A high voltage passive inductive temperature monitoring system

Abstract

This invention relates to the field of temperature monitoring and discloses a high-voltage passive sensing temperature monitoring system. The system includes: several temperature sensors deployed in different cabinets and aisles within a computer room; a control unit that sends detection commands to each temperature sensor and receives feedback temperature signals; the control unit generates a temperature change curve for each monitoring point based on the received temperature signals; the control unit constructs a three-dimensional temperature distribution model of the entire computer room based on the temperature change curve of each monitoring point, identifies potential hotspot areas based on the three-dimensional temperature distribution model, and predicts the temperature change trend of the potential hotspot areas; and a cooling unit that regulates the temperature of its area; the cooling unit communicates with the control unit; wherein the control unit adjusts the power of each cooling unit based on the identified potential hotspot areas.

Description

Technical Field

[0001] This invention relates to the field of temperature monitoring, and more particularly to a high-voltage passive sensing temperature monitoring system. Background Technology

[0002] The safety and reliability of power equipment is a crucial aspect of ensuring the safety of ultra-large-scale power transmission and distribution and the power grid. Therefore, real-time monitoring of the safe operation of power grid equipment is essential. Long-term power grid operation data shows that most power grid electrical equipment failures are caused by factors such as high-current operation, equipment aging, and decreased insulation levels, leading to equipment operating under high-temperature conditions and subsequently causing serious consequences such as combustion and explosion.

[0003] In power systems, temperature changes in high-voltage electrical equipment are a crucial indicator, directly impacting their safe and stable operation. During operation, factors such as manufacturing defects, contact oxidation, and arcing can increase contact resistance at the equipment's joints, leading to temperature rise. When the temperature exceeds a certain level, the mechanical and electrical strength of the equipment decreases, potentially causing short circuits or even equipment damage, severely threatening the safe operation of the power grid. Real-time monitoring of electrical equipment temperature helps on-duty personnel identify problems early, eliminate potential hazards, and ensure the safe operation of the power system.

[0004] Traditional temperature monitoring and control systems mostly use wired temperature sensors for data acquisition, which suffers from problems such as complex wiring and poor scalability. At the same time, existing systems often rely on fixed thresholds to detect temperature anomalies, lacking the ability to predict temperature change trends and failing to effectively intervene in potential hotspot areas in advance.

[0005] Solving the above-mentioned technical problems is a technical challenge that needs to be overcome by those skilled in the art. Summary of the Invention

[0006] This invention provides a high-voltage passive sensing temperature monitoring system to at least partially solve the above-mentioned technical problems.

[0007] To address the aforementioned technical problems, this invention provides a high-voltage passive sensing temperature monitoring system, comprising:

[0008] Several temperature sensors; the several temperature sensors are deployed in different cabinets and aisles within the computer room;

[0009] Control unit; the control unit is used to send detection commands to each temperature sensor and receive feedback temperature signals; the control unit generates a temperature change curve for each monitoring point based on the received temperature signals; the control unit constructs a three-dimensional temperature distribution model of the entire computer room based on the temperature change curve of each monitoring point, identifies potential hotspot areas based on the three-dimensional temperature distribution model, and predicts the temperature change trend of the potential hotspot areas.

[0010] A refrigeration unit; the refrigeration unit is used to regulate the temperature of the area it is in; the refrigeration unit communicates with the control unit;

[0011] The control unit adjusts the power of each cooling unit based on identified potential hotspot areas.

[0012] In one optional implementation, the control unit generates a temperature change curve for each monitoring point based on the received temperature signal, including:

[0013] The temperature signals sent by each temperature sensor are timestamped to synchronize the temperature data and establish a unified time series.

[0014] Interpolation is used to smoothly fit the temperature data and generate a temperature change curve.

[0015] The control unit constructs a three-dimensional temperature distribution model of the entire computer room based on the temperature change curve of each monitoring point, including:

[0016] Extract the rate of temperature change based on the temperature change curve;

[0017] Establish a three-dimensional spatial mesh based on the rack coordinates;

[0018] The temperature change rate extracted from each monitoring point is mapped to the corresponding grid node of the three-dimensional spatial grid to obtain a discrete feature field;

[0019] Spatial interpolation of discrete feature fields generates a three-dimensional temperature distribution model;

[0020] Potential hotspot areas were identified based on a three-dimensional temperature distribution model, including:

[0021] Based on the three-dimensional temperature distribution model, regions whose temperature change rate exceeds the first temperature change rate are extracted and denoted as the first region.

[0022] The first area is further filtered based on real-time temperature values, and areas where the real-time temperature values ​​exceed a set threshold are retained as potential hotspot areas.

[0023] In one optional implementation, predicting the temperature change trend of the potential hotspot area includes:

[0024] A first-order autoregressive model was constructed based on the extracted temperature change rate features;

[0025] Input the current temperature value and temperature change rate into the first-order autoregressive model to calculate the temperature prediction value for the first time period in the future to obtain the temperature change trend curve; the predicted range of the temperature change rate is limited to a preset change rate, and the temperature prediction value exceeding the preset change rate is truncated.

[0026] In one alternative implementation, the control unit adjusts the power of each cooling unit based on identified potential hotspot areas, including:

[0027] Determine the difference between the predicted temperature and the target temperature for potential hotspot areas; the target temperature is the optimal operating temperature for the server equipment.

[0028] The amount of heat that needs to be removed is calculated using a heat calculation formula based on the spatial volume of the potential hotspot area, the specific heat capacity of the air, and the time required for the predicted temperature to reach the target temperature.

[0029] The increased cooling capacity is distributed to the surrounding cooling units in inverse proportion to the spatial distance;

[0030] The power of each refrigeration unit is adjusted by a PID controller;

[0031] The increased cooling capacity is distributed to surrounding cooling units in inverse proportion to spatial distance, including:

[0032] Determine the location of each refrigeration unit around the potential hotspot area, and calculate the straight-line distance between each refrigeration unit and the center of the potential hotspot area.

[0033] The allocation weight of each refrigeration unit is calculated based on the straight-line distance; where the closer the distance, the higher the allocation weight.

[0034] The power of each refrigeration unit is adjusted using a PID controller, including:

[0035] Based on the deviation between the current actual power and the target power, as well as the deviation between the real-time temperature data and the target temperature, the power adjustment amount is calculated using a PID control algorithm; whereby the target power is determined by the allocated cooling capacity.

[0036] In one optional implementation, establishing a three-dimensional spatial mesh based on rack coordinates further includes:

[0037] Increase the mesh density in regions with large temperature gradients to improve model resolution;

[0038] Reduce mesh density in regions with gradual temperature changes to reduce computational load;

[0039] Adjusting the grid distribution based on real-time temperature data changes; wherein, adjusting the grid distribution based on real-time temperature data changes includes:

[0040] Using the geometric center coordinates of the server rack as a reference point, the three-dimensional space of the server room is divided into an initial grid and the side length of the initial grid is set.

[0041] Assign a unique identifier to each grid node and establish a table to map the grid nodes to the locations of racks and aisles;

[0042] For the temperature data collected by each temperature sensor, calculate its temperature gradient vector in three-dimensional space;

[0043] For adjacent grid nodes, the temperature gradient between nodes is obtained by interpolation to calculate the temperature difference between the nodes;

[0044] Configure a temperature gradient intensity index for each grid node; the temperature gradient intensity index is the weighted average of the temperature gradients of the adjacent grids around the node.

[0045] When the temperature gradient intensity index of a grid node exceeds the first threshold, the grid and its surrounding adjacent grids are subdivided into quadtrees, reducing the grid side length to half of its original size.

[0046] When the temperature gradient intensity index of a grid node is less than the second threshold, the grid is merged with the surrounding grids that have similar temperature gradient intensities, thus doubling the grid's side length.

[0047] After each adjustment, update the identifier number and corresponding relationship table of the grid nodes.

[0048] In one optional embodiment, the control unit, when adjusting the power of the refrigeration unit, further includes:

[0049] Based on the layout of the computer room, cooling units with a straight-line distance less than the first preset distance are marked as adjacent, and an undirected graph G=(V, E) is constructed, where node V represents a cooling unit and edge E represents an adjacent relationship;

[0050] Each edge is assigned a weight value W; the weight value is calculated based on the correlation between the temperature regulation of the two refrigeration units in historical operating data, and the higher the correlation, the greater the weight.

[0051] After initially allocating the cooling capacity, the power difference between adjacent cooling units is calculated. If the power difference exceeds the preset difference, a particle swarm optimization algorithm is used to obtain the target power value of each cooling unit with the objectives of minimizing the mean square error of temperature regulation, minimizing the power difference, and minimizing the total energy consumption.

[0052] The particle swarm optimization algorithm is employed, including:

[0053] Each particle is defined as a power allocation scheme for a set of cooling units, and the particle dimension is equal to the number of cooling units. The initial position and velocity of the particles are randomly generated. The position range is from the minimum power to the maximum power of each cooling unit, and the velocity range is set to the maximum adjustment step size.

[0054] For each particle, calculate the root mean square error of temperature regulation, the sum of power differences between adjacent units, and the total energy consumption of the system, and substitute them into the multi-objective function to calculate the fitness value;

[0055] The particle velocity and position are updated based on the particle's own optimal position and the global optimal position; after each iteration, the particle position is checked to see if it meets the power constraint, and if not, boundary processing is performed.

[0056] When the maximum number of iterations is reached or the convergence condition is met, the power allocation scheme corresponding to the globally optimal particle is output as the target power value for each cooling unit.

[0057] In one optional implementation, the system further includes an alarm module. The control unit judges the temperature value of the received temperature signal. When the temperature value is greater than the set temperature, the control unit sends an alarm command to the alarm module, and the alarm module emits an audible alert to the outside world.

[0058] In one alternative implementation, the temperature sensor is a wireless passive temperature sensor.

[0059] Compared to existing technologies, this invention offers at least the following advantages: By deploying passive temperature sensors within the server room, real-time temperature acquisition of different server racks and aisles is achieved. The control unit performs in-depth processing on the acquired temperature signals, constructs a three-dimensional temperature distribution model, identifies potential hotspot areas, and predicts their temperature change trends. Based on the prediction results, the control unit adjusts the power of the cooling unit to regulate the server room temperature. Attached Figure Description

[0060] Figure 1 This is a system block diagram of a high-voltage passive sensing temperature monitoring system provided by the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Reference Figure 1 The first embodiment of the present invention provides a high-voltage passive sensing temperature monitoring system, comprising:

[0063] A high-voltage passive sensing temperature monitoring system includes:

[0064] Several temperature sensors; these temperature sensors are deployed in different cabinets and aisles within the computer room. High-voltage passive temperature sensors: These are sensors that do not require an external power supply and obtain energy from the environment through electromagnetic induction, thermoelectric generation, or other methods to achieve temperature sensing. In high-voltage environments, these sensors can avoid safety hazards introduced by power lines, while also reducing wiring costs and maintenance difficulty.

[0065] The control unit sends detection commands to each temperature sensor and receives feedback temperature signals. Based on the received temperature signals, the control unit generates a temperature change curve for each monitoring point. Based on the temperature change curves of each monitoring point, the control unit constructs a three-dimensional temperature distribution model of the entire computer room and identifies potential hotspot areas and predicts the temperature change trend of these potential hotspot areas. The control unit interacts with the temperature sensors and cooling units, sending data and commands. By processing the signals fed back from the temperature sensors, comprehensive monitoring and analysis of the computer room's temperature status are achieved. The temperature change curve is obtained by arranging the temperature data collected by the temperature sensors in chronological order and smoothing and fitting it using mathematical methods such as interpolation. It can intuitively reflect the temperature change of the monitoring points over time. The three-dimensional temperature distribution model is based on the rack coordinates, dividing the three-dimensional space of the computer room into a grid, mapping the temperature change rate of each monitoring point to the grid nodes, and generating the model through spatial interpolation. This model can comprehensively and three-dimensionally display the temperature distribution status within the computer room. Potential hotspot areas are areas in the three-dimensional temperature distribution model that are filtered by setting the temperature change rate and real-time temperature threshold, indicating areas where the temperature may rise rapidly or is already at a high level, posing a risk of equipment overheating.

[0066] A refrigeration unit; the refrigeration unit is used to regulate the temperature of the area it is in; the refrigeration unit communicates with the control unit;

[0067] The control unit adjusts the power of each cooling unit based on identified potential hotspot areas. Cooling units, such as fan units, are devices used to regulate the temperature of specific areas within the computer room. They communicate with the control unit to receive power adjustment commands and achieve cooling.

[0068] This application utilizes passive temperature sensors deployed within the computer room to achieve real-time temperature acquisition of different cabinets and aisles. The control unit performs in-depth processing of the acquired temperature signals, constructing a three-dimensional temperature distribution model and identifying potential hotspot areas to predict their temperature change trends. Based on the prediction results, the control unit adjusts the power of the cooling units to regulate the computer room temperature. The use of passive temperature sensors avoids the safety risks and wiring challenges associated with external power supplies in high-voltage environments. The three-dimensional temperature distribution model constructed by the control unit based on temperature change curves, combined with potential hotspot area identification and temperature trend prediction, can detect abnormal temperature areas in advance and predict their development, offering greater foresight compared to traditional fixed threshold judgments. The control unit adjusts the cooling unit power based on the prediction results, considering the collaborative work between cooling units to avoid power conflicts and overcompensation, effectively reducing equipment failure rates and extending equipment lifespan.

[0069] In one embodiment, the control unit generates a temperature change curve for each monitoring point based on the received temperature signal, including:

[0070] The temperature signals sent by each temperature sensor are time-stamped and synchronized to establish a unified time series to obtain temperature data. Specifically, precise time stamps are added to the signals collected by different temperature sensors, and the clock deviation of each sensor is calibrated by an algorithm to ensure that all temperature data are aligned on the same time scale and to avoid data distortion caused by sampling time differences.

[0071] Interpolation methods are used to smooth and fit temperature data to generate temperature change curves. Specifically, between known discrete temperature data points, temperature values ​​at intermediate moments are estimated using methods such as Lagrange interpolation and spline interpolation to smooth data fluctuations and generate continuous temperature change curves.

[0072] The control unit constructs a three-dimensional temperature distribution model of the entire computer room based on the temperature change curve of each monitoring point, including:

[0073] Extract the rate of temperature change based on the temperature change curve;

[0074] A three-dimensional spatial mesh is established based on the rack coordinates. Specifically, the three-dimensional space of the data center is divided into regular or irregular cubic mesh units based on the actual coordinates of the racks. Each mesh node corresponds to a location in the space and is used to carry data such as the temperature change rate.

[0075] The temperature change rate extracted from each monitoring point is mapped to the corresponding grid nodes in a three-dimensional spatial grid to obtain a discrete feature field. Specifically, the temperature change rate of each monitoring point is mapped to the three-dimensional spatial grid nodes to form a discretely distributed dataset that describes the spatial characteristics of temperature changes in the computer room.

[0076] A three-dimensional temperature distribution model is generated by spatial interpolation of a discrete feature field. Specifically, based on known data points in the discrete feature field, the rate of temperature change at unmonitored locations in the grid is estimated using algorithms such as Kriging interpolation, thus generating a continuous three-dimensional temperature distribution model.

[0077] Potential hotspot areas were identified based on a three-dimensional temperature distribution model, including:

[0078] Based on the three-dimensional temperature distribution model, regions whose temperature change rate exceeds the first temperature change rate are extracted and denoted as the first region.

[0079] The first area is further filtered based on real-time temperature values, and areas where the real-time temperature values ​​exceed a set threshold are retained as potential hotspot areas.

[0080] Specifically, the original temperature signal is timestamped and interpolated to eliminate time errors and data noise; the temperature change rate is extracted based on the curve, and a three-dimensional spatial grid is constructed based on the rack coordinates. The temperature change rate is mapped to the grid nodes to form a discrete feature field; the discrete data is transformed into a continuous three-dimensional temperature distribution model through spatial interpolation algorithm, and potential hotspot areas are selected by combining the temperature change rate threshold and the real-time temperature threshold.

[0081] In one implementation, predicting the temperature change trend of the potential hotspot area includes:

[0082] A first-order autoregressive model is constructed based on the extracted temperature change rate features. Specifically, a first-order autoregressive model is a time series forecasting model that predicts future values ​​based on the lagged values ​​of variables (i.e., values ​​at past times). In this approach, the dynamic trend of temperature change is captured by establishing a linear relationship between the temperature change rate and its previous value.

[0083] Input the current temperature value and temperature change rate into the first-order autoregressive model to calculate the temperature prediction value for the first time period in the future to obtain the temperature change trend curve; the predicted range of the temperature change rate is limited to a preset change rate, and the temperature prediction value exceeding the preset change rate is truncated.

[0084] Specifically, by employing a first-order autoregressive model and utilizing the time-series characteristics of temperature change rate to establish a predictive model, it can effectively capture short-term trends in temperature changes. Compared to simple empirical judgments or fixed threshold warnings, it can identify abnormal temperature changes in potential hotspot areas in advance. By using the current temperature value and temperature change rate as inputs and combining them with historical data patterns to dynamically calculate predicted values, it avoids the lag caused by static predictions, effectively reducing the risk of equipment overheating and minimizing energy waste.

[0085] In one implementation, the control unit adjusts the power of each cooling unit based on identified potential hotspot areas, including:

[0086] Determine the difference between the predicted temperature and the target temperature for potential hotspot areas; the target temperature is the optimal operating temperature for the server equipment.

[0087] The amount of heat that needs to be removed is calculated using a heat calculation formula based on the spatial volume of the potential hotspot area, the specific heat capacity of the air, and the time required for the predicted temperature to reach the target temperature.

[0088] The increased cooling capacity is distributed to the surrounding cooling units in inverse proportion to the spatial distance;

[0089] The power of each refrigeration unit is adjusted by a PID controller;

[0090] The increased cooling capacity is distributed to surrounding cooling units in inverse proportion to spatial distance, including:

[0091] Determine the location of each refrigeration unit around the potential hotspot area, and calculate the straight-line distance between each refrigeration unit and the center of the potential hotspot area.

[0092] The allocation weight of each refrigeration unit is calculated based on the straight-line distance; where the closer the distance, the higher the allocation weight.

[0093] The power of each refrigeration unit is adjusted using a PID controller, including:

[0094] Based on the deviation between the current actual power and the target power, as well as the deviation between the real-time temperature data and the target temperature, the power adjustment amount is calculated using a PID control algorithm; whereby the target power is determined by the allocated cooling capacity.

[0095] Specifically, by determining the difference between the predicted temperature and the target temperature in potential hotspot areas, the target deviation for temperature regulation can be clearly identified, avoiding blind adjustments. The amount of heat to be removed is calculated based on the spatial volume of the potential hotspot area, the specific heat capacity of the air, and the time required for the predicted temperature to reach the target temperature. This combination of actual physical parameters and time factors makes the calculated cooling capacity more accurate. The increased cooling capacity is distributed to surrounding cooling units inversely proportional to spatial distance. The location of surrounding cooling units is determined using the potential hotspot area as the center, and the straight-line distance is calculated to determine the allocation weight. The closer the distance, the higher the weight. This considers spatial relationships, prioritizing cooling units closer to the potential hotspot area to undertake more cooling tasks, enabling more efficient cooling of the hotspot area, reducing energy waste during the cooling process, and improving cooling efficiency. A PID controller calculates the power adjustment amount based on the deviation between the current actual power and the target power, as well as the deviation between the real-time temperature data and the target temperature. The PID control algorithm enables dynamic adjustment of the cooling unit power. PID controllers can adjust the power regulation in real time based on factors such as the magnitude and rate of change of the deviation, so that the power of the cooling unit can quickly and stably reach the target power. This allows for more effective real-time temperature regulation to the target temperature, ensuring that the server equipment operates at the most suitable temperature, reducing the probability of equipment failure due to overheating, and extending the service life of the equipment.

[0096] In one implementation, establishing a three-dimensional spatial mesh based on rack coordinates further includes:

[0097] Increase the mesh density in regions with large temperature gradients to improve model resolution;

[0098] Reduce mesh density in regions with gradual temperature changes to reduce computational load;

[0099] Adjusting the grid distribution based on real-time temperature data changes; wherein, adjusting the grid distribution based on real-time temperature data changes includes:

[0100] Using the geometric center coordinates of the server rack as a reference point, the three-dimensional space of the server room is divided into an initial grid and the side length of the initial grid is set.

[0101] Assign a unique identifier to each grid node and establish a table to map the grid nodes to the locations of racks and aisles;

[0102] For the temperature data collected by each temperature sensor, calculate its temperature gradient vector in three-dimensional space;

[0103] For adjacent grid nodes, the temperature gradient between nodes is obtained by interpolation to calculate the temperature difference between the nodes;

[0104] Configure a temperature gradient intensity index for each grid node; the temperature gradient intensity index is the weighted average of the temperature gradients of the adjacent grids around the node.

[0105] When the temperature gradient intensity index of a grid node exceeds the first threshold, the grid and its surrounding adjacent grids are subdivided into quadtrees, reducing the grid side length to half of its original size.

[0106] When the temperature gradient intensity index of a grid node is less than the second threshold, the grid is merged with the surrounding grids that have similar temperature gradient intensities, thus doubling the grid's side length.

[0107] After each adjustment, update the identifier number and corresponding relationship table of the grid nodes.

[0108] Specifically, the system divides the initial grid based on the geometric center coordinates of the cabinet, assigns unique numbers to nodes, and establishes a location correspondence table. Temperature gradient vectors are calculated from temperature sensor data, and combined with temperature gradient values ​​obtained through interpolation between nodes, a temperature gradient intensity index based on a weighted average of adjacent grids is configured for each grid node. This quantifies the severity of temperature changes in each region. When the temperature gradient intensity index of a grid node exceeds a first threshold, a quadtree subdivision strategy is used to reduce the grid side length. This increases grid density in areas with large and complex temperature gradients, improving the resolution of the 3D temperature distribution model in key areas and preventing missed detections of hotspots. When the index is less than a second threshold, grids are merged to expand the side length. This reduces grid density in areas with gentle temperature changes, effectively reducing unnecessary computation. Each grid adjustment updates the numbering and correspondence table, enabling the system to dynamically optimize the grid distribution based on real-time temperature data.

[0109] In one embodiment, the control unit further includes, when adjusting the power of the refrigeration unit:

[0110] Based on the server room layout, cooling units with a straight-line distance less than a first preset distance are marked as adjacent, constructing an undirected graph G=(V, E), where node V represents a cooling unit and edge E represents an adjacency relationship. Specifically, the undirected graph G=(V, E) is a mathematical model composed of a set of nodes V and a set of edges E, where the edges have no direction. In this scheme, each cooling unit is abstracted as a node V. If the straight-line distance between two cooling units is less than the first preset distance, then there is an undirected edge E between them, representing an adjacency relationship.

[0111] Each edge is assigned a weight value W; the weight value is calculated based on the correlation between the temperature regulation of the two refrigeration units in historical operating data, and the higher the correlation, the greater the weight.

[0112] After initially allocating the cooling capacity, the power difference between adjacent cooling units is calculated. If the power difference exceeds the preset difference, a particle swarm optimization algorithm is used to obtain the target power value of each cooling unit with the objectives of minimizing the mean square error of temperature regulation, minimizing the power difference, and minimizing the total energy consumption.

[0113] The particle swarm optimization algorithm is employed, including:

[0114] Each particle is defined as a power allocation scheme for a set of cooling units, and the particle dimension is equal to the number of cooling units. The initial position and velocity of the particles are randomly generated. The position range is from the minimum power to the maximum power of each cooling unit, and the velocity range is set to the maximum adjustment step size.

[0115] For each particle, calculate the root mean square error of temperature regulation, the sum of power differences between adjacent units, and the total energy consumption of the system, and substitute them into the multi-objective function to calculate the fitness value;

[0116] The particle velocity and position are updated based on the particle's own optimal position and the global optimal position. After each iteration, it is checked whether the particle position meets the power constraint; if not, boundary processing is performed. Specifically, boundary processing is used to ensure that the particle position (power allocation scheme) is within the physically feasible range. When the updated position of the particle exceeds the minimum or maximum power limit of the cooling unit, it is forcibly constrained to the boundary value.

[0117] When the maximum number of iterations is reached or the convergence condition is met, the power allocation scheme corresponding to the globally optimal particle is output as the target power value for each cooling unit.

[0118] Specifically, by constructing an undirected graph with cooling units as nodes and adjacent relationships as edges, and assigning edge weights based on historical operating data, the degree of temperature regulation correlation between cooling units can be quantified. After initial allocation of cooling capacity, the power difference is calculated and compared with a preset difference to promptly identify potential power conflicts or imbalances between adjacent cooling units. A particle swarm optimization algorithm is introduced, abstracting each cooling unit power allocation scheme as a particle. By randomly initializing particle positions and velocities, a search is performed within the feasible power range. During the iteration process, a multi-objective function is constructed based on the mean square error of temperature regulation, the sum of power differences, and total energy consumption to calculate the fitness value, prompting the algorithm to optimize towards high temperature uniformity, balanced power, and low energy consumption. Velocity and position are updated based on the particle's own optimal position and the global optimal position, and boundary handling is performed for cases exceeding power constraints to ensure that the generated power allocation scheme satisfies physical limitations while achieving optimal balance among multiple objectives. The final output power allocation scheme corresponding to the globally optimal particle effectively avoids power conflicts and overcompensation between cooling units, achieving precise temperature regulation in the computer room and significantly reducing system energy consumption while ensuring stable equipment operation.

[0119] In one embodiment, the system further includes an alarm module. The control unit judges the temperature value of the received temperature signal. When the temperature value is greater than the set temperature, the control unit sends an alarm command to the alarm module, and the alarm module emits an audible alert to the outside world.

[0120] In one embodiment, the temperature sensor is a wireless passive temperature sensor.

[0121] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A high-voltage passive sensing temperature monitoring system, characterized in that, include: Several temperature sensors; the several temperature sensors are deployed in different cabinets and aisles within the computer room; Control unit; The control unit is used to send detection commands to each temperature sensor and receive feedback temperature signals; the control unit generates a temperature change curve for each monitoring point based on the received temperature signals; the control unit constructs a three-dimensional temperature distribution model of the entire computer room based on the temperature change curve of each monitoring point, identifies potential hotspot areas based on the three-dimensional temperature distribution model, and predicts the temperature change trend of the potential hotspot areas. A refrigeration unit; the refrigeration unit is used to regulate the temperature of the area it is in; the refrigeration unit communicates with the control unit; The control unit adjusts the power of each cooling unit based on identified potential hotspot areas; Establishing a three-dimensional spatial mesh based on rack coordinates also includes: Increase the mesh density in regions with large temperature gradients to improve model resolution; Reduce mesh density in regions with gradual temperature changes to reduce computational load; Adjusting the grid distribution based on real-time temperature data changes; wherein, adjusting the grid distribution based on real-time temperature data changes includes: Using the geometric center coordinates of the server rack as a reference point, the three-dimensional space of the server room is divided into an initial grid and the side length of the initial grid is set. Assign a unique identifier to each grid node and establish a table to map the grid nodes to the locations of racks and aisles; For the temperature data collected by each temperature sensor, calculate its temperature gradient vector in three-dimensional space; For adjacent grid nodes, the temperature gradient between nodes is obtained by interpolation to calculate the temperature difference between the nodes; Configure a temperature gradient intensity index for each grid node; the temperature gradient intensity index is the weighted average of the temperature gradients of the adjacent grids around the node. When the temperature gradient intensity index of a grid node exceeds the first threshold, the grid and its surrounding adjacent grids are subdivided into quadtrees, reducing the grid side length to half of its original size. When the temperature gradient intensity index of a grid node is less than the second threshold, the grid is merged with the surrounding grids that have similar temperature gradient intensities, thus doubling the grid's side length. After each adjustment, update the grid node identifiers and their corresponding relationship tables; When adjusting the power of the refrigeration unit, the control unit further includes: Based on the layout of the computer room, cooling units with a straight-line distance less than the first preset distance are marked as adjacent, and an undirected graph G=(V, E) is constructed, where node V represents a cooling unit and edge E represents an adjacent relationship; Each edge is assigned a weight value W; the weight value is calculated based on the correlation between the temperature regulation of the two refrigeration units in historical operating data, and the higher the correlation, the greater the weight. After initially allocating the cooling capacity, the power difference between adjacent cooling units is calculated. If the power difference exceeds the preset difference, a particle swarm optimization algorithm is used to obtain the target power value of each cooling unit with the objectives of minimizing the mean square error of temperature regulation, minimizing the power difference, and minimizing the total energy consumption. The particle swarm optimization algorithm is employed, including: Each particle is defined as a power allocation scheme for a set of cooling units, and the particle dimension is equal to the number of cooling units. The initial position and velocity of the particles are randomly generated. The position range is from the minimum power to the maximum power of each cooling unit, and the velocity range is set to the maximum adjustment step size. For each particle, calculate the root mean square error of temperature regulation, the sum of power differences between adjacent units, and the total energy consumption of the system, and substitute them into the multi-objective function to calculate the fitness value; The particle velocity and position are updated based on the particle's own optimal position and the global optimal position; after each iteration, the particle position is checked to see if it meets the power constraint, and if not, boundary processing is performed. When the maximum number of iterations is reached or the convergence condition is met, the power allocation scheme corresponding to the globally optimal particle is output as the target power value for each cooling unit.

2. The high-voltage passive sensing temperature monitoring system according to claim 1, characterized in that, The control unit generates a temperature change curve for each monitoring point based on the received temperature signal, including: The temperature signals sent by each temperature sensor are timestamped to synchronize the temperature data and establish a unified time series. Interpolation is used to smoothly fit the temperature data and generate a temperature change curve. The control unit constructs a three-dimensional temperature distribution model of the entire computer room based on the temperature change curve of each monitoring point, including: Extract the rate of temperature change based on the temperature change curve; Establish a three-dimensional spatial mesh based on the rack coordinates; The temperature change rate extracted from each monitoring point is mapped to the corresponding grid node of the three-dimensional spatial grid to obtain a discrete feature field; Spatial interpolation of discrete feature fields generates a three-dimensional temperature distribution model; Potential hotspot areas were identified based on a three-dimensional temperature distribution model, including: Based on the three-dimensional temperature distribution model, regions whose temperature change rate exceeds the first temperature change rate are extracted and denoted as the first region. The first area is further filtered based on real-time temperature values, and areas where the real-time temperature values ​​exceed a set threshold are retained as potential hotspot areas.

3. The high-voltage passive sensing temperature monitoring system according to claim 2, characterized in that, Predicting temperature change trends in the potential hotspot areas includes: A first-order autoregressive model was constructed based on the extracted temperature change rate features; Input the current temperature value and temperature change rate into the first-order autoregressive model to calculate the temperature prediction value for the first time period in the future to obtain the temperature change trend curve; the predicted range of the temperature change rate is limited to a preset change rate, and the temperature prediction value exceeding the preset change rate is truncated.

4. The high-voltage passive sensing temperature monitoring system according to claim 3, characterized in that, The control unit adjusts the power of each cooling unit based on identified potential hotspot areas, including: Determine the difference between the predicted temperature and the target temperature for potential hotspot areas; the target temperature is the optimal operating temperature for the server equipment. The amount of heat that needs to be removed is calculated using a heat calculation formula based on the spatial volume of the potential hotspot area, the specific heat capacity of the air, and the time required for the predicted temperature to reach the target temperature. The increased cooling capacity is distributed to the surrounding cooling units in inverse proportion to the spatial distance; The power of each refrigeration unit is adjusted by a PID controller; The increased cooling capacity is distributed to surrounding cooling units in inverse proportion to spatial distance, including: Determine the location of each refrigeration unit around the potential hotspot area, and calculate the straight-line distance between each refrigeration unit and the center of the potential hotspot area. The allocation weight of each refrigeration unit is calculated based on the straight-line distance; where the closer the distance, the higher the allocation weight. The power of each refrigeration unit is adjusted using a PID controller, including: Based on the deviation between the current actual power and the target power, as well as the deviation between the real-time temperature data and the target temperature, the power adjustment amount is calculated using a PID control algorithm; whereby the target power is determined by the allocated cooling capacity.

5. The high-voltage passive sensing temperature monitoring system according to claim 4, characterized in that, The system also includes an alarm module. The control unit judges the temperature value of the received temperature signal. When the temperature value is greater than the set temperature, the control unit sends an alarm command to the alarm module, and the alarm module emits an audible alert to the outside world.

6. The high-voltage passive sensing temperature monitoring system according to claim 5, characterized in that, The temperature sensor is a wireless passive temperature sensor.

Images