GIL through pipe multi-parameter monitoring intelligent sensor device

Abstract

The invention relates to a GIL through pipe multi-parameter monitoring intelligent sensor device. Comprising a shell, a microcontroller, an A / D conversion module, a motion posture acquisition sensor, a temperature and humidity sensor for detecting environment temperature and humidity, an electrochemical sensor for detecting abnormal gas in the environment, a piezoelectric sensor for detecting stress of a GIL through pipe, a power supply module for supplying power and a communication module for transmitting sensing data, the microcontroller is respectively connected with the motion posture acquisition sensor, the temperature and humidity sensor, the electrochemical sensor and the piezoelectric sensor through the A / D conversion module; an AI chip is carried on the microcontroller, and a sensor detection parameter adaptive adjustment module and an abnormity identification module are integrated in the AI chip. Compared with the prior art, multi-parameter monitoring of the GIL through pipe can be achieved, and the detection result is more accurate and reliable.

Description

Technical Field

[0001] The present invention relates to the technical field of power sensors, and in particular to an intelligent sensor device for multi-parameter monitoring of GIL through-tubes. Background Art

[0002] With the rapid development of urban social economy, the contradiction between the increasing demand for power resources and the aging of existing facilities has become prominent. Some old and aging substation equipment has frequent defects, and the equipment capacity is difficult to meet the load growth in the power supply area. Moreover, there is no redundant land for the expansion and capacity increase of substations in the city center or sub-center, so the power grid can only be upgraded and transformed on the original site. However, the new and old equipment and facilities coexist in large substations, and the operating environment is complex. The potential safety hazards caused by the mutual influence during the near-electric construction and transformation are huge, resulting in an increased risk during the transformation process.

[0003] The GIL through-tubes in substations cover a wide range, large area, and have various layout methods. During the transformation process, they may be affected by external factors such as mechanical vibration, temperature change, or accidental accidents, resulting in deformation or damage. Conducting multi-parameter comprehensive monitoring of the new and old GIL through-tubes during the expansion and renovation of large substations is crucial for the safe and reliable operation of the power system.

[0004] Currently, existing research has used multi-sensor integrated devices for substation monitoring. However, there is no device that integrates parameters such as temperature and humidity, dynamic response, and gas concentration into a single sensor for multi-parameter monitoring of the new and old GIL through-tubes during the expansion and renovation of large substations, and the sensor lacks intelligent calculation and cannot achieve self-operation at the sensor end and adaptive adjustment of the sensor monitoring parameters. Summary of the Invention

[0005] The purpose of the present invention is to overcome the above-mentioned defects existing in the prior art and provide an intelligent sensor device for multi-parameter monitoring of GIL through-tubes, which integrates parameters such as temperature and humidity, dynamic response, and gas concentration into a single sensor for multi-parameter monitoring, integrates an AI chip on a microcontroller, realizes the adaptive adjustment of the sensor monitoring parameters through the built-in sensor detection parameter adaptive adjustment module, and timely discovers the deformation and damage of the GIL through-tubes and environmental abnormalities through the built-in anomaly recognition module, with higher reliability.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] According to the first aspect of the present invention, there is provided a GIL through-tube multi-parameter monitoring intelligent sensor device, characterized in that it includes a housing, a microcontroller, an A / D conversion module, a motion attitude acquisition sensor, a temperature and humidity sensor for detecting the ambient temperature and humidity, an electrochemical sensor for detecting abnormal gases in the environment, a piezoelectric sensor for detecting the force on the GIL through-tube, a power supply module for power supply, and a communication module for transmitting sensing data; the microcontroller is respectively connected to the motion attitude acquisition sensor, the temperature and humidity sensor, the electrochemical sensor, and the piezoelectric sensor through the A / D conversion module;

[0008] An AI chip is mounted on the microcontroller, and a sensor detection parameter adaptive adjustment module and an anomaly recognition module are integrated in the AI chip.

[0009] Preferably, a PID control model is integrated in the sensor detection parameter adaptive adjustment module for dynamically adjusting the sensor detection parameters according to the error output by the sensor, and the sensor detection parameters include measurement frequency, range, and sensitivity parameters.

[0010] Preferably, the PID control parameters in the PID control model are dynamically adjusted, and the dynamic adjustment expression is specifically:

[0011] ΔV t (t) = α·ΔV a +(1 - α)·ΔV t0

[0012] K p (t) = K p0 ·(1 + βR(t))

[0013] K d (t) = K d0 ·(1 + βR(t))

[0014]

[0015] In the formula: ΔV t is the adjusted rate of change; ΔV a is the average rate of change over a period of time; ΔV t0 is the current target rate of change; ΔV t is the adjusted target rate of change; α is the smoothing coefficient; K p is the adjusted proportional gain coefficient, K d is the adjusted derivative gain coefficient; K p0 is the initial proportional gain coefficient, K d0 is the initial derivative gain coefficient; β is the adjustment sensitivity coefficient; R is the rate of change influence factor; ΔV max is the maximum rate of change over a period of time.

[0016] Preferably, the abnormal recognition process of the abnormal recognition module is specifically as follows:

[0017] 1) Data preprocessing: including data denoising, normalization, and time window segmentation;

[0018] 2) Abnormal detection: Use the LSTM model to learn the time correlation of normal data, input the sensor time series data, and output the reconstruction error or prediction error. When the error is greater than the set threshold, determine that the current time window is an abnormal time window;

[0019] 3) Abnormal classification: Input the abnormal time window obtained by abnormal detection into the pre-trained random forest model, and output the abnormal category to which the current abnormal time window belongs. The abnormal categories include equipment failure, environmental fluctuation, and data acquisition error.

[0020] Preferably, perform incremental learning and dynamic expansion on the random forest model, and perform incremental update using the new data by adding category output nodes.

[0021] Preferably, a three-axis gyroscope, a three-axis accelerometer, and a three-axis magnetometer are integrated in the motion posture acquisition sensor for monitoring the deformation and position of the GIL pipe.

[0022] Preferably, the sensing elements in the device are arranged in a vertical layered manner, specifically as follows:

[0023] The motion posture module and the piezoelectric sensor are arranged at the bottom layer, and the temperature and humidity sensor and the electrochemical sensor are arranged at the top layer; among them, the temperature and humidity sensor and the electrochemical sensor are separated by a partition board.

[0024] Preferably, a bridge module is also mounted on the A / D conversion module. The input ends of the bridge module are respectively connected to the motion posture acquisition sensor, the temperature and humidity sensor, the electrochemical sensor, and the piezoelectric sensor, and the output end of the bridge module is connected to the A / D conversion module.

[0025] Preferably, the electrochemical sensor is a dual-channel electrochemical gas sensor with temperature compensation function, and / or the piezoelectric sensor is a piezoelectric ceramic sensor.

[0026] Preferably, the device further includes a photovoltaic module connected to the microcontroller for charging the power module.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] (1) The present invention integrates an AI chip on a microcontroller, enabling the entire multi-parameter monitoring intelligent sensor device to endow the sensor itself with intelligent computing capabilities, capable of preprocessing monitoring data, supporting the deployment of neural network models, and having a built-in sensor detection parameter adaptive adjustment module to achieve the adaptive adjustment of sensor monitoring parameters. The built-in anomaly recognition module can timely detect the deformation and damage of the GIL pipe and environmental anomalies, with higher reliability.

[0029] (2) In the sensor detection parameter adaptive adjustment module integrated with a PID control model in the present invention, sensor detection parameters such as measurement frequency, range, and sensitivity parameters are dynamically adjusted according to the sensor sensing data acquisition situation, with strong environmental response capabilities.

[0030] (3) The anomaly recognition module of the present invention uses an LSTM model to learn the temporal correlation of normal data, makes anomaly judgments according to the sensor time series error threshold, and uses a random forest model to judge the categories of abnormal data. The abnormal category output structure is more accurate and reliable. At the same time, in actual scenarios, new abnormal categories may appear, and incremental learning is performed on the random forest model for dynamic expansion. By adding category output nodes and using new data for incremental updates, the scenario applicability is strong.

[0031] (4) The present invention integrates sensing modules such as motion posture, temperature and humidity, gas concentration, and stress on a microcontroller, enabling multi-parameter comprehensive monitoring of new and old GIL pipes during the reconstruction and expansion process of large substations, and ensuring the safe and reliable operation of the power system.

[0032] (5) The motion posture acquisition sensor in the present invention integrates a three-axis gyroscope, a three-axis accelerometer, and a three-axis magnetometer, which can more accurately and reliably monitor the deformation and position of the GIL pipe.

[0033] (6) The sensing elements are arranged in a vertical layered manner. The motion posture module and the piezoelectric sensor are arranged at the bottom layer, and the temperature and humidity sensor and the electrochemical sensor are arranged at the top layer, which can effectively reduce the interference between different sensors and improve the measurement accuracy. In addition, the temperature and humidity sensor and the electrochemical sensor are separated by a partition plate to avoid the influence of airflow disturbance on the measurement results of the temperature and humidity sensor, and a closed airflow channel is used to ensure that the gas detected by the electrochemical sensor is not affected by temperature and humidity changes.

[0034] (7) Through the external connection of a photovoltaic module in the present invention, when the output power exceeds the battery demand, it charges the battery, and when the output power is insufficient, it directly powers the microcontroller through the battery module, making the power supply more flexible. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a schematic structural diagram of the GIL pipe multi-parameter monitoring intelligent sensor device of the present invention;

[0036] Figure 2 Schematic diagram of the sensing data transmission process

[0037] Reference numerals in the drawings: 1 - microcontroller; 2 - AI chip; 3 - power supply module; 4 - A / D conversion module; 5 - motion attitude acquisition sensor; 6 - temperature and humidity sensor; 7 - electrochemical sensor; 8 - piezoelectric sensor; 9 - bridge module; 10 - ; 11 - photovoltaic module; 12 - housing. Detailed implementation manners

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] Embodiment 1

[0040] As Figure 1 shown, this embodiment provides a GIL through-tube multi-parameter monitoring intelligent sensor device, including a housing 12, a microcontroller 1, an A / D conversion module 4, a motion attitude acquisition sensor 5, a temperature and humidity sensor 6 for detecting the environmental temperature and humidity, an electrochemical sensor 7 for detecting abnormal gases in the environment, a piezoelectric sensor 8 for detecting the force on the GIL through-tube, a power supply module 3 for power supply, and a communication module 10 for transmitting sensing data; the microcontroller 1 is respectively connected to the motion attitude acquisition sensor 5, the temperature and humidity sensor 6, the electrochemical sensor 7, and the piezoelectric sensor 8 through the A / D conversion module 4.

[0041] In this embodiment, the microcontroller 1 uses a Raspberry Pi 4B, and by integrating with the A / D conversion module 4, the power supply module 3, the motion attitude acquisition sensor 5, the temperature and humidity sensor 6, the electrochemical sensor 7, and the piezoelectric sensor 8, the acquisition, processing, and transmission of sensor data are realized. Further, in this embodiment, a heat dissipation copper foil is added to the surface of the Raspberry Pi 4B board to reduce the local temperature rise of the sensor.

[0042] An AI chip 2 is mounted on the microcontroller 1. The AI chip 2 uses a CSK 6002 chip, and an adaptive adjustment module 201 for sensor detection parameters and an abnormal recognition module 202 are integrated therein for the adaptive adjustment of sensor monitoring parameters and the abnormal recognition of monitoring data.

[0043] Specifically, in this embodiment, a PID control model is integrated in the sensor detection parameter adaptive adjustment module 201, which is used to dynamically adjust the sensor detection parameters according to the error output by the sensor. The sensor detection parameters include measurement frequency, range, and sensitivity parameters. The PID control model includes three links: proportional control (P), integral control (I), and derivative control (D). It dynamically adjusts parameters such as measurement frequency, range, and sensitivity according to the error output by the sensor. Proportional control (P) responds to the error between the sensor measurement value and the control target, integral control (I) responds to the long-term steady-state error, and derivative control (D) responds to the change rate of the error.

[0044] Taking the measurement frequency as an example, when the monitored physical quantity changes greatly, the PID control model increases the sampling frequency to improve the data response speed, and when the monitored physical quantity changes little, it reduces the sampling frequency to reduce the component power consumption and system burden. Suppose the current sampling frequency of the sensor is f0, and the adjusted sampling frequency is f. The PID control formula is:

[0045]

[0046] e(t) = ΔV s -ΔV t

[0047]

[0048] In the formula: e(t) is the difference between the actual change rate ΔV s and the target change rate ΔV t . The target change rate can be considered as the data change rate under normal circumstances. K p , K i , and K d are the proportional, integral, and derivative gain coefficients respectively. V(t) is the measurement value at the current moment, V(t - Δt) is the measurement value at the previous moment, and N is the window size for data smoothing.

[0049] To adapt to different environmental changes, the PID control parameters (K p , K d , and ΔV t ) are dynamically adjusted on the AI chip 2. When the data changes violently, K p and K d are increased to make the controller more sensitive. When the data changes smoothly, K p and K d are decreased to make the controller smoother. The specific dynamic adjustment expression is specifically:

[0050] ΔV t (t) = α·ΔV a +(1 - α)·ΔVt0

[0051] K p (t) = K p0 ·(1 + βR(t))

[0052] K d (t) = K d0 ·(1 + βR(t))

[0053]

[0054] Where: ΔV t is the adjusted rate of change; ΔV a is the average rate of change over a period of time; ΔV t0 is the current target rate of change; ΔV t is the adjusted target rate of change; α is the smoothing coefficient (between 0 and 1); K p is the adjusted proportional gain coefficient, K d is the adjusted derivative gain coefficient; K p0 is the initial proportional gain coefficient, K d0 is the initial derivative gain coefficient; β is the adjustment sensitivity coefficient; R(t) is the rate of change impact factor; ΔV max is the maximum rate of change over a period of time.

[0055] In this embodiment, the anomaly recognition module 202 integrates machine learning methods represented by random forests and deep learning methods represented by LSTM. The specific anomaly recognition process is as follows:

[0056] 1) Data preprocessing: including data denoising, normalization, and time window segmentation;

[0057] 2) Anomaly detection: Use the LSTM model to learn the time correlation of normal data, input the sensor time series data, and output the reconstruction error or prediction error. When the error is greater than the set threshold, determine that the current time window is an abnormal time window;

[0058] 3) Anomaly classification: Input the abnormal time window obtained from anomaly detection into the pre-trained random forest model, and output the abnormal category to which the current abnormal time window belongs. The abnormal categories include equipment failure, environmental fluctuation, and data acquisition error.

[0059] In the actual scenario, new abnormal categories may appear. This implementation also includes incremental learning and dynamic expansion of the random forest model. By adding category output nodes, incremental updates are performed using the new data.

[0060] In this embodiment, the A / D conversion module 4 uses an ADS1115 analog-to-digital conversion module to convert the analog signals output by various sensor elements into digital signals, facilitating the data processing of the microcontroller 1.

[0061] In this embodiment, the communication module 10 uses a Wi-Fi module ESP8266 for data transmission of the sensors. In addition, the sensing data and sensing monitoring data can be uploaded to the cloud through the communication module for further analysis.

[0062] Embodiment 2

[0063] In this embodiment, a vertical layered layout is adopted inside the housing of the intelligent sensor device. The motion attitude acquisition sensor 5 and the piezoelectric sensor 8 are arranged at the bottom layer, and the temperature and humidity sensor 6 and the electrochemical sensor 7 are arranged at the upper layer to reduce the mutual influence between them.

[0064] The piezoelectric sensor 8 uses a piezoelectric ceramic sensor to monitor the force on the GIL pipe. The motion attitude acquisition sensor 5 uses an MPU9250 module, which integrates a three-axis gyroscope, a three-axis accelerometer, and a three-axis magnetometer, to monitor the deformation and position of the GIL pipe.

[0065] In this embodiment, the temperature and humidity sensor 6 uses a DHT11 digital temperature and humidity sensor to monitor the environmental temperature and humidity. The electrochemical sensor 7 uses a CN0396 dual-channel electrochemical gas sensor with temperature compensation function to monitor the toxic gases in the environment. A separator is used to separate the temperature and humidity sensor 6 and the electrochemical sensor 7 to avoid the influence of airflow disturbance on the measurement results of the temperature and humidity sensor 6, and a closed airflow channel is used to ensure that the gases detected by the electrochemical sensor 7 are not affected by temperature and humidity changes.

[0066] Other settings of this embodiment are the same as those of Embodiment 1.

[0067] Embodiment 3

[0068] In this embodiment, the power supply module 3 uses a rechargeable lithium battery, equipped with a TP4056 charging management system to manage the charging process of the battery and prevent overcharging. A DC-DC converter (boost or buck converter) is used to adjust the output voltage according to the voltage of the battery and the working voltage of the microcontroller 2 to ensure the stable operation of the microcontroller 2.

[0069] Such as Figure 2As shown, a photovoltaic module 11 is also externally connected. The photovoltaic module 11 is connected to the microcontroller 2 using a Micro-USB interface, ensuring that the photovoltaic module 11 can be flexibly installed in different positions or environments and charge the lithium battery of the power module 3 in the presence of sunlight. When the output power of the photovoltaic module 11 exceeds the battery's demand, the battery will be charged. When the power of the externally connected photovoltaic module 11 is insufficient, the battery will supply power to the microcontroller 1.

[0070] Other settings in this embodiment are the same as those in Embodiment 1.

[0071] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An intelligent sensor device for multi-parameter monitoring of GIL through-tubes, characterized in that, It includes a housing, a microcontroller, an A / D conversion module, a motion attitude acquisition sensor, a temperature and humidity sensor for detecting the environmental temperature and humidity, an electrochemical sensor for detecting abnormal gases in the environment, a piezoelectric sensor for detecting the force on the GIL pipe, a power supply module for power supply, and a communication module for transmitting sensing data; the microcontroller is connected to the motion attitude acquisition sensor, the temperature and humidity sensor, the electrochemical sensor, and the piezoelectric sensor through the A / D conversion module respectively; An AI chip is mounted on the microcontroller, and a sensor detection parameter adaptive adjustment module and an anomaly recognition module are integrated in the AI chip.

2. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, wherein, A PID control model is integrated in the sensor detection parameter adaptive adjustment module for dynamically adjusting the sensor detection parameters according to the error output by the sensor, and the sensor detection parameters include measurement frequency, range, and sensitivity parameters.

3. The intelligent sensor for multi-parameter monitoring of GIL through-tubes according to claim 2, characterized in that, The PID control parameters in the PID control model are dynamically adjusted, and the dynamic adjustment expression is specifically: ΔV t (t) = α·ΔV a + (1 - α)·ΔV t0 K p K(t) = p0 ·(1 + βR(t)) K d K(t) = d0 ·(1 + βR(t)) Where: ΔV t is the adjusted rate of change; ΔV a is the average rate of change over a period of time; ΔV t0 is the current target rate of change; ΔV t is the adjusted target rate of change; α is the smoothing coefficient; K p is the adjusted proportional gain coefficient, K d is the adjusted derivative gain coefficient; K p0 is the initial proportional gain coefficient, K d0 is the initial derivative gain coefficient; β is the adjustment sensitivity coefficient; R is the rate of change influence factor; ΔV max is the maximum rate of change over a period of time.

4. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, wherein The anomaly recognition process of the anomaly recognition module is specifically: 1) Data preprocessing: including data denoising, normalization, and time window segmentation; 2) Anomaly detection: The LSTM model is used to learn the time correlation of normal data, the sensor time series data is input, and the reconstruction error or prediction error is output. When the error is greater than the set threshold, the current time window is determined as an abnormal time window; 3) Anomaly classification: The abnormal time window obtained by anomaly detection is input into the pre-trained random forest model, and the anomaly category to which the current abnormal time window belongs is output. The anomaly categories include equipment failure, environmental fluctuation, and data acquisition error.

5. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 4, characterized in that, Incremental learning and dynamic expansion are performed on the random forest model by adding category output nodes and using the new data for incremental update.

6. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, characterized in that, A three-axis gyroscope, a three-axis accelerometer, and a three-axis magnetometer are integrated in the motion attitude acquisition sensor for monitoring the deformation and position of the GIL pipe.

7. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, characterized in that, The sensing elements in the device are arranged in a vertical hierarchical manner, specifically: The motion attitude module and the piezoelectric sensor are arranged on the bottom layer, and the temperature and humidity sensor and the electrochemical sensor are arranged on the top layer; wherein, the temperature and humidity sensor and the electrochemical sensor are separated by a partition plate.

8. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, characterized in that, A bridge module is also mounted on the A / D conversion module. The input ends of the bridge module are respectively connected to the motion attitude acquisition sensor, the temperature and humidity sensor, the electrochemical sensor, and the piezoelectric sensor, and the output end of the bridge module is connected to the A / D conversion module.

9. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, characterized in that, The electrochemical sensor is a dual-channel electrochemical gas sensor with temperature compensation function, and / or the piezoelectric sensor is a piezoelectric ceramic sensor.

10. The intelligent sensor device for multi-parameter monitoring of GIL through-tubes according to claim 1, characterized in that, The device further includes a photovoltaic module connected to the microcontroller for charging the power supply module.

Images