Bridge suspender pin shaft sensor for health monitoring and use method

Abstract

The invention relates to a bridge suspender pin shaft sensor for health monitoring and a use method, and the bridge suspender pin shaft sensor for health monitoring comprises a pin shaft main body, a plurality of eddy current sensors for detecting deformation displacement, and a plurality of corresponding temperature sensors for detecting temperature. The pin shaft body is provided with a plurality of annular grooves, the eddy current sensors and the temperature sensors are integrally arranged in the annular grooves, and each eddy current sensor and the corresponding temperature sensor form a sensor group. And the stress value of the pin shaft is deduced by combining detection data of the eddy current sensor and the temperature sensor. The invention provides a sensor capable of directly measuring the mechanical condition of a suspender pin shaft, the mechanical condition of the pin shaft of a bridge sling and a rod can be monitored for a long time, the clear degree of monitoring the bridge suspender pin shaft is improved, and the stability and the anti-interference capability of the monitoring sensor are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of pin monitoring, and in particular to a bridge hanger pin sensor for health monitoring and a use method thereof. Background Art

[0002] At present, large bridges such as arch bridges and suspension bridges often have hanger components. These large bridges are hyperstatic structures, and the local mechanical conditions are affected by the overall stress conditions. In these large bridges, the hangers transfer the load of the bridge deck main beam to the supporting bodies such as the cables and arches. These supporting bodies are key force-transmitting components. Among them, the pins on the bridge cables and rods provide stable support for the cables and rods and other components.

[0003] In the related art, the vibration or strain sensor used to monitor the vibration or strain of the pin shaft establishes a correlation with the health status of the bridge, and uses the correlation and the monitoring data of the sensor to monitor the pin shaft. However, such correlation is usually more complicated, resulting in unclear monitoring of the pin shaft, and more obvious characteristics can only be seen after the pin shaft structure has deteriorated to a higher degree. Summary of the invention

[0004] In order to provide a sensor that can directly measure the mechanical condition of the hanger pin, and can monitor the pin mechanical state of the bridge cables and rods for a long time, improve the clarity of bridge hanger pin monitoring, and enhance the stability and anti-interference ability of the monitoring sensor, the present application provides a bridge hanger pin sensor for health monitoring and a method of use.

[0005] In the first aspect, the above-mentioned invention objective of the present application is achieved through the following technical solutions: A bridge hanger pin sensor for health monitoring, the bridge hanger pin sensor for health monitoring comprising: a pin body, a plurality of eddy current sensors for detecting deformation displacement and a corresponding plurality of temperature sensors for detecting temperature, the pin body being provided with a plurality of annular grooves, the eddy current sensors and the temperature sensors being integrated in the annular grooves, each of the eddy current sensors and the corresponding temperature sensor forming a sensor group, the detection data of the eddy current sensors and the temperature sensors being combined to infer the pin force value.

[0006] By adopting the above technical scheme, the bridge hanger pin sensor for health monitoring includes a pin body and several sensors, the sensors include several eddy current sensors for detecting deformation displacement and corresponding several temperature sensors for detecting temperature, these sensors are connected to the pin through several annular grooves provided on the pin body, so that the mechanical state of the pin can be measured more accurately and directly, reducing the interference of other factors on the sensor, and because the sensor is installed in the annular groove of the pin and integrated with the pin, the probability of sensor damage or falling off is reduced, the service life of the sensor is extended, and the sensor can monitor the mechanical state of the pin of the bridge sling and rod for a long time; in addition, the eddy current sensor is mainly used for non-contact measurement of the displacement of the pin. When the pin vibrates or deforms during operation, the eddy current sensor can accurately capture its displacement change. This displacement change is an important manifestation of the mechanical state of the pin and can It reflects whether the pin is subjected to excessive external force and whether there are problems such as wear or fatigue. The temperature sensor can measure the temperature of the pin in real time. The temperature change is closely related to the mechanical state of the pin. For example, when the pin is subjected to excessive friction or load, its temperature may rise, thereby indirectly reflecting the force condition, friction state and potential damage risk of the pin. There is a more prominent correlation between the monitoring data of the eddy current sensor and the temperature sensor. For example, when the pin is subjected to excessive friction, the monitoring data of the eddy current sensor and the temperature sensor will increase. Therefore, the eddy current sensor is used in combination with the temperature sensor to improve the clarity of monitoring the bridge hanger pin. In addition, the mechanical state of the pin can be more comprehensively understood by comprehensively analyzing the displacement and temperature information, thereby improving the accuracy and reliability of the measurement. It can also promptly discover possible problems with the pin, providing an important basis for preventing and solving pin failures.

[0007] In a preferred example, the present application can be further configured as follows: the temperature sensor is a thermocouple sensor, and each of the eddy current sensors is integrated with the corresponding thermocouple sensor and arranged in the annular groove.

[0008] By adopting the above technical solution, the temperature sensor adopts a thermocouple sensor, and the thermocouple is in direct contact with the object to be measured, avoiding the influence of the intermediate medium on the measurement result, thereby ensuring higher measurement accuracy. The thermocouple sensor can perform continuous measurement in an extremely wide temperature range, from extremely low temperature to extremely high temperature, and can maintain good performance, so that the thermocouple can adapt to various complex and extreme working environments, and the thermal mass of the thermocouple sensor is low, so it can respond to temperature changes more quickly; in addition, each eddy current sensor is integrated with the corresponding thermocouple sensor, so that each eddy current sensor and the corresponding thermocouple sensor can be detected simultaneously to ensure the accuracy and synchronization of the detection data.

[0009] In a preferred example, the present application can be further configured as follows: a plurality of the eddy current sensors and the thermocouple sensors are arranged on both sides of the annular groove, the eddy current sensors arranged on both sides of the same annular groove are staggered with each other, and a metal shell is arranged in the annular groove, and the metal shell is used to cover the eddy current sensors and the thermocouple sensors.

[0010] By adopting the above technical solution, the annular groove on the pin shaft surrounds the pin shaft, and a number of eddy current sensors and thermocouple sensors are arranged on both sides of the annular groove. The eddy current sensors and the corresponding thermocouple sensors are evenly arranged in the entire annular groove. The sensors are arranged on the free surface of the groove side, which can avoid the influence of creep of the main stress-bearing parts of the pin shaft on the sensor readings, and realize all-round monitoring of the pin shaft. When displacement or temperature change occurs in different parts of the pin shaft, the eddy current sensor and the corresponding thermocouple sensor can capture it in time, thereby ensuring a comprehensive detection of the mechanical state of the pin shaft, and the uniform distribution of multiple groups of sensors makes it possible to obtain data from multiple angles and positions, which helps to reduce the errors that may be generated by a single sensor, thereby improving the overall measurement accuracy, and the setting of multiple sensors provides data redundancy, thereby ensuring the reliability of the entire monitoring system; in addition, a metal shell is arranged in the annular groove, and the metal shell is used to cover the eddy current sensor and the thermocouple sensor, further reducing the interference of external environmental factors on sensor detection, thereby further improving the overall measurement accuracy.

[0011] In a preferred example, the present application can be further configured as follows: the eddy current sensors and the thermocouple sensors in the same sensor group take readings simultaneously, and the eddy current sensors and the thermocouple sensors in each sensor group take readings in sequence in a scanning manner.

[0012] By adopting the above technical solution, the eddy current sensor and the thermocouple sensor take readings at the same time, which can ensure that the displacement and temperature data of the pin shaft are obtained at the same time point. The data synchronization helps to more accurately analyze the working status of the pin shaft and avoid errors caused by time differences. Moreover, since the eddy current and thermocouple data obtained at the same time are paired, data analysis and comparison can be conveniently performed, which makes it easy to discover the correlation between the detection data of the eddy current sensor and the thermocouple sensor, thereby discovering more details and characteristics of the working status of the pin shaft.

[0013] In the second aspect, the above invention objective of the present application is achieved through the following technical solutions: A method for using a bridge hanger pin sensor for health monitoring, using the above-mentioned bridge hanger pin sensor for health monitoring, the method for using the bridge hanger pin sensor for health monitoring includes: Obtain eddy current detection data and corresponding temperature detection data; Performing initialization processing on the eddy current detection data and the corresponding temperature detection data to obtain initialization eddy current detection data and initialization temperature detection data; Inputting the initialization eddy current detection data and the initialization temperature detection data into a preset neural network model, and obtaining a plurality of model output values ​​based on the neural network model; The sensor test result is obtained by combining several of the model output values ​​for calculation.

[0014] By adopting the above technical solution, the eddy current detection data of each eddy current sensor and the temperature detection data of the corresponding thermocouple sensor are obtained, and the eddy current detection data of each eddy current sensor and the temperature detection data of the corresponding thermocouple sensor are initialized. The initialization processing refers to the calibration processing of the eddy current detection data and the temperature detection data. For example, the eddy current data and the temperature detection data measured when the pin shaft is in a stationary or known state and is stable without the influence of external heat sources are subtracted from the eddy current detection data and the temperature detection data obtained by the detection, so as to perform the initialization processing to ensure the accuracy of the detection. The preset neural network A network model refers to a model used to analyze eddy current detection data and temperature detection data and then determine the bearing force value of the pin shaft. For example, the correlation between the eddy current detection data and the temperature detection data and the mechanical state of the pin shaft is analyzed through a neural network algorithm. In addition, since the eddy current detection data and the temperature detection data include detection data of multiple groups of eddy current sensors and temperature sensors in multiple annular grooves on the pin shaft, different positions on the pin shaft may affect each other. Therefore, the neural network model performs multiple analyses, inputs multiple results, and then calculates the average value of multiple output values ​​of the neural network model to obtain the final test result and improve the accuracy of the test.

[0015] In a preferred example, the present application can be further configured as follows: the neural network model is obtained by: obtaining the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, and obtaining several groups of training data sets under calibration test conditions, wherein each group of the training data sets includes the calibration detection data of each eddy current sensor, the corresponding calibration detection data of each temperature sensor, and the corresponding pin bearing force value, and each group of the training data sets is sorted according to the calibration test sequence of each group of the training data sets, and the calibration detection data of each eddy current sensor and the corresponding calibration detection data of the temperature sensor in each group of the training data sets are sorted according to a preset sensor sequence; Based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, initialization processing is performed on the calibrated detection data of each eddy current sensor and the calibrated detection data of each temperature sensor in each group of the training data set; A neural network training is performed based on each set of the training data sets to generate a neural network model.

[0016] By adopting the above technical solution, the neural network model is the main means of analyzing the mechanical state of the pin. The prediction accuracy of the neural network model determines the accuracy of the test results. Therefore, it is necessary to improve the robustness of the neural network model. Based on this, the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load are obtained. The initial detection data is the recorded reading of each sensor when it is factory set under no load. The initial detection data is used as the value of the initialization processing. Since the initial detection data is recorded when it is factory set under no load, it directly reflects the initial state of the sensor on the pin, making the initial detection data more accurate. It can accurately represent the actual initial conditions of the sensor on the pin shaft. Compared with the commonly used method of taking the detection data under standard conditions as the value for initialization, it can avoid introducing errors caused by the difference between standard conditions and actual working conditions. In this way, the accuracy of neural network training is improved by improving the accuracy of training samples. In addition, each training data set is sorted according to the calibration test sequence of each training data set, and the calibration detection data of each eddy current sensor and the corresponding calibration detection data of the temperature sensor in each training data set are sorted according to the preset sensor sequence, so as to facilitate finding abnormal data during neural network training.

[0017] In a preferred example, the present application can be further configured as follows: performing neural network training based on each set of the training data sets to generate a neural network model specifically includes: Each time when the neural network training is performed based on each group of the training data sets, the calibration detection data of any one of the eddy current sensors and the calibration detection data of the corresponding temperature sensor in the training data sets are randomly selected; The calibration detection data of the randomly selected eddy current sensor and the calibration detection data of the corresponding temperature sensor are excluded from the training data set of the current neural network training, and the neural network training is performed based on the remaining data in the training data set to generate a neural network model.

[0018] By adopting the above technical scheme, in order to improve the robustness and anti-interference ability of the neural network model, it is necessary to improve the neural network model, each set of training data sets represents the data obtained by testing under calibration test conditions, and each time the neural network training is performed based on each set of training data sets, the calibration detection data of any eddy current sensor in the training data set and the calibration detection data of the corresponding temperature sensor are randomly selected, and the selected calibration detection data of the eddy current sensor and the calibration detection data of the corresponding temperature sensor are frozen and excluded in the training data set of the current neural network training, that is, by the random truncation method, a group of eddy current sensor and temperature sensor data are randomly frozen in the data obtained by testing under calibration test conditions each time before training, so that the effect of improving the robustness and anti-interference ability of the neural network model is achieved by randomly truncating each set of training data sets.

[0019] In a preferred example, the present application can be further configured as: the eddy current detection data and the temperature detection data are initialized to obtain the initialized eddy current detection data and the initialized temperature detection data, specifically including: based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, the eddy current detection data and the temperature detection data are initialized to obtain the initialized eddy current detection data and the initialized temperature detection data.

[0020] By adopting the above technical scheme, since the training samples of the neural network model are also processed by the initial detection data of the eddy current sensor under no load and the initial detection data of the temperature sensor, the eddy current detection data and the temperature detection data are initialized based on the initial detection data of the eddy current sensor under no load and the initial detection data of the temperature sensor, so that the current actual detection data can correspond to the training data of the neural network model, avoiding errors in the model prediction results due to different data formats and ensuring the accuracy of the detection.

[0021] In a preferred example, the present application can be further configured as follows: the initialization eddy current detection data and the initialization temperature detection data are input into a preset neural network model, and based on the neural network model, a number of model output values ​​are obtained, specifically including: Based on the sensor sequence, one initialization eddy current detection data and the corresponding initialization temperature detection data are sequentially excluded, and the remaining initialization eddy current detection data and initialization temperature detection data are combined to generate a plurality of detection data groups corresponding to the sensor sequence; Several detection data groups are input into a preset neural network model, and several model output values ​​are obtained based on the neural network model.

[0022] By adopting the above technical scheme, since the training of the neural network model is through the random truncation method, a group of eddy current sensor and temperature sensor data is randomly frozen in the data obtained by each test under the calibration test conditions before training, so as to improve the anti-interference ability of the neural network model. Therefore, similarly, in order to minimize the interference that may exist in the initialization eddy current detection data and the corresponding initialization temperature detection data, one initialization eddy current detection data and the corresponding initialization temperature detection data in the actual detected data are frozen, and each initialization eddy current detection data and the corresponding initialization temperature detection data are frozen in sequence according to the sensor sequence, and multiple groups of detection data groups after freezing part of the data are obtained. The number of detection data groups corresponds to the number of sensor groups formed by the eddy current sensors and the temperature sensors. Through the analysis of the multiple detection data groups by the neural network model, multiple model output values ​​are obtained, thereby ensuring the correspondence between the detection data and the model analysis method, thereby ensuring the accuracy of the detection.

[0023] In a preferred example, the present application can be further configured as follows: the calculation is performed in combination with a plurality of the model output values ​​to obtain the sensor test result, specifically including: Calculate the average value and the standard deviation based on the output values ​​of the model to obtain the average value and the standard deviation; The sensor test results include average value calculation results and standard deviation calculation results By adopting the above technical solution, after obtaining multiple model output values, the average value and standard deviation calculation are performed in combination with each model output value to obtain the final test result, thereby achieving high-accuracy measurement of the mechanical state of the pin shaft.

[0024] In summary, the present application includes at least one of the following beneficial technical effects: 1. The bridge hanger pin sensor for health monitoring includes a pin body and several sensors. The sensors include several eddy current sensors for detecting deformation displacement and corresponding several temperature sensors for detecting temperature. These sensors are connected to the pin through several annular grooves provided on the pin body, so that the mechanical state of the pin can be measured more accurately and directly, reducing the interference of other factors on the sensor. Moreover, since the sensor is installed in the annular groove of the pin and integrated with the pin, the probability of sensor damage or falling off is reduced, and the service life of the sensor is extended, so that the sensor can monitor the mechanical state of the pin of the bridge sling and rod for a long time. In addition, the eddy current sensor is mainly used for non-contact measurement of the displacement of the pin. When the pin vibrates or deforms during operation, the eddy current sensor can accurately capture its displacement change. This displacement change is an important manifestation of the mechanical state of the pin and can reflect the pin. Whether the shaft is subjected to excessive external force and whether there are problems such as wear or fatigue, and the temperature sensor can measure the temperature of the pin in real time. The temperature change is closely related to the mechanical state of the pin. For example, when the pin is subjected to excessive friction or load, its temperature may rise, thereby indirectly reflecting the force condition, friction state and potential damage risk of the pin. There is a more prominent correlation between the monitoring data of the eddy current sensor and the temperature sensor. For example, when the pin is subjected to excessive friction, the monitoring data of the eddy current sensor and the temperature sensor will rise; therefore, the eddy current sensor is used in combination with the temperature sensor to improve the clarity of the monitoring of the bridge hanger pin. In addition, the mechanical state of the pin can be more comprehensively understood by comprehensively analyzing the displacement and temperature information, which improves the accuracy and reliability of the measurement, and can also promptly discover possible problems with the pin, providing an important basis for preventing and solving pin failures. 2. Obtain the eddy current detection data of each eddy current sensor and the temperature detection data of the corresponding thermocouple sensor, and perform initialization processing on the eddy current detection data of each eddy current sensor and the temperature detection data of the corresponding thermocouple sensor. The initialization processing refers to the calibration processing of the eddy current detection data and the temperature detection data. For example, the eddy current data and the temperature detection data measured when the pin shaft is in a stationary or known state and stable without the influence of external heat sources are subtracted from the eddy current detection data and the temperature detection data obtained by the detection, so as to perform initialization processing to ensure the accuracy of the detection. The preset neural network model refers to A model for analyzing eddy current detection data and temperature detection data to determine the bearing force value of the pin shaft, for example, by analyzing the correlation between the eddy current detection data and the temperature detection data and the mechanical state of the pin shaft through a neural network algorithm. In addition, since the eddy current detection data and the temperature detection data include detection data of multiple groups of eddy current sensors and temperature sensors in multiple annular grooves on the pin shaft, different positions on the pin shaft may affect each other. Therefore, the neural network model performs multiple analyses, inputs multiple results, and then calculates the average value of multiple output values ​​of the neural network model to obtain the final test result, thereby improving the accuracy of the test; 3. The neural network model is the main means of analyzing the mechanical state of the pin. The prediction accuracy of the neural network model determines the accuracy of the test results. Therefore, it is necessary to improve the robustness of the neural network model. Based on this, the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load are obtained. The initial detection data is the recorded reading of each sensor when it is factory set under no load. The initial detection data is used as the value of the initialization processing. Since the initial detection data is recorded when it is factory set under no load, it directly reflects the initial state of the sensor on the pin, making the initial detection data more accurate. It represents the actual initial condition of the sensor on the pin shaft. Compared with the commonly used method of taking the detection data under standard conditions as the value for initialization, it can avoid introducing errors caused by the difference between standard conditions and actual working conditions. In this way, the accuracy of neural network training is improved by improving the accuracy of training samples. In addition, each training data set is sorted according to the calibration test sequence of each training data set, and the calibration detection data of each eddy current sensor and the corresponding calibration detection data of the temperature sensor in each training data set are sorted according to the preset sensor sequence, so as to facilitate finding abnormal data during neural network training. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the annular groove of the bridge hanger pin sensor used for health monitoring in an embodiment of the present application.

[0026] Figure 2It is a schematic diagram of an eddy current sensor and a temperature sensor of a bridge hanger pin sensor used for health monitoring in an embodiment of the present application.

[0027] Figure 3 It is a flowchart of an implementation method of using a bridge hanger pin sensor for health monitoring in an embodiment of the present application.

[0028] Figure 4 It is a flow chart of the implementation method of obtaining the neural network model in the embodiment of the present application.

[0029] Figure 5 It is a schematic diagram of the training method of the neural network model in the embodiment of the present application.

[0030] Figure 6 It is a flow chart for implementing S03 of the method for using the bridge hanger pin sensor for health monitoring in an embodiment of the present application.

[0031] Figure 7 It is a flow chart for implementing S20 of the method for using the bridge hanger pin sensor for health monitoring in an embodiment of the present application.

[0032] Figure 8 It is a flow chart for implementing S30 of the method for using the bridge hanger pin sensor for health monitoring in an embodiment of the present application.

[0033] Fig. 9 It is a flow chart for implementing S40 of the method for using the bridge hanger pin sensor for health monitoring in an embodiment of the present application.

[0034] Fig.10 It is a principle block diagram of a device for using a bridge hanger pin sensor for health monitoring in an embodiment of the present application. DETAILED DESCRIPTION

[0035] The following is combined with Figure 1-10 This application is described in further detail.

[0036] In one embodiment, if Figure 1 and Figure 2 As shown, the present application discloses a bridge hanger pin sensor for health monitoring, which includes: a pin body, a plurality of eddy current sensors for detecting deformation displacement and a corresponding plurality of temperature sensors for detecting temperature, and the pin body is provided with a plurality of annular grooves. In this embodiment, the pin body is a cylinder, and the annular groove is arranged around the outer surface of the pin body, and the probe of the eddy current sensor and the temperature sensor are arranged in the annular groove. Preferably, the eddy current sensor can also be used to measure parameters such as vibration, speed and density of the pin, and the number of annular grooves is preferably set to two.

[0037] Each eddy current sensor forms a sensor group with a corresponding temperature sensor, and multiple eddy current sensors form multiple sensor groups with multiple corresponding temperature sensors. Each sensor group includes an eddy current sensor and a temperature sensor, and the eddy current sensor and temperature sensor in the same sensor group are integrated. In this embodiment, the integrated setting of the eddy current sensor and the temperature sensor refers to merging or integrating these two sensors into the same system or device to realize the function of implementing eddy current measurement and temperature measurement using one system or device. In this embodiment, the pin shaft body is also configured or connected with a data transmission interface for transmitting the detection data of the eddy current sensor and the temperature sensor to an external monitoring system. The measurement range, accuracy and output signal type (such as voltage, current or digital signal) of the eddy current sensor and the temperature sensor meet the predetermined standard requirements. The eddy current sensor and the temperature sensor are connected to the same external monitoring system or controller. In addition, the electromagnetic compatibility of the eddy current sensor and the temperature sensor in the same sensor group meets the predetermined standard requirements.

[0038] Combining the detection data of eddy current sensors and temperature sensors can infer the force value of the pin, because eddy current sensors and temperature sensors are complementary in measurement, and the information they provide can jointly reflect the mechanical properties of the pin. Specifically, eddy current sensors are mainly used for non-contact measurement of the displacement of the pin. When the pin vibrates or deforms during operation, the eddy current sensor can accurately capture its displacement change. This displacement change is an important manifestation of the mechanical state of the pin, and can reflect whether the pin is subjected to excessive external force, wear or fatigue, etc. The temperature sensor can measure the temperature of the pin in real time. The temperature change is closely related to the mechanical state of the pin. For example, when the pin is subjected to excessive friction or load, its temperature may rise. This temperature change can indirectly reflect the force condition, friction state and potential damage risk of the pin. Therefore, the correlation between displacement change and temperature can be determined, for example, by analyzing through machine learning models, or calculating statistics such as correlation coefficients or covariance between displacement and temperature data, to analyze the degree of influence of temperature on pin displacement, and then based on pre-established evaluation indicators and thresholds, such as displacement range, temperature change range, maximum allowable deviation between displacement and temperature, etc., the mechanical state of the pin can be detected; or the changing trends of displacement and temperature data can be analyzed through machine learning models to determine whether the working state of the pin is normal, and whether there are problems such as overload, wear or thermal expansion.

[0039] In one embodiment, the temperature sensor is a thermocouple sensor, and each eddy current sensor is integrated with the corresponding thermocouple sensor and then arranged in the annular groove. Preferably, the probe of the thermocouple sensor and the probe of the eddy current sensor can be integrated together to achieve the coordinated work of the two.

[0040] In one embodiment, a plurality of eddy current sensors and thermocouple sensors are arranged on both sides of the annular groove, the number of eddy current sensors and thermocouple sensors on both sides of the annular groove is the same, and the intervals between the integrated eddy current sensors and thermocouple sensors on the same side of the annular groove are the same. In this embodiment, the two sides of the annular groove refer to the two surfaces facing each other in the length direction of the pin body in the annular groove, that is, when the pin body is placed vertically to the ground in the length direction of the pin body, the annular groove includes two side surfaces and a covering surface surrounding the inner core of the pin body, and the covering surface is located between the upper and lower side surfaces. The eddy current sensors arranged on both sides of the same annular groove are staggered with each other, that is, the sensor groups formed by the eddy current sensors and thermocouple sensors arranged on both sides of the same annular groove are staggered with each other, so that the detection areas of each eddy current sensor and thermocouple sensor on both sides of the same annular groove are different, thereby realizing more accurate and comprehensive monitoring of a larger area of ​​the pin.

[0041] A metal shell is arranged in the annular groove, and the metal shell is used to cover the integrated eddy current sensor and the thermocouple sensor to further reduce the interference effect of external environmental factors on sensor detection, thereby further improving the accuracy of the overall measurement.

[0042] In one embodiment, the eddy current sensors and thermocouple sensors of the same sensor group take readings simultaneously, and the eddy current sensors and thermocouple sensors of each sensor group take readings in turn in a scanning manner, that is, after the eddy current sensors and thermocouple sensors in the same sensor group complete the reading, the eddy current sensors and thermocouple sensors in the next sensor group in sequence are read again. In this embodiment, the order in which the eddy current sensors and thermocouple sensors of each sensor group take readings in turn in a scanning manner is preset. For example, the reading detection order of any sensor group on one side of the annular groove can be set as the first, and the reading detection order of the sensor group adjacent to the first-ranked sensor group on the other side of the annular groove can be set as the second, so that the scanning reading order is set according to the positional relationship.

[0043] In one embodiment, if Figure 3 As shown, the present application discloses a method for using a bridge hanger pin sensor for health monitoring, and the method for using the bridge hanger pin sensor for health monitoring includes: S10: Acquire eddy current detection data and corresponding temperature detection data.

[0044] In this embodiment, the eddy current detection data refers to the detection data actually detected by the eddy current sensor on the pin body. The temperature detection data refers to the detection data actually detected by the temperature sensor on the pin body.

[0045] Specifically, during the use and operation of the bridge hanger pin sensor used for health monitoring, when it is necessary to detect the bearing force value of the bridge hanger pin, the detection data actually obtained by the eddy current sensors, i.e., the eddy current detection data, and the detection data actually obtained by the temperature sensors, i.e., the temperature detection data, are obtained.

[0046] S20: performing initialization processing on the eddy current detection data and the corresponding temperature detection data to obtain the initialization eddy current detection data and the initialization temperature detection data.

[0047] Specifically, the eddy current detection data and the corresponding temperature detection data are initialized. In this embodiment, the eddy current detection data and the corresponding temperature detection data are initialized by subtracting a preset initial value from the eddy current detection data and the corresponding temperature detection data at the same time. The preset initial value may refer to detection data under standard conditions, historical average or median data, or calibration data. Based on this, after the eddy current detection data and the corresponding temperature detection data are initialized, the initialized eddy current detection data and the initialized temperature detection data are obtained.

[0048] S30: Inputting the initialization eddy current detection data and the initialization temperature detection data into a preset neural network model, and obtaining a plurality of model output values ​​based on the neural network model.

[0049] Specifically, the initialization eddy current detection data and the initialization temperature detection data are input into a preset neural network model, which refers to a model for analyzing eddy current detection data and temperature detection data to determine the bearing force value of the pin shaft. The training sample of the neural network model can be the detection data obtained by the actual detection of the eddy current sensor and the temperature sensor on the pin shaft body in the past or the test data under standard test conditions. Therefore, the initialization eddy current detection data and the initialization temperature detection data are analyzed multiple times by the neural network model to obtain multiple model output values, and the model output values ​​represent the bearing force value of the pin shaft. In this embodiment, the initialization eddy current detection data and the initialization temperature detection data include the detection data of multiple groups of eddy current sensors and temperature sensors in multiple annular grooves on the pin shaft. Due to the fact that different positions on the pin shaft may affect each other, resulting in a decrease in the accuracy of the detection data, the aging of some sensors, resulting in a decrease in the accuracy of the detection data, and the lack of robustness of the neural network model, the accuracy of the output value of the model may be reduced. Therefore, multiple analyses are performed based on the neural network model to obtain multiple model output values, so as to facilitate the judgment of whether these factors exist according to multiple model output values, thereby improving the accuracy of the test results.

[0050] S40: Calculate and combine several model output values ​​to obtain sensor test results.

[0051] Specifically, multiple model output values ​​output by the neural network model are analyzed. For example, it is calculated whether the difference between every two model output values ​​is greater than a preset deviation threshold. If it is judged to be yes, the larger model output value is removed, and then the average value of the remaining model output values ​​is calculated to obtain the final sensor test result representing the pin bearing force value; if it is judged to be no, the average value of all model output values ​​can be directly calculated to obtain the sensor test result.

[0052] In one embodiment, if Figure 4 As shown, the neural network model is obtained in the following way: S01: Obtain initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, and obtain several groups of training data sets under calibration test conditions, wherein each group of training data sets includes calibration detection data of each eddy current sensor, calibration detection data of each corresponding temperature sensor and corresponding pin bearing force value, and sort each group of training data sets according to the calibration test sequence of each group of training data sets, and sort the calibration detection data of each eddy current sensor and the calibration detection data of the corresponding temperature sensor in each group of training data sets according to a preset sensor sequence.

[0053] Specifically, the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load are obtained, and the initial detection data are the recorded readings of each sensor when it is factory set under no load, and several groups of training data sets are obtained under calibration test conditions, that is, multiple calibration tests are carried out, and the training data sets of each calibration test are obtained, wherein each group of training data sets includes the calibration detection data of each eddy current sensor, the corresponding calibration detection data of each temperature sensor and the corresponding pin bearing force value, and each group of training data sets is sorted according to the calibration test sequence of each group of training data sets, and the calibration detection data of each eddy current sensor and the calibration detection data of the corresponding temperature sensor in each group of training data sets are sorted according to the preset sensor sequence, that is, the calibration detection data of any eddy current sensor in the training data set and the calibration detection data of the corresponding temperature sensor are sorted in the same order. For example, the data in each training dataset can be represented as follows: XF j , wx ij , Tx ij ,....,i=1,2,3,....,n,j=1,2,3,....,m, Where j is the number of the calibration test sequence, XF j is the bearing force value of the pin during the jth test; wx ij is the reading of the i-th eddy current sensor in the j-th test; Tx ij is the reading of the i-th temperature sensor during the j-th test, and the serial number i is the preset sensor order; In addition, the eddy current detection data actually detected by the eddy current sensor on the pin body and the temperature detection data actually detected by the temperature sensor on the pin body can be expressed as follows: w i , T i ,....,i=1,2,3,....,n,where,w i is the eddy current detection data obtained by the actual detection of the i-th eddy current sensor, T i It is the temperature detection data actually obtained by the i-th temperature sensor, and the sequence number i is the preset sensor sequence.

[0054] S02: Based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, the calibrated detection data of each eddy current sensor and the calibrated detection data of each temperature sensor in each training data set are initialized.

[0055] Specifically, based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, the calibration detection data of each eddy current sensor and the calibration detection data of each temperature sensor in each group of training data sets are initialized. In this embodiment, the initialization method refers to subtracting the initial detection data of the eddy current sensor from the calibration detection data of each eddy current sensor in each group of training data sets to obtain the first initialization data, subtracting the initial detection data of the temperature sensor from the calibration detection data of each temperature sensor to obtain the second initialization data, and then dividing the first initialization data by the second initialization data to obtain the third initialization data. In this way, each group of training data sets finally obtained after the initialization processing includes the third initialization data and the corresponding pin bearing force value.

[0056] For example, the recorded readings of each sensor at the factory setting under no load can be expressed as follows: w0 i , T0 i ,....,i=1,2,3,....,n, Among them, w0 i is the initial detection data of the eddy current sensor, T0 i is the initial detection data of the temperature sensor; The input training data set of the neural network model can be represented as follows: XF j ,(wx ij -w0 i ) / (Tx ij -T0 i ), ...., i = 1, 2, 3, ...., n, j = 1, 2, 3, ...., m, where (wx ij -w0 i ) is the first initialization data, (Tx ij -T0 i ) is the second initialization data, (wx ij -w0 i ) / (Tx ij -T0 i ) is the third initialization data.

[0057] S03: Perform neural network training based on each set of training data sets to generate a neural network model.

[0058] Specifically, each set of training data sets is used to perform neural network training, and after the training is completed, a neural network model is obtained.

[0059] In this embodiment, if Figure 5 As shown, the training method of the neural network model is carried out by completing the training data set after the initialization processing.

[0060] In one embodiment, if Figure 6 As shown, in step S03, neural network training is performed based on each set of training data sets to generate a neural network model, which specifically includes: S031: Each time when the neural network training is performed based on each set of training data sets, the calibration detection data of any eddy current sensor and the calibration detection data of the corresponding temperature sensor in the training data sets are randomly selected.

[0061] Specifically, in order to improve the robustness and anti-interference ability of the neural network model, the neural network model needs to be improved. Therefore, based on the random truncation method, each time the neural network training is performed based on each set of training data sets, the calibration detection data of any eddy current sensor in the training data set and the calibration detection data of the corresponding temperature sensor are randomly selected.

[0062] For example, when the jth training data set is input, the input training data set of the neural network model includes XF j ,(wx ij -w0 i ) / (Tx ij -T0 i ), ...., i = 1, 2, 3, ...., n, j = 1, 2, 3, ...., m, the random number randomly selected from i = 1, 2, 3, ...., n is k, then the randomly selected data is (wx kj -w0 k ) / (Tx kj -T0 k ).

[0063] S032: Excluding the randomly selected calibration test data of the eddy current sensor and the corresponding calibration test data of the temperature sensor from the training data set of the current neural network training, performing neural network training based on the remaining data in the training data set to generate a neural network model.

[0064] Specifically, after completing the random selection of data, the calibration detection data of the randomly selected eddy current sensor and the corresponding calibration detection data of the temperature sensor are excluded from the training data set of the current neural network training, and then the neural network training is performed based on the remaining data in the training data set to generate a neural network model.

[0065] For example, the remaining data in the training dataset includes XF j ,(wx ij -w0 i ) / (Tx ij -T0 i ),....,i=1,2,3,....,n,j=1,2,3,....,m,i≠k.

[0066] In one embodiment, if Figure 7 As shown, in step S20, the eddy current detection data and the temperature detection data are initialized to obtain the initialized eddy current detection data and the initialized temperature detection data, which specifically includes: S21: Based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, the eddy current detection data and the temperature detection data are initialized to obtain the initialized eddy current detection data and the initialized temperature detection data.

[0067] Specifically, since the training samples of the neural network model are also processed by the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, similarly, based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, the eddy current detection data and the temperature detection data are initialized so that the current actual detection data can correspond to the training data of the neural network model, thereby obtaining the initialized eddy current detection data and the initialized temperature detection data.

[0068] For example, the recorded readings of each sensor at the factory setting under no load can be expressed as follows: w0 i , T0 i ,....,i=1,2,3,....,n, Among them, w0 i is the initial detection data of the eddy current sensor, T0 i is the initial detection data of the temperature sensor; Eddy current test data and temperature test data can be expressed as follows: xx i , cT i ,....,i=1,2,3,....,n, Among them, cx i is the eddy current testing data, cT i It is the temperature detection data; The initialization eddy current detection data and the initialization temperature detection data can be expressed as follows: xx i -w0 i , cT i -w0 i ,....,i=1,2,3,....,n,where cx i -w0 i is the initialization eddy current detection data, cT i -w0 i To initialize the temperature detection data.

[0069] Furthermore, since it is necessary to make the current actual detection data correspond to the training data of the neural network model, the initialized eddy current detection data and the initialized temperature detection data are further processed, that is, the initialized eddy current detection data is divided by the initialized temperature detection data, so that the data input into the neural network model is the data after the initialized eddy current detection data is divided by the initialized temperature detection data.

[0070] For example, the data obtained by dividing the initialization eddy current detection data by the initialization temperature detection data can be expressed as follows: (cx i -w0 i ) / (cT i -w0 i ), ...., i=1, 2, 3, ...., n.

[0071] In one embodiment, if Figure 8 As shown, in step S30, the initialized eddy current detection data and the initialized temperature detection data are input into a preset neural network model, and based on the neural network model, a number of model output values ​​are obtained, specifically including: S31: Based on the sensor sequence, one initialized eddy current detection data and the corresponding initialized temperature detection data are excluded in turn, and the remaining initialized eddy current detection data and the initialized temperature detection data are combined to generate a number of detection data groups corresponding to the sensor sequence.

[0072] Specifically, since the training of the neural network model is carried out through the random truncation method, a group of eddy current sensor and temperature sensor data is randomly frozen in the data obtained from each test under the calibration test conditions before training, so as to improve the anti-interference ability of the neural network model. Therefore, similarly, in order to minimize the possible interference in the initialized eddy current detection data and the corresponding initialized temperature detection data, based on the sensor sequence, the initialized eddy current detection data and the corresponding initialized temperature detection data with the first sequence number are excluded, and the remaining initialized eddy current detection data and the initialization temperature detection data are used as the first detection data group, and then the initialized eddy current detection data and the corresponding initialization temperature detection data with the second sequence number are excluded, and the remaining initialized eddy current detection data and the initialization temperature detection data are used as the second detection data group, so as to obtain multiple detection data groups by excluding based on the sensor sequence.

[0073] For example, the data in the detection data set can be represented as follows: The first test data set: (cx i -w0 i ) / (cT i -w0 i), ...., i=2, 3, ...., n; The second test data set: (cx i -w0 i ) / (cT i -w0 i ), ...., i=1, 3, ...., n; The nth detection data set: (cx i -w0 i ) / (cT i -w0 i ), ...., i=1, 2, 3, ...., n-1.

[0074] S32: Inputting a plurality of detection data groups into a preset neural network model, and obtaining a plurality of model output values ​​based on the neural network model.

[0075] Specifically, each detection data group is input into a preset neural network model, and analysis is performed based on the neural network model to obtain a number of model output values ​​corresponding to each detection data group.

[0076] For example, the model output value can be expressed as follows: xx i ,i=1,2,3,....,n.

[0077] In one embodiment, if Fig. 9 As shown, in step S40, the sensor test results include the average value calculation result and the standard deviation calculation result, which are calculated in combination with several model output values ​​to obtain the sensor test results, specifically including: S41: Calculate the average value and the standard deviation based on the output values ​​of several models to obtain the average value calculation result and the standard deviation calculation result.

[0078] Specifically, the average value calculation and the standard deviation calculation are performed in combination with several model output values ​​to obtain the average value calculation result and the standard deviation calculation result, and the average value calculation result and the standard deviation calculation result are both sensor test results.

[0079] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0080] In one embodiment, a device for using a bridge hanger pin sensor for health monitoring is provided, and the device for using the bridge hanger pin sensor for health monitoring corresponds to the method for using the bridge hanger pin sensor for health monitoring in the above embodiment. Fig.10As shown, the device for using the bridge hanger pin sensor for health monitoring includes a data acquisition module, an initialization processing module, a model output module and a statistical analysis module. The detailed description of each functional module is as follows: A data acquisition module, used to acquire eddy current detection data and corresponding temperature detection data; An initialization processing module is used to perform initialization processing on the eddy current detection data and the corresponding temperature detection data to obtain the initialization eddy current detection data and the initialization temperature detection data; A model output module, used to input the initialization eddy current detection data and the initialization temperature detection data into a preset neural network model, and obtain a number of model output values ​​based on the neural network model; The statistical analysis module is used to combine several model output values ​​for calculation and obtain sensor test results.

[0081] Optionally, the device for using the bridge hanger pin sensor for health monitoring also includes: A data collection module, used to obtain the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, and to obtain several groups of training data sets under calibration test conditions, wherein each group of training data sets includes the calibration detection data of each eddy current sensor, the corresponding calibration detection data of each temperature sensor, and the corresponding pin bearing force value, and to sort each group of training data sets according to the calibration test sequence of each group of training data sets, and to sort the calibration detection data of each eddy current sensor and the corresponding calibration detection data of the temperature sensor in each group of training data sets according to a preset sensor sequence; An initialization module, used for initializing the calibration detection data of each eddy current sensor and the calibration detection data of each temperature sensor in each training data set based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load; The model generation module is used to perform neural network training based on each set of training data sets and generate a neural network model.

[0082] Optionally, the model generation module includes: A random selection submodule, used for randomly selecting calibration detection data of any eddy current sensor and the corresponding calibration detection data of the temperature sensor in the training data set each time the neural network training is performed based on each set of training data sets; The training submodule is used to exclude the randomly selected calibration test data of the eddy current sensor and the corresponding calibration test data of the temperature sensor in the training data set of the current neural network training, perform neural network training based on the remaining data in the training data set, and generate a neural network model.

[0083] Optionally, the initialization processing module includes: The initialization processing submodule is used to perform initialization processing on the eddy current detection data and the temperature detection data based on the initial detection data of the eddy current sensor under no load and the initial detection data of the temperature sensor, so as to obtain the initialization eddy current detection data and the initialization temperature detection data.

[0084] Optionally, the model output module includes: An exclusion submodule, for excluding an initialization eddy current detection data and a corresponding initialization temperature detection data in turn based on the sensor sequence, and collecting the remaining initialization eddy current detection data and initialization temperature detection data to generate a plurality of detection data groups corresponding to the sensor sequence; The model analysis submodule is used to input a number of detection data groups into a preset neural network model, and obtain a number of model output values ​​based on the neural network model.

[0085] Optionally, the sensor test results include average value calculation results and standard deviation calculation results, and the statistical analysis module includes: The statistical analysis submodule is used to calculate the average value and the standard deviation by combining several model output values ​​to obtain the average value calculation results and the standard deviation calculation results.

[0086] For specific limitations on the use device of the bridge hanger pin sensor for health monitoring, please refer to the limitations on the use method of the bridge hanger pin sensor for health monitoring mentioned above, which will not be repeated here. Each module in the above-mentioned use device of the bridge hanger pin sensor for health monitoring can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0087] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0088] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A bridge hanger pin sensor for health monitoring, characterized in that: The bridge hanger pin sensor for health monitoring includes: a pin body, a plurality of eddy current sensors for detecting deformation displacement and a corresponding plurality of temperature sensors for detecting temperature. The pin body is provided with a plurality of annular grooves. The eddy current sensors and the temperature sensors are integrated in the annular grooves. Each eddy current sensor forms a sensor group with the corresponding temperature sensor. The detection data of the eddy current sensor and the temperature sensor are combined to infer the pin force value.

2. The bridge hanger pin sensor for health monitoring according to claim 1, characterized in that: The temperature sensor is a thermocouple sensor, and each of the eddy current sensors is integrated with the corresponding thermocouple sensor and arranged in the annular groove.

3. The bridge hanger pin sensor for health monitoring according to claim 2, characterized in that: A plurality of eddy current sensors and thermocouple sensors are arranged on both sides of the annular groove, and the eddy current sensors arranged on both sides of the same annular groove are staggered with each other. A metal shell is arranged in the annular groove, and the metal shell is used to cover the eddy current sensors and the thermocouple sensors.

4. The bridge hanger pin sensor for health monitoring according to claim 2, characterized in that: The eddy current sensors and the thermocouple sensors in the same sensor group take readings simultaneously, and the eddy current sensors and the thermocouple sensors in each sensor group take readings in sequence in a scanning manner.

5. A method for using a bridge hanger pin sensor for health monitoring, characterized in that: The bridge hanger pin sensor for health monitoring according to claim 1 is used, and the method for using the bridge hanger pin sensor for health monitoring includes: Obtain eddy current detection data and corresponding temperature detection data; Performing initialization processing on the eddy current detection data and the corresponding temperature detection data to obtain initialization eddy current detection data and initialization temperature detection data; Inputting the initialization eddy current detection data and the initialization temperature detection data into a preset neural network model, and obtaining a plurality of model output values ​​based on the neural network model; The sensor test result is obtained by combining several of the model output values ​​for calculation.

6. The method for using the bridge hanger pin sensor for health monitoring according to claim 5, characterized in that: The neural network model is obtained in the following way: Acquire initial detection data of the eddy current sensor and initial detection data of the temperature sensor under no load, and acquire several groups of training data sets under calibration test conditions, wherein each group of the training data sets includes calibration detection data of each eddy current sensor, calibration detection data of each corresponding temperature sensor, and corresponding pin bearing force value, and sort each group of the training data sets according to the calibration test sequence of each group of the training data sets, and sort the calibration detection data of each eddy current sensor and the calibration detection data of the corresponding temperature sensor in each group of the training data sets according to a preset sensor sequence; Based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, initialization processing is performed on the calibrated detection data of each eddy current sensor and the calibrated detection data of each temperature sensor in each group of the training data set; A neural network training is performed based on each set of the training data sets to generate a neural network model.

7. The method for using the bridge hanger pin sensor for health monitoring according to claim 6, characterized in that: The performing neural network training based on each set of the training data sets to generate a neural network model specifically includes: Each time when the neural network training is performed based on each group of the training data sets, the calibration detection data of any one of the eddy current sensors and the calibration detection data of the corresponding temperature sensor in the training data sets are randomly selected; The calibration detection data of the randomly selected eddy current sensor and the calibration detection data of the corresponding temperature sensor are excluded from the training data set of the current neural network training, and the neural network training is performed based on the remaining data in the training data set to generate a neural network model.

8. The method for using the bridge hanger pin sensor for health monitoring according to claim 6, characterized in that: The initialization processing of the eddy current detection data and the temperature detection data to obtain the initialization eddy current detection data and the initialization temperature detection data specifically includes: Based on the initial detection data of the eddy current sensor and the initial detection data of the temperature sensor under no load, the eddy current detection data and the temperature detection data are initialized to obtain the initialized eddy current detection data and the initialized temperature detection data.

9. The method for using the bridge hanger pin sensor for health monitoring according to claim 6, characterized in that: The initialization eddy current detection data and the initialization temperature detection data are input into a preset neural network model, and a plurality of model output values ​​are obtained based on the neural network model, specifically including: Based on the sensor sequence, one initialization eddy current detection data and the corresponding initialization temperature detection data are sequentially excluded, and the remaining initialization eddy current detection data and initialization temperature detection data are combined to generate a plurality of detection data groups corresponding to the sensor sequence; Several detection data groups are input into a preset neural network model, and several model output values ​​are obtained based on the neural network model.

10. The method for using the bridge hanger pin sensor for health monitoring according to claim 5, characterized in that: The calculation is performed in combination with the output values ​​of the model to obtain the sensor test result, specifically including: Calculate the average value and the standard deviation based on the output values ​​of the model to obtain the average value and the standard deviation; The sensor test results include an average value calculation result and a standard deviation calculation result.

Images