A method and system for detecting anomalies in a temperature sensor of a restaurant exhaust duct

Abstract

The application relates to the technical field of electric data processing, and discloses an abnormality detection method and system for a temperature sensor of a smoke exhaust pipeline in the catering industry, which comprises a data acquisition module, a data processing module, a thermal change standard module, an abnormality detection module and a log management module. The data acquisition module acquires a sensor temperature value and constructs a data set; the data processing module judges and counts outliers, establishes a sensitive data set and a second data set; the thermal change standard module acquires a thermal radiation image through an infrared camera and obtains a time-pixel value curve; the abnormality detection module judges the accuracy of the sensor according to the data set and the curve; and the log management module stores data. The application can effectively detect the abnormality of the temperature sensor, and ensures the accuracy and reliability of temperature monitoring of the smoke exhaust pipeline.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric data processing, in particular to an abnormality detection method and system for a temperature sensor of a smoke exhaust pipeline in the catering industry. BACKGROUND

[0002] Due to the abnormality of the temperature of the oil pan of the chef's stove, the oil pan catches fire, and the burning stove fire and high temperature are easy to ignite the oil dirt in the smoke exhaust pipeline, causing the whole smoke exhaust pipeline to catch fire, and further causing serious secondary disasters. Therefore, it is of great significance to detect the temperature of the kitchen smoke exhaust pipeline to prevent smoke flue fire accidents and public building safety. However, this field has been blank for a long time, so it is of great significance to develop an abnormality detection method or system for a temperature sensor of a smoke exhaust pipeline in the catering industry to accurately find the temperature abnormality of the smoke flue and timely prevent and handle the smoke flue fire accident.

[0003] Traditional temperature monitoring methods rely on a single temperature sensor to detect the temperature change in the smoke exhaust pipeline in real time. However, these methods often face the risk of temperature sensor failure or misalignment, resulting in inaccurate monitoring data. In order to improve the accuracy and reliability of temperature monitoring, especially in complex working environments, a scheme using a multi-point sensor array for temperature monitoring has emerged.

[0004] However, the existing technology often lacks effective abnormality detection means to timely find the abnormal state of the temperature sensor, such as sensor failure or deviation.

[0005] Therefore, how to accurately detect and identify the abnormality of the temperature sensor through data processing and intelligent analysis, and further improve the accuracy and reliability of the temperature monitoring system of the smoke exhaust pipeline, has become a problem to be solved. SUMMARY

[0006] In view of this, the present application proposes an abnormality detection method and system for a temperature sensor of a smoke exhaust pipeline in the catering industry to solve the problem of lack of effective abnormality detection means for temperature sensors in the prior art.

[0007] On the one hand, the present application proposes an abnormality detection system for a temperature sensor of a smoke exhaust pipeline in the catering industry, comprising:

[0008] The data acquisition module is electrically connected with the sensor array, and the acquisition module is configured to acquire the temperature values of each temperature sensor in the sensor array, construct a data set of the temperature values, acquisition time and position coordinates of the corresponding temperature sensor, denoted as a first data set, and the first data set is arranged according to the acquisition time;

[0009] A data processing module is electrically connected with the data acquisition module, and is configured to judge and count outliers in the first data set, establish a sensitive data set according to the outliers, and establish a second data set according to the remaining data in the first data set.

[0010] A thermal change standard module is electrically connected with the data processing module, and is configured to acquire a thermal radiation map of the temperature sensor according to the infrared camera, acquire a change relationship between time and pixel value at each temperature sensor in the thermal radiation map, and obtain a time-pixel value curve.

[0011] An anomaly detection module is electrically connected with the data processing module and the thermal change standard module, and is configured to judge the temperature sensor accuracy according to the sensitive data set, the second data set and the time-pixel value curve.

[0012] A log management module is electrically connected with the data acquisition module, the data processing module, the thermal change standard module and the anomaly detection module, and is configured to store data.

[0013] Further, the outlier judgment step includes:

[0014] The position coordinates of each temperature sensor in the sensor array are acquired, the position relationship of each temperature sensor is determined according to the position coordinates, each temperature sensor is selected as a center one by one, the radius length is determined according to the thermal diffusion coefficient in the smoke exhaust duct and the standard deviation of the temperature values between the center temperature sensor and all adjacent temperature sensors at the same collection time, the domain range of the current temperature sensor is formed, the temperature sensors in the domain range are extracted, and the following relationship is used to judge whether it is an outlier:

[0015] ;

[0016] ;

[0017] ;

[0018] ;

[0019] ;

[0020] wherein, represents a temperature sensor i, i.e. the current temperature sensor, represents a temperature sensor j, i.e. the remaining temperature sensors in the neighborhood range except represents the temperature difference between two temperature sensors, is the temperature value of the current temperature sensor,​ is a temperature value of a remaining sensor; is a standard deviation of temperature difference, is a number of remaining sensors; is a temperature increment of a current sensor at time t, is a temperature value of a current temperature sensor at time t, is a collection period set for a temperature sensor; is a critical value of temperature difference, is an average value of temperature difference, is a critical coefficient; is a temperature increment of a remaining temperature sensor at time t; is a proportional threshold value;

[0021] if the following condition is met or , the temperature value of the current temperature sensor is determined as an outlier.

[0022] Further, the radius length is obtained by the following relationship:

[0023] ;

[0024] wherein, is a radius length, is a thermal diffusivity, is a collection period set for a temperature sensor, is a maximum interval distance between two temperature sensors.

[0025] Further, the proportional threshold value is obtained by the following method:

[0026] Obtaining correct historical data of a plurality of groups of sensor arrays, calculating the proportion of the maximum value and the minimum value collected by the temperature sensor in each group of correct historical data, and determining the proportion as the proportional threshold value.

[0027] Further, when judging and counting outliers in the first data set, and establishing a sensitive data set according to the outliers, comprising:

[0028] When it is judged that there is an outlier, the collection time and position coordinates corresponding to the outlier are uniformly input into the sensitive data set.

[0029] Further, when obtaining the time-pixel value curve, comprising:

[0030] According to each temperature sensor position coordinate, an infrared thermal radiation image is obtained by an infrared camera in each acquisition cycle, the infrared thermal radiation image is denoised based on a Gaussian filter or a wavelet transform, and pixel values at each temperature sensor are obtained according to the coordinate position; and the time-pixel value curve is established according to the acquisition cycle and the pixel value.

[0031] Further, when judging the temperature sensor accuracy according to the sensitive data set, the second data set and the time-pixel value curve, the method comprises the following steps.

[0032] The time-pixel value curve and the temperature value corresponding to each temperature sensor in the sensitive data set are obtained, the pixel value and the temperature value are normalized respectively according to the min-max normalization method, and the time-temperature value curve and the time-pixel value correction curve are obtained.

[0033] The similarity of the time-temperature value curve and the time-pixel value correction curve is calculated, the similarity is compared with a similarity threshold value, when the similarity is greater than or equal to the similarity threshold value, it is judged that the current temperature sensor is accurate, and when the similarity is less than the similarity threshold value, it is judged that the current temperature sensor is abnormal.

[0034] Further, when judging the temperature sensor accuracy according to the sensitive data set, the second data set and the time-pixel value curve, the method further comprises the following steps.

[0035] The temperature value and the coordinate position of each temperature sensor in the sensor array at the same time and the denoised thermal radiation image are obtained, the pixel value at the corresponding coordinate position in the thermal radiation image is obtained, the temperature value and the pixel value are normalized respectively, the normalized temperature value and the normalized pixel value are obtained, the normalized temperature value and the normalized pixel value at the same coordinate position are compared, if the comparison result is equal, it is judged that the temperature sensor is accurate, and if the comparison result is unequal, it is judged that the temperature sensor is abnormal.

[0036] Further, when calculating the similarity, the similarity is calculated and obtained through the following relationship:

[0037] After aligning the first temperature value and the pixel value in the time-temperature value curve and the time-pixel value correction curve, the ratio of each temperature value and pixel value is calculated in turn, and the product of the ratios is recorded as the similarity.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] The application adopts a multi-point sensor array combined with an infrared thermal radiation image for temperature monitoring. Through real-time data analysis of the sensor and the thermal radiation image, it can effectively determine whether the temperature sensor is abnormal. This multi-level, multi-dimensional monitoring method greatly enhances the accuracy of the temperature monitoring system, especially in complex kitchen environments, it can timely identify the failure or deviation of the temperature sensor, avoiding safety hazards caused by sensor failure.

[0040] The application can accurately identify abnormal values in sensor data through a scientific outlier detection method, and input the corresponding time and location data into the sensitive data set. This mechanism can effectively help the system identify temperature sensors that may have faults, and take maintenance or replacement measures in a timely manner to ensure the accuracy and stability of temperature monitoring data.

[0041] Through the thermal radiation image obtained by the infrared camera and the time-pixel value curve based on pixel value changes, the system can dynamically monitor the thermal changes of the temperature sensor and analyze the performance changes of the sensor in real time. This thermal change standard module provides an important basis for anomaly detection, further improving the accuracy and response speed of detection.

[0042] The application can accurately judge the accuracy of the temperature sensor through the minimum-maximum normalization method and the comparison of the time-temperature value curve and the time-pixel value correction curve. When the similarity reaches the set threshold, the system can confirm that the sensor is accurate; if the similarity does not meet the standard, it will be determined that the sensor is abnormal, ensuring the reliability of the temperature monitoring result.

[0043] The system is equipped with a log management module that can record and store data at each link, including temperature sensor acquisition data, anomaly detection process, and thermal change curve information. In this way, users can trace the monitoring process comprehensively, facilitating historical data analysis and fault diagnosis, and improving the maintainability of the system.

[0044] The system of the application combines data acquisition, anomaly detection, log management and other modules to form an intelligent temperature monitoring and anomaly detection system. The system not only automates data processing and anomaly judgment, but also automatically takes appropriate measures such as alarm or maintenance prompt according to the monitoring results, thereby realizing full automation of the temperature monitoring process and reducing the need for manual intervention.

[0045] The method of the application can flexibly adapt to different types of exhaust ducts and complex working environments, especially in the catering industry, where the environmental factors of exhaust ducts are complex and varied. Through accurate temperature monitoring and anomaly detection, the system can operate stably under different conditions to meet the special needs of the catering industry.

[0046] In another aspect, the present application provides a method for detecting abnormality of a temperature sensor of a kitchen exhaust duct, comprising:

[0047] S1: obtaining temperature values of each temperature sensor in the sensor array, constructing a data set of temperature values, collection time and position coordinates of the corresponding temperature sensor, denoted as a first data set, the first data set is arranged according to the collection time;

[0048] S2: judging and counting outliers in the first data set, establishing a sensitive data set according to the outliers, and establishing a second data set according to the remaining data in the first data set;

[0049] S3: obtaining a thermal radiation map of the temperature sensor according to the infrared camera, collecting the change relationship between time and pixel value of each temperature sensor in the thermal radiation map, and obtaining a time-pixel value curve;

[0050] S4: judging the accuracy of the temperature sensor according to the sensitive data set, the second data set and the time-pixel value curve;

[0051] S5: storing the first data set, the sensitive data set and the second data set.

[0052] It can be understood that the method and system for detecting abnormality of a temperature sensor of a kitchen exhaust duct provided by the present application have the same beneficial effects, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0053] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of illustrating the preferred embodiments and are not considered as limiting the present application. Moreover, the same reference signs are used to represent the same parts throughout the drawings. In the drawings:

[0054] Figure 1 The application scenario diagram of a system for detecting abnormality of a temperature sensor of a kitchen exhaust duct is provided for an embodiment of the present application.

[0055] Figure 2 The functional framework diagram of a system for detecting abnormality of a temperature sensor of a kitchen exhaust duct is provided for an embodiment of the present application.

[0056] Figure 3 The flowchart of a method for detecting abnormality of a temperature sensor of a kitchen exhaust duct is provided for an embodiment of the present application.

[0057] In the figure, 11 is a hood synchronous duct, 12 is a sensor layout area, 21 is a first fireproof valve, 22 is a second fireproof valve, 3 is an exhaust fan, and 4 is a property exhaust main duct. DETAILED DESCRIPTION

[0058] Exemplary embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood, and the scope of the present disclosure can be accurately conveyed to those skilled in the art. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0059] The application scenario of the embodiment of the present application is shown in Figure 1 The arrow in the figure indicates the direction of the flue gas flow. A plurality of flue gas sources in the kitchen first enter the hood synchronization pipeline 11. The sensor arrangement area 12 is provided with a sensor array. The flue gas sequentially passes through the sensor array, the first fire damper 21, the exhaust fan 3, and the second fire damper 22 from the hood synchronization pipeline 11, and enters the property exhaust main pipeline 4. The first fire damper 21 and the second fire damper 22 are arranged on both sides of the exhaust fan 3. The first fire damper 21 and the second fire damper 22 are used to perform a closing operation when a fire occurs, so as to achieve the effect of protecting the exhaust fan 3 and preventing the spread of flue gas or fire.

[0060] Referring to Figure 2 The embodiment of the present application provides an abnormality detection system for a temperature sensor of a kitchen exhaust pipeline, which comprises:

[0061] A data acquisition module is electrically connected with the sensor array. The acquisition module is configured to obtain the temperature values of each temperature sensor in the sensor array, construct a data set of the temperature values, the collection time, and the position coordinates of the corresponding temperature sensors, denoted as a first data set, and arrange the first data set according to the collection time;

[0062] A data processing module is electrically connected with the data acquisition module. The data processing module is configured to judge and count outliers in the first data set, establish a sensitive data set according to the outliers, and establish a second data set according to the remaining data in the first data set;

[0063] A thermal change standard module is electrically connected with the data processing module. The thermal change standard module is configured to obtain a thermal radiation image of the temperature sensor according to the infrared camera, collect the change relationship between the time and the pixel value of each temperature sensor in the thermal radiation image, and obtain a time-pixel value curve;

[0064] An abnormality detection module is electrically connected with the data processing module and the thermal change standard module. The abnormality detection module is configured to judge the accuracy of the temperature sensor according to the sensitive data set, the second data set, and the time-pixel value curve;

[0065] A log management module is electrically connected with the data acquisition module, the data processing module, the thermal change standard module and the anomaly detection module respectively, and is used for storing data.

[0066] It should be noted that the sensor array is a plurality of temperature sensors with known position coordinates pre-deployed in the smoke exhaust duct, which are used to detect temperature data of the duct, and each temperature sensor is uniformly distributed.

[0067] For ease of calculation and processing, all parameters involved in formula calculation are dimensionless before calculation.

[0068] Specifically, the data acquisition module can acquire data of the temperature sensor in real time, and construct a complete data set (first data set) according to the acquisition time and the sensor position, to ensure the timeliness and accuracy of the monitoring data. This enables the system to reflect the temperature change of the smoke exhaust duct in real time, facilitating immediate monitoring and maintenance. The data processing module can identify and count outliers, and then construct a sensitive data set according to the outliers, to effectively identify faulty sensors or abnormal temperature data. Through intelligent filtering and processing of data, the accuracy of temperature monitoring is ensured, and errors caused by sensor failure or abnormal data are reduced. The thermal change standard module combines with the infrared camera to analyze the thermal radiation change of the temperature sensor through the time-pixel value curve, to assist in judging the accuracy of the sensor. The reliability of the temperature sensor can be verified by additional thermal radiation data, enhancing the comprehensiveness and accuracy of the system for temperature monitoring. The anomaly detection module comprehensively analyzes the sensitive data set, the second data set and the time-pixel value curve, to judge the accuracy of the temperature sensor from multiple dimensions, further improving the accuracy of anomaly detection. The system can intelligently identify and process potential problems of the sensor, thereby avoiding adverse consequences caused by temperature monitoring errors. The log management module records detailed information of all data acquisition, processing, analysis and anomaly detection, providing comprehensive support for later fault tracing, data analysis and system optimization. The storage and management of data not only improve the traceability of the system, but also ensure that the system can be quickly diagnosed and recovered when a fault occurs. The system realizes a fully automated workflow from data acquisition, processing to anomaly detection, reducing the need for manual intervention and improving monitoring efficiency. Through intelligent data analysis and anomaly recognition, the system can automatically process problems when they occur, improving the safety and reliability of operation.

[0069] In some embodiments of the present application, the step of judging outliers includes:

[0070] The position coordinates of each temperature sensor in the sensor array are acquired, the positional relationship of each temperature sensor is determined according to the position coordinates, each temperature sensor is selected as a center one by one, the radius length is determined according to the thermal diffusion coefficient in the smoke exhaust duct and the standard deviation of the temperature values between the center temperature sensor and all adjacent temperature sensors at the same collection time, the field range of the current temperature sensor is formed, the temperature sensors in the field range are extracted, and whether it is an outlier is judged through the following relationship:

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] wherein, represents the temperature sensor i, i.e. the current temperature sensor, represents the remaining temperature sensors in the neighborhood range except represents the temperature difference between two temperature sensors, is the temperature value of the current temperature sensor, is the temperature value of the remaining sensor; is the standard deviation of the temperature difference, is the number of remaining sensors; is the temperature increment of the current sensor at t, is the temperature value of the current temperature sensor at t, is the collection period set for the temperature sensor; is the critical value of the temperature difference, is the average value of the temperature difference, is the critical coefficient; is the temperature increment of the remaining temperature sensor at t; is the proportional threshold value;

[0077] If or , the temperature value of the current temperature sensor is judged as an outlier.

[0078] Specifically, the positional relationship is that the current temperature sensor and the remaining temperature sensors are adjacent or separated by how many temperature sensors.

[0079] ​​​​​​It should be noted that by considering the thermal diffusion coefficient in the smoke exhaust duct and the positional coordinate relationship between the sensors, the standard deviation is used to determine the range of the current temperature sensor. The judgment radius can be dynamically adjusted according to the actual environmental heat conduction characteristics, ensuring that the identification of outliers is more in line with the actual physical laws. Compared with the traditional judgment method based on fixed radius or simple distance, this judgment method based on thermal diffusion is more accurate and can better adapt to the temperature variation characteristics in different environments. In the outlier judgment, each temperature sensor is selected as the center one by one, and the temperature difference with the surrounding adjacent sensors is used to determine whether it is an outlier. The system does not rely on a static global model, but dynamically adjusts the judgment standard according to the local environment and changes of each sensor, and has strong adaptability. By considering multiple factors such as temperature difference, temperature increment and standard deviation, combined with the data of adjacent temperature sensors, this judgment method can accurately capture the instantaneous abnormal fluctuations, and not only relies on a single data point. This comprehensive use of multiple statistical characteristics can effectively avoid false judgments caused by environmental noise or errors of a single sensor, and improves the robustness of outlier identification. In the judgment process, the proportion threshold is calculated and compared with the temperature difference, so that the system can adaptively adjust the sensitivity. This dynamic adjustment mechanism can flexibly adjust the threshold of anomaly detection according to different environments, sensor performance and historical data changes, enhancing the flexibility and precision of the system.

[0080] In some embodiments of the present application, the radius length is obtained by the following relationship:

[0081] ;

[0082] wherein, is the radius length, is the thermal diffusion coefficient, is the collection period set for the temperature sensor, is the maximum interval distance between two temperature sensors.

[0083] It should be noted that by introducing the thermal diffusion coefficient into the radius calculation, the judgment range can be dynamically adjusted according to the heat conduction characteristics. The thermal diffusion coefficient takes into account the propagation speed and range of heat in the pipeline, making the judgment process more in line with the actual heat conduction law. This makes the selection of radius length not only dependent on distance, but also reflects the actual diffusion trend of temperature change, improving the accuracy of anomaly detection. Combined with the collection period of the temperature sensor, the system can automatically adjust the sensitivity of outlier judgment according to the collection frequency and data change period of different sensors. When the collection period is short, it can identify fast-changing anomalies more finely, and when the collection period is long, it can reduce false positives caused by short-term fluctuations. Through adaptive adjustment of the collection period, the response speed and flexibility of the system are enhanced. Introducing the maximum interval distance as a parameter for radius calculation ensures that even when the physical distance between sensors in the pipeline is large, the radius length can still reasonably cover the relevant area, avoiding missing anomaly detection due to too far apart sensors. By reasonably setting the maximum interval, the comprehensiveness of the detection area can be ensured, reducing the risk of missed detection. By considering multiple factors (thermal diffusion coefficient, collection period, maximum interval distance), the size of the outlier detection area can be automatically adjusted according to different environments and different sensor configurations. In this way, different environmental conditions and sensor layouts can be adapted, greatly improving the accuracy of anomaly detection and avoiding errors that may be caused by fixed radius judgment.

[0084] In some embodiments of the present application, the proportion threshold is obtained by the following method:

[0085] Obtain correct historical data of several groups of sensor arrays, calculate the proportion of the maximum value and the minimum value collected by the temperature sensor in each group of correct historical data, and determine the proportion as the proportion threshold.

[0086] By analyzing historical data, the ratio of the maximum value to the minimum value in each set of data can be calculated to provide a threshold for temperature sensor anomaly detection based on actual data. This can reflect the temperature variation range of the sensor under normal working conditions, providing an actual and feasible reference standard for subsequent anomaly detection. According to the actual temperature fluctuations in historical data, the proportional threshold is dynamically determined, avoiding the false positives that may be caused by fixed thresholds. Over time, the system can automatically adjust the proportional threshold based on new historical data, ensuring that the anomaly detection standard matches the actual working environment and data fluctuations, improving the adaptability and robustness of the system. Traditional anomaly detection methods often rely on manually set thresholds, which may produce errors due to changes in different environments, sensor types, or working conditions. By automatically calculating the proportional threshold based on historical data, the errors caused by human intervention are reduced, improving the accuracy and consistency of anomaly detection. By calculating the ratio of the maximum value to the minimum value, the actual amplitude of temperature variation can be captured to determine whether there is an anomaly. The system can effectively identify situations where temperature fluctuations exceed a reasonable range, improving the sensitivity and detection accuracy of the system to sudden anomalies. In addition, it can adapt to the temperature fluctuation characteristics of different sensor arrays and different working conditions. When the configuration of the sensor array or the working environment changes, the system can flexibly adjust the detection standard through the proportional threshold calculation of historical data, ensuring high accuracy in different situations.

[0087] In some embodiments of the present application, when judging and counting outliers in the first data set, and establishing a sensitive data set according to the outliers, it includes:

[0088] When it is judged that there is an outlier, the collection time and position coordinates corresponding to the outlier are uniformly input into the sensitive data set.

[0089] In some embodiments of the present application, when obtaining the time-pixel value curve, it includes:

[0090] According to the position coordinates of each temperature sensor, an infrared thermal radiation image is obtained by the infrared camera in each collection period, the infrared thermal radiation image is denoised based on Gaussian filtering or wavelet transform, and the pixel value at each temperature sensor is obtained according to the coordinate position; the time-pixel value curve is established according to the collection period and the pixel value.

[0091] It is necessary to note that by using Gaussian filtering or wavelet transform to denoise the infrared thermal radiation image, the environmental noise and interference can be effectively reduced, ensuring that the collected thermal radiation image is clearer and more stable. This denoising process helps to improve the accuracy of subsequent temperature measurement and data analysis, ensuring that the pixel values obtained by the temperature sensor accurately reflect the actual temperature changes. According to the position coordinates of each temperature sensor, the corresponding pixel values are extracted from the infrared thermal radiation image, which can realize the precise association of the temperature sensor and the infrared thermal radiation image. In this way, the data of the temperature sensor can be directly corresponded to the thermal changes in the environment, enhancing the spatial perception ability of the system to temperature changes. By combining the acquisition period and the pixel value to construct the time-pixel value curve, the system can track and record the temperature data of the temperature sensor over time. This system can capture the dynamic trend of temperature, thereby more effectively identifying abnormal fluctuations or sudden changes in temperature changes, which helps to detect potential system failures or abnormal states in a timely manner. The combination of infrared thermal radiation images and pixel values can calibrate and verify the measurement accuracy of the sensor, especially in complex environments or multiple temperature sensors coexist. Through the association of temperature sensor positions and infrared image data, the limitations of single sensor data can be effectively compensated, and the overall accuracy and environmental adaptability of the system can be improved. The time-pixel value curve provides a time series data perspective, allowing the system to perform more complex data analysis, such as trend analysis, anomaly detection, model prediction, etc. By analyzing the pattern of temperature changes, the system can more sensitively identify sudden temperature anomalies and provide early warning of potential failures or safety issues. Combining the position coordinates of the temperature sensor with the pixel values in the infrared image allows the system to accurately locate the temperature abnormal area in space and time. When a failure or anomaly occurs, the time-pixel value curve and thermal radiation image can be used to quickly locate the fault, improving the efficiency of system fault diagnosis and maintenance. Through the combination of infrared thermal radiation images and pixel value data of temperature sensors, a visual thermal map or temperature change curve can be generated, which can intuitively display the system running state. This visualization effect helps managers quickly understand the dynamic characteristics of temperature changes and make timely decisions.

[0092] In some embodiments of the present application, when judging the accuracy of the temperature sensor according to the sensitive data set, the second data set and the time-pixel value curve, it includes:

[0093] The time-pixel value curve and the acquisition time and temperature value of each temperature sensor in the sensitive data set are obtained, and the pixel value and the temperature value are normalized respectively according to the min-max normalization method, to obtain the time-temperature value curve and the time-pixel value correction curve;

[0094] The similarity of the time-temperature value curve and the time-pixel value correction curve is calculated, and the similarity is compared with a similarity threshold value. When the similarity is greater than or equal to the similarity threshold value, it is judged that the current temperature sensor is accurate. When the similarity is less than the similarity threshold value, it is judged that the current temperature sensor is abnormal.

[0095] It should be noted that by calculating the similarity of the time-temperature value curve and the time-pixel value correction curve, the accuracy of the temperature sensor can be judged based on the similarity between the two data sets. Using the min-max normalization method to standardize the pixel value and the temperature value respectively can eliminate the deviation caused by different data scales or dimensions, making the similarity calculation more objective and accurate. The system can effectively reduce the interference of external factors on the accuracy judgment of the temperature sensor, and improve the reliability of the accuracy judgment. By processing the time-pixel value and the time-temperature value through min-max normalization, it is ensured that the two data sets are within the same range, avoiding comparison errors between different scale data. This normalization process helps to improve the consistency of the data, providing more accurate and reliable input for subsequent similarity calculation, effectively improving the detection ability of the system. According to the comparison of the calculated similarity and the similarity threshold value, when the similarity is lower than the set threshold value, the system can timely judge that the current temperature sensor is abnormal. This real-time accuracy detection mechanism can quickly identify the sensors with problems and issue an early warning, avoiding temperature monitoring failure caused by sensor failure, and improving the early warning ability and response speed of the system. By using the similarity calculation and threshold comparison method, the system can sensitively capture subtle temperature change abnormalities. When the data of the sensor and the data of the infrared thermal radiation image are significantly different, even small temperature fluctuations can be detected in time. This not only helps to find abnormalities caused by sensor failure, calibration errors or external interference, but also avoids false positives and false negatives, enhancing the accuracy of detection. Based on similarity judgment and threshold comparison, the system can dynamically adjust and optimize. If some temperature sensors have long-term accuracy problems, the system can compensate and optimize through historical data analysis or re-calibration strategies to ensure that the sensors always maintain a high level of accuracy, improving the long-term stability of the entire system. When a temperature sensor is found to be abnormal, the system can record the abnormal information of the sensor through the log management module, helping maintenance personnel quickly locate and solve the problem. The system reduces the need for manual intervention and improves the system's automated monitoring capabilities, making the system more fault-tolerant and maintainable. In a multi-sensor array, the accuracy of multiple sensors can be judged to ensure the coordinated work of the entire sensor array. When the accuracy of some sensors is found to be low, the system can automatically adjust or redistribute the acquisition tasks to ensure the overall consistency and effectiveness of the data. This not only improves the overall accuracy of the sensors, but also ensures the stability of the system during long-term operation.

[0096] In some embodiments of the present application, when judging the accuracy of the temperature sensor according to the sensitive data set, the second data set and the time-pixel value curve, further comprising:

[0097] Obtaining the temperature values and coordinate positions of each temperature sensor in the sensor array at the same time, and the denoised thermal radiation image, normalizing the temperature values and the pixel values at the corresponding coordinate positions in the thermal radiation image respectively to obtain normalized temperature values and normalized pixel values, comparing the normalized temperature values and the normalized pixel values at the same coordinate positions, and if the comparison result is equal, judging that the temperature sensor is accurate, and if the comparison result is not equal, judging that the temperature sensor is abnormal.

[0098] It should be noted that the normalization process can eliminate the difference in the dimension of temperature value and pixel value, so that the data of two different sources can be compared on the same scale. By normalizing the temperature value and the pixel value, it can be ensured that the two have the same weight when compared, thereby avoiding the deviation or error that may be caused by the different dimensions of the data, and ensuring the accuracy of the comparison. The normalized temperature value and pixel value are directly compared, and if they are equal, it is judged to be accurate, and if they are not equal, it is judged to be abnormal. By simplifying the judgment standard, the complexity of the judgment process is reduced, making the accuracy detection of the temperature sensor more direct and efficient. Avoiding complex algorithms and multiple calculation steps, the detection speed is improved. The temperature value is directly related to the pixel value in the thermal radiation image, and the comparison after normalization enables the system to sensitively capture the accuracy deviation caused by sensor failure, external environmental influence or calibration error. The system can also respond to slight abnormal changes, enhancing the system's ability to identify sensor accuracy problems. The system can compare the temperature value of the temperature sensor with the corresponding pixel value of the thermal radiation image in real time, and make judgments based on each acquisition cycle. By monitoring and judging the accuracy of the temperature sensor in real time, abnormalities can be detected in time, thereby reducing data distortion or monitoring errors caused by sensor failure and improving the response speed and accuracy of the system. Due to the standardization of the comparison data after normalization, the abnormal judgment of the temperature sensor is more objective, which can reduce the false alarm or missed alarm caused by data fluctuations, environmental changes and other factors. This makes the system more stable in long-term operation, reducing the maintenance cost caused by inaccurate alarms. By synchronously comparing the temperature value of the temperature sensor with the pixel value obtained by the infrared camera, the data consistency between multiple sensors can be effectively ensured. When some sensors have accuracy errors, the system can discover and adjust in real time to ensure the coordination and data consistency of the overall temperature monitoring system, avoiding the influence of local sensor failure on the accuracy of global data. The system combines the comparison of temperature sensor and infrared camera image data, making the system have higher intelligent level. The system can automatically detect accuracy and issue abnormal alarms according to the matching of temperature and pixel data, thereby reducing manual intervention and operation and maintenance cost, and improving the automation ability of the system.

[0099] In some embodiments of the present application, when calculating the similarity, it is calculated by the following relationship:

[0100] After aligning the first temperature value and pixel value in the time-temperature value curve and the time-pixel value correction curve, the ratio of each temperature value and pixel value is calculated in turn, and the product of the ratios is recorded as the similarity.

[0101] It needs to be explained that the relationship between the two kinds of data is quantified into a specific numerical value by ratio calculation and multiplication of the temperature value and the pixel value. The greater the similarity, the higher the accuracy of the temperature sensor. The system effectively avoids the calculation complexity brought by the use of complex distance measurement or other algorithms, making the similarity calculation more concise, direct and efficient. The ratio calculation between each temperature value and the corresponding pixel value can capture the small changes of the temperature sensor and the thermal radiation image pixel value at different collection times. By multiplying these ratio products to obtain the similarity, the system can more sensitively detect changes in sensor accuracy and timely detect sensor abnormalities, avoiding problems caused by data deviation or sensor failure. The similarity calculation uses the ratio product method, which can effectively eliminate the influence between different data magnitudes and scales, avoiding errors caused by differences in numerical ranges. For example, temperature values and pixel values may change in different ranges. By calculating the ratio and then multiplying the products, the scale can be unified, thereby reducing the calculation errors caused by differences in data itself. By comparing each data point of the time-temperature value curve and the time-pixel value correction curve, the correlation between the temperature change and the thermal radiation image pixel value change in the time dimension can be effectively captured. The trend changes in the time series data are fully considered, ensuring accurate identification and judgment of dynamic changes. Compared with other similarity calculation methods (such as Euclidean distance, cosine similarity, etc.), the ratio product calculation is more concise and has lower computational complexity. Therefore, the system can provide efficient and accurate similarity evaluation without increasing the computational burden, which is suitable for scenarios that require real-time or large-scale data processing. Through this ratio product calculation, the value of the similarity can directly reflect the degree of accurate matching between the temperature sensor and the thermal radiation image of the infrared camera. The system provides a clear quantitative standard for subsequent accuracy judgment, facilitating subsequent decision-making and alarm mechanisms. The system provides a simple and intuitive detection mechanism that can quickly locate and identify accuracy abnormal temperature sensors. This allows staff to quickly find potential faulty sensors in daily maintenance, reducing troubleshooting time and cost and improving system maintainability.

[0102] Referring to Figure 3 The embodiment of the application also provides an abnormality detection method for a temperature sensor of a smoke exhaust duct in the catering industry, comprising:

[0103] S1: Obtain the temperature values of each temperature sensor in the sensor array, construct a data set of temperature values, collection times and position coordinates of the corresponding temperature sensors, denoted as a first data set, and arrange the first data set according to the collection times;

[0104] S2: Determine and count the outliers in the first data set, establish a sensitive data set according to the outliers, and establish a second data set according to the remaining data in the first data set;

[0105] S3: obtaining a thermal radiation image of the temperature sensor according to the infrared camera, collecting a time-pixel value curve of each temperature sensor in the thermal radiation image;

[0106] S4: judging the accuracy of the temperature sensor according to the sensitive data set, the second data set and the time-pixel value curve;

[0107] S5: storing the first data set, the sensitive data set and the second data set.

[0108] Finally, it should be noted that: the above examples are used to illustrate the technical solutions of the present application, but not to limit it, although the present application has been described in detail with reference to the above examples, those skilled in the art should understand that: the specific embodiments of the present application can be modified or replaced by the same, without departing from the spirit and scope of the present application, any modification or equivalent replacement, which should be covered within the protection scope of the claims of the present application.

Claims

1. An anomaly detection system for a temperature sensor in a catering industry exhaust duct, characterized in that, include: A data acquisition module is electrically connected to the sensor array. The acquisition module is configured to acquire the temperature values ​​of each temperature sensor in the sensor array and construct a dataset of the temperature values, acquisition time, and corresponding temperature sensor position coordinates, denoted as the first dataset. The first dataset is arranged according to the acquisition time. The data processing module is electrically connected to the data acquisition module. The data processing module is configured to identify and count outliers in the first dataset, establish a sensitive dataset based on the outliers, and establish a second dataset based on the remaining data in the first dataset. A thermal change standard module is electrically connected to the data processing module. The thermal change standard module is configured to acquire the thermal radiation map of the temperature sensor obtained by the infrared camera, collect the time-pixel value change relationship at each temperature sensor in the thermal radiation map, and obtain a time-pixel value curve. An anomaly detection module is electrically connected to the data processing module and the thermal change standard module, respectively. The anomaly detection module is configured to determine the accuracy of the temperature sensor based on the sensitive dataset, the second dataset and the time-pixel value curve. The log management module is electrically connected to the data acquisition module, data processing module, thermal change standard module and anomaly detection module respectively, and the log management module is used to store data. The steps for identifying outliers include: Obtain the position coordinates of each temperature sensor in the sensor array. Determine the positional relationship of each temperature sensor based on the position coordinates. Select each temperature sensor as the center. Determine the radius length based on the thermal diffusivity in the exhaust duct and the standard deviation of the temperature values ​​between the central temperature sensor and all adjacent temperature sensors at the same acquisition time. This forms the current temperature sensor's neighborhood. Extract the temperature sensors within the neighborhood and determine whether they are outliers using the following relationship: ； ； ； ； ； in, This refers to temperature sensor i, i.e., the current temperature sensor. Represents temperature sensor j, i.e., except for The remaining temperature sensors outside the neighborhood range, This indicates the temperature difference between the two temperature sensors. The current temperature value from the temperature sensor. The remaining sensor temperature values; The standard deviation of the temperature difference. This represents the number of remaining sensors. This represents the temperature increment of the sensor at time t. This represents the temperature value of the current temperature sensor at time t. The sampling period set for the temperature sensor; This is the critical value for temperature difference. This represents the average temperature difference. This is the critical coefficient; This represents the temperature increment of the remaining temperature sensor at time t. This is the proportional threshold; If the following conditions are met or then the temperature value of the current temperature sensor is judged as an outlier. The radius length is obtained through the following relationship: ； wherein, is a radius length, is a thermal diffusivity, is a collection period set for the temperature sensor, is a maximum separation distance between the two temperature sensors; The ratio threshold is obtained through the following method: Obtain correct historical data from several sets of sensor arrays, calculate the ratio of the maximum to the minimum value collected by the temperature sensor in each set of correct historical data, and determine the ratio as the ratio threshold. When acquiring the time-pixel value curve, the following is included: Based on the location coordinates of each temperature sensor, an infrared thermal radiation map is obtained by an infrared camera in each acquisition cycle. The infrared thermal radiation map is then denoised using Gaussian filtering or wavelet transform. The pixel value at each temperature sensor is obtained based on the coordinate position. A time-pixel value curve is then established based on the acquisition cycle and the pixel value. When determining the accuracy of the temperature sensor based on the aforementioned sensitive dataset, the second dataset, and the time-pixel value curve, the following is included: The time-pixel value curve and the acquisition time and temperature value of each temperature sensor in the sensitive dataset are obtained. The pixel value and temperature value are normalized according to the minimum-maximum normalization method to obtain the time-temperature value curve and the time-pixel value correction curve. The similarity of the time-temperature value curve and the time-pixel value correction curve is calculated, and the similarity is compared with a similarity threshold value. When the similarity is greater than or equal to the similarity threshold value, it is determined that the current temperature sensor is accurate. When the similarity is less than the similarity threshold value, it is determined that the current temperature sensor is abnormal.

2. The restaurant exhaust duct temperature sensor abnormality detection system according to claim 1, characterized by, When the outliers in the first data set are determined and counted, and the sensitive data set is established according to the outliers, the following steps are included: When it is determined that there are outliers, the acquisition time and position coordinates corresponding to the outliers are uniformly input into the sensitive data set.

3. The restaurant exhaust duct temperature sensor abnormality detection system according to claim 1, characterized by, When the accuracy of the temperature sensor is determined according to the sensitive data set, the second data set, and the time-pixel value curve, the following steps are further included: The temperature values and coordinate positions of each temperature sensor in the sensor array at the same time, and the denoised thermal radiation image are obtained. According to the pixel values at the corresponding coordinate positions in the thermal radiation image, the temperature values and pixel values are normalized respectively to obtain normalized temperature values and normalized pixel values. The normalized temperature values and normalized pixel values at the same coordinate positions are compared. If the comparison result is equal, it is determined that the temperature sensor is accurate. If the comparison result is not equal, it is determined that the temperature sensor is abnormal.

4. The restaurant exhaust duct temperature sensor abnormality detection system according to claim 3, characterized by When the similarity is calculated, it is calculated by the following relationship: After aligning the first temperature value and pixel value in the time-temperature value curve and the time-pixel value correction curve, the ratio of each temperature value and pixel value is calculated in turn. The product of the ratios is recorded as the similarity.

5. A method for detecting an anomaly in a temperature sensor of a restaurant exhaust duct, applied to the system of any one of claims 1 to 4, characterized in that, It includes: S1: Obtain the temperature values of each temperature sensor in the sensor array, construct a data set of temperature values, acquisition time, and position coordinates of the corresponding temperature sensor, denoted as a first data set, and the first data set is arranged according to the acquisition time; S2: Determine and count the outliers in the first data set, establish a sensitive data set according to the outliers, and establish a second data set according to the remaining data in the first data set; S3: Obtain the thermal radiation image of the temperature sensor according to the infrared camera, and acquire the change relationship of time and pixel value at each temperature sensor in the thermal radiation image to obtain a time-pixel value curve; S4: Determine the accuracy of the temperature sensor according to the sensitive data set, the second data set, and the time-pixel value curve; S5: Store the first data set, the sensitive data set, and the second data set.

Images