An energy consumption intelligent online monitoring method and system

Abstract

The application discloses an energy consumption intelligent online monitoring method and system, and relates to the technical field of energy management. The method comprises the following steps: privacy protection is performed on a privacy gateway between each local node and a global overall node; a local energy knowledge graph is acquired, a local energy network topology is deployed and configured based on the local energy knowledge graph, and a local energy consumption monitoring model is constructed; a global energy knowledge graph is acquired, an intelligent contract is defined according to global operation monitoring indexes in combination with the local energy knowledge graph, a global energy consumption monitoring model is constructed; the global energy consumption monitoring model sends corresponding security aggregation requests to the local energy consumption monitoring models of the local nodes through the privacy gateway, and receives returned monitoring data; and the global energy consumption monitoring model evaluates energy efficiency health degrees according to the received monitoring data. The application aims to realize energy consumption intelligent management which strictly follows physical laws of an energy system and supports regional-level collaborative monitoring, accurate traceability and automatic optimization.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to an intelligent online monitoring method and system for energy consumption. Background Technology

[0002] Currently, the need for refined regional energy management is increasingly urgent. However, existing technological solutions have significant limitations. Traditional centralized monitoring methods require detailed operational data to be uploaded locally, raising concerns about the leakage of core local business secrets and production and operational privacy, resulting in high data silos and difficulty in obtaining a comprehensive perspective. On the other hand, current digital solutions mostly focus on energy visualization for individual enterprises, lacking cross-enterprise regional collaborative analysis capabilities. Even some advanced attempts to build regional energy consumption models often deviate from the operating laws of the physical world because the models are purely data-driven, leading to predictions that deviate from reality, poor interpretability, and an inability to effectively support accurate energy efficiency diagnosis and optimization decisions. Especially in the event of regional anomalies, how to quickly trace the source and coordinate a response while protecting the data privacy of all parties remains a challenge that existing technologies have not yet solved. Summary of the Invention

[0003] This invention provides an intelligent online monitoring method for energy consumption, comprising:

[0004] Configure a privacy gateway between each local node and the global coordinating node to protect privacy.

[0005] Obtain a local energy knowledge graph, deploy and configure the local energy network topology based on the local energy knowledge graph, and build a local energy consumption monitoring model;

[0006] Obtain a global energy knowledge graph, define smart contracts based on global operation monitoring indicators and local energy knowledge graph, and build a global energy consumption monitoring model;

[0007] The global energy consumption monitoring model sends corresponding security aggregation requests to the local energy consumption monitoring models of each local node through the privacy gateway, and receives the monitoring data returned by the local energy consumption monitoring models of each local node.

[0008] The global energy consumption monitoring model aggregates and assesses energy efficiency health based on the received monitoring data.

[0009] The aforementioned intelligent online energy consumption monitoring method acquires a local energy knowledge graph, deploys and configures a local energy network topology based on the local energy knowledge graph, and constructs a local energy consumption monitoring model, including:

[0010] Collect local multi-source data, establish multi-dimensional relationships based on physical connections, logical dependencies and energy flow, and construct a local energy knowledge graph;

[0011] Based on the local energy knowledge graph, a local energy digital twin model is deployed and an energy network topology and correlation model is configured to build an independent and refined local energy monitoring model.

[0012] The aforementioned intelligent online energy consumption monitoring method, based on a local energy knowledge graph, deploys a local energy digital twin model and configures an energy network topology and correlation model to construct an independent and refined local energy monitoring model, including:

[0013] Establish a high-precision energy digital twin to accurately map the local power distribution network, energy-consuming equipment, and the entire process.

[0014] This twin is used to conduct panoramic visual monitoring of local energy consumption at the equipment and production line levels, and to monitor and trace the internal causes of abnormal energy consumption events.

[0015] The aforementioned intelligent online energy consumption monitoring method acquires a global energy knowledge graph, defines smart contracts based on global operational monitoring indicators and the local energy knowledge graph, and constructs a global energy consumption monitoring model, including:

[0016] Collect multi-source static data and define the relationships between them to construct a global energy knowledge graph;

[0017] A global benchmark energy consumption model is trained by combining an energy knowledge graph with local operating data.

[0018] The aforementioned intelligent online monitoring method for energy consumption, which combines an energy knowledge graph and acquires local operating data to train a global benchmark energy consumption model, includes:

[0019] Based on the regional energy knowledge graph and local operation data, monitoring indicators are defined, and smart contract templates with executable monitoring rules are generated.

[0020] A global benchmark energy consumption model is trained based on a monitoring rule base and smart contract templates.

[0021] The aforementioned intelligent online energy consumption monitoring method, based on received monitoring data, aggregates and assesses energy efficiency health using a global energy consumption monitoring model, including:

[0022] The global benchmark energy consumption model periodically sends secure aggregation requests to the privacy gateway to obtain macro indicators. After aggregation, the energy efficiency health is assessed, and an energy efficiency benchmarking report and in-depth insights are generated.

[0023] When the regional total load reaches a critical point or local energy efficiency is abnormal, the model triggers a collaborative analysis between the local and regional levels.

[0024] The aforementioned intelligent online monitoring method for energy consumption, when the regional total load reaches a critical point or local energy efficiency is abnormal, triggers a collaborative analysis between the local and regional systems, including:

[0025] The relevant privacy gateway is triggered to calculate the local contribution and correlation of multiple parties and return encrypted intermediate results;

[0026] The decryption results locate the affected parties and trigger the smart contract, generate anomaly liability determination, and automatically carry out collaborative optimization actions.

[0027] An intelligent online monitoring system for energy consumption includes:

[0028] The privacy gateway building module is used to configure the privacy gateway between each local node and the global coordinating node for privacy protection;

[0029] The local energy consumption monitoring module is used to acquire the local energy knowledge graph, deploy and configure the local energy network topology based on the local energy knowledge graph, and build a local energy consumption monitoring model.

[0030] The global energy consumption monitoring module is used to obtain a global energy knowledge graph, define smart contracts based on global operation monitoring indicators and local energy knowledge graph, and build a global energy consumption monitoring model.

[0031] The aggregation monitoring module is used to send corresponding security aggregation requests from the global energy consumption monitoring model to the local energy consumption monitoring models of each local node through the privacy gateway, and to receive the monitoring data returned by the local energy consumption monitoring models of each local node.

[0032] The energy efficiency health assessment module is used by the global energy consumption monitoring model to aggregate and assess energy efficiency health based on the received monitoring data.

[0033] The beneficial effects achieved by this invention are as follows:

[0034] By employing a privacy-preserving computation gateway mechanism, secure aggregation of regional energy consumption and model training are achieved without the need for data collection, fundamentally resolving the conflict between data sharing and privacy protection. Deeply integrating knowledge graphs and differentiable physics engines into the model kernel enables the global benchmark model to not only possess high-precision prediction capabilities but also strictly adhere to physical conservation laws, resulting in highly reliable and insightful outputs. A complete closed-loop system has been constructed, encompassing micro-level monitoring, macro-level regional collaboration, and automatic anomaly optimization, significantly improving the intelligence level and operational efficiency of regional energy management. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0036] Figure 1This is a flowchart of an intelligent online monitoring method for energy consumption provided in Embodiment 1 of this application.

[0037] Figure 2 This is a schematic diagram of an intelligent online monitoring system for energy consumption provided in Embodiment 2 of this application. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example 1

[0040] like Figure 1 As shown, Embodiment 1 of this application provides an intelligent online monitoring method for energy consumption, including:

[0041] S1: Configure a privacy gateway between each local node and the global coordinating node for privacy protection;

[0042] Specifically, a privacy gateway is deployed at the network boundary between the local internal network and the external regional model, serving as the sole authorized data entry and exit point.

[0043] This gateway employs a two-way authentication mode. First, it generates unique quantum keys for both the local and regional models based on quantum encryption, ensuring the verifiability and trustworthiness of endpoint identities. Specifically, both the privacy gateway and the regional authentication center integrate quantum key distribution terminals. During key generation, the quantum light source at the authentication center first randomly generates a series of single-photon states, with each photon encoding one qubit of information. These single photons then flow to the quantum detectors of each privacy gateway. After successful transmission of the quantum sequence, a processing procedure is executed, including but not limited to key error correction and privacy amplification, to eliminate any small amounts of information that may have been leaked during transmission. Finally, a unique, absolutely random long-sequence symmetric key is extracted.

[0044] The core of this process lies in utilizing the indivisibility of a single photon and the quantum no-cloning theorem. Any eavesdropping will inevitably interfere with the quantum state of the photon, which can be detected by both parties in the post-processing step through bit error rate analysis. Once the bit error rate exceeds the security threshold, the key generation is terminated and an alarm is triggered, thus ensuring unconditional security of key distribution from a physical perspective.

[0045] For the energy consumption data to be transmitted, a national cryptographic algorithm suite is integrated. An algorithm suite that supports additive homomorphic characteristics is used for the original energy consumption data. All data payloads to be transmitted are encrypted end-to-end and sent out through a secure link to form a ciphertext transmission channel.

[0046] Specifically, after encrypting the plaintext data, a specific mathematical operation is performed on the resulting ciphertext. The decryption result is equivalent to performing the corresponding addition operation on the original plaintext. Specifically, the public key provided by the regional model is used to encrypt energy consumption readings at various time granularities locally, generating corresponding ciphertext. It transforms a single energy consumption value into an unreadable ciphertext token, but this token retains the aggregatable mathematical properties.

[0047] Meanwhile, the gateway's built-in differential engine injects calibrated noise that meets the preset privacy budget before data aggregation. Then, it performs deterministic computation in the ciphertext domain, performing a specified homomorphic addition operation on the encrypted energy consumption data and the encrypted noise to generate encrypted energy consumption data carrying noise, so that independent local real data cannot be inferred from the aggregation result.

[0048] S2: Obtain the local energy knowledge graph, deploy and configure the local energy network topology based on the local energy knowledge graph, and build a local energy consumption monitoring model;

[0049] S21: Collect local multi-source data, establish multi-dimensional relationships based on physical connections, logical dependencies and energy flow, and construct a local energy knowledge graph;

[0050] The system collects multi-source local data, covering power distribution facility attributes, energy-consuming equipment parameters, production line processes, and real-time operational data. Based on this, entity recognition and relationship extraction technologies are used to abstract elements such as equipment, instruments, and processes into knowledge nodes. Multi-dimensional relationships are then established based on physical connections, logical dependencies, and energy flow, forming a local energy knowledge graph with semantic reasoning capabilities. This graph structurally describes the static topological relationships at the equipment and production line levels within the energy network and deeply binds energy consumption data with entity behaviors through dynamic attribute mapping.

[0051] S22: Based on the local energy knowledge graph, deploy a local energy digital twin model and configure the energy network topology and association model to build a local independent and refined local energy monitoring model;

[0052] This includes deploying a local energy digital twin model based on an energy knowledge graph and configuring an energy network topology and correlation model to construct a local, independent, and refined local energy monitoring model, including:

[0053] S221: Establish a high-precision energy digital twin to accurately map the local power distribution network, energy-consuming equipment, and the entire process.

[0054] A three-dimensional high-precision model layer is constructed based on the local energy knowledge graph and real-time collected local operation data.

[0055] First, power distribution facilities and energy-consuming equipment are modeled to form a static mapping. Then, multimodal operational data is dynamically bound to the 3D model entities based on the relationships defined in the knowledge graph. Local operational data includes at least energy consumption data such as current, voltage, and power.

[0056] After dynamic binding, business logic is defined according to rules and states to simulate the entire dynamic process of local independent device operation and local energy flow. The business logic includes at least device start-up and shutdown, energy consumption conversion, etc.

[0057] S222: Utilize this twin to perform panoramic visualization monitoring of local energy consumption at the equipment and production line levels, and to monitor and trace the internal causes of abnormal energy consumption events.

[0058] Based on the energy digital twin and the topological relationships defined by the knowledge graph, energy consumption data at the equipment level and production line level are dynamically mapped into visual elements with spatial location and logical relationship.

[0059] Next, a visualized energy digital twin is used for anomaly monitoring. Specifically, the digital twin model incorporates a hybrid monitoring module that combines a rule engine with a lightweight time-series pattern recognition algorithm. By using real-time energy consumption time-series data provided by the twin model and a rule base, it quickly identifies obvious out-of-bounds behaviors while continuously learning the inherent energy consumption patterns of local operation.

[0060] If an anomaly is detected, the cause-finding reasoning chain is immediately initiated, using predefined device associations and other connections in the knowledge graph to trace the propagation path of the anomaly signal in the spatiotemporal dimensions. Then, the production work order and equipment status logs are linked, and the root cause, scope of impact, and details of the anomaly are highlighted in a visual interface.

[0061] S3: Obtain the global energy knowledge graph, define smart contracts based on global operation monitoring indicators and local energy knowledge graph, and build a global energy consumption monitoring model;

[0062] S31: Collect multi-source static data and define relationships to construct a global energy knowledge graph;

[0063] The system proactively scans and collects heterogeneous data sources from local information systems, including building information models, equipment asset databases, power distribution network topology diagrams, and process flow diagrams. Semantic modeling of entities and relationships is then performed, defining core associations such as "belongs to," "supplyes to," and "consumes energy," automatically mapping the collected heterogeneous data into a unified semantic network. Based on this, graph computing algorithms are used to automatically construct a global energy knowledge graph, with equipment nodes as vertices and energy flows and logical relationships as edges.

[0064] S32: Combine energy knowledge graphs and acquire local operating data to train a global benchmark energy consumption model;

[0065] This includes training a global benchmark energy consumption model by combining energy knowledge graphs and acquiring local operational data, including:

[0066] S321: Define monitoring indicators based on regional energy knowledge graph and local operation data, and generate smart contract templates with executable monitoring rules;

[0067] Based on the entity relationships and attributes in the regional energy knowledge graph, abstract indicators such as energy consumption and load factor are parsed into computable expressions using domain ontology. Then, business knowledge such as industry energy efficiency benchmarks is transformed into production rules. The antecedent of each rule is the context state based on the knowledge graph, and the consequent is the computational logic and control action to be triggered.

[0068] Based on this, the indicator calculation expression is dynamically bound to business rules, and automatically assembled into a complete, deployable smart contract template according to a predefined contract architecture. The predefined contract architecture includes, but is not limited to, participant roles, data input sources, conditional judgment logic, and execution functions.

[0069] S322: Training a global benchmark energy consumption model based on a monitoring rule base and smart contract template;

[0070] A global benchmark energy consumption model is trained based on a monitoring rule base and smart contract templates combined with a regional energy knowledge graph.

[0071] The global baseline energy consumption model uses a differentiable physical engine as its core, encoding physical rules such as grid topology and power balance constraints in the regional energy knowledge graph into a series of long, differentially grouped constraints. Then, a graph-structured network is used to extract the dynamic energy flow relationships between local nodes from the knowledge graph as the engine's initial state and boundary conditions.

[0072] Meanwhile, the time-series coding part processes historical aggregated energy consumption data and inputs its hidden state into a solver based on neural differential equations. Under the constraints of the physics engine, it performs forward simulation and deduction of the dynamics of the regional energy system, thereby combining the representation learning ability of neural networks with physical laws.

[0073] During model training, the regional center distributes model parameters, and each privacy gateway calculates the loss between the output and the true value locally using the privacy computing gateway. Simultaneously, it calculates an additional physics conservation loss term and combines this with the joint loss function to train the model to determine if the inference results satisfy fundamental physical laws. The formula for the joint loss function is as follows:

[0074] Indicates the model in time Predicted regional energy consumption values; Indicates a point in time Actual regional energy consumption measurements obtained through secure aggregation; S represents the time series length; It is a hyperparameter used to balance the weight of physical constraints in the total loss function and control the degree of influence of physical laws on model training. From regional energy knowledge graph The power flow tensor derived from it encodes the power grid topology and energy flow relationships; E represents the system energy state, which changes over time; This represents the rate of change of energy over time. This term is a physical conservation term, ensuring that the model output conforms to the physical conservation laws; It is a regularization weight hyperparameter that controls the strength of regularization; The parameter matrix represents the model, including the weights in the neural differential equation solver; The Frobenius norm is used to measure the complexity of the parameter matrix. Used to adjust the weight of the privacy robustness term in the loss function; For privacy noise, let represent the noise level from the normal distribution. Medium-sampled random noise; It is a data loss function that evaluates the predicted value of the noisy version. Deviation from the true value.

[0075] After aggregating encrypted energy consumption data, the total loss function simultaneously optimizes data fitting accuracy and physical consistency, making the model accurate while strictly adhering to the underlying laws of the energy network.

[0076] S4: The global energy consumption monitoring model sends corresponding security aggregation requests to the local energy consumption monitoring models of each local node through the privacy gateway, and receives the monitoring data returned by the local energy consumption monitoring models of each local node.

[0077] Based on the trained global baseline energy consumption model, a secure aggregation command is automatically triggered and transmitted to each local model via a privacy computing gateway. Its core lies in the use of a dynamic aggregation path selection algorithm, which intelligently plans the most efficient and secure data aggregation order based on the network topology and relationships defined in the knowledge graph.

[0078] After receiving the instruction, each privacy gateway calls the homomorphic encryption and differential privacy fusion processing module to encrypt and inject noise into the local energy consumption data at a specified time granularity, generating a data packet with "encryption-disturbance" dual protection.

[0079] Next, the data packets are hierarchically aggregated along a preset path. Intermediate nodes can only perform addition operations on the ciphertext but cannot decrypt it, ultimately forming an aggregated ciphertext representing the total local energy consumption at the regional level. After decryption using a private key, privacy-protected macroeconomic energy consumption data is obtained. Then, the aggregation result is correlated with contextual information such as model prediction bias and real-time weather events to dynamically generate new load pattern entities and their relationships with other nodes, thereby achieving intelligent evolution and updating of the regional energy knowledge graph under strict protection of individual privacy.

[0080] S5: The global energy consumption monitoring model aggregates and assesses energy efficiency health based on the received monitoring data.

[0081] S51: The global benchmark energy consumption model periodically sends security aggregation requests to each privacy gateway to obtain macro indicators. After aggregation, it assesses the energy efficiency health and generates an energy efficiency benchmarking report and in-depth insights.

[0082] Using the total load sequence obtained through real-time secure aggregation as the core input, an energy efficiency health assessment is conducted using a global benchmark energy consumption model. Specifically, by comparing the actual sequence with the self-generated benchmark prediction probability distribution, a health index reflecting the current energy efficiency status of the region is automatically calculated. The health index is automatically compared with historical benchmark values ​​and energy consumption target thresholds in the knowledge graph. After energy efficiency benchmarking, benchmarking results and in-depth insights are output, and specific information is visualized.

[0083] The core of generating the health index lies in quantifying the abnormal level of the current load by analyzing the deviation between real-time data points and the forecast distribution confidence interval, as well as the pattern of deviation within a continuous time window (e.g., consistently high, drastic fluctuations, etc.). The formula is as follows:

[0084] The wavelet basis function is represented by the scale parameter. With time translation parameters It is used to perform continuous wavelet transform on the regional total load sequence obtained by secure aggregation, and extract local features of the signal at multiple scales; Indicates the time shift parameter; Indicates the scale parameter; The power spectral density function represents the power density of the reference load sequence at frequency . The energy distribution at a given location reflects the predictive characteristics of the baseline model; It is the kernel function of the inverse Fourier transform, used to convert a frequency domain signal back to the time domain; Indicates the current point in time, at a specific time. The regional energy efficiency health status; Indicates the positive weighting coefficient; The normalized Laplace matrix represents the regional energy knowledge graph (representing the network topology in graph theory). It is its largest eigenvalue, reflecting the network's connectivity and stability; Represents the positive weighting coefficient, and Jointly optimize health sensitivity; N represents the total number of local models participating in the aggregation; This represents the load variance of the v-th model data, quantifying the total volatility and uncertainty of the system.

[0085] S52: When the regional total load is critical or the local energy efficiency is abnormal, the model triggers a local and regional collaborative analysis.

[0086] When the regional total load reaches a critical point or local energy efficiency is abnormal, the model triggers a joint analysis of local and regional data, including:

[0087] S521: Trigger the relevant privacy gateway to perform multi-party calculations of local contribution and correlation and return encrypted intermediate results;

[0088] The global baseline energy consumption model monitors in real time and triggers a collaborative analysis process when it identifies a critical regional total load or a local energy efficiency anomaly.

[0089] First, the relationships in the regional energy knowledge graph are transformed into a task topology graph through secure multi-party computation, determining the computational paths that each participant needs to interact with. Next, the secure multi-party computation protocol stack corresponding to the collaborative analysis instruction set is sent to the relevant privacy gateways, receiving contribution and correlation vectors calculated by each gateway.

[0090] In this process, each gateway performs two core calculations in parallel without exchanging raw data locally: first, it calculates its own contribution to macro-level anomaly indicators based on local historical energy consumption and production data; second, it calculates the correlation strength with anomaly events by simulating changes in the status of related devices through a local digital twin model. All calculations are performed in an encrypted state, and gateways only exchange necessary encrypted intermediate results (such as contribution and correlation vectors) according to the protocol.

[0091] S522: Decrypt the results to locate the affected party and trigger the smart contract, generate anomaly liability determination, and automatically carry out collaborative optimization actions.

[0092] Using encrypted intermediate results as direct input, the global baseline energy consumption model decrypts the aggregated encrypted results using a private key to obtain the quantitative contribution and statistical correlation indicators of each local model to abnormal events.

[0093] Next, semantic enhancement analysis is performed using the regional energy consumption knowledge graph. This involves associating the decrypted numerical indicators with enterprise attributes and energy consumption thresholds within the graph to accurately pinpoint the core influencing parties and their root causes of the anomalies. For example, it identifies enterprises whose energy consumption has increased due to a cascading effect caused by a critical equipment malfunction. Then, a smart contract matching the event type is automatically triggered, generating an energy efficiency anomaly warning evidence chain based on contribution ratios and automatically issuing collaborative optimization instructions.

[0094] Once the command is issued, an anomaly warning will be issued and complete information will be displayed in the regional visualization and local digital twin display, such as energy consumption details and fault details.

[0095] Using newly acquired monitoring data sequences obtained through continuous secure aggregation, along with the parameters of the current global baseline energy consumption model, as input, a privacy-preserving model iterative optimization process is initiated. Without centralizing the original data, model updates are performed through encrypted gradient exchange. Specifically, the regional center distributes existing model parameters to each privacy-preserving computation gateway; each gateway uses locally added, differentially privacy-processed time-series energy consumption data to calculate the gradient of the model's loss function locally, encrypts the gradient using homomorphic encryption, and then uploads it. The regional center aggregates all encrypted gradients and decrypts them, then updates the global model using an adaptive optimization algorithm. Its core lies in dynamically adjusting the weight of local nodes in the knowledge graph based on their freshness and correlation with recent anomalies, ensuring that model optimization focuses more on currently active or key data contributors.

[0096] The system uses energy efficiency anomaly warning evidence chains generated by collaborative causal analysis, macro-load pattern change trends discovered through continuous safety aggregation, and new equipment-level association rules reported by local digital twin models, as well as real-time monitoring data, as inputs to dynamically evolve the regional energy knowledge graph. The core lies in automatically identifying entity relationships hidden in new data that are not defined in the existing graph, based on association rule mining and graph structure similarity calculation. Specifically, the input collaborative analysis results are structurally parsed to extract new knowledge assertions such as "the correlation coefficient between compressor start-up and shutdown in region A and regional total load fluctuations increases" or "a certain type of production process adjustment is strongly correlated with energy efficiency anomalies during a specific period," and their significance is verified using statistical hypothesis testing. These verified new assertions are then semantically mapped and conflict detected with the existing graph. The graph structure consistency after embedding new relationships is calculated using a graph neural network to evaluate the overall quality of the update. Finally, an update operation is automatically performed, creating new virtual entities or establishing new relationship edges, recording version traceability information, and outputting an adaptively evolved regional energy consumption knowledge graph.

[0097] Example 2

[0098] like Figure 2 As shown, Embodiment 2 of this application provides an intelligent online monitoring method for energy consumption, including:

[0099] Privacy Gateway Module 21: Configure privacy gateways between local nodes and the global coordinating node for privacy protection;

[0100] Local energy consumption monitoring module 22: acquires local energy knowledge graph, deploys and configures local energy network topology based on local energy knowledge graph, and builds local energy consumption monitoring model;

[0101] Global Energy Consumption Monitoring Module 23: Obtain the global energy knowledge graph, define smart contracts based on global operation monitoring indicators and local energy knowledge graph, and construct a global energy consumption monitoring model;

[0102] Aggregation monitoring module 24: The global energy consumption monitoring model sends corresponding security aggregation requests to the local energy consumption monitoring models of each local node through the privacy gateway, and receives the monitoring data returned by the local energy consumption monitoring models of each local node.

[0103] Energy Efficiency Health Assessment Module 25: The global energy consumption monitoring model aggregates and assesses energy efficiency health based on the received monitoring data.

[0104] Corresponding to the above embodiments, the present invention provides a computer storage medium, including: at least one memory and at least one processor;

[0105] The memory is used to store one or more program instructions;

[0106] A processor is used to run one or more program instructions to execute an intelligent online monitoring method for energy consumption.

[0107] Corresponding to the above embodiments, this embodiment of the invention provides a computer-readable storage medium containing one or more program instructions, which are executed by a processor to provide an intelligent online monitoring method for energy consumption.

[0108] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions, which, when executed on a computer, cause the computer to perform the aforementioned intelligent online monitoring method for energy consumption.

[0109] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0110] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0111] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0112] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0113] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0114] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0115] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0116] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent online monitoring method for energy consumption, characterized in that, The method comprises the following steps: configuring privacy protection between each local node and the global unified node through a privacy gateway; obtaining a local energy knowledge graph, deploying and configuring a local energy network topology based on the local energy knowledge graph, and constructing a local energy consumption monitoring model; obtaining a global energy knowledge graph, defining an intelligent contract according to global operation monitoring indexes and the local energy knowledge graph, and constructing a global energy consumption monitoring model; sending a corresponding security aggregation request to the local energy consumption monitoring model of each local node through the privacy gateway by the global energy consumption monitoring model, and receiving monitoring data returned by the local energy consumption monitoring model of each local node; evaluating energy efficiency health degree according to the received monitoring data by the global energy consumption monitoring model; obtaining a global energy knowledge graph, defining an intelligent contract according to global operation monitoring indexes and the local energy knowledge graph, and constructing a global energy consumption monitoring model, comprising: collecting multi-source static data and defining a correlation relationship to construct a global energy knowledge graph; training a global benchmark energy consumption model in combination with the energy knowledge graph and local operation data; wherein training a global benchmark energy consumption model in combination with the energy knowledge graph and local operation data comprises: defining monitoring indexes based on the regional energy knowledge graph and local operation data to generate an intelligent contract template with executable monitoring rules; training a global benchmark energy consumption model based on a monitoring rule library and an intelligent contract template; The global benchmark energy consumption model takes a differentiable physical engine as the core, encodes the power grid topology and power balance constraint physical rules in the regional energy knowledge graph into a series of long differential group constraint conditions, then extracts the dynamic energy flow relationship between local nodes from the knowledge graph as the initial state and boundary conditions of the engine using a graph structure network; at the same time, the time series coding part processes the historical aggregated energy consumption data, inputs its hidden state into the solver based on the neural differential equation, and under the constraint of the physical engine, the dynamics of the regional energy system is forward simulated and deduced, thereby combining the representation learning ability of the neural network with the physical law; evaluating energy efficiency health degree according to the received monitoring data by the global energy consumption monitoring model, comprising: The global benchmark energy consumption model sends a security aggregation request to the local gateway to obtain macro indexes regularly, evaluates energy efficiency health degree after aggregation, generates an energy efficiency benchmark report and deep insight; When the regional total load is critical or the local energy efficiency is abnormal, the model triggers local and regional collaborative analysis; wherein when the regional total load is critical or the local energy efficiency is abnormal, the model triggers local and regional collaborative analysis, comprising: triggering relevant local privacy gateways to calculate local calculation contribution and correlation and return encrypted intermediate results; decrypting the results to locate the influencing parties and triggering the intelligent contract to generate abnormal responsibility definition and automatically perform collaborative optimization action; taking the encrypted intermediate results as direct input, the global benchmark energy consumption model decrypts the aggregated encrypted results using a private key to obtain the quantitative contribution of each local model to the abnormal event and statistical correlation indexes; then calling the regional energy consumption knowledge graph for semantic enhancement analysis, correlating the decrypted numerical indexes with enterprise attributes, energy consumption thresholds and other information in the graph for reasoning, and accurately locating the core influencing parties and the root cause of the abnormality.

2. The method according to claim 1, wherein, Obtaining a local energy knowledge graph, deploying and configuring a local energy network topology based on the local energy knowledge graph, and constructing a local energy consumption monitoring model, including: Collecting local internal multi-source data, establishing multi-dimensional association relationships according to physical connections, logical dependencies and energy flow directions, and constructing a local energy knowledge graph; Based on the local energy knowledge graph, deploying a local energy digital twin model and configuring an energy network topology and associated model, and constructing an independent and refined local energy monitoring model.

3. The method according to claim 2, wherein, Based on the local energy knowledge graph, deploying a local energy digital twin model and configuring an energy network topology and associated model, and constructing an independent and refined local energy monitoring model, including: Establishing a high-precision energy digital twin, accurately mapping the local internal power distribution network, energy-using equipment and entire process; Using the twin to conduct panoramic visual monitoring of local energy consumption at the device level and production line level, and monitoring and internal tracing of abnormal energy consumption events.

4. An energy consumption intelligent online monitoring system, which executes an energy consumption intelligent online monitoring method according to any one of claims 1-3, characterized in that, Including: A privacy gateway construction module for configuring a privacy gateway between each local node and a global overall node for privacy protection; A local energy consumption monitoring module for obtaining a local energy knowledge graph, deploying and configuring a local energy network topology based on the local energy knowledge graph, and constructing a local energy consumption monitoring model; A global energy consumption monitoring module for obtaining a global energy knowledge graph, defining an intelligent contract in combination with the local energy knowledge graph according to global operation monitoring indicators, and constructing a global energy consumption monitoring model; An aggregation monitoring module for sending a corresponding secure aggregation request to the local energy consumption monitoring model of each local node through the privacy gateway by the global energy consumption monitoring model, and receiving monitoring data returned by the local energy consumption monitoring model of each local node; An energy efficiency health assessment module for aggregating and assessing energy efficiency health degree according to the received monitoring data by the global energy consumption monitoring model.

5. A computer readable storage medium, characterized in that, One or more program instructions for executing an energy consumption intelligent online monitoring method according to any one of claims 1-3 by a processor.

Images