of the energy consumption value and the time series associated implicit feature vector of the emission level, including: The time series of the temperature values, the time series of the pressure values, the time series of the flow values, the time series of the energy consumption values and the time series of the emission levels are respectively input into a dual-branch structure sequence encoder including a 1DCNN model and a bidirectional gated recurrent unit to obtain the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, the energy consumption value time series associated implicit feature vector and the emission level time series associated implicit feature vector.
4. The comprehensive energy and carbon management method based on digital twin technology according to claim 3, It is characterized in that In the digital twin model, the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector are input into the emission level reasoner to obtain the emission level time series reasoning representation vector, including: The temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector are input into the emission level reasoner based on the deep belief network to obtain the emission level time series reasoning representation vector.
5. According to claim 4, the comprehensive energy and carbon management method based on digital twin technology, It is characterized in that In the digital twin model, based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, a detection result is determined, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection, including: Inputting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector into an associated interaction analysis network guided by a prior distribution to obtain an emission level reasoning-emission level detection interaction analysis representation vector; The emission level reasoning-emission level detection interaction analysis representation vector is input as the interaction feature into a classifier-based anomaly detector to obtain the detection result.
6. The comprehensive energy and carbon management method based on digital twin technology according to claim 5, It is characterized in that Inputting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector into an associated interactive analysis network guided by a prior distribution to obtain an emission level reasoning-emission level detection interactive analysis representation vector, including: Interacting the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector to obtain a first indexed emission level reasoning-emission level detection time series interaction representation vector and a second indexed emission level reasoning-emission level detection time series interaction representation vector; determining a first random coefficient and a second random coefficient; Based on the first random coefficient and the second random coefficient, the first indexed emission level reasoning-emission level detection timing interaction representation vector and the second indexed emission level reasoning-emission level detection timing interaction representation vector are associated and fused based on the guidance of prior distribution to obtain the emission level reasoning-emission level detection interaction analysis representation vector.
7. The comprehensive energy and carbon management method based on digital twin technology according to claim 6, It is characterized in that The emission level time series reasoning representation vector and the emission level time series associated implicit feature vector are interacted to obtain a first indexed emission level reasoning-emission level detection time series interaction representation vector and a second indexed emission level reasoning-emission level detection time series interaction representation vector, including: Calculate the positional division of the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, and multiply the vector obtained by the positional division by the first weight coefficient and the second weight coefficient respectively to obtain a first emission level reasoning-emission level detection time series interaction representation vector and a second emission level reasoning-emission level detection time series interaction representation vector; Calculate an exponential function value with a natural constant as a base and each eigenvalue of the first emission level reasoning-emission level detection time series interaction representation vector as an exponent to obtain the first indexed emission level reasoning-emission level detection time series interaction representation vector; An exponential function value is calculated with a natural constant as the base and each eigenvalue of the second emission level reasoning-emission level detection timing interaction representation vector as the exponent to obtain the second exponential emission level reasoning-emission level detection timing interaction representation vector.
8. According to the comprehensive energy and carbon management method based on digital twin technology according to claim 7, It is characterized in that Based on the first random coefficient and the second random coefficient, the first indexed emission level reasoning-emission level detection time series interaction representation vector and the second indexed emission level reasoning-emission level detection time series interaction representation vector are associated and fused based on prior distribution guidance to obtain the emission level reasoning-emission level detection interaction analysis representation vector, including: Calculate the position-weighted sum of the first indexed emission level reasoning-emission level detection timing interaction representation vector and the second indexed emission level reasoning-emission level detection timing interaction representation vector to obtain the emission level reasoning-emission level detection interaction analysis representation vector, wherein the weight of the first indexed emission level reasoning-emission level detection timing interaction representation vector is the first random coefficient, and the weight of the second indexed emission level reasoning-emission level detection timing interaction representation vector is the second random coefficient.
9. Comprehensive energy and carbon management system based on digital twin technology, It is characterized in that include: A data collection module, for collecting time series of energy usage and carbon emission data collected by a sensor network deployed in the physical system, wherein the energy usage and carbon emission data include temperature values, pressure values, flow values, energy consumption values and emission levels; A digital twin module for integrating the time series of energy usage and carbon emission data into a digital twin model; A sequence association coding module is used to perform sequence association coding on the time series of the energy use and carbon emission data in the digital twin model to obtain a temperature value time series association implicit feature vector, a pressure value time series association implicit feature vector, a flow value time series association implicit feature vector, an energy consumption value time series association implicit feature vector, and an emission level time series association implicit feature vector; The emission level time series reasoning representation module is used to input the temperature value time series associated implicit feature vector, the pressure value time series associated implicit feature vector, the flow value time series associated implicit feature vector, and the energy consumption value time series associated implicit feature vector into the emission level reasoner in the digital twin model to obtain the emission level time series reasoning representation vector.
10. The comprehensive energy and carbon management system based on digital twin technology according to claim 9, It is characterized in that It also includes a detection result determination module, which is used to determine the detection result in the digital twin model based on the interaction characteristics between the emission level time series reasoning representation vector and the emission level time series associated implicit feature vector, and the detection result is used to indicate whether there is an abnormality in the carbon emission level detection.
Citation Information
Patent Citations
Urban traffic carbon emission prediction method and device based on deep learning
CN117172138A
Carbon transaction management system based on industrial park
CN117455594A
Carbon emission detection and early warning method and system
CN118171771A
Network data flow anomaly detection system and method based on deep learning
CN118353678A
Regional carbon emission accounting system and method
CN118377787A
Cited By
Blast furnace carbon emission anomaly prediction method based on digital twinning
CN119623103A
Intelligent equipment operation and maintenance method and system based on railway operation and maintenance
CN120011832A
Factory carbon emission monitoring method and system based on electricity-carbon calculation model
CN120746030A
