Energy-saving and carbon-reducing optimization method and system for low-carbon park

Abstract

The invention relates to the technical field of smart energy low-carbon management, and discloses an energy-saving and carbon-reducing optimization method and system for a low-carbon park. The method comprises the following steps: defining and constructing a hierarchical control architecture consisting of a park level, a region level and an equipment level according to a park energy supply network; and deploying a feature extraction node at each level to sense the energy flow and carbon emission state of the level. And synchronously collecting operation characterization data of each hierarchy, and carrying out cross-hierarchy association mapping so as to form a space-time association map for describing the whole energy and carbon flow of the park. Based on the map, a key energy-carbon coupling node and a potential energy-carbon imbalance link in an energy flow path are identified. And starting an adaptive optimization process of the park energy system for the identified potential energy-carbon imbalance link. According to the invention, systematic panoramic insight and accurate optimization positioning of energy and carbon flow in the park are realized, and the collaboration and effectiveness of energy saving and carbon reduction work are improved.

Description

Technical Field

[0001] This invention relates to the field of smart energy and low-carbon management technology, specifically to energy-saving and carbon-reduction optimization methods and systems for low-carbon industrial parks. Background Technology

[0002] Currently, low-carbon industrial parks generally rely on building layered energy monitoring systems to achieve their energy conservation and carbon reduction goals. These systems typically deploy monitoring nodes at multiple levels, including the park itself, specific areas, and individual equipment, collecting operational data such as energy consumption, power output, and production volume. Based on this data, common optimization strategies involve analyzing the independent indicators at each level or evaluating the aggregated overall data to develop energy-saving measures for identified high-energy-consuming equipment or areas, or to schedule energy supply.

[0003] The aforementioned conventional technical solutions have shortcomings. Layered monitoring can only provide isolated, localized data snapshots, lacking effective correlation and fusion between data from different levels. This prevents management from accurately depicting the dynamic flow paths and transformation relationships of energy and carbon emissions within the complex network of the entire industrial park, making it difficult to understand the synergistic and coupling mechanisms of energy and carbon flows from a holistic system perspective. Due to the inability to perceive the inherent interconnectedness of the system, optimization measures often target surface phenomena or end-user equipment, failing to accurately pinpoint the pivotal links that have a crucial impact on overall energy efficiency and carbon emissions, as well as potential structural bottlenecks where energy and carbon flows grow asynchronously and uncoordinated. Optimization efforts are prone to becoming localized and reactive, making it difficult to achieve systematic and preventative coordinated carbon reduction. Summary of the Invention

[0004] The purpose of this invention is to provide an energy-saving and carbon-reduction optimization method and system for low-carbon industrial parks to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an energy-saving and carbon-reduction optimization method for low-carbon industrial parks, the method comprising: Based on the energy supply network of the low-carbon park, a hierarchical control architecture for the park's energy system is defined and constructed, which consists of park-level, regional-level, and equipment-level control architectures. At each level of the hierarchical control architecture, feature extraction nodes are deployed to sense the energy flow and carbon emission status of that level. The feature extraction nodes synchronously collect operational characterization data at the corresponding levels, and perform cross-level correlation mapping on the operational characterization data obtained from different levels of the hierarchical control architecture to form a spatiotemporal correlation map describing the overall energy and carbon flow of the park. Based on the constructed spatiotemporal correlation map, key energy-carbon coupling nodes and potential energy-carbon imbalance links in the energy flow path of the park are identified. For the identified potential energy-carbon imbalance links, an adaptive optimization process for the park's energy system is initiated.

[0006] Preferably, a hierarchical control architecture for the park's energy system is defined and constructed. This hierarchical control architecture comprises park-level, regional-level, and equipment-level components, specifically including: The park-level unit corresponds to the total energy inlet, total outlet, and core conversion facilities of the entire low-carbon park, and is responsible for formulating energy dispatch strategies and total carbon emission targets. The regional level corresponds to sub-regions within the low-carbon park that have independent functions or physical boundaries. It receives and executes park-level control instructions and manages energy-consuming units within the sub-regions. The equipment level corresponds to the specific energy-consuming equipment, production-producing equipment, and energy storage equipment under the regional level, serving as the final execution unit for energy consumption and conversion. Establish two-way data reporting and command issuance channels between the park level, regional level, and equipment level.

[0007] Preferably, at each level of the hierarchical control architecture, feature extraction nodes are deployed to sense the energy flow and carbon emission status of that level, specifically including: The feature extraction node in the park-level deployment center is used to sense the inlet parameters of the park's total power grid, total gas grid, total heating grid, and total emission outlet monitoring parameters; Distributed feature extraction nodes are deployed at the regional level to sense the energy distribution hubs, internal microgrid operation status, and regional boundary carbon emission data of the corresponding sub-regions. Deploy terminal feature extraction nodes at the device level to perceive the real-time power, operating mode, energy efficiency level, and operating emission factors of a single device; The central feature extraction node, distributed feature extraction nodes, and terminal feature extraction nodes are logically connected according to the hierarchical relationship of the layered control architecture.

[0008] Preferably, the feature extraction node synchronously collects operational characterization data at the corresponding level, and performs cross-level correlation mapping on the operational characterization data obtained from different levels of the hierarchical control architecture to form a spatiotemporal correlation map describing the overall energy and carbon flow of the park, specifically including: The operational characterization data includes total energy input, energy conversion efficiency, load dynamics curve, and direct carbon emission intensity. Driven by a unified timestamp, the data acquisition actions of the central feature extraction node, distributed feature extraction nodes and terminal feature extraction nodes are triggered synchronously. The total energy input data of the park collected by the central feature extraction node is correlated with the energy receiving data of the sub-regions collected by the relevant distributed feature extraction nodes to determine the flow direction. The sub-region load data collected by the distributed feature extraction nodes is aggregated, verified, and traced back to its source with the device-level power data collected by all terminal feature extraction nodes within its jurisdiction. The equipment emission factor data collected by the terminal feature extraction node is aggregated upwards level by level according to the energy flow path and cross-validated with regional and park-level carbon emission monitoring data. Using energy flow and carbon flow paths as edges and feature extraction nodes and key facilities at each level as vertices, a spatiotemporal correlation graph with time attributes is constructed.

[0009] Preferably, based on the constructed spatiotemporal correlation map, key energy-carbon coupling nodes and potential energy-carbon imbalance links in the energy flow path of the park are identified, specifically including: Traverse all vertices in the spatiotemporal correlation graph and calculate the correlation between the energy flux intensity and the carbon flux intensity at each vertex; Vertices with a correlation degree exceeding a preset threshold are identified and marked as key energy carbon coupling nodes; Traverse all edges in the spatiotemporal correlation graph and analyze the synchronicity of energy flow and carbon flow carried on each edge; Identify the edges where the trends of energy flow and carbon flow continue to diverge, and mark them as potential energy-carbon imbalance links.

[0010] Preferably, for the identified potential energy-carbon imbalance links, an adaptive optimization process for the park's energy system is initiated, specifically including: Extract the upstream and downstream vertex information associated with the potential energy-carbon imbalance link; Backtrack and obtain the level to which the upstream and downstream vertices belong in the hierarchical control architecture and the corresponding feature extraction node identifiers; Based on the hierarchical relationship between the upstream and downstream vertices, the generation and execution levels of the optimization instructions are determined. At the level of generating optimization instructions, based on the deviation characteristics of the potential energy-carbon imbalance link, an optimization instruction set containing adjustment objectives and constraints is generated.

[0011] Preferably, at the optimization instruction generation level, after generating an optimization instruction set containing adjustment objectives and constraints based on the deviation characteristics of the potential energy-carbon imbalance link, the process further includes: The optimized instruction set is sent to the execution level of the optimized instruction through the preset instruction delivery channel in the hierarchical control architecture; At the execution level of the optimization instructions, the received set of optimization instructions is parsed and converted into control parameter adjustment amounts that can be executed at the execution level; The feature extraction nodes of the execution level are used to provide real-time feedback of operational representation data during the control parameter adjustment process; The feedback operational performance data is transmitted back to the optimization instruction generation level through the preset data reporting channel in the hierarchical control architecture.

[0012] Preferably, after transmitting the feedback operational performance data back to the optimization instruction generation level through a preset data reporting channel in the hierarchical control architecture, the method further includes: At the level of generating optimization instructions, based on the returned operational characterization data, the local spatiotemporal correlation map involving the potential energy-carbon imbalance link is reconstructed. By comparing the local spatiotemporal correlation maps before and after optimization, the effect of the optimized instruction set on the correction of the potential energy-carbon imbalance link is evaluated. If the correction effect does not meet expectations, a new optimized instruction set will be generated iteratively based on the latest local spatiotemporal correlation map.

[0013] Preferably, after comparing the local spatiotemporal correlation maps before and after optimization to evaluate the corrective effect of the optimized instruction set on the potential energy-carbon imbalance link, the method further includes: If the correction effect achieves the expected result, record the current optimization strategy and final control parameters for the potential energy-carbon imbalance link. The recorded optimization strategies and control parameters are encapsulated into standardized optimization cases for such potential energy-carbon imbalance links. Standardized optimization cases are stored in the park's knowledge base to provide initial templates for optimization strategies for newly identified potential energy-carbon imbalance links.

[0014] Preferably, the present invention also includes an energy-saving and carbon-reduction optimization system for low-carbon industrial parks. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the energy-saving and carbon-reduction optimization method for low-carbon industrial parks as described above.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By deploying feature extraction nodes at each level of the hierarchical architecture and simultaneously collecting operational characterization data, cross-level correlation mapping of these operational data from different levels was performed to construct a spatiotemporal correlation map describing the overall energy and carbon flow in the park. This changes the isolated state of data in traditional monitoring, connecting discrete hierarchical data into a dynamic network model that reflects spatial topology and time series. It enables a systematic understanding of the park's energy and carbon flow, allowing managers to intuitively grasp the entire process of energy from input, conversion, transmission to consumption, as well as the associated carbon emission generation, transfer, and accumulation paths. It clearly reveals the mutual influence and coupling relationships between equipment and systems at different locations and levels in the dimensions of energy and carbon emissions.

[0016] Based on the constructed spatiotemporal correlation map, key energy-carbon coupling nodes and potential energy-carbon imbalance links in the energy flow path of the park are further identified. Utilizing the network analysis capabilities of the map, not only are individual energy hotspots or major emitters identified, but also hub nodes that connect multiple links and have a leverage effect on overall energy-carbon performance, as well as weak links in the system where energy efficiency changes and carbon emission changes are mismatched or uncoordinated. This allows optimization targets to be precisely focused from generalized equipment or regions to key interconnected structures and contradictory links within the system. This enables subsequent optimization resources to be precisely allocated to a few key links that most significantly impact overall performance, and allows for early warning and intervention against potential systemic imbalance risks, thereby achieving precise and coordinated regulation from a broader perspective to specific points, and from addressing existing problems to preventing future ones. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the working principle of the energy-saving and carbon-reduction optimization method for low-carbon industrial parks described in this invention. Figure 2 A flowchart for constructing a hierarchical control architecture; Figure 3 A flowchart for constructing a spatiotemporal correlation graph; Figure 4 A bar chart comparing the number of nodes at different levels in a low-carbon industrial park; Figure 5 A multi-dimensional comparative bar chart of the hierarchical control architecture for low-carbon industrial parks. Detailed Implementation

[0018] 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 embodiments of the present invention, and not all embodiments. 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.

[0019] Please see Figure 1This invention provides an energy-saving and carbon-reduction optimization method for low-carbon industrial parks. The method includes: defining and constructing a hierarchical control architecture for the park's energy system based on the park's energy supply network. This hierarchical control architecture consists of park-level, regional-level, and equipment-level architectures. At each level of the hierarchical control architecture, feature extraction nodes are deployed to sense the energy flow and carbon emission status of that level. The feature extraction nodes synchronously collect operational characterization data for the corresponding level. Cross-level correlation mapping is performed on the operational characterization data obtained from different levels of the hierarchical control architecture to form a spatiotemporal correlation map describing the overall energy and carbon flow of the park. Based on the constructed spatiotemporal correlation map, key energy-carbon coupling nodes and potential energy-carbon imbalance links in the park's energy flow paths are identified. For the identified potential energy-carbon imbalance links, an adaptive optimization process for the park's energy system is initiated.

[0020] In one embodiment of the present invention, see [reference] Figure 2 The park-level system corresponds to the overall energy inlet, outlet, and core conversion facilities of the entire low-carbon park, responsible for formulating energy dispatch strategies and carbon emission targets. The regional level corresponds to sub-regions within the low-carbon park with independent functions or physical boundaries, receiving and executing park-level control commands while managing energy-consuming units within those sub-regions. The equipment level corresponds to specific energy-consuming, energy-producing, and energy-storage equipment under the regional level, serving as the final execution unit for energy consumption and conversion. A two-way data reporting and command issuance channel is established between the park-level, regional-level, and equipment-level systems. At the park-level, a central feature extraction node is deployed to sense the inlet parameters of the park's overall power grid, gas grid, and heating grid, as well as the monitoring parameters of the overall emission outlet. At the regional level, distributed feature extraction nodes are deployed to sense the energy distribution hubs, internal microgrid operating status, and regional boundary carbon emission data of the corresponding sub-regions. At the equipment level, terminal feature extraction nodes are deployed to sense the real-time power, operating mode, energy efficiency level, and operating emission factors of individual devices. The central feature extraction node, distributed feature extraction node, and terminal feature extraction node are logically connected according to the hierarchical relationship of the layered control architecture.

[0021] In practical implementation, this method defines and constructs a hierarchical control architecture for the park's energy system. This hierarchical control architecture consists of park-level, regional-level, and equipment-level components. For example, a high-tech industrial park encompassing manufacturing, R&D testing, and office / living functions has an energy supply network covering the external power grid, natural gas grid, and internal photovoltaic power generation and energy storage systems. The park-level architecture corresponds to the overall energy inlet, outlet, and core conversion facilities of the entire low-carbon park, and is responsible for formulating energy dispatch strategies and total carbon emission targets. In practical implementation, the overall energy inlet includes the 10 kV main incoming line from the city power grid, the municipal natural gas pipeline inlet valve, and the park's centralized energy station. The main outlet is the park's networked carbon emission monitoring system. The core conversion facilities include centralized gas-fired combined cooling, heating, and power (CCHP) units and high-voltage substations.

[0022] In some embodiments, the region level corresponds to a sub-region within a low-carbon park that has independent functions or physical boundaries. It receives and executes park-level control commands and manages the energy-consuming units within the sub-region. For example, the entire park may be divided into a production and manufacturing area, a research and development testing area, and a comprehensive office and living area. Each sub-region has an independent energy metering and distribution subsystem. The equipment level corresponds to the specific energy-consuming equipment, production-producing equipment, and energy storage equipment under the region level. As the final execution unit for energy consumption and conversion, in specific implementations, the equipment-level units in the production and manufacturing area include CNC machine tools, air compressors, and circulating water pumps; the equipment-level units in the research and development testing area include environmental testing chambers and precision instruments; and the equipment-level units in the comprehensive office and living area include air conditioning units, lighting circuits, and electric vehicle charging piles.

[0023] A two-way data reporting channel and command issuance channel are established between the park level, regional level, and equipment level. The data reporting channel is used to collect the operating status of each level, and the command issuance channel is used to transmit optimization control commands. For example, by combining the park's industrial Ethernet with a 5G private network, bidirectional, low-latency transmission of data and commands can be achieved. In specific implementation, a central feature extraction node is deployed at the park level to sense the inlet parameters of the park's total power grid, gas network, and heating network, as well as the monitoring parameters of the total emission outlet. For example, the central feature extraction node collects in real time the active power of the 10 kV main incoming line, the instantaneous flow and pressure of the natural gas main pipeline, the total supply and return water temperature and flow of the park's energy station, and the instantaneous carbon dioxide equivalent concentration and cumulative emissions recorded by the carbon emission monitoring master table. Distributed feature extraction nodes are deployed at the regional level to sense the energy distribution hubs, internal microgrid operation status, and regional boundary carbon emission data of the corresponding sub-regions. For example, distributed feature extraction nodes are deployed in the power distribution room of the production and manufacturing area to collect the three-phase current, voltage, and power factor of the low-voltage side output line of the transformer in that area. At the same time, the output power of the photovoltaic inverter connected to the sub-region and the charging and discharging power of the energy storage system are collected, as well as the data of the carbon emission monitoring instrument installed at the ventilation opening of the sub-region.

[0024] It is understandable that terminal feature extraction nodes are deployed at the device level to sense the real-time power, operating mode, energy efficiency level, and operating emission factors of individual devices. For example, terminal feature extraction nodes can be integrated or connected within the control cabinet of an air compressor inverter, the controller of an air conditioning unit, or the communication module of an electric vehicle charging pile. These nodes collect the device's instantaneous power, operating frequency or set temperature, real-time energy efficiency ratio calculated based on power and output, and corresponding values ​​from a carbon emission factor database matched to the device type and operating status. The central feature extraction node, distributed feature extraction nodes, and terminal feature extraction nodes are logically connected according to the hierarchical relationship of a layered control architecture. For example, the central feature extraction node establishes a star-shaped logical connection with all distributed feature extraction nodes, and each distributed feature extraction node establishes a star-shaped logical connection with all terminal feature extraction nodes within its jurisdiction, forming a logical tree-like topology network. Optionally, data interaction between distributed feature extraction nodes can be forwarded through the central feature extraction node or, under authorization, directly through edge-to-edge communication.

[0025] In some embodiments, the data sensed by the feature extraction nodes logically constitute a strict hierarchical relationship. The total load data collected by the distributed feature extraction nodes in a production area should be approximately equal to the sum of the device-level power data collected by all associated terminal feature extraction nodes in that area. The matching degree can be verified by the formula: in: This indicates the degree of aggregation and matching between region-level and device-level data. This represents the total load data of the region collected by the distributed feature extraction nodes. Indicates the first Device-level power data collected by each terminal feature extraction node. This indicates the total number of devices associated with this region. Logical connections ensure that this data aggregation and verification relationship can be clearly established and maintained.

[0026] In one embodiment of the present invention, see [reference] Figure 3The operational characterization data includes total energy input, energy conversion efficiency, load dynamic curves, and direct carbon emission intensity. Driven by a unified timestamp, data acquisition actions are synchronously triggered at the central feature extraction node, distributed feature extraction nodes, and terminal feature extraction nodes. The total energy input data of the park collected by the central feature extraction node is correlated with the energy reception data of the sub-regions collected by the relevant distributed feature extraction nodes. The sub-region load data collected by the distributed feature extraction nodes is aggregated, verified, and traced back to its source with the equipment-level power data collected by all terminal feature extraction nodes within its jurisdiction. The equipment emission factor data collected by the terminal feature extraction nodes is aggregated upwards according to the energy flow path and cross-validated with regional and park-level carbon emission monitoring data. A spatiotemporal correlation map with time attributes is constructed, using energy flow and carbon flow paths as edges and feature extraction nodes and key facilities at each level as vertices.

[0027] In practical implementation, operational characterization data includes total energy input, energy conversion efficiency, load dynamic curves, and direct carbon emission intensity. For example, at the park level, total energy input is represented by the daily cumulative electricity purchased from the city power grid and the daily cumulative natural gas consumption; at the regional level, energy conversion efficiency is represented by the power generation efficiency of the gas-fired internal combustion engine in the sub-regional energy station; at the equipment level, the load dynamic curve is represented by the minute-by-minute active power change sequence of an air compressor over 24 hours; and direct carbon emission intensity is represented by the mass of carbon dioxide output per unit calorific value in the flue gas of a boiler. Driven by a unified timestamp, the data acquisition actions of the central feature extraction node, distributed feature extraction nodes, and terminal feature extraction nodes are synchronously triggered. This timestamp is distributed by a precision clock source deployed in the park through a network time synchronization protocol.

[0028] In some embodiments, the total energy input data of the park collected by the central feature extraction node is correlated with the energy reception data of the sub-area collected by the relevant distributed feature extraction nodes. For example, at the same collection time, the total power input of the park recorded by the central feature extraction node is 5 MW, while the power received by the distributed feature extraction node in the production and manufacturing area is 3.2 MW, the power received by the R&D and testing area is 1.1 MW, and the power received by the comprehensive office and living area is 0.6 MW. The energy flow distribution relationship of "total input of the park (5 MW) → production area (3.2 MW), R&D area (1.1 MW), office area (0.6 MW)" is established through flow correlation, where the 0.1 MW difference is related to the park-level public facilities and line losses.

[0029] It is understandable that the sub-regional load data collected by the distributed feature extraction nodes is aggregated, verified, and traced back to its source with the equipment-level power data collected by all terminal feature extraction nodes within its jurisdiction. For example, if the total regional load collected by the distributed feature extraction nodes in the production and manufacturing area at a certain moment is 850 kW, and at this time the power reported by the CNC machine tool terminal feature extraction node in this area is 200 kW, the power reported by the air compressor terminal feature extraction node is 300 kW, the power reported by the circulating water pump terminal feature extraction node is 280 kW, and the total power reported by the lighting and other terminal feature extraction nodes is 65 kW, the aggregation verification will find that there is a 5 kW deviation between the total reported equipment-level power (845 kW) and the total regional load (850 kW). Source tracing will then need to check whether the terminal feature extraction node data of critical equipment (such as a 75 kW ventilation fan) has not been successfully collected or whether there is a communication delay.

[0030] The equipment emission factor data collected by the terminal feature extraction nodes are aggregated upwards according to the energy flow path and cross-validated with regional and park-level carbon emission monitoring data. For example, the terminal feature extraction node of a gas boiler retrieves its operating emission factor of 56.1 kg CO2 / GJ from the preset factor library based on its combustion state and fuel type. The boiler output thermal power is 2 MW, and the instantaneous carbon emission flow rate of the equipment is calculated to be 0.403 kg / s. At the same time, the distributed feature extraction node of the sub-region to which the boiler belongs reads an instantaneous value of 0.41 kg / s from the regional boundary carbon emission monitor, and the central feature extraction node at the park level reads an instantaneous contribution value of approximately 0.398 kg / s from the total discharge monitor. By cross-validating these three data, the consistency of monitoring and the reliability of data can be assessed.

[0031] Optionally, when constructing associations, the consistency of flow direction can be quantitatively evaluated. One evaluation method uses the deviation rate parameter, the calculation formula of which is: in: This indicates the deviation rate between the aggregated data and the parent data. This represents the algebraic sum of the data reported by all relevant feature extraction nodes at the next level. This represents the total amount of data collected by the feature extraction node at the next higher level. This parameter is used to quantify the degree of fit between cross-level data mappings.

[0032] In one embodiment of the present invention, all vertices in the spatiotemporal correlation graph are traversed, and the correlation between the energy flow intensity and carbon flow intensity of each vertex is calculated. Vertices with a correlation exceeding a preset threshold are identified and marked as key energy-carbon coupling nodes. All edges in the spatiotemporal correlation graph are traversed, and the synchronicity of the energy flow and carbon flow carried on each edge is analyzed. Edges where the trends of energy flow and carbon flow change continuously diverge are identified and marked as potential energy-carbon imbalance links. The upstream and downstream vertex information associated with the potential energy-carbon imbalance links is extracted. The hierarchy of the upstream and downstream vertices in the hierarchical control architecture and the corresponding feature extraction node identifiers are obtained by backtracking. Based on the hierarchical relationship between the upstream and downstream vertices, the generation level and execution level of optimization instructions are determined. At the optimization instruction generation level, based on the divergence characteristics of the potential energy-carbon imbalance links, an optimization instruction set containing adjustment targets and constraints is generated.

[0033] In practice, all vertices in the spatiotemporal correlation graph are traversed, and the correlation between the energy flow intensity and carbon flow intensity of each vertex is calculated. For example, a vertex in the spatiotemporal correlation graph represents a "gas-fired internal combustion engine in the production area". Its energy flow intensity is the sum of the electrical power and thermal power output by the internal combustion engine (unit: kilowatts), and its carbon flow intensity is the carbon emission rate calculated in real time based on fuel consumption (unit: kilograms per hour). The correlation calculation aims to quantify the tightness of the coupling between energy conversion and carbon emissions at this vertex, identify vertices with a correlation exceeding a preset threshold, and mark them as key energy-carbon coupling nodes. For example, when the correlation calculation result of the "gas-fired internal combustion engine in the production area" vertex is 0.92, and the preset threshold is 0.85, this vertex is marked as a key energy-carbon coupling node.

[0034] In some embodiments, all edges in the spatiotemporal correlation graph are traversed, and the synchronicity of the energy flow and carbon flow carried on each edge is analyzed. For example, an edge connects the vertex of "10kV substation in the park" and the vertex of "energy storage system charging interface". The energy flow carried by this edge is the charging power, and the carbon flow intensity is the indirect carbon emission flow calculated based on the current marginal carbon emission factor of the power grid. The synchronicity analysis is to check whether the corresponding indirect carbon emission flow shows a synchronous upward trend when the charging power increases in the time series. Edges where the trends of energy flow and carbon flow change continuously diverge are identified and marked as potential energy-carbon imbalance links. For example, when it is detected that the charging power continuously increases during the midday peak of photovoltaic power generation, but because the proportion of clean energy in the power source of the power grid is high at this time, the indirect carbon emission factor decreases, resulting in a decrease in carbon flow intensity. This continuous reverse change trend makes the edge marked as a potential energy-carbon imbalance link.

[0035] It is understandable that the upstream and downstream vertex information associated with the potential carbon imbalance link is extracted. For example, for the marked edge from "10kV substation in the park" to "energy storage system charging interface", the extracted upstream vertex information is "10kV substation in the park" and the downstream vertex information is "energy storage system charging interface". The upstream and downstream vertices are traced back and the level to which they belong in the hierarchical control architecture and the corresponding feature extraction node identifier is obtained. By querying the metadata of the graph, the vertex "10kV substation in the park" belongs to the park level and the corresponding central feature extraction node identifier is "CTL_001", and the vertex "energy storage system charging interface" belongs to the device level and the corresponding terminal feature extraction node identifier is "TTL_energy storage_07".

[0036] Based on the hierarchical relationship between upstream and downstream vertices, the generation and execution levels of optimization instructions are determined. For example, the upstream vertex belongs to the park level, and the downstream vertex belongs to the device level. When the optimization objective involves adjusting the park-level energy dispatch strategy to affect device-level behavior, the generation level of optimization instructions can be determined as the park level, and the execution level can be determined as the device level. At the generation level of optimization instructions, based on the deviation characteristics of the potential energy carbon imbalance link, an optimization instruction set containing adjustment objectives and constraints is generated. For example, for the characteristic of "deviation between energy storage charging and grid carbon intensity", the generated optimization instruction set may include the adjustment objective: "When the grid marginal carbon intensity is lower than a set threshold, increase the charging power of the energy storage system to the upper limit", and the constraints: "The state of charge of the energy storage system is not lower than the safe lower limit" and "The rate of change of charging power does not exceed the maximum allowable value of the device".

[0037] Optionally, the vertex correlation degree can be calculated using a method based on the time series covariance and standard deviation, and the calculation formula is defined as: in: This indicates the correlation between vertex energy flux intensity and carbon flux intensity. Indicates a point in time The energy flux intensity value, Indicates the energy flow intensity over a time period The average value within, Indicates a point in time The carbon flux intensity value, Indicates carbon flow intensity over a time period The average value within, This represents the total number of sampling points within the selected analysis time period. The formula calculates... The closer the value is to 1 or -1, the stronger the synchronization or anti-synchronization.

[0038] See Figure 4This is a bar chart comparing the number of nodes at different levels in a low-carbon industrial park, primarily showcasing the distribution of feature extraction nodes at different levels. The number increases significantly with each level, with the equipment level (48 nodes) far exceeding the park level (5) and regional level (12), reflecting the characteristic of a hierarchical control architecture characterized by "many end-point devices and few top-level nodes." The proportions of each level differ considerably, with the regional level (5 nodes) exceeding the park level (2) and equipment level (3), indicating that the regional level is the core level for energy-carbon coupling. The purpose of this chart is to visually present the node layout characteristics of the hierarchical control architecture of the energy system in a low-carbon industrial park, helping to identify the key energy-carbon monitoring areas at each level. The equipment level is a dense area of ​​monitoring terminals, while the regional level is a concentrated area of ​​key energy-carbon coupling nodes.

[0039] In one embodiment of the present invention, the optimized instruction set is sent to the execution level of the optimized instruction through a preset instruction delivery channel in the hierarchical control architecture. At the execution level, the received optimized instruction set is parsed and converted into control parameter adjustment quantities executable at the execution level. Using the feature extraction nodes at the execution level, operational characterization data during the control parameter adjustment process is fed back in real time. This feedback operational characterization data is then transmitted back to the generation level of the optimized instruction through a preset data reporting channel in the hierarchical control architecture.

[0040] In practical implementation, the optimized instruction set is sent to the execution level of the optimized instruction through a pre-defined instruction delivery channel in the hierarchical control architecture. For example, an optimized instruction set generated at the park level, whose goal is to adjust the load distribution of the production and manufacturing area during the midday peak electricity consumption period, is encapsulated into a data message of a specific format and transmitted downlink to the corresponding regional control unit in the production and manufacturing area through the industrial Ethernet link between the park-level central controller and the regional-level distributed controller. This regional control unit is the execution level of the optimized instruction. In some embodiments, the received set of optimization instructions is parsed at the execution level of the optimization instructions and converted into control parameter adjustment quantities that can be executed at the execution level. For example, the set of optimization instructions received by the regional control unit includes the adjustment objective "to reduce the peak value of the total regional electrical load by no less than 15% between 13:00 and 15:00" and the constraints "the power supply reliability of the critical production line shall not be affected" and "the indoor temperature setpoint of the air conditioning system shall be adjusted within a range of ±2 degrees Celsius". The strategy parser inside the regional control unit converts these abstract objectives and constraints into a sequence of control parameter adjustment quantities for specific equipment within the jurisdiction. See Table 1 for the conversion results.

[0041] Table 1: Equipment-level Control Parameter Adjustment Table Equipment identification Control parameters Adjusted previous value Adjusted value Execution Time Window Priority Air conditioning unit A Temperature setpoint 24°C 26°C 13:00-15:00 Low Air compressor group B Output pressure setting 0.75MPa 0.72MPa 13:00-15:00 middle Interruptible charging station C Charging power 60kW 30kW 13:00-15:00 high Lighting circuit D Illuminance percentage 100% 80% 13:00-15:00 Low The system utilizes feature extraction nodes at the execution level to provide real-time feedback on operational data during control parameter adjustments. For example, when the temperature setpoint of air conditioning unit A is adjusted from 24 degrees Celsius to 26 degrees Celsius, the associated terminal feature extraction nodes begin to collect and report real-time power, compressor operating frequency, return air temperature, and other data of the air conditioning unit at a higher frequency. Simultaneously, distributed feature extraction nodes in the production area continuously report data such as total regional load and total electricity consumption to reflect the aggregation effect of control actions. It is understandable that the feedback operation performance data is transmitted back to the generation level of optimization instructions through the preset data reporting channel in the hierarchical control architecture. For example, the terminal feature extraction nodes of the aforementioned air conditioning units, air compressor groups and other equipment will first report the adjusted data collected to the distributed feature extraction nodes in the production and manufacturing area for preliminary aggregation. Then, the distributed feature extraction nodes will upload the summary data packet containing the changes in the total load of the area and the actual operating status of each device to the central feature extraction node and central controller at the park level through the data reporting channel.

[0042] Optionally, the completeness and timeliness of feedback data can be evaluated using a feedback delay parameter, the formula of which is defined as: in: Indicates the feedback delay parameter. This indicates the timestamp at which the central controller at the optimization instruction generation level received the feedback data packet. This indicates the timestamp at which the hierarchical control action is confirmed to have been executed. The feedback delay parameter is used to quantify the time interval from instruction execution to the generation of hierarchical perception results.

[0043] In some embodiments, the operational characterization data returned includes not only direct measurement data at the device and regional levels, but also derived evaluation data obtained through preliminary calculations at the execution level. For example, in the returned data, the regional control unit, in addition to including the value that the total regional load decreased from 1200 kW to 1000 kW before adjustment, also calculated the peak load reduction ratio as 16.7%. This ratio indicates that the target of "reduction of no less than 15%" in the optimization instruction set has been achieved. Such derived data is encapsulated together in the feedback data packet for return. In one embodiment of the present invention, at the generation level of optimization instructions, a local spatiotemporal correlation map involving potential energy-carbon imbalance links is reconstructed based on the returned operational characterization data. The local spatiotemporal correlation maps before and after optimization are compared to evaluate the corrective effect of the optimization instruction set on the potential energy-carbon imbalance links. If the corrective effect does not meet expectations, a new optimization instruction set is iteratively generated based on the latest local spatiotemporal correlation map. If the corrective effect meets expectations, the current optimization strategy and final control parameters for the potential energy-carbon imbalance links are recorded. The recorded optimization strategy and control parameters are encapsulated into standardized optimization cases for such potential energy-carbon imbalance links. The standardized optimization cases are stored in the park's knowledge base to provide optimization strategy initialization templates for newly identified similar potential energy-carbon imbalance links in the future.

[0044] In specific implementation, at the generation level of optimization instructions, the local spatiotemporal correlation graph involving potential energy carbon imbalance links is reconstructed based on the returned operational characterization data. For example, in the optimization case of "discrepancy between energy storage charging and grid carbon intensity" in the embodiment, the park-level central controller receives the adjusted data from the execution level, including the energy storage system charging power sequence and the real-time grid marginal carbon emission factor sequence at different time points. Based on these latest time series data, a local spatiotemporal correlation graph is redrawn with "park 10kV substation" and "energy storage system charging interface" as the core vertices and electrical flow and indirect carbon flow as the edges, and this graph is marked as "optimized state". By comparing the local spatiotemporal correlation maps before and after optimization, the effect of the optimized instruction set on the correction of potential energy-carbon imbalance links can be evaluated. For example, comparing the divergence map of increased midday charging power and decreased carbon flow before optimization with the map of further increased midday charging power and further decreased carbon flow due to the continued decline in grid carbon intensity after optimization, the goal of the analysis is to determine whether the original "divergence" feature has been weakened or eliminated due to the execution of the optimization instructions. The correlation coefficient of the changing trends of energy flow and carbon flow on the two sides is calculated. If the correlation coefficient changes from a negative value before optimization to a positive value after optimization or the absolute value of the negative value decreases significantly, it indicates that the divergence feature has been corrected.

[0045] In some embodiments, if the correction effect does not meet expectations, a new set of optimized instructions is generated iteratively based on the latest local spatiotemporal correlation map. For example, if the evaluation finds that although the midday deviation is weakened, the energy storage system is still executing the maximum power charging strategy at midday when the grid carbon intensity rises due to the load peak in the evening, resulting in a new energy-carbon imbalance, the park-level central controller will generate a new set of optimized instructions containing time-sharing strategies based on the latest local map containing data from the evening. The new set of instructions may include the adjustment target of "limiting the energy storage charging power below the baseline power Y when the real-time grid carbon intensity exceeds the set threshold X".

[0046] Understandably, if the correction effect meets expectations, the current optimization strategy and final control parameters for potential energy-carbon imbalance links will be recorded. For example, if the optimization for "midday photovoltaic peak period energy storage charging" is evaluated as effective, the system will record the complete optimization strategy: triggering conditions, control objectives, device-level control parameters, and related constraints. The recorded optimization strategy and control parameters will be encapsulated into standardized optimization cases for such potential energy-carbon imbalance links. The encapsulation process includes generating a unique index number for the case, defining the energy-carbon imbalance mode feature description applicable to the case, the associated vertex and edge types, and a complete control logic parameter package.

[0047] The quantitative evaluation of the corrective effect can be achieved using the effect attainment parameter, the calculation formula of which is defined as: in: This parameter represents the degree to which the effect is achieved. This represents the correlation coefficient between the time series of energy flow and carbon flow in the potential energy-carbon imbalance link before optimization. This represents the correlation coefficient between the energy flow and carbon flow time series on the same link after optimization. The closer the effect achievement parameter is to 1, the more significant the effect of eliminating deviation features. When the effect achievement parameter remains below the preset success threshold, it is determined that the correction effect has not met expectations.

[0048] Standardized optimization cases are stored in the park's knowledge base to provide initialization templates for optimization strategies for newly identified potential energy carbon imbalance links. For example, when the system identifies a new potential energy carbon imbalance link on a future operating day where "during peak clean energy output, the trend of a certain adjustable load deviates from the trend of grid carbon intensity," the system will query the park's knowledge base, match the case characteristics, and automatically extract the control logic framework from the stored standardized optimization case of "midday photovoltaic peak-hour energy storage charging" as an initialization template for generating a specific set of optimization instructions for the new link, thereby accelerating the optimization decision-making process.

[0049] See Figure 5 This is a multi-dimensional comparative bar chart of the hierarchical control architecture of low-carbon industrial parks, showing the differences in three indicators: "number of feature extraction nodes," "total energy flow," and "total carbon emissions" among the park-level, regional-level, and equipment-level. At the park level, the hierarchical architecture logic of "top-level overall energy dispatch and lower-level tiered allocation" is consistent; it is positively correlated with total energy flow, reflecting the coupling relationship between energy consumption and carbon emissions; only the equipment level has a few nodes, while the park-level and regional-level nodes are close to zero, indicating that the equipment level is the core level for terminal data collection. This chart can help low-carbon industrial parks identify the core energy and carbon control points at each level: the park level needs to focus on monitoring the macro-balance between total energy and carbon emissions, the equipment level needs to strengthen the refinement of terminal data collection, and the regional level is the intermediate control layer connecting the top level and the terminal.

[0050] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0051] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An energy-saving and carbon-reduction optimization method for low-carbon industrial parks, characterized in that, The method includes: Based on the energy supply network of the low-carbon park, a hierarchical control architecture for the park's energy system is defined and constructed, which consists of park-level, regional-level, and equipment-level control architectures. At each level of the hierarchical control architecture, feature extraction nodes are deployed to sense the energy flow and carbon emission status of that level. The feature extraction nodes synchronously collect operational characterization data at the corresponding levels, and perform cross-level correlation mapping on the operational characterization data obtained from different levels of the hierarchical control architecture to form a spatiotemporal correlation map describing the overall energy and carbon flow of the park. Based on the constructed spatiotemporal correlation map, key energy-carbon coupling nodes and potential energy-carbon imbalance links in the energy flow path of the park are identified. For the identified potential energy-carbon imbalance links, an adaptive optimization process for the park's energy system is initiated.

2. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 1, characterized in that, Define and construct a hierarchical control architecture for the park's energy system. This hierarchical control architecture consists of park-level, region-level, and equipment-level components, specifically including: The park-level unit corresponds to the total energy inlet, total outlet, and core conversion facilities of the entire low-carbon park, and is responsible for formulating energy dispatch strategies and total carbon emission targets. The regional level corresponds to sub-regions within the low-carbon park that have independent functions or physical boundaries. It receives and executes park-level control instructions and manages energy-consuming units within the sub-regions. The equipment level corresponds to the specific energy-consuming equipment, production-producing equipment, and energy storage equipment under the regional level, serving as the final execution unit for energy consumption and conversion. Establish two-way data reporting and command issuance channels between the park level, regional level, and equipment level.

3. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 2, characterized in that, At each level of the hierarchical control architecture, feature extraction nodes are deployed to sense the energy flow and carbon emission status of that level, specifically including: The feature extraction node in the park-level deployment center is used to sense the inlet parameters of the park's total power grid, total gas grid, total heating grid, and total emission outlet monitoring parameters; Distributed feature extraction nodes are deployed at the regional level to sense the energy distribution hubs, internal microgrid operation status, and regional boundary carbon emission data of the corresponding sub-regions. Deploy terminal feature extraction nodes at the device level to perceive the real-time power, operating mode, energy efficiency level, and operating emission factors of a single device; The central feature extraction node, distributed feature extraction nodes, and terminal feature extraction nodes are logically connected according to the hierarchical relationship of the layered control architecture.

4. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 3, characterized in that, The feature extraction nodes synchronously collect operational characterization data at corresponding levels, and perform cross-level correlation mapping on the operational characterization data obtained from different levels of the hierarchical control architecture to form a spatiotemporal correlation map describing the overall energy and carbon flow of the park, specifically including: The operational characterization data includes total energy input, energy conversion efficiency, load dynamics curve, and direct carbon emission intensity. Driven by a unified timestamp, the data acquisition actions of the central feature extraction node, distributed feature extraction nodes and terminal feature extraction nodes are triggered synchronously. The total energy input data of the park collected by the central feature extraction node is correlated with the energy receiving data of the sub-regions collected by the relevant distributed feature extraction nodes to determine the flow direction. The sub-region load data collected by the distributed feature extraction nodes is aggregated, verified, and traced back to its source with the device-level power data collected by all terminal feature extraction nodes within its jurisdiction. The equipment emission factor data collected by the terminal feature extraction node is aggregated upwards level by level according to the energy flow path and cross-validated with regional and park-level carbon emission monitoring data. Using energy flow and carbon flow paths as edges and feature extraction nodes and key facilities at each level as vertices, a spatiotemporal correlation graph with time attributes is constructed.

5. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 4, characterized in that, Based on the constructed spatiotemporal correlation map, key energy-carbon coupling nodes and potential energy-carbon imbalance links in the energy flow path of the park are identified, specifically including: Traverse all vertices in the spatiotemporal correlation graph and calculate the correlation between the energy flux intensity and the carbon flux intensity at each vertex; Vertices with a correlation degree exceeding a preset threshold are identified and marked as key energy carbon coupling nodes; Traverse all edges in the spatiotemporal correlation graph and analyze the synchronicity of energy flow and carbon flow carried on each edge; Identify the edges where the trends of energy flow and carbon flow continue to diverge, and mark them as potential energy-carbon imbalance links.

6. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 5, characterized in that, For the identified potential energy-carbon imbalance links, an adaptive optimization process for the park's energy system will be initiated, which specifically includes: Extract the upstream and downstream vertex information associated with the potential energy-carbon imbalance link; Backtrack and obtain the level to which the upstream and downstream vertices belong in the hierarchical control architecture and the corresponding feature extraction node identifiers; Based on the hierarchical relationship between the upstream and downstream vertices, the generation and execution levels of the optimization instructions are determined. At the level of generating optimization instructions, based on the deviation characteristics of the potential energy-carbon imbalance link, an optimization instruction set containing adjustment objectives and constraints is generated.

7. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 6, characterized in that, At the optimization instruction generation level, after generating an optimization instruction set containing adjustment objectives and constraints based on the deviation characteristics of the potential energy-carbon imbalance link, the following steps are also included: The optimized instruction set is sent to the execution level of the optimized instruction through the preset instruction delivery channel in the hierarchical control architecture; At the execution level of the optimization instructions, the received set of optimization instructions is parsed and converted into control parameter adjustment amounts that can be executed at the execution level; The feature extraction nodes of the execution level are used to provide real-time feedback of operational representation data during the control parameter adjustment process; The feedback operational performance data is transmitted back to the optimization instruction generation level through the preset data reporting channel in the hierarchical control architecture.

8. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 7, characterized in that, After the feedback operational performance data is transmitted back to the optimization instruction generation level through the preset data reporting channel in the hierarchical control architecture, the process also includes: At the level of generating optimization instructions, based on the returned operational characterization data, the local spatiotemporal correlation map involving the potential energy-carbon imbalance link is reconstructed. By comparing the local spatiotemporal correlation maps before and after optimization, the effect of the optimized instruction set on the correction of the potential energy-carbon imbalance link is evaluated. If the correction effect does not meet expectations, a new optimized instruction set will be generated iteratively based on the latest local spatiotemporal correlation map.

9. The energy-saving and carbon-reduction optimization method for low-carbon industrial parks according to claim 8, characterized in that, After comparing the local spatiotemporal correlation maps before and after optimization, and evaluating the effect of the optimized instruction set on correcting the potential energy-carbon imbalance link, the method further includes: If the correction effect achieves the expected result, record the current optimization strategy and final control parameters for the potential energy-carbon imbalance link. The recorded optimization strategies and control parameters are encapsulated into standardized optimization cases for such potential energy-carbon imbalance links. Standardized optimization cases are stored in the park's knowledge base to provide initial templates for optimization strategies for newly identified potential energy-carbon imbalance links.

10. An energy-saving and carbon-reduction optimization system for a low-carbon industrial park, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the energy-saving and carbon-reduction optimization method for low-carbon parks as described in any one of claims 1 to 9.

Images