temperature sensor

Abstract

This invention relates to a temperature sensor, comprising: a package, and multiple temperature measuring units with different temperature measurement principles disposed within the package, each of the temperature measuring units being thermally coupled to the object being measured via a thermally conductive structure; and a processor electrically connected to each of the temperature measuring units, the processor being configured to determine anomalies in each of the temperature measuring units based on measurement differences between the temperature measurements of the individual temperature measuring units, and / or based on a reference temperature value determined from the temperature measurements; and to determine a temperature output result based on the temperature measurements of the temperature measuring units determined to be non-abnormal. This approach can integrate multiple different types of sensing elements, utilizing the complementary advantages of the characteristics of different sensing elements to achieve high-precision temperature measurement.

Description

Technical Field

[0001] This invention relates to the field of sensor technology, and in particular to a temperature sensor. Background Technology

[0002] Temperature sensors are one of the fundamental and critical measuring instruments in industrial process control, air conditioning systems, energy management, and other fields. Currently, commonly used temperature sensors mainly include platinum resistance temperature sensors, thermocouple temperature sensors, and semiconductor temperature sensors.

[0003] In practical industrial applications, regardless of the type of temperature sensor used, accuracy drift occurs after long-term use, requiring periodic external calibration or sensor replacement. This leads to high maintenance costs, long downtime, and impacts production continuity and system reliability. This problem is particularly prominent in applications requiring high temperature measurement accuracy, such as air conditioning systems and industrial process control. Summary of the Invention

[0004] Based on this, the present invention provides a temperature sensor that can integrate multiple different types of sensing elements, and utilizes the complementary advantages of the characteristics of different sensing elements to achieve high-precision temperature measurement.

[0005] This invention provides a temperature sensor, comprising: The package includes a package body and multiple temperature measuring units with different temperature measuring principles disposed within the package body, and each of the temperature measuring units forms a thermal coupling with the object being measured through a thermally conductive structure. A processor, electrically connected to each of the temperature measuring units, is configured to determine an anomaly for each of the temperature measuring units based on the measurement difference between the temperature measurements of each of the temperature measuring units and / or based on a reference temperature value determined from the temperature measurements; and to determine a temperature output result based on the temperature measurement values ​​of the temperature measuring units determined to be non-abnormal. The processor is configured to perform anomaly detection on each temperature measuring unit based on a reference temperature value determined by the temperature measurement value, including: in response to acquiring the temperature measurement value of each temperature measuring unit, determining the temperature measurement sequence of each temperature measuring unit within a preset time window, as well as the spatial location information and thermal coupling parameters of each temperature measuring unit; constructing a reference temperature field reconstruction model based on the spatial location information of each temperature measuring unit, the thermal coupling parameters, and the temperature measurement values ​​of neighboring temperature measuring units; determining the reference temperature value corresponding to each temperature measuring unit based on the reference temperature field reconstruction model; and calculating the temperature residual index of each temperature measuring unit based on the difference between the temperature measurement value of each temperature measuring unit and the corresponding reference temperature value. A spatial association graph model is constructed based on the spatial adjacency relationship between each temperature measuring unit. A local residual consistency index is calculated for each temperature measuring unit based on the temperature residual distribution within its neighborhood. Temperature measuring units with temperature residual indices exceeding a preset threshold are identified as candidate temperature measuring units and temporarily removed from the reference temperature field reconstruction model. Temperature estimation is performed on the location of the candidate temperature measuring unit based on its neighboring temperature measuring units, and the change in reconstruction error of the reference temperature field before and after removal is calculated. When a candidate temperature measuring unit meets the conditions of a temperature residual index exceeding a preset threshold, a local residual consistency index below a preset threshold, and a reduction in reconstruction error after removal, the candidate temperature measuring unit is determined to be an abnormal temperature measuring unit.

[0006] Optionally, the processor is configured to determine anomalies for each of the temperature measuring units based on measurement differences between the temperature measurements of each of the temperature measuring units, including: Calculate the temperature difference between any two temperature measurement values ​​from the temperature measuring units; In response to the presence of a temperature difference between any two of the temperature measuring units exceeding a first preset threshold, and the most frequent occurrence of a temperature difference between one of the temperature measuring units and other temperature measuring units among the plurality of temperature measuring units exceeding the first preset threshold, the temperature measuring unit is determined to be an abnormal temperature measuring unit. If the temperature difference between any two of the temperature measuring units does not exceed the first preset threshold, it is determined that all temperature measuring units are normal.

[0007] Optionally, the step of constructing a reference temperature field reconstruction model based on the spatial location information of each temperature measuring unit, the thermal coupling parameters, and the temperature measurement values ​​of neighboring temperature measuring units includes: Based on the spatial location information of each temperature measuring unit, the initial spatial adjacency relationship and spatial distance are determined; Based on the spatial distance, the thermal coupling parameters, and the degree of synchronization of temperature changes within the preset time window, the adjacency weights between the corresponding temperature measurement units are determined. The initial spatial adjacency relationship is corrected based on each adjacency weight to obtain a thermally coupled adjacency relationship; Obtain the temperature measurement values ​​of the neighboring temperature measurement units corresponding to each temperature measurement unit, and construct a local temperature propagation constraint relationship based on the thermal coupling adjacency relationship; Based on the spatial location information, the thermal coupling adjacency relationship, and the local temperature propagation constraint relationship, spatial smoothing constraint terms, thermal propagation constraint terms, and local observation constraint terms are established respectively, and joint modeling is performed to obtain the initial reference temperature field model. Based on the initial reference temperature field model, an objective optimization function is constructed, and the reference temperature field reconstruction model is obtained by iteratively solving the objective optimization function.

[0008] Optionally, determining the reference temperature value corresponding to each temperature measuring unit based on the reference temperature field reconstruction model includes: The temperature measurement value of the neighboring temperature measurement unit corresponding to each temperature measurement unit at the current sampling time is input into the reference temperature field reconstruction model to obtain the initial reference temperature value. Extract the historical temperature measurement sequence of the temperature measuring unit within the preset time window, and determine the time-corrected temperature value based on the historical temperature measurement sequence; Based on the initial reference temperature value and the time-corrected temperature value, a candidate reference temperature value is determined; Calculate the temperature difference between the temperature measuring unit and each neighboring temperature measuring unit, and determine the direction of the local temperature gradient based on the temperature difference and the spatial orientation relationship of each neighboring temperature measuring unit. The gradient compensation coefficient is determined based on the direction of the local temperature gradient and the corresponding thermal coupling parameters. The candidate reference temperature value is corrected using the gradient compensation coefficient to obtain the reference temperature value corresponding to the temperature measuring unit.

[0009] Optionally, the step of constructing a spatial association graph model based on the spatial adjacency relationship between each temperature measuring unit, and calculating the local residual consistency index of each temperature measuring unit according to the temperature residual distribution in its neighborhood, includes: Each temperature measuring unit is constructed as a graph node, and the connection relationships between temperature measuring units that have spatial adjacency are constructed as graph edges; Based on the spatial distance between the temperature measurement units at both ends of the graph edge, the thermal coupling parameters, and the degree of synchronization of temperature changes, the edge weight of the corresponding graph edge is determined, and the spatial association graph model is constructed. Extract the temperature residual distribution corresponding to the target temperature measuring unit and its neighboring temperature measuring units; Calculate the residual difference between the target temperature measuring unit and each neighboring temperature measuring unit, and determine the local deviation of the target temperature measuring unit relative to each neighboring temperature measuring unit. A weighted aggregation is performed based on the local deviation values ​​and the corresponding graph edge weights to obtain a consistent result. Based on the degree of residual dispersion among the neighboring temperature measurement units within the target temperature measurement unit's neighborhood, the consistency result is compensated and corrected to obtain the local residual consistency index.

[0010] Optionally, the step of estimating the temperature at the location of the candidate temperature measuring unit based on its neighboring temperature measuring units, and calculating the change in reconstruction error of the reference temperature field before and after removal, includes: While shielding the temperature measurement values ​​corresponding to the candidate temperature measurement units from the reference temperature field reconstruction model, the spatial location information and thermal coupling parameters corresponding to the candidate temperature measurement units are retained. Based on the temperature measurement value corresponding to the first-order neighboring temperature measurement unit of the candidate temperature measurement unit, the first estimated temperature value is determined; The first estimated temperature value is corrected based on the temperature measurement value corresponding to the second-order neighboring temperature measurement unit of the candidate temperature measurement unit, and an alternative temperature estimate is obtained by combining the historical temperature evolution trend of the candidate temperature measurement unit within the preset time window. Reconstruct the removed reference temperature field based on the alternative temperature estimate; The amount of change in reconstruction error is determined based on the difference between the measured temperature value of each temperature measuring unit in the predetermined evaluation set and the reconstructed temperature value corresponding to the reference temperature field before and after removal.

[0011] Optionally, determining the temperature output result based on the temperature measurement values ​​of the temperature measuring units determined to be non-abnormal includes: In response to the assumption that all temperature measuring units are non-abnormal, the temperature contribution value of each temperature measuring unit is determined based on the temperature measurement value of each temperature measuring unit and a pre-set weighting coefficient; the temperature contribution values ​​of all temperature measuring units are accumulated to determine the temperature output result. In response to an abnormal temperature measuring unit, the temperature measurement value of that unit is removed, and the actual weighting coefficient of each non-abnormal temperature measuring unit is determined. The actual weighting coefficient of each non-abnormal temperature measuring unit and its temperature measurement value are used to determine the actual temperature contribution value of each non-abnormal temperature measuring unit. The actual temperature contribution values ​​of all non-abnormal temperature measuring units are accumulated to determine the temperature output result.

[0012] Optionally, the temperature sensor satisfies one or more of the following: The plurality of temperature measuring units include: a platinum resistance temperature measuring unit, a thermocouple temperature measuring unit, and a semiconductor temperature measuring unit; The platinum resistance temperature measurement unit uses a Pt100 or Pt1000 platinum resistance thermometer. The thermocouple temperature measuring unit uses type K, type E, or type T thermocouples; The semiconductor temperature measurement unit uses an NTC thermistor or a digital temperature sensor. Multiple temperature measuring units are disposed on the same temperature sensing end face or adjacent areas of the package body, wherein the same temperature sensing end face is the same physical contact surface or adjacent areas with a distance of less than 5mm; The package is a metal shell with an insert-type structure, including a probe part and a connecting cable. The probe part is cylindrical or needle-shaped with a diameter of 3mm to 15mm.

[0013] Optionally, an automatic calibration procedure is executed in response to a preset trigger condition being met, wherein the preset trigger condition includes at least one of the following: reaching a preset time point or the cumulative running time reaching a preset threshold. The automatic calibration procedure includes: Obtain the current measurement value of each temperature measuring unit and determine whether the measurement deviation between each temperature measuring unit is within a preset tolerance range; When the measurement deviation of a temperature measuring unit exceeds the preset tolerance range, a reference unit is selected from the multiple temperature measuring units; wherein the reference unit meets the following conditions: no abnormal records within a preset time window; the measurement deviation from the other temperature measuring units is less than a preset threshold; and it is currently within the effective operating temperature range of the corresponding sensing element. Using the measured value of the reference unit as a reference, deviation calculations are performed on the remaining temperature measuring units and compensation corrections are executed. The calibration time, reference unit information, and compensation parameters for this automatic calibration are recorded in the data storage module. The automatic calibration program dynamically prioritizes each temperature measurement unit based on the differences in calibration cycles of different types of sensing elements, and selects the reference unit based on the priority.

[0014] In summary, the package can simultaneously encapsulate multiple temperature sensing units based on different temperature measurement principles, enabling complementary advantages of different sensing elements. Furthermore, each temperature sensing unit forms thermal coupling with the object being measured through a thermally conductive structure. This allows the processor to determine anomalies for each temperature sensing unit based on measurement differences between the individual units and / or a reference temperature value determined from the measured values. In this way, the processor can determine the temperature output based on the temperature measurements from the non-abnormal temperature sensing units, achieving high-precision temperature measurement. Moreover, this process eliminates the need to replace the temperature sensing units, reducing maintenance costs and downtime. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the structure of a temperature sensor provided in an embodiment of the present invention; Figure 2 A flowchart of the first anomaly determination method provided in the embodiments of the present invention; Figure 3 A flowchart of the second anomaly determination method provided in this embodiment of the invention; Figure 4 This is a schematic diagram illustrating the dynamic adjustment of weighting coefficients according to an embodiment of the present invention. Detailed Implementation

[0016] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the specific details described below are only a part of the embodiments of the present invention, and the present invention can be implemented in many other embodiments different from those described herein. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0017] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly attached to the other element or there may be an intervening element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0019] As described in the background section, single-type temperature sensors suffer from accuracy drift after long-term use, specifically: Platinum resistance thermometers (PTS) offer high measurement accuracy and good long-term stability, but they suffer from slow response, large size, and a tendency for resistance drift that leads to accuracy degradation during prolonged use at high temperatures (above 500°C). Furthermore, PTS are sensitive to lead and contact resistance, requiring three- or four-wire wiring to eliminate errors, which increases system complexity.

[0020] Thermocouples offer advantages such as a wide measurement range, fast response speed, and high temperature resistance. However, their measurement accuracy is lower in the low-temperature region (below 200℃), and they suffer from cold junction compensation issues. The thermocouple's thermoelectric potential-temperature relationship is non-linear, requiring complex linearization processing. Furthermore, different types of thermocouples exhibit significant differences in long-term stability, and some types may experience thermoelectric potential drift during use.

[0021] Semiconductor temperature sensors are small, inexpensive, and fast-responding, but their accuracy is relatively low, their temperature range is limited, and they suffer from significant aging drift over long-term use, requiring periodic calibration.

[0022] Based on this, the present invention provides a temperature sensor that can simultaneously encapsulate multiple temperature sensing units with different temperature measurement principles within a package, achieving complementary advantages of the characteristics of different sensing elements. Furthermore, each temperature sensing unit forms thermal coupling with the object being measured through a thermally conductive structure, enabling the processor to determine anomalies for each temperature sensing unit based on the measurement differences between the temperature measurements of each unit and / or based on a reference temperature value determined from the temperature measurements. In this way, the processor can determine the temperature output result based on the temperature measurement values ​​of the temperature sensing units determined to be non-abnormal, achieving high-precision temperature measurement. Moreover, this process eliminates the need to replace the temperature sensing units, reducing maintenance costs and downtime.

[0023] To enable those skilled in the art to better understand and implement this solution, the following detailed description of the specific solution, principles, advantages, and effects of the present invention is provided with reference to the accompanying drawings and specific embodiments.

[0024] See Figure 1 , Figure 1 A schematic diagram of the structure of a temperature sensor provided in an embodiment of the present invention is shown below. Figure 1 As shown, the temperature sensor may include: The package 100, and a plurality of temperature measuring units with different temperature measuring principles disposed within the package 100 (e.g., Figure 1 The diagram shows a first temperature measuring unit 110, a second temperature measuring unit 120, and a third temperature measuring unit 130, and each of the temperature measuring units forms a thermal coupling with the object being measured through a heat-conducting structure (not shown in the figure). A processor 140 is electrically connected to each of the temperature measuring units (e.g., the processor 140 is coupled to a first temperature measuring unit 110, a second temperature measuring unit 120, and a third temperature measuring unit 130, respectively). The processor 140 is configured to determine an anomaly for each of the temperature measuring units based on the measurement difference between the temperature measurements of each of the temperature measuring units and / or based on a reference temperature value determined from the temperature measurements; and to determine a temperature output result based on the temperature measurements of the temperature measuring units determined to be non-abnormal.

[0025] Specifically, when the temperature sensor is working, the first temperature measuring unit 110, the second temperature measuring unit 120, and the third temperature measuring unit 130 can detect the temperature of the same object being measured, thereby enabling the processor 140 to acquire the temperature measurement values ​​of the first temperature measuring unit 110, the second temperature measuring unit 120, and the third temperature measuring unit 130. By processing the temperature measurement values, the processor 140 can determine whether there are any abnormal temperature measuring units.

[0026] In this way, the processor 140 can determine the temperature output result based on the temperature measurement value of the non-abnormal temperature sensing unit.

[0027] In some embodiments, the package is a metal housing and has an insert-type structure, including a probe portion and a connecting cable. The probe portion is cylindrical or needle-shaped and has a diameter of 3 mm to 15 mm.

[0028] In some embodiments, multiple temperature sensing units are disposed on the same temperature sensing end face 102 of the package 100 or in adjacent areas, wherein the same temperature sensing end face 102 is the same physical contact surface or adjacent areas with a distance of less than 5 mm.

[0029] In some embodiments, the plurality of temperature measuring units include: a platinum resistance temperature measuring unit, a thermocouple temperature measuring unit, and a semiconductor temperature measuring unit.

[0030] The platinum resistance temperature measuring unit uses Pt100 or Pt1000 platinum resistance thermometers; the thermocouple temperature measuring unit uses K-type, E-type or T-type thermocouples; and the semiconductor temperature measuring unit uses NTC thermistors or digital temperature sensors.

[0031] See Figure 2 , Figure 2 The flowchart of the first anomaly determination method provided in the embodiments of the present invention includes steps S201 to S203: S201, calculate the temperature difference between the temperature measurements of any two of the temperature measuring units.

[0032] S202, in response to the existence of a temperature difference between any two of the temperature measuring units exceeding a first preset threshold, and the number of times that the temperature difference between one of the temperature measuring units and the other temperature measuring units among the plurality of temperature measuring units exceeds the first preset threshold is the most, the one of the temperature measuring units is determined to be an abnormal temperature measuring unit.

[0033] S202, in response to the fact that the temperature difference between any two of the temperature measuring units does not exceed the first preset threshold, it is determined that all temperature measuring units are normal.

[0034] For example, suppose there are multiple temperature measuring units, including a first temperature measuring unit, a second temperature measuring unit, and a third temperature measuring unit, and the temperature measurement value of the first temperature measuring unit is T1.i The temperature measurement value of the second temperature measuring unit is T2. i The temperature measurement value of the third temperature measuring unit is T3. i .

[0035] Calculate the difference between the temperature measurements of any two temperature measuring units in the first, second, and third temperature measuring units respectively, and obtain: T1 i -T2 i T1 i -T3 i T3 i -T2 i .

[0036] In one scenario, by comparing the above differences, it is determined that the first temperature measuring unit is abnormal; or, the first, second, and third temperature measuring units are all normal.

[0037] It should be noted that when an abnormal temperature measuring unit is identified, multiple additional abnormality judgment logics can be executed. If the judgment structure is abnormal multiple times (e.g., 3 times), the temperature measuring unit is determined to be an abnormal temperature measuring unit.

[0038] See Figure 3 , Figure 3 The flowchart of the second anomaly determination method provided in the embodiments of the present invention includes steps S301 to S307: S301, in response to obtaining the temperature measurement value of each of the temperature measuring units, determine the temperature measurement sequence of each of the temperature measuring units within a preset time window, as well as the spatial location information and thermal coupling parameters of each of the temperature measuring units.

[0039] In some embodiments, each temperature measuring unit can periodically output temperature measurement values. After continuously acquiring these temperature measurement values, the processor can extract the temperature measurement values ​​according to a preset time window (as a non-limiting example, it can be 5 seconds, 10 seconds or 1 minute) to form a corresponding temperature measurement sequence.

[0040] Simultaneously, each temperature measuring unit possesses spatial location information, represented as two-dimensional or three-dimensional coordinates of the temperature measuring unit within the target system. For example, during the packaging of the temperature measuring units, the installation position of each unit can be manually calibrated and recorded to establish a mapping relationship between the temperature measuring units and spatial coordinates. Alternatively, the position of the temperature measuring units can be automatically acquired using positioning devices (such as visual positioning systems, laser rangefinders, or radio frequency-based positioning modules).

[0041] Thermal coupling parameters are used to characterize the degree of thermal influence between different temperature measuring units. In one embodiment, the thermal coupling parameters can be determined by experimental calibration. Specifically, under controlled environmental conditions, a known thermal excitation is applied to a certain temperature measuring unit, and its influence on the temperature changes of adjacent temperature measuring units is recorded, thereby calculating the thermal coupling coefficient between each temperature measuring unit.

[0042] It should be noted that the thermal coupling parameters can be represented in matrix form, where the matrix elements are used to represent the thermal influence weight between any two temperature measuring units; in some embodiments, only the thermal coupling relationship between temperature measuring units within a preset neighborhood range can be considered to reduce computational complexity.

[0043] In summary, by reasonably acquiring and modeling the spatial location information and thermal coupling parameters of the temperature measurement unit, accurate basic data can be provided for subsequent temperature field reconstruction, thereby improving the reliability and accuracy of anomaly detection.

[0044] S302, Based on the spatial location information of each temperature measuring unit, the thermal coupling parameters, and the temperature measurement values ​​of neighboring temperature measuring units, a reference temperature field reconstruction model is constructed.

[0045] In some embodiments, all temperature measurement units can be regarded as discrete sampling points, the temperature field can be regarded as a continuous physical field, and a reference temperature field reconstruction model can be constructed based on spatial interpolation methods or physical constraint-based models.

[0046] In this reference temperature field reconstruction model, spatial location information is used to characterize the geometric relationship between temperature measuring units, thermal coupling parameters are used to reflect the degree of thermal influence between each temperature measuring unit, and the temperature measurement values ​​of neighboring temperature measuring units are used as model inputs, thereby enabling the estimation of temperature at unobserved locations or under ideal conditions.

[0047] In one embodiment, step S302 may include: A1) Determine the initial spatial adjacency relationship and spatial distance based on the spatial location information of each temperature measuring unit.

[0048] Specifically, the spatial location information of each temperature measuring unit can be two-dimensional planar coordinates or three-dimensional spatial coordinates.

[0049] After obtaining the spatial location information, the spatial distance between any two temperature measuring units can be calculated. This spatial distance can be calculated as a straight line distance, or, depending on the actual application scenario, a distance calculation method considering structural obstacles or path limitations can be used.

[0050] Based on this, initial spatial adjacency relationships can be established according to preset adjacency rules. As a non-limiting example, when the distance between two temperature measuring units is less than a preset threshold, they are determined to be adjacent; for each temperature measuring unit, the nearest several temperature measuring units are selected as its adjacent units; or the adjacency relationship is determined based on a spatial structure partitioning method.

[0051] The above method can be used to obtain the initial spatial adjacency relationship that reflects the spatial proximity of the temperature measuring units.

[0052] A2) Based on the spatial distance, the thermal coupling parameters, and the degree of synchronization of temperature changes within the preset time window, determine the adjacency weights between the corresponding temperature measurement units.

[0053] Specifically, after determining the initial spatial adjacency relationship, for any pair of temperature measuring units that have an adjacency relationship, their adjacency weight can be further calculated to characterize the thermal correlation strength between the two temperature measuring units.

[0054] The adjacency weight can be determined by considering the following factors: Spatial distance is a factor; generally speaking, the closer the temperature measuring units are, the stronger the heat conduction between them, and the corresponding weight should be relatively large.

[0055] Thermal coupling parameters, which are used to characterize the thermal conductivity between different locations.

[0056] The degree of synchronization in temperature changes refers to the consistency of the temperature change trends between two temperature measuring units within a preset time window. As a non-limiting example, the degree of synchronization can be determined by analyzing the correlation or consistency of the change trends between temperature sequences.

[0057] After considering the above factors, the adjacency weights between each temperature measurement unit can be obtained and normalized to make the weight distribution of each temperature measurement unit comparable, thereby reflecting the non-uniformity of actual thermal coupling.

[0058] A3) Based on each of the adjacency weights, the initial spatial adjacency relationship is modified to obtain a thermally coupled adjacency relationship.

[0059] Specifically, after obtaining the adjacency weights, the initial spatial adjacency relationships can be modified to obtain an adjacency structure that better reflects the actual heat conduction characteristics.

[0060] As a non-restrictive example, the following modifications can be made: delete adjacency relationships with weights below a preset threshold to remove weak or noisy connections; convert the original binary adjacency relationships into weighted adjacency relationships so that different adjacent edges have different levels of importance; retain only a few adjacent units with higher weights for each temperature measurement unit to reduce computational complexity.

[0061] Through the above processing, thermal coupling adjacency relationships can be obtained, which not only reflect spatial adjacency, but also reflect the heat conduction path and intensity.

[0062] A4) Obtain the temperature measurement value of the neighboring temperature measurement unit corresponding to each of the temperature measurement units, and construct a local temperature propagation constraint relationship in combination with the thermal coupling adjacency relationship.

[0063] Specifically, for any temperature measuring unit, its set of neighboring temperature measuring units can be determined based on thermal coupling adjacency, and the temperature measurement values ​​of these neighboring temperature measuring units at the current time or within a preset time window can be obtained.

[0064] Based on this, a local temperature propagation constraint relationship can be constructed to describe the temperature change pattern between adjacent temperature measurement units.

[0065] As a non-limiting example, it can be assumed that there is a certain balance between the temperature of a certain temperature measuring unit and the temperature of its neighboring temperature measuring units, that is, the temperature of the temperature measuring unit is affected by the combined temperature of its neighboring units.

[0066] By introducing adjacency weights, the differences in the degree of influence of different neighboring temperature measurement units on the current temperature measurement unit can be further reflected, thereby establishing a constraint relationship that is more in line with the actual heat conduction law, so as to effectively describe the diffusion and transfer characteristics of temperature in space.

[0067] A5) Based on the spatial location information, the thermal coupling adjacency relationship, and the local temperature propagation constraint relationship, establish spatial smoothing constraint terms, thermal propagation constraint terms, and local observation constraint terms respectively, and perform joint modeling to obtain the initial reference temperature field model.

[0068] Specifically, after completing the above modeling foundation, multiple constraint terms can be further constructed and modeled uniformly, including: A spatial smoothing constraint term is established to ensure the continuity of the temperature field in space, that is, the temperature change between adjacent temperature measuring units should be relatively gradual; a heat propagation constraint term is established to reflect the temperature conduction law based on thermal coupling adjacency relationship, so that the temperature distribution conforms to the actual heat diffusion process; a local observation constraint term is established to ensure that the temperature value calculated by the model is consistent with or as close as possible to the measured value of the actual temperature measuring unit.

[0069] Based on this, the above-mentioned multiple constraints can be weighted and combined to form a unified temperature field modeling framework, thereby obtaining the initial reference temperature field model.

[0070] A6) Construct an objective optimization function based on the initial reference temperature field model, and obtain the reference temperature field reconstruction model by iteratively solving the objective optimization function.

[0071] Specifically, the above joint modeling results are transformed into an optimization problem, and the optimal temperature field distribution is obtained by minimizing the overall error.

[0072] During the solution process, the temperature field estimate can be updated iteratively. As a non-limiting example, gradient-based optimization methods or matrix-based iterative methods can be used.

[0073] In each iteration, the temperature is adjusted based on the current temperature field estimation results and various constraints to gradually meet the requirements of spatial smoothness, heat propagation consistency, and observation consistency.

[0074] When the preset convergence condition is met (e.g., the temperature change is below a threshold or the maximum number of iterations is reached), the iteration stops and the final result is output.

[0075] The final result is the reference temperature field reconstruction model, which can reflect the temperature distribution continuously in space and reasonably in physical terms.

[0076] Using the above embodiments, the adjacency weights between temperature measuring units are determined by fusing spatial distance, thermal coupling parameters, and the degree of synchronization of temperature changes. Based on this, the initial spatial adjacency relationship is corrected to obtain a thermally coupled adjacency relationship that better reflects actual heat conduction characteristics, thereby improving the physical accuracy of adjacency modeling. By introducing temperature measurements from neighboring temperature measuring units and constructing local temperature propagation constraints, the model can effectively characterize the propagation law of temperature within a local range, enhancing its ability to express dynamic temperature changes. Simultaneously, by constructing spatial smoothing constraints, thermal propagation constraints, and local observation constraints and performing joint modeling, the model balances spatial continuity, consistency between the heat conduction mechanism and actual observation data, and improves the overall stability and robustness of the modeling. Finally, by iteratively solving the objective optimization function, a converged reference temperature field reconstruction model is obtained, significantly improving the accuracy and reliability of temperature field reconstruction.

[0077] S303, Based on the reference temperature field reconstruction model, determine the reference temperature value corresponding to each temperature measuring unit.

[0078] In some embodiments, after constructing the reference temperature field reconstruction model, the reference temperature value of each temperature measuring unit within the current time window can be obtained by solving the model or performing interpolation calculations. This reference temperature value can be understood as the temperature level that the temperature measuring unit should exhibit under normal operating conditions or without abnormal interference. It should be noted that the reference temperature value can be a single-moment value, or an average value or a smoothed value within a time window; this embodiment of the invention does not impose any limitations on this.

[0079] In one embodiment, step S303 may include: B1) Input the temperature measurement value of the neighboring temperature measurement unit corresponding to each temperature measurement unit at the current sampling time into the reference temperature field reconstruction model to obtain the initial reference temperature value.

[0080] Specifically, at the current sampling moment, each temperature measuring unit can acquire not only its own temperature measurement value, but also the temperature measurement values ​​of its neighboring temperature measuring units, which can be predetermined based on their spatial location. Based on the temperature measurement values ​​of these neighboring temperature measuring units, these values ​​are input into the reference temperature field reconstruction model for calculation, thereby outputting the initial reference temperature value corresponding to that temperature measuring unit at the current moment.

[0081] B2) Extract the historical temperature measurement sequence of the temperature measuring unit within the preset time window, and determine the time-corrected temperature value based on the historical temperature measurement sequence.

[0082] Specifically, each temperature measurement unit has continuous historical temperature measurement data within a preset time window. By extracting the historical temperature measurement sequence within this time window, the trend of temperature change over time can be reflected.

[0083] Based on this, time series analysis methods (such as moving average, exponential weighting, or trend fitting) can be used to process historical temperature measurement sequences to obtain time-corrected temperature values ​​that characterize time-varying properties. It should be noted that different time correction methods may affect the calculation results of the time-corrected temperature values, but all are within the scope of this embodiment.

[0084] B3) Based on the initial reference temperature value and the time-corrected temperature value, determine the candidate reference temperature value.

[0085] Specifically, the initial reference temperature value obtained in step B1 and the time-corrected temperature value obtained in step B2 are fused together using methods such as weighted summation, nonlinear mapping, or rule combination to obtain a candidate reference temperature value. This candidate reference temperature value takes into account both spatial distribution information and temporal variation characteristics to a certain extent, and has higher accuracy than temperature estimation from a single source.

[0086] B4) Calculate the temperature difference between the temperature measuring unit and each neighboring temperature measuring unit, and determine the direction of the local temperature gradient based on the temperature difference and the spatial orientation relationship of each neighboring temperature measuring unit.

[0087] Specifically, by calculating the temperature difference between the temperature measuring unit and each neighboring temperature measuring unit, the temperature distribution differences around the temperature measuring unit can be obtained.

[0088] By further combining the spatial orientation relationship (such as azimuth angle, relative position coordinates, etc.) of each neighboring temperature measuring unit relative to the temperature measuring unit, the directional characteristics of local temperature changes can be determined, thereby obtaining the direction of local temperature gradient.

[0089] As a non-restricted example, when the temperature of a neighboring temperature measuring unit in a certain direction is significantly higher than that in other directions, that direction can be determined as the direction of increasing temperature gradient.

[0090] B5) Determine the gradient compensation coefficient based on the direction of the local temperature gradient and the corresponding thermal coupling parameters.

[0091] Specifically, after obtaining the direction of the local temperature gradient, the influence of the temperature gradient can be quantified by combining the corresponding thermal coupling parameters (such as thermal conductivity or coupling strength coefficient in different directions), thereby calculating the gradient compensation coefficient. This gradient compensation coefficient is used to characterize the influence of local temperature distribution non-uniformity on the temperature of the target temperature measuring unit.

[0092] B6) The candidate reference temperature value is corrected using the gradient compensation coefficient to obtain the reference temperature value corresponding to the temperature measuring unit.

[0093] Specifically, the candidate reference temperature value obtained in step B3 is corrected using the gradient compensation coefficient obtained in step B5, for example, through multiplicative or additive compensation, to obtain the final reference temperature value. This reference temperature value is optimized under the combined effect of multiple factors such as spatial distribution, temporal variation, and the influence of local gradients, and can more accurately reflect the true temperature state of the temperature measuring unit.

[0094] In the above scheme, the temperature data of the neighboring temperature measurement units are processed by a reference temperature field reconstruction model, so that the initial reference temperature value can reflect the overall spatial temperature distribution characteristics, thereby improving spatial consistency. At the same time, time correction is performed by combining the historical temperature measurement sequence within a preset time window, effectively introducing temperature change trend information, reducing the impact of instantaneous noise on the measurement results, and enhancing the smoothness and anti-interference ability of the results. On this basis, candidate reference temperature values ​​are obtained by fusing spatial and temporal information, making the temperature estimation more comprehensive and reliable. Furthermore, the local temperature gradient direction is determined by calculating the temperature difference between the temperature measurement unit and the neighboring temperature measurement units and combining the spatial orientation relationship. Then, a gradient compensation coefficient is constructed based on thermal coupling parameters to correct the candidate reference temperature value, thereby effectively eliminating the deviation caused by the uneven distribution of local temperature, making the final reference temperature value closer to the true temperature.

[0095] S304, calculate the temperature residual index of each temperature measuring unit based on the difference between the temperature measurement value of each temperature measuring unit and the corresponding reference temperature value.

[0096] In one embodiment, step S304 may include: determining the instantaneous residual value of each temperature measuring unit at the current sampling moment based on the difference between the temperature measurement value of each temperature measuring unit and the corresponding reference temperature value; determining the residual mean value representing the overall deviation level, the residual fluctuation degree representing the residual dispersion, and the residual change trend representing the dynamic change characteristics of the residual based on the difference sequence corresponding to multiple consecutive sampling moments of the temperature measuring unit within a preset time window; constructing a residual feature set for the corresponding temperature measuring unit based on the instantaneous residual value, the residual mean value, the residual fluctuation degree, and the residual change trend; and processing the instantaneous residual value... The difference, the mean residual, the degree of residual fluctuation, and the trend of residual change are normalized to map residual features of different dimensions to a unified numerical range. Based on the stability of each residual feature under historical normal operating conditions, the basic weights corresponding to each residual feature are determined, and based on the contribution of each residual feature to the discrimination of abnormal temperature measurement units in the current sampling period, the dynamic weights corresponding to each residual feature are determined. Based on the basic weights and the dynamic weights, each residual feature is jointly weighted to obtain the weighted residual characterization value of each temperature measurement unit. The weighted residual characterization value is determined as the temperature residual index of each temperature measurement unit.

[0097] S305, construct a spatial association graph model based on the spatial adjacency relationship between each temperature measuring unit, and calculate the local residual consistency index of each temperature measuring unit according to the temperature residual distribution in its neighborhood.

[0098] In some embodiments, each temperature measuring unit can be abstracted as a node in a graph model, and edge connections can be established between spatially adjacent or thermally coupled temperature measuring units to form a spatial association graph.

[0099] In this spatial correlation graph model, for any temperature measurement unit, the temperature residual index of other temperature measurement units in its neighborhood can be obtained, and the local residual consistency index can be obtained through statistical analysis (such as calculating the mean, variance, or similarity index). When the residual distribution of a certain temperature measurement unit differs greatly from that of its neighborhood residual distribution, its local consistency index is low, and vice versa.

[0100] In one embodiment, step S305 may include: C1), each of the temperature measuring units is constructed as a graph node, and the connection relationship between temperature measuring units that have spatial adjacency is constructed as a graph edge.

[0101] Specifically, each temperature measuring unit has a clear spatial distribution location, and each temperature measuring unit can be abstracted as a node in a graph structure. At the same time, for temperature measuring units that are adjacent in space, a connection relationship is established based on their relative positional relationship, thereby constructing graph edges to form a basic graph structure expression.

[0102] It should be noted that the spatial adjacency relationship can be determined based on a distance threshold, topological structure, or preset neighborhood rules, and this embodiment does not impose any restrictions on this.

[0103] C2) Based on the spatial distance between the temperature measurement units at both ends of the graph edge, the thermal coupling parameters, and the degree of synchronization of temperature changes, the edge weight of the corresponding graph edge is determined, and the spatial association graph model is constructed.

[0104] Specifically, based on the graph structure, the spatial distance between the temperature measurement units at both ends of the graph edge, the thermal coupling parameters, and the degree of synchronization of temperature changes are further considered to assign corresponding weight values ​​to each graph edge.

[0105] Among them, spatial distance is used to reflect the influence of physical location, thermal coupling parameters are used to characterize the strength of thermal conduction correlation, and the degree of synchronization of temperature changes is used to characterize dynamic consistency features.

[0106] By integrating the above multiple factors, a spatial correlation diagram model that better reflects actual heat conduction characteristics can be constructed.

[0107] C3) Extract the temperature residual distribution corresponding to the target temperature measuring unit and its neighboring temperature measuring units.

[0108] Specifically, for the target temperature measuring unit and its neighboring temperature measuring units, their respective temperature residual data are obtained. The temperature residual can be expressed as the difference between the measured temperature and the reference temperature or the prediction error, thereby forming a corresponding residual distribution to reflect the deviation of each temperature measuring unit.

[0109] C4), calculate the residual difference between the target temperature measuring unit and each neighboring temperature measuring unit respectively, and determine the local deviation of the target temperature measuring unit relative to each neighboring temperature measuring unit.

[0110] Specifically, by calculating the residual differences between the target temperature measuring unit and each neighboring temperature measuring unit, the degree of deviation of the target temperature measuring unit relative to its neighbors can be quantified. Based on these residual differences, the local deviation values ​​of the target temperature measuring unit relative to each neighboring temperature measuring unit can be further determined, which can be used to characterize its anomalies or consistency within a local area.

[0111] C5), based on the local deviation value and the corresponding graph edge weight, performs weighted aggregation to obtain a consistent result.

[0112] Specifically, the local deviation value obtained in step C4 is fused with the graph edge weights determined in step C2 to perform weighted aggregation of the influence of each neighboring temperature measurement unit, thereby obtaining the preliminary consistency result of the target temperature measurement unit; among them, the neighboring temperature measurement unit with a larger edge weight contributes more to the consistency result, making the result more consistent with the actual physical relationship.

[0113] C6), based on the degree of residual dispersion among the neighboring temperature measurement units in the neighborhood of the target temperature measurement unit, the consistency result is compensated and corrected to obtain the local residual consistency index.

[0114] Specifically, after obtaining the consistency results, the residual dispersion among neighboring temperature measuring units within the target temperature measuring unit's neighborhood is further analyzed, for example, by using variance or coefficient of variation. When the residual distribution within the neighborhood is relatively dispersed, the consistency results are compensated and corrected to reduce the impact of outliers, thereby obtaining the final local residual consistency index. This index can more accurately reflect the temperature consistency level of the target temperature measuring unit within a local area.

[0115] This embodiment introduces a graph-based spatial correlation modeling method to uniformly model the spatial relationships, thermal coupling characteristics, and temperature change synchronization among temperature measurement units. Compared to traditional methods that independently analyze individual temperature measurement points, this approach can more comprehensively reflect the spatial correlation characteristics of the temperature field. Simultaneously, by analyzing temperature residuals and constructing local deviation values, it can effectively identify abnormal deviations of the target temperature measurement unit relative to its neighborhood, improving the sensitivity of anomaly detection. Furthermore, weighted aggregation using graph edge weights makes the consistency results more consistent with actual physical conduction laws, enhancing the reliability of the results. In addition, by introducing the dispersion of neighborhood residuals to compensate for and correct the consistency results, the impact of local anomalies or noisy data on the results is effectively reduced, improving the robustness and stability of the indicators. Therefore, this embodiment can achieve high-precision evaluation of the local consistency of temperature measurement units under complex spatial distribution and dynamic operating conditions.

[0116] S306, the temperature measuring unit whose temperature residual index exceeds the preset threshold is identified as a candidate temperature measuring unit, and the candidate temperature measuring unit is temporarily removed from the reference temperature field reconstruction model. The temperature of the candidate temperature measuring unit is estimated based on its neighboring temperature measuring units, and the change in the reconstruction error of the reference temperature field before and after removal is calculated.

[0117] In some embodiments, a temperature residual threshold can be set first, and temperature measuring units with residuals exceeding this threshold can be designated as candidate anomalous units. Subsequently, during the temperature field reconstruction process, the data of the candidate temperature measuring unit is temporarily removed, and the temperature value at its location is re-estimated based solely on the data of its neighboring temperature measuring units. The overall temperature field reconstruction error (e.g., mean square error) is then recalculated. By comparing the change in error before and after removal, the degree of influence of the temperature measuring unit on the temperature field reconstruction can be determined.

[0118] In one embodiment, step S306 may include: D1) While shielding the temperature measurement value corresponding to the candidate temperature measurement unit from the reference temperature field reconstruction model, the spatial location information and thermal coupling parameters corresponding to the candidate temperature measurement unit are retained.

[0119] Specifically, once a candidate temperature measurement unit is determined, the temperature measurement value corresponding to the candidate temperature measurement unit is shielded from the reference temperature field reconstruction model, thereby obtaining the reconstruction model in the candidate removal state.

[0120] It should be noted that while masking the temperature measurement values, the spatial location information, thermal coupling parameters, and adjacency topology of the candidate temperature measurement unit are still retained, so that the temperature measurement unit still participates in the calculation as a structural node in the model, in order to maintain its positional constraints and thermal correlation constraints in the reference temperature field and avoid destroying the overall structural continuity of the temperature field due to the complete removal of nodes.

[0121] D2), based on the temperature measurement value corresponding to the first-order neighboring temperature measurement unit of the candidate temperature measurement unit, determine the first estimated temperature value.

[0122] Specifically, after the temperature measurement value of the candidate temperature measurement unit is masked, the temperature of the candidate temperature measurement unit is estimated based on the temperature measurement value of the first-order neighboring temperature measurement unit of the candidate temperature measurement unit, thereby obtaining the first estimated temperature value.

[0123] Furthermore, by analyzing the dispersion of the temperature distribution of the first-order neighborhood temperature measurement unit (e.g., using indicators such as variance, range, or standard deviation to characterize it), the reliability of the first estimated temperature value is marked. When the neighborhood temperature distribution is relatively concentrated, it indicates that the estimation result has high reliability; otherwise, it indicates that the estimation result has large uncertainty.

[0124] D3), ​​based on the temperature measurement value corresponding to the second-order neighboring temperature measurement unit of the candidate temperature measurement unit, corrects the first estimated temperature value, and combines the historical temperature evolution trend of the candidate temperature measurement unit within the preset time window to obtain an alternative temperature estimate.

[0125] Specifically, based on the first estimated temperature value, the temperature measurement value corresponding to the second-order neighboring temperature measurement unit of the candidate temperature measurement unit is introduced to further correct the first estimated temperature value, resulting in a second estimated temperature value. Subsequently, the second estimated temperature value is corrected for time consistency by combining the historical temperature evolution trend of the candidate temperature measurement unit within a preset time window (such as the slope of temperature change or the time series fitting result), thereby obtaining a more stable alternative temperature estimate that conforms to the actual change law.

[0126] D4), based on the alternative temperature estimate, reconstructs the removed reference temperature field.

[0127] Specifically, based on the reconstruction model under the candidate removal state and the alternative temperature estimate, the reference temperature field is recalculated to obtain the reference temperature field after removal. At the same time, the local temperature distribution characteristics (such as temperature gradient, distribution uniformity, etc.) of the reference temperature field before removal and the reference temperature field after removal in the neighborhood of the candidate temperature measurement unit can be extracted to reflect the impact of the removal operation on the local temperature field structure.

[0128] D5), based on the difference between the measured temperature value of each temperature measuring unit in the predetermined evaluation set and the reconstructed temperature value corresponding to the reference temperature field before and after removal, the amount of change in reconstruction error is determined.

[0129] Specifically, the predetermined evaluation set includes at least first-order neighboring temperature measurement units, second-order neighboring temperature measurement units, and verification temperature measurement units located in the same thermal zone as the candidate temperature measurement units.

[0130] A first reconstruction error is determined based on the difference between the measured temperature value of each temperature measuring unit in the predetermined evaluation set and the reconstructed temperature value corresponding to the reference temperature field before removal. A second reconstruction error is determined based on the difference between the measured temperature value of each temperature measuring unit in the predetermined evaluation set and the reconstructed temperature value corresponding to the reference temperature field after removal.

[0131] Finally, by combining the difference, ratio, direction of change, and change in local temperature distribution characteristics between the first and second reconstruction errors, the reconstruction effect is comprehensively evaluated to determine the amount of change in reconstruction error, which is used to characterize the degree of influence of the candidate temperature measurement unit on the reconstruction result of the reference temperature field.

[0132] By introducing a mechanism that combines candidate temperature measurement unit removal with alternative estimation, we can effectively identify and assess the impact of abnormal or unreliable temperature measurement units. Compared with the traditional method of directly using all measurement data, this method can significantly improve the accuracy and robustness of the reference temperature field reconstruction and accurately assess the degree of influence of candidate temperature measurement units on the temperature field reconstruction results.

[0133] S307, when the candidate temperature measuring unit meets the conditions that the temperature residual index exceeds the preset threshold, the local residual consistency index is lower than the preset threshold, and the reconstruction error is reduced after removal, the candidate temperature measuring unit is determined to be an abnormal temperature measuring unit.

[0134] In some embodiments, setting multiple judgment conditions can improve the accuracy of anomaly detection. Wherein: First, the temperature residual index is used to determine whether there is a significant deviation in the temperature measuring unit. Second, the local residual consistency index is used to determine whether the deviation is consistent with the neighborhood. Third, the reconstruction error change is used to verify whether the temperature measuring unit has a negative impact on the overall temperature field. When the above conditions are met simultaneously, the temperature measuring unit can be considered to be abnormal (e.g., sensor failure, poor contact, or local interference), thus being identified as an abnormal temperature measuring unit, and further processing operations such as alarm, rejection, or correction can be performed.

[0135] By employing the above scheme, a reference temperature field reconstruction model is constructed by acquiring the temperature measurement sequence, spatial location information, and thermal coupling parameters of the temperature measuring units. A temperature residual index is then generated based on the difference between the reference temperature and the actual measured value. Furthermore, local residual consistency constraints are introduced in conjunction with spatial correlations, and the model is validated by verifying the changes in reconstruction error after removing candidate temperature measuring units. This enables a multi-dimensional comprehensive judgment of anomalies in temperature measuring units. Therefore, this approach not only fully considers the spatial coupling effects between temperature measuring units, improving the accuracy of the reference temperature, but also effectively reduces false alarms and false negatives, enhancing the accuracy and robustness of anomaly detection, thereby strengthening the overall reliability and data validity of the temperature monitoring system.

[0136] Accordingly, based on the temperature measurement values ​​of the temperature measuring units determined to be non-abnormal, the temperature output results are determined, including: In response to the assumption that all temperature measuring units are non-abnormal, the temperature contribution value of each temperature measuring unit is determined based on the temperature measurement value of each temperature measuring unit and a pre-set weighting coefficient; the temperature contribution values ​​of all temperature measuring units are accumulated to determine the temperature output result.

[0137] In some embodiments, the weighting of each temperature measuring unit can be adjusted dynamically based on the temperature measurement range and the type of temperature measuring unit.

[0138] See Figure 4 , Figure 4 This is a schematic diagram illustrating the dynamic adjustment of weighting coefficients according to an embodiment of the present invention.

[0139] like Figure 4As shown, the temperature sensor measures temperatures from -50℃ to 800℃. Higher weight is assigned to the platinum resistance temperature measurement unit in the low-temperature range, to the thermocouple temperature measurement unit in the high-temperature range, and to the thermocouple temperature measurement unit in the rapid temperature change range.

[0140] For example, in the range of -50℃ to 200℃, the weighting coefficient of the first temperature measuring unit is 0.5, the weighting coefficient of the second temperature measuring unit is 0.3, and the weighting coefficient of the first temperature measuring unit is 0.2; in the range of 200℃ to 500℃, the weighting coefficient of the first temperature measuring unit is 0.4, the weighting coefficient of the second temperature measuring unit is 0.4, and the weighting coefficient of the first temperature measuring unit is 0.2; in the range of 500℃ to 800℃, the weighting coefficient of the first temperature measuring unit is 0.2, the weighting coefficient of the second temperature measuring unit is 0.6, and the weighting coefficient of the first temperature measuring unit is 0.2.

[0141] It should be noted that, Figure 4 The relationship between temperature and weighting coefficient shown is merely an illustrative example to illustrate how the weighting coefficient changes with temperature, and should not be construed as a limitation of the present invention.

[0142] In response to an abnormal temperature measuring unit, the temperature measurement value of that unit is removed, and the actual weighting coefficient of each non-abnormal temperature measuring unit is determined. The actual weighting coefficient of each non-abnormal temperature measuring unit and its temperature measurement value are used to determine the actual temperature contribution value of each non-abnormal temperature measuring unit. The actual temperature contribution values ​​of all non-abnormal temperature measuring units are accumulated to determine the temperature output result.

[0143] Specifically, if an abnormal temperature measuring unit is identified, the measurement values ​​of the remaining normal temperature measuring units are used for weighted calculation.

[0144] For example, each temperature measuring unit has its own corresponding initial weighting coefficient. When an abnormal temperature measuring unit is identified, the weighting coefficients are divided according to the initial weighting coefficients of each of the remaining temperature measuring units, so that the sum of the weighting coefficients after division is still 1.

[0145] Assume there are 3 temperature measuring units with initial weighting coefficients of 0.15, 0.4, and 0.45, respectively. If the temperature measuring unit with an initial weighting coefficient of 0.4 is faulty, then the weighting coefficients for the remaining temperature measuring units will be 0.25 and 0.75.

[0146] In some embodiments, when an abnormal temperature measuring unit is determined to exist, the following actions are also performed: outputting an alarm signal, the alarm signal including the abnormal temperature measuring unit identifier, the time of the abnormality, and the type of abnormality; and storing historical temperature measurement data, measurement deviation records of each temperature measuring unit, and abnormal event records.

[0147] In some embodiments, an automatic calibration procedure is executed in response to a preset trigger condition being met, the preset trigger condition including at least one of the following: reaching a preset time point, or the cumulative running time reaching a preset threshold.

[0148] For example, the preset time is the 1st of each month, and the preset threshold is 500 hours of cumulative operation.

[0149] Accordingly, the automatic calibration procedure includes: acquiring the current measurement value of each temperature measuring unit and determining whether the measurement deviation between each temperature measuring unit is within a preset tolerance range; when the measurement deviation of a temperature measuring unit exceeds the preset tolerance range, selecting a reference unit from multiple temperature measuring units; wherein the reference unit meets the following conditions: no abnormal records within a preset time window; measurement deviation from other temperature measuring units is less than a preset threshold; currently within the effective operating temperature range of the corresponding sensing element; using the measurement value of the reference unit as a reference, performing deviation calculation and compensation correction on the other temperature measuring units; recording the calibration time, reference unit information, and compensation parameters of this automatic calibration to the data storage module; wherein the automatic calibration procedure performs dynamic priority management of each temperature measuring unit based on the calibration cycle differences of different types of sensing elements, and selects the reference unit based on the priority.

[0150] By employing a cross-calibration mechanism with multiple sensing elements, the drift error of a single element can be effectively offset, thus extending the sensor's lifespan. Furthermore, mutual calibration is achieved by utilizing the differences in calibration cycles of different sensing elements. Through dynamic selection of reference units, cross-verification of deviation trends, and optimization of calibration strategies, self-organizing, adaptive, long-term maintenance-free operation is realized.

[0151] In some embodiments, the processor further shortens the calibration cycle of a temperature measuring unit when it detects that the deviation growth rate of a temperature measuring unit exceeds a preset threshold; and extends the calibration cycle of a temperature measuring unit when the deviation of a temperature measuring unit is less than a preset threshold for multiple consecutive calibration cycles.

[0152] In some embodiments, the processor further includes a communication interface for data interaction with external devices, wherein the communication interface is at least one of a 4-20mA analog interface, an RS485 digital interface, or a wireless communication interface.

[0153] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0154] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications, substitutions, and improvements without departing from the concept of the present invention, and these should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of this invention should be determined by the claims.

Claims

1. A temperature sensor, characterized in that, include: The package includes a package body and multiple temperature measuring units with different temperature measuring principles disposed within the package body, and each of the temperature measuring units forms a thermal coupling with the object being measured through a thermally conductive structure. A processor, electrically connected to each of the temperature measuring units, is configured to determine anomalies in each of the temperature measuring units based on measurement differences between temperature measurements of the respective temperature measuring units and / or based on a reference temperature value determined from the temperature measurements. The temperature output result is determined based on the temperature measurement value of the temperature measuring unit that is determined to be non-abnormal; The processor is configured to perform anomaly detection on each temperature measuring unit based on a reference temperature value determined by the temperature measurement values, including: in response to acquiring the temperature measurement values ​​of each temperature measuring unit, determining the temperature measurement sequence of each temperature measuring unit within a preset time window, as well as the spatial location information and thermal coupling parameters of each temperature measuring unit; constructing a reference temperature field reconstruction model based on the spatial location information of each temperature measuring unit, the thermal coupling parameters, and the temperature measurement values ​​of neighboring temperature measuring units; and determining the reference temperature value corresponding to each temperature measuring unit based on the reference temperature field reconstruction model; according to... The temperature residual index of each temperature measuring unit is calculated based on the difference between its measured temperature value and the corresponding reference temperature value. A spatial correlation graph model is constructed based on the spatial adjacency relationship between each temperature measuring unit, and a local residual consistency index is calculated for each temperature measuring unit according to the temperature residual distribution within its neighborhood. Temperature measuring units with temperature residual indices exceeding a preset threshold are identified as candidate temperature measuring units and temporarily removed from the reference temperature field reconstruction model. Temperature estimation is performed on the location of each candidate temperature measuring unit based on its neighboring temperature measuring units. The change in reconstruction error of the reference temperature field before and after removal is calculated. When the candidate temperature measurement unit meets the conditions that the temperature residual index exceeds a preset threshold, the local residual consistency index is lower than a preset threshold, and the reconstruction error decreases after removal, the candidate temperature measurement unit is determined to be an abnormal temperature measurement unit. The step of estimating the temperature at the location of the candidate temperature measurement unit based on its neighboring temperature measurement units and calculating the change in reconstruction error of the reference temperature field before and after removal includes: while masking the temperature measurement value corresponding to the candidate temperature measurement unit from the reference temperature field reconstruction model, retaining the spatial location information and thermal coupling of the candidate temperature measurement unit. The parameters are combined; based on the temperature measurement values ​​corresponding to the first-order neighboring temperature measurement units of the candidate temperature measurement unit, a first estimated temperature value is determined; based on the temperature measurement values ​​corresponding to the second-order neighboring temperature measurement units of the candidate temperature measurement unit, the first estimated temperature value is corrected, and combined with the historical temperature evolution trend of the candidate temperature measurement unit within the preset time window, an alternative temperature estimate is obtained; based on the alternative temperature estimate, the reference temperature field after removal is reconstructed; based on the difference between the measured temperature values ​​of each temperature measurement unit in the predetermined evaluation set and the reconstructed temperature values ​​corresponding to the reference temperature field before and after removal, the amount of change in reconstruction error is determined.

2. The temperature sensor according to claim 1, characterized in that, The processor is configured to determine anomalies for each of the temperature measuring units based on the measurement differences between the temperature measurements of each of the temperature measuring units, including: Calculate the temperature difference between any two temperature measurement values ​​from the temperature measuring units; In response to the presence of a temperature difference between any two of the temperature measuring units exceeding a first preset threshold, and the most frequent occurrence of a temperature difference between one of the temperature measuring units and the other temperature measuring units among the plurality of temperature measuring units exceeding the first preset threshold, the one of the temperature measuring units is determined to be an abnormal temperature measuring unit. If the temperature difference between any two of the temperature measuring units does not exceed the first preset threshold, it is determined that all temperature measuring units are normal.

3. The temperature sensor according to claim 1, characterized in that, The reference temperature field reconstruction model is constructed based on the spatial location information of each temperature measuring unit, the thermal coupling parameters, and the temperature measurement values ​​of neighboring temperature measuring units, including: Based on the spatial location information of each temperature measuring unit, the initial spatial adjacency relationship and spatial distance are determined; Based on the spatial distance, the thermal coupling parameters, and the degree of synchronization of temperature changes within the preset time window, the adjacency weights between the corresponding temperature measurement units are determined. The initial spatial adjacency relationship is corrected based on each adjacency weight to obtain a thermally coupled adjacency relationship; Obtain the temperature measurement values ​​of the neighboring temperature measurement units corresponding to each temperature measurement unit, and construct a local temperature propagation constraint relationship based on the thermal coupling adjacency relationship; Based on the spatial location information, the thermal coupling adjacency relationship, and the local temperature propagation constraint relationship, spatial smoothing constraint terms, thermal propagation constraint terms, and local observation constraint terms are established respectively, and joint modeling is performed to obtain the initial reference temperature field model. Based on the initial reference temperature field model, an objective optimization function is constructed, and the reference temperature field reconstruction model is obtained by iteratively solving the objective optimization function.

4. The temperature sensor according to claim 3, characterized in that, The step of determining the reference temperature value corresponding to each temperature measuring unit based on the reference temperature field reconstruction model includes: The temperature measurement value of the neighboring temperature measurement unit corresponding to each temperature measurement unit at the current sampling time is input into the reference temperature field reconstruction model to obtain the initial reference temperature value. Extract the historical temperature measurement sequence of the temperature measuring unit within the preset time window, and determine the time-corrected temperature value based on the historical temperature measurement sequence; Based on the initial reference temperature value and the time-corrected temperature value, a candidate reference temperature value is determined; Calculate the temperature difference between the temperature measuring unit and each neighboring temperature measuring unit, and determine the direction of the local temperature gradient based on the temperature difference and the spatial orientation relationship of each neighboring temperature measuring unit. The gradient compensation coefficient is determined based on the direction of the local temperature gradient and the corresponding thermal coupling parameters. The candidate reference temperature value is corrected using the gradient compensation coefficient to obtain the reference temperature value corresponding to the temperature measuring unit.

5. The temperature sensor according to claim 1, characterized in that, The process involves constructing a spatial association graph model based on the spatial adjacency relationships between each temperature measuring unit, and calculating the local residual consistency index for each temperature measuring unit based on the temperature residual distribution within its neighborhood, including: Each temperature measuring unit is constructed as a graph node, and the connection relationships between temperature measuring units that have spatial adjacency are constructed as graph edges; Based on the spatial distance between the temperature measurement units at both ends of the graph edge, the thermal coupling parameters, and the degree of synchronization of temperature changes, the edge weight of the corresponding graph edge is determined, and the spatial association graph model is constructed. Extract the temperature residual distribution corresponding to the target temperature measuring unit and its neighboring temperature measuring units; Calculate the residual difference between the target temperature measuring unit and each neighboring temperature measuring unit, and determine the local deviation of the target temperature measuring unit relative to each neighboring temperature measuring unit. A weighted aggregation is performed based on the local deviation values ​​and the corresponding graph edge weights to obtain a consistent result. Based on the degree of residual dispersion among the neighboring temperature measurement units within the target temperature measurement unit's neighborhood, the consistency result is compensated and corrected to obtain the local residual consistency index.

6. The temperature sensor according to claim 1, characterized in that, The determination of the temperature output result based on the temperature measurement values ​​of the temperature measuring units determined to be non-abnormal includes: In response to the assumption that all temperature measuring units are non-abnormal, the temperature contribution value of each temperature measuring unit is determined based on the temperature measurement value of each temperature measuring unit and a pre-set weighting coefficient; the temperature contribution values ​​of all temperature measuring units are accumulated to determine the temperature output result. In response to an abnormal temperature measuring unit, the temperature measurement value of that unit is removed, and the actual weighting coefficient of each non-abnormal temperature measuring unit is determined. The actual weighting coefficient of each non-abnormal temperature measuring unit and its temperature measurement value are used to determine the actual temperature contribution value of each non-abnormal temperature measuring unit. The actual temperature contribution values ​​of all non-abnormal temperature measuring units are accumulated to determine the temperature output result.

7. The temperature sensor according to claim 1, characterized in that, Meet one or more of the following conditions: The plurality of temperature measuring units include: a platinum resistance temperature measuring unit, a thermocouple temperature measuring unit, and a semiconductor temperature measuring unit; The platinum resistance temperature measurement unit uses a Pt100 or Pt1000 platinum resistance thermometer. The thermocouple temperature measuring unit uses type K, type E, or type T thermocouples; The semiconductor temperature measurement unit uses an NTC thermistor or a digital temperature sensor. Multiple temperature measuring units are disposed on the same temperature sensing end face or adjacent areas of the package body, wherein the same temperature sensing end face is the same physical contact surface or adjacent areas with a distance of less than 5mm; The package is a metal shell with an insert-type structure, including a probe part and a connecting cable. The probe part is cylindrical or needle-shaped with a diameter of 3mm to 15mm.

8. The temperature sensor according to claim 1, characterized in that, The automatic calibration procedure is executed in response to the fulfillment of a preset trigger condition, which includes at least one of the following: reaching a preset time point or the cumulative running time reaching a preset threshold. The automatic calibration procedure includes: Obtain the current measurement value of each temperature measuring unit and determine whether the measurement deviation between each temperature measuring unit is within a preset tolerance range; When the measurement deviation of a temperature measuring unit exceeds the preset tolerance range, a reference unit is selected from the multiple temperature measuring units; wherein the reference unit meets the following conditions: no abnormal records within a preset time window; the measurement deviation from the other temperature measuring units is less than a preset threshold; and it is currently within the effective operating temperature range of the corresponding sensing element. Using the measured value of the reference unit as a reference, deviation calculations are performed on the remaining temperature measuring units and compensation corrections are executed. The calibration time, reference unit information, and compensation parameters for this automatic calibration are recorded in the data storage module. The automatic calibration program dynamically prioritizes each temperature measurement unit based on the differences in calibration cycles of different types of sensing elements, and selects the reference unit based on the priority.

Images