Method and system for monitoring carbon emission efficiency of green ecological energy

Abstract

The invention relates to a green ecological energy carbon emission efficiency monitoring method and system, and relates to the technical field of green energy monitoring and data processing, and the method comprises the steps: obtaining multi-source heterogeneous data of a green ecological energy system in a preset time window, and constructing a high-dimensional point cloud based on the data; applying a main filtering function to the high-dimensional point cloud, and processing the high-dimensional point cloud based on the main filtering function to generate a reference topology network representing the operation form of the energy system; and monitoring the carbon emission efficiency of the green ecological energy based on the reference topology network. The system operation mode can be revealed from the structural level, state drift can be found in time, a deep diagnosis basis is provided, the problem that in the prior art, accurate evaluation and root positioning are difficult to conduct on the complex operation mode is solved, the deep analysis and diagnosis capacity of complex system faults is enhanced, and the system reliability is improved. And a decision basis is provided for subsequent energy consumption structure optimization.

Description

Technical Field

[0001] This application relates to the field of green energy monitoring and data processing technology, and in particular to a method and system for monitoring the carbon emission efficiency of green ecological energy. Background Technology

[0002] Green and eco-friendly energy systems, such as microgrids integrating photovoltaics, wind power, energy storage, and diverse loads, exhibit high-dimensional, nonlinear, and strongly coupled characteristics in their operational data. Accurate monitoring of the carbon emission efficiency of these systems is fundamental to achieving their optimized operation and energy conservation and emission reduction goals.

[0003] Current technologies for monitoring green energy systems typically employ methods based on key performance indicator thresholds or statistical models. These methods simplify the system's operating status into several independent numerical indicators for analysis. However, this approach struggles to reveal the overall structural characteristics inherent in high-dimensional data when analyzing complex operating patterns formed by the interaction of multiple variables. Therefore, when a system exhibits specific operating modes under different energy combinations and load demands, current technologies have limitations in providing in-depth, structured assessments of the carbon emission efficiency of these modes.

[0004] Furthermore, existing monitoring technologies primarily identify anomalies by determining whether specific parameters exceed preset static limits. This method is not sensitive enough to gradual or structural shifts in system operating modes. The system state may have already begun to evolve towards an inefficient or unstable mode, but as long as individual indicators have not yet reached their thresholds, this potential morphological change cannot be detected in time. This results in insufficient ability to monitor the dynamic evolution of the system's operating state and identify structural changes in advance.

[0005] Meanwhile, when existing monitoring methods detect performance degradation or anomalies, they typically only output a general result or alarm, failing to provide specific clues about the root cause of the problem. In a complex energy system where units are tightly coupled, this lack of in-depth diagnostic information makes it difficult for maintenance personnel to quickly locate the specific physical components or parameters causing performance degradation, thus affecting the efficiency of troubleshooting and system optimization. Summary of the Invention

[0006] The technical problem this application aims to solve is that existing green energy carbon emission monitoring methods are difficult to reveal the inherent structural characteristics of system operation when processing high-dimensional, nonlinear system data, and have limitations in their ability to identify and diagnose unknown abnormal patterns.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution.

[0008] The first aspect of this invention provides a method for monitoring the carbon emission efficiency of green ecological energy, the method comprising: Acquire multi-source heterogeneous data of a green ecological energy system within a preset time window, and construct a high-dimensional point cloud based on the multi-source heterogeneous data; Apply the master filter function to the high-dimensional point cloud; The high-dimensional point cloud is processed based on the main filtering function to generate a benchmark topology network that characterizes the operating mode of the energy system. The carbon emission efficiency of the green ecological energy is monitored based on the aforementioned benchmark topology network.

[0009] In an optional implementation, the step of constructing the high-dimensional point cloud includes combining N standardized measurements into an N-dimensional state vector v(t) at each time t within the preset time window: v(t)=[x′1(t),x′2(t),…,x′ N (t)]; Where v(t) is the state vector at time t, and x′ i (t) represents the i-th standardized measurement value; Furthermore, the high-dimensional point cloud is a set of all the state vectors within the preset time window.

[0010] In an optional implementation, the step of generating the baseline topology network includes: applying the master filtering function to the high-dimensional point cloud to obtain a set of filtered values, and covering the range of the filtered values ​​with multiple overlapping intervals; clustering each interval into a subset of the preimage in the high-dimensional point cloud to obtain multiple clusters; using the multiple clusters as nodes, and connecting an edge between the corresponding nodes if any two clusters share at least one common data point, thereby constructing the baseline topology network.

[0011] In an optional implementation, the method further includes: generating a series of high-dimensional point clouds for multiple consecutive time windows, and computing a persistent graph corresponding to the topology of each high-dimensional point cloud; calculating the distance between the persistent graphs of two adjacent time windows to obtain a topology drift index to quantify the dynamic changes in the operating form.

[0012] In an optional implementation, the topology drift index is determined by the following formula: TDI k =W q (D k-1 D k ); Where k is the index of the current time window, TDI k Let D be the current topological drift index.k and D k-1 W represents the persistence charts for the current and previous time windows, respectively. q The q-order Wasserstein distance is used to calculate the distance between the persistent graphs.

[0013] In an optional implementation, the method further includes: triggering an adaptive multi-view diagnostic process when the topology drift index exceeds a preset drift threshold.

[0014] In one optional implementation, the adaptive multi-view diagnostic process includes: selecting multiple diagnostic filtering functions from a preset filtering function library; and using the multiple diagnostic filtering functions to generate multiple diagnostic topology networks in parallel for the same high-dimensional point cloud that triggers the adaptive multi-view diagnostic process.

[0015] In an optional implementation, the adaptive multi-view diagnostic process further includes: comparing the structural differences among the plurality of diagnostic topologies, identifying the diagnostic topology with the most significant structural changes, to indicate the root cause of anomalies associated with the diagnostic filtering function corresponding to that diagnostic topology.

[0016] In an optional implementation, the diagnostic filtering functions in the filtering function library are selected from at least one of the following: grid voltage imbalance, energy storage system charge and discharge rate, or renewable energy penetration volatility.

[0017] In one optional implementation, the multi-source heterogeneous data includes at least one of the following: energy production data, energy consumption data, energy storage data, environmental parameters, or power grid interaction data.

[0018] A second aspect of this invention provides a monitoring system for the carbon emission efficiency of green ecological energy, the system comprising: The data acquisition and point cloud construction module is used to acquire multi-source heterogeneous data of a green ecological energy system within a preset time window, and construct a high-dimensional point cloud based on the multi-source heterogeneous data. The benchmark topology network generation module is used to apply a master filter function to the high-dimensional point cloud and process the high-dimensional point cloud based on the master filter function to generate a benchmark topology network that characterizes the operating mode of the energy system. The efficiency monitoring module is used to monitor the carbon emission efficiency of the green ecological energy based on the benchmark topology network.

[0019] In summary, this application includes at least one of the following beneficial technical effects: 1. The monitoring method provided by this invention constructs a high-dimensional point cloud from multi-source heterogeneous data and generates a baseline topology network characterizing the system's operating mode using a master filter function, thereby revealing the internal geometric structure and connectivity of the high-dimensional data. Compared with existing methods that rely solely on numerical statistics or function fitting, this invention can characterize and distinguish different energy consumption and production modes at the structural level, providing a deeper analytical dimension for carbon emission efficiency assessment and solving the problem that existing technologies struggle to accurately assess specific operating modes. 2. This invention calculates the persistence graph of a high-dimensional point cloud within a continuous time window and quantifies the distances between them to obtain the topological drift index, enabling the monitoring of the dynamic evolution rate of the system's operating mode. Compared to traditional methods that rely on whether specific values ​​exceed static thresholds, this method can identify state drift in the early stages of structural changes in the system's operating mode, thereby improving the timeliness of monitoring and the responsiveness to dynamic changes in the system, and overcoming the limitations of existing technologies in terms of accuracy and real-time performance. 3. This invention, through an adaptive multi-perspective diagnostic process, enables parallel analysis of the same data using multiple diagnostic filtering functions from a filter function library after detecting significant topology drift. By comparing structural changes in the diagnostic topology network from different perspectives, potential root causes leading to system instability can be identified. It provides specific directions for fault diagnosis, enhancing the ability to deeply analyze and diagnose complex system faults compared to existing technologies that only output single performance results, thus providing a basis for subsequent energy consumption structure optimization. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method used in this application; Figure 2 This is the system architecture diagram of this application. Detailed Implementation

[0021] The following is in conjunction with the appendix Figure 1 This application will be described in further detail below.

[0022] Example 1: A method for monitoring the carbon emission efficiency of green ecological energy, comprising the following steps: S1. Acquire multi-source heterogeneous data of the green ecological energy system within a preset time window, and construct a high-dimensional point cloud based on the multi-source heterogeneous data; This step aims to transform raw, multi-source measurement data from green ecological energy systems into a unified mathematical object suitable for subsequent topology analysis: a high-dimensional point cloud.

[0023] Specifically, this step first involves synchronously collecting or acquiring timestamped data from multiple data sources within the green eco-energy system to form a multi-source heterogeneous dataset. Data sources may include, for example, energy production units, energy consumption units, energy storage units, and environmental monitoring units.

[0024] Furthermore, energy production data may include the output power of photovoltaic arrays, the speed and power of wind turbines, etc.; energy consumption data may include industrial load, residential electricity consumption, charging pile power, etc.; energy storage data may include the state of charge (SOC) and charging / discharging current of energy storage batteries, etc.; environmental parameters may include light intensity, wind speed, ambient temperature, etc.

[0025] Because the raw data collected directly has different physical dimensions and numerical ranges, it is impossible to directly perform distance comparisons and structural analyses in geometric space. Therefore, it is necessary to standardize the collected data from each dimension to construct a unified, dimensionless state space.

[0026] In one specific implementation, the Z-score normalization method is used to process the measurements for each dimension. First, the historical mean and standard deviation of the time series of each measurement are calculated. Then, for the i-th raw measurement x at time t... i (t) Perform the following calculation to obtain the standardized measurement value x′. i (t): in: x i (t) represents the i-th original measurement value at time t; μ i The historical mean of the time series to which the i-th measurement belongs; σ i The historical standard deviation of the time series to which the i-th measurement belongs; x′ i (t) represents the i-th measurement value after standardization.

[0027] After standardization, all N standardized measurements of v(t) at the same time are combined into an N-dimensional state vector v(t). This vector resides in N-dimensional Euclidean space. The state vector v(t) represents the complete operating state of the energy system at that instant. The state vector v(t) is constructed as follows: v(t)=[x′1(t),x′2(t),…,x′ N (t)]; To analyze the dynamic behavior of the system over a period of time rather than isolated instantaneous states, this step further defines a rolling time window with a preset time length W. Within the k-th time window, all state vectors v(t) contained in that period are aggregated to construct a high-dimensional point cloud P. k .

[0028] High-dimensional point cloud P k It is a set of state vectors, defined as follows: P k ={v(t)|t∈[t]} start ,t end ]}; Among them, t start and t end Let t be the start and end times of the k-th time window, respectively, and t be the start and end times of the k-th time window. end -t start =W.

[0029] This step, which contains all the operational status information of the system over a period of time, will serve as the direct input for subsequent steps S2 and S3 to extract its inherent topological features.

[0030] S2. Apply the master filter function to the high-dimensional point cloud; The purpose of this step is to calculate a corresponding low-dimensional value that represents a specific physical meaning for each N-dimensional state vector in the high-dimensional point cloud. This process is achieved by applying a predefined mathematical function, namely the master filter function.

[0031] The main filter function is a function that transforms an N-dimensional state space... Mapped to one-dimensional real space The function is a continuous function. The selection of this function is determined based on the monitoring target of this invention, and its output value is used for the structured segmentation and organization of the high-dimensional point cloud in the subsequent step S3.

[0032] In one specific implementation, the main filter function f p It is set as an indicator that can directly reflect the carbon emission efficiency of the system. For example, the main filter function f p It can be a function of instantaneous carbon emission efficiency.

[0033] The instantaneous carbon emission efficiency function takes an N-dimensional state vector v(t) as input and calculates a scalar value representing the relationship between total energy output and carbon emissions at the current moment. Its specific calculation method can be given by the following formula: in: f p,eff(v(t)) is the instantaneous carbon emission efficiency filtering value corresponding to the state vector v(t); This is the set of indices for all energy production units in the system. P i (v(t)) represents the output power of the i-th energy production unit extracted from the state vector v(t); This is a set of indices for all units in the system that directly or indirectly generate carbon emissions. E j (v(t)) is the carbon emission proxy value of the j-th unit extracted from the state vector v(t); C0 is a preset constant bias term used to avoid the denominator being zero.

[0034] In another alternative implementation, the main filter function f p It can be set as an energy imbalance function, which measures the degree of coordination between energy generation, consumption and storage within the system, and indirectly reflects the operating efficiency of the system.

[0035] The energy imbalance function also receives the state vector v(t) as input, and its calculation method can be given by the following formula: in: f p,imb (v(t)) is the energy imbalance filter value corresponding to the state vector v(t); This is the set of indices for all energy production units in the system. P i (v(t)) represents the output power of the energy production unit with index i extracted from the state vector v(t); This is a set of indexes for all energy-consuming loads in the system. L j (v(t)) represents the power consumption of the load with index j extracted from the state vector v(t); P S (v(t)) represents the net power of the energy storage system extracted from the state vector v(t).

[0036] After executing step S2, the high-dimensional point cloud P k Each N-dimensional state vector v(t) in the dataset receives a corresponding scalar filter value. This generates a set of filter values. This set of filter values ​​is then compared with the original high-dimensional point cloud P. k Together, they serve as input data for executing step S3.

[0037] S3. Based on the main filtering function, the high-dimensional point cloud is processed to generate a benchmark topology network that represents the operating mode of the energy system. The purpose of this step is to transform the discrete high-dimensional point cloud and its one-dimensional projection (i.e., the set of filtered values) into a structured, low-dimensional graph theory object, namely, a baseline topological network. This network can intuitively represent the intrinsic operational form of the energy system within this time window.

[0038] Specifically, this step involves applying a topological data analysis algorithm, exemplified by the Mapper algorithm, to analyze the high-dimensional point cloud P. k The filtered values ​​are then processed. This process includes the following sequential sub-steps.

[0039] First, the set of filtered values ​​generated in step S2 is overlaid. This process aims to divide the one-dimensional range of filtered values ​​into manageable segments for local analysis. In one implementation, the maximum and minimum values ​​of the set of filtered values ​​are determined, and a set of intervals with overlapping portions is used. To cover the entire value range.

[0040] Set of intervals It consists of m intervals, that is The generation of intervals is controlled by two preset parameters: the number of intervals m and the overlap rate o between adjacent intervals. The overlap setting ensures that in subsequent clustering, data points near the interval boundaries can be included by two or more intervals simultaneously, thus providing a basis for building network connections.

[0041] Secondly, for each interval Local clustering is performed on a subset of the preimage. A subset of the preimage refers to the high-dimensional point cloud P. k In the middle, its filtered value falls within a specific interval I j The set of all state vectors within a subset P. k,j The definition is as follows: in: P k,j This is the subset of the preimage corresponding to the j-th interval; f p This refers to the main filter function applied in step S2; I j Let j be the coverage interval; v represents the high-dimensional point cloud P. k Any state vector in; P k This is the high-dimensional point cloud for the current time window.

[0042] Subsequently, for each preimage subset P k,jA clustering algorithm can be applied independently. For example, the DBSCAN (density-based noisy spatial clustering application) algorithm can be used because it does not require pre-specifying the number of clusters. This algorithm will cluster each subset P... k,j Decompose into one or more numbers Clusters with densely packed locations are denoted as {C} j,1 C j,2 ,…}.

[0043] Finally, a baseline topology network G is constructed based on all the aforementioned clusters. k In this network, each cluster C obtained through clustering is abstracted as a network node. The connection between two nodes in the network, i.e., the edge, is determined based on whether their corresponding clusters share common data points.

[0044] Specifically, the baseline topology network G k =(V k ,ε k ), its node set V k and edge set ε k The definition is as follows: V k For the baseline topology network G k The set of nodes; m is the total number of coverage areas; C j,s This represents the subset P of the j-th preimage. k,j The generated s-th cluster.

[0045] in: ε k For the baseline topology network G k The set of edges; C a C b For node set V k Any two distinct clusters in; This indicates that two clusters, which are sets of state vectors, share at least one common state vector. This connectivity exists due to the coverage interval. This is due to the overlapping characteristics.

[0046] After completing step S3, the output baseline topology network G k The geometry of a high-dimensional point cloud is compressed and displayed in graph form. The nodes of this network represent dense regions of system operating states, while the edges represent continuous transitions between these state regions. This network will serve as input to step S4 for monitoring carbon emission efficiency.

[0047] S4. Based on the baseline topology network, monitor the carbon emission efficiency of green ecological energy.

[0048] This step aims to continuously and quantitatively monitor and diagnose the carbon emission efficiency of green energy based on the baseline topology network and its derived dynamic characteristics.

[0049] In one specific implementation, this step not only analyzes the static baseline topology network within a single time window, but also establishes a dynamic monitoring process to quantify the evolution of the system's operational state over time. This process aims to identify structural transitions in the system state from one stable region to another.

[0050] To achieve dynamic monitoring, this method targets a series of high-dimensional point clouds {…,P} generated from multiple consecutive time windows. k-1 ,P k The process involves processing each high-dimensional point cloud P. k Persistent cohomology calculations are applied to generate a corresponding persistence graph D. k .

[0051] Persistence Chart D k It is a two-dimensional scatter plot, where each point (b, d) represents a point in a high-dimensional point cloud P. k In this context, topological features are generated (born) when the scale parameter is b and disappear when the scale parameter is d. These features include zero-dimensional connected components and one-dimensional holes, etc.

[0052] Subsequently, to quantify the magnitude of structural changes in the system's operational behavior between two adjacent time windows (e.g., the (k-1)th and the kth windows), their corresponding persistence graphs D are calculated. k-1 and D k The distance between them. This distance is defined as the Topological Drift Index (TDI). k .

[0053] In one specific implementation, the Topology Drift Index (TDI) k By calculating the q-order Wasserstein distance W between the two persistent graphs q The formula for determining this is as follows: in: TDI k Let be the topological drift index for the k-th time window; D k With D k-1 These are persistent charts for the current and previous time windows, respectively. W qFor a Wasserstein distance operator of order q, where q is a preset integer, which can be 1 or 2 for example; η is a value from D k-1 Point D k Optimal bijective matching of points in; p is D k-1 Any point in it; ||·|| ∞ It is an infinite norm used to calculate the distance between matching point pairs.

[0054] The calculated topological drift index TDI k The sequence forms a time series curve. This method further sets a configurable drift threshold θ. TDI When the latest Topology Drift Index (TDI) is detected... k When this threshold is exceeded, i.e., TDI k >θ TDI The system determines that the operating mode of the energy system has undergone significant structural changes and automatically triggers an adaptive multi-perspective diagnostic process.

[0055] The adaptive multi-view diagnostic workflow aims to perform root cause analysis on identified significant changes. The workflow begins with a pre-built library of filtering functions F = {f1, f2, ..., f...}. M Multiple diagnostic filter functions are selected from}. Each diagnostic filter function f j Each corresponds to a specific physical diagnostic perspective, such as grid voltage imbalance or energy storage system degradation rate.

[0056] Subsequently, the process uses multiple diagnostic filtering functions to analyze the same high-dimensional point cloud P that triggered the diagnosis. k The processing steps in step S3 are executed in parallel and repeatedly. This process will generate M different diagnostic topologies {G}. k,1 G k,2 ,…,G k,M Each network is a structured representation of the system state from a specific diagnostic perspective.

[0057] Finally, by comparing the structural properties of these M diagnostic topologies, such as the number of nodes, the number of edges, or the differences in the graph Laplace spectrum, the diagnostic topology with the most drastic structural changes is identified. Diagnostic filtering function corresponding to this network The physical dimension represented is indicated as the potential root cause of significant drift in the system state, thereby enabling the monitoring and diagnosis of anomalies related to carbon emission efficiency.

[0058] Combined with appendix Figure 2 Another embodiment of this application provides a monitoring system for the carbon emission efficiency of green ecological energy, comprising: The data acquisition and point cloud construction module is used to acquire multi-source heterogeneous data of a green ecological energy system within a preset time window, and construct a high-dimensional point cloud based on the multi-source heterogeneous data. The benchmark topology network generation module is used to apply a master filter function to a high-dimensional point cloud and process the high-dimensional point cloud based on the master filter function to generate a benchmark topology network that represents the operating mode of the energy system. The efficiency monitoring module is used to monitor the carbon emission efficiency of green ecological energy based on a benchmark topology network.

[0059] The system in this embodiment can be used to execute the above method embodiments, and its principle and technical effect are similar, so they will not be described again here.

[0060] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Identical components are represented by the same reference numerals. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for monitoring the carbon emission efficiency of green ecological energy, characterized in that, Includes the following steps: S1. Acquire multi-source heterogeneous data of the green ecological energy system within a preset time window, and construct a high-dimensional point cloud based on the multi-source heterogeneous data; S2. Apply the master filter function to the high-dimensional point cloud; S3. Process the high-dimensional point cloud based on the main filtering function to generate a benchmark topology network that characterizes the operating mode of the energy system; S4. Based on the aforementioned baseline topology network, monitor the carbon emission efficiency of the green ecological energy.

2. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 1, characterized in that, The steps for constructing a high-dimensional point cloud include: At each time t within the preset time window, the N standardized measurements are combined into an N-dimensional state vector v(t): v(t)=[x ′ 1(t),x ′ 2(t),…,x ′ N (t)]; Where v(t) is the state vector at time t, x i ′ (t) represents the i-th standardized measurement value; Furthermore, the high-dimensional point cloud is a set of all the state vectors within the preset time window.

3. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 1, characterized in that, The steps for generating the baseline topology network include: The main filtering function is applied to the high-dimensional point cloud to obtain a set of filtered values, and the range of the filtered values ​​is covered by multiple overlapping intervals; Clustering is performed on the preimage subset of each interval in the high-dimensional point cloud to obtain multiple clusters; Using the multiple clusters as nodes, if any two clusters share at least one common data point, then an edge is connected between the corresponding nodes to construct the baseline topology network.

4. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 1, characterized in that, Also includes: A series of high-dimensional point clouds are generated for multiple consecutive time windows, and a persistent graph corresponding to the topology of each high-dimensional point cloud is generated. The distance between the persistence charts of two adjacent time windows is calculated to obtain the topology drift index, which quantifies the dynamic changes in the operating pattern.

5. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 4, characterized in that, The topology drift index is determined by the following formula: TDI k =W q (D k-1 ,D k ); Where k is the index of the current time window, TDI k Let D be the current topological drift index. k and D k-1 W represents the persistence charts for the current and previous time windows, respectively. q The q-order Wasserstein distance is used to calculate the distance between the persistent graphs.

6. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 1, characterized in that, Also includes: When the topology drift index exceeds a preset drift threshold, an adaptive multi-view diagnostic process is triggered.

7. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 6, characterized in that, The adaptive multi-view diagnostic process includes: Select multiple diagnostic filter functions from the preset filter function library; Using the aforementioned diagnostic filtering functions, multiple diagnostic topology networks are generated in parallel for the same high-dimensional point cloud that triggers the adaptive multi-view diagnostic process.

8. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 7, characterized in that, The adaptive multi-view diagnostic process also includes: By comparing the structural differences among the multiple diagnostic topologies, the diagnostic topology with the most significant structural changes is identified to indicate the root cause of anomalies associated with the diagnostic filter function corresponding to that topology.

9. The method for monitoring the carbon emission efficiency of green ecological energy according to claim 1, characterized in that, The multi-source heterogeneous data includes at least one of the following: energy production data, energy consumption data, energy storage data, environmental parameters, or power grid interaction data.

10. A monitoring system for the carbon emission efficiency of green ecological energy, comprising a monitoring method for the carbon emission efficiency of green ecological energy according to any one of claims 1-9, characterized in that, include: The data acquisition and point cloud construction module is used to acquire multi-source heterogeneous data of the green ecological energy system within a preset time window, and construct a high-dimensional point cloud based on the multi-source heterogeneous data; The benchmark topology network generation module is used to apply a master filtering function to the high-dimensional point cloud and process the high-dimensional point cloud based on the master filtering function to generate a benchmark topology network that characterizes the operating mode of the energy system. The efficiency monitoring module is used to monitor the carbon emission efficiency of the green ecological energy based on the benchmark topology network.

Images