An integrated temperature sensing method and system based on high performance fusion architecture

Abstract

The application discloses an integrated temperature sensing method and system based on a high-efficiency fusion architecture, relates to the technical field of intelligent sensing, and dynamically increases the sensor deployment density of an abnormal area of temperature gradient change by carrying out grid division on a target monitoring area, identifying the abnormal area of temperature gradient change in real time; the distance weight concept is also introduced, the distance weight of adjacent sensors is calculated according to the sensor array layout, the sensors with unreasonable positions are dynamically adjusted, and the abnormal area is continuously identified, the sensor layout is adjusted, the data deviation is corrected and the collection frequency is adaptively adjusted in the monitoring process. By adopting the embodiment of the application, the abnormal area of temperature can be dynamically identified, the data fusion logic can be optimized, the restoration precision of the temperature field distribution is improved, and the data error caused by unreasonable sensor layout or transmission delay is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent sensing, in particular to an integrated temperature sensing method and system based on a high-efficiency fusion architecture. BACKGROUND

[0002] At present, with the rapid development of sensor technology, the application scenarios of temperature sensing systems are increasingly complex, especially in environments with uneven temperature distribution and drastic changes. Precise capture and reconstruction of temperature fields have become key requirements. The integration of temperature sensing and data transmission is an important direction in the field of intelligent sensing. The core is to achieve high-precision monitoring of complex environments through effective fusion of multi-source data, which is crucial for industrial production, environmental monitoring, and intelligent device operation.

[0003] However, existing methods often expose significant defects when faced with complex temperature environments. Most multi-source signal fusion schemes lack dynamic adaptability, leading to decreased sensing accuracy, especially in areas with drastic temperature gradients, where the system struggles to accurately reflect the true temperature distribution. Furthermore, existing methods do not fully consider the spatial layout and response differences of sensor arrays during data fusion, resulting in error accumulation and affecting overall monitoring effectiveness. SUMMARY

[0004] The purpose of the embodiments of the present application is to provide an integrated temperature sensing method and system based on a high-efficiency fusion architecture, focusing on complex temperature environment monitoring needs, capable of dynamically identifying temperature abnormal areas, optimizing data fusion logic, and improving the restoration accuracy of temperature field distribution, reducing data errors caused by unreasonable sensor layout or transmission delays.

[0005] The embodiments of the present application provide an integrated temperature sensing method based on a high-efficiency fusion architecture, comprising:

[0006] Grid division is performed on the target monitoring area, and temperature sensors are preliminarily deployed according to the division results;

[0007] Real-time acquisition of temperature data in each grid cell, identification of a first abnormal area based on the temperature data, increase in the deployment density of temperature sensors in the first abnormal area, and obtaining of a sensor array layout distribution map of the target monitoring area;

[0008] According to the sensor array layout distribution map, the distance weight between adjacent temperature sensors is calculated, and the grid cell position of the sensor is adjusted based on the distance weight;

[0009] Obtaining the response time of each temperature sensor after adjustment, calculating the delay degree of each temperature sensor, and performing weighted fusion on the temperature data according to the delay degree to obtain preliminary temperature field distribution data;

[0010] According to the preliminary temperature field distribution data, a second abnormal area is identified, and the preliminary temperature field distribution data of the second abnormal area is corrected for deviation to obtain optimized temperature field distribution data;

[0011] When the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, the data acquisition frequency is adjusted according to the temperature change rate to obtain an optimal temperature sensing result of the target monitoring area.

[0012] As an improvement of the above scheme, the target monitoring area is divided into grids, and temperature sensors are preliminarily deployed according to the division result, comprising:

[0013] The target monitoring area is equally divided according to a preset grid size to obtain grid cells;

[0014] Temperature sensors are deployed at the center point positions of each grid cell.

[0015] As an improvement of the above scheme, the temperature data in each grid cell is obtained in real time, the first abnormal area is identified according to the temperature data, and the deployment density of temperature sensors in the first abnormal area is increased to obtain a sensor array layout distribution map of the target monitoring area, comprising:

[0016] The temperature data in each grid cell is obtained in real time to form an initial temperature distribution matrix;

[0017] According to the initial temperature distribution matrix, the temperature difference between adjacent grid cells is calculated;

[0018] According to the temperature difference and the spacing of the grid cells, a temperature gradient value of the boundary between the grid cells is obtained;

[0019] When the temperature gradient value is greater than a preset gradient threshold, the grid cells on both sides of the boundary are marked as the first abnormal area;

[0020] The first abnormal area is divided into a secondary grid, and the deployment density of temperature sensors is increased according to the secondary grid division result to obtain a sensor array layout distribution map of the target monitoring area.

[0021] As an improvement of the above scheme, the first abnormal area is divided into a secondary grid, and the deployment density of temperature sensors is increased according to the secondary grid division result to obtain a sensor array layout distribution map of the target monitoring area, comprising:

[0022] The grid cells in the first abnormal area are divided into a secondary grid to obtain sub-grid cells;

[0023] Temperature sensors are added at the center point positions of the sub-grid cells;

[0024] According to the position information of all the temperature sensors, a sensor array layout distribution map of the target monitoring area is formed.

[0025] As an improvement of the above scheme, the distance weight between adjacent temperature sensors is calculated according to the sensor array layout distribution map, and the grid cell position where the sensor is located is adjusted based on the distance weight, comprising:

[0026] According to the sensor array layout distribution map, the spatial position coordinates of each temperature sensor are obtained;

[0027] According to the spatial position coordinates, the Euclidean distance between each temperature sensor and its adjacent temperature sensors is calculated;

[0028] According to the Euclidean distance between adjacent temperature sensors and the monitoring radius range of the temperature sensor, the distance weight is calculated;

[0029] When the distance weight of the target temperature sensor is less than the preset weight threshold, the number of temperature sensors deployed in the adjacent grid cells is counted, and the grid cell with the smallest number of temperature sensors is regarded as the target grid cell;

[0030] The target temperature sensor is adjusted from the original grid cell to the target grid cell.

[0031] As an improvement of the above scheme, the response time of each temperature sensor after adjustment is obtained, the delay degree of each temperature sensor is calculated, the temperature data is weighted and fused according to the delay degree, and preliminary temperature field distribution data is obtained, comprising:

[0032] The response time of each temperature sensor after adjustment is obtained, and the response time is standardized to obtain the response time deviation coefficient of each temperature sensor;

[0033] According to the preset deviation coefficient threshold and the response time deviation coefficient, the temperature sensor is divided into high delay sensor and low delay sensor;

[0034] According to the response time deviation coefficient and importance of the low delay sensor and the high delay sensor, the preset initial weight of the high delay sensor is de-weighted, and the weight correction coefficient of the high delay sensor is calculated;

[0035] According to the weight correction coefficient, the weight of each temperature sensor in the first abnormal area is dynamically adjusted to obtain a high delay fusion temperature value;

[0036] According to the high delay fusion temperature value and the preset initial weight of the low delay sensor, preliminary temperature field distribution data covering the target monitoring area is formed.

[0037] As an improvement of the above scheme, the preset initial weight of the high-delay sensor is reduced according to the response time deviation coefficient and importance of the low-delay sensor and the high-delay sensor, and a weight correction coefficient of the high-delay sensor is calculated, comprising:

[0038] According to the position of the high-delay sensor in the target monitoring area, the importance level of the high-delay sensor is obtained;

[0039] According to the response time deviation coefficient and the importance level of the high-delay sensor, a weight reduction factor is obtained;

[0040] The preset initial weight of the high-delay sensor is reduced by using the weight reduction factor, and a weight value after weight reduction is obtained;

[0041] The correlation between the historical data sequence of the high-delay sensor and the low-delay sensor is calculated by using the Pearson correlation coefficient;

[0042] According to the weight value after weight reduction and the correlation, a weight correction coefficient of the high-delay sensor is calculated.

[0043] As an improvement of the above scheme, the second abnormal area is identified according to the preliminary temperature field distribution data, and the preliminary temperature field distribution data of the second abnormal area is corrected to obtain optimized temperature field distribution data, comprising:

[0044] According to the preliminary temperature field distribution data, the temperature difference of each grid unit at consecutive time points is calculated to obtain the temperature change rate;

[0045] According to the temperature change rate, the second abnormal area is identified;

[0046] The normal temperature values of the normal grid cells adjacent to each abnormal grid cell in the second abnormal area are extracted, and the interpolation temperature of the abnormal grid cell is calculated by using the bilinear interpolation algorithm;

[0047] The interpolation temperature is used to replace the temperature value of the abnormal grid cell to obtain the optimized temperature field distribution data.

[0048] As an improvement of the above scheme, when the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, the data acquisition frequency is adjusted according to the temperature change rate to obtain the optimal temperature sensing result of the target monitoring area, comprising:

[0049] According to the optimized temperature field distribution data, the temperature change rate of each grid unit is obtained;

[0050] when the temperature change rate of the grid unit is greater than a preset rate threshold, increasing a data collection frequency of the temperature sensor in the grid unit to a preset high-frequency collection frequency, to obtain high-frequency temperature sampling data;

[0051] According to the high-frequency temperature sampling data, updating the optimized temperature field distribution data to obtain an optimal temperature sensing result of the target monitoring area.

[0052] The embodiment of the application further provides an integrated temperature sensing system based on the high-efficiency fusion architecture, comprising:

[0053] a grid division module, configured to perform grid division on a target monitoring area, and to preliminarily deploy temperature sensors according to a division result;

[0054] a first abnormality processing module, configured to acquire temperature data in each grid unit in real time, to identify a first abnormal area according to the temperature data, to increase the deployment density of temperature sensors in the first abnormal area, and to obtain a sensor array layout distribution diagram of the target monitoring area;

[0055] a sensor position adjustment module, configured to calculate distance weights between adjacent temperature sensors according to the sensor array layout distribution diagram, and to adjust the position of a grid unit where a sensor is located based on the distance weights;

[0056] a delay correction module, configured to acquire the response time of each temperature sensor after adjustment, to calculate the delay degree of each temperature sensor, to perform weighted fusion on the temperature data according to the delay degree, and to obtain preliminary temperature field distribution data;

[0057] a second abnormality processing module, configured to identify a second abnormal area according to the preliminary temperature field distribution data, to perform deviation correction on the preliminary temperature field distribution data of the second abnormal area, and to obtain optimized temperature field distribution data;

[0058] a sampling frequency adjustment module, configured to, when the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, adjust the data collection frequency according to the temperature change rate, to obtain an optimal temperature sensing result of the target monitoring area.

[0059] Compared with existing technologies, this invention discloses an integrated temperature sensing method and system based on a high-efficiency fusion architecture. The method involves dividing the target monitoring area into grids and initially deploying temperature sensors based on the grid division results. Temperature data within each grid cell is acquired in real time. A first abnormal region is identified based on the temperature data, and the deployment density of temperature sensors in the first abnormal region is increased to obtain a sensor array layout distribution map of the target monitoring area. Based on the sensor array layout distribution map, distance weights between adjacent temperature sensors are calculated, and the grid cell positions of the sensors are adjusted based on these distance weights. The response time of each adjusted temperature sensor is acquired, and the delay level of each temperature sensor is calculated. The temperature data is then weighted and fused based on the delay level to obtain preliminary temperature field distribution data. A second abnormal region is identified based on the preliminary temperature field distribution data, and the preliminary temperature field distribution data for the second abnormal region is corrected for deviations to obtain optimized temperature field distribution data. When the temperature change rate of the optimized temperature field distribution data exceeds a preset rate threshold, the data acquisition frequency is adjusted based on the temperature change rate to obtain the optimal temperature sensing result for the target monitoring area. By employing the embodiments of the present invention, abnormal temperature regions can be dynamically identified, and data fusion logic can be optimized to improve the accuracy of temperature field distribution restoration and reduce data errors caused by unreasonable sensor layout or transmission delay. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating the steps of an integrated temperature sensing method based on a high-performance fusion architecture provided in an embodiment of the present invention.

[0061] Figure 2 This is a schematic diagram of an integrated temperature sensing system based on a high-efficiency fusion architecture provided in an embodiment of the present invention. Detailed Implementation

[0062] 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.

[0063] In the description and claims of the specification, the terms "first", "second", and the like, if any, are used merely as identifiers that identify different technical features, and are not intended to signify relative importance or a number of the technical features. The terms "first", "second", and the like, if any, are not necessarily used to describe a number or order of importance, but are used to identify one feature from another. The terms "first", "second", and the like, if any, are interchangeable under appropriate circumstances. Thus, a feature specified as "first" can also be termed, either explicitly or implicitly, as "second".

[0064] In the multi-source temperature sensor data acquisition and fusion process, the spatial layout of the temperature sensor array directly affects the comprehensiveness and accuracy of data acquisition. Unreasonable layout can easily lead to the loss of temperature information in some areas, especially in the case of large temperature deviation between the edge and the center area, it is difficult to capture the complete temperature change trend.

[0065] The problem of unreasonable sensor layout exacerbates the difference in response time of sensors at different positions, making the reliability of each sensor data different during data fusion, increasing the difficulty of error correction. At the same time, the difference in response time also affects the real-time tracking ability of the dynamic temperature field, and further causes deviation or delay in the temperature field reconstruction process.

[0066] Based on the above thinking, the embodiment of the present application provides an integrated temperature sensing method based on a high-efficiency fusion architecture. Please refer to Figure 1 In this embodiment, the integrated temperature sensing method based on the high-efficiency fusion architecture is specifically executed through steps S1 to S6:

[0067] S1, grid division is performed on a target monitoring area, and temperature sensors are preliminarily deployed according to the division result.

[0068] Grid division is a basic link of temperature monitoring. By abstracting the target monitoring area into a regular grid, a spatial reference can be provided for sensor layout, and support can be provided for subsequent abnormal area positioning and temperature sensor position optimization.

[0069] S2, real-time temperature data in each grid unit is acquired, a first abnormal area is identified according to the temperature data, the deployment density of temperature sensors in the first abnormal area is increased, and a sensor array layout distribution map of the target monitoring area is obtained.

[0070] By identifying the first abnormal area and dynamically adjusting the sensor deployment density, high-precision detection of temperature change areas can be achieved, and the monitoring blind area caused by traditional uniform point deployment can be effectively avoided.

[0071] S3, according to the sensor array layout distribution map, the distance weight between adjacent temperature sensors is calculated, and the grid unit position where the sensor is located is adjusted based on the distance weight.

[0072] A reasonable sensor layout makes the data collected by each sensor more spatially representative, enhances the complementarity between data, and provides richer and more accurate information in the data fusion process, thereby improving the overall monitoring system's ability to perceive the temperature field. In the embodiments of the present application, through distance weight calculation and position adjustment, the signal interference and data redundancy between sensors are effectively reduced.

[0073] S4, the response time of each temperature sensor after adjustment is obtained, the delay degree of each temperature sensor is calculated, the temperature data is weighted and fused according to the delay degree, and preliminary temperature field distribution data is obtained.

[0074] The data is weighted and fused based on the response time of the temperature sensor, the sensor information with fast response and accurate data is preferentially utilized, so that the preliminary temperature field distribution data can more truly reflect the actual temperature condition of the monitoring area, and the data error and distortion caused by sensor transmission delay are effectively suppressed. The data fusion mode given in the embodiments of the present application can enhance the fault tolerance of the monitoring, even if individual sensors are delayed or fail, the weights of the sensors can be adjusted to ensure the availability of the temperature field distribution data.

[0075] S5, a second abnormal area is identified according to the preliminary temperature field distribution data, deviation correction is performed on the preliminary temperature field distribution data of the second abnormal area, and optimized temperature field distribution data is obtained.

[0076] Through identification and deviation correction of the second abnormal area, the accuracy and integrity of the temperature data are effectively improved.

[0077] S6, when the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, the data acquisition frequency is adjusted according to the temperature change rate to obtain optimal temperature perception results of the target monitoring area.

[0078] For the area with rapid temperature change, the data acquisition frequency is increased to obtain temperature data more frequently and capture temperature change details in time; for the area with stable temperature change, the data acquisition frequency is reduced to save system resources, realizing a win-win of monitoring performance and resource utilization.

[0079] In the above scheme, in an environment with rapid temperature gradient change, the fusion weights of the data of each sensor are dynamically adjusted based on the spatial layout optimization and response time difference of the sensor array, thereby effectively improving the accuracy of temperature field reconstruction and reducing the system response delay.

[0080] As a preferred embodiment, step S1, the target monitoring area is divided into grids, and the temperature sensors are preliminarily deployed according to the division result, including:

[0081] According to the preset grid size, the target monitoring area is evenly divided to obtain a grid unit;

[0082] A temperature sensor is disposed at the center point of each grid unit.

[0083] For example, for a 100m x 100m industrial plant monitoring area, it can be divided according to a 5m x 5m grid size to form 400 grid units. The center point of each grid unit is used as the deployment position of the temperature sensor. This regular arrangement facilitates subsequent data collection and processing.

[0084] As a preferred embodiment, in step S2, real-time temperature data of each grid unit is obtained, a first abnormal area is identified according to the temperature data, the deployment density of the temperature sensor in the first abnormal area is increased, and a sensor array layout distribution map of the target monitoring area is obtained. Steps S21-S24 are performed:

[0085] S21, real-time temperature data of each grid unit is obtained to form an initial temperature distribution matrix.

[0086] In some preferred embodiments, when the temperature sensor obtains the temperature data, the temperature data is arranged according to the row and column positions of the grid unit to form a two-dimensional matrix structure as an initial temperature distribution matrix. This matrix data organization makes the temperature relationship between adjacent grid units clear and visible, laying a data foundation for subsequent gradient calculation.

[0087] S22, according to the initial temperature distribution matrix, the temperature difference between adjacent grid units is calculated.

[0088] S23, according to the temperature difference and the spacing of the grid unit, the temperature gradient value of the boundary between the grid units is obtained.

[0089] The calculation of the temperature gradient value reflects the temperature change rate between adjacent grid units.

[0090] In some preferred embodiments, the temperature difference is divided by the grid spacing to obtain the temperature change per unit distance, that is, the temperature gradient of the boundary between the grid units.

[0091] S24, when the temperature gradient value is greater than a preset gradient threshold, the grid units on both sides of the boundary are marked as a first abnormal area.

[0092] It can be understood that when an industrial equipment overheats, the temperature difference between the grid unit and the surrounding grid units will increase significantly, resulting in a gradient value exceeding the normal range. The gradient threshold is usually based on historical data and safety standards. When the threshold is exceeded, it indicates that there is a temperature abnormality problem in the grid unit.

[0093] S25, performing secondary grid division on the first abnormal area, increasing the deployment density of temperature sensors according to the secondary grid division result, and obtaining a sensor array layout distribution map of the target monitoring area.

[0094] Further, preferably, step S25, performing secondary grid division on the first abnormal area, increasing the deployment density of temperature sensors according to the secondary grid division result, and obtaining a sensor array layout distribution map of the target monitoring area, comprises:

[0095] performing secondary grid division on the grid cells in the first abnormal area to obtain sub-grid cells;

[0096] increasing temperature sensors at the center points of the sub-grid cells;

[0097] forming a sensor array layout distribution map of the target monitoring area according to the position information of all the temperature sensors.

[0098] In the embodiment of the present application, the first abnormal area is divided by using a bisection strategy, for example, an abnormal grid originally 5m x 5m is divided into four sub-grids 2.5m x 2.5m, and a new sensor deployment point is added at the center of each sub-grid.

[0099] It should be noted that after the secondary grid division on the first abnormal area, the target monitoring area will be divided into a plurality of grid cells with two different sizes, and in the subsequent calculation and collection process, the grid cells with the smallest size will be used as the basis. For example, if a 5m x 5m grid cell A is marked as an abnormal grid cell, and it is divided into 2.5m x 2.5m grid cells A1-A4, then in the subsequent data collection of grid cell A, it will be analyzed whether there are sub-grid cells, and then the data of sub-grid cells A1-A4 will be analyzed; for a normal 5m x 5m grid cell B, the data analysis will be directly performed on the grid cell B without the need for grid division.

[0100] In the above scheme, the temperature monitoring accuracy of the abnormal area is significantly improved through encrypted deployment, and the heat source location and temperature distribution characteristics can be more accurately located. The formation process of the high-density arrangement coordinate set embodies the idea of differentiated deployment, that is, the normal area maintains the original sensor density, and the abnormal area realizes key monitoring by increasing the number of sensors. This strategy not only ensures the comprehensiveness of monitoring, but also avoids waste of resources, achieving a balance between monitoring efficiency and cost. The final sensor array layout distribution map presents a well-proportioned feature of sparseness and density. In the area with stable temperature, the sensor spacing is large; in the area with abnormal temperature gradient, the sensors are densely distributed. This adaptive layout can dynamically respond to the temperature distribution characteristics of the monitoring area, ensuring the focus on potential risk areas and improving the reliability and early warning capability of the entire temperature monitoring system.

[0101] As a preferred embodiment, step S3, according to the sensor array layout distribution map, calculates the distance weight between adjacent temperature sensors, and adjusts the grid cell position where the sensor is located based on the distance weight, which is specifically implemented through steps S31-S35:

[0102] S31, according to the sensor array layout distribution map, obtains the spatial position coordinates of each temperature sensor.

[0103] In some preferred embodiments, the sensor array layout distribution map includes a unique identifier and a two-dimensional coordinate value of each temperature sensor. The two-dimensional coordinate value reflects the spatial position coordinates of the sensor in the monitoring area.

[0104] And it should be noted that the spatial position coordinates of the temperature sensor are also associated with its data acquisition frequency. For example, the data acquisition frequency of the key area can be once per minute, while the ordinary area may be once every five minutes.

[0105] S32, according to the spatial position coordinates, calculate the Euclidean distance between each temperature sensor and its adjacent temperature sensor.

[0106] S33, according to the Euclidean distance between adjacent temperature sensors and the monitoring radius range of the temperature sensor, calculate the distance weight.

[0107] It can be understood that if the sum of the monitoring radii of adjacent temperature sensors is greater than the Euclidean distance between them, it means that the monitoring areas overlap. Although this overlap improves the reliability of local monitoring, it also causes waste of resources, and the distance weight will decrease as the overlap degree increases. The design of the distance weight makes the sensor combination with serious overlap obtain a lower score, thereby triggering the position adjustment mechanism.

[0108] S34, when the distance weight of the target temperature sensor is less than a preset weight threshold, counting the number of temperature sensors deployed in adjacent grid cells, and regarding the grid cell with the least number of temperature sensors as a target grid cell.

[0109] S35, adjusting the target temperature sensor from the original grid cell to the target grid cell.

[0110] Exemplarily, when the distance weight is lower than 0.3, it indicates that the sensor layout is too dense and needs to be adjusted. In the embodiment of the present application, the identification of the adjacent grid cell adopts an eight-neighborhood method, that is, the eight grids around the target grid are regarded as adjacent cells. By counting the number of sensors in these grids one by one, a distribution density atlas is formed. The grid with the least number of sensors is selected as the new deployment position, realizing the migration of sensors from high-density areas to low-density areas, achieving the dynamic balance of sensor distribution in the whole monitoring area, and improving the overall efficiency of the temperature monitoring network. This adjustment is not a simple movement, but a comprehensive consideration of the balance of the overall layout.

[0111] As a preferred embodiment, step S4, obtaining the response time of each temperature sensor after adjustment, calculating the delay degree of each temperature sensor, and weighting and fusing the temperature data according to the delay degree to obtain preliminary temperature field distribution data, is specifically implemented through steps S41-S45:

[0112] S41, obtaining the response time of each temperature sensor after adjustment, and performing standardization processing on the response time to obtain the response time deviation coefficient of each temperature sensor.

[0113] It should be noted that in the embodiment of the present application, the response time is the time interval from the beginning of the sensor sensing temperature change to the final output of stable measurement data. This time interval includes the internal heat conduction process, signal conversion process and data processing process of the sensor.

[0114] For thermocouple type temperature sensors, the response time is usually between 0.5 seconds and 5 seconds, while the response time of infrared sensors can be shorter, reaching the level of 0.1 seconds. The recording of the response time data forms the basis for evaluation.

[0115] S42, according to a preset deviation coefficient threshold and the response time deviation coefficient, dividing the temperature sensors into high-delay sensors and low-delay sensors.

[0116] S43, according to the response time deviation coefficient and importance of the low-delay sensors and the high-delay sensors, reducing the preset initial weight of the high-delay sensors, and calculating the weight correction coefficient of the high-delay sensors.

[0117] It should be noted that the importance is determined based on the monitoring area where the temperature sensor is located. In the embodiments of the present application, the importance of the temperature sensor is represented by an importance level, and the higher the value of the importance level is, the more important it is.

[0118] For example, the area near the key production equipment can be assigned the highest level 5, the general storage area level 3, and the auxiliary passage area level 1. This grading mechanism enables the system to manage differentiated sensor weights according to actual business needs.

[0119] S44, dynamically adjusting the weight of each temperature sensor in the first abnormal area according to the weight correction coefficient, to obtain a high-delay fusion temperature value.

[0120] S45, forming preliminary temperature field distribution data covering the target monitoring area according to the high-delay fusion temperature value and the preset initial weight of the low-delay sensor.

[0121] It should be noted that the preset initial weight is determined based on the sensor model, accuracy level, etc. The initial weight value of a high-precision sensor is higher.

[0122] In some preferred embodiments, step S41, obtaining the response time of each adjusted temperature sensor, the response time is standardized to obtain the response time deviation coefficient of each temperature sensor, comprising:

[0123] Obtaining the response time of each adjusted temperature sensor, calculating the average response time and the standard deviation;

[0124] Subtracting the average response time from the response time and dividing by the standard deviation to obtain the standardized response time as the response time deviation coefficient.

[0125] By subtracting the average value from the response time of each sensor, the deviation of the sensor from the average level can be obtained. Dividing by the standard deviation converts this deviation into a dimensionless standardized value. A positive value indicates that the response time is higher than the average level, and a negative value indicates that the response time is lower than the average level. The absolute value of the number reflects the severity of the deviation. The physical meaning of the response time deviation coefficient is to quantify the response delay of the sensor.

[0126] Exemplarily, when the response time deviation coefficient is 2, it means that the response time of the sensor is higher than the average value by two standard deviations. According to the normal distribution theory, the probability of this situation is low, which usually means that there is a performance problem with the sensor. The selection of the preset threshold is usually based on the requirements of the system on the response speed. For a fire warning system that requires fast response, the threshold can be set to 1.5, while for general environmental temperature monitoring, the threshold can be appropriately relaxed to 2.5.

[0127] Further, preferably, step S43, according to the response time deviation coefficient and the importance of the low-delay sensor and the high-delay sensor, the preset initial weight of the high-delay sensor is reduced, and the weight correction coefficient of the high-delay sensor is calculated, comprising:

[0128] According to the position of the high-delay sensor in the target monitoring area, the importance level of the high-delay sensor is obtained;

[0129] According to the response time deviation coefficient and the importance level of the high-delay sensor, a weight reduction factor is obtained;

[0130] The preset initial weight of the high-delay sensor is reduced by the weight reduction factor to obtain a reduced weight value;

[0131] The Pearson correlation coefficient is used to calculate the correlation between the historical data sequences of the high-delay sensor and the low-delay sensor;

[0132] According to the reduced weight value and the correlation, the weight correction coefficient of the high-delay sensor is calculated.

[0133] In some preferred embodiments, the response time deviation is multiplied by the importance level to obtain the weight reduction factor.

[0134] Exemplarily, when the response time deviation coefficient of a certain temperature sensor is 2.5 and the importance level of the area is 4, the weight reduction factor is calculated to be 10. When reducing the weight, the preset initial weight of the temperature sensor needs to be divided by 10 to obtain the reduced weight value, which greatly reduces its contribution in temperature data fusion.

[0135] It should be noted that the historical data sequence needs to consider the periodic characteristics of temperature changes, and generally 24 hours or more containing complete day and night temperature difference changes are selected. The historical data of the low-delay sensor is regarded as a reliable reference for temperature changes in the area, and the data sequence reflects the real temperature change trend. By calculating the correlation between the high-delay sensor and the low-delay sensor, the linear correlation degree of the two data sequences can be measured.

[0136] In some preferred embodiments, the correlation coefficient ranges from -1 to 1, and a value close to 1 indicates that the temperature measurements of the two sensors are highly consistent, and a value close to 0 indicates that the correlation is weak. If the correlation coefficient of the high-delay sensor and the low-delay sensor is high, it means that although there is a response delay, the measurement trend is still reliable. The determination of the weight correction coefficient takes into account the delay characteristics and data reliability of the sensor. In the embodiments of the present application, the dynamic adjustment of the weight based on data quality is realized by multiplying the correlation coefficient by the weight value after weight reduction.

[0137] Preferably, step S44, dynamically adjusting the weight of each temperature sensor in the first abnormal area according to the weight correction coefficient, to obtain a high-delay fusion temperature value, comprises:

[0138] Multiplying the weight correction coefficient by the preset initial weight to obtain a fusion weight value;

[0139] Using a weighted average algorithm to fuse and calculate the temperature data of the temperature sensors in the first abnormal area to obtain a high-delay fusion temperature value.

[0140] In one possible implementation, the weight correction coefficient of the high-delay sensor is usually less than 1, such as 0.6 or 0.7, which reflects the loss of timeliness of its data. The low-delay sensor keeps its preset weight value unchanged, which is usually 1.0. This differentiated weight allocation ensures that the sensor with fast response dominates in data fusion, while not completely excluding the contribution of the high-delay sensor.

[0141] Specifically, the preset weight value of the high-precision sensor can be 1.2, the standard sensor can be 1.0, and the economic sensor can be 0.8. When these preset values are multiplied by the correction coefficient, the adjusted fusion weight value reflecting the comprehensive performance of the sensor is obtained.

[0142] Exemplarily, a high-precision sensor with a preset weight of 1.2, if its correction coefficient is 0.6, the adjusted fusion weight value is 0.72. The application of the weighted average algorithm in temperature data fusion ensures the reasonable integration of multi-source data.

[0143] In the above scheme, different contribution ratios are assigned to each sensor according to its reliability. Assuming there are three sensors in a certain grid cell, with temperature readings of 25.3 degrees, 25.8 degrees and 26.1 degrees, and the corresponding fusion weight values are 0.72, 1.0 and 0.5, respectively, the fusion calculation process is as follows: first, calculate the products 18.216, 25.8 and 13.05, sum them up to get 57.066, then divide by the weight sum 2.22, and finally get the fusion temperature value of the grid 25.7 degrees. The processing of temperature gradient change abnormal area and normal area embodies the idea of classification management. In the abnormal area, due to the sharp change of temperature, the sensor density is higher, and the fusion calculation involves more data sources. In the normal area, the sensor distribution is relatively sparse, and the fusion process is relatively simple. This differential treatment not only ensures the monitoring accuracy of the key area, but also avoids the waste of computing resources. After the fusion temperature value of the grid cell is calculated, it needs to be organized according to the spatial position.

[0144] As a preferred embodiment, step S5, identifying a second abnormal area according to the preliminary temperature field distribution data, correcting the preliminary temperature field distribution data of the second abnormal area to obtain optimized temperature field distribution data, is executed through steps S51-S54:

[0145] S51, according to the preliminary temperature field distribution data, calculate the temperature difference value of each grid cell at consecutive time points to obtain the temperature change rate.

[0146] Preferably, the temperature difference value of each grid cell at adjacent time points is divided by the time interval to obtain the temperature change rate.

[0147] S52, according to the temperature change rate, identify a second abnormal area.

[0148] For example, temperature data is collected every 5 minutes, if the temperature of a certain grid cell at two consecutive time points is 25 degrees and 28 degrees respectively, the temperature difference is 3 degrees, and divided by the time interval of 5 minutes, the temperature change rate is 0.6 degrees per minute. This rapid temperature rise may indicate overheating of equipment or fire hazard and needs special attention. The determination of the preset threshold needs to consider the characteristics of different monitoring scenes.

[0149] Specifically, in a chemical production workshop, due to the exothermic characteristics of chemical reactions, the normal temperature change rate threshold may be set to 0.3 degrees per minute; while in a general warehouse environment, the threshold may be only 0.1 degrees per minute. When the actual change rate exceeds the corresponding threshold, the system will mark the area as abnormal in change trend, triggering the subsequent data correction process.

[0150] S53, extract the normal temperature values of the normal grid cells adjacent to the four directions of each abnormal grid cell in the second abnormal area, and calculate the interpolation temperature of the abnormal grid cell using a bilinear interpolation algorithm.

[0151] In some preferred embodiments, a two-dimensional interpolation calculation is performed by using the temperature values of four adjacent normal grids. Assuming that the temperatures of the four grids adjacent to the abnormal grid are 24 degrees, 26 degrees, 25 degrees, and 25.5 degrees, respectively, the interpolation process will perform a weighted calculation according to the relative distances from the abnormal grid to these four points, and finally obtain a smooth transition interpolation temperature, such as 25.2 degrees.

[0152] The selection of the four adjacent grids follows the principle of spatial adjacency. In the gridded monitoring area, each internal grid has four directly adjacent grids above, below, to the left, and to the right. This neighborhood structure makes the interpolation calculation have a clear spatial reference, avoiding the interference of distant grids on local temperature estimation. When the abnormal grid is located at the boundary, the number of available adjacent grids will decrease, and the interpolation algorithm will adjust the calculation accordingly. The determination of the correction value embodies the idea of data quality control.

[0153] S54, replace the temperature value of the abnormal grid cell with the interpolation temperature to obtain the optimized temperature field distribution data.

[0154] For example, if the temperature of a certain abnormal grid in the preliminary fusion data set is 35 degrees, and the reasonable temperature obtained by interpolation calculation is 25.2 degrees, then the correction value is -9.8 degrees.

[0155] It can be understood that such a large correction usually means that the sensor at this location may be malfunctioning or affected by an abnormal heat source. The optimized temperature field distribution data achieves data smoothness and continuity. By replacing the temperature values of all abnormal areas with the interpolation results, the data discontinuity caused by sensor failure or local interference is eliminated. This optimization not only improves the accuracy of temperature field representation, but also provides a more reliable data basis for subsequent temperature field analysis and warning decisions, ensuring that even in the case of partial sensor abnormalities, complete and reasonable temperature field distribution information can still be provided.

[0156] As a preferred implementation, step S6, when the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, the data acquisition frequency is adjusted according to the temperature change rate to obtain the optimal temperature sensing result of the target monitoring area, which is implemented by steps S61-S63:

[0157] S61, obtain the temperature change rate of each grid cell according to the optimized temperature field distribution data.

[0158] In some preferred embodiments, the temperature change rate of each grid cell is obtained by dividing the temperature difference between two adjacent temperature acquisition times by the time interval.

[0159] Specifically, if the temperature of a grid cell at the last acquisition time is 25 degrees, the current temperature is 27 degrees, and the time interval is 1 minute, the temperature change rate is 2 degrees per minute. This rate value directly reflects the speed of temperature rise in this area and is an important basis for determining whether intensive monitoring is needed. The determination of the preset rate threshold needs to consider the safety requirements of different application scenarios.

[0160] S62, when the temperature change rate of the grid cell is greater than the preset rate threshold, the data acquisition frequency of the temperature sensor in the grid cell is increased to a preset high-frequency acquisition frequency, and high-frequency temperature sampling data is obtained.

[0161] It should be noted that in the monitoring of battery energy storage stations, due to the risk of thermal runaway, the rate threshold may be set to 1 degree per minute; while in ordinary office areas, the rate threshold can be relaxed to 3 degrees per minute.

[0162] It can be understood that when the actual temperature change rate exceeds the rate threshold, it means that there may be an abnormal heat source or a risk of equipment failure in this area. The sending of the frequency adjustment instruction realizes the dynamic allocation of monitoring resources. The conventional acquisition frequency may be once per minute, while the high-frequency acquisition frequency is increased to once every 10 seconds. This six-fold frequency increase enables the system to more accurately track the rapid temperature change process. After receiving the adjustment instruction, the sensor immediately changes its internal timer settings and performs data acquisition and reporting according to the new frequency. Intensive temperature sampling data provides more fine-grained temperature change information.

[0163] S63, according to the high-frequency temperature sampling data, updating the optimized temperature field distribution data to obtain the optimal temperature sensing result of the target monitoring area.

[0164] The updated optimized temperature field distribution data can help accurately depict the rising curve of temperature, help determine whether the temperature change is continuously accelerated or gradually slowed down, and provide a basis for subsequent early warning decisions. The real-time updating mechanism ensures the timeliness of the monitoring results. Whenever new temperature sampling data is received, the system immediately updates the temperature value of the corresponding grid cell. This immediate update is not only reflected in the numerical update of a single grid, but also includes the dynamic refreshing of the entire temperature field distribution map. In this way, monitoring personnel can real-time master the temperature distribution condition of the entire monitoring area, especially those high-risk areas with rapid temperature changes. The entire adaptive monitoring process forms a closed-loop control mechanism. From continuous monitoring to rate calculation, from threshold judgment to frequency adjustment, to data updating, each link is closely connected, and together they form an intelligent monitoring system that can automatically respond to abnormal temperature changes.

[0165] The integrated temperature sensing method based on the high-efficiency fusion architecture provided by the embodiment of the application can effectively improve the accuracy of temperature field reconstruction and reduce system response delay by dynamically adjusting the fusion weight of each sensor data based on the spatial layout optimization and response time difference of the sensor array. In addition, by continuously identifying abnormal areas, adjusting sensor layout, correcting data deviation and adaptively adjusting the sampling frequency, the environmental changes can be responded to in real time, and the monitoring performance can be continuously improved.

[0166] The embodiment of the application provides an integrated temperature sensing system based on a high-efficiency fusion architecture. Figure 2 The integrated temperature sensing system based on the high-efficiency fusion architecture comprises a grid division module 11, a first abnormality processing module 12, a sensor position adjustment module 13, a delay correction module 14, a second abnormality processing module 15 and a sampling frequency adjustment module 16, wherein:

[0167] The grid division module 11 is used for grid division of a target monitoring area, and temperature sensors are preliminarily deployed according to the division result.

[0168] The first abnormality processing module 12 is used for real-time acquisition of temperature data in each grid cell, identification of a first abnormal area according to the temperature data, increase of the deployment density of temperature sensors in the first abnormal area, and obtaining of a sensor array layout distribution map of the target monitoring area.

[0169] The sensor position adjustment module 13 is used for calculating the distance weight between adjacent temperature sensors according to the sensor array layout distribution map, and adjusting the position of the grid cell where the sensor is located based on the distance weight.

[0170] The delay correction module 14 is configured to obtain an adjusted response time of each temperature sensor, calculate a delay degree of each temperature sensor, and perform weighted fusion on the temperature data according to the delay degree to obtain preliminary temperature field distribution data.

[0171] The second anomaly processing module 15 is configured to identify a second anomaly region according to the preliminary temperature field distribution data, and correct deviation of the preliminary temperature field distribution data of the second anomaly region to obtain optimized temperature field distribution data.

[0172] The sampling frequency adjustment module 16 is configured to adjust a data acquisition frequency according to a temperature change rate of the optimized temperature field distribution data when the temperature change rate is greater than a preset rate threshold, so as to obtain an optimal temperature sensing result of the target monitoring region.

[0173] As a preferred implementation, the grid division module 11 comprises:

[0174] The region equal division unit is configured to equally divide the target monitoring region according to a preset grid size to obtain a grid unit.

[0175] The first sensor deployment unit is configured to deploy a temperature sensor at a center point of each grid unit.

[0176] As a preferred implementation, the first anomaly processing module 12 comprises:

[0177] The initial temperature distribution matrix generation unit is configured to obtain temperature data in each grid unit in real time to form an initial temperature distribution matrix.

[0178] The temperature difference calculation unit is configured to calculate a temperature difference between adjacent grid units according to the initial temperature distribution matrix.

[0179] The gradient calculation unit is configured to obtain a temperature gradient value of a boundary between grid units according to the temperature difference and a spacing of the grid units.

[0180] The first anomaly region identification unit is configured to mark grid units on both sides of the boundary as a first anomaly region when the temperature gradient value is greater than a preset gradient threshold.

[0181] The second sensor deployment unit is configured to perform secondary grid division on the first anomaly region, increase a deployment density of temperature sensors according to a secondary grid division result, and obtain a sensor array layout distribution map of the target monitoring region.

[0182] Further, preferably, the second sensor deployment unit is specifically configured to:

[0183] The grid cells in the first abnormal area are subdivided to obtain sub-grid cells;

[0184] A temperature sensor is added at the center point of each sub-grid cell;

[0185] According to the position information of all the temperature sensors, a sensor array layout distribution map of the target monitoring area is formed.

[0186] As a preferred embodiment, the sensor position adjustment module 13 comprises:

[0187] A spatial position coordinate acquisition unit is configured to obtain the spatial position coordinates of each temperature sensor according to the sensor array layout distribution map;

[0188] An Euclidean distance calculation unit is configured to calculate the Euclidean distance between each temperature sensor and its adjacent temperature sensors according to the spatial position coordinates;

[0189] A distance weight calculation unit is configured to calculate the distance weight according to the Euclidean distance between adjacent temperature sensors and the monitoring radius range of the temperature sensor;

[0190] A target grid cell screening unit is configured to, when the distance weight of a target temperature sensor is less than a preset weight threshold, count the number of temperature sensors deployed in adjacent grid cells, and regard the grid cell with the least number of temperature sensors as a target grid cell;

[0191] A third sensor deployment unit is configured to adjust the target temperature sensor from the original grid cell to the target grid cell.

[0192] As a preferred embodiment, the delay correction module 14 comprises:

[0193] A response time deviation coefficient calculation unit is configured to obtain the response time of each adjusted temperature sensor, perform standardization processing on the response time, and obtain the response time deviation coefficient of each temperature sensor;

[0194] A delay level division unit is configured to divide the temperature sensors into high-delay sensors and low-delay sensors according to a preset deviation coefficient threshold and the response time deviation coefficient;

[0195] A weight correction coefficient calculation unit is configured to reduce the preset initial weight of the high-delay sensor according to the response time deviation coefficient and importance of the low-delay sensor and the high-delay sensor, and calculate the weight correction coefficient of the high-delay sensor;

[0196] a high-delay fusion temperature value calculation unit configured to dynamically adjust the weight of each temperature sensor in the first abnormal region according to the weight correction coefficient to obtain a high-delay fusion temperature value;

[0197] a preliminary temperature field distribution data generation unit configured to form preliminary temperature field distribution data covering the target monitoring region according to the high-delay fusion temperature value and the preset initial weight of the low-delay sensor.

[0198] Further, preferably, the weight correction coefficient calculation unit is specifically configured to:

[0199] obtain the importance level of the high-delay sensor according to the position of the high-delay sensor in the target monitoring region;

[0200] obtain a weight reduction factor according to the response time deviation coefficient and the importance level of the high-delay sensor;

[0201] reduce the preset initial weight of the high-delay sensor by the weight reduction factor to obtain a weight value after weight reduction;

[0202] calculate the correlation between the historical data sequence of the high-delay sensor and the low-delay sensor by using a Pearson correlation coefficient;

[0203] calculate the weight correction coefficient of the high-delay sensor according to the weight value after weight reduction and the correlation.

[0204] As a preferred embodiment, the second abnormality processing module 15 comprises:

[0205] a temperature change rate calculation unit configured to calculate the temperature difference value of each grid cell at consecutive time points to obtain a temperature change rate according to the preliminary temperature field distribution data;

[0206] a second abnormal region identification unit configured to identify a second abnormal region according to the temperature change rate;

[0207] an interpolation temperature calculation unit configured to extract the normal temperature value of the normal grid cell adjacent to the abnormal grid cell in the second abnormal region in four directions, and calculate the interpolation temperature of the abnormal grid cell by using a bilinear interpolation algorithm;

[0208] an optimized temperature field distribution data generation unit configured to replace the temperature value of the abnormal grid cell with the interpolation temperature to obtain optimized temperature field distribution data.

[0209] As a preferred embodiment, the sampling frequency adjustment module 16 comprises:

[0210] a temperature change rate calculation unit configured to obtain a temperature change rate of each grid unit according to the optimized temperature field distribution data;

[0211] a sampling frequency raising unit configured to raise a data collection frequency of the temperature sensor in the grid unit to a preset high-frequency collection frequency when the temperature change rate of the grid unit is greater than a preset rate threshold, to obtain high-frequency temperature sampling data;

[0212] a perception result generation unit configured to update the optimized temperature field distribution data according to the high-frequency temperature sampling data, to obtain an optimal temperature perception result of the target monitoring area.

[0213] The integrated temperature perception system based on the high-efficiency fusion architecture provided by the embodiment of the present application can effectively improve the accuracy of temperature field reconstruction and reduce system response delay by dynamically adjusting the fusion weight of each sensor data based on the spatial layout optimization and response time difference of the sensor array. In addition, the system can respond to environmental changes in real time and continuously improve monitoring performance by continuously identifying abnormal areas, adjusting sensor layout, correcting data deviation and adaptively adjusting collection frequency.

[0214] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when the program is executed, the processes of the above-mentioned embodiments can be included. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM), etc.

[0215] The above is the preferred embodiment of the present application. It should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the present application. These improvements and refinements are also considered within the scope of protection of the present application.

Claims

1. An integrated temperature-aware approach based on a high-performance fusion architecture, characterized by, The method comprises the following steps: grid division is performed on a target monitoring area, and temperature sensors are preliminarily deployed according to the division result; temperature data in each grid unit is acquired in real time, a first abnormal area is identified according to the temperature data, the deployment density of the temperature sensors in the first abnormal area is increased, and a sensor array layout distribution map of the target monitoring area is obtained; distance weights between adjacent temperature sensors are calculated according to the sensor array layout distribution map, and the positions of the grid units where the sensors are located are adjusted based on the distance weights; the response time of each temperature sensor after adjustment is acquired, the delay degree of each temperature sensor is calculated, the temperature data is weighted and fused according to the delay degree, and preliminary temperature field distribution data is obtained; a second abnormal area is identified according to the preliminary temperature field distribution data, deviation correction is performed on the preliminary temperature field distribution data of the second abnormal area, and optimized temperature field distribution data is obtained; when the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, the data acquisition frequency is adjusted according to the temperature change rate, and optimal temperature sensing results of the target monitoring area are obtained.

2. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 1, wherein, The method of grid division on a target monitoring area and preliminary deployment of temperature sensors according to the division result comprises the following steps: The target monitoring area is equally divided according to a preset grid size, and a grid unit is obtained. A temperature sensor is deployed at the center point position of each grid unit.

3. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 1, wherein, The method of real-time acquisition of temperature data in each grid unit, identification of a first abnormal area according to the temperature data, and increase of the deployment density of temperature sensors in the first abnormal area to obtain a sensor array layout distribution map of the target monitoring area comprises the following steps: The temperature data in each grid unit is acquired in real time, and an initial temperature distribution matrix is formed. The temperature difference between adjacent grid units is calculated according to the initial temperature distribution matrix. The temperature gradient value of the boundary between grid units is obtained according to the temperature difference and the spacing of the grid units. When the temperature gradient value is greater than a preset gradient threshold, the grid units on both sides of the boundary are marked as the first abnormal area. The first abnormal area is subjected to secondary grid division, the deployment density of temperature sensors is increased according to the secondary grid division result, and a sensor array layout distribution map of the target monitoring area is obtained.

4. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 3, wherein, The method of secondary grid division on the first abnormal area and increase of the deployment density of temperature sensors according to the secondary grid division result to obtain a sensor array layout distribution map of the target monitoring area comprises the following steps: The grid units in the first abnormal area are subjected to secondary grid division, and sub-grid units are obtained. Temperature sensors are added at the center point positions of the sub-grid units. The spatial position coordinates of each temperature sensor are obtained according to the sensor array layout distribution map.

5. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 1, wherein, The spatial position coordinates of each temperature sensor are obtained according to the sensor array layout distribution map. ​ According to the spatial position coordinates, the Euclidean distance between each temperature sensor and its adjacent temperature sensor is calculated; According to the Euclidean distance between adjacent temperature sensors and the monitoring radius range of the temperature sensor, the distance weight is calculated; When the distance weight of the target temperature sensor is less than the preset weight threshold, the number of temperature sensors deployed in the adjacent grid cell is counted, and the grid cell with the smallest number of temperature sensors is regarded as the target grid cell; The target temperature sensor is adjusted from the original grid cell to the target grid cell.

6. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 1, wherein, The response time of each temperature sensor after adjustment is obtained, the delay degree of each temperature sensor is calculated, the temperature data is weighted and fused according to the delay degree, and preliminary temperature field distribution data is obtained, including: The response time of each temperature sensor after adjustment is obtained, the response time is standardized to obtain the response time deviation coefficient of each temperature sensor; According to the preset deviation coefficient threshold and the response time deviation coefficient, the temperature sensor is divided into high delay sensor and low delay sensor; According to the response time deviation coefficient and importance of the low delay sensor and the high delay sensor, the preset initial weight of the high delay sensor is reduced, and the weight correction coefficient of the high delay sensor is calculated; According to the weight correction coefficient, the weight of each temperature sensor in the first abnormal area is dynamically adjusted to obtain a high delay fusion temperature value; According to the high delay fusion temperature value and the preset initial weight of the low delay sensor, preliminary temperature field distribution data covering the target monitoring area is formed.

7. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 6, wherein, According to the response time deviation coefficient and importance of the low delay sensor and the high delay sensor, the preset initial weight of the high delay sensor is reduced, and the weight correction coefficient of the high delay sensor is calculated, including: According to the position of the high delay sensor in the target monitoring area, the importance level of the high delay sensor is obtained; According to the response time deviation coefficient and importance level of the high delay sensor, a weight reduction factor is obtained; The preset initial weight of the high delay sensor is reduced by using the weight reduction factor to obtain a weight value after weight reduction; The Pearson correlation coefficient is used to calculate the correlation between the historical data sequences of the high delay sensor and the low delay sensor; According to the weight value after weight reduction and the correlation, the weight correction coefficient of the high delay sensor is calculated.

8. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 1, wherein, According to the preliminary temperature field distribution data, the temperature difference of each grid cell at consecutive time points is calculated to obtain a temperature change rate; According to the temperature change rate, a second abnormal area is identified; The normal temperature values of the normal grid cells adjacent to each abnormal grid cell in the second abnormal area are extracted, and the bilinear interpolation algorithm is used to calculate the interpolation temperature of the abnormal grid cell. ​ The interpolation temperature is used to replace the temperature value of the abnormal grid cell to obtain optimized temperature field distribution data.

9. An integrated temperature sensing method based on a high performance fusion architecture as claimed in claim 1, wherein, When the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, the data acquisition frequency is adjusted according to the temperature change rate to obtain an optimal temperature sensing result of the target monitoring area, including: According to the optimized temperature field distribution data, the temperature change rate of each grid cell is obtained. When the temperature change rate of the grid cell is greater than the preset rate threshold, the data acquisition frequency of the temperature sensor in the grid cell is increased to a preset high-frequency acquisition frequency to obtain high-frequency temperature sampling data. According to the high-frequency temperature sampling data, the optimized temperature field distribution data is updated to obtain the optimal temperature sensing result of the target monitoring area.

10. An integrated temperature-aware system based on a high-performance fusion architecture, comprising: Including: The grid division module is configured to divide the target monitoring area into grids and preliminarily deploy temperature sensors according to the division result. The first abnormality processing module is configured to acquire temperature data in each grid cell in real time, identify a first abnormal area according to the temperature data, increase the deployment density of temperature sensors in the first abnormal area, and obtain a sensor array layout distribution map of the target monitoring area. The sensor position adjustment module is configured to calculate the distance weight between adjacent temperature sensors according to the sensor array layout distribution map, and adjust the position of the grid cell where the sensor is located based on the distance weight. The delay correction module is configured to acquire the response time of each temperature sensor after adjustment, calculate the delay degree of each temperature sensor, weight and fuse the temperature data according to the delay degree, and obtain preliminary temperature field distribution data. The second abnormality processing module is configured to identify a second abnormal area according to the preliminary temperature field distribution data, correct the deviation of the preliminary temperature field distribution data of the second abnormal area, and obtain optimized temperature field distribution data. The sampling frequency adjustment module is configured to adjust the data acquisition frequency according to the temperature change rate when the temperature change rate of the optimized temperature field distribution data is greater than a preset rate threshold, so as to obtain an optimal temperature sensing result of the target monitoring area.

Images