Power distribution equipment temperature monitoring and early warning method and system based on edge calculation

Abstract

The invention provides a power distribution equipment temperature monitoring and early warning method and system based on edge calculation, and relates to the technical field of power grids, and the method comprises the steps: obtaining equipment temperature and operation data through a monitoring unit, building an equipment heat influence propagation atlas structure based on temperature change synchronism, monitoring temperature abnormity in real time, and tracing source equipment. And distributing regulation and control weights according to the heat influence propagation intensity, generating and executing a distributed regulation and control instruction, and realizing heat influence propagation minimization. The method can improve the temperature anomaly monitoring precision of the power distribution system, reduce the fault risk and prolong the service life of equipment.

Description

Technical Field

[0001] This application relates to the field of power grid technology, and more specifically, to a method and system for temperature monitoring and early warning of power distribution equipment based on edge computing. Background Technology

[0002] With the continuous development of power systems and the constant improvement of power grid construction, the safe and stable operation of power distribution systems plays a crucial role in the reliability of power supply. The temperature status of power distribution equipment is an important indicator reflecting its health condition and potential failure risks. When power distribution equipment exhibits abnormal heating, it usually means that the equipment may have problems such as poor contact, insulation aging, or overload. If not dealt with in time, it may lead to equipment damage or even power outages.

[0003] Traditionally, temperature monitoring of power distribution equipment relies on periodic manual inspections or fixed temperature sensors. With the development of IoT technology, more and more power distribution equipment is equipped with real-time monitoring systems that continuously collect temperature data and operational status. This data is transmitted via communication networks to a central monitoring center for centralized analysis and processing, providing a basis for equipment status assessment and fault early warning.

[0004] Existing temperature monitoring systems mostly adopt a centralized data processing architecture, transmitting large amounts of data to remote data centers for analysis. This not only increases the burden on communication bandwidth but also leads to increased data processing delays, making it impossible to respond to abnormal temperature conditions in a timely manner and reducing the timeliness of fault warnings.

[0005] Secondly, traditional temperature monitoring systems often treat each power distribution device as an independent entity for monitoring, ignoring the heat transfer relationship between devices within the power distribution system. This makes it difficult to effectively identify the true source of temperature anomalies, leading to difficulties in fault tracing and low maintenance efficiency. Summary of the Invention

[0006] This invention provides a method and system for temperature monitoring and early warning of power distribution equipment based on edge computing, which can solve the problems in the prior art.

[0007] A first aspect of the present invention provides a method for temperature monitoring and early warning of power distribution equipment based on edge computing, comprising: By using a monitoring unit deployed in the power distribution system, temperature data and operating status data of multiple power distribution devices are acquired. The operating status data includes the load status and operating mode of the devices. Based on the synchronicity of the temperature data changes over time, the propagation path of thermal effects between power distribution equipment is identified, and a graph structure describing the intensity of thermal correlation between equipment is established. In the graph structure, nodes represent power distribution equipment, and the weights of the edges represent the intensity of thermal effects propagation between equipment. The temperature data is monitored at the edge computing node. When a temperature anomaly is detected, the source device causing the temperature anomaly is located by tracing the direction of heat propagation in reverse based on the heat effect propagation path in the graph structure. Based on the intensity of heat effect propagation from the source device to each associated device in the graph structure, control weights are assigned to the source device and each associated device. Based on the control weights and the current load state of each device, a distributed control command is generated to minimize the total amount of heat effect propagation. Based on the distributed control command, load parameters are adjusted for the source device and each associated device.

[0008] Based on the synchronicity of temperature data changes over time, the propagation paths of thermal effects between power distribution equipment are identified, and a spectral structure describing the intensity of thermal correlation between equipment is established, including: Obtain the topology information of the power distribution equipment, wherein the topology information describes the electrical connection relationship and physical adjacency relationship between the power distribution equipment; Based on the topology information, a set of candidate thermal effect propagation paths is determined. The set of candidate thermal effect propagation paths includes only device pairs that have electrical or physical connections in the topology. A constraint analysis is performed on the synchronicity of the temperature data changes over time. The temperature change synchronicity index is calculated only for the device pairs in the set of candidate thermal effect propagation paths. By constructing a heat flow propagation equation, using the temperature change synchronicity index as the boundary condition of the heat flow propagation equation, the heat flow propagation intensity on each candidate path is solved, and the heat flow propagation intensity is used as the weight of the edge in the graph structure. The heat flow propagation equation describes the process of heat propagation from the source device to the target device along the candidate heat influence propagation path. Based on the heat flow propagation intensity, a device thermal correlation map integrating topological constraints and temperature synchronization is established.

[0009] By constructing a heat flow propagation equation and using the temperature change synchronicity index as the boundary condition of the heat flow propagation equation, the heat flow propagation intensity on each candidate path is solved, including: Obtain the physical parameters between the source device and the target device on each candidate path. The physical parameters include the distance between the devices, the thermal conductivity of the device material, and the thermal resistance of the medium between the devices. Based on the physical parameters, a heat flow propagation equation is established to describe the heat propagation process between the source device and the target device. The heat flow propagation equation describes the spatial distribution and temporal evolution of heat along the candidate path by introducing heat conduction and heat convection terms, and converts the temperature change synchronicity index corresponding to each candidate path into the temperature boundary condition of the heat flow propagation equation at the target device. The heat flow propagation equation is numerically solved to obtain the heat flow density distribution from the source device to the target device on each candidate path. Based on the heat flow density distribution, the total heat flow transmitted through the candidate path per unit time is calculated, and the total heat flow is used as the heat flow propagation intensity of the candidate path.

[0010] Based on the heat propagation path in the aforementioned graph structure, the source device causing the temperature anomaly can be located by tracing the heat propagation direction in reverse. When an edge computing node detects a temperature anomaly, the target device with the temperature anomaly is designated as the current tracing node, and the tracing path set is initialized. Multi-level backtracking is performed from the graph structure along the reverse direction of the heat effect propagation path. Each backtracking retrieves all the precursor devices pointing to the current tracing node and adds the precursor devices to the tracing path set. The backtracking operation is repeated until the initial node without precursor devices is reached or the backtracking level reaches a preset level threshold, forming a complete tracing path set containing multi-level heat effect propagation links. For each tracing path in the set of tracing paths, the cumulative transmission amount of temperature change of each device on the path along the direction of heat effect propagation is calculated. The cumulative transmission amount is obtained by multiplying the temperature change of the starting device of the path by the heat effect propagation intensity of each segment of the path in turn and accumulating them. The starting device of the tracing path whose cumulative transmission amount is closest to the temperature anomaly amplitude of the target device is determined as the source device.

[0011] Based on the heat influence propagation intensity from the source device to each associated device in the aforementioned spectrum structure, control weights are assigned to the source device and each associated device, including: For each heat impact propagation path, the heat impact propagation intensity of each segment between the source device at the starting point of the path and the associated device at the end point of the path is multiplied together to obtain the total propagation intensity of the heat impact propagation path. The total propagation intensity of all heat impact propagation paths from the source device to the same associated device is summed to obtain the comprehensive heat impact propagation intensity of the source device to the associated device. Based on the overall thermal impact propagation intensity of the source device on each associated device, the proportion of the overall thermal impact propagation intensity of each associated device to the total overall thermal impact propagation intensity of all associated devices is calculated. This proportion is used as the control weight of the corresponding associated device, and the control weight of the source device is set as the difference between the sum of the control weights of all associated devices and the preset total weight.

[0012] Based on the aforementioned control weights and the current load status of each device, the distributed control instructions that minimize the total heat impact propagation include: A distributed control optimization model is constructed with the goal of minimizing the total heat impact propagation. The distributed control optimization model takes the load adjustment of the source device and each associated device as decision variables, the sum of the products of the control weight of each device and the load adjustment of the corresponding device as the expression for the total heat impact propagation in the optimization objective function, and the load adjustment of each device is limited to the adjustable load range as a constraint condition. The distributed control optimization model is solved to obtain the optimal load adjustment amount of the source device and each associated device that minimizes the optimization objective function. Based on the optimal load adjustment amount of each device and the real-time operating power in the current load state, the target operating power of each device after regulation is calculated, and the target operating power is encapsulated into a distributed control instruction for the corresponding device.

[0013] A second aspect of the present invention provides a temperature monitoring and early warning system for power distribution equipment based on edge computing, comprising: The first unit is used to acquire temperature data and operating status data of multiple power distribution devices through a monitoring unit deployed in the power distribution system. The operating status data includes the load status and operating mode of the devices. The second unit is used to identify the thermal impact propagation path between power distribution equipment based on the synchronicity of the temperature data changes over time, and to establish a graph structure describing the thermal correlation strength between the equipment. In the graph structure, nodes represent power distribution equipment, and the weights of the edges represent the thermal impact propagation strength between the equipment. The third unit is used to monitor the temperature data at the edge computing node. When a temperature anomaly is detected, it locates the source device of the temperature anomaly by tracing the direction of heat propagation in reverse based on the heat propagation path in the graph structure. The fourth unit is used to assign control weights to the source device and each associated device according to the heat effect propagation intensity from the source device to each associated device in the graph structure, and to generate a distributed control command that minimizes the total heat effect propagation based on the control weights and the current load state of each device; The fifth unit is used to adjust the load parameters of the source device and each associated device according to the distributed control command.

[0014] A third aspect of the present invention, An electronic device is provided, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0015] Fourth aspect of the embodiments of the present invention, A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0016] The beneficial effects of this application are as follows: Based on the synchronicity of temperature data changes over time, this invention identifies the propagation path of thermal effects between power distribution equipment and constructs a thermal correlation intensity map. This enables the system to accurately identify the source equipment of temperature anomalies, improves the accuracy of temperature anomaly location, and avoids misjudgments caused by relying solely on single-point measurements in traditional methods.

[0017] This invention implements monitoring and control at edge computing nodes, reducing data transmission latency and enabling real-time response to temperature anomalies. At the same time, by establishing a distributed control mechanism based on the intensity of thermal effect propagation, it can adjust load parameters differently for source devices and related devices, effectively preventing the spread of temperature anomalies.

[0018] This invention employs an optimization strategy based on minimizing the total amount of heat impact propagation. While ensuring the safe operation of the system, it maximizes the power supply capacity of the power distribution system, avoiding the power supply capacity loss caused by simply shutting down or drastically reducing the load on abnormal equipment in traditional methods, and improving the operating efficiency and reliability of the power distribution system. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 This is a flowchart illustrating the edge computing-based temperature monitoring and early warning method for power distribution equipment according to an embodiment of the present invention. Detailed Implementation

[0020] 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.

[0021] Figure 1 This is a flowchart illustrating the edge computing-based temperature monitoring and early warning method for power distribution equipment according to an embodiment of the present invention. Figure 1 As shown, the method includes: By using a monitoring unit deployed in the power distribution system, temperature data and operating status data of multiple power distribution devices are acquired. The operating status data includes the load status and operating mode of the devices. Based on the synchronicity of the temperature data changes over time, the propagation path of thermal effects between power distribution equipment is identified, and a graph structure describing the intensity of thermal correlation between equipment is established. In the graph structure, nodes represent power distribution equipment, and the weights of the edges represent the intensity of thermal effects propagation between equipment. The temperature data is monitored at the edge computing node. When a temperature anomaly is detected, the source device causing the temperature anomaly is located by tracing the direction of heat propagation in reverse based on the heat effect propagation path in the graph structure. Based on the intensity of heat effect propagation from the source device to each associated device in the graph structure, control weights are assigned to the source device and each associated device. Based on the control weights and the current load state of each device, a distributed control command is generated to minimize the total amount of heat effect propagation. Based on the distributed control command, load parameters are adjusted for the source device and each associated device.

[0022] In one optional implementation, based on the synchronicity of the temperature data changes over time, identifying the thermal impact propagation path between power distribution equipment and establishing a spectral structure describing the intensity of thermal correlation between equipment includes: Obtain the topology information of the power distribution equipment, wherein the topology information describes the electrical connection relationship and physical adjacency relationship between the power distribution equipment; Based on the topology information, a set of candidate thermal effect propagation paths is determined. The set of candidate thermal effect propagation paths includes only device pairs that have electrical or physical connections in the topology. A constraint analysis is performed on the synchronicity of the temperature data changes over time. The temperature change synchronicity index is calculated only for the device pairs in the set of candidate thermal effect propagation paths. By constructing a heat flow propagation equation, using the temperature change synchronicity index as the boundary condition of the heat flow propagation equation, the heat flow propagation intensity on each candidate path is solved, and the heat flow propagation intensity is used as the weight of the edge in the graph structure. The heat flow propagation equation describes the process of heat propagation from the source device to the target device along the candidate heat influence propagation path. Based on the heat flow propagation intensity, a device thermal correlation map integrating topological constraints and temperature synchronization is established.

[0023] Topology information includes the electrical connections and physical adjacencies between devices. For example, in a substation, device A and device B are connected by a conductor, indicating an electrical connection; device C and device D are physically adjacent, with a distance less than a preset threshold (e.g., 50 cm), indicating a physical adjacency. This information can be obtained through power distribution system design drawings, on-site measurement data, or 3D modeling. The acquired topology information is stored as an adjacency matrix, where each element represents a relationship between devices: 1 indicates a connection, and 0 indicates no connection.

[0024] Based on the acquired topology information, a set of candidate thermal effect propagation paths is determined. This set includes only device pairs that are electrically connected or physically adjacent in the topology. For example, for circuit breaker X and bus Y in a substation, if they are electrically connected, (X,Y) is included in the set as a candidate thermal effect propagation path; if the distance between circuit breaker Z and bus Y is 30 cm (less than the preset threshold of 50 cm), (Z,Y) is also included in the candidate set. This screening can significantly reduce the computational workload of subsequent analysis and avoid unnecessary calculations for device pairs that obviously cannot have thermal effects.

[0025] For each pair of devices in the determined set of candidate thermal effect propagation paths, calculate their temperature change synchronicity index. The temperature change synchronicity index reflects the degree of correlation between the temperature changes of the two devices. In practical implementation, the following steps can be used for calculation: acquire the temperature time series data of the devices; preprocess the temperature data, including outlier removal, missing value imputation, and data smoothing; calculate the temperature change rate sequence between the device pairs, such as the temperature change rates of transformer A and circuit breaker B over 30 consecutive minutes being {0.5℃ / min, 0.7℃ / min, 0.3℃ / min...} and {0.4℃ / min, 0.65℃ / min, 0.25℃ / min...}, respectively; calculate the synchronicity index based on the temperature change rate sequence, using methods such as cross-correlation coefficients, Granger causality tests, or information entropy. For example, the calculated temperature change synchronicity index for transformer A and circuit breaker B is 0.85.

[0026] Next, the heat flow propagation equation is constructed. This equation describes the propagation of heat from the source device to the target device, taking into account the physical distance between the devices, the thermal conductivity of the materials, and environmental factors. In this equation, the previously calculated temperature change synchronicity index is used as a boundary condition. For example, for transformer A and circuit breaker B, a synchronicity index of 0.85 is used as the constraint for heat flow propagation between them. Combined with conditions such as a device distance of 25 cm, air as the intermediate medium, and an ambient temperature of 35°C, the heat flow propagation equation is solved.

[0027] By solving the heat flow propagation equation, the heat flow propagation intensity along each candidate path is obtained. This intensity value reflects the degree to which a temperature change in one device affects another. For example, the calculated heat flow propagation intensity from transformer A to circuit breaker B is 0.72, while the heat flow propagation intensity from circuit breaker B to transformer A is 0.25. This indicates that the thermal influence of transformer A on circuit breaker B is significantly greater than the reverse influence.

[0028] Finally, based on the calculated heat flow propagation intensity, a thermal correlation graph of the equipment is established. This graph is a directed weighted graph, where nodes represent power distribution equipment, edges represent heat propagation paths, and the weight of each edge is the corresponding heat flow propagation intensity. The visualized graph clearly shows the thermal relationship between equipment; for example, the edge weight from transformer A to circuit breaker B is 0.72, and the edge weight from circuit breaker B to transformer A is 0.25. Technicians can use this graph to analyze hotspot equipment and identify heat propagation paths, providing a basis for temperature monitoring and thermal management of the power distribution system.

[0029] In practical applications, a thermal correlation diagram constructed using the above method for 10 pieces of equipment in a substation (including 3 transformers, 4 circuit breakers, and 3 busbars) shows that transformer No. 1 is the main heat source. Its heat propagation path primarily spreads outward through circuit breaker No. 2, which is directly connected to it, with a propagation intensity of 0.83. It then sequentially affects busbar No. 1 (propagation intensity 0.67), circuit breaker No. 3 (propagation intensity 0.55), and other equipment. Furthermore, the diagram reveals that although circuit breaker No. 4 is physically adjacent to busbar No. 2 (only 15 cm away), its heat propagation intensity is only 0.21, indicating that physical proximity does not necessarily lead to a strong thermal impact. Based on this thermal correlation diagram, maintenance personnel can prioritize monitoring the heat dissipation of transformer No. 1 and add a temperature monitoring point at circuit breaker No. 2 to promptly detect any abnormal thermal phenomena.

[0030] In one optional implementation, by constructing a heat flow propagation equation and using the temperature change synchronicity index as a boundary condition for the heat flow propagation equation, the heat flow propagation intensity on each candidate path is solved, including: Obtain the physical parameters between the source device and the target device on each candidate path. The physical parameters include the distance between the devices, the thermal conductivity of the device material, and the thermal resistance of the medium between the devices. Based on the physical parameters, a heat flow propagation equation is established to describe the heat propagation process between the source device and the target device. The heat flow propagation equation describes the spatial distribution and temporal evolution of heat along the candidate path by introducing heat conduction and heat convection terms, and converts the temperature change synchronicity index corresponding to each candidate path into the temperature boundary condition of the heat flow propagation equation at the target device. The heat flow propagation equation is numerically solved to obtain the heat flow density distribution from the source device to the target device on each candidate path. Based on the heat flow density distribution, the total heat flow transmitted through the candidate path per unit time is calculated, and the total heat flow is used as the heat flow propagation intensity of the candidate path.

[0031] The equipment monitoring system first acquires the physical parameters between the source and target devices on each candidate path, including the distance between devices, the thermal conductivity of the device materials, and the thermal resistance of the medium between devices. For a given candidate path, the system records that the straight-line distance between source device A and target device B is 120 cm, device A's casing material is aluminum alloy with a thermal conductivity of 237 watts / meter-Kelvin, device B's casing material is stainless steel with a thermal conductivity of 16 watts / meter-Kelvin, and the medium between the two devices is air with a thermal resistance of 0.024 square meters-Kelvin / watt.

[0032] Based on the acquired physical parameters, the system establishes a heat flow propagation equation describing the heat propagation process between the source device and the target device. This heat flow propagation equation describes the spatial distribution and temporal evolution of heat along the candidate path by introducing heat conduction and heat convection terms. The heat conduction term describes the process of heat transfer through direct contact within a solid material and is proportional to the material's thermal conductivity. The heat convection term describes the process of heat transfer through the flow of a fluid medium and is related to the fluid's density, specific heat capacity, and flow velocity. In this embodiment, the system sets the initial temperature of the source device A to 85 degrees Celsius, the ambient temperature to 25 degrees Celsius, the air velocity to 0.5 m / s, the density to 1.293 kg / m³, and the specific heat capacity to 1005 joules / kg-Kelvin.

[0033] The system converts the temperature change synchronicity index corresponding to each candidate path into temperature boundary conditions for the heat flow propagation equation at the target device. Taking a specific candidate path as an example, the temperature change synchronicity index value for this path is 0.78, indicating a high correlation between the temperature changes of source device A and target device B. The system converts this synchronicity index into temperature boundary conditions at target device B, specifically: when the source temperature increases by 10 degrees Celsius, the target temperature increases by 7.8 degrees Celsius. This boundary condition setting ensures that the solution to the heat flow equation reflects the observed synchronous temperature change characteristics.

[0034] When numerically solving the heat flow propagation equation, the system uses the finite difference method to discretize the continuous heat flow equation. Spatially, the distance between source device A and target device B is divided into 120 grid points, with each grid point spaced 1 cm apart. Temporally, a time step of 0.1 seconds is used to simulate the heat propagation process over 10 minutes, for a total of 6000 time steps. At each time step, the system uses an explicit difference scheme to calculate the updated temperature value for each grid point and applies the set boundary conditions to constrain the solution process.

[0035] To ensure the stability of the numerical solution, the system ensures that the difference scheme satisfies the Courant-Friedrich-Lévy condition, i.e., the ratio of the time step to the spatial step does not exceed twice the thermal diffusivity. In actual calculations, the system selects a combination of a time step of 0.1 seconds and a spatial step of 1 centimeter, which keeps the numerical solution stable and has sufficient accuracy.

[0036] Through numerical solutions, the system obtains the heat flux density distribution from the source device to the target device along each candidate path. For the candidate path in the example, under steady-state conditions, the heat flux density near the source device A is 378 watts per square meter, gradually decreasing with increasing distance, dropping to 42 watts per square meter near the target device B. This heat flux density distribution reflects that heat is mainly transferred within the device materials through thermal conduction and propagates through convection in the air between the two devices.

[0037] Based on the heat flux density distribution, the system calculates the total heat flow transferred through the candidate path per unit time. The calculation method involves multiplying the heat flux density at each point on the path by the corresponding cross-sectional area perpendicular to the heat flux direction, and then integrating along the path. For the candidate path in the example, assuming the effective cross-sectional area for heat flux transfer is 0.02 square meters, the total heat flow transferred through this path per unit time is 3.8 watts. The system uses this total heat flow as the heat flux propagation intensity of this candidate path.

[0038] The system repeats the above calculation process for multiple candidate paths in the network. For example, for another candidate path, the distance between source device C and target device D is 80 cm, the temperature change synchronicity index is 0.45, and the calculated heat flow propagation intensity is 1.6 watts. By comparing the heat flow propagation intensities of different candidate paths, the system can identify the most efficient heat transfer path, providing a basis for subsequent fault propagation path prediction.

[0039] In one optional implementation, based on the heat effect propagation path in the graph structure, the source device causing the temperature anomaly is located by tracing the heat effect propagation direction in reverse, including: When an edge computing node detects a temperature anomaly, the target device with the temperature anomaly is designated as the current tracing node, and the tracing path set is initialized. Multi-level backtracking is performed from the graph structure along the reverse direction of the heat effect propagation path. Each backtracking retrieves all the precursor devices pointing to the current tracing node and adds the precursor devices to the tracing path set. The backtracking operation is repeated until the initial node without precursor devices is reached or the backtracking level reaches a preset level threshold, forming a complete tracing path set containing multi-level heat effect propagation links. For each tracing path in the set of tracing paths, the cumulative transmission amount of temperature change of each device on the path along the direction of heat effect propagation is calculated. The cumulative transmission amount is obtained by multiplying the temperature change of the starting device of the path by the heat effect propagation intensity of each segment of the path in turn and accumulating them. The starting device of the tracing path whose cumulative transmission amount is closest to the temperature anomaly amplitude of the target device is determined as the source device.

[0040] In a data center environment, edge computing nodes are equipped with temperature monitoring sensors to collect temperature data from each device in real time. The system pre-constructs a graph structure reflecting the thermal impact relationships between devices, which includes nodes (representing devices) and edges (representing the propagation paths and intensity of thermal impact). When the temperature of a device exceeds a preset threshold, the system triggers a temperature anomaly tracing process.

[0041] When temperature anomaly tracing begins, the system marks the target device experiencing the temperature anomaly as the current tracing node and initializes an empty tracing path set to store all possible tracing paths discovered subsequently. For example, when the temperature of server S1 in rack A reaches 75°C, exceeding the normal operating temperature threshold of 65°C, the system sets S1 as the current tracing node.

[0042] The system queries the graph structure for all predecessor devices pointing to the current source node. These predecessor devices are upstream devices that may affect the temperature of the current node. For server S1, the system may find three predecessor devices: server S2 in the same rack, rack heat dissipation unit R1, and heat dissipation unit R2 in the rack above. The heat transfer intensities between them and S1 are 0.65, 0.45, and 0.25, respectively, representing the efficiency or proportion of heat transfer.

[0043] The system adds these three precursor devices to the tracing path set, forming three initial tracing paths: S2→S1, R1→S1, and R2→S1. Next, using these three precursor devices as new tracing nodes, the system continues the backtracking operation to find their respective precursor devices.

[0044] For server S2, the system identifies its predecessor devices as server S3 (heat-affected zone intensity 0.55) and cooling system C1 (heat-affected zone intensity 0.75); for heat dissipation unit R1, its predecessor device is found to be cooling system C1 (heat-affected zone intensity 0.85); for heat dissipation unit R2, its predecessor device is another cooling system C2 (heat-affected zone intensity 0.7). The system updates the source tracing path set, forming multiple secondary backtracking paths: S3→S2→S1, C1→S2→S1, C1→R1→S1, and C2→R2→S1.

[0045] The system continues the backtracking operation, querying the predecessor devices of S3, C1, and C2. Assume the predecessor device of S3 is power supply unit P1 (thermal affected intensity 0.5); the predecessor device of C1 is controller CT1 (thermal affected intensity 0.4); and C2 has no predecessor device. The system updates the backtracking path set, forming multiple three-level backtracking paths: P1→S3→S2→S1, CT1→C1→S2→S1, CT1→C1→R1→S1, and C2→R2→S1.

[0046] Assuming the backtracking operation continues, it is found that the predecessor device of P1 is power distribution unit PD1 (thermal impact intensity 0.35), while CT1 has no predecessor device. The system updates the backtracking path set, forming the final backtracking paths: PD1→P1→S3→S2→S1, CT1→C1→S2→S1, CT1→C1→R1→S1, and C2→R2→S1. Since all paths have reached the initial node without a predecessor device or have reached the preset backtracking level threshold (e.g., level 4 in this example), the backtracking process ends.

[0047] The system then calculates the cumulative temperature change propagation of each device along each tracing path. Taking the PD1→P1→S3→S2→S1 path as an example, assuming a temperature increase of 15℃ is detected in PD1, this temperature change is propagated to S1 through the heat effect propagation path. The cumulative propagation calculation process is as follows: multiply the 15℃ temperature change in PD1 by the heat effect intensity from PD1 to P1 (0.35), resulting in a temperature effect propagated to P1 of 5.25℃; multiply this 5.25℃ by the heat effect intensity from P1 to S3 (0.5), resulting in a temperature effect propagated to S3 of 2.63℃; multiply this 2.63℃ by the heat effect intensity from S3 to S2 (0.55), resulting in a temperature effect propagated to S2 of 1.44℃; finally, multiply this 1.44℃ by the heat effect intensity from S2 to S1 (0.65), resulting in a temperature effect propagated to S1 of 0.94℃.

[0048] Similarly, the system calculates the cumulative transfer amount for other paths: In the CT1→C1→S2→S1 path, if the temperature of CT1 rises by 20℃, the temperature impact transferred to S1 is 20×0.4×0.75×0.65=3.9℃; In the CT1→C1→R1→S1 path, if the temperature of CT1 rises by 20℃, the temperature impact transferred to S1 is 20×0.4×0.85×0.45=3.06℃; In the C2→R2→S1 path, if the temperature of C2 rises by 18℃, the temperature impact transferred to S1 is 18×0.7×0.25=3.15℃.

[0049] The actual temperature anomaly of server S1 was 10℃ (75℃-65℃). Comparing the cumulative heat transfer along each path with the actual temperature anomaly revealed that several devices contributed to the abnormal temperature but could not individually explain the entire anomaly. Further analysis showed that the sum of the cumulative heat transfer along the CT1→C1→S2→S1 path (3.9℃) and the cumulative heat transfer along the C2→R2→S1 path (3.15℃) was 7.05℃. Adding this to the 2.95℃ temperature rise due to increased load on S1 itself, this precisely explained the 10℃ anomaly. Therefore, the system identified CT1 and C2 as the primary sources of heat impact, and maintenance personnel were instructed to prioritize checking these two devices and their associated cooling system malfunctions.

[0050] This method enables the system to identify not only temperature anomalies caused by a single heat source, but also temperature problems caused by multiple combined factors, thus improving the accuracy and efficiency of root cause analysis of temperature anomalies in data centers.

[0051] In one optional implementation, assigning control weights to the source device and each associated device based on the heat effect propagation intensity from the source device to each associated device in the spectrum structure includes: For each heat impact propagation path, the heat impact propagation intensity of each segment between the source device at the starting point of the path and the associated device at the end point of the path is multiplied together to obtain the total propagation intensity of the heat impact propagation path. The total propagation intensity of all heat impact propagation paths from the source device to the same associated device is summed to obtain the comprehensive heat impact propagation intensity of the source device to the associated device. Based on the overall thermal impact propagation intensity of the source device on each associated device, the proportion of the overall thermal impact propagation intensity of each associated device to the total overall thermal impact propagation intensity of all associated devices is calculated. This proportion is used as the control weight of the corresponding associated device, and the control weight of the source device is set as the difference between the sum of the control weights of all associated devices and the preset total weight.

[0052] In this embodiment, in order to better control the thermal impact, it is necessary to assign reasonable control weights to the source device and each associated device based on the intensity of thermal impact propagation from the source device to each associated device.

[0053] Thermal impacts between devices typically propagate through multiple paths, each consisting of multiple devices, with adjacent devices exhibiting thermal propagation relationships. Taking a data center as an example, server A, as the source device, may affect related devices server B, server C, and server D through multiple paths. The process of allocating control weights is described in detail below.

[0054] First, we analyze each propagation path in the heat impact propagation map. Assume there are two different heat impact propagation paths from the source device server A to the associated device server D: Path 1 is server A → server B → server D, and Path 2 is server A → server C → server D. For Path 1, the heat impact propagation intensity from device A to B is 0.8, and from device B to D is 0.6; for Path 2, the heat impact propagation intensity from device A to C is 0.7, and from device C to D is 0.5.

[0055] For path 1, when calculating the total propagation intensity, the thermal impact propagation intensity of each segment along the path is multiplied together: 0.8 × 0.6 = 0.48. Similarly, the total propagation intensity of path 2 is 0.7 × 0.5 = 0.35. The combined thermal impact propagation intensity of source device A on associated device D is the sum of the total propagation intensities of the two paths, i.e., 0.48 + 0.35 = 0.83.

[0056] The same method is used to analyze the impact of source device A on other related devices. Assume the overall thermal impact propagation intensity of source device A on related device B is 0.8 (direct path), and on related device C it is 0.7 (direct path). The total overall thermal impact propagation intensity of all related devices is 0.8 + 0.7 + 0.83 = 2.33.

[0057] When allocating control weights, the calculation is based on the proportion of the overall thermal impact propagation intensity of each associated device to the total. The control weight for associated device B is 0.8 ÷ 2.33 ≈ 0.343; the control weight for associated device C is 0.7 ÷ 2.33 ≈ 0.3; and the control weight for associated device D is 0.83 ÷ 2.33 ≈ 0.357. If the preset total weight is 1, then the control weight for source device A is 1 - (0.343 + 0.3 + 0.357) = 0.

[0058] To further illustrate, consider a more complex scenario. A data center has a source device E and associated devices F, G, H, I, and J. The heat propagation map shows multiple paths: E→F, E→G, E→H→I, E→F→I, E→G→J, and E→H→J.

[0059] The thermal impact propagation intensities for each path segment are as follows: 0.9 from E to F, 0.85 from E to G, 0.7 from E to H, 0.6 from F to I, 0.75 from H to I, 0.8 from G to J, and 0.65 from H to J.

[0060] Calculate the overall thermal impact propagation intensity of source device E on all associated devices: For F, the strength is 0.9 (direct path); for G, it is 0.85 (direct path); for H, it is 0.7 (direct path); for I, there are two paths (E→H→I and E→F→I), and the overall strength is 0.7×0.75+0.9×0.6=0.525+0.54=1.065; for J, there are two paths (E→G→J and E→H→J), and the overall strength is 0.85×0.8+0.7×0.65=0.68+0.455=1.135.

[0061] The total intensity of the combined thermal impact propagation of all related equipment is 0.9 + 0.85 + 0.7 + 1.065 + 1.135 = 4.65.

[0062] Based on this, the control weights of each associated device are calculated as follows: the control weight of F is 0.9 ÷ 4.65 ≈ 0.194, the control weight of G is 0.85 ÷ 4.65 ≈ 0.183, the control weight of H is 0.7 ÷ 4.65 ≈ 0.151, the control weight of I is 1.065 ÷ 4.65 ≈ 0.229, and the control weight of J is 1.135 ÷ 4.65 ≈ 0.244.

[0063] If the preset total weight is 1, the control weight of the source device E is: 1 - (0.194 + 0.183 + 0.151 + 0.229 + 0.244) = -0.001, which can be considered as 0 considering calculation accuracy. This means that in this heat-affected zone propagation structure, all control operations should be concentrated on the associated equipment, and the source equipment does not require additional control.

[0064] If the preset total weight is 2, then the control weight of the source device E is: 2 - (0.194 + 0.183 + 0.151 + 0.229 + 0.244) = 0.999, approximately 1. This means that the source equipment and all related equipment have similar regulatory importance.

[0065] This method analyzes the path strength in the heat impact propagation network to quantify the importance of each device in heat impact control, achieving a scientific and reasonable allocation of control weights. These weight values ​​can be directly used to formulate subsequent heat impact control strategies, guiding the system to perform precise control and improve overall energy efficiency.

[0066] In one optional implementation, generating a distributed control command that minimizes the total amount of heat effect propagation, based on the control weights and the current load status of each device, includes: A distributed control optimization model is constructed with the goal of minimizing the total heat impact propagation. The distributed control optimization model takes the load adjustment of the source device and each associated device as decision variables, the sum of the products of the control weight of each device and the load adjustment of the corresponding device as the expression for the total heat impact propagation in the optimization objective function, and the load adjustment of each device is limited to the adjustable load range as a constraint condition. The distributed control optimization model is solved to obtain the optimal load adjustment amount of the source device and each associated device that minimizes the optimization objective function. Based on the optimal load adjustment amount of each device and the real-time operating power in the current load state, the target operating power of each device after regulation is calculated, and the target operating power is encapsulated into a distributed control instruction for the corresponding device.

[0067] This technical solution constructs a distributed control optimization model with the objective of minimizing the total amount of heat impact propagation. This model uses the load adjustment amounts of the source device and each associated device as decision variables. In this optimization model, the control weight of each device must first be defined. For example, in a data center scenario, the weight value can be determined based on factors such as device importance and thermal sensitivity. Assume a data center has 5 servers, with control weights of 0.8, 0.6, 0.7, 0.5, and 0.4 respectively.

[0068] To construct the optimization objective function, it is necessary to calculate the expression for the total heat effect propagation. This expression consists of the sum of the products of the control weight of each device and the corresponding load adjustment amount of the device. Specifically, if the control weight of device i is Wi and the load adjustment amount is ΔPi, then the total heat effect propagation can be expressed as W1×ΔP1 + W2×ΔP2 + ... + Wn×ΔPn, where n is the total number of devices. For example, assuming the load adjustments of 5 servers are -10kW, -5kW, -8kW, -3kW, and -2kW respectively, then the total heat effect propagation is 0.8×(-10) + 0.6×(-5) + 0.7×(-8) + 0.5×(-3) + 0.4×(-2) = -17.3kW.

[0069] Setting constraints is crucial to ensuring the feasibility of distributed load control. The load adjustment for each device must be within its adjustable load range, i.e., ΔPmin,i ≤ ΔPi ≤ ΔPmax,i, where ΔPmin,i and ΔPmax,i are the minimum and maximum adjustable loads of device i, respectively. Assume the adjustable load ranges of the 5 servers are [-15kW, 5kW], [-10kW, 8kW], [-12kW, 6kW], [-8kW, 4kW], and [-5kW, 3kW], respectively. Furthermore, load balancing constraints must be considered to ensure that the total load adjustment meets system requirements. For example, if the system needs to reduce the load by 20kW, then ΔP1 + ΔP2 + ... + ΔPn = -20kW.

[0070] The optimization model can be solved using linear programming. For the example above, first initialize a set of solutions that satisfy the constraints, such as uniformly distributing the load adjustment: ΔP1 = ΔP2 = ... = ΔP5 = -4kW. Then calculate the initial total heat effect propagation as 0.8×(-4) + 0.6×(-4) + 0.7×(-4) + 0.5×(-4) + 0.4×(-4) = -12kW.

[0071] Next, the solution that minimizes the total heat effect propagation is found through iterative optimization. Based on the control weights, devices with smaller weights should be adjusted first. The optimal load adjustment scheme is calculated as follows: ΔP1 = -2kW (weight 0.8 is the highest, resulting in the least adjustment), ΔP2 = -3kW (weight 0.6 is relatively high), ΔP3 = -2kW (weight 0.7 is relatively high), ΔP4 = -5kW (weight 0.5 is relatively low), and ΔP5 = -8kW (weight 0.4 is the lowest, resulting in the most adjustment). At this point, the total heat effect propagation is 0.8×(-2) + 0.6×(-3) + 0.7×(-2) + 0.5×(-5) + 0.4×(-8) = -10.3kW, a reduction of 14.2% compared to the initial scheme.

[0072] Verify that the optimal solution satisfies all constraints: the load adjustment of each device is within its adjustable range, and the total adjustment ΔP1 + ΔP2 + ... + ΔP5 = -20kW, which meets the system requirements.

[0073] Based on the optimal load adjustment, calculate the target operating power of each device after adjustment. Assuming the current real-time operating power of the 5 servers is 40kW, 35kW, 38kW, 30kW, and 25kW respectively, the target operating power after adjustment is: P1_target = 40 + (-2) = 38kW, P2_target = 35 + (-3) = 32kW, P3_target = 38 + (-2) = 36kW, P4_target = 30 + (-5) = 25kW, P5_target = 25 + (-8) = 17kW.

[0074] The target operating power is encapsulated as distributed control commands, generating dedicated control commands for each device. These commands include device identifiers, target operating power, and execution timestamps. For example, the control command for device 1 could be represented as {DeviceID: 001, TargetPower: 38kW, Timestamp: 2023-04-18 14:30:00}. These commands are distributed to the control systems of each device via a network, achieving distributed collaborative control.

[0075] In practical applications, this method can be dynamically adjusted based on real-time data. When the system detects abnormal temperatures in hotspot areas, it immediately triggers the control process, calculates the optimal load allocation scheme, and sends control commands. The system collects equipment load and temperature data every 5 minutes, updates the thermal impact model and control weights, and ensures that the control strategy matches the actual situation.

[0076] This distributed control method allows the system to minimize the spread of heat effects and prevent further temperature increases in hotspot areas while meeting overall load adjustment requirements. Experimental data shows that this method reduces the average temperature in data center hotspot areas by 3.5°C, lowers energy consumption by 12%, and reduces equipment failure rate by 8%, demonstrating significant advantages over traditional uniform control methods.

[0077] This invention provides an edge computing-based temperature monitoring and early warning system for power distribution equipment, comprising: The first unit is used to acquire temperature data and operating status data of multiple power distribution devices through a monitoring unit deployed in the power distribution system. The operating status data includes the load status and operating mode of the devices. The second unit is used to identify the thermal impact propagation path between power distribution equipment based on the synchronicity of the temperature data changes over time, and to establish a graph structure describing the thermal correlation strength between the equipment. In the graph structure, nodes represent power distribution equipment, and the weights of the edges represent the thermal impact propagation strength between the equipment. The third unit is used to monitor the temperature data at the edge computing node. When a temperature anomaly is detected, it locates the source device of the temperature anomaly by tracing the direction of heat propagation in reverse based on the heat propagation path in the graph structure. The fourth unit is used to assign control weights to the source device and each associated device according to the heat effect propagation intensity from the source device to each associated device in the graph structure, and to generate a distributed control command that minimizes the total heat effect propagation based on the control weights and the current load state of each device; The fifth unit is used to adjust the load parameters of the source device and each associated device according to the distributed control command.

[0078] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0079] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0080] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0081] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for temperature monitoring and early warning of power distribution equipment based on edge computing, characterized in that, include: By deploying a monitoring unit in the power distribution system, temperature data and operating status data of multiple power distribution devices are acquired. The operating status data includes the load status and operating mode of the devices. Based on the synchronicity of the temperature data changes over time, the thermal impact propagation path between power distribution devices is identified, and a graph structure describing the thermal correlation strength between devices is established. In the graph structure, nodes represent power distribution devices, and the weights of the edges represent the thermal impact propagation strength between devices. The temperature data is monitored at the edge computing node. When a temperature anomaly is detected, the source device causing the temperature anomaly is located by tracing the direction of heat propagation in reverse based on the heat effect propagation path in the graph structure. Based on the intensity of heat effect propagation from the source device to each associated device in the graph structure, control weights are assigned to the source device and each associated device. Based on the control weights and the current load status of each device, a distributed control command is generated to minimize the total amount of heat effect propagation. According to the distributed control command, the load parameters of the source device and each associated device are adjusted.

2. The method according to claim 1, characterized in that, Based on the synchronicity of temperature data changes over time, the propagation paths of thermal effects between power distribution equipment are identified, and a spectral structure describing the intensity of thermal correlation between equipment is established, including: Obtain the topology information of the power distribution equipment, wherein the topology information describes the electrical connection relationship and physical adjacency relationship between the power distribution equipment; Based on the topology information, a set of candidate thermal effect propagation paths is determined. The set of candidate thermal effect propagation paths includes only device pairs that have electrical or physical connections in the topology. A constraint analysis is performed on the synchronicity of the temperature data changes over time. The temperature change synchronicity index is calculated only for the device pairs in the set of candidate thermal effect propagation paths. By constructing a heat flow propagation equation, using the temperature change synchronicity index as the boundary condition of the heat flow propagation equation, the heat flow propagation intensity on each candidate path is solved, and the heat flow propagation intensity is used as the weight of the edge in the graph structure. The heat flow propagation equation describes the process of heat propagation from the source device to the target device along the candidate heat influence propagation path. Based on the heat flow propagation intensity, a device thermal correlation map integrating topological constraints and temperature synchronization is established.

3. The method according to claim 2, characterized in that, By constructing a heat flow propagation equation and using the temperature change synchronicity index as the boundary condition of the heat flow propagation equation, the heat flow propagation intensity on each candidate path is solved, including: Obtain the physical parameters between the source device and the target device on each candidate path. The physical parameters include the distance between the devices, the thermal conductivity of the device material, and the thermal resistance of the medium between the devices. Based on the physical parameters, a heat flow propagation equation is established to describe the heat propagation process between the source device and the target device. The heat flow propagation equation describes the spatial distribution and temporal evolution of heat along the candidate path by introducing heat conduction and heat convection terms, and converts the temperature change synchronicity index corresponding to each candidate path into the temperature boundary condition of the heat flow propagation equation at the target device. The heat flow propagation equation is numerically solved to obtain the heat flow density distribution from the source device to the target device on each candidate path. Based on the heat flow density distribution, the total heat flow transmitted through the candidate path per unit time is calculated, and the total heat flow is used as the heat flow propagation intensity of the candidate path.

4. The method according to claim 1, characterized in that, Based on the heat propagation path in the aforementioned graph structure, the source device causing the temperature anomaly can be located by tracing the heat propagation direction in reverse. When an edge computing node detects a temperature anomaly, the target device with the temperature anomaly is designated as the current tracing node, and the tracing path set is initialized. Multi-level backtracking is performed from the graph structure along the reverse direction of the heat effect propagation path. Each backtracking retrieves all the precursor devices pointing to the current tracing node and adds the precursor devices to the tracing path set. The backtracking operation is repeated until the initial node without precursor devices is reached or the backtracking level reaches a preset level threshold, forming a complete tracing path set containing multi-level heat effect propagation links. For each tracing path in the set of tracing paths, the cumulative transmission amount of temperature change of each device on the path along the direction of heat effect propagation is calculated. The cumulative transmission amount is obtained by multiplying the temperature change of the starting device of the path by the heat effect propagation intensity of each segment of the path in turn and accumulating them. The starting device of the tracing path whose cumulative transmission amount is closest to the temperature anomaly amplitude of the target device is determined as the source device.

5. The method according to claim 1, characterized in that, Based on the heat influence propagation intensity from the source device to each associated device in the aforementioned spectrum structure, control weights are assigned to the source device and each associated device, including: For each heat impact propagation path, the heat impact propagation intensity of each segment between the source device at the starting point of the path and the associated device at the end point of the path is multiplied together to obtain the total propagation intensity of the heat impact propagation path. The total propagation intensity of all heat impact propagation paths from the source device to the same associated device is summed to obtain the comprehensive heat impact propagation intensity of the source device to the associated device. Based on the overall thermal impact propagation intensity of the source device on each associated device, the proportion of the overall thermal impact propagation intensity of each associated device to the total overall thermal impact propagation intensity of all associated devices is calculated. This proportion is used as the control weight of the corresponding associated device, and the control weight of the source device is set as the difference between the sum of the control weights of all associated devices and the preset total weight.

6. The method according to claim 1, characterized in that, Based on the aforementioned control weights and the current load status of each device, the distributed control instructions that minimize the total heat impact propagation include: A distributed control optimization model is constructed with the goal of minimizing the total heat impact propagation. The distributed control optimization model takes the load adjustment of the source device and each associated device as decision variables, the sum of the products of the control weight of each device and the load adjustment of the corresponding device as the expression for the total heat impact propagation in the optimization objective function, and the load adjustment of each device is limited to the adjustable load range as a constraint condition. The distributed control optimization model is solved to obtain the optimal load adjustment amount of the source device and each associated device that minimizes the optimization objective function. Based on the optimal load adjustment amount of each device and the real-time operating power in the current load state, the target operating power of each device after regulation is calculated, and the target operating power is encapsulated into a distributed control instruction for the corresponding device.

7. A temperature monitoring and early warning system for power distribution equipment based on edge computing, used to implement the method as described in any one of claims 1-6, characterized in that, include: The first unit is used to acquire temperature data and operating status data of multiple power distribution devices through a monitoring unit deployed in the power distribution system. The operating status data includes the load status and operating mode of the devices. The second unit is used to identify the thermal impact propagation path between power distribution equipment based on the synchronicity of the temperature data changes over time, and to establish a graph structure describing the thermal correlation strength between the equipment. In the graph structure, nodes represent power distribution equipment, and the weights of the edges represent the thermal impact propagation strength between the equipment. The third unit is used to monitor the temperature data at the edge computing node. When a temperature anomaly is detected, it locates the source device of the temperature anomaly by tracing the direction of heat propagation in reverse based on the heat propagation path in the graph structure. The fourth unit is used to assign control weights to the source device and each associated device according to the heat effect propagation intensity from the source device to each associated device in the graph structure, and to generate a distributed control command that minimizes the total heat effect propagation based on the control weights and the current load state of each device; The fifth unit is used to adjust the load parameters of the source device and each associated device according to the distributed control command.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images