Provincial energy supply risk early warning method and system and processing device

Abstract

The application relates to a provincial energy supply risk early warning method, a system and processing equipment, the method obtains multi-source data of a corresponding provincial energy system; risk identification is performed on the multi-source data to obtain a risk node, a node feature corresponding to the risk node, a risk transmission edge and a risk transmission strength corresponding to the risk transmission edge; a dynamic risk transmission graph is obtained according to the risk node, the risk transmission edge, the node feature and the risk transmission strength; the dynamic risk transmission graph is processed based on a preset graph intelligent model to obtain risk information of a corresponding risk node; an influence factor is obtained according to the multi-source data, the influence factor is processed based on a preset threshold algorithm to obtain a target dynamic threshold; risk early warning information is obtained according to the risk information and the target dynamic threshold, accurate early warning of provincial energy supply risk is realized, reliable early warning of provincial energy supply and emergency disposal is provided, and the reliability and accuracy of risk early warning are improved.

Description

Technical Field

[0001] This application relates to the field of energy supply risk early warning technology, and in particular to a provincial energy supply risk early warning method, system and processing equipment. Background Technology

[0002] With the ongoing energy revolution and the rapid development of digital technology, provincial energy systems are shifting from a traditional single-energy supply model to a multi-energy coordinated security model. The supply, transmission, storage, conversion, and consumption of various energy sources, including electricity, coal, natural gas, oil, and new energy, are becoming increasingly interconnected. The operational status of the energy system no longer depends on a single energy source or a single indicator, but is influenced by multiple energy types, multiple operational stages, and various external factors. For provinces like Guangdong, with large energy consumption, limited local primary energy supply capacity, and a high degree of dependence on external energy imports, energy security faces greater uncertainty and complexity.

[0003] The risks to provincial energy supply are characterized by significant transmission and cumulative effects. On the one hand, the supply of primary energy sources such as coal, natural gas, and oil is affected by factors such as production, transportation, imports, inventory, and prices, and these fluctuations can further transmit to power supply, industrial production, and residential energy consumption. On the other hand, the output of new energy sources is greatly affected by meteorological conditions. When extreme weather events such as high temperatures, typhoons, heavy rainfall, and cold waves occur, they may simultaneously trigger a rapid increase in electricity load, disrupt port transportation, increase the operational risks of transmission lines, and cause instability in fuel supply. For example, typhoons may restrict port operations, affecting the arrival of coal for power generation and the inventory of thermal power plants; tight natural gas supply may limit the output of gas-fired peak-shaving units, further exacerbating the pressure on power supply and demand; and sustained high temperatures may push up electricity load and amplify the system regulation pressure caused by fluctuations in new energy output. Therefore, provincial energy risks often do not occur in isolation, but rather are gradually transmitted and amplified along the energy supply, transmission, storage, and consumption links.

[0004] Existing provincial energy systems have accumulated a wealth of operational, transaction, inventory, load, meteorological, economic, and historical risk event data. However, this data is typically scattered across different business systems and management platforms, resulting in issues such as multiple data sources, inconsistent time scales, inconsistent spatial granularity, significant differences in indicator definitions, and insufficient semantic correlation within business contexts. Traditional energy monitoring methods often focus on a single energy type or a single business indicator; for example, electricity focuses on load and supply capacity monitoring, coal on inventory and arrival volume monitoring, and natural gas on import volume and receiving terminal inventory monitoring. These methods struggle to comprehensively depict the interconnectedness between different energy types, different operational stages, and external risk events and the energy system. They also fail to promptly identify potential risks that have not yet manifested but already exhibit a transmission trend.

[0005] Currently, existing provincial energy supply risk monitoring methods mainly fall into the following categories: 1. Risk monitoring schemes based on fixed thresholds or manual rules. For example, existing energy monitoring systems typically set fixed thresholds for indicators such as load, inventory, supply, price, and equipment status. When an indicator exceeds the preset threshold, the system outputs an alarm. This type of scheme is simple to implement, but it can usually only identify single-point anomalies that have already occurred. It is difficult to express the process of risk transmission along the energy industry chain and spatial regions, and it is also difficult to identify potential risks that have not yet exceeded the threshold but are rapidly deteriorating. 2. Risk assessment schemes based on load forecasting or supply and demand forecasting, such as multi-dimensional load forecasting, weather data coupled with multi-energy load forecasting, demand response potential forecasting, and regional energy system multi-dimensional load forecasting. This type of scheme can predict future load changes, energy demand changes, or supply and demand balance, providing a reference for energy operation management. However, its focus is usually on the target indicators themselves, mainly focusing on outcome indicators such as load, demand, supply, or supply and demand gaps. It does not directly depict the risk transmission path between energy supply, transmission, storage, consumption, and external events, and it is also difficult to explain how a certain risk factor spreads to other energy links and forms a systemic supply risk. 3. Energy system operation and management solutions based on situational awareness and optimized scheduling, such as those that monitor the energy system status to assist in generating scheduling strategies or optimizing operation plans, are valuable for improving the efficiency of energy system operation. However, if the goal is only optimized scheduling, it is easy to overlook the linkage and propagation process of risks from external events, upstream supply, transmission channels, inventory reserves to end-user consumption. 4. Risk assessment solutions based on single energy subsystems, such as in power systems and oil and gas pipeline networks, existing solutions assess fault risks, cascading failure risks, or pipeline leakage risks based on system topology and operating status. These solutions are usually geared towards single energy systems, with nodes and edges mostly built on a single physical network, making it difficult to simultaneously cover multiple energy categories such as electricity, coal, gas, and oil, as well as external risk factors such as macroeconomic factors, extreme disasters, and social events. 5. Risk management solutions based on static risk databases or manually maintained risk lists, such as compiling historical risk events, typical risk types, and emergency response measures into static risk lists or risk databases, are helpful for the accumulation of risk knowledge. However, risk factors, risk weights, and risk relationships often rely on manual maintenance and are difficult to update dynamically based on real-time data, making it difficult to support real-time risk diffusion identification and early warning. Summary of the Invention

[0006] The purpose of this application is to provide a method, system and processing equipment for early warning of provincial energy supply risks, in order to solve the problems of insufficient utilization of multi-source data, insufficient expression of risk transmission relationships, insufficient identification of risk diffusion paths, delayed early warning of potential risks and insufficient interpretability of early warning results in the existing provincial energy supply risk monitoring methods.

[0007] Firstly, this application provides a method for early warning of provincial energy supply risks, including:

[0008] Obtain multi-source data from the corresponding provincial energy system; Risk identification is performed on multi-source data to obtain risk nodes, node characteristics of corresponding risk nodes, risk transmission edges, and risk transmission strength of corresponding risk transmission edges; Based on the risk nodes, risk transmission edges, node characteristics, and risk transmission intensity, a dynamic risk transmission graph is obtained; The dynamic risk transmission graph is processed based on the preset graph intelligent model to obtain the risk information of the corresponding risk nodes; Based on multi-source data, the impact factors are obtained. Based on a preset threshold algorithm, the impact factors are processed to obtain the target dynamic threshold. Risk warning information is obtained based on risk information and target dynamic thresholds.

[0009] In one embodiment, the step of obtaining risk warning information based on risk information and a target dynamic threshold includes: Based on the risk information and the target dynamic threshold, the actual risk information and / or the potential risk information of the corresponding risk node are obtained; Risk warning information is obtained based on actual risk information and / or potential risk information.

[0010] In one embodiment, the risk information includes node risk probability and risk path; The steps for obtaining the actual risk information and / or potential risk information of the corresponding risk node based on risk information and target dynamic thresholds include: When the node risk probability of the corresponding risk node is greater than or equal to the target dynamic threshold, the actual risk information of the corresponding risk node is obtained. When the node risk probability of the corresponding risk node is less than the target dynamic threshold, and the corresponding risk node meets the preset potential risk conditions, the potential risk information of the corresponding risk node is obtained; wherein, the preset potential risk conditions include at least one of the following: the node risk probability gradually increases in at least two consecutive preset time windows and the upward slope of the corresponding node risk probability exceeds the preset trend threshold; the difference between the node risk probability of the corresponding risk node and the dynamic threshold is less than the preset proximity threshold; the corresponding risk node has a risk path from the upstream risk node; the upstream risk node or the same type of adjacent risk node of the corresponding risk node has been confirmed as actual risk information.

[0011] In one embodiment, the influencing factors include historical risk distribution information, operational scenario information, intensity of external risk events, and risk tolerance. The steps for processing influencing factors based on a preset threshold algorithm to obtain the target dynamic threshold include: Based on the operational scenario information, historical risk distribution information, the intensity of external risk events, and risk tolerance are input into a preset threshold algorithm for processing to obtain the target dynamic threshold.

[0012] In one embodiment, the preset graph intelligence model includes at least one of the following network models: graph attention network model, graph convolutional network model, spatiotemporal graph convolutional network model, graph convolutional recurrent neural network model, and relation-aware graph neural network model.

[0013] In one embodiment, the step of identifying risk nodes from multi-source data includes: Risk identification is performed on multi-source data to obtain regional risk factors, energy category risk factors, energy sector risk factors, and risk objects; Based on regional risk factors, energy category risk factors, energy sector risk factors, and risk objects, the corresponding risk nodes are obtained.

[0014] In one embodiment, the step of identifying risks from multi-source data to obtain risk propagation edges includes: Based on the preset risk relationships, a transmission edge is constructed for the corresponding two risk nodes to obtain the corresponding risk transmission edge; among them, the preset risk relationships include physical dependence, supply and demand balance, spatial adjacency, facility association, historical co-occurrence, external influence, and time lag.

[0015] In one embodiment, the step of identifying risks in multi-source data and obtaining the risk transmission strength of the corresponding risk transmission edge includes: Based on the preset risk relationships, the risk relationship strength of the corresponding risk transmission edges is processed to obtain the corresponding risk relationship strength value; Based on the corresponding risk relationship strength value and the preset weight of the corresponding risk relationship strength value, the risk transmission strength of the corresponding risk transmission edge is obtained.

[0016] In one embodiment, prior to the step of risk identification of multi-source data, the following steps are included: Preprocessing of multi-source data yields the first intermediate data; Risk factors are labeled on the first intermediate data to obtain the target multi-source data.

[0017] In one embodiment, after obtaining risk warning information based on risk information and target dynamic threshold, the process includes: When the risk warning information is consistent with the actual risk result, the weight of the corresponding risk transmission edge is increased based on the first step value and the weight of the corresponding risk node is increased based on the second step value. When the number of information items in the risk warning information is less than the number of information items in the actual risk result, the dynamic threshold of the corresponding risk node is reduced based on the third step value or the weight of the corresponding risk node is increased based on the fourth step value. When one of the information items in the risk warning information is inconsistent with the information items in the actual risk result, the weight of the corresponding risk node is reduced based on the fifth step value, or the dynamic threshold of the relevant risk node is increased based on the sixth step value.

[0018] Secondly, this application also provides a provincial energy supply risk early warning system, including: The data acquisition module is used to acquire multi-source data from the corresponding provincial energy system; The risk identification module is used to identify risks in multi-source data and obtain risk nodes, node characteristics of corresponding risk nodes, risk transmission edges, and risk transmission strength of corresponding risk transmission edges. The risk transmission graph construction module is used to generate a dynamic risk transmission graph based on risk nodes, risk transmission edges, node characteristics, and risk transmission intensity. The graph intelligent risk processing module is used to process the dynamic risk transmission graph based on the preset graph intelligent model to obtain the risk information of the corresponding risk nodes. The dynamic threshold acquisition module is used to obtain the impact factor based on multi-source data, process the impact factor based on a preset threshold algorithm, and obtain the target dynamic threshold. The early warning output module is used to obtain risk warning information based on risk information and target dynamic thresholds.

[0019] Thirdly, this application also provides a processing device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the provincial energy supply risk early warning method described above.

[0020] Fourthly, this application also provides a computer storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the steps of the provincial energy supply risk early warning method described above.

[0021] One of the above technical solutions has the following advantages and beneficial effects: The aforementioned provincial energy supply risk early warning method involves acquiring multi-source data from the corresponding provincial energy system; identifying risks in the multi-source data to obtain risk nodes, node characteristics of the corresponding risk nodes, risk transmission edges, and risk transmission intensity of the corresponding risk transmission edges; generating a dynamic risk transmission map based on the risk nodes, risk transmission edges, node characteristics, and risk transmission intensity; processing the dynamic risk transmission map using a preset intelligent model to obtain risk information for the corresponding risk nodes; obtaining influencing factors based on the multi-source data; processing the influencing factors using a preset threshold algorithm to obtain a target dynamic threshold; and obtaining risk early warning information based on the risk information and the target dynamic threshold, thereby achieving accurate early warning of provincial energy supply risks. This application fully utilizes multi-source data acquired for provincial energy supply security, and identifies risk nodes, risk transmission edges, risk node characteristics, and risk transmission intensity based on the multi-source data. Then, based on these risk nodes, risk transmission edges, risk node characteristics, and risk transmission intensity, a dynamic risk transmission map is constructed that reflects the risk correlation and transmission path of provincial energy supply security. This dynamic risk transmission map is input into a pre-set intelligent model, which outputs risk information for each risk node. Combined with a dynamic threshold mechanism, it enables early identification of risks that have not exceeded the threshold but have a worsening trend and the potential for propagation, thereby outputting risk warning information. This provides reliable early warning for provincial energy supply security and emergency response, improving the reliability and accuracy of risk warnings. Attached Figure Description

[0022] Figure 1 This is a schematic diagram illustrating the application environment of the provincial energy supply risk early warning method in the embodiments of this application; Figure 2 This is a flowchart illustrating the provincial energy supply risk early warning method in the embodiments of this application; Figure 3 This is a flowchart illustrating the risk identification steps in an embodiment of this application; Figure 4 This is a flowchart illustrating the risk node acquisition steps in an embodiment of this application. Figure 5 This is a flowchart illustrating the parameter correction steps in an embodiment of this application; Figure 6 This is a block diagram of the provincial energy supply risk early warning system in the embodiments of this application. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] In addition, the term "multiple" should mean two or more.

[0026] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0027] The provincial energy supply risk early warning method provided in this application can be applied to, for example... Figure 1The application environment shown is illustrated. The processing device includes a memory 10 and a processor 20. The memory 10 is connected to the processor 20 and can store multi-source data, risk nodes, risk transmission edges, node characteristics and risk transmission strength, dynamic risk transmission graphs, preset graph intelligent models, risk information, target dynamic thresholds, and risk warning information. The processor 20 can acquire multi-source data of the corresponding provincial energy system; perform risk identification on the multi-source data to obtain risk nodes, node characteristics of the corresponding risk nodes, risk transmission edges, and risk transmission strength of the corresponding risk transmission edges; obtain a dynamic risk transmission graph based on the risk nodes, risk transmission edges, node characteristics, and risk transmission strength; process the dynamic risk transmission graph based on the preset graph intelligent model to obtain risk information for the corresponding risk nodes; obtain influence factors based on the multi-source data; process the influence factors based on a preset threshold algorithm to obtain target dynamic thresholds; and obtain risk warning information based on the risk information and target dynamic thresholds, thereby achieving accurate early warning of provincial energy supply risks. The processing device may also include a display 30, which can display information such as multi-source data, risk nodes, risk transmission edges, node characteristics and risk transmission strength, dynamic risk transmission graphs, preset intelligent models, risk information, target dynamic thresholds, and risk warning information through a graphical interface. For example, the processing device may be a server, cloud platform, edge computing platform, or energy and economic data platform.

[0028] In one example, a provincial energy system may involve multiple energy sources such as electricity, coal, natural gas, oil, and new energy, with operational stages including energy supply, inter-regional import, transmission and transportation, storage and inventory, conversion and utilization, and end-use consumption. For instance, in provinces like Guangdong, with large energy consumption, limited local primary energy supply capacity, and high dependence on external energy input, energy security risks are often not due to anomalies in a single energy type or a single operational indicator, but rather the result of multiple factors interacting, gradually transmitting, and amplifying. This application is applicable to analyzing the risk correlations among multiple energy types and operational stages in a provincial energy system, identifying risk factors such as insufficient energy supply, low inventory, transportation disruptions, rapid load increases, abnormal price fluctuations, fluctuations in new energy output, and the impact of external disasters, and determining the potential transmission paths, scope of impact, and early warning levels of these risks.

[0029] This application can be applied to provincial energy economic data platforms, provincial energy operation monitoring platforms, provincial energy supply risk early warning platforms, and comprehensive energy dispatch auxiliary decision-making platforms. It can also be used by energy authorities, energy companies, energy trading institutions, and emergency response agencies to deploy provincial energy supply risk early warning systems for energy security monitoring and decision support. Typical application scenarios include energy security monitoring during peak summer and winter seasons, major event support, and holiday supply guarantees; energy supply risk early warning under extreme weather conditions such as typhoons, ice storms, high temperatures, heavy rainfall, cold waves, and droughts; and risk identification and early warning under multiple overlapping factors such as reduced coal deliveries, port shutdowns, disrupted railway transportation, delayed natural gas import shipments, insufficient LNG receiving terminal inventory, abnormal oil prices, and rapid increases in electricity load.

[0030] In one embodiment, such as Figure 2 As shown, a method for early warning of energy supply risks in a province is provided, including: Step S210: Obtain multi-source data of the corresponding provincial energy system.

[0031] For example, multi-source data for a provincial energy system may include energy operation data, energy reserve data, energy transmission and circulation data, market and price data, macroeconomic and industrial data, environmental and disaster data, and historical risk event data. Specifically, energy operation data may include electricity load, power generation output, coal arrivals, natural gas supply, oil supply, new energy output, and energy consumption. Energy reserve data may include coal inventory, natural gas reserves, oil inventory, available reserve capacity, and the number of days of coal reserves at key power plants. Energy transmission and circulation data may include transmission line load rates, pipeline pressure, port throughput, railway or waterway transport capacity, road traffic status, and energy flow data. Market and price data may include coal prices, natural gas prices, oil prices, electricity trading prices, and energy procurement price indices. Macroeconomic and industrial data may include GDP, PMI, industrial added value, key industry capacity, key enterprise operating rates, and regional economic prosperity. Environmental and disaster data may include typhoon paths, wind speed, rainfall, temperature, humidity, ice storms, severe convective weather, and geological disaster warnings. Historical risk event data may include records of historical supply disruptions, abnormal loads, inventory alerts, transportation disruptions, sharp price fluctuations, equipment failures, and disaster impacts.

[0032] For example, multi-source data can be accessed through database interfaces, real-time data stream interfaces, file interfaces, network interfaces, or manual input interfaces. Data from different data sources can be normalized using unified timestamps, unified regional codes, unified energy category codes, and unified risk object codes.

[0033] Step S220: Perform risk identification on multi-source data to obtain risk nodes, node characteristics of corresponding risk nodes, risk transmission edges, and risk transmission strength of corresponding risk transmission edges.

[0034] In this context, a risk node refers to a combination of risk factors within a provincial energy system. A node feature is the characteristic vector of a risk node at a given time. For example, the node feature of each risk node at time t is denoted as xi(t), which may include current value, rate of change, deviation from historical mean, volatility, risk event markers, trend characteristics, scenario characteristics, and business labels. A risk transmission edge represents the probability that one risk node will influence, transmit, or spread to another risk node. The risk transmission strength reflects the probability and degree of impact of risk transmission from risk node i to risk node j.

[0035] For example, multiple risk factors are obtained by identifying risks from multi-source data; corresponding risk nodes are constructed based on each risk factor, resulting in multiple risk nodes; features are collected from the corresponding risk nodes to obtain the node features of the corresponding risk nodes; corresponding risk transmission edges are constructed based on each risk node, resulting in multiple risk transmission edges; and the risk transmission intensity of the corresponding risk transmission edges is obtained by calculating the intensity of the corresponding risk transmission edges.

[0036] Step S230: Based on the risk nodes, risk transmission edges, node characteristics, and risk transmission intensity, obtain the dynamic risk transmission graph.

[0037] For example, based on multiple risk nodes, a risk node set V is obtained; based on the node characteristics at multiple corresponding times, a node feature matrix Xt is obtained; based on multiple risk transmission edges, a risk transmission edge set E is obtained; and based on the risk transmission intensity at multiple corresponding times, a risk transmission intensity matrix Wt is obtained. Based on the risk node set, risk transmission edge set, node feature matrix, and risk transmission intensity matrix, a dynamic risk transmission graph Gt is constructed for the corresponding provincial energy system at time t, reflecting the correlation and transmission path of provincial energy supply risks. That is, Gt = (V, E, Xt, Wt). Here, the node feature matrix Xt represents the operating status, changing trend, and risk characteristics of each risk node at the current time; the risk transmission intensity matrix Wt represents the transmission intensity and impact degree between different risk nodes.

[0038] In one example, as real-time data is continuously updated, node characteristics and risk transmission intensity dynamically change over time, thus forming a continuous sequence of dynamic risk transmission graphs, i.e., Gt. ={Gt m+1, Gt {m+2, ..., Gt}, where m represents the length of the historical time window used for model inference. This dynamic risk transmission graph sequence can serve as input for subsequent graph intelligent models to learn risk transmission patterns and perform risk inference.

[0039] Step S240: Process the dynamic risk transmission map based on the preset intelligent model to obtain the risk information of the corresponding risk nodes.

[0040] For example, the dynamic risk transfer graph sequence Gt ={Gt m+1, Gt The system inputs {m+2, ..., Gt} into a pre-defined graph intelligent model, learns the transmission patterns and temporal evolution trends of risks among different nodes in the corresponding provincial energy system, and thus obtains the risk information of the corresponding risk nodes. It should be noted that the pre-defined graph intelligent model uses a dynamic risk transmission graph sequence Gt. Using Xt as input, the node feature matrix and Wt risk transmission intensity matrix at each time point are jointly modeled, and the risk information of each risk node within the preset time window is output.

[0041] In one example, risk information includes risk probability, risk type, and risk path. The risk inference result of risk node vi at future time t+τ can be expressed as: (pi(t+τ), yi(t+τ), Pi(t+τ))=F(Gt). Where pi(t+τ) represents the risk probability of risk node vi at future time t+τ, yi(t+τ) represents the risk type, Pi(t+τ) represents the risk path associated with this risk node, and F(Gt) represents the preset graph intelligent model.

[0042] Step S250: Based on multi-source data, obtain the influence factor, process the influence factor based on the preset threshold algorithm, and obtain the target dynamic threshold.

[0043] The preset threshold algorithm can be determined by a combination of historical statistical methods, machine learning regression methods, quantile prediction methods, or expert rules. For example, when the provincial energy system is in a period of high temperature, typhoon, low inventory, or critical protection, the dynamic threshold can be automatically lowered to improve early warning sensitivity; when the provincial energy system is in a low-load or high-inventory safe state, the dynamic threshold can be appropriately raised to reduce false alarms. Influencing factors can be determined based on different regions, energy types, energy links, seasons, weather conditions, protection periods, and operational scenarios.

[0044] For example, for regions, energy categories, energy processes, and operational scenarios in multi-source data, influencing factors are identified. Then, based on a preset threshold algorithm, these influencing factors are processed to obtain a target dynamic threshold. It should be noted that the target dynamic threshold differs from a fixed threshold; it can be adjusted according to factors such as season, weather, holidays, peak summer and winter demand, major event support, economic operating status, and the intensity of external risk events.

[0045] Step S260: Obtain risk warning information based on risk information and target dynamic threshold.

[0046] Risk warning information can include warning level, risk type, risk level, risk probability, risk occurrence time or prediction time window, affected area, affected energy category, key risk nodes, risk path, explanation of risk source, and recommended areas of concern. Risk levels can be classified into four levels: blue, yellow, orange, and red, or into five levels: normal, attentive, tense, abnormal, and severely abnormal. Warning levels can be determined by the node risk probability, risk transmission intensity, affected area, risk duration, and importance of key nodes. Risk paths can be used to explain the reasons for the warning. For example, in a typhoon scenario, the system can output risk paths for typhoon risk nodes, port transportation disruption nodes, coal arrival decline nodes, thermal power inventory decline nodes, thermal power output limitation nodes, and power supply security risk nodes.

[0047] For example, based on risk information and target dynamic thresholds, it outputs risk warning information including risk type, risk level, affected area, affected energy type, key transmission path and warning time window, providing technical support for provincial energy supply security and emergency response.

[0048] In the above embodiments, by fully utilizing the acquired multi-source data for provincial energy supply security, and identifying risk nodes, risk transmission edges, risk node characteristics, and risk transmission intensity based on the multi-source data, a dynamic risk transmission map that reflects the risk correlation and transmission path of provincial energy supply security is constructed based on the risk nodes, risk transmission edges, risk node characteristics, and risk transmission intensity. The dynamic risk transmission map is input into a preset intelligent model, which outputs the risk information of each risk node. Combined with a dynamic threshold mechanism, it enables early identification of risks that have not exceeded the threshold but have a worsening trend and the possibility of propagation, and then outputs risk warning information to provide reliable early warning for provincial energy supply security and emergency response, thereby improving the reliability and accuracy of risk warning.

[0049] In one embodiment, the step of obtaining risk warning information based on risk information and a target dynamic threshold includes: Based on risk information and target dynamic thresholds, the actual risk information and / or potential risk information of the corresponding risk nodes are obtained; based on the actual risk information and / or potential risk information, risk warning information is obtained.

[0050] Among them, actual risk information is used to indicate that the probability of a risk occurring at a corresponding risk node is relatively high; potential risk information is used to indicate that a corresponding risk node has not yet reached the threshold condition but has the potential for spread, deterioration, or chain effect.

[0051] Based on the target dynamic threshold and the risk information output by the preset intelligent model, the system identifies the actual risks exceeding the dynamic threshold at corresponding risk nodes, as well as potential risks that have not yet exceeded the threshold but exhibit a continuous upward trend, strong upstream transmission paths, or external risk amplification effects. This yields the actual risk information and / or potential risk information for the corresponding risk nodes. When actual and / or potential risk information is identified, risk warning information is generated, providing reliable early warnings for provincial energy supply security and emergency response, thus improving the reliability and accuracy of risk warnings.

[0052] In one example, the system provides a risk warning for energy supply security in a province under the influence of a typhoon. Taking the typhoon impact scenario in the coastal areas of Guangdong Province as an example, the system obtains typhoon path, wind speed, rainfall and warning level from meteorological data sources, and obtains data such as port coal arrivals, coal inventory, thermal power output and power load from the energy system. First, the risk values ​​of external risk nodes of the typhoon are identified as increasing, and the risk enhancement effect on coastal port transportation nodes is calculated based on the external impact edges. Then, the pre-set intelligent model performs risk propagation inference along the path of typhoon risk, port transportation obstruction, coal arrival decline, thermal power inventory decline, thermal power output limitation, and power supply security risk. If the risk probability of thermal power inventory nodes or power supply security nodes exceeds the target dynamic threshold, a risk warning corresponding to the actual risk information is generated. If the risk probability has not yet exceeded the threshold but continues to rise and there is a strong upstream risk path, a risk warning corresponding to the potential risk is generated. Then, risk warning information such as warning level, affected cities, affected energy categories, key risk nodes, risk paths, and warning time windows are output. This provides a basis for energy supply security management departments to carry out inventory replenishment, transportation coordination, load monitoring, and emergency support in advance, and provides reliable early warning for provincial energy supply security and emergency response, improving the reliability and accuracy of risk warnings.

[0053] In one embodiment, risk information includes node risk probability and risk path. Node risk probability refers to the probability of a corresponding risk node at a predetermined future time; risk path refers to the risk contribution path associated with the corresponding risk node. It should be noted that risk information may also include risk type.

[0054] In one example, such as Figure 3As shown, the steps for obtaining the actual risk information and / or potential risk information of a corresponding risk node based on risk information and target dynamic thresholds include: Step S310: When the node risk probability of the corresponding risk node is greater than or equal to the target dynamic threshold, the actual risk information of the corresponding risk node is obtained.

[0055] For example, let pi(t+τ) be the node risk probability of risk node i, and θi(t) be the target dynamic threshold. When pi(t+τ) ≥ θi(t), the corresponding risk node is determined as an actual risk node, and the actual risk information for that node is generated. For instance, the actual risk information may include node risk probability, impact range, and risk path. The risk level of the corresponding risk node can be determined based on its node risk probability, impact range, and risk path.

[0056] Step S320: When the node risk probability of the corresponding risk node is less than the target dynamic threshold and the corresponding risk node meets the preset potential risk conditions, the potential risk information of the corresponding risk node is obtained.

[0057] Among them, the preset potential risk conditions include at least one of the following: the node risk probability gradually increases in at least two consecutive preset time windows and the upward slope of the corresponding node risk probability exceeds the preset trend threshold; the difference between the node risk probability of the corresponding risk node and the dynamic threshold is less than the preset proximity threshold; the corresponding risk node has a risk path from the upstream risk node; the upstream risk node or the same adjacent risk node of the corresponding risk node has been confirmed as actual risk information.

[0058] It should be noted that if the difference between the node risk probability of the corresponding risk node and the dynamic threshold is less than the preset proximity threshold, then the node risk probability is determined to be close to the target dynamic threshold, for example, pi(t+τ)≥λθi(t), where λ is the proximity coefficient. Preset potential risk conditions also include: the external risk event enhancement factor increases by a preset value, and there is an external influence edge between the node and the corresponding risk node.

[0059] When the node risk probability of the corresponding risk node has not yet reached the target dynamic threshold, and the corresponding risk node meets at least one of the preset potential risk conditions, the corresponding risk node is identified as a potential risk node, and potential risk information of the corresponding risk node is generated.

[0060] In one embodiment, the influencing factors include historical risk distribution information, operational scenario information, the intensity of external risk events, and risk tolerance. The step of processing the influencing factors based on a preset threshold algorithm to obtain the target dynamic threshold includes: Based on the operational scenario information, historical risk distribution information, the intensity of external risk events, and risk tolerance are input into a preset threshold algorithm for processing to obtain the target dynamic threshold.

[0061] In one example, the target dynamic threshold θi(t) of risk node i at time t is determined based on historical risk distribution information, operational scenario information, the intensity of external risk events, and risk tolerance, i.e., θi(t) = Qi(c, q) + βσi(c). γEi(t). Where Qi(c, q) represents the q quantile of the historical risk distribution information of risk node i in operating scenario c, σi(c) represents the risk tolerance in operating scenario c, Ei(t) represents the intensity of external risk events, and β and γ are preset adjustment coefficients.

[0062] In one embodiment, the preset graph intelligence model includes at least one of the following network models: graph attention network model, graph convolutional network model, spatiotemporal graph convolutional network model, graph convolutional recurrent neural network model, and relation-aware graph neural network model.

[0063] For example, a pre-defined graph intelligent model is obtained based on a graph attention network model and a temporal neural network model. The graph attention network model is used to weighted aggregate information of neighboring nodes of a target risk node based on node characteristics, risk propagation edges, and risk propagation intensity, calculating the risk contribution of different upstream risk nodes to the target risk node. The temporal neural network model is used to model the risk representation of risk nodes at continuous time points, learning the trend of risk changes over time. It should be noted that the graph attention mechanism can assign different attention weights to different neighboring risk nodes based on node characteristics and risk propagation intensity, enabling the pre-defined graph intelligent model to identify the upstream risk sources that have the greatest impact on the target risk node. For example, when typhoon risk, port transportation disruptions, declining coal inventories, and rising electricity loads coexist, the pre-defined graph intelligent model can identify key risk nodes that contribute significantly to the risk of thermal power supply or power supply security, and form corresponding risk paths.

[0064] In one example, the training data for the pre-defined graph intelligent model can consist of historical dynamic risk transmission graph sequences and historical risk event labels. Risk event labels can include whether a risk has occurred, the type of risk, the risk level, the affected area, the duration, the risk path, and the actual handling outcome. The training objective of the pre-defined graph intelligent model can comprehensively consider the risk classification loss, risk probability prediction loss, risk level prediction loss, risk path identification loss, and risk transmission intensity regularization term for risk nodes, enabling the pre-defined graph intelligent model to both predict the probability of risk occurrence and identify the main transmission paths of risk within the provincial energy system.

[0065] In one embodiment, such as Figure 4As shown, the steps for identifying risk nodes from multi-source data include: Step S410: Perform risk identification on multi-source data to obtain regional risk factors, energy category risk factors, energy sector risk factors, and risk objects.

[0066] Among them, regional risk factors may include risk factors such as provinces, cities, industrial parks, ports, receiving stations or key enterprises; energy category risk factors may include risk factors such as electricity, coal, natural gas, oil products, and new energy; energy link risk factors may include risk factors such as supply, transmission, storage, consumption or external events; risk objects may include risk factors such as load, inventory, price, transportation capacity (such as port transportation), power generation output, and external disaster events (such as typhoons).

[0067] Step S420: Based on regional risk factors, energy category risk factors, energy sector risk factors, and risk objects, obtain the corresponding risk nodes.

[0068] By combining four attributes—region, energy type, energy link, and risk object—corresponding risk nodes can be obtained. For example, a risk node could be located in the Pearl River Delta region, with electricity as the energy type, consumption as the energy link risk, and load as the risk object; a risk node could also be located in western Guangdong, with coal as the energy type, transmission as the energy link risk, and port transportation as the risk object; a risk node could also be located in the entire province, with natural gas as the energy type, storage as the energy link risk, and inventory as the risk object; and a risk node could also be located in a coastal area, with oil as the energy type, external events as the energy link risk, and typhoons as the risk object.

[0069] In one embodiment, the step of identifying risks from multi-source data and obtaining risk propagation edges includes: Based on the preset risk relationships, a transmission edge is constructed for the corresponding two risk nodes to obtain the corresponding risk transmission edge; among them, the preset risk relationships include physical dependence, supply and demand balance, spatial adjacency, facility association, historical co-occurrence, external influence, and time lag.

[0070] Risk transmission edges can be categorized into physical dependence edges, supply-demand linkage edges, spatial adjacency edges, facility linkage edges, historical co-occurrence edges, external impact edges, and time lag edges based on predefined risk relationships. In one example, a physical dependence edge represents the physical dependence between energy production, transmission, storage, and consumption; for instance, coal supply affects thermal power output, and natural gas supply affects gas-fired power generation. A supply-demand linkage edge represents the supply-demand balance between the supply and demand sides; for example, industrial load growth affects the tightness of power supply. A spatial adjacency edge represents the risk linkage between adjacent regions or regions with energy transmission relationships. A facility linkage edge represents the linkage and impact between facilities such as ports, pipelines, transmission lines, gas storage facilities, oil depots, and power plants. A historical co-occurrence edge represents the co-occurrence or sequential occurrence of two risk events or risk nodes in historical data. An external impact edge represents the impact of external risks such as typhoons, ice storms, high temperatures, price fluctuations, and macroeconomic changes on energy system nodes. A time lag edge represents the relationship where an upstream risk node affects a downstream risk node after a certain time delay.

[0071] In one embodiment, the step of identifying risks from multi-source data and obtaining the risk transmission strength of corresponding risk transmission edges includes: Based on the preset risk relationships, the risk relationship strength of the corresponding risk transmission edge is processed to obtain the corresponding risk relationship strength value; according to the preset weight of the corresponding risk relationship strength value and the corresponding risk relationship strength value, the risk transmission strength of the corresponding risk transmission edge is obtained.

[0072] In one example, the risk transmission intensity wij(t) from risk node i to risk node j is calculated as follows: wij(t) = α1Dij(t) + α2Cij(t) + α3Sij(t) + α4Lij(t) + α5Rij(t) + α6Bij(t).

[0073] Wherein, Dij(t) represents physical dependence strength, Cij(t) represents historical risk co-occurrence strength, Sij(t) represents spatial correlation strength, Lij(t) represents time lag impact coefficient, Rij(t) represents external risk enhancement factor, Bij(t) represents real-time business status impact factor, and α1 to α6 are preset weights. For example, physical dependence strength can be determined based on energy conversion relationships, supply chain dependencies, or business expert rules; historical co-occurrence strength can be determined based on the frequency of co-occurrence of historical risk events, conditional probability, or temporal causal relationships; spatial correlation strength can be determined based on geographical distance, energy transmission routes, regional adjacency relationships, or administrative hierarchy relationships; the time lag impact coefficient can be determined based on the historical lag time of risk transmission from upstream to downstream; and the external risk enhancement factor can be calculated based on typhoon level, rainfall intensity, duration of high temperatures, price fluctuation amplitude, etc. The real-time business status impact factor can be determined based on electricity load, coal arrival volume, natural gas supply, oil supply, energy consumption, etc.

[0074] In one example, the strength of risk transmission from one risk node to another is calculated based on physical dependence strength, historical co-occurrence strength, spatial correlation strength, time lag effect, external risk amplification factor, and real-time business status influence factor. The corresponding strength is dynamically updated as real-time operational data and external environment changes, thereby characterizing the transmission effect of risk between energy supply, transmission, storage, consumption, and external events.

[0075] In one embodiment, prior to the step of risk identification of multi-source data, the following is included: The multi-source data is preprocessed to obtain the first intermediate data; the first intermediate data is then labeled with risk factors to obtain the target multi-source data.

[0076] Preprocessing may include at least one of the following: missing value imputation, outlier identification, deduplication, unit conversion, time scale alignment, spatial scale mapping, and data format conversion. For example, performing missing value imputation, outlier identification, deduplication, unit conversion, time scale alignment, spatial scale mapping, and data format conversion on multi-source data yields first intermediate data. For example, if the multi-source data includes outliers, noise-type outliers can be preprocessed by replacement, correction, or deletion; risk signal-type outliers can be retained as risk factors and enhanced, thus obtaining the target multi-source data.

[0077] Risk factors can be categorized according to the formation mechanism of energy supply security risks. For example, risk factors may include supply risk factors, transmission risk factors, storage risk factors, consumption risk factors, market risk factors, and external risk factors. In one example, the initial screening of risk factors can combine business rules, statistical correlation, and model contribution. Business rules are used to ensure that risk factors conform to the operating mechanism of the energy system, statistical correlation is used to identify the strength of the correlation between risk factors and supply security results, and model contribution is used to measure the degree to which risk factors contribute to the output of the risk prediction model.

[0078] In one embodiment, such as Figure 5 As shown, after obtaining risk warning information based on risk information and target dynamic thresholds, the steps include: Step S510: When the risk warning information is consistent with the actual risk result, increase the weight of the corresponding risk transmission edge based on the first step value and increase the weight of the corresponding risk node based on the second step value.

[0079] Here, the first step value refers to the increment of the corresponding weight, and the second step value refers to the increment of the corresponding weight. Both the first and second step values ​​can be obtained from system presets, and can be default constants set by the system. For example, the first step value can be set to 2%, 5%, or 10%; the second step value can be set to 2%, 5%, or 10%. For example, risk warning information is matched with actual risk results. If the matching degree reaches the corresponding preset threshold, it is determined that the risk warning result is consistent with the actual risk result. Then, the weight of the relevant risk transmission edge is increased based on the first step value and the weight of the corresponding risk node is increased based on the second step value.

[0080] Step S520: When the number of information items in the risk warning information is less than the number of information items in the actual risk result, reduce the dynamic threshold of the corresponding risk node based on the third step value or increase the weight of the corresponding risk node based on the fourth step value.

[0081] For example, the number of information items in the risk warning information is compared with the number of information items in the actual risk result. If the number of information items in the risk warning information is less than the number of information items in the actual risk result, it is determined that the risk warning information has been missed. Then, the dynamic threshold of the corresponding risk node is reduced based on the third step value or the weight of the corresponding risk node is increased based on the fourth step value.

[0082] Step S530: When one of the information items in the risk warning information is inconsistent with the information items in the actual risk result, reduce the weight of the corresponding risk node based on the fifth step value or increase the dynamic threshold of the relevant risk node based on the sixth step value.

[0083] For example, the information items of the risk warning information are compared with the information items of the actual risk result. If one of the information items of the risk warning information is inconsistent with the information items of the actual risk result, it is determined that the risk warning information is a false alarm. Then, the weight of the corresponding risk node is reduced based on the fifth step value or the dynamic threshold of the relevant risk node is increased based on the sixth step value.

[0084] In the above embodiments, the system records risk warning information such as warning results, actual risk events, handling measures and handling effects, and corrects the risk node weights, risk transmission edge weights and dynamic threshold graphs based on the risk warning information, so as to realize the closed loop of early warning feedback optimization for provincial energy supply. It can further improve the accuracy and interpretability of risk warnings as the operating status of the provincial energy system and the external environment change.

[0085] It should be understood that, although Figures 2 to 5 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figures 2 to 5 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0086] In one embodiment, such as Figure 6 As shown, this application also provides a provincial energy supply risk early warning system, including: The data acquisition module 610 is used to acquire multi-source data from the corresponding provincial energy system.

[0087] The risk identification module 620 is used to identify risks in multi-source data and obtain risk nodes, node characteristics of corresponding risk nodes, risk transmission edges, and risk transmission strength of corresponding risk transmission edges.

[0088] The risk transmission graph construction module 630 is used to obtain a dynamic risk transmission graph based on risk nodes, risk transmission edges, node characteristics, and risk transmission intensity.

[0089] The graph intelligent risk processing module 640 is used to process the dynamic risk transmission graph based on the preset graph intelligent model to obtain the risk information of the corresponding risk nodes.

[0090] The dynamic threshold acquisition module 650 is used to obtain the influence factor based on multi-source data, process the influence factor based on a preset threshold algorithm, and obtain the target dynamic threshold.

[0091] The early warning output module 660 is used to obtain risk early warning information based on risk information and target dynamic thresholds.

[0092] Specific limitations regarding the provincial energy supply risk early warning system can be found in the limitations of the provincial energy supply risk early warning method described above, and will not be repeated here. Each module in the aforementioned provincial energy supply risk early warning system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the processing device in hardware form or independently of it, or stored in the memory of the processing device in software form, so that the processor can call and execute the corresponding operations of each module.

[0093] In one embodiment, such as Figure 1 As shown, this application also provides a processing device, including a memory 10 and a processor 20. The memory 10 stores a computer program, and the processor 20 executes the computer program to implement the steps of the provincial energy supply risk early warning method described above.

[0094] The processor 20 acquires multi-source data of the corresponding provincial energy system; performs risk identification on the multi-source data to obtain risk nodes, node characteristics of the corresponding risk nodes, risk transmission edges, and risk transmission intensity of the corresponding risk transmission edges; obtains a dynamic risk transmission graph based on the risk nodes, risk transmission edges, node characteristics, and risk transmission intensity; processes the dynamic risk transmission graph based on a preset graph intelligent model to obtain risk information of the corresponding risk nodes; obtains influencing factors based on the multi-source data, processes the influencing factors based on a preset threshold algorithm to obtain a target dynamic threshold; and obtains risk warning information based on the risk information and the target dynamic threshold to achieve accurate early warning of provincial energy supply security risks.

[0095] In the above embodiments, the processor 20 fully utilizes the acquired multi-source data for provincial energy supply security, and identifies risk nodes, risk transmission edges, risk node characteristics, and risk transmission intensity based on the multi-source data. Then, based on the risk nodes, risk transmission edges, risk node characteristics, and risk transmission intensity, it constructs a dynamic risk transmission map that reflects the risk correlation and transmission path of provincial energy supply security. The dynamic risk transmission map is input into a preset intelligent model, and the risk information of each risk node is output. Combined with a dynamic threshold mechanism, it can identify risks that have not exceeded the threshold but have a deterioration trend and the possibility of propagation in advance, and then output risk warning information to provide reliable early warning for provincial energy supply security and emergency response, thereby improving the reliability and accuracy of the processing equipment in providing early warning of provincial energy supply security risks.

[0096] In one embodiment, this application also provides a computer storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the steps of the provincial energy supply risk early warning method described above.

[0097] For example, when a computer program is executed by a processor, it performs the following steps: The system acquires multi-source data of the corresponding provincial energy system; identifies risks in the multi-source data to obtain risk nodes, node characteristics of corresponding risk nodes, risk transmission edges, and risk transmission intensity of corresponding risk transmission edges; generates a dynamic risk transmission map based on risk nodes, risk transmission edges, node characteristics, and risk transmission intensity; processes the dynamic risk transmission map based on a preset intelligent model to obtain risk information of corresponding risk nodes; obtains influencing factors based on multi-source data, processes the influencing factors based on a preset threshold algorithm to obtain target dynamic thresholds; and obtains risk warning information based on risk information and target dynamic thresholds to achieve accurate early warning of provincial energy supply security risks.

[0098] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the division operations described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), direct memory bus RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0099] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0100] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for early warning of energy supply risks in a province, characterized in that, include: Obtain multi-source data from the corresponding provincial energy system; Risk identification is performed on the multi-source data to obtain risk nodes, node characteristics corresponding to the risk nodes, risk transmission edges, and risk transmission strength corresponding to the risk transmission edges; A dynamic risk transmission graph is obtained based on the risk nodes, the risk transmission edges, the node characteristics, and the risk transmission intensity. The dynamic risk transmission graph is processed based on a preset graph intelligent model to obtain the risk information of the corresponding risk nodes; Based on the multi-source data, an impact factor is obtained. The impact factor is then processed using a preset threshold algorithm to obtain a target dynamic threshold. Risk warning information is obtained based on the risk information and the target dynamic threshold.

2. The provincial energy supply risk early warning method according to claim 1, characterized in that, The step of obtaining risk warning information based on the risk information and the target dynamic threshold includes: Based on the risk information and the target dynamic threshold, the actual risk information and / or potential risk information of the corresponding risk node are obtained; The risk warning information is obtained based on the actual risk information and / or the potential risk information.

3. The provincial energy supply risk early warning method according to claim 2, characterized in that, The risk information includes node risk probability and risk path; The step of obtaining the actual risk information and / or the potential risk information of the corresponding risk node based on the risk information and the target dynamic threshold includes: When the node risk probability of the corresponding risk node is greater than or equal to the target dynamic threshold, the actual risk information of the corresponding risk node is obtained. When the node risk probability of the corresponding risk node is less than the target dynamic threshold, and the corresponding risk node meets the preset potential risk conditions, the potential risk information of the corresponding risk node is obtained; wherein, the preset potential risk conditions include at least one of the following: the node risk probability gradually increases in at least two consecutive preset time windows and the upward slope of the corresponding node risk probability exceeds the preset trend threshold; the difference between the node risk probability of the corresponding risk node and the dynamic threshold is less than the preset proximity threshold; the corresponding risk node has a risk path from the upstream risk node; the upstream risk node or the same type of adjacent risk node of the corresponding risk node has been confirmed as actual risk information.

4. The provincial energy supply risk early warning method according to claim 1, characterized in that, The influencing factors include historical risk distribution information, operational scenario information, intensity of external risk events, and risk tolerance. The step of processing the influencing factors based on a preset threshold algorithm to obtain the target dynamic threshold includes: Based on the operational scenario information, the historical risk distribution information, the intensity of external risk events, and the risk tolerance are input into a preset threshold algorithm for processing to obtain the target dynamic threshold.

5. The provincial energy supply risk early warning method according to claim 1, characterized in that, The preset graph intelligent model includes at least one of the following network models: graph attention network model, graph convolutional network model, spatiotemporal graph convolutional network model, graph convolutional recurrent neural network model, and relation-aware graph neural network model.

6. The provincial energy supply risk early warning method according to claim 1, characterized in that, The step of identifying risk nodes from the multi-source data includes: Risk identification is performed on the multi-source data to obtain regional risk factors, energy category risk factors, energy sector risk factors, and risk objects. Based on the regional risk factor, the energy category risk factor, the energy sector risk factor, and the risk object, the corresponding risk node is obtained.

7. The provincial energy supply risk early warning method according to claim 6, characterized in that, The step of identifying risks from the multi-source data and obtaining risk propagation edges includes: Based on the preset risk relationships, a transmission edge is constructed for the corresponding two risk nodes to obtain the corresponding risk transmission edge; wherein, the preset risk relationships include physical dependency relationships, supply and demand balance relationships, spatial adjacency relationships, facility association relationships, historical co-occurrence relationships, external influence relationships, and time lag relationships.

8. The provincial energy supply risk early warning method according to claim 7, characterized in that, The step of identifying risks in the multi-source data and obtaining the risk transmission strength of the corresponding risk transmission edge includes: Based on the preset risk relationship, the risk relationship strength is processed on the corresponding risk transmission edge to obtain the corresponding risk relationship strength value; The risk transmission strength of the corresponding risk transmission edge is obtained based on the corresponding risk relationship strength value and the preset weight of the corresponding risk relationship strength value.

9. The provincial energy supply risk early warning method according to claim 1, characterized in that, Prior to the step of risk identification of the multi-source data, the following steps are included: The multi-source data is preprocessed to obtain the first intermediate data; Risk factors are labeled on the first intermediate data to obtain target multi-source data.

10. The provincial energy supply security risk early warning method according to any one of claims 1 to 9, characterized in that, After the step of obtaining risk warning information based on the risk information and the target dynamic threshold, the following steps are included: When the risk warning information is consistent with the actual risk result, the weight of the corresponding risk transmission edge is increased based on the first step value and the weight of the corresponding risk node is increased based on the second step value. When the number of information items in the risk warning information is less than the number of information items in the actual risk result, the dynamic threshold of the corresponding risk node is reduced based on the third step value or the weight of the corresponding risk node is increased based on the fourth step value. When one of the information items in the risk warning information is inconsistent with the information items in the actual risk result, the weight of the corresponding risk node is reduced based on the fifth step value or the dynamic threshold of the relevant risk node is increased based on the sixth step value.

11. A provincial energy supply risk early warning system, characterized in that, include: The data acquisition module is used to acquire multi-source data from the corresponding provincial energy system; The risk identification module is used to identify risks in the multi-source data and obtain risk nodes, node features corresponding to the risk nodes, risk transmission edges, and risk transmission strength corresponding to the risk transmission edges. The risk transmission graph construction module is used to obtain a dynamic risk transmission graph based on the risk nodes, the risk transmission edges, the node characteristics, and the risk transmission intensity. The graph intelligent risk processing module is used to process the dynamic risk transmission graph based on a preset graph intelligent model to obtain the risk information of the corresponding risk nodes. The dynamic threshold acquisition module is used to obtain the influence factor based on the multi-source data, process the influence factor based on the preset threshold algorithm, and obtain the target dynamic threshold. The early warning output module is used to obtain risk early warning information based on the risk information and the target dynamic threshold.

12. A processing device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the provincial energy supply risk early warning method according to any one of claims 1 to 10.

Images