Intelligent detection method for grease injection effect of strain clamp based on electric-thermal coupling response and AI analysis

Abstract

The invention discloses a strain clamp grease injection effect intelligent detection method based on electric-thermal coupling response and AI analysis, which combines conductive tracer electrical resistance tomography (ERT) and active thermal excitation infrared imaging, verifies the grease injection effect of a strain clamp through electrical and thermal characteristics, can effectively judge whether grease injection in the strain clamp is sufficient or not, and can effectively detect the grease injection effect of the strain clamp. A machine learning algorithm is utilized to establish a dynamic relation model of the grease injection state and the electric-thermal response, accurate positioning and quantitative analysis of defects are realized, the structure of the strain clamp does not need to be damaged, the grease injection effect of the strain clamp can be detected in a short time, and the method is suitable for on-site rapid detection and high in detection efficiency. And the detection efficiency of the strain clamp is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-voltage line tool detection, and in particular relates to an intelligent detection method for grease injection effect of a tension clamp based on electric-thermal coupling response and AI analysis. Background Art

[0002] As a key component for fixing conductors and bearing line tension in transmission lines, the stability of the long-term service performance of the tension clamp is directly related to the safe and reliable operation of the power grid. However, with the continuous deterioration of the service environment, the internal structure of the tension clamp faces increasingly serious risks of corrosion and mechanical damage, posing a significant threat to the safety of the transmission line. Specifically, the infiltration of rainwater or corrosive media will cause corrosion of the steel anchor and steel core, resulting in significant degradation of the mechanical properties of the material. Under the synergistic effect of external loads such as wind load and icing, the risk of conductor falling off is significantly increased, which in turn affects the long-term service reliability of the clamp. In addition, in a low temperature environment, the freezing of water that penetrates into the clamp will induce structural damage such as bulging or even cracking of the clamp due to the volume expansion effect, further exacerbating the safety hazards of the transmission line. Especially under harsh environmental conditions such as high humidity, high salt fog or low temperature, the above risks are further exacerbated. In response to the problem of rainwater or corrosive media infiltration, the power industry generally adopts a cavity internal grease injection strategy to effectively block the intrusion of external media. This technology fills the cavity of the wire clamp with special grease at a high density to form a continuous and dense physical barrier on the surface of the cavity, thereby isolating rainwater and corrosive media from contacting the internal metal parts. It can be seen that the quality of the grease injection process is a key factor in determining the protective effect. Insufficient grease injection will lead to the formation of local unfilled areas inside the cavity, providing a channel for the infiltration of rainwater and corrosive media. At the same time, the presence of pores will destroy the continuity and density of the grease, reduce the integrity of its physical barrier, and thus weaken the protective effect. Therefore, strict control of the grease injection process parameters, including grease injection amount, grease injection pressure and grease injection uniformity, is crucial to ensure that the filling degree reaches more than 95% and avoid pore defects. Only through high-quality grease injection technology can the protective effect of the filling grease be fully exerted and the service life of the tension clamp be effectively extended.

[0003] At present, the detection of fat injection effect in the industry mainly relies on two types of methods, but both have certain limitations. The first method is the process monitoring method, which indirectly evaluates the fat filling degree by real-time monitoring of parameters such as pressure and flow of the fat injection equipment. However, this method cannot directly detect the distribution state of the internal fat body, such as key defects such as bubbles and unfilled areas, and it is difficult to fully reflect the actual fat injection quality. The second method is the offline sampling method, which evaluates the fat injection effect by destructive dissection or X-ray imaging of a small number of samples. Although this method can intuitively detect internal defects, it has significant disadvantages such as high detection cost, low efficiency, and radiation hazards, which makes it difficult to meet the quality control needs in large-scale production. Non-invasive detection technologies, such as ultrasonic detection and infrared thermal imaging, have gradually been explored and applied in many fields. However, these technologies still face the following technical bottlenecks in practical applications: First, the single limitation of physical field detection significantly restricts its detection effect. For example, ultrasonic detection is easily interfered by multiple reflections of metal structures, which makes it difficult to accurately determine the location and type of defects; while infrared thermal imaging is highly dependent on ambient temperature differences and is not sensitive enough to deep defects. Secondly, the lack of quantitative analysis capabilities further limits the application value of these technologies. Existing methods mostly rely on manual experience and judgment, which makes it difficult to accurately locate tiny defects.

[0004] Therefore, developing an efficient, non-destructive detection method that can comprehensively evaluate the quality of injected fat has become a technical problem that the industry urgently needs to solve. As inherent properties of materials, electrical and thermal properties are highly sensitive to the distribution of the medium, and the two show good complementarity. Among them, the conductive property can effectively reflect the macroscopic distribution of liposomes, while the thermal conductivity can characterize the microscopic contact state. Based on this principle, the contrast of electrical signals can be enhanced by introducing conductive tracers, and the signal-to-noise ratio of thermal imaging can be improved by combining active thermal excitation, thereby constructing a dual-modal detection system. Furthermore, with the help of artificial intelligence technology, especially deep learning algorithms, multi-source data can be fused and modeled, which can break through the limitations of traditional threshold criteria and achieve a leap from "empirical judgment" to "data-driven". This method can not only improve detection accuracy and reliability, but also provide scientific basis and technical support for the intelligent operation and maintenance of power equipment, and promote the industry to develop in a more efficient and intelligent direction. Summary of the invention

[0005] The purpose of the present invention is to provide an intelligent detection method for the grease injection effect of a tension clamp based on electro-thermal coupling response and AI analysis, so as to solve the problem that the quality of the grease injection inside the tension clamp cannot be accurately determined in the prior art.

[0006] To achieve the above object, the technical solution of the present invention is: an intelligent detection method for the grease injection effect of a tension clamp based on electric-thermal coupling response and AI analysis, comprising the following steps: Manufacture strain clamps with different filling types of artificial defects, collect images of the strain clamps with artificial defects through X-ray real-time imaging technology, use the defective strain clamps as reference models, and use the fully filled strain clamps as standard models; S1.1 Add the filling grease with a small amount of conductive nanomaterials into the cavity of the crimped strain clamp through a grease injection device; S1.2 Arrange multiple groups of microelectrodes around the outer surface of the grease-injected strain clamp to form an ERT sensor ring; S1.3 Apply a small alternating current to the ERT sensor ring, measure the boundary voltage value, and invert the distribution of the internal conductivity based on Ohm's law and electromagnetic field theory; S1.4 Defect identification: There is a conductivity difference between the greased area and the non-greased area, forming different conductivity distributions, and judge the quality status through the standard model and the reference model; S2.1 Arrange electric heating sheets outside the grease-injected strain clamp for local heating; S2.2 Use an infrared image to collect the heated strain clamp to generate a thermal imaging image; S2.3 Defect identification: There is a difference in thermal conductivity between the greased area and the non-greased area, forming different thermal distribution images, and judge the quality status through the standard model and the reference model; S3.1 Input the data of ERT and thermal imaging into a convolutional neural network to extract multi-modal features; S3.2 Train the model based on a large number of laboratory samples, learn the mapping relationship between the electro-thermal signal and the greasing state, and judge the internal quality status; S3.3 Output a quantitative report on the grease filling rate, void position and size based on AI, and mark the defect area through a visualization interface; S4 Use portable ERT and infrared imaging equipment to complete the detection of the strain clamp within 10 minutes, and the AI model analyzes and marks the defect location in real time.

[0007] Preferably, the filling grease is a carbon nanotube conductive nanomaterial with a mass percentage of 0.1% to 0.5%, and the rest is a conductive paste.

[0008] Preferably, the filling grease is a graphene conductive nanomaterial with a mass percentage of 0.1% to 0.5%, and the rest is a conductive paste.

[0009] Preferably, in S1.1, a high-precision pressure sensor is used to detect the change of the grease injection pressure in real time, and the grease injection speed is feedback-regulated in combination with a flow meter to generate a pressure-time curve as a preliminary quality index for identifying abnormal grease injection.

[0010] Preferably, the strain clamps with artificial defects include fully filled standard samples, partially non-greased samples with a simulated filling rate of 50% to 80%, and defective samples containing air holes and cracks.

[0011] The advantages of the present invention are as follows: 1. By combining conductive tracer electrical resistance tomography (ERT) with active thermal excitation infrared imaging, the greasing effect of the strain clamp is verified through dual verification of electrical and thermal characteristics, which can effectively judge whether the strain clamp is fully greased, thus ensuring the safe and reliable operation of the power grid; 2. Using machine learning algorithms to establish a dynamic relationship model between the greasing state and the electro-thermal response, realizing precise positioning and quantitative analysis of defects; 3. There is no need to damage the structure of the strain clamp, and the greasing effect of the strain clamp can be detected in a short time, which is suitable for on-site rapid detection and improves the detection efficiency of the strain clamp. Specific embodiments

[0012] The present invention will be further described below in conjunction with embodiments.

[0013] Embodiment 1 Manufacture strain clamps with different types of artificial defects, collect images of the strain clamps with artificial defects through X-ray real-time imaging technology, use the defective strain clamps as reference models, and use the fully greased strain clamps as standard models; S1.1 Add filling grease with a small amount of conductive nanomaterials into the cavity of the crimped strain clamp through a greasing device, and the filling rate is 50%; S1.2 Arrange 16 microelectrode silver electrodes around the outer surface of the greased strain clamp. The diameter of the electrodes is 5 mm, and the gap between the electrodes is 20 mm to form a circular ERT sensor array; S1.3 Apply a small alternating current to the ERT sensor ring, apply an alternating current with a frequency of 1 kHz and an amplitude of 10 mA to the electrodes through a multi-channel current source, measure the boundary voltage value, invert the distribution of the internal conductivity based on Ohm's law and electromagnetic field theory, and generate a conductivity distribution map of ERT; S1.4 Defect identification: There is a conductivity difference between the greased area and the non-greased area, forming different conductivity distributions. The quality status is judged through the standard model and the reference model. The conductivity of the standard greased area is extremely high, and the ERT image shows an extremely high conductivity area (warm color, such as red). The conductivity of the area with a filling rate of 50% shows a local average effect, and the ERT image may show an intermediate color (yellow or green). The conductivity of the greased area is 100%, and the conductivity of the non-greased area is 50%; S2.1 Install a flexible electric heating sheet on the surface of the greased strain clamp, apply a pulsed heating for 5 seconds to stimulate the internal heat conduction process; S2.2 Use a high-resolution infrared thermal imager to record the change of the surface temperature field of the strain clamp and generate a thermal imaging image; S2.3 Defect identification: There is a difference in thermal conductivity between the greased area and the non-greased area, forming different thermal distribution images. The quality status is judged through the standard model and the reference model. The greased area has a high thermal conductivity and shows rapid thermal decay. The air in the non-greased area has a low thermal conductivity and shows the characteristic of delayed heat dissipation. The greasing defect of the strain clamp is identified by comparing with the standard model. S3.1 Input the conductivity distribution map and infrared thermal imaging map data of the obtained ERT into the convolutional neural network, and extract spatial features and temporal features respectively. S3.2 Train the model based on a large number of laboratory samples, learn the mapping relationship between the electro-thermal signal and the greasing state, and judge the internal quality status. S3.3 Based on the AI output, a quantitative report of the greasing filling rate of 50%, the void position and size is generated, and the defect area and the non-greased area are marked through the visualization interface. S4 Use portable ERT and infrared imaging equipment to complete the clamp detection within 10 minutes, and the AI model analyzes and marks the defect location in real time.

[0014] It can be judged from this that the strain clamp with a greasing filling rate of 50% has obvious defects, and there are obvious conductivity differences between its greased area and non-greased area; the greased area has a high thermal conductivity and shows rapid thermal decay, while the air in the non-greased area has a low thermal conductivity and shows the characteristic of delayed heat dissipation. This strain clamp has obvious defects and should be replaced in time to ensure the safe and reliable operation of the power grid.

[0015] Embodiment 2 Manufacture strain clamps with different types of artificial defects, collect images of the strain clamps with artificial defects through X-ray real-time imaging technology, use the defective strain clamps as the reference model, and use the fully greased strain clamps as the standard model. S1.1 Add the filling grease with a small amount of conductive nanomaterials into the cavity of the crimped strain clamp through the greasing device, and the filling rate is 98%. S1.2 Arrange 16 microelectrode silver electrodes around the outer surface of the greased strain clamp. The diameter of the electrode is 5 mm, and the gap between the electrodes is 20 mm to form a ring-shaped ERT sensor array. S1.3 Apply a small alternating current to the ERT sensor ring, apply an alternating current with a frequency of 1 kHz and an amplitude of 10 mA to the electrodes through a multi-channel current source, measure the boundary voltage value, and invert the internal conductivity distribution based on Ohm's law and electromagnetic field theory, and generate a conductivity distribution map of the ERT. S1.4 Defect Identification: There is a conductivity difference between the greased area and the non-greased area, forming different conductivity distributions. The quality status is judged through the standard model and the reference model. The conductivity of the standard greased area is extremely high, and the ERT winter image shows an extremely high conductivity area (warm color tone, such as red). The area with a filling rate of 98% shows a uniformly high conductivity area as a whole (bright red or orange), which is almost visually indistinguishable from the 100% greased area. If there are scattered micron-sized bubbles or unfilled points, the ERT image may show sporadic cold color spots in the warm color background. The conductivity of the greased area is 98%, and the conductivity of the non-greased area is 2%; S2.1 Install a flexible electric heating sheet on the surface of the strain clamp after greasing, and apply a pulsed heating for 5 seconds to stimulate the internal heat conduction process; S2.2 Use a high-resolution infrared thermal imager to record the change of the surface temperature field of the strain clamp and generate a thermal imaging image; S2.3 Defect Identification: There is a difference in the thermal conductivity coefficient between the greased area and the non-greased area, forming different thermal distribution images. The quality status is judged through the standard model and the reference model. The greased area has a high thermal conductivity coefficient and shows a rapid thermal decay. The air in the non-greased area has a low thermal conductivity coefficient, showing the characteristic of delayed heat dissipation. The greasing defects of the strain clamp are identified by comparing with the standard model; S3.1 Input the obtained conductivity distribution map of ERT and the data of the infrared thermal imaging map into a convolutional neural network, and extract spatial features and temporal features respectively; S3.2 Train the model based on a large number of laboratory samples, learn the mapping relationship between the electro-thermal signal and the greasing state, and judge the internal quality status; S3.3 Based on the AI output, generate a report on the greasing filling rate of 98%, the void position and size quantification, and mark the defect area and the non-greased area through the visualization interface; S4 Use a portable ERT and infrared imaging device to complete the detection of the clamp within 10 minutes, and the AI model analyzes and marks the defect position in real time.

[0016] It can be judged therefrom that the strain clamp with a greasing filling rate of 98% has no obvious defects. Through the conductivity, it is judged that the non-greased area of this strain clamp is extremely small and can be used continuously; the overall thermal decay amplitude of the strain clamp is the same, and no obvious difference is found, so it is judged that this strain clamp has no obvious defects and can still be used continuously.

[0017] The above are only the preferred examples of the present invention. The technical solutions of the present invention are not limited thereto. It should be pointed out that for those of ordinary skill in the art, under the technical inspiration provided by the present invention, as the common general knowledge in the art, other equivalent deformations and improvements can also be made, which should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent detection method for the greasing effect of strain clamps based on electro-thermal coupling response and AI analysis, characterized in that It includes the following steps: Manufacture strain clamps with different filling types of artificial defects, collect images of the strain clamps with artificial defects through X-ray real-time imaging technology, use the defective strain clamps as reference models, and use the fully filled strain clamps as standard models; S1.1 Add the filling grease with a small amount of conductive nanomaterials into the cavity of the crimped strain clamp through a grease injection device; S1.2 Arrange multiple groups of microelectrodes around the outer surface of the grease-injected strain clamp to form an ERT sensor ring; S1.3 Apply a small alternating current to the ERT sensor ring, measure the boundary voltage value, and invert the distribution of the internal conductivity based on Ohm's law and electromagnetic field theory; S1.4 Defect identification: There is a conductivity difference between the greased area and the non-greased area, forming different conductivity distributions, and judge the quality status through the standard model and the reference model; S2.1 Arrange heating sheets outside the grease-injected strain clamp for local heating; S2.2 Use an infrared image to collect the heated strain clamp to generate a thermal imaging image; S2.3 Defect identification: There is a difference in the thermal conductivity coefficient between the greased area and the non-greased area, forming different thermal distribution images, and judge the quality status through the standard model and the reference model; S3.1 Input the data of ERT and thermal imaging into a convolutional neural network to extract multi-modal features; S3.2 Train the model based on a large number of laboratory samples, learn the mapping relationship between the electro-thermal signal and the greasing state, and judge the internal quality status; S3.3 Output a quantitative report on the grease filling rate, void position and size based on AI, and mark the defect area through a visualization interface; S4 Use portable ERT and infrared imaging equipment to complete the detection of the clamp within 10 minutes, and the AI model analyzes and marks the defect position in real time.

2. The intelligent detection method for the grease injection effect of the strain clamp based on electro-thermal coupling response and AI analysis according to claim 1, wherein: The filling grease is a carbon nanotube conductive nanomaterial with a mass percentage of 0.1% - 0.5%, and the rest is conductive paste.

3. The intelligent detection method for the injection effect of strain clamps based on electro-thermal coupling response and AI analysis according to claim 1, characterized in that: The filling grease is a graphene conductive nanomaterial with a mass percentage of 0.1% - 0.5%, and the rest is conductive paste.

4. The intelligent detection method for the grease injection effect of the strain clamp based on electro-thermal coupling response and AI analysis according to claim 1, characterized in that: In S1.1, a high-precision pressure sensor is used to detect the change of the grease injection pressure in real time, and the grease injection speed is feedback-regulated in combination with a flow meter to generate a pressure-time curve as a preliminary quality index for identifying abnormal grease injection.

5. The intelligent detection method for the grease injection effect of strain clamps based on electro-thermal coupling response and AI analysis according to claim 1, characterized in that: The strain clamps with artificial defects include fully filled standard samples, partially non-greased samples with a simulated filling rate of 50% - 80%, and defective samples containing air holes and cracks.